Welcome to the Eleventh

EFEE World Conference

on Explosives and Blasting
Proceedings

Edited by
R. Holmberg et al

To view individual papers, simple click on

the relevant description in the following
Table of Contents section

-1-
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Table of Contents

Foreword by the Chairman of the Technical Committee 4

Acknowledgement 5

1. Blast design management

Blasting a huge overhanging rock formation in an emergency 8
J. Wallace & W. Gates

Blast vibration optimisation at a limestone quarry by using E*STAR electronic detonators

and advanced modelling methods 22
V. Kala & E. Kubat

Impact of fragmentation on the operational conditions of loading and haulage equipment

- a case study 28
V. Miranda, F. Leite, A. Oliveira, T. Kouvonen & B. Saraiva

Modelling and determining the effective parameters in ore dilution in Sarcheshmeh copper mine 50
V. Kala, E. Kubat, G.R. Jabalbarzai, A. Nikkhah, G.R. Saeedi, J. Hakimi
& P. Tahernejad Javazm

Understanding single salt, dual salt, and multiple salt bulk emulsions and emulsion blends 62
D.S. Scovira, B. Victor & Q. Steyl

2. Blasting work experiences

Domestic water well decommissioning during Covid 72
J. Wallace & C. Breeds
Wireless blasting techniques to improve ore extraction and remove people from dangerous
areas in secondary stopes 80
C. Walker, J.A. Kreivi, J.S.C. Marijnissen & S. Barnett

3. Blast vibration and seismology

A comparison of recent and previous pipeline blasting studies 90
C. Dowding, T.A. Davidsavor, D.A. Provost & Y. Gou

Building damage due to vibration from rock blasting – instrumented test in Norway 100
K.M. Norén-Cosgriff, C. Madshus & N. Ramstad

Effects on buildings from blast induced vibrations – a numerical pre-study 106

P.D. Folkow, M. Johansson, J. Leppänen, G. Edefors & A. Naseri

-1-
Elevated pipeline response to shaft blasting 116
C.T. Aimone-Martin, J. Redyke, B.M. Meins & O.C. Meins

Magnitude 2.8 Earthquake Caused by 90 Pounds (41 kg) Per Delay! Not So Fast 126
T.A. Davidsavor, P.E. & J. Aiken, R.E. Burnham, J.K. Ratliff, I.G. Wong
& C.T. Aimone-Martin

New approaches for evaluating building response to blast vibrations 140

M.I. Álvarez-Fernández, M.B. Prendes-Gero, L. Conde-Fernández
& C. González-Nicieza

Proposed methodology to reduce microfractures in working stopes utilising vibration modelling 152
N. Valencia

Rock blast induced pore pressures in quick clay 164

J. Johansson, S. Gjengedal, J. Park, K.H. Andersen, C. Madshus, & S. Rønning

4. Construction, mining and quarrying (blasting)

A study of the fragmentation of rock masses by explosives
- Application to four Belgian case studies 174
J-P. Tshibangu, N. Gonze, P. Van Landschoot, V. Vanoverbeke & L. Rosini

Rock mass characterisation and ore grade assessment through drill monitoring
for open pit applications 184
J. Navarro, P. Couceiro, J.A. Sanchidrián, P. Segarra, T. Seidl, P. Hartlieb & P. Schimek

Studying blast induced rock movement in open-pit mining with small benches 196
I. Koprev, I. Garkov, D. Kaykov, & D. Terziyski

5. Demolition blasting
Analysis of air blast overpressures within a negative pressure boundary 212
F.B. Kuhnow

Demolition and reuse of a 152 m chimney at Amager Resource Center, Copenhagen 224
C.A. Andersen, S. Qvist, P. Brask, E.C. Thyrring & J. Schneider

Technical and safety aspects for the demolition of a chimney with a height of 150 m
in a difficult location 236
R. Laszlo, E. Ghicioi, C. Jitea, A. Nicola & L. Miklos

6. EU directives & harmonisation work

Implementation of track & trace solutions in existing production environments 250
F. Hirthammer

PECCS – Pan-European Competency Certificate for Shot firers /blast designers 256
T. Tuuna

-2-
7. Health, safety & environment
Cold case investigation - Who blasted the transmission tower and killed four people? 264
M. Ruspeckhofer

Key learning from Beirut ammonium nitrate accident

- Evaluation of warehouse safety management in Indonesia 272
A. Wibowo, & A. Sutoyo

The effects of dynamic pressure in blasting 280

G. Cavanough & A. Torrance

8. New applications and training

Blasting activities with fast-combusting energetic compositions in dimension stone
quarries and for demolition of structural elements in civil engineering 290
P. Shishkov, N. Stoycheva & V. Penev

Optimising blast hole loading with MWD and 3D image analysis 300
B. Gyngell, T. Buschjost, T. Worsey & G. Diehr

The art of detonation – energetic creation 308

P. Bosch

9. Technical development
Automatic 3D Fragmentation Analysis from Drone Imagery 316
A. Gaich, M. Pötsch, & M. Baumgartner

Automatic blast design based on an autonomous drone 326

T. Bernard

Density monitoring of bulk emulsion explosives in confined boreholes 336

F.F. Sitonio & J.C. Koppe

Fundamental study of reactive ground chemistry 344

G. Conroy & K. Robinson

Ground improvement blasting - further development of the BDR classification scheme 364
T.N. Little

Implementation of Technology to Fully Digitalize the Drill & Blast Process in a Remote Setting 384
A. Bihlar, L. Bermingham & R. Farnfield

Multiple blasthole dynamic pressure (MBDP) model to alleviate dynamic pressure in blastholes 396
R. Yang

Optimisation of surface blasting by photogrammetric 3D models from drone imagery 408

A. Gaich, M. Pötsch & M. Baumgartner

Thermo-mechanical behaviour of high strength concrete containing

polypropylene under thermal loading 416
Z. Kurdo & L.G. Clough

-3-
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Foreword by the Chairman of the Technical Committee

As Chairman for the EFEE Technical committee I would like to thank all members of the Organisation
and Technical Committee for their dedication and commitment to this 11th EFEE World Conference on
Explosives and Blasting held in Maastricht, Netherlands, 2022.

Together with my colleagues from the Technical Committee we have selected contributions you will
find in these proceedings. We had many abstracts to evaluate and we have been very much impressed
by the high quality of the papers submitted. I would like to thank all the authors for their time and effort
to submit the technical papers presented in the proceedings.

The papers in these proceedings cover many important subjects like EU Directives and Harmonisation
Work, HSE, Blast Design Management, Blasting Work Experiences, New Applications and Training,
Blast Vibration and Seismology, Construction Mining and Quarrying, Demolition Blasting and
Technical Development that influences our industry. This makes the EFEE World Conference on
Explosives and Blasting so important and valuable, namely to bring together manufacturers, shotfirers,
explosives engineers, managers and researchers from the blasting industry to discuss new developments
in our business.

As the papers published in these proceedings of the 11th EFEE World Conference on Explosives and
Blasting show, the conference has fulfilled its mandate admirably and the material will be a good
platform for additional challenging blasting engineering developments and experiences that we can
share in future EFEE conferences. I encourage all of you to take the time to document your coming
experiences and share the findings with all of us by sending in abstracts to the coming EFEE
conferences.

EFEE and Nederlandse Vereniging van Explosieven Experts (NVEE) are very proud of arranging this
event in 2022 and I trust that you will enjoy the conference in Maastricht, Netherlands.

Roger Holmberg
EFEE Technical Committee Chairman and EFEE Secretary General

-4-
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Acknowledgement

As Chairman for the EFEE Technical Committee I would like to thank all members of the EFEE
Conference Committee and the EFEE Technical Committee for their interest, dedication and
commitment to this 11th EFEE World Conference on Explosives and Blasting held in Maastricht, 2022.

Specifically, I would like to thank the EFEE Technical Committee for their help with the selection
procedure and with the editing of the various papers.

Roger Holmberg, EFEE Technical Committee Chairman.

Technical Committee
Roger Holmberg, Sweden (Chairman)
Johan Finsteen Gjødvad, Denmark
Robert Laszlo, Romania
Joerg Rennert, Germany
Janeen Smith, USA
Jacob Uittenboogaard, Netherlands

Conference Committee
Igor Kopal, Slovakia (Chairman)
Doru Anghelache, Romania
Alan Dolan, Ireland
Johan Finsteen Gjødvad, Denmark
José Carlos Gois, Portugal
Bernard Vercouteren van den Berge, Netherlands
Roger Holmberg, Sweden
Jari Honkanen, Finland
Donald Jonson, Sweden
James Tyler, UK

-5-
 1. Blast design management

-6-
-7-
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Blasting a huge overhanging rock formation in an emergency

J. Wallace
Wallace Technical Blasting, Woodland, Washington, USA

W. Gates
McMillen Jacobs Associates, Seattle, Washington, USA

ABSTRACT: On 3 and 10 July 2020, two large sequential rockslides fell near Riggins, Idaho, USA onto
highway US-95. It closed the state’s only major north-south artery for Idaho, requiring a ten-hour detour
to transport personnel, goods, and services. The blast was designed to drop the overhanging rock mass
while protecting the remaining backwall, and while keeping the material from spilling into the Little
Salmon River just across the highway. All this work took place as emergency work during Covid-19
restrictions and during the time of massive wildfires in the region. Factors critical to project success
included Idaho Transportation Department (ITD) engaging experts immediately to address the
emergency: a geological engineer with rock engineering experience and Blaster-in-Charge with extensive
experience in technical blasting. Blasting was safely and successfully completed on 28 August 2020, the
muck pile landed, as designed, out of the river. The project was completed, and the road repaved in
November 2020.

1 INTRODUCTION points needed to monitor slope movement. Based

on our assessment, US-95 was reopened
1.1 Project location temporarily using a shoofly to bypass the slide
rock debris and under a diligent slope-monitoring
This project is approximately 430 km north of protocol. Traffic was limited to daytime only.
Boise, Idaho, in the Northwest USA. This steep canyon along the Salmon River near
Riggins is historically prone to rockslides because
1.2 Geologic history of the complex structural geology. Rock mass
consists of strong hard green serpentinite with
On 3 July 2020, a rockslide occurred on a slope weak bands of schist. The slope is about 76 m tall
along US-95, approximately 5 miles (8 km) south and 244 m long. ITD has reported ravelling and
of Riggins, Idaho. The slide narrowly missed rockslides from this location for over 60 years.
vehicles, and the debris-covered highway was Our first time responding to a rockslide on this
closed to traffic. The Idaho Transportation slope was in December 2012. At that time, MJA
Department (ITD) immediately contacted Bill established and briefed ITD that the entire slide
Gates with McMillen Jacobs Associates (MJA) to mass was potentially unstable because of the
assess the failure and provide guidance on evidence of large tension fractures between several
reopening the road. Gates’ team promptly blocks in the main rock mass, which we identified
mobilised a rope-access geotechnical team to as Blocks A through D (Figure 2). At the direction
assess slope stability and identify key tracking of ITD, MJA developed a blasting design that

-8-
 Figure 1. Project location, Riggins rockslide, 10 July 2020, Riggins, Idaho, USA.

allowed quick removal by cushion (trim) blasting the rest of C failed along the tension fracture,
of one of the most critical and threatening of the fouling the road, exposing the massive underlying
four blocks forming a large-scale wedge on the wedge failure surface, and leaving a large unstable
slope (Block A). The road was reopened to rock mass in the top throat of the wedge.
Christmas traffic. After the initial emergency The MJA geotechnical field team immediately
response in 2012, ITD elected to monitor the slope remobilised to the site to assess the stability of the
for additional movement. ITD located some survey remaining rock mass of about 12,000+ cubic
targets on the rock mass and installed PVC metres, which weighed approximately 31,232
extensometers over the largest tension fracture. metric tons. The team instituted slope monitoring
In Figure 2, the left photo shows the rockslide and safety protocols. MJA also surveyed the
on December 2012, block A removed by trim unstable rock slope using our in-house unmanned
blasting; middle photo portions of Block C failed aerial vehicle (UAV drone) to collect aerial
on 3 July 2020; right photo rockslide on 10 July imagery, then ran the data though photogrammetry
2020, all of blocks B and C and D failed to the software to produce topographic plans and slope
tension fracture noted in December 2012 leaving cross sections for blasting designs, rock anchor
the large rock mass above to be removed by trim layout and drape systems.
blasting. Failure occurred along the tension fracture. The
Eight years later, the investigation of the slope unstable upper rock mass was removed by
after the July 2020 event revealed the original blasting. Note the crushed Connex containers
tension fracture behind Blocks B and D displayed which were used as rockfall barriers. The large
clear evidence of significant opening and boulder above the Connex container is about 10 m
movement since 2012. Between 3 July and 8 July, high. Together the rockslides dumped about
survey data across Block B showed the slope was 15,300 cubic metres of rock onto the highway but
creeping and the tension fracture was continuing to left a large unstable menacing remnant block of
open. Based on our recommendations, ITD closed about 12,000+ cubic metres hanging above the
the road on July 9. On July 10, Blocks B, D, and highway.

-9-
 Figure 2. Riggins rockslide sequence of failures.

Figure 3. The third rockslide event on 10 July 2020.

to gauge the potential movement of the rock mass.

1.3 Monitoring potential movement of the rock On the 3 and 8 of August 2020, MJA worked with
mass Bridge Diagnostics Incorporated (BDI) and
surveyors with David Evans Associates (DEA) to
Because of the concern for personnel safety install high-precision slope monitoring telemetry
working on and below the large unstable rock equipment, which included crack (Figure 4). After
mass, MJA developed a rigorous monitoring plan its installation, field personnel remained to

- 10 -
 Figure 4. Installation of instrumentation (crack gauges, tilt meters, prisms) total station targets, extensometers.

monitor the rock mass 24/7. gauges, tilt meters, present. Meanwhile, MJA initiated the following
prisms, extensometers, and survey targets at key mitigation measures: avoid disturbance of areas
locations on the unstable rock mass. with suspected NOA, initiate dust control, air
monitoring and personnel training. Analytical
1.4 Naturally occurring asbestos (NOA) results demonstrated that NOA was limited to
shear zones and schistose areas within the lower
Twenty-one days after the rockslide, ITD zone of the rockslide scarp. Most samples were
advertised the project and conducted a pre-bid less than 1% regulated NOA. However, one
meeting to select a contractor. The bid was sample out of the excavation area displayed
delayed because naturally occurring asbestos concentrations of NOA > 5%. The contractor was
(NOA) was discovered on the rockslope; a first able to avoid this area and mitigate the large
known case of NOA on ITD projects. NOA is overhanging rock block by drilling and blasting.
different to asbestos associated with building Lessons learned about NOA: metamorphic
materials. There are several fibrous mineral greenstones are suspect; NOA is not a project
species, the regulated species include actinolite killer.
and tremolite. Non-regulated species include
antigorite. Asbestos can cause several forms of 2 TECHNICAL TRIM BLASTING OF THE
cancer. To understand NOA, ITD contracted a LARGE ROCK MASS
subject matter expert (SME) to assess and mitigate
the problem. Guided by the SME, MJA established Mitigation measures for the rockslide included
sampling protocol for seven distinct locations on trim blasting (cushion blasting) of the main rock
the rockslide scarp (Figure 5). Samples were sent mass overhang and other associated rock blocks,
to Asbestos TEM Laboratories of Berkeley, scaling of loose rock and debris, installation of
California for identification of potential asbestos rock anchor reinforcement, hanging of cable-net-
mineral species and quantification of the amounts drape and pinned mesh and installation of a

- 11 -
 Figure 5. Jim Struthers with MJA using rope access safety techniques sampling for NOA at the contact between
the serpentinite rock and the schistose rock units. Photo on the left is a nonregulated fibrous mineral antigorite
like the regulated NOA species actinolite and tremolite.

rockfall fence at the toe of the slope in the main company, Wallace Technical Blasting, Inc., had a
fall line. contractor’s license for private works in Idaho and
this project required a public works contractor’s
2.1 Decision to blast license. During the few days it took Jerry to
complete the required training and pass the test for
After the emergency assessment of the menacing the additional license, ITD had the rock in the
overhanging block, Gates immediately contacted roadway blasted and removed by a local operator.
Jerry Wallace to join the team as the blaster-in- In the meantime, Gates and crew developed a
charge. Jerry had a wealth of experience removing conceptual blasting plan as a starting point for
similar overhanging rock blocks on other projects. consideration to remove the block.
Therefore, the goal was to come up with a blasting The overall contract was awarded on 17 August
plan to remove the large unstable overhanging 2020 to Scarsella Brothers, Inc. of Seattle,
rock that was threating public safety. It was Washington for $3 million. Typically, the blaster-
concluded the best way to remove the threat was in-charge (BIC) and the drilling for the blasting
by trim blasting in one shot. holes would be subcontracted under the general
Jerry was available but unfortunately his contractor. However, because this was a special

- 12 -
technical blasting problem, Wallace contracted intersected forming a plunge line and a classic
directly to ITD. The hybrid contract meant that the wedge for sliding failure.
general contractor would be hiring a drilling
contractor that could drill to the depths, angle, and 2.3 Blasting design
hole diameters in the blast plan provided by
Wallace as BIC. Ryno Works of McCall, Idaho, After assessing the dimensions of the wedge
was awarded the drilling contract. While forming the overhanging rock mass it was
exceptionally challenging and on a tight schedule established that trim holes would have to be drilled
the Ryno Works performed admirably which went subparallel to the two slide planes to depths
a long way to making the blast a success. exceeding 30 metres. The goal was to release the
rock mass from the slide plane and let gravity
2.2 Onsite inspection for blasting control the debris slide to the bottom of the slope.
Because of the attitude and plunge of the wedge,
Immediately after acquiring the proper license, the left and right trim holes had to be drilled at
Wallace drove to the site and met with Gates and different depths and angles to avoid running into
MJA engineering geologist James Struthers. each other and stay within the rock mass. It was
Conversation with a local ITD maintenance established that production holes would be mostly
personnel indicated that access to the top of the vertical and shorter than the trim holes, with a
slope was possible via rudimentary roads on the single row of angled production holes interspersed
ranch property abutting the top of the slope. between the second row of production holes to
Wallace took advantage of that information and pull the broad ‘nose’ of the mass.
went to the top of the slope in the company of the From a safety standpoint, it was ascertained
ITD person and evaluated the site, access, and that it would also be necessary to fire the blast as a
hazards. It was evident that the overhanging rock single trim shot (cushion blast), on short delay
block would have to be brought down in a single timing using electronic detonators, to ensure that
blast. Benching to bring it down in lifts was not the explosives fired in the rock mass while still on
feasible and was dangerous because of the the wall rather than while the mass was dropping
unstable conditions. Figure 6 displays the down the slope. The production holes would be
dimensions of the overhanging rock block. Note, loaded lightly simply to fracture the mass, while
the rock block rests on two slide planes that the trim row would be hit at about twice the

Figure 6. Face view of the 12,000 cubic metre overhanging block, 80 m above the highway, showing the two
major angles meeting at the plunge line.

- 13 -
 Figure 7. Birds-eye view of production rows and trim rows meeting at the plunge line of the wedge block.

Figure 8. Blasthole layout and angles; angled production holes were drilled between vertical holes in the
second row for safety purposes (slope face at the bottom of drawing).

standard trim blast load per square meter of wall to of the blasting design and Figure 9 displays the
lift the rock off the slide plane and move the cross-sectional view. Figure 10 shows the firing
material off the wall by gravity sliding. sequence in plan-view.
The final blasting concept was drafted on CAD
with the assistance of MJA engineering geologist, 2.4 Blasting challenges and limitations
Thomas Pallua in the Seattle office in preparation
for submittal of the design package to ITD. The shot design for this large rock block presented
Preparation of plans and cross-sections were several blasting challenges and limitations. For
facilitated by using imagery developed by flying a example, we had to remove all the rock material in
UAV drone. At the same time, Wallace oversaw a a single blast because of the instability of
ground crew to lay out the blast pattern on the formation and lack of future access to a new
surface of the rock block. Figure 7 displays the bench. Because the rock mass was unstable and
blast hole lay out. Figure 8 presents the plan view perched on the slide plane, safety for crew and

- 14 -
 Figure 9. Typical cross-section displaying vertical and angled blastholes sub-parallel to the failure surface.

Figure 10. Delay timing showing the ‘fast’ timing to ensure that the explosives fired within the mass.
(Slope face of block is at the bottom of the schematic).

equipment when drilling and loading was crews. The 29 m perimeter holes along trim line
paramount. The block was telemetered for early were tough angles to drill and to keep sub-parallel
warning of any movement. All equipment was to the failure plane. Moreover, we had to shorten
anchored to fixed rock dowels behind the slide the initiation delay-time to ensure all explosives
plane and fall restraints were provided for the fired while the rock mass was still on the wall. In

- 15 -
 Figure 11. Ryno Works drilling under tough conditions, at 40+C with wildfire smoke filling the horizon.
Trim blastholes were drilled at 56 to 60 degrees sub-parallel to the failure surface of the block. Depth of the
blastholes were about 30 m.

addition, multiple priming was required for timing for firing to ensure all charges fired on the
explosive columns over 6 m in length. Finally, we bench or block. Break the production rock without
had to keep the blasted rock material from spilling heaving significantly to keep it out of the river.
into the Little Salmon River. Figure 11 That is, take advantage of gravity. Lightly load the
demonstrates the drilling challenges for this production holes, at near boulder-reduction charge
project. Note the drilling rigs are perched out on weights, about 15-20% of normal bench blast load.
the overhanging block. All drilling rigs were tied Blast the trim wall with enough energy to move
off to a safety bar installed uphill from the block the material slightly, again allowing gravity to take
and the rock block was telemetered for potential over. The charge weight on trim wall was
movement. 130 blastholes were drilled in seven approximately 0.50 kg per square metre of wall, or
days, with more than half drilled sub-parallel to roughly double the usual quantity.
the failure plane of the block.
2.6 Explosives and loading of the blastholes
2.5 Planning the solution
Figure 12 displays the typical loading scheme of
The solution to the blasting design problem was the trim and production blast holes. Explosives
accomplished as follows: First, fire as a trim shot were supplied by a regional major supplier,
(post-split, not pre-split design). Shorten the delay Northwest Energetic Services located in Deer

- 16 -
 Figure 12. Typical loading scheme for trim (cushion) and production blastholes.

Figure 13. Blasthole loading in progress. Two crews on the trim row at the rear of the block, and one crew on
the production holes. Crews on the outer brow of the overhanging rock were tethered off to a safety anchor.

Park, Washington, about a 4 hour drive distant. in the long trim holes, took a substantial crew of
Loading took one full day. Figures 13 and 14 show skilled blasters and helpers. The loading chores
the crews loading the blastholes. The crews loaded were split among three crews, with two crews
±2725 kg of explosives. loading the trim holes. One trim crew was
The loading of the blastholes, with a minimum composed of drilling contractor employees, the
of two primers in each hole and with three primers second composed of employees of the explosive’s

- 17 -
Figure 14. The Shot at 1730 hours on 28 August 2020. Note the catchment berms at the toe of the slope.

Figure 15. The image displays the muck pile immediately following the dust clearing from the blast.
None of the material ended up in the Little Salmon River, a good success.

- 18 -
supplier. Jerry and his team of MJA helpers loaded Figure 16. Crews loading the production holes
the production holes. at the face of the overhanging rock block. Jerry
The pre-determined delay timing was installed and Ethan Guzek are tethered to a safety anchor
into the electronic detonators and the blast was while loading the holes. Alexis Jane with MJA is
ready to fire about an hour later than originally documenting the explosive loads in real time.
scheduled, but well within daylight hours. James Struthers with MJA is supplying the
powder.
2.7 The Shot, 28 August 2020
3 FINAL SLOPE MITIGATION
Once all the blastholes were loaded and ready for
firing, the shot was detonated at 1730 hours on 28 Once the rock slope relaxed after blasting (about
August 2020. Maximum explosives per delay was two days) Scarcella Brothers, the general
65.77 kg. Actual peak particle velocity measured contractor began removing rock debris to reopen
at Night Force Ranch (nearest facility) was 2.0 the temporary shoofly road at the toe of the slope.
mm/sec. Figure 14 is a photo of the shot and Following the blast, MJA personnel assessed the
Figure 15 displays the aftermath of the shot. engineering geology and stability of the blasted
Among other challenges, ITD wanted the shot slope and the extension of the slope to the south
rock to not end up in the Little Salmon River. A using rope access safety techniques (Figure 16).
combination of a catchment berm and shoofly at The teams subdivided the rockslope into zones and
the bottom of the slope coupled with a proper blast identified areas that needed rock scaling, rock
design made that possible. Following the large anchors or drapes. In addition, plans and drawings
blast on 28 August 2020 another small trim blast were developed to implement the mitigation plans.
further south was fired on 22 September 2020 to MJA worked with Rock Supremacy to mitigate the
close out the blasting work. rockslope to accomplish the following: scaling of

Figure 16. Ethan Guzek and Luke Ferguson with MJA assessing the stability of the rockslope using rope
access safety techniques after the primary blast on 28 August 2020.

- 19 -
 Figure 17. Completed mitigation of the rockslope, rockfall fence and US-95 highway, December 2020.

about 775 cubic metres of rock; drilling and company involved from the start to ensure that the
installation of about 1525 metres of rock anchors; geology of the rock mass was characterised
installation of 90 metres of drains; installation of employing special safety equipment and
5575 square metres of wire drape; installation of techniques, properly instrumented and stable
185 square metres of Tecco Mesh and installation enough to allow for the work to be accomplished
of the rockfall fence. Scarsella Brothers moved on the rock block and for traffic to flow below.
about 26750 cubic metres of rock to include the Second, ITD required a qualified blasting
rock berm and repaired US-95. The project was contractor to perform the design, management of
completed on 14 November 2020 except for drilling and performance of the critical blasting.
installation of the rockfall fence which was Third, ITD awarded the general contract to a
compete in December of 2020. Figure 17 shows highly qualified company who managed and
the completed project with the repaired highway, conducted all the other operations, and who
drape on the south end of the slope and rockfall worked efficiently with the blasting contractor, the
fence at the base of the slope. civil and geotechnical engineer.
Fourth, naturally occurring asbestos (NOS) was
4 CHALLENGES AND LESSONS-LEARNED dealt with, but it was not a project killer.
The blasting was safely and successfully
This was a very intense and challenging project.
accomplished through the efforts of all involved,
The emergency required highspeed turnaround of
including keeping the muck pile out of the river.
designs and the designs changed on a continuing
Cleaning up the rock slope after the shot involved
basis. Use of drone imagery was key to assessing
rock scaling, installation of rock anchors and cable
and mapping the rockslope and developing plan
net drape. Once completed the highway alignment
sheets and cross-sections for the blasting,
was corrected and paving completed just prior to
anchorage, and drape designs. Monitoring stability
winter 2020.
of rockslope with telemetry, survey locations and
The Owner, Idaho Transportation Department,
visual observation was key to tracking the
and the general contractor, Scarsella Bros, along
potential movement of menacing block. These
with WTB and MJA, were recently honored with a
observations acted as early warning for protecting
Top Gold Award for Partnering on this
the workers on the drilling rigs and for closing the
challenging project.
road to protect the public. One large trim blast was
key to safely removing large unstable rock block. 6 RECOMMENDATIONS
And finally, naturally occurring asbestos (NOA) When working with unique challenges it is
can create a challenge to the project, but NOA is important to have a team made up of experts in all
not a project killer. the fields that comprise the challenges.
Geotechnical engineering and environmental
5 CONCLUSIONS
firms, excavation contractors, scaling and bolting
Several factors were critical to project success. contractors, drilling contractors, and blasting
First, Idaho Transportation Department (ITD), contractors all played significant roles in making
the Owner, engaged a qualified geotechnical this a successful response to a rockfall emergency.

- 20 -
7 ACKNOWLEDGEMENTS
We extend thanks to following companies and
personnel that were associated with this very
challenging project:

Asbestos TEM Laboratories, Inc.-Mark Bailey

Bridge Diagnostics Incorporated (BDI)
David Evans and Associates, Inc.-Duffy Haggarty
and Darrel Ramus
Erskine Environmental Consulting-Dr. Erskine
McMillen Jacobs – Wm. Gates, Jim Struthers,
Jamie Schick, Ethan Guzek, Luke Ferguson,
Alexis Judy, Justin Reeves, Carol Ravano, Thomas
Pallua
HDR, Inc.-Ty Bardwell and Dustin Lenz
Wallace Technical Blasting-Jerry Wallace
Scarsella Brothers Inc.
Rock Supremacy – Rowen Anderegg and crew
Ryno Works – Ryan Miller and crew
WHPacific, Inc. (now NV5) -Jeremy Robbins and
Dawson Sigman
Idaho Transportation Department-Dan
McElhinney, Doral Hoff, Jared Hopkins, Janet
Zarate, Brian Bannon, Bud Converse, Megan
Sausser, Mel Moore, D2 and D3 Operations staff

- 21 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Blast vibration optimisation at a limestone quarry by using

E*STAR electronic detonators and advanced modelling methods

V. Kala & E. Kubat

Austin Detonator s.r.o., Vsetin, Czech Republic

ABSTRACT: This paper presents a modern blasting approach based on a combination of advanced
software technologies and E*STAR electronic initiation system. A unique geomorphological
phenomenon, historical Koněprusy caves, are situated in Central Bohemia, and in close proximity to one
of the biggest quarries in the Czech Republic, Čertovy schody. To protect the historical caves and their
ecosystems, the ground vibration limits for blasting were set and the stationary vibration monitoring
devices were placed inside the caves. Non-electric detonators were used but vibration results were
unpredictable due to slight timing inaccuracies, fixed delay times, and no vibration analysis. The
innovative blast optimisation software is used to achieve vibration goals as well as to keep the production
balance. Since that time, every single blast has been analysed in detail, and the optimum solution for PPV
a PVS reduction is chosen. Among the major advantages of a complete process was the accuracy and
timing variability of E*STAR detonators.

1 INTRODUCTION So far, this is one of the largest limestone quarries

in the Czech Republic. The average yearly
1.1 Location and History production of limestone products is approximately
2 million tons.
In the 1950s, started the investment plans of
building the limestone processing plant ‘Vápenka 1.2 Koněprusy Caves
Čertovy schody a.s.’ and the quarry ‘Čertovy
schody’. The quarry and the processing plant are Koněprusy Caves are a cave system in the heart of
near the city of Beroun, approximately 40 km the limestone region known as Bohemian Karst.
southwest of Prague (Figure 1). The name derives from the nearby village of
Since 1962, the processing plant and the quarry Koněprusy. With a length of 2 km and a vertical
have been in full operation. The production range of 70 m, it is the largest cave system in
initially took place in a quarry opened at the site of Bohemia. Golden Horse hill conceals the most
the so-called ‘Císařský’ (today known as a quarry extensive cave system in Bohemia, which was
‘Čertovy schody west’). At the end of the 1970s, accidentally discovered after the blasting in a
an opening project for the eastern part of the nearby limestone quarry in 1950, and subsequently
quarry was created, and production started in was made accessible for the public in 1959. The
1987. The Koněprusy deposit consists of blasting times at the quarry are set on the exact
high-quality limestones with a density of 2.6 – hour, which highly depends on the guided tours for
2.71 g/cm3. Limestone products are mostly used visitors, which are available every day from April
for agriculture, as building stone, disinfectant, etc. to November.

- 22 -
 Figure 1. Location of the quarry.

1.3 Blasting Issues blasts, a detailed analysis of the previous non-

The problems with the blast vibrations started in electric blasts was completed. The most important
2014 when the vibration limits were set to a blasting parameters as well as the vibration results
maximum of 3 mm/s. The caves are approximately were taken into account. The following table
300 m away from the west part of the quarry, and represents the short summary of the non-electric
as a result, the quarry management had a blasts (Table 1).
significant issue to consider. Various blasting
parameters were applied to solve the problem, Table 1. Non-electric blasts analysis.
including non-electric detonators in combination
Non-electric blasts E*STAR
with multi-decking, but without any meaningful
improvement.
Number of blasts 22 -
2 TRANSITION TO ELECTRONIC
INITIATION Benches 300, 320, 338 -

At the beginning of 2015, the decision was made Average exp. per
13.25 kg -
to try out electronic detonators from Austin interval
Powder. The most important advantages of the
Distance to caves 346 m -
E*STAR electronic detonators compared to
non-electric detonators are high accuracy, delay Average PPV 2.6 mm/s -
range from 1 – 20,000 ms, minimum delay
increment of 1 ms, reliability as well as multi-level
PPV Range 1.1 – 4.1 mm/s -
detonator verification prior to each blast.
% of blast >= 3
2.1 Initial Plan mm/s
50% -

Before we made any initial plan for the E*STAR

- 23 -
 Figure 2. Signature hole parameters.

According to the initial plan, the first step in a GPS position of each signature hole was
blast vibration analysis is to obtain as much measured together with the GPS position of
information as possible about the local conditions, seismographs in the caves. Packaged cartridge
the rock characteristics, surrounding geology, and explosives were used for the charging, and the
the propagation characteristics of seismic waves. stemming material was gravel (Figure 3). The
After detailed analysis, the charging plan signature holes were blasted individually, so each
combined with the optimum delay times were hole was analysed separately based on the hole
applied to the first electronic blast. That would parameters and the vibrations waveform.
make a proper foundation for the systematic
optimisation, where K and n factors were going to 2.3 Paradigm (Blasting Solutions)
be adjusted depending on the blast results.
Blasting Solutions analytical tools were the
2.2 Signature Hole foundation of the Paradigm, state-of-the-art blast
optimisation software, which is an essential tool
By shooting a single representative borehole used to enhance the successful design and
(Figure 2), a signature waveform can be isolated execution of each blast. Rock properties analysis
for each deck and analysed. Multiple signature and geology-based attribute adjustments provide
holes were blasted on different bench levels. the basis for vibrations optimisation modelling.
Signature hole parameters (number of decks, By importing the seed wave (Figure 4), exact
stemming, a charge per deck, etc.) were different GPS positions of the signature hole and
depending on the bench characteristics, except the seismograph, and blast parameters with holes GPS
diameter, which was 92 mm. positions, the software is calculating the ideal

- 24 -
 Figure 3. Signature hole prior to shot.

Figure 4. Seedwave attributes.

timings to achieve the lowest seismic impact for PPV, PVS, frequencies, perception index, and
the environment and the critical zones. Thousands total blast duration are justsome of the parameters
of combinations allow users to choose the which are taken into account before the final
optimum delay times based on preferences. decision.

- 25 -
3 OPTIMISATION Powder E*STAR electronic initiation system and
22 blasts with non-electric detonators is shown in
3.1 First results the table 2.

Before each analysis, the GPS positions of each

blast hole were measured and imported to the blast
optimising software. After analysing the signature
holes results, the optimum delay time was chosen
based on the software solutions and experience of
the blasters.
In the beginning, to keep continuity and to not
make so many changes, the first optimized blasts
utilised the same charging parameters as well as
number of decks as were blasted with previous
non-electric blasts. Only delay times were Figure 6. Vibration results including the
changed. The average delay between two decks in pattern changes.
the single hole was between 11 – 13 ms, with an
average inter-hole delay around 120 ms, all with
single row blasts. The average explosive amount per interval was
After the first couple of blasts, the results were increased to approximately 18.5 kg. The number
more than satisfying (Figure 5). With 5-7 decks of decks, delay times, and charging plan mainly
per hole, 7.5 – 9 kg of explosives per interval the depends on the blast orientation, distance to
average PPV was reduced to 1.09 mm/s (values measurement points, or receiver priorities.
ranged from 0.97 mm/s to 1.30 mm/s).
4 CONCLUSION

Since 2015, more than 500 blasts have been done.

Blast optimisation software, with specific blast
vibration analysis, was used for each of those
blasts. Several measurement instruments are
constantly monitoring the vibrations which are
kept below the maximum limit of 3 mm/s.
Nowadays, the delay range between the decks in
the single hole is between 6 – 20 ms, while hole to
hole delay is 12-35 ms, depending on the specific
field conditions. The reduced vibrations resulted in
Figure 5. Vibration results non-electric vs electronic. increased blast size from single row to multi-row,
where inter-row delay varies from 65 to 120 ms.
The average blast size has increased by 300%
3.2 Blast pattern changes since the start of the project.
Depending on the specific direction of
The outstanding vibration results made space for propagation (longitudinal, transferal, or vertical)
significant changes in the blasting parameters. and the certain seismograph location, the K factor
From the previous 5-7 decks per hole, the number range is from 200 to 600 while the ‘n’ coefficient
decreased to three decks per single hole. As is 1.5. Due to the fact that the explosive amount
expected, the PPV got higher, though it was still per interval has been almost doubled since 2015
not over the limit (Figure 6). The most important (13.25 kg with non-electric compared to 25 kg
thing at that point was to keep the PPV constant, today), it can be said that the analytical modelling
below the limit, and to avoid unpredicted peaks. and vibration predictions software were the game
By performing the blasting optimisation on changers in the blasting industry.
three different benches, an excellent outcome has As a consequence of using modern blasting
been achieved during the trial period. Regardless tools, quarries are still able to keep production
of the reduction of seismic impact, the quarry level and simultaneously protect the environment.
production and overall blast size were increased. A One of the most important facts, during the entire
comparison between 29 trial blasts with the Austin operation, was the exceptionally user-friendly

- 26 -
 Table 2. Results after trial period.

Non-electric blasts E*STAR

Number of blasts 22 29
Benches 300, 320, 338 300, 320, 338
Average exp. per interval 13.25 kg 18.5 kg
Distance to caves 346 m 343 m
PPV Range 1.1 – 4.1 mm/s 0.76 – 3.7 mm/s
Average PPV 2.6 mm/s 1.8 mm/s
% of blast >= 3 mm/s 50% 10%

E*STAR system and the highly skilled blasting

crew which got used to it in a very short period.
The future of blasting should be built on
sustainable and environmental-friendly usage of
civil explosives, detonators, and blasting
equipment.

- 27 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Impact of fragmentation on the operational conditions of loading

and haulage equipment - a case study

V. Miranda, F. Leite, A. Oliveira, T. Kouvonen & B. Saraiva

O-Pitblast Lda., Porto, Portugal

T. Kouvonen1
FORCIT International Ltd., Hanko, Finland

ABSTRACT: The degree of fragmentation of rock using industrial explosives is one of the most relevant
factors in blasting results. Even if the fragmentation curve obtained after this phase is further altered by a
subsequent crushing process, the particle size after blasting has direct implications on cost, productivity
and the operational conditions of loading and haulage equipment. To make production viable and
correctly organise operations at the different fronts of a mine, it is necessary and essential that the loading
and haulage fleet be stationed along the length of the mine, being dependent on its efficiency, hourly
productivity, number of equipment, and the compatibility between them. Normally the production
plans are often prepared based on deterministic modelling, when in fact they are stochastic events.
Implementing this type of methodology can lead to non-optimal intuitive choices that can undermine
long-term objectives. Based on this, this study was carried out in an open-pit mine where the cycle times
of the loading and haulage equipment were collected through an app, in which the data was processed to
evaluate its statistical behaviour and the respective fleet operational indicators. A queue theory analysis
was applied to the data in order to minimise equipment idleness and find the best configuration that would
maximise the mine hourly productivity.

1 INTRODUCTION In the search for improving the production

process, the queue theory has created an
Due to the opening of the market and global opportunity to eliminate bottlenecks, in the case of
competition, producing more with fewer resources, this study applied to a quarry in Portugal, an
and better quality than your competitors, have adequate haulage model was reached, reducing the
become a major challenge. Creating internal waiting time of trucks near the loading queue and
conditions that ensure the survival of organisations maximising the hourly productivity of the mine.
has become the focus of total quality control. In
this situation, it is essential to implement tools and 2 BACKGROUND
methods that help in the organisations decision
making. 2.1 Fragmentation
One of these methods is the theory of queues,
which emerged in the early twentieth century with Bhandari (1997) defines that the main objective of
the aim of maximising the results of operations fragmentation is to obtain the largest possible
and minimise the complexity existing in volume of rock at minimum cost, guaranteeing the
production scenarios. quantity and quality of the material, so that the

- 28 -
 Figure 1. Existing methods for secondary blasting - (A) Use of explosives; (B) Hydraulic breaker; (C) Drop ball.

profits of the subsequent operations of this process fragmentation is not predicted and controlled, it
are maximised. In blasting, this fragmentation can make a mining project unfeasible.
process occurs when an explosive is inserted into a
borehole and detonated, creating a state of stress 2.2 Effects of blasting on loading and haulage
followed by a violent gaseous expansion that can
reach the order of 10 GPa, which causes the rock Drilling and blasting is the process used within the
to be fragmented and displaced. mining cycle to fragment rock for excavation.
A poorly sized blasting operation may result in Cornelius (2017) states that through efficient
the generation of fine materials and/or also the blasting it is possible to create excellent digging
generation of boulders, which may require costly rates by providing high production cycles between
secondary blasting operations, such as the use of the loading and haulage equipments. The term
explosives, hydraulic breakers or drop balls, ‘Diggability’ refers to the measure of ease of
significantly increasing the overall cost of the excavation under specific operating conditions
blasting operation. (Hall & Khorzoughi 2016).
Floyd (2000) elucidates that optimal blasting The size of the particles generated by the blast
can be achieved if adequate fragmentation results is directly related to the time it takes for the loader
from it, in order to achieve the lowest overall cost to fill the bucket (Giles 2016). The bucket fill
by combining the stages of drilling, blasting, factor can be calculated as follows:
loading, haulage, and mineral processing.

(1)

Where Bucket fill factor; = Volume of

material loaded in bucket (m3); = Nominal
volume of bucket (m3).

Table 1. Diggability classification based on fill

factor (Hall & Khorzoughi 2016).

Figure 2. Mining cost relation (adapted from Material Diggability Bucket Fill Factor
Efficient Blasting Techniques, (Floyd 2000). Easily Digging 95 to 105%
Medium Digging 90 to 100%
Hard Digging 85 to 95%
Authors, like Shim et al. (2009) explain that the Very Hard Digging 80 to 90%
blasting stage is one of the most important in the
entire chain, because even if the fragmentation
curve obtained after this process is altered by In a qualitative way, the bucket fill factor can
comminution in the treatment plant, the particle be interpreted according to table 1:
size after the blasting process will have direct The productivity of the loading equipment is
implications on the productivity cost and directly affected by the uniformity index (Brunton,
efficiency of the processes following this stage. Thornton, Hodson, & Sprott 2003). If the particles
Cunningham (2005) also mentions that if this that make up the muckpile present a high

- 29 -
uniformity index, it means that the particles are In Figure 4, it is visible how particles fit better
all similar, i.e., they have a very narrow into the loader bucket when there are regions of
granulometry, which greatly affects the filling granulometry (low uniformity index) that are
efficiency of the bucket as more voids are created different from more homogeneous granulometry
during loading. This can lead to lower productivity (high uniformity index). Some studies have
as each loading cycle decreases the tonnage shown that reducing the uniformity index from 1.5
loaded. Therefore, for better loading, it is to 0.5 increases the fill factor by 15% (Cottee
advisable that the material being loaded has a 2001).
relatively large particle size range with a low It is worth pointing out that there is a certain
degree of uniformity. This reduces the voids in the limit to which the uniformity index can be
bucket and increases the amount of material reduced, otherwise it may make the loading and
loaded in each cycle. This effect can be seen in haulage operation impractical.
Figure 3 based on the studies of Singh & Haulage is designed to transport multiple
Narendrula (2007): bucket loads of rock. Therefore, this equipment is
better suited for oversized material than the loader
itself. However, if the rock blasting is done poorly
and creates large boulders, it can cause problems
for the haulage equipment.
Owolabi (2019) states that in addition to
ensuring that a mine’s production requirements are
met and production costs are minimised, one of the
biggest challenges for mine designers is making
the right fleet choices. Figure 5 shows the range of
recommended pass numbers for filling a given
compatible truck:
The authors Silva (2009) and Jimeno et al.
(2014) agree if the haulage units are well
dimensioned, the loading of these cyclic machines
should be done with a number of buckets that is
between 3 and 5. In this way queues and idling can
be avoided. The expression below shows how to
obtain this result:

(2)
Figure 3. %Voids (first) and production rate
(second) in relation to the uniformity index
(Singh & Narendrula 2007), adapted. Where Number of buckets (passes).

Figure 4. Effect of ‘n’ in fill factor (Cottee 2001), adapted.

- 30 -
 Figure 5. Number of passes from the loading equipment to the truck, (Peroni 2015).

2.3 Queueing theory

Queue Theory is an analytical method, developed

by Agner, Krarup & Erlang (1909), which studies
the formation of queues and their characteristics
through mathematical formulas. Queues occur in
the production process, whether in the form of
material waiting to enter a machine to be
processed, whether trucks waiting for the
availability of a wheel loader to carry out the Figure 6. Truck and loader queuing system
(Meredith 2012).
loading of a truck. Queues will happen whenever
the demand exceeds the capacity to supply of the
service. However, Moreira (2010) argues that In cyclic queuing system, the haul route of a
queue formation is not always associated with truck can be divided into four parts: being loaded
service capacity, but also the variability in the at the loader position, loaded travelling route,
interval between the arrival and the service time. unloading at the crusher and unloaded travelling
Queue theory aims to provide data that helps in route. These four stages are repeated in sequence
modelling a system to find an ideal balance as shown in the Figure 7:
between the costs of performing a particular
service and the costs of delays caused by queues.
For Prado (1999, p.33) studying the behaviour of
queues is a way to modify systems in which
bottlenecks exist.
2.3.1 Queuing system in mining

In mining operations, queuing systems normally

originate from haulage process when trucks arrive
at the loading position, crushing area or dumping
sites and wait for their turns in line. In these
queuing systems, trucks play a role as the Figure 7. Cyclic queuing system (Meredith 2012).
customers of the system and the servers here are
the loaders or crushers. A basic mining queuing In some mining operations, there are multiple
system included trucks and loader can be loaders working at the same time, the cyclic
illustrated as in Figure 6:

- 31 -
queuing system for these operations can be set of short notations in the form of (a/b/c) where
adjusted as a system with multiple loaders. The each variable represents a characteristic of the
queuing system here is a typical of a queuing system as listed in the Table 3.
system with multiple servers, Figure 8: Symbols ‘a’ and ‘b’ represent both types of
distributions and may contain codes representing
either of the common distributions. Symbol ‘c’ is a
value and is represented with an appropriate
number (when greater than 1 ‘S’ is used) or ∞ is
generally used if there is no limit on the size of the
system or population source.

Table 2. Description of the loading and hauling

machines and their respective capacities.
EQUIPMENT MODEL Nº CAPACITY
Komatsu PC 360 1 2.66 m3
EXCAVATOR Komatsu PC 490 1 2.5 m3
Volvo A30C 1 27 ton.
TRUCK Volvo A30D 1 28 ton.
Figure 8. Cyclic queuing system with parallel Volvo A35D 1 32.5 ton.
loaders (Meredith 2012) Volvo A35E 1 33.5 ton.

2.3.2 Description of the Gouvães da Serra

quarry Table 3. Queuing notation abbreviations (Meredith,
Augusta & May 2012), adapted.
The quarry in this study is located in Portugal and
Symbol Explanation
belongs to the Domingos da Silva Teixeira Group
a Arrival distribution
(DST Group) working on behalf of the Spanish
b Service distribution
energy company Iberdrola Generación. The
c Number of parallel serves
purpose of this enterprise is to supply aggregates
specifically for the construction of the Tâmega
Generating System (SET), one of the largest
Table 4. Distribution abbreviations (Meredith,
hydroelectric projects in Europe. Augusta & May 2012).
The information about the loading and haulage Symbol Explanation
equipment operating at the quarry is in Table 2. M Exponential (Markovian) or Poisson arrival
process
2.3.3 Notation D Deterministic: constant distribution
El Erlang distribution with parameter l
G General Distribution
Queueing processes are often referred to using a

Figure 9. Overview of quarry Gouvães da Serra.

- 32 -
 Figure 10. Screens of app used to collect date.

The loading and transport machines in this outliers were removed to ensure non-dispersion
study both show a Poisson distribution for the and to reach a 95% confidence level of the
recorded random variables of the diggability time validated data. These data, when organized,
and cycle time, as will be shown later in the showed a behaviour that approximates the
article. Knowing that we have two loading probabilistic models, where state that the data
machines (see Table 2) to serve the trucks, then from the experiments describe a Poisson
the notation that describes the system is M/M/S. distribution. Through this distribution, and based
on different blasting, it was finally possible to
3 DEVELOPMENT describe the loading and haulage activity with
more reliable approximations, which facilitated the
3.1 Field Procedure
understanding of each cycle phenomenon as a
To understand what would be the ideal excavation probability distribution.
and haulage fleet that would minimise idleness and Therefore, it can be recognised that they occur
generate good productivity for a given muckpile in a certain interval of time.
generated by a blasting, the O-Pitflow – a Through Table 5 it is possible to verify the
smartphone application developed by O-Pitblast – amount of sample data collected from loading
was used to collect data from the actual fleet. The cycle of both excavator (Komatsu PC 360 NLC
version used, was made to record the arrival and & Komatsu PC 490 NLC). After performing
departure times of the trucks for loading, as well statistical tests and removing outliers it was found
as the diggability time of the excavators. As shown that the diggability time of Komatsu PC 360 NLC
in Figure 10, the application allowed the input of ranges from 2 to 40 seconds, and Komatsu PC 490
the fill factor which was stipulated by the operator, NLC ranges from 7 to 29 seconds per cycle. The
with photos of the bucket whenever possible. figures 11 and 12 confirm through the frequency
Under these circumstances a statistical analysis distribution, the loading time behave for both
was carried out divided into the following parts: excavators. The first excavator showed validated
Observed data and statistical behaviour to data with standard deviation of 5.4 and a mean of
loading/haulage; medium fragmentation size (x50) 19.1 seconds per cycle (λ =17.5 seconds per cycle)
vs. loading time; truck cycle time; and choice the and second a standard deviation of 3.9 and mean
best configuration of loading/haulage equipment. of 17 seconds per cycle (λ =16.5 seconds per
cycle).
3.2 Statistical Analysis
Table 5. Excavator: information about total
Observed data and statistical behaviour to sample collected.
loading/haulage. Exacavator Collected data
Statiscally
validated data
The statistical study was performed to Komatsu PC 360 NLC
Komatsu PC 490 NLC
2031
4253
1950
4002
understand the behaviour of the sample data. The

- 33 -
Figure 11. Relative frequencies and Poisson distribution for Komatsu PC 360 NLC.

Figure 12. Relative frequencies and Poisson distribution for Komatsu PC 490 NLC.

Table 6 shows the amount of collected data as

Table 6. Truck information about total sample collected.
well as validated data related to the cycle time of
Truck
Volvo A30C
Collected data
322
Statiscally validated data
304
trucks equipment.
Volvo A30D 26 25 Figures 13 to 16 shows the relative frequencies
Volvo A35D 247 230 of trucks cycle time (Volvo A30C, Volvo A30D,
Volvo A35E 196 187
Volvo A35D and Volvo A35E respectively).

- 34 -
Figure 13. Relative frequencies and Poisson distribution for Volvo A30C.

Figure 14. Relative frequencies and Poisson distribution for Volvo A30D.

- 35 -
Figure 15. Relative frequencies and Poisson distribution for Volvo A35D.

Figure 16. Relative frequencies and Poisson distribution for Volvo A35E.

- 36 -
 Figure 17. Particle size analysis of one of the muckpiles in this study.

Due to the high variability of the trucks cycle line, we were able to determine the confidence
time, a visual inspection was carried out to remove interval for the constant on the line and thus
the outliers. When analysing the model behaviour, determine the standard deviation of the sample.
it was also confirmed that the data follows a This made it possible to simulate a normally
Poisson distribution. Through this analyse the distributed number with these parameters around
following conclusions could be made: the regression line. This value was then used as the
λ parameter to simulate the loading time of the
− Volvo A30C: cycle time ranging from 4 to 13 excavators. The random relationship obtained by
minutes per cycle (λ = 5.32 minutes per the confidence level allowed a more realistic
cycle); approximation to the phenomenon of cycle activity
− Volvo A30D: cycle time ranging from 4 to 13 of the monitored equipment. Below is represented
minutes per cycle (λ = 5.50 minutes per the regression used to estimate the diggability time
cycle); of the loading machines:
− Volvo A35E: cycle time ranging from 4 to 15
minutes per cycle (λ = 5.68 minutes per
cycle);
− Volvo A35E: cycle time ranging from 4 to 16
minutes per cycle (λ = 5.22 minutes per
cycle);

3.3 Mean particle size & loading time

After the particle size analysis using the WipFrag

software, it was possible to observe that there is a
relationship between the loading time and the
average fragment size (X50).
Following this, an algorithm was developed to
simulate the interaction between these two
variables. Once this information was available, the
objective of the study was to anticipate and predict
loading and haulage times for a given muckpile.
Generated modelling was used to analyse a
linear regression between mean particle size and
loading time. Although the data is not statistically
significant by Fisher's F-test, the literature shows a
strong linear association between these two Figure 18a and 18b. Diagram representing the
variables, as cited by Giles (2016). approximation of a straight line with upper and
On the basis of 𝑋𝑋50 it was possible to estimate lower limitations conferred by a confidence
the average loading time, but not precisely. interval for the normal distribution of
Assuming that the linear regression had normally diggability.
distributed residuals around the point found on the

- 37 -
 The application of the method described, made products applied and the characteristics of the
it possible to calculate the random time as many geology would be constant, i.e. the same situation
times as necessary for different 𝑋𝑋50. The data for all cases. Currently the quarry works as
presented in Figure 18 was based on 6 blasts using follows:
the equipment listed in Table 2.
− Type of explosive: Emulsion
3.4 Truck cycle time − Type of ignition system: Non-electric
− Burden, spacing, stemming, subdrilling and
For the simulation process of the truck cycle time, bench height: 2.8 m, 3.7 m, 1.9 m, 0.9 m and
a simple procedure was elaborated. Once the λ of 10 m
the Poisson distribution and the standard deviation − Drill hole diameter: 76.2 mm
were known after the outlier removal process, a − Type of drill rig: Top hammer drilling
random number was generated to determine what − Powder factor: 0.41 kg/m3
would be the total cycle time of the truck. This − Volume of rock per blast: 18.000 m3
process was done for all the haulage machines. − Rock factor: 14.3

3.5 Simulation Next, the combinations of equipment to be

evaluated were defined, as shown in Table 7:
Initially it was established that the blast design, the Based on the current operational model of the

Table 7. Combinations defined for the simulations of the loading and haulage operation.
#Simulation Nº of Nº of trucks Description
excavators
1 2 4 Two excavators with 2.66 and 3.5 m³ buckets
and four trucks with 30 tons.
2 1 4 A single excavator with a 2.66 m³ bucket and
four trucks with a capacity of 30 tons.
3 1 4 A single excavator with a 3.5 m³ bucket and four
trucks with a capacity of 30 tons.
4 2 4 Two excavators with 3.5 m³ buckets and four
trucks with 30 tons.
5 2 3 Two excavators with 2.66 and 3.5 m³ buckets
and three trucks with 30 tons.
6 2 3 Two excavators with 3.5 m³ buckets and three
trucks with 30 tons.
7 1 3 A single excavator with a 3.5 m³ bucket and three
trucks with a capacity of 30 tons.
8 1 3 A single excavator with a 2.66 m³ bucket and
three trucks with a capacity of 30 tons.
9 1 2 A single excavator with a 2.66 m³ bucket and two
trucks with a capacity of 30 tons.
10 1 2 A single excavator with a 3.5 m³ bucket and two
trucks with a capacity of 30 tons.
11 2 2 Two excavators with 3.5 m³ buckets and two
trucks with 30 tons.
12 2 2 Two excavators with 2.65 and 3.5 m³ buckets
and two trucks with 30 tons.
13 2* 4 Two excavators with 3.5 and 4.5* m³ buckets
and four trucks with 30 tons.
14 2* 3 Two excavators with 3.5 and 4.5* m³ buckets and
three trucks with 30 tons.
15 2 5** Two excavators with 3.5 and 4.5* m³ buckets and
five trucks with 30 tons.
* Larger bucket capacity for simulation effects.
**Extra truck in operation.

- 38 -
quarry, described previously, we sought to find we would reach a situation where the cost per
arrangements that would lead to a reduction in cubic metre of rock would be 2.80€ (0.17€ less
related costs, as well as the best performance for than the current one, resulting in 82,062€/year), in
the same annual production. For simulation addition to enabling greater blasting, since it
purposes, we defined that the trucks (Volvo A30C, would take 17.38 hours to complete the removal of
A30D, A35D and A35E) would have the same the material, making it the best simulation of all if
capacity of 30 tons and for the excavators the two this scenario is possible.
existing ones (Komatsu PC 360 NLC and PC 490
NLC) and one ‘supposed’ to have a 4.5 m³ bucket 4 CONCLUSIONS
were used.
With the help of the O-PitFlow application, it was
3.6 Results possible to collect more detailed information about
the excavation capacity, muckpile loading cycle
Table 8 shows the summary of the outcome of the and the trucks.
simulations indicated previously, assuming an Statistical treatment for both the excavation
annual production of 500,000 m³ with shift costs and transport cycle time data revealed a
equivalent to 1.000€. In the Tables found in good correlation between the time-class
Appendix A, the information and results of each distribution of these parameters and the Poisson
proposed combination are presented in detail. In distribution.
Table 8, under the current conditions (two loading Data collection allowed the application of
machines and trucks), the best alternative would simulation strategies that included various
be the 7th, with savings of 13,474.48€/year. But scenarios to obtain the most viable option for the
when we take into account the work time spent drilling, blasting, loading, and hauling phases.
(10.38 hours more) it would not be interesting, as This type of analysis is of utmost importance to
it could disturb the quarry operations. engineers and technicians who, in their daily work,
If we add the possibility of changes in the try to make estimates based on these types of
capacity of the equipment, preliminarily, a simple operations, and allows them to create a plan
change of the bucket of the Komatsu PC360, if taking into account operational variability. In
possible, from 2.66 m³ to 3.5 m³, would lead to a this study it was seen that it is possible to achieve
reduction in the estimated cost by 38.340€/year in very significant annual savings of more than
the expedient (21.89 versus 23.86 hours), as 80,000€.
explained in the 4th arrangement. For the simulation results presented throughout
If we increase the bucket capacity to 4.5 m³, we this paper we have not taken into consideration the
can arrive at scenarios 13 and 14, where it is extra investment (acquisition), fuel and
notable that a larger number of trucks would maintenance costs for a new truck or for
benefit the operation. This is confirmed by adding increasing the capacity of an excavator.
an extra truck (scenario 15), totalling five, where

Table 8. General summary of the results obtained in the simulation of the equipment combinations.
Time to Total cost Cost per
Nº of Shift Truck Loading Blasting Total cost of Estimated
# remove the of operations cubic meter
shifts cost costs cost cost operations annual cost
muckpile + shifts of rock
1 23.86h 2.98 2 982.64 € 9 571.00 € 5 968.00 € 40 807.68 € 56 346.68 € 59 329.32 € 2.97€/m3 1 483 233.09 €
2 39.46h 4.93 4 932.92 € 15 782.00 € 3 946.00 € 40 807.68 € 60 535.68 € 65 468.60 € 3.27€/m3 1 636 715.03 €
3 30.07h 3.76 3 758.16 € 12 043.00 € 3 007.00 € 40 807.68 € 55 857.68 € 59 615.84 € 2.98€/m3 1 490 396.11 €
4 21.89h 2.74 2 736.04 € 8 775.00 € 5 477.00 € 40 807.68 € 55 059.68 € 57 795.73 € 2.89€/m3 1 444 893.16 €
5 30.69h 3.84 3 835.66 € 9 237.00 € 7 680.00 € 40 807.68 € 57 724.68 € 61 560.34 € 3.08€/m3 1 539 008.61 €
6 28.60h 3.57 3 574.69 € 8 595.00 € 7 150.00 € 40 807.68 € 56 552.68 € 60 127.37 € 3.01€/m3 1 503 184.30 €
7 34.25h 4.28 4 280.66 € 10 277.00 € 3 425.00 € 40 807.68 € 54 509.68 € 58 790.34 € 2.94€/m3 1 469 758.61 €
8 40.79h 5.10 5 098.26 € 12 245.00 € 4 079.00 € 40 807.68 € 57 131.68 € 62 229.95 € 3.11€/m3 1 555 748.71 €
9 53.37h 6.67 6 671.08 € 10 685.00 € 5 337.00 € 40 807.68 € 56 829.68 € 63 500.76 € 3.18€/m3 1 587 519.02 €
10 47.24h 5.91 5 905.42 € 9 458.00 € 4 724.00 € 40 807.68 € 54 989.68 € 60 895.10 € 3.04€/m3 1 522 377.53 €
11 43.05h 5.38 5 381.63 € 8 633.00 € 10 773.00 € 40 807.68 € 60 213.68 € 65 595.32 € 3.28€/m3 1 639 882.91 €
12 43.26h 5.41 5 407.43 € 8 661.00 € 10 818.00 € 40 807.68 € 60 286.68 € 65 694.12 € 3.28€/m3 1 642 352.88 €
13 21.05h 2.63 2 631.25 € 8 458.00 € 5 265.00 € 40 807.68 € 54 530.68 € 57 161.93 € 2.86€/m3 1 429 048.37 €
14 27.15h 3.39 3 394.20 € 8 184.00 € 6 798.00 € 40 807.68 € 55 789.68 € 59 183.89 € 2.96€/m3 1 479 597.15 €
15 17.38h 2.17 2 172.15 € 8 722.00 € 4 345.00 € 40 807.68 € 53 874.68 € 56 046.84 € 2.80€/m3 1 401 170.93 €

- 39 -
REFERENCES Prado, D. S. 1999. Teoria das Filas e da Simulação.
Belo Horizonte: Editora de Desenvolvimento
Bhandari, Sushil. 1997. Engineering rock blasting Gerencial.
operations. Rotterdam Brookfield: A.A.Balkema.
Brunton, I., D. Thornton, D. Hodson, & D. Sprott. Shim, H. J., Ryu, D. W., Chung, S. K., Synn, J. H. &
2003. Impact of blast fragmentation on hydraulic Song, J.J.. 2009. Optimised blasting design for
excavator dig time. Fifth Large Open Pit Mining large-scale quarrying based on a 3-D spatial
Conference 10. distribution of rock factor. International Journal
of Rock Mechanics and Mining Sciences.
Camelo, G., Coelho, A., Borges, A. & Souza, R.
2010. Teoria das filas e da simulacao applicada Silva, V. C. 2009. Apostila de Carregamento e
ao embarque de minerio de ferro e manganes no transporte de rochas. Ouro Preto: Escola de
terminal Maritimo de Ponta da Madeira. XXX Minas da Universidade de Ouro Preto.
Encontro Nacional de Engenharia de Producao
14. Singh, S. & Narendrula, R. 2007. Factors affecting
the productivity of loaders in surface mines.
Cornelius, M. 2017. Mine production improvement International Journal of Mining, Reclamation
through haulage optimisation. Queensland: The and Environment, 14.
University of Queensland.
APPENDIX
Cottee, S. 2001. Impact of fragmentation on truck and Data from the proposed combinations for
loader productivity. The University of simulation are provided on the following pages.
Queensland.

Cunningham, C. 2005. The Kuz-Ram fragmentation

model – 20 years on. European Federation of
Explosives Engineers.

Floyd, J. 2000. Efficient blasting techniques. Blast

Dynamics.

Giles, E. 2016. Improving quarry production rates by

reducing bucket fill time. International Society of
Explosives Engineers.

Hall, R. & Khorzoughi, M.B. 2016. Diggability

assessment in open pit mines: a review. Mining
and Mineral Engineering.

Jimeno, C., Jimeno, E.,. Bérmudez, P. & Degea, A.

2014. Manual de transporte con voloquetes y
diseño de pistas mineras. Madrid: Gráficas Arias
Montano, S. A.
May, M. 2012. Applications of Queuing Theory for
open-pit truck/shovel haulage systems.

Moreira, D. A. 2010. Pesquisa Operacional: Curso

Introdutório. São Paulo: Cengage Learning.

Owolabi, A. 2019. Loading and haulage equipment

selection for optimum production in a granite
quarry. International Journal of Mining Science
(IJMS) 6.

Peroni, R. L. 2015. Introdução da lavra à céu aberto.

Rio Grande do Sul: Universidade Federal do Rio
Grande do Sul.

- 40 -
1. Times and costs resulting from the simulation of the combination 1.
Excavator Truck
Combination 1
2.66 m3 3.5 m3 A30D A30C A30E A35D
71106 39205 28729 27447 26316 27819
Loading (s)
(82.78%) (45.61%) (33.87%) (31.87%) (30.55%) (32.18%)
14794 46756
Idle (s) - - - -
(17.22%) (54.39%)
2403 4262 3005 4201
In queue (s) - -
(2.80%) (4.95%) (3.49%) (4.86%)
54720 54420 56820 54420
Travelling (s) - -
(63.74%) (63.18%) (65.96%) (62.96%)
Average in queue (s) 44 140 14 25 18 26
Average load
214 117 169 163 157 173
completion (s)
Average delivery of
- - 322 324 338 338
material (s)
Total cost (€) 2,386.00 € 3,582.00 € 2,385.00 € 2,392.00 € 2,393.00 € 2,401.00 €
Final Result
Excavators cost Trucks cost Blasting cost Total
5,968.00 € 9,571.00 € 40,807.68 € 56,346.68 €

2. Times and costs resulting from the simulation of the combination 2.

Excavator Truck
Combination 2
2.66 m3 A30C A35D A30D A35E
141871 35922 34353 36031 35565
Loading (s)
(99.86%) (25.38%) (24.18%) (25.35%) (24.97%)
Idle (s) 197 (0.14%) - - - -
51992 50337 52035 50643
In queue (s) -
(36.73%) (35.44%) (36.61%) (35.56%)
53640 57360 54060 56220
Travelling (s) -
(37.89%) (40.38%) (38.04%) (39.47%)
Average in queue (s) 0 308 311 308 303
Average load completion
213 213 212 213 213
(s)
Average delivery of
- 317 354 320 337
material (s)
Total cost (€) 3,946.00 € 3,932.00 € 3,946.00 € 3,948.00 € 3,956.00 €
Final Result
Excavators cost Trucks cost Blasting cost Total

- 41 -
3. Times and costs resulting from the simulation of the combination 3.
Excavator Truck
Combination 3
3.5 m3 A30D A30C A35D A35E
106657 26830 25705 26476 27646
Loading (s)
(98.54%) (24.84%) (23.70%) (24.40%) (25.46%)
1578
Idle (s) - - - -
(1.46%)
28061 27950 27440 28149
In queue (s) -
(25.98%) (25.78%) (25.29%) (25.92%)
53100 54780 54600 52800
Travelling (s) -
(49.17%) (50.52%) (50.31%) (48.62%)
Average in queue (s) 2 167 173 166 164
Average load completion
160 160 159 160 161
(s)
Average delivery of
- 316 338 331 307
material (s)
Total cost (€) 3,007.00 € 3,000.00 € 3,012.00 € 3,014.00 € 3,017.00 €
Final Result
Excavators cost Trucks cost Blasting cost Total
3 00 00 € 12 043 00 € 40 80 68 € 8 68 €

4 Times and costs resulting from the simulation of the combination 4.

Excavator Truck
Combination 4
3.5 m3 3.5 m3 A30C A30D A35E A35D
53326 39089 23343 22390 24304 22378
Loading (s)
(67.67%) (49.54%) (29.60%) (28.37%) (30.80%) (28.25%)
25472 39816
Idle (s) - - - -
(32.33%) (50.46%)
1525 2228 2175
In queue (s) - - 2227 (2.81%)
(1.93%) (2.82%) (2.75%)
54000 54300 52440 54600
Travelling (s) - -
(68.47%) (68.81%) (66.45%) (68.94%)
Average in queue
76 119 9 14 12 14
(s)
Average load
160 117 137 137 139 141
completion (s)
Average delivery of
- - 318 333 300 343
material (s)
Total cost (€) 2,189.00 3,288.00 2,191.00 2,192.00 2,192.00 2,200.00
Final Result
Excavators cost Trucks cost Blasting cost Total
5,477.00 € 8,775.00 € 40,807.68 € 55,059.68 €

- 42 -
5. Times and costs resulting from the simulation of the combination 5.
Excavator Truck
Combination 5
2.66 m3 3,5 m3 A30C A35E A30D
70892 39399 36937 36917 36437
Loading (s)
(64,17%) (35,60%) (33,38%) (33,29%) (32,84%)
39575 71262
Idle (s) - - -
(35,83%) (64,38%)

In queue (s) - - 1843 (1,66%) 2404 (2,17%) 1670 (1,51%)

71880 71580 72840

Travelling (s) - -
(64,96%) (64,54%) (65,65%)
Average in queue (s) 119 213 8 11 7
Average load completion (s) 213 118 169 164 163
Average delivery of
- - 330 318 325
material (s)
Total cost (€) 3 069,00 4 611,00 3 074,00 3 081,00 3 082,00
Final Result
Excavators cost Trucks cost Blasting cost Total
7 680,00 € 9 237,00 € 40 807,68 € 57 724,68 €

6. Times and costs resulting from the simulation of the combination 6.

Excavator Truck
Combination 6
3.5 m3 3.5 m3 A30D A35E A30C
39066 52783 29053 31982 30814
Loading (s)
(37.95%) (51.26%) (28.19%) (31.01%) (29.86%)
63885 50189
Idle (s) - - -
(62.05%) (48.74%)
In queue (s) - - 943 (0.91%) 889 (0.86%) 1118 (1.08%)
73080 70260 71280
Travelling (s) - -
(70.90%) (68.13%) (69.06%)
Average in queue (s) 191 151 4 4 5
Average load completion
117 159 138 139 136
(s)
Average delivery of
- - 346 305 315
material (s)
Total cost (€) 4,290.00 2,860.00 2,863.00 2,865.00 2,867.00
Final Result
Excavators cost Trucks cost Blasting cost Total
7,150.00 € 8,595.00 € 40,807.68 € 56,552.68 €

- 43 -
7. Times and costs resulting from the simulation of the combination 7.
Excavator Truck
Combination 7
3.5 m3 A30C A30D A35E
Loading (s) 106397 (86.30%) 36651 (29.76%) 33883 (27.48%) 35863 (29.03%)
Idle (s) 16886 (13.70%) - - -
In queue (s) - 16173 (13.13%) 14961 (12.13%) 18000 (14.57%)
Travelling (s) - 70320 (57.10%) 74460 (60.39%) 69660 (56.39%)
Average in queue (s) 25 71 70 80
Average load completion
160 161 158 159
(s)

Average delivery of
- 308 348 310
material (s)
Total cost (€) 3,425.00 3,421.00 3,425.00 3,431.00
Final Result
Excavators cost Trucks cost Blasting cost Total
3,425.00 € 10,277.00 € 40,807.68 € 54,509.68 €

8. Times and costs resulting from the simulation of the combination 8.

Excavator Truck
Combination 8
2.66 m3 A35E A30C A30D
46917 47903 46682
Loading (s) 141502 (96.37%)
(31.96%) (32.63%) (31.70%)
Idle (s) 5328 (3.63%) - - -
29846 30150 29528
In queue (s) -
(20.33%) (20.54%) (20.05%)
70020 68760 71040
Travelling (s) -
(47.70%) (46.83%) (48.24%)
Average in queue (s) 8 136 133 134
Average load completion (s) 212 213 211 212
Average delivery of material (s) - 318 303 323
Total cost (€) 4,079.00 4,077.00 4,078.00 4,090.00
Final Result
Excavators cost Trucks cost Blasting cost Total
4,079.00 € 12,245.00 € 40,807.68 € 57,131.68 €

- 44 -
9. Times and costs resulting from the simulation of the combination 9.
Excavator Truck
Combination 9
2.66 m3 A30D A35E
Loading (s) 142105 (73.96%) 69633 (36.24%) 72472 (37.65%)
Idle (s) 50022 (26.04%) - -
In queue (s) - 12189 (6.34%) 12615 (6.55%)
Travelling (s) - 110340 (57.42%) 107400 (55.80%)
Average in queue (s) 75 37 37
Average load completion (s) 213 214 213
Average delivery of material (s) - 338 315
Total cost (€) 5,337.00 5,338.00 5,347.00
Final Result
Excavators cost Trucks cost Blasting cost Total
5,337.00 € 10,685.00 € 40,807.68 € 56,829.68 €

10. Times and costs resulting from the simulation of the combination 10.
Excavator Truck
Combination 10
3.5 m3 A35E A30D
Loading (s) 106593 (62.67%) 54595 (32.09%) 51998 (30.53%)
Idle (s) 63483 (37.33%) - -
In queue (s) - 7324 (4.30%) 8038 (4.72%)
Travelling (s) - 108240 (63.61%) 110280 (64.75%)
Average in queue (s) 95 22 25
Average load completion (s) 160 161 159
Average delivery of material (s) - 318 337
Total cost (€) 4,724.00 4,727.00 4,731.00
Final Result
Excavators cost Trucks cost Blasting cost Total
4,724.00 € 9,458.00 € 40,807.68 € 54,989.68 €

- 45 -
11. Times and costs resulting from the simulation of the combination 11.
Excavator Truck
Combination 11
3.5 m3 3.5 m3 A30D A35E
Loading (s) 52971 (34.18%) 39141 (25.21%) 46130 (29.72%) 45982 (29.55%)
102020 116101
Idle (s) - -
(65.82%) (74.79%)
In queue (s) - - 0 0
109080 109620
Travelling (s) - -
(70.28%) (70.45%)
Average in queue (s) 306 348 0 0
Average load completion (s) 159 117 139 137
Average delivery of
- - 329 327
material (s)
Total cost (€) 4,305.00 6,468.00 4,311.00 4,322.00
Final Result
Excavators cost Trucks cost Blasting cost Total
10,773.00 € 8,633.00 € 40,807.68 € 60,213.68 €

12. Times and costs resulting from the simulation of the combination 12.
Excavator Truck
Combination 12
2.66 m3 3.5 m3 A35E A30D
Loading (s) 51722 (33.21%) 53683 (34.45%) 53614 (34.40%) 51791 (33.21%)
104012 102148
Idle (s) - -
(66.79%) (65.55%)
In queue (s) - - 0 0
102240 104160
Travelling (s) - -
(65.60%) (66.79%)
Average in queue (s) 312 306 0 0
Average load completion (s) 155 161 158 158
Average delivery of material (s) - - 301 319
Total cost (€) 6,489.00 4,329.00 4,329.00 4,332.00
Final Result
Excavators cost Trucks cost Blasting cost Total
10,818.00 € 8,661.00 € 40,807.68 € 60,286.68 €

- 46 -
13. Times and costs resulting from the simulation of the combination 13.
Excavator Truck
Combination 13
3.5 m3 4.5 m3 A35D A30D A30C A35E
39059 41098 19462 20050 20627 20018
Loading (s)
(51.54%) (54.18%) (25.67%) (26.38%) (27.05%) (26.21%)
36721 34749
Idle (s) - - - -
(48.46%) (45.82%)
1285 1302 1820 1342
In queue (s) - -
(1.69%) (1.71%) (2.38%) (1.76%)
55080 54660 53820 55020
Travelling (s) - -
(72.64%) (71.91%) (70.57%) (72.03%)
Average in queue (s) 110 104 8 8 11 8
Average load completion
117 123 121 120 121 119
(s)
Average delivery of
- - 342 327 315 328
material (s)
2,106.0 2,119.0
Total cost (€) 3,158.00 2,107.00 2,111.00 2,122.00
0 0
Final Result
Excavators cost Trucks cost Blasting cost Total
5,265.00 € 8,458.00 € 40,807.68 € 54,530.68 €

14. Times and costs resulting from the simulation of the combination 14.
Excavator Truck
Combination 14
3.5 m3 4.5 m3 A30C A35E A30D
39477 41402 27610 28177 25092
Loading (s)
(40.29%) (42.35%) (28.18%) (28.68%) (25.50%)
58507 56351
Idle (s) - - -
(59.71%) (57.65%)

In queue (s) - - 1075 (1.10%) 516 (0.52%) 592 (0.60%)

69300 69540 72720

Travelling (s) - -
(70.72%) (70.79%) (73.90%)
Average in queue (s) 175 169 5 2 3
Average load completion (s) 118 124 121 121 121
Average delivery of
- - 304 300 351
material (s)
Total cost (€) 4,083.00 2,715.00 2,722.00 2,729.00 2,733.00
Final Result
Excavators cost Trucks cost Blasting cost Total
6,798.00 € 8,184.00 € 40,807.68 € 55,789.68 €

- 47 -
15. Times and costs resulting from the simulation of the combination 15.
Excavator Truck
Combination 15
3.5 m3 4.5 m3 A35E A30D A30D A30C A35D
38971 41548 17089 15647 15973 16494 15316
Loading (s)
(62.30%) (66.39%) (27.33%) (24.99%) (25.43%) (26.24%) (24.25%)
23587 and 21031
Idle (s) - - - - -
(37.70%) (33.61%)
2777 1677 1786 1841 1822
In queue (s) - -
(4.44%) (2.68%) (2.84%) (2.93%) (2.88%)
42660 45300 45060 44520 46020
Travelling (s) - -
(68.23%) (72.33%) (71.73%) (70.83%) (72.87%)
Average in queue
71 63 19 13 13 14 14
(s)
Average load
117 124 120 120 120 123 121
completion (s)
Average delivery
- - 298 348 339 332 362
of material (s)
Total cost (€) 2,607.00 1,738.00 1,737.00 1,740.00 1,745.00 1,746.00 1,754.00
Final Result
Excavators cost Trucks cost Blasting cost Total
4,345.00 € 8,722.00 € 40,807.68 € 53,874.68 €

- 48 -
- 49 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Modelling and determining the effective parameters in ore dilution

in Sarcheshmeh copper mine

V. Kala & E. Kubat

Austin Detonator s.r.o., Vsetin, Czech Republic

G.R. Jabalbarzai
Master of Sarcheshmeh Copper Mine Ore Control

A. Nikkhah
Supervisor of mining projects, Tehran, Iran

G.R. Saeedi
Mining Engineering Department, Shahid Bahonar University of Kerman, Kerman, Iran

J. Hakimi
Manager of Sarcheshmeh Copper Mine

P. Tahernejad Javazm
Manager of Sarcheshmeh Copper Mining Engineering

ABSTRACT: The grade control process in mines is aimed at maximising the value of extracted ore
deposits. In this research, the sources of dilution and also strategies in drilling, blasting and loading units
were studied in Sarcheshmeh copper mine. The result of this study was to determine the 18 factors of
dilution in five groups including geometry, uncertainty in knowledge of the in situ ore boundaries, blast
movement, mining errors and weather conditions. Most of the blasting is controlled by bulk blasting type,
but due to blast patterns and time delay sequences, the maximum dilution rate is 13.17%. The parameters,
including stemming length and spacing, increase the dilution, however the hole diameter, specific drilling
and ratio of length to width of blast block decreases the dilution. In order to control dilution, strategies
such as switching blast types, post blasting geological control and loading fleet dispatching will be
effective.
Keywords: blasting, ore dilution, specific drilling, SPSS software, Sarcheshmeh copper mine.

1 INTRODUCTION product sales progressively (Little 2015, Smith et

al. 2015, Domingo et al. 2015, Wang et al. 2011,
Dilution in mines, such as underground and open Najafi et al. 2014). Dilution is one of the mining
pits, disrupts the process of mineral operations, problems that, in addition to economically
and increases all mining costs from extraction to imposing more mining costs, puts a lasting

- 50 -
negative impact on the lives of humans and block of minerals, waste or mineral in a very low
creatures of later generations. The degradation and grade, which is not possible to be separated, which
environmental pollutants resulting from the is known as the inner dilution. The amount of
dilution of mining are considered to be a life- internal dilution varies in different types of ore
threatening factor. In order to prevent the negative deposits. External dilution is referred to the waste
effects, the dilution of a means is not limited to that is extracted from the deposit and extracted
identifying and preventing it in the mining from the mineral block. External dilution is based
industry (Little 2015). on geology, ore formation, drilling and blasting
Mining dilution is one of the most important techniques, extraction operation scale and
factors affecting the economy of mining projects. equipment size (Ebrahimi 2013). In Figure 1, the
While all efforts are made to identify and calculate risk area is shown.
all the cost items of a project, ignoring the
phenomenon of dilution or lack of accurate
calculation of this quantity results in incomplete
economic estimates. This is usually due to lack of
sufficient funding, time to study, and the use of an
inadequate method. Instead of the amount of
dilution in mining studies, it is usually assumed to
be a general improvement, such as 5% for huge
and massive deposits, and 10% for small, tabular
deposits. While these figures may be a good
starting point in the early stages of mining studies,
these numbers are not used in complex issues. In
addition, the direct impact on short-term mining
revenue and also the upswing have led to
significant changes in other factors that in the long
run lead to a reduction in the overall value of the
project. For example, by reducing the effective
capacity of the processing plant, the life of the
mine will be prolonged. The dilution also reduces
the grade of the input product and the output of the
plant. In most cases, lower grade means less
processing recovery. The dilution also increases Figure 1. Risk area (Domingo et al. 2015).
cut-off grades, which in turn reduces the use of a
mine’s storage (Ebrahimi 2013). The aim of the process of mine control is to
The dilution in different mines varies according maximise the amount of minerals extracted for
to the characteristics of the deposits, the processing or crushing for use in downstream
operational and financial aspects of the economic products. This definition provides an
boundary. For example, the shape of the ore, the understanding of the concepts of ‘mine to factory’
height of the benches, the size of the equipment (Little 2015, Agoshkov et al. 1998).
and the market conditions can affect the degree of
dilution in a mine. In order to better produce a 2 THE BACKGROUND
project, dilution studies should be considered as an
inseparable part of any project. Up to now, extensive research has been conducted
in order to comprehensively understand and
1.1 The concept of dilution in mining operations comprehend the dilution phenomenon. A study of
the research conducted in addition to identifying
The dilution refers to waste materials that are the importance of the dilution problem, suggests
mined from the minerals during the extraction that various factors can contribute to dilution. In
process and combined with minerals, which reduce 1966, an analytical model was presented by
the final grade of minerals. These waste materials Augushkös with the assumption that valuable
are combined with mineral matter and sent to the metals in precious minerals were found, valuable
processing plant (Saeedi 2012). metal cassettes in the extruded materials and
In an inorganic block, the dilution occurs in valuable metal cassettes in lateral rocks were
two different regions; sometimes it is found in a associated with the dilution (Agoshkov et al.

- 51 -
1998). In 1986, Liuyong and Zhang Vijing devised floor layer and the slope of the coal bed are
studies on the economic destructive effects of considered (Saeedi et al. 2009). In 2011, Wigong
dilution for thin vein conditions in tungsten mines Wang and colleagues conducted studies on the
in China, and provided some strategies to reduce calculation and management of mineral waste and
dilution for these mining conditions (Yihong & dilution under the conditions of three-dimensional
Zhang 1986). In 2002, Ran presented a model for technical visualisation. In their research, they
the development of open-source mining methods. introduced the principles of operation and the
In this model, the height of the workshop is process of calculating mineral dissipation and
considered as a beam, if the load rate is lower than dilution in the mining process, which is based on
the resistance, then in this case the workshop is the adoption of geological models (Wang et al.
unstable and, based on the depth of fall, the 2011). The dilution control was studied by
amount of dilution is calculated (Ran 2002). In measuring the displacements in the explosion at
2003, Noppe presented a method for behavioural the New Hamad AHFU gold mine in Ghana. The
analysis, reporting and dilution control in coal purpose of this study was to investigate the effect
underground mines (Noppe 2003). of explosion displacements on the control of its
In 2004, Chugh et al. considered a dilution dilution and its reduction (Engmann et al. 2013).
model to study the impact of recovery on In 2014, Tommila explored the effect of
extraction speed, cost and profits from a tonne of dilution profitability. The author believes that
coal mine in Illinois, USA. The model is a there are many reasons for the dilution, in fact, the
function of the thickness of the extraction layer, reasons for the dilution are divided into three
the thickness of the ceiling and the floor, the effect categories: 1- drilling and exploding; 2- errors in
on the dilution, the specific gravity of the coal and planning; 3- geotechnical problems (Tommila
the specific weight of the ceiling and the floor. 2014). In the last quarter of 2015, an explosives
This model assumes continuous production in such classification system was dilution to control the
a way that the installation cycle and other cases do degree to which classification of explosives can be
not cause any interruptions in the speed of the classified and dilution based on this classification
advance. It is also assumed that the average speed system (Little 2015). In 2019, Masoumi presented
of cutting the coal and the lower and lower legs is a self-correlated minimally-invariant invocative
the same (Chugh et al. 2004). method for simulating continuous viable variables
In 2005, Mubita conducted a research effort to and the calculated dilution using this method based
reduce the rate of recovery and increase income in on lithology logs of the drill holes was a good
the Zambia Concola mine (Mubita 2005). In 2006, result (Masoumi et al. 2019). Distribution of
Soyer presented a method for estimating mineral crushing and dilution size is effective on Asia's
depletion. The combination of mine design efficiency (Xingwana 2016). In the case of wet
software and mine control systems has been used mining conditions due to incomplete and
and aims to compare new methods of calculating re-explosion, the mineralisation and tailings must
dilution with old methods. The results of new be increased (Nikkhah & Taji 2018). For the
methods used to estimate the amount of control of the dilution of the Ajupa mine, an
extractable minerals, 1.6% for tonnage, and 1.6% explosive displacement monitoring system was
for grade, indicate the proximity to their actual used which cost $116,031 during the study. The
values compared with the old methods (Soyer mine’s revenue from the site was $866,869,
2006). In 2007, Canada’s Open Workshop mines indicating that the mining savings were due to an
provided numerical methods using Map3D increase in control of $753,835 (Xingwana 2016).
software. In this method, the effects of parameters The direction and displacement of the explosion
such as mining depth, workshop geometry, slope affects the cost of dilution and the process of
angle, tensions and their direction on the dilution mining to the plant (Rogers & Kanchibotla 2012;
have been investigated (Henning & Mitri 2007). Gaunt et al. 2015). Dilution and loss of ore are
Also, in 2009 Saeedi et al. Using data collected important factors that can affect the cost and
from the Tabas coal mine and using the FLAC 2D profitability of a mine. Considering the dilution is
software, provided a model for out-of-layer dependent on a number of factors that vary from
dilutions in longitudinal fuselage mines. In this period to period as a general rule, for larger
modelling, the parameters affecting the dilution of periods, it might become easier. for estimate
the layer, including the depth of the mineral, the dilution, but the operation needs to deal with
stress, the geometry of the ceiling immediately monthly, weekly or even daily production (Camara
above the shale, and the quality of the roof and et al. 2018).

- 52 -
 Marinin developed guidelines for assessing affect the economic profit of the mines, therefore,
how losses and dilution indicators affect the cost the calculation and management of the extent of
chain from ore mining to processing. Economic the dilution as mine strategies are considered in
evaluation method as a combination of factor the long run and are continuously studied and
analysis and engineering value, which saves solved (Little 2015, Gaunt et al. 2015, Smith et al.
operating costs. The basis of continuous planning 2015, Masoumi et al. 2019).
is the analysis and control of the production
process from the ore extraction stage to the 4 EFFECT OF DILUTION ON THE
refining of the mineral product (mine to mill). PROFITABILITY OF OPEN PIT MINES
Based on the structure of the total reserves as well
as the comparison between grade and metal Improvement affects profitability in two ways
performance due to changes in loss and dilution (Wang et al. 2011):
indices, the main results of the experimental study
are that there is a reduction in mineral losses. Ore − Reduce direct costs by controlling more
inventory changes in ore loss and dilution values materials;
(3.1% losses and 17% dilution in the baseline case − Reduction of production (or product grade)
scenario. 2.1% losses and 13.0% dilution under the with fixed factory capacity.
project scenario) result in a 30 tonne reduction in
processed ore values. The analysis of project With higher dilution, the mineral is less
sensitivity indicators shows that provided that the processed, resulting in less crop production. The
amount of ore mined and shipped fluctuates relationship between dilution and profitability can
between 5%, the project efficiency of reducing the be illustrated in Figure 2.
annual workload of mixing and loading ore varies
in the range of $191.25-1147.5 (Marinin et al.
2021).
In open-cast extraction methods in porphyry
mines, because of the high volume of extraction
and the extent of the extractive operations front,
dilution is due to several factors such as geological
conditions and the presence of various dikes in
minerals, design factors and engineering factors of
operating factors, and the volume of buckets of
loading machines and mine supervision. Given
that the Sarcheshmeh copper mine is no exception
to these conditions. Due to the fact that dilution is
directly related to lowering the cost of producing
and producing high-grade products, for this Figure 2. Impact of dilution on mining profit
purpose, in this research, dilution has been studied (Wang et al. 2011)
in Sarcheshmeh copper mine.
One of the consequences of a dilution is the
3 THE EFFECT OF DILUTION IN MINING reduction in the feed grade of the processing plant.
PROCESSES A lower-grade feed means less income. On the
borderline levels, the dilution may reduce the
The amount of dilution must be carefully grade to the degree that it is for non-economic
measured and determined. Dilution control in processing. In other words, the dilution may cause
active mines, which are in operation, are often the mineral to be deposited (Saeedi et al. 2008).
controlled by a grade, which is the most
controllable part of the mining profit. Mines are 5 STUDY OF DILUTION IN
seeking to mechanise mine operations and increase SARCHESHMEH COPPER MINE
production capacity to reduce costs and increase
profits. Achieving these goals is only possible if Sarcheshmeh Copper Mine is one of the most
you can maintain the degree of dilution. Operating important copper marble porphyry Mills in the
mechanisation is always better than unmanaged world. The mine is located 160 km southwest of
and manual operations. But care should be taken Kerman and 50 km southwest of Rafsanjan and at
that the cost savings in work will not be offset by a height of 2600 meters above sea level. The mine
increasing costs. The size and amount directly is stocked with a margin of 0.25% and an average

- 53 -
 Table 1. Dilution resources type in the Sarcheshmeh copper mine.
Dilution type Dilution resources Sarcheshmeh Copper mine
1. Dilution due to the interaction of ore deposit with the central dike
Type 1. Dilution due to
2. Dilution due to the loading machine bucket
geometry
3. Dilution due to waste rock in berm
Type 2. Dilution due to 4. Dilution due to mining development projects in ramps
uncertainty in the 5. Dilution due to geological conditions of the mine
knowledge of the in situ 6. Dilution due to uncertainty of block model
ore boundaries
Type 3. Dilution due to 7 Dilution due to incomplete blast
blast movement 8. Dilution due to back-break and side-break blasting block
9. Dilution due to the design of ore and waste blasting blocks in one
blast
10. Spreading waste as a cover in the benchs and ramps
11. Colour similarity of ore with waste rock
Type 4. Dilution due to 12. Dilution due to cutting the floor of working benchs
mining errors 13. Dilution due to reblast
14. Dilution due to instability of bench
15. Dilution due to mining operations at upper bench
16. Dilution due to waste stemming in production hole stemming
17. Dilution due to inadequate visibility of working night operator
Type 5. Dilution due to
18. Undesirable weather conditions
weather conditions

grade of 0.66% and is 1763 million tons. In the loading due to mineral and tailings interlocking. If
development plan of the mine, the annual the detonator blast is used, along with the use of
extraction of 25 million tons of minerals and 35 electronic detergents to control the development, it
million tons of waste is set. Sarcheshme copper will be very effective. The sources of dilution in
mine steps are 12.5 m and the width of the stairs is the Sarcheshmeh copper mine were classified into
8.75 m. Periodically and periodically, for each 4 five categories and control and modification
steps, a large staircase with a width of 75.23 m is strategies (Table 1).
placed. The final slope of the mine wall is
currently 34 to 36 degrees. Also, the width of the 5.2 Database dilution control
mine roads is 30 metres and the slope is 8%
(Sarcheshmeh Copper Complex 2015). In order to effectively investigate the dilution of
Sarcheshmeh copper mine, the databases of
5.1 Type and dilution control drilling, blasting and loading sections were created
as follows. The description of the database
In open pit mines dilution due geometry, dilution parameters is given in Table 2.
due uncertainty in knowledge of the in situ ore
boundaries, dilution due blast movement, dilution − Drilling Database: Including bench height,
due mining errors (Little 2015) and dilution due length, angle, and diameter of each hole,
weather conditions. drilling pattern and total length drilling blast
Most of the production explosions in the block.
Sarcheshmeh copper mine are of a bulk type for − Blasting Database: Including burden, spacing,
controlling the amount of explosive blocks that are specific drilling, special charge, calculation of
multi-tile or mined or waste, the recovery rate is blast block dilution.
greater than the predicted value in the − Loading Database: Including performance,
development plans, so that in explosions 7 and 10, preparedness and movement of the loading
the amount of dilution 13.17% and 10.45%, machine (Nikkhah & Taji 2017), type and
respectively. The main reasons for the fuel of loading device and specific device.
development of this type of explosion are the non-
explosion of mineral and tailings in the form of Specific loading with the symbol (Slj), specific
explosive separation and regeneration during loading of the jth blasting block (hours per cubic

- 54 -
metre), which is calculated by dividing the total
Table 3. Database unit drilling and blasting 7
time available for loading machineries of jth blast blocks.
blasting block (per hour) to the total volume of the
j-th blasting block (cubic metre) (Taji et al. 2012). Parameters symbol Unit Range
In Table 3 and Table 4, the data related to the Burden B m 5-7
drilling and exploding section and the loading Spacing S m 6-9.5
section are presented. Diameter D inch 6.5-10
Hole length H m 14.5-15
Table 2. Description of collected parameters to
create database. Bench height m 12.5
Sub-drilling J m 2.5
Parameters Symbol Unit Stemming
Burden B m m 6.5-8.5
length
Spacing S m Ratio of length
Diameter D inch to width of - 1.36-4.95
Hole length H m blast block
Bench height m Volume of blast 67718.04-
Sub-drilling J m block 412139.77
Blast block
Stemming length m ton 48000-108438
tonnage
Ratio of length to width of
-
blast block
Volume of blast block theories proposed in the dilution prediction. Also,
different linear and nonlinear models have been
Blast block tonnage ton tested and selected by using two Watson Camera
Specific drilling and F-model tests. To compare the accuracy of
nonlinear regression prediction, F test was used. In
Specific charge
this test, if the Sig value obtained at 95%
Loading specific confidence level is less than 5%, it is indicative
Dilution % that the accuracy of the estimate for the obtained
relationship is acceptable. Also, in the Watson
Blast block tonnage ore ton camera test, the value obtained is between 0-4.
Blast block tonnage waste ton The predicted model is acceptable (Mirzadeh
2010). In Table 5, the matrix of the correlation
coefficients of the studied parameters are show.
6 DILUTION MODELLING The values of the correlation coefficients
shown in Table 6 are variables in the range 1 to 1.
Studies in previous sections show that many The positive sign of these coefficients expressing
parameters can be effective in dilution. But these the direct relationship with the negative sign
parameters were ranked so that the control could expressing their inverse relationship with each
be controlled and reduced. For this purpose, the other and their magnitude indicates their effect.
data was studied using Spss v.21 software. By The coefficients of greater absolute magnitude
studying the dependent and independent have a higher impact and therefore have a stronger
parameters, one can get the relationship between relationship. The coefficients in the matrix
each one together and get informed about the diameter are usually 1. Since the relationship of
process of their changes. This helps to compare the each variable with itself is a completely linear
accuracy and overall trend of the changes with the relationship. The same is true of the matrix.

Table 4. Key database 7 blast blocks.

Parameters Symbol Unit Range Average
Specific drilling 0.013-0.04 0.021
Specific charge 0.256-0.612 0.446
Specific loading 0.004-0.009 0.007
Dilution % 0.000-13.7 6.73

- 55 -
Table 5. Matrix of correlation coefficients.

- 56 -
Table 6. Proposed models and analysis of variance of projected desalination models with independent parameters.
Predict model
1
2
3
4

Analysis of variance (ANOVA)

Model R Adj. Std. error F change Sig F change Durbin-Watson

No.
1 0.988 0.977 0.953 0.949 41.928 0.006 2.159
2 0.937 0.877 0.877 1.878 14.281 0.02 1.654
3 0.892 0.796 0.796 2.431 7.816 0.04 2.335
4 0.915 0838 0.757 2.168 10.333 0.03 2.258

Table 7. Multivariate nonlinear regression coefficients and statistical parameters of coefficients for
models 1, 2, 3, and 4.

Unstandardised Standardised 95% Confidence Collinearity

Model Independent coefficients coefficients interval for B statistics
No. variables Lower Upper t val-
B Std. Error Beta Tolerance VIF
Bound Bound ues
Constant -81.163 9.460 -111.270 -51.056 -8.579
30.332 4.420 0.621 16.267 44.397 6.863 0.948 1.055
1
-3.418 0.955 -0.325 -6.456 -0.380 -3.581 0.944 1.059
-7.962 1.173 -0.613 -11.694 -4.230 -6.789 0.953 1.05
Constant -88.247 18.398 -139.327 -37.166 -4.797
2 33.282 8.635 0.682 9.307 57.258 3.854 0.982 1.018
-7.236 2.297 -0.577 -13.613 -0.858 -3.150 0.982 1.018
Constant 284.534 116.443 -38.944 607.652 2.442
3 -275.412 113.729 -0.558 -59.173 40.349 -2.422 0.959 1.042
0.002 0.001 0.593 0.000 0.04 2.574 0.959 1.042
Constant -5.989 5.385 -20.940 8.961 -1.112
0.273 0.090 0.627 0.022 0.523 3.021 0.941 1.063
4
- 1870.645 -0.532 - 402.616 -2.561 0.941 1.063
4791.127 9984.869

Table 8. Proposed models and analysis of variance of suggested dilution models with independent parameters.
Predict model
1
2

Analysis of variance (ANOVA)

Model No. Adj. Std. error F change Sig F change Durbin-Watson
R
1 0.985 0.970 0.939 1.082 31.815 0.009 2.228
2 0.999 0.999 0.997 0.230 728.994 0.000 1.765

- 57 -
 Table 9. Multivariate nonlinear regression coefficients and statistical parameters of coefficients for models 1 and 2.
Unstandardised Standardised 95% Confidence Collinearity
Model Independent coefficients coefficients interval for B statistics
No. variables Lower Upper t values
B Std. Error Beta Tolerance VIF
Bound Bound
Constant -91.238 10.599 -124.968 -57.508 -8.608
S 35.679 5.875 1.282 16.982 54.376 6.073 0.228 4.388
1
D -18.144 4.05 -0.936 -31.208 -5.08 -4.420 0.227 4.411
30.559 5.389 0.626 13.407 47.711 5.670 0.834 1.199
Constant -18.627 0.573 -20.450 -16.803 -32.506
D -0.155 0.007 -1.024 -0.176 -0.134 -23.249 0.235 4.249
2
0.253 0.01 0.584 0.220 0.285 24.411 0.806 1.241
S 0.326 0.011 1.353 0.291 0.360 30.242 0.228 4.383

Among the parameters mentioned, the length of variable. Two models 1 and 2 were proposed, as
the choral flowering fabric with a correlation indicated in Table 8.
coefficient of 0.957 had the highest effect and the Accordingly, the correlation coefficient of
weight ratio with the coefficient of correlation of 0.970 and 0.999, respectively, for the dilution
0.204 had the least effect on the dilution. models 1 and 2, show a significant correlation
between the dilution parameter and its independent
6.1 Relationship between , with variables, which is also the basis for determining
dilution the validity of the model. According to the
variance inflation factor test, the independent
In order to investigate the effect of effective variables of model 1 and 2 are presented in Table
parameters on dilution, using multi-parameter 9. There is no solid correlation between the
linear and non-linear regression, four models were independent input variables of models 1 and 2.
obtained in Table 7 of the models. In the proposed models, according to the
In order to control the validity of the models, or standardised coefficient column (Borcard 2002), it
in other words, to control its significance, the error can be seen that the standardised coefficients of
independence test was used from the Watson applying regression by evaluating this column can
camera and the F test. The results of both tests are be concluded which factor has a positive effect
shown in Table 6. Since the value of both tests is and which negative effect is the diameter of the
acceptable at the 95% confidence level, the chamber, the special drilling and the length to
correlation coefficient of the obtained models for width ratio The explosive block reduces depletion,
dilution indicates that the dilution parameter but the length of the flowering and the spacing will
has a significant relation with its independent increase it. Since the values of their standard
variables at a suitable level. This test is coefficients were the highest values.
another basis for determining the validity of the
model. 7 CONCLUSIONS
Variable inflation factor has been used to
confirm the lack of relation between independent In this research, the study of dilution was
input variables. So that the variance inflation investigated in Sarcheshmeh copper mine. Most of
factor for each independent variable is less than the explosions are of precise type flare explosions.
10. This means that there is no significant According to the calculations, the highest dilution
correlation between the independent input rate in the Sarcheshmeh copper mine is 13.17, and
variables (Faramarzi et al. 2013). The variance the average dilution is 6.73. In order to study the
inflation factor results for the models 1, 2, 3, and 4 effective parameters on dilution, we used
are shown (Table 7). statistical analysis. The very high correlation
coefficient of the obtained models indicates the
6.2 Relation between spacing, diameter and effect of independent parameters on dilution. By
stemming with dilution comparing the results, we find that stemming
length, spacing, hole diameter and specific drilling
In the analysis of nonlinear regression, logarithmic are the most important parameters affecting
and invariant variables of hole spacing, diameter dilution. By controlling some of the parameters
of chalk and clipping length were considered as that affect the dilution, the changing of blasting
independent variable and dilution of dependent pattern design can minimise the amount of

- 58 -
dilution, which will result in profitability and Henning, J. G. & Mitri, H. S. 2007. Numerical
mining costs reduction in Sarcheshmeh Copper modelling of ore dilution in blasthole stoping,
Complex. International Journal of Rock Mechanics &
Mining Sciences 44: 692-703.
8 ACKNOWLEDGMENTS
Little, T.N. 2015. Classification and development in
The authors appreciate Iranian copper industry grade control blasting for surface mines, Proc. Of
national company. FRAGBLAST 11, Fragmentation by blasting.

REFERENCES Masoumi, I., Kamali, G.R. & Asghari, O. 2019.

Assessment of an ore body internal dilution based
Agoshkov, M., Borisov, S. & Boyarski, V. 1998. on multivariate geostatistical simulation using
Mining of ores and non-metalic minerals (Trans. exploratory drill hole data, Journal of Mining and
V. Kochin), Mir Publishers Moscow, Russia. Environment (JME) 10(1): 271-286.

Borcard, D. 2002. Multiple and partial regression and Mirzadeh, M. R. 2010. Statistical analysis with SPSS,
correlation Partial , contribution and fraction, Teheran, Taymaz Pub.
Université de Montréal Département de sciences
Mubita, D.M. 2005. Recent initiatives in reducing
biologiques, Canada.
dilution at Konkola mine, Zambia, Journal-
South African Institute of Mining and Metallurgy
Chugh, Y. P., Moharana, A. & Patwardhan, A. 2004.
105(2):107-112.
An analysis of the effect of out-of-seam dilution
on coal utilisation, International Conference on
Marinin, M., Marinina, O. & Wolniak, R. 2021.
Clean Technologies for the Mining Industry,
Assessing of losses and dilution impact on the
University of Concepción, Chile.
cost chain: case study of gold ore deposits,
Journal of Sustainability, DOI: doi.org/10.3390
Camara, T.R., Leal, R.S., Peroni., R.L. & Capponi,
L.N. 2018. Controlling operational dilution in /su13073830.
open-pit mining, Mining Technology, DOI:
10.1080/25726668.2018.1470275. Najafi, B., Saeedi, G.S. & Ebrahimi Farsangi, M.A.
2014. Risk analysis and prediction of out-of-seam
Domingo, J.F.D., Leite, F.S., Mirada, V.G. & dilution in longwall mining, International
Carrasco, I.C. 2015. Dilution, ore grade and blast Journal of Rock Mechanics & Mining Sciences
movement calculation model, Proc. Of 70: 115–122.
FRAGBLAST 11, Fragmentation by blasting.
Nikkhah, A. & Taji, M. 2017. Investigating the
Ebrahimi, A. 2013. An attempt to standardise the impacts of fragmentation caused by blasting
estimation of dilution factor for open pit mining operation on machinery efficiency of
project, World Mining Congress, Montreal, Sarcheshmeh Copper Mine, Journal of
Canada. Research-Papers Mineral Resources Engineering
28: 267-272.
Engmann, E., Ako, S., Bisiaux, B., Rogers, W. &
Kanchibotla, S. 2013. Measurement and Nikkhah, A. & Taji, M. 2018. The effect of the
modelling of blast movement to reduce ore losses underground water on production costs and
and dilution at Ahafo gold mine in Ghan, Ghana mining operation in open-pit mines, Case Study,
Mining Journal 27-36. Sarcheshmeh Copper Mine 28: 267-272.

Faramarzi, F., Mansouri, H. & Ebrahimi Farsangi, Noppe, M. 2003. The measurement and control of
M.A. 2013. A rock engineering systems-based dilution in an underground coal operation, 5th
model to predict rock fragmentation by blasting, International Mining Geology Conference,
International Journal of Rock Mechanics and Bendigo, Australia.
Mining Sciences 60: 82-94.
Ran, J. 2002. Hangingwall sloughing mechanism in
Gaunt, J., Symonds, D., McNamara, G., Adiyansyah, open stope mining, CIM Bulletin 95: 74-77.
B., Kennelly, L., Sellers, E.J. & Kanchibotla, S.S.
2015. Optimisation of drill and blast for mill Rogers, W.D. & Kanchibotla, S.S. 2012. Application
throughput improvement at Ban Houayxai mine, of stochastic approach to predict blast movement,
Proc. Of FRAGBLAST 11, Fragmentation by Proceedings of the 10th international symposium
blasting. rock fragmentation by blasting, New Delhi, India.

- 59 -
Saeedi, G., Rezai, B., Shareiar, K., Oraee, K. 2008.
Quantifying level of out-of-seam dilution for
longwall mining method and its impact on yield
of coal washing plant in Tabas coal mine,
Proceedings of the International Seminar on
Mineral Processing Technology, Trivandrum,
India.

Saeedi, Gh. 2012. Dilution control in mines. Ministry

of Industry, Mine and Trade, Office of Oversight
and Utilisation of Mining Rules and Standards.

Sarcheshmeh Copper Complex 2015. Report of

Sarcheshmeh copper mine development plan,
Sarcheshmeh Copper Mine Engineering, Design
Department.

Smith, C.B., Nehring, M. & Bosompem, M.K. 2015.

A critique of selective mining unit sizing at
Century Mine to optimise productivity with
dilution, Proc. Of FRAGBLAST 11,
Fragmentation by blasting.

Soyer, N. 2006. An approach on dilution and ore

recovery/ loss calculations in mineral reserve
estimations at the Cayeli mine, Turkey, A Thesis
submitted to the Graduate School of Natural and
Applied Sciences of Middle East Technical
University.

Taji, M., Ataei, M., Goshtasbi, K. & Osanloo M.

2012. A new approach for open pit mine blasting
evaluation, Journal of Vibration and Control
2(3): 25-36.

Tommila, E. 2014. Mining method evaluation and

dilution control in Kittilä mine, Master’s Thesis.

Wang, W., Huang, S., Wu, X. & Ma, Q. 2011.

Calculation and management for mining loss and
dilution under 3D visualisation technical
condition, Journal of Software Engineering and
Applications 4: 329-334.

Xingwana, L. 2016. Control of ore loss and dilution

at AngloGold Ashanti, Iduapriem Mine using
blast movement monitoring system, Ghana
Mining Journal 16 (1): 49-59.

Xingwana, L. 2016. Monitoring ore loss and dilution

for mine-to-mill integration in deep gold mines: a
survey-based investigation, Southern African
Institute of Mining and Metallurgy 116(2): 149-
160.

Yihong, L. & Zhang, W. 1986. Reducing waste-rock

dilution in narrow-vain conditions at tungsten
mines in China, Mining Science and Technology
4: 1-7.

- 60 -
- 61 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Understanding single salt, dual salt, and multiple salt bulk

emulsions and emulsion blends

D.S. Scovira, B. Victor & Q. Steyl

BME, a division of Omnia Group (Pty) Ltd, Gauteng, South Africa

ABSTRACT: Bulk emulsions and emulsion blends may be formulated with a single salt, dual salts, or
multiple salts. Single salt bulk emulsions and emulsion blends, based on ammonium nitrate, are familiar
to many blasting professionals as they are manufactured and employed on a global basis. Blasting
professionals may be less acquainted with dual and multiple salt emulsion and emulsion blends that may
include calcium nitrate, sodium nitrate, and urea in their formulations.

1 INTRODUCTION nitrate, sodium nitrate, and urea in their

formulations. These additional salts can impart
Bulk emulsions and emulsion blends may be various beneficial qualities to address end user
formulated with a single salt, dual salts, or needs and blasting results.
multiple salts. Single salt bulk emulsions and
emulsion blends, most based on ammonium 2 EMULSION STABILITY AND
nitrate, are familiar to many blasting professionals CRYSTALLISATION
as they are manufactured and employed on a
global basis. Blasting professionals may be less The process of heterogenous seed driven
acquainted with dual and multiple salt emulsion crystallisation primarily effects single salt bulk
and emulsion blends that may include calcium emulsions. Seed needle crystal grows and

Figure 1. Heterogenous (seed) driven crystallisation in a AN single salt emulsion.

- 62 -
penetrate adjacent oxidiser phase cells. The emulsion composition and contributes to keeping
compromised oxidiser cell crystallises. The cycle the total emulsion water content low enough to
repeats and eventually compromises the emulsion. ensure detonator sensitivity. Specific to small
The result is commonly called ‘salting out’ or diameter bulk emulsion products, lesser amounts
‘phase separation’. See Figure 1 - Heterogenous of sodium nitrate addition be used to enhance
(seed) driven crystallisation. sensitivity without crossing the threshold of
Dual and multiple salt emulsions are inherently detonator sensitivity. During manufacture, sodium
chemically resistant to heterogenous seed driven nitrate acts as a eutectic that depresses the
crystallisation. Dual ANCN salt emulsions have a crystallisation temperature (fudge point) and
chemistry advantage over single salt AN results in stable finished emulsion high in
emulsions in keeping the AN in solution. The energetic salt content and low water content.
crystallisation point is depressed due to increased Dual salt ANCN emulsion technology is also
solubility, and this stabilises the oxidiser phase. commonly called ‘cold emulsion technology’. The
Crystals in the oxidiser phase of dual salt ANCN meaning behind cold emulsion technology is that
emulsion are cubic and less likely to form needles. the emulsion matrix is manufactured at a lower
Accordingly, they are less likely to drive the temperature of 65 versus 85 [149 versus
homogenous seed crystallisation process
185 ] for a single salt AN emulsion. CN
previously described. See Figure 2 – Dual salt
promotes emulsification and therefore less
emulsion cubic crystals.
emulsifier is required for stable emulsion
Sodium nitrate (NaNO3, SN) is also a salt that
formation.
is also used in multiple salt emulsion formulations.
Sodium nitrate is commonly found in detonator Dual salt ANCN emulsions have a finished
sensitive packaged emulsion formulations and temperature of 50 [122 and no cooling time
some bulk emulsions aimed at small diameter is required before field use. The lower
applications. Specific to packaged products, manufacturing temperature eliminates the need for
sodium nitrate does not introduce water to the additional storage capacity at the plant to allow

Figure 2. Electron microscope image of cubic crystals of a dual salt and water solution.

- 63 -
enough product cooling time before use. The However, when looking at the oxygen balance
lower manufacturing temperature also requires less of ammonium nitrate versus calcium nitrate, it is
energy intensity and cost at the plant. shown that calcium nitrate contributes more than
In remote mining locations, where installation twice as much to the oxygen balance than
of a bulk emulsion manufacturing plant is not ammonium nitrate does:
economically justified or desired, bulk emulsion
products may need to be delivered over OB% = -1600/molecular weight x (2x +(y/2) + m-
considerable distances and stored and consumed z). Where: x = number of atoms of carbon(C); y =
over extended periods of time. Dual and multiple number of atoms of hydrogen (H); m = number of
salt emulsions are highly stable and have been atoms of metal (metallic oxide produced); and z =
demonstrated to resist oxidiser and fuel phase number of atoms of oxygen (O).
separation during the rigor of transport, resistance
to breakdown by repeated re-pumping, and salting Oxygen balance for NH4NO3 = +0.20 (+20%)
out during extended storage times. Oxygen balance for Ca(NO3)2 = + 0.49 (+49%)

3 OXYGEN BALANCE AND POST BLAST

FUME

Oxygen balance (OB, or OB%) is the term used to

indicate the degree that an explosive can be
oxidised. If an explosive contains sufficient
oxygen to form only CO2, H2O and N2, then the
explosive is at zero oxygen balance. Due to
chemical reaction kinetics, the detonation of
explosives based on nitrogen molecules may
generate toxic fumes of nitrogen dioxide (NO2),
nitric oxide (NO), and carbon monoxide (CO).
Oxygen balance has an influence on the
detonation products and is illustrated by the
following reactions: See Equation 1 and Equation Figure 3. Relationship of Oxygen balance to
NOx generation.
2: Negative and positive oxygen balances.

Negative oxygen balance (Excess fuel): This results in ANCN dual salt emulsions
2NH4NO3 + CH2 → 2N2 + CO + 5H2O formulations being more closely oxygen balanced
Equation 1. as compared to single salt AN emulsions. In field
practice, single salt AN bulk emulsions and bulk
Positive oxygen balance (Excess oxidiser): emulsion blends have been observed to generate
5NH4NO3 + CH2 → 4N2 + CO2 + 11H2O + 2NO considerable post blast NOx fume while dual salt
Equation 2. ANCN bulk emulsions and emulsion blends have
been noted to produce lesser degrees of NOx.
Figure 3 is an illustration of NOx production The low finished product temperature of dual
with AN/fuel blends (ANFO) with increasing salt ANCN emulsions may also reduce the heat
oxygen balance. It can be seen from the figure that erosion of AN prill in emulsions blends and this
NOx production is rapidly increased and most can also aid in reducing post blast NOx fume.
affected by excess oxidiser (oxygen positive). It is
therefore vital when formulating an emulsion 4 AMMONIA AND NITRATE IN MINE
explosive that there is enough oxygen for the WATER
available fuel. See Figure 3, the relationship of
oxygen balance to NOx. Chemical contents of mine water discharged to the
When looking at the actual percentage of environment continues to draw attention and
oxygen in ammonium nitrate versus calcium additional regulation. NH4+ (ammonium or ionised
nitrate there is not much of a difference: ammonia) is essentially harmless to marine life;
NH30 (un-ionised or free ammonia) is known to be
% Oxygen in NH4NO3 = 59.97% toxic to marine life. Ammonia and nitrate ions
% Oxygen in Ca(NO3)2 = 58.51% have been shown to be especially harmful to fish

- 64 -
and fish eggs. Very low levels have been shown to mg [0.23 lb] of NH4+ to the environment. =
consequential and hence the mine water discharge 102 300 ppm. This is a +67.84% difference in
can be highly monitored by regulators. Many NH4+ release.
mines are required to capture and/or treat their If the above ammonium quantities are dissolved
mine wastewater before discharge to the outside in 1 litre of water, then:
environment. In Canada, the Metal and Diamond
Mining Effluent Regulations (MDMER) effective 1 kg [2.204 lb] AN emulsion can result in 679 mg
01 June 2021 established a federal effluent limit [0.0015 lb] of NH30 (679 ppm, 679 mg/l [0.0057
of unionised ammonia at a monthly mean lb/gal]) released in water at pH 7 at temperature of
concentration limit of 0.5 mg-N/L [0.5 ppm]. 20 .
Residual bulk emulsions or emulsion blends
originating from product that migrates into
cracks/voids in the blast hole, occurs as spillage at
the collar of the hole, and other causes may end up
in blasted rock. In Canada, the US Pacific
Northwest, and other wet-dry-wet cyclical
environments, muck piles and waste rock may be
repeatedly washed by rain and melting snow (679 mg/l) / 0.5 ppm = 1358 litres [358.75 gal] of
between dry periods. This wet-dry-wet cycling water required to dilute to 5 ppm concentration.
environment promotes decomposition of the [0.0057 lb/gal]
emulsion with subsequent leaching out of the
component salts into the mine water. 1 kg [2.204 lb] of ANCN emulsion can result in
Regarding bulk explosive applications in 405 mg [0.001 lb] of NH30 (405 ppm, 405 mg/l
sensitive/regulated areas, consider the differences [0.0034 lb/gal]) released in water at pH 7 at
for the two compositions of emulsions if broken temperature of 20 .
down and then dissolved in water. Broken down
emulsion results in the liberation of NH4+.
Liberated NH4+ from emulsion when introduced to
environmental water, and dependent upon
environmental water pH and temperature, can
be converted to NH30. Equation 3 shows the
ammonium-ammonia equilibrium.
(405 mg/l) / 0.5 ppm = 810 litres [214.98 gal]
water required to dilute to 0.5 ppm concentration.
NH3 + H2O NH4+ + OH-
[0.0034 lb/gal]
Equation 3.
Therefore, it takes + 71.5% more water to
The Iowa Calculator can be used to determine dilute 1 kilogram [2.204 lb] of disassociated single
the NH30 amount created/present based on water salt AN emulsion down to 0.5 ppm concentration
pH and temperature. as compared to 1 kilogram of disassociated dual
The following example is presented for salt ANCN emulsion.
unionised ammonia at a limit of 0.5 mg/L or 0.5 By design, some mining and dumping
ppm and compares industry standard AN single operations are constructed so that the mine water
salt emulsion with an ANCN dual salt emulsion drains into a central pit lake. This pit lake water
formulations. Both emulsions contain a total of gets treated and helps to avoid the issue of
76% oxidiser salts. Assume that the salts from the discharge release ‘pulses’ that can be caused by
emulsion oxidiser phase are fully broken periods of dry (where emulsion breaks down)
down/fully separated from the fuel phase and other followed by periods of wet (broken down
components of the emulsion. The single salt AN emulsion washes out). Scrubbing out ammonia and
emulsion contains 17.17% NH4+ and the dual salt nitrate from mine discharge water may be
ANCN emulsion contains 10.23% NH4+. This done by chemical treatment, filtering, and/or
equates to a 6.94% difference in the NH4+ quantity. bioremediation, and these treatments of large
One kilogram of broken down AN emulsion quantities of mine discharge water can be costly.
can release 171 700 mg [0.38 lb] of NH4+ to the There is no significant difference to the nitrate
environment = 171 700 ppm. One kilogram of (NO3-) quantity liberated when comparing single
broken down ANCN emulsion can release 102 300 salt AN emulsions to dual salt ANCN emulsions.

- 65 -
 TM
Figure 4. Example of critical density and void space calculation from BME’s DensDepth program.

The single salt AN emulsion contains 59.03% have computer programs to aid in the calculation
NO3- and the dual salt ANCN emulsion contains of the relationship using site specific conditions.
59.44% NO3- By manipulating the cup density, the end user can
be ensured that there is sufficient critical void
5 CRITICAL DENSITY AND VOID SPACE space and that the critical density is not exceeded
FOR GASSED EMULSIONS for the drilled hole diameter and depth. See Figure
4, an example of critical density and void space
To realise reliable performance and optimal results calculation.
with gas sensitised single salt, dual salt, and
multiple salt emulsions and emulsion blends, it is 6 ENERGY VALUES AND IDEAL
important to understand the relationship between DETONATION CODE
hole diameter, product densities, and maximum
hole depth. This relationship must be understood Manufacturers set standard ANFO to a Relative
by the user to ensure there is enough critical void Weight Strength (RWS) = 100 and Relative Bulk
space in the bulk product at the bottom of the hole Strength (RBS) = 100. The ANFO RWS and RBS
to ensure detonation sensitivity. To operate in the are typically published to establish baseline values
clear of exceeding critical densities, many on each manufacturer’s Technical Data Sheet
manufacturers/suppliers of gassed bulk emulsions (TDS). A literature search shows that the Absolute

Table 1. Survey of standard ANFO and Ideal Detonation Codes – 2019.

- 66 -
Weight Strength (AWS), Absolute Bulk Strength again generate values that are comparable only
(ABS), and density values used to define standard within the house running the code.
ANFO can vary among manufacturers. Handbooks For the end user, it does raise the question of
published by manufacturers and professional how does one go about understanding and
societies also show variability. See Table 1 – comparing explosive energies between the
Survey of standard ANFO values and ideal different manufacturers?
detonation codes. Underwater bubble energy testing can be useful
Company propriety computer ideal detonation for making relative and comparable physical
codes are often used to calculate and publish measurements of total, shock and gas energies
Relative Weight Strength (RWS) and Relative realised by explosive products. UTEC Lab
Bulk Strength (RBS) product energy values (Kansas, USA) is a reputable 3rd party explosives
relative to their standard ANFO. The ideal energy testing facility that may be engaged by
values generated by these various computer codes manufacturers and end users to conduct
while useful in comparing products within a given underwater bubble energy, confined VOD testing,
manufacturer’s product line, often creates and other tests on commercial explosives. As an
confusion for the end user when trying to cross effort to understand and differentiate between
compare published energy values between various manufacturer explosive offerings,
manufacturers. some major mining houses are now having
Ideal detonation codes calculate explosive manufacturers conduct and report physical
energies based on the inputs of the formulation measurements from UTEC as part of their tender
composition and a database of theoretical process.
thermodynamic properties for the various
composites. Ideal detonation codes do not consider 7 PHYSICAL TESTING AND EXPLOSIVE
detonation efficiency realised through explosive PERFORMANCE
material particle size and manufacturing
refinement. As a simplified example, a given Recent underwater bubble energy tests conducted
composition of salts, fuel, and water input into an at UTEC with single salt, dual salt and urea
ideal detonation code will produce the same inhibited emulsions and emulsion blends record
energy values whether the raw materials are in interesting differences in total energy, and shock
solid blocks or highly processed. Likewise, ideal and gas energy partitions between formulations.
detonation codes do not calculate comparable Measurements from underwater bubble energy
energies realised by say a straight emulsion with tests show the total energy difference on a volume
highly refined particle versus a blend of 70% basis (cal/cc) comparing a straight single salt AN
straight emulsion with highly refined particles emulsion to a straight dual salt ANCN emulsion is
blended with 30% AN/ANFO coarse grained 1%, and the total energy difference on a weight
particles. basis (cal/g) is 2%. These measurements confirm
Military, commercial, and academic institutions the total energies realised by a straight dual salt
have continued to develop non-ideal detonation ANCN emulsion is essentially equivalent to a
codes that strive to correct the shortcomings of straight single salt AN emulsion. The dual salt
calculations by ideal detonation codes. Non-ideal ANCN emulsion tested generated nearly equal
detonation codes are usually proprietary and once amounts of shock and gas energies, while the

Figure 5. Underwater energy test measurements for straight single salt and dual salt emulsion.

- 67 -
 Figure 6. Underwater energy test measurements for single salt and dual salt emulsion blend.

single salt AN emulsion developed slightly greater against reactive ground. Urea inhibits the
shock energy and slightly less gas energy. See exothermic reactive ground chain of reaction by
Figures 5 – Energy of straight single salt and trapping the vital catalyst nitrous acid. See
straight dual salt emulsion. Equation 4, Urea-nitrous acid inhibition.
Measurements from underwater bubble energy
tests show the total energy difference on a volume
basis (cal/cc) comparing a 70%EP/30% ANFO
emulsion blend made with single salt AN emulsion
blend and a dual salt ANCN emulsion blend is 1%,
and the total energy difference on a weight basis
(cal/g) is also 1%. These measurements confirm Equation 4.
the total energies realised by a dual salt ANCN
emulsion blend is essentially equivalent to a single Urea is an inhibiting agent only and it does get
salt AN emulsion blend. The dual salt ANCN consumed, and possibly exhausted, during
emulsion blend tested generated slightly greater exposure to reactive ground and the inhibition
gas energy as compared to the single salt AN process. Accordingly, it is extremely important to
emulsion blend. See Figures 6 – Energy of single have urea inhibited emulsion formulations
salt and dual salt emulsion blend. qualified for use in reactive ground by isothermal
testing. Isothermal testing provides an
8 REACTIVE GROUND INHIBITION BY understanding of how resistant an inhibited AN
UREA based explosive product is to onset reaction and
the duration of the safe loading window before the
Urea is a nitrogen rich organic chemical inhibition is no longer effective.
compound that is not a salt. While not a salt, urea Urea inhibition incrementally reduces energy
has won its place in bulk emulsion explosive release and sensitivity. Most mild to moderately
formations as the most effective chemical inhibitor reactive grounds may be managed with 3% to 6%
in explosives that contain ammonium nitrate (AN) urea inhibition. Highly reactive ground may

Figure 7. Underwater energy test measurements for dual salt emulsions w/wo urea inhibition.

- 68 -
require urea inhibition up to 10%. To ensure − Energy values (RWS and RBS) generated by
reliable initiation with primers assemblies made manufacturer ideal detonations codes vary
using standard detonators or detonating cord from code to code. Energy values for
and/or cast primers or detonator sensitive products stated by a manufacturer are directly
emulsions, the urea content in inhibited emulsions comparable to only to formulations and
typically is usually no more than 10% of the total products made by that manufacturer. Directly
formulation. Understanding the urea content of an comparing published energy values (RWS
inhibited product is important when comparing the and RBS) between different manufacturers
energies of inhibited emulsion products. See may not be completely meaningful.
Figure 7, energy of dual salt emulsions with and − Physical testing by underwater bubble energy
without urea inhibition. tests confirm that single salt (AN) emulsions
There is a considerable amount of public and dual salt (ANCN) emulsions produce
knowledge available describing explosives and similar bulk and volumetric energies, and
blasting applications in reactive ground types for comparable gas and shock energy partitions.
those interested or needing it. − Urea is the most effective chemical inhibitor
for AN based explosives against reactive
9 CONCLUSIONS ground. Urea inhibited emulsion realise
reduced energy as compared to their non-
The purpose of this paper has been to acquaint
inhibited counterparts.
blasting professionals with bulk emulsions and
emulsion blends formulated with single salt, dual People interested in expanding their knowledge
salts, and multiple salts. on any one of the topics presented in this paper are
Summary of key takeaways and learnings for highly encouraged to explore the public domain
the reader: and interact with other experienced chemists,
researchers, and end users.
− Product stability. Dual and multiple salt
emulsions are inherently chemically resistant 10 ACKNOWLEDGEMENTS
to crystallisation and phase separation.
Emulsion stability should be considered when The authors appreciate the support and efforts of
risk assessing product selection for all BME personnel that contributed to building the
transportation survivability, resistance to body of knowledge hosted in this paper. The
breakdown by repeated re-pumping, and authors also express gratitude to the explosive
salting out during extended storage times. physical testing professionals at UTEC
− In using gassed emulsions and emulsion Corporation, R&D Lab for their all support over
blends, the maximum hole depth loaded the past several years. Special thanks to Gary Eck,
depends on product selection, hole diameter, Special Project Manager for being our key contact.
product densities, critical void space and the
presence of water. End users should consult REFERENCES
manufacturer specialists for additional
guidance and site-specific calculations before Alleman, J.E. 1998. Free Ammonia-Nitrogen
loading holes to ensure reliable application. Calculator & Information.https://www. http://
− Environmental considerations. Dual salt home.eng.iastate.edu/~jea/w3-research/free-
ANCN emulsions require less energy ammonia/nh3.html
intensity to manufacture as compared to
Atlas Power Company Basic Technical Training,
single salt AN emulsions. Dual salt ANCN
circa mid-1980s.
emulsions formulations are closely oxygen
balanced and realise minimise generation of Oluwoye, I., Dlugogorski, B.Z., Gore, J., Oskierski,
post blast NOx. Dual salt ANCN salt H.C., & Altarawneh, M. 2017. Atmospheric
emulsions formulations can significantly emission of NOx from mining explosives: A
reduce the concentration of toxic ammonia in critical review, Atmospheric Environment (2017),
mine-water thus reduce the amount of doi: 10.1016/j.atmosenv.2017.08.006.
treatment mine-water requires before release
to the environment. With concern to nitrates Unpublished internal reports. UTEC Corporation,
in mine-water, dual salt ANCN emulsions do R&D Laboratory. BME Underwater Anergy and
not offer any advantage over a single salt AN Confined VOD testing results. 2018, 2019, and
emulsion. 2020.

- 69 -
 2. Blasting work experiences

- 70 -
- 71 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Domestic water well decommissioning during Covid

J. Wallace & C. Breeds

ABSTRACT: A domestic water well, drilled and fully cased in 1946 to a depth of 340+ m in glacial
gravels, had been abandoned decades ago but never properly decommissioned. The well was now located
within the active footprint of a 30-metre deep foundation in alluvial gravels for two 15-storey high rise
buildings being constructed in the centre of the City of Bellevue, a suburb of Seattle, and the second
largest city centre in Washington State, USA. The well had to be properly decommissioned to meet
Environmental Protection Agency (EPA) and Washington Department of Ecology (WDOE) requirements.
Those requirements included perforating and pressure grouting the pipe from the bottom up to ensure that
no future inflows could travel down the old pipe into the aquifer.

Figure 1. The project site – surrounded by residential and commercial structures.

- 72 -
 Figure 2. Location of the 1946 well inside the site footprint.

1 INTRODUCTION WAC 173-160-381 [l(i)], the Washington

Department of Ecology (WDOE) well
The well was located in the southeast corner of abandonment requirements.
the site close to the soldier pile and lagging Figure 4 contains the rationale and important
shoring wall that had been installed prior to conclusions made by the WDOE when deciding to
excavating the basement structure, as shown in grant the variance required from WAC-173-160-
Figure 2. Its discovery was a surprise to the 381 [l(i)] to proceed with the work.
contractor when it was uncovered during early site It was also believed important to create a
preparations. separation between the 187 metres (612 feet) and
Subsequent research and contact with the higher and 297 metres (974 feet) perforations
Washington Department of Ecology (WDOE) which were inaccessible. The plan therefore
provided the well log, shown in Figure 3, which included setting an explosive charge and pressure
identified sequences of sand, sandstone, sandy grouting to adequately seal the lower aquifer from
clay, gravel and streaks of shale. water migration from any higher perforation
The well log also identified ‘as-built’ zones.
perforated zones at 75 to 112 metres (247 to 370 The team assembled to plan and complete the
feet), 162 to 189 metres (530 to 621 feet) and 297 work included Tacoma Pump Drilling (well
to 340 metres (974 to 1115 feet). abandonment contractor), Wallace Technical
However, a video conducted shortly after Blasting (licensed blasting contractor), and
discovery revealed a 45 cm diameter (18-inch) SubTerra, Inc. (blast consultant and vibration
well casing to a depth of 180 metres (590-feet) that specialist).
then necked down to 30 cm (12-inch) to a depth of
270 metres (890-feet) below which the conditions 1.1 Covid-19 Appears
were unknown.
It quickly became clear that the lower A major complication arose after preliminary work
perforated zone was beyond the reach of the began and actual work could be conducted. That
planned well abandonment program. complication was Covid-19. Not only did it arrive
Furthermore, the casing was so deteriorated but one of the early USA epicenters was at a
that blasting the casing with explosives was retirement home just a few kilometres from the
determined to be the best way to meet the intent of project site.

- 73 -
 Figure 3. The original 1946 well log.

There was a universal lockdown instigated and

work on the project came to a screeching halt,
which meant that a huge excavation was setting
idle with only temporary shoring installed. This
created a huge hazard to surrounding buildings and
utilities given that either a small earthquake or a
large rain event could cause the temporary wall
support to fail.
An emergency-provision exemption was
granted after a week or two of serious
correspondence between the builder, engineers and
the health authorities to allow construction to
resume, with Covid precautions being drafted on
the fly.
Among those provisions were what we now
call ‘social distancing’, with workers required to
stay 2 metres apart, a tough standard to meet when
constructing a perforating gun at the site. Other Figure 4.
WDOE
provisions included one-way stairwells for ingress Assessment
and egress on the project, morning temperature of the well
checks and symptoms questions-and-answers hazard.
daily, and sitting apart during breaks and lunch.

- 74 -
 Figure 5: Well truck in the excavation near far wall, SE View.

2 MOBILISATION AND EXECUTION OF A set of heavy tyre mats were placed over the
WORK top of the well to deviate any out flow. A small
Mobilisation for the decommissioning of the 1001 trench and or pit was also utilised to contain and
Office Towers well consisted of 10T pump hoist, pump off any water purged from the well during
Rosewall grouter, a cable winch containing 300 decommissioning. After the charges detonated the
metres of 10 mm cable (3/8-inch) and 268 metres well was prepared for the following lift. Before the
of 50 mm (880 feet of 2-inch) threaded and next set of charges were installed the hole was
coupled PVC and related tooling. This equipment tagged for depth, and the next lift then resumed;
was lowered into the excavation and positioned at this procedure was followed to within 15 metres
the well head. (50-feet) of the surface.
A 15-23 metres (50 to 75 feet) exclusion zone When the decommissioning procedure was at +
was set up around the well site and only authorised 15 metres (50-feet) below grade a shape charge
personnel that needed to be there could enter. was used to separate the 450 mm (18-inch) casing
approximately 9 metres (30-feet) below grade. The
2.1 Blasting Program remainder of the hole was filled with cement grout
leaving a solid cement plug between the top of the
Decommissioning was done in stages or lifts.
450 mm pipe (18-inch) and the bottom of the 600
After the 10T was set in place and towered up the
mm (24-inch), approximately 7.5 metres (25-feet)
spiral perforating charges were lowered into the
thick.
well via 10 mm cable. The first lift was at the
bottom with subsequent 10 to 30 metres vertical 2.2 Concurrent work
lifts (35 to 100-feet) planned at the discretion of
the blaster in charge working in consultation with The effort to stabilise the excavation meant that
the well decommissioning contractor. Some of the multiple concurrent activities were going on
tools that were used are shown in Figures 6 nearby on a daily basis (see Figure 9). Fortunately,
through 8. the general contractor was very helpful in

- 75 -
Figure 6. Scalloped gun system.

Figure 7. Unassembled guns in shipping crates. Figure 8. Retrieving successfully fired gun.

- 76 -
 Figure 9. Jobsite ongoing activities.

scheduling work to the most efficient for all, either

postponing a blast for a short while to allow for 3 CONCLUSION
completion of a concrete pour, or conversely, Perforating charges were required at a depth of as
postponing other activities for a short while to much as 340+ metres, resulting in significant static
allow us to fire the blast. pressure on the explosives. Therefore, the use of
molecular explosives was necessary. These
2.3 Monitoring program charges were designed in collaboration with the
two selected explosives manufacturers and tested
Figure 2 shows the layout of blast vibration at their sites prior to shipment.
monitoring stations set up around the site to Once onsite, the charges were lowered to the
monitor compliance with the State of Washington, designated depths where no prior perforations
WAC 296-52 Blasting Regulations. Blasting existed in the pipe, and fired. Those perforation/no
operations were designed to minimise vibration perforation locations were accurately marked on
levels and maintain vibration levels below the 1946 drillers log, a huge advantage. Pressure
regulatory requirements within the site boundaries grouting then followed the perforating as the work
to protect the surrounding structures shown in progressed upwards. The job was a success, but
Figure 1. made all the more challenging by Covid-19
arriving between the time the contract was
2.4 Evaluating impacts to the adjacent awarded, project blasting design and charge
structures manufacturing began and the actual work on the
ground.
Table 1 contains estimates of vibration magnitude
and frequency along with induced strains at 4 RECOMMENDATIONS
various distances from the blasting site that were
developed during the planning process. This Using proper explosives that would work at depth
process demonstrated that vibration levels would was a key to success. Testing offsite, with the
be below regulated criteria using conservative results documented, combined with careful record
parameters in the attenuation equation and keeping on grout consumption and rise in the pipe
provided the basis for the blasting permit that was facilitated EPA approval.
issued.

- 77 -
Table 1. Estimated stresses, strains and displacements.

This prediction was for the uppermost 30-metre zone (100-feet) in the program with 6 kg (13.5-pounds) loaded.

Figure 10 shows a plot of the actual PPV vs scaled distance for the project demonstrating the conservative
nature of the predictions and completion of the work while maintaining vibrations below regulated limits at
locations inside the site boundary.

- 78 -
 We do however live in changed times, both
from a working standpoint and a contracting
standpoint. Challenges that arose were unique in
that the cost to perform the work changed
significantly between contract award and actual
work. Negotiating the cost changes due to Covid
was as challenging, if not more so, than
performing the work. Fortunately, the emergency
nature of the work was an assist in the
negotiations.

- 79 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Wireless blasting techniques to improve ore extraction and remove

people from dangerous areas in secondary stopes

C. Walker
Chief Mine Engineer, Boliden Tara, Ireland

J.A. Kreivi
Senior WebGen Specialist, Orica, Finland

J.S.C. Marijnissen
WebGen Manager EMEA, Orica

S. Barnett
Technical Service Lead UK, Orica, United Kingdom

ABSTRACT: Ore recovery is the driving force in the success of a mine. Drill and blast challenges, ore
dilution, geological uncertainty and geotechnical issues can result in low recovery and the loss of
thousands of tons of valuable ore. Tara-Boliden is a mine that primarily produces zinc. Open-stoping
methods are used to mine the ore. Secondary stopes are mined after the primary stopes are mined and
backfilled with paste fill. Tara-Boliden’s 1835C1 stope was such a secondary stope. It was the 2nd last
remaining stope in the section, surrounded by backfill. The pillars of the stope were merely two metres
thick, and a large fault crossed the stope. Geotechnical assessment showed it was not safe to go back into
the stope to perform drilling and blasting activities once blasting started. However, with a grade of 14%
Zn it was economically attractive to mine out the stope. Tara-Boliden mine teamed up with Orica’s
WebGen™ (wireless blasting) team to find a solution. Using traditional wired blasting systems, it would
have been impossible to design a single blast that would avoid re-entry. Tara-Boliden mine took a
different approach, using WebGen and auto-loader remote mucking. After careful consideration of the
risks, it was decided to blast the raise traditionally using i-kon™ electronic detonators (blast #1). After
this, the void was increased by slashing with WebGen™ pre-charged rings (blast #2). A mass blast going
through the fault zone was taken after the void increase (blast #3). The remaining WebGen™ charged
portion of the stope was designed so it could be blasted in four separate blasts to further improve recovery
(i.e. separate the up-holes from the side/wall holes) but depended on time to muck the stope whilst
remaining inside the 30 days sleep time of the WebGen™ primers. Due to time pressures the remaining
WebGen™ primers were fired in one shot (blast #4). The last part of the stope (blast #5) was safe to
access and was fired with i-kon™ electronic detonators. Besides the design with preloaded blasts,
adjustments to the timing design were used to accommodate the best breakout of the blast holes. The
blasting result was excellent, with good fragmentation and clean break out with very limited overbreak
and underbreak. Recovery was greater than 92%. Pre-charging the stope and remote mucking avoided
re-entry to the dangerous area, making the mining operation safer while increasing productivity. This
paper will focus on the practical challenge of blasting a secondary stope and the wireless blasting
techniques that provided the solution.

- 80 -
 Figure 4. Evolution of initiation systems

1 THE WEBGEN™ WIRELESS people further from the danger zone. However,
BLASTING SYSTEM progress is limited by connections using wires and
tubes because they require people to re-enter
Blast initiation systems began with the invention dangerous areas to connect blasts and are
of safety fuse by William Bickford in the 1830s. attributable to some types of misfires. The
Since then, the three major revolutions in initiation WebGen™ system is the first truly wireless rock
systems were electric detonators, shock tube and blasting system (Marijnissen & Pereira 2019).
electronic detonators (Figure 1). Each brought Eliminating the wires from blasting enables new
better safety and productivity and helped to move mining methods on the surface and underground

Figure 5. WebGen™ uses low frequency magnetic induction to communicate with wireless primers in blastholes.

- 81 -
that were previously impossible with wired and challenging, and the secondary stopes can be
shock tube systems. difficult to mine. The 1835C1 was one such stope.
WebGen™ uses low frequency magnetic It was the second last stope of a large mining area
induction signals (1800 Hz) that can penetrate after all nearby stopes were mined out and paste
solid rock up to 800m and initiate sleeping filled.
and charged WebGen™ primers located in In addition to the high mining induced stress
blastholes. conditions around the stope, the wall pillars to the
two neighbouring stopes were less than two metres
2 BOLIDEN'S TARA MINE thick due to overdevelopment of their slot
cross-cuts. In a nearby area a few months earlier, a
Boliden’s Tara mine located at Navan, Ireland, is
similar geometry (of narrow development pillars)
Europe's largest underground zinc mine and one of
and high mining stresses resulted in the rock
the largest mines on a global scale. The Tara
pressure exceeding the rock strength, leading to a
deposit was discovered in 1970 by Northgate
sudden rock burst. Tara wanted to remove this risk
Exploration. Mine development began in 1973,
when mining the 1835C1 stope. As a further
followed by the first ore in 1977. Since this time
complication mapping had revealed a large fault
more than 85 million tonnes of ore have been
through the stope. Based on this information, it
extracted; with Outokumpu operating the mine
was concluded that mining the1835C1 stope with
from 1986 until it was acquired by Boliden in
conventional drill and blast methods was too risky
2004. Through Boliden’s continuous exploration
because after blasting the raise, the thin wall
and acquisition the mineral reserve and resources
pillars might collapse. By using conventional
have grown over time. Around 2.6 million tonnes
blasting methods, lead wires would most certainly
of ore are mined annually to produce zinc and
be cut-off when blasting the stope in multiple
lead concentrates. Boliden Tara has a strong
sections. Pre-charging the full stope was therefore
focus on cost improvement through productivity
not an option. Blasting would make the stope
enhancing investments and savings related
unstable, so it would not be safe to send in
projects.
charging crews after the first blast.
3 BACKGROUND OF THE WIRELESS Boliden Tara mine teamed up with Orica
BLASTING PROJECT AT BOLIDEN and came up with the idea to charge the stope
TARA MINE using wireless blasting technology. The
WebGen™ wireless blasting initiation system
Boliden Tara mine uses open-stoping as the main would allow the mine to pre-charge the
mining method. Primary stopes are mined and stope completely in pre-designed sections. This
backfilled with paste. The secondary stopes are would remove the risk of damaging detonator
mined after the paste fill has settled. As paste fill wires, and the risk of sending people back into the
has different properties to virgin rock, the danger zone to reconnect wires between each
geotechnical situation of secondary stopes is often blast.

Figure 6. Boliden Tara Mine.

- 82 -
4 BOLIDEN TARA AND ORICA WORKING increased the uncharged collar length on the very
TOGETHER ON STOPE BLAST DESIGN last ring of each blast section (Figure 4). Also, the
WITH WEBGEN™ collars of nearby holes were too close to each
other, so uncharged columns and delay times
The stope was already drilled when Orica’s between the holes were adjusted to avoid
WebGenTM team arrived on site. A team of unnecessary damage.
experienced underground mining specialists To improve toe accuracy and provide better
reviewed the drill and blast design and conformity, a one metre pillow charge was
recommended alternative delay timing. One of the introduced. The charging method included a one
recommendations was to speed up the delay times metre pillow charge in the toe of the hole, before
for the raise to avoid overbreak of the raise. Raise placing the primer. This method was introduced to
blasting used i-konTM electronic detonators. The ensure good breakout at the toe.
total blast duration for the raise was decreased by Some of the design peer review
2 seconds to speed up the blast and to improve the recommendations around blast sequencing and
blast result. This timing was discussed with double priming could not be implemented due to
Boliden Tara, and they decided to use the geotechnical concerns and partly blocked holes.
suggested timing design. The original recommendation from the review was
With the introduction of the alternative raise to sequence the stope so the up-holes (a maximum
timing, an improved blast sequence for the stope of 5 to 6 rings) were blasted first. The side slash
was also suggested. Traditionally the roof and wall rings would be fired after mucking to provide
sections were blasted at the same time. With more relief and improve mucking of the upper
wireless technology, a new sequence was holes by creating a funnelling effect. However,
recommended to improve fragmentation and this was not possible because the side pillars on
produce better wall and roof control. The idea was both sides of the stope were only 1-2 m wide
to first blast the up-holes with WebGen™ and (Figure 4.). Boliden Tara mine’s Geotechnical
blast the side rings into the void using WebGen™ Engineer recommended to pass through this area
after completely mucking the stope out, to create a with one big blast rather than slice it to smaller
void. sections and risk a failure.
The plan was to blast the stope in multiple With this recommendation in mind, Orica
pre-charged sections, made possible due to the suggested that the final design should be sliced
flexibility of WebGen™. The first ring of each into eight blasts, including two i-kon™ electronic
section was double primed to reduce the risks of detonator blasts. The other six blasts were
failure. Pre-charging with wireless primers must separated so each section could be fired
be done with caution. Backbreak can cause unfired independently or combined if required. The blast
primers from the next unblasted ring to end up in sections are shown in Figure 5. (GID stands for
the muck. This is called a liberated primer. With Group Identification. GID is used to group
this in mind, the planning team introduced a longer charged holes that form a blast. They can be
delay time for these blast intersections and merged with other GIDs to make larger blasts).

Figure 7. Picture of thin pillars on both sides of the stope.

- 83 -
 Figure 8 Blast sections

− Blast 1 (i-kon) and any re-drills exposing personnel to potentially

− Blast 2 (GID 1 & 2) unstable ground. With WebGen™ these costly,
− Blast 3 (GID 3) time consuming processes and unnecessary risks
− Blast 4 (GID 4) are eliminated.
− Blast 5 (GID 5) Charging the WebGen™ rings took about one
− Blast 6 (GID 6) week including a few interruptions due to mine
scheduling. The stope was released for one day for
− Blast 7 (GID 7)
the auto-remote mucking crew to train their
− Blast 8 (ikon)
equipment for that specific stope, and on another
day the mobile charging unit was required by
5 CHARGING THE FIRST WEBGEN™ another blast crew (Figure 6).
BLAST AT BOLIDEN TARA MINE Most of the holes were charged according to
the charge plan, with a few exceptions. Some
In total 495 WebGen™ primers were charged in
holes were dislocated, shorter than planned,
the 1835C1 stope. The whole stope (except for the
blocked, or drilled into the backfill. On some holes
final blast 8 – ikon) was pre-charged before any
the WebGen™ primer’s location was adjusted due
blasting (including the i-kon™ raise blast). The
to a blockage or hole dislocation, and on some
capability to pre-charge is crucial for extracting
holes the second primer had to be left out.
stopes like 1835C1, where geotechnical concerns
Exceptions from the design were noted on 31
limit stability and recovery using conventional
holes. (Figure 7.)
blasting methods.
Even though the stope was divided into eight 6 BLASTING WITH WEBGEN™ AT
blasts with seven GID’s, no additional time delays BOLIDEN TARA MINE
were applied between sections. The stope blasting
sequence was designed as if it was one single blast WebGen™ blasts can be initiated using LAN or
in case it became necessary to fire it all at once. WLAN communications. If such a network is not
With conventional methods, the time between available, blasts can be fired from the transmitter
blasts would have required the brow area to be location or by running a hardwired ethernet cable
inspected (plus possibly supported with shotcrete), from the antenna to a safe location.

Figure 9. Charging table.

- 84 -
 Figure 10. Blasthole exceptions.

All WebGen™ blasts were successfully fired an auto-operated loading unit, but after
using Boliden Tara’s WLAN network. This Blast 3 Boliden Tara faced scheduling issues
created more flexibility, as WebGen™ could be that caused an unexpected delay for mucking.
easily and reliably blasted from the surface control To comply with the maximum allowed
room. sleep time of WebGen™ it was decided
The first blasts (1-3) were fired as originally to merge blasts 4-7 together, as one big blast.
planned. The raise was blasted on the 27th This showed the advantage and flexibility of
of September 2021, and after few days of WebGen™ and a problematic scenario was
mucking, GIDs 1 and 2 were merged and avoided. The last WebGen™ blast was fired on
blasted on 2nd of October 2021. Blast no. 3 the 22nd of October 2021. Boliden Tara blasted
(GID 3) was blasted on the 5th of October 2021. the final rings of the stope with i-kon™ once
It was planned to muck the whole stope with everything was mucked out.

- 85 -
7 BLAST RESULTS have prevented safe drill & blast operations in this
particular stope.
Overall, charging and blasting with WebGen™ During 2021, the average compliance to plan in
worked very well, creating approximately 1.5 m of the mine was 86%. Compliance in the WebGen™
over break in the up-holes, but all the rest broke to stope was 92.2%, more than 6 percentage points
design (Table 1). No misfires or liberated boosters above the 2021 average. The non-compliance of
were detected. All measures were done to prevent 7.8% was mostly from the walls of the stope due
elevated risks during charging and blasting. to a combination of: drill hole stand-off to adjacent
filled stopes, blocked holes and open joints at the
Table 1. Stope 1835C1 reconciliation report. back of the stope.
Original design solid 8 FIGURES OF RECONCILIATION FILES
volume 18,361 m3
As-built solid volume 21,680 m3 Shown in figures 8 to 11 below.
Design solid volume
After mining 17,957 m3 REFERENCES
As-built solid volume Marijnissen, J.S.C. & Pereira, N.B. 2019. Wireless
After Mining 18,754 m3 Blasting in Europe. Tenth EFEE World
Overbreak volume 2,206 m3 Conference on Explosives and Blasting 2019.
Underbreak volume 1,409 m3
Designed volume
actually mined
calculated 16,548 m3
Designed volume
actually mined cut 16,548 m3
Compliance to plan 92.2 %
Overbreak percentage 12.3 %

Underbreak percentage 7.8 %

It is worth noting that this stope could not have

been charged and blasted without using WebGen.
Geological and geotechnical challenges would

Figure 8. Final Stope scan of 1835C1 reflected to the original design.

- 86 -
Figure 9. Final Stope scan of 1835C1 reflected to the original design.

Figure 10. Final stope scan of 1835C1 on Orica Shotplus™ software with drill holes and neighbour stopes.
(WebGen™ holes visible).

Figure 11. Sideview of 1835C1. Orica Shotplus™ software. (WebGen™ holes visible).

- 87 -
 3. Blast vibration and seismology

- 88 -
- 89 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

A comparison of recent and previous pipeline blasting studies

C. Dowding
Northwestern University

T.A. Davidsavor & D.A. Provost

Barr Engineering Co.

Y. Gou
Central South University

ABSTRACT: When blasting is conducted adjacent to operating pipelines the main concern for those
pipeline operators is that the pipeline is not subjected to ground motion that causes damage or over
stresses the pipeline. Several studies have been conducted over the past few decades that have recorded
and analysed the strains on the pipelines during blasting. These studies include surface mining blasts and
response of several segments of steel and plastic test pipeline, trench blasts with small number of holes
and response of water lines and decommissioned steel, and a Barr 2016 study involving trench blasting
with over 50 holes parallel to and in between to two operating pipelines. This paper reviews the difference
in the studies, compares observed peak particle velocity measured at surface seismographs to measured
pipe strain and stress, as well as strain and particle velocity time histories. Similarities and differences of
the findings are discussed. Each study indicated blasting was performed successfully without apparent
damage or overstress of the pipe even at relatively low scaled distances as long as the pipe being
monitored was outside of the crater zone of the blast.

1 INTRODUCTION above 10 µstrain and peak particle velocities

above 1 inch per second (ips) (12.7 mm/s). Some
This paper compares measurements of strains in of the blasts were detonated as close as 4 ft (1.3 m)
large diameter welded steel pipelines produced by from the pipeline and produced strains as high as
actual trenching blasts having more than 50 holes 300 µstrain. Strain and peak particle velocity
per blast with three other theoretical studies to attenuation relationships are compared on the
confirm the measured and calculated strains. same scaled distance graphs for all four studies to
Additionally, the paper comments on the illustrate the effects of geology and shot type.
intricacies of strain measurement as well as the Measured strains are compared between the
fact that damage in the pipelines was not observed. studies to investigate the proper interpretation of
The four studies evaluated in this paper were strain time histories. Maximum stresses are
conducted at scaled distances between 1.0 and 30 calculated for comparison with the specified
ft/lb1/2 (0.5 to 14 m/kg1/2) and produced stains maximum yield stress of the pipes.

- 90 -
Table 1. Summary of blast parameters.

Study
Parameter
Unit SwRI USBM VME Barr 2016
Number holes Average 15 50 6 39 61
Depth holes ft 16 40-65 10 12
Diameter holes inch 3 10-12 ? 3
Burden ft 6 15 varies 5
Spacing ft 6 15 varies 2.5
Total explosive wt. lb ~225 ~900 15-50 255 501
Typical no. of delays 15 20 2 78 122
Delay interval ms 25 126-rw/ 25-hl ~1000 8
Explosive type 60% Ex Gel ANFO? ? Emulsion
W per delay min. lbs 15 668 5 2.5 3.75
W per delay max. lbs 20 964 19.5 4.16 5.83
Rock type Igneous Shale ? Limestone
Depth of trench ft ~5 ~5 ~5 ~10
Soil thickness ft <5 7 2 0
Pipeline diameter in 30 20 34 20 & 36
Pipe wall thickness in 0.47 0.26 0.31 0.34
Pipe steel yield, σy ksi ? 56 42 46

Figure 1. Blast hole and pipe measurement locations (top) and elevation view showing the stand-off distance
and pipe depth (bottom) for the Barr 2016 study.

- 91 -
 Figure 2. Cross sections showing stand-off distances for the four studies at the same scale.

2 BACKGROUND AND BLAST DESIGN trenching study did not identify the rock type, but
AND GEOLOGY DIFFERENCES photographs indicate that it was unlikely to be
sedimentary and was most likely igneous. The
Details of the four studies are compared in Table 1 VME study also did not identify the rock type, and
and the shot geometries are compared in Figures 1 unfortunately, there were no photographs provided
& 2. More details on the Southwest Research to allow visual identification. However, this study
Institute (SwRI), United States Bureau of Mines was conducted for a Seattle, Washington area
(USBM) and VME studies can be found in municipality, so the rock type is likely to be either
Dowding and Akkaya (2001) and the more recent igneous or metamorphic. The Barr 2016 study was
Barr 2016 study in Davidsavor and Provost conducted in limestone in support of active
(2020). pipeline construction.
As can be seen in Figure 1, the two shots in the Shot designs across the four studies varied
Barr 2016 study differed in their design. Holes in considerably. As shown in Figure 2, hole depths
the Barr 2016 verification blast 1 were designed as for the mining study were some 66 ft (20 m) deep
a 5 ft (1.5 m) ‘5 spot or 5-of-dice’ pattern while while those for the trenching studies were only
Barr verification blast 2 was designed as a 10 ft (3 some 6 to 10 ft (2 to 3 m) deep. None of the SwRI,
m) wide reoccurring 3 then 2 holes per row down USBM or VME shots involved any decking. The
the length of the blast. The change in pattern was Barr 2016 shots were decked with 1 ft to 1½ ft (0.3
to assess the geometric responses in preparation m to 0.5 m) of crushed stone stemming. Only the
for future blasting on the project. VME study involved more than one hole being
As can be seen in Figure 2, the blasting detonated at the same delay, although the use of ¾
geometry of the USBM surface mining study to 1 second delay intervals in the VME study
differs radically from that of the trenching resulted in non-simultaneous detonation in several
geometry of the SwRI, VME, and Barr 2016 instances. The USBM and SwRI studies employed
studies. The 800 to 900 lb (360 to 410 kg) per standard 25 millisecond delay intervals between
delay detonations associated with mining in the single blast hole detonations. The Barr 2016 study
USBM study result in large stand-off distances employed nonelectric initiation with 17 ms
from the pipeline. This would be expected as the interhole delays and 25 ms decking delays, all with
USMB study focused on the effects of 350 ms down hole delays. This resulted in nominal
mining-related blasting produced by mining 8 ms separation between any individual charge.
pattern blasts on pipe stress, whereas the other The Barr 2016 study represents present-day large-
studies focused on the effect of pipeline- scale pipeline blasting practices.
construction trench blasting on adjacent pipelines.
Even though the scaled distances are similar for 3 INSTRUMENTATION DIFFERENCES
the three previous studies, the stand-off distances ACROSS THE STUDIES
for the mining shots are large in an absolute and
wavelength scale. Instrumentation across the studies was relatively
The rock type across the four studies varied similar. Strains were measured with wire strain
considerably. The USBM mining study involved gages in all four studies. Both weldable and epoxy
shots at a coal mining area in shale. The SwRI mounted gauges were employed in the USBM

- 92 -
study. Although not described explicitly, it appears gauges were located at the crown and each side of
that the strain gages were epoxy mounted for the the pipe. Only the largest of the longitudinal and
SwRI study. The VME study employed weldable hoop strains is reported herein as was the case for
gages. The Barr 2016 study employed epoxy the other studies.
mounted strain gages. All four studies employed The cross sections in Figure 2 do not reveal
standard blast vibration seismographs. Unless sufficiently the differences in the full shot
modified, standard seismographs may not return geometries as well as does a plan view of the blast
sufficiently accurate readings in particle velocity hole locations relative to the distance to the
environments above 5 ips, which compromised the pipelines being monitored. Figure 3 presents blast
measured PPVs in the SwRI study at small hole geometries at the same scale and
stand-off distances. demonstrates uniqueness of the Barr 2016 study.
The more recent Barr 2016 study involved While the other two studies that involve trench
more intensive instrumentation of strain gages and geometry are similar in location, they involve
seismographs. Above each strain gauged location, fewer blast holes in much shorter segments.
two, three-axis velocity transducers were located; While the USBM blasts involved a similar
one just above the pipe and the other 10 to 12 in number of holes, the spacing, burden, and
(25 to 30 cm) below the ground surface. All stand-off distance of the surface mining
seismographs were placed on and in native soil blast are much greater. The overall location,
that was compacted with significant effort. Particle geometry, and arrangement of the holes and
velocities reported herein are those of the shallow explosives in the Barr study are more typical
burial velocity transducers, as opposed to the of a production blast for a pipeline trench
transducers nearest the pipe. Two sets of strain excavation blast.

Figure 3. Example of blast geometrics.

- 93 -
 kg) per delay, while the SwRI, VME, and Barr
studies reflected trench type construction with
some 4 to 15 lbs (2 to 7 kg) per delay. Figure 4
provides a comparison of the attenuation of peak
particle velocity (grey diamonds) and strain
(black squares) of the single and multiple pulse
interpretations of the Barr study (rightmost two
graphs) and the other three (left three graphs).
Differences in PPV attenuation rates and
amplitudes result from the differences in the
geology, excavation and blast hole geometry,
explosive type, as well as initiation sequence, and
powder factor. In addition, some velocity
transducers from the previous studies may not
have responded properly at high PPVs. The effect
of shot design will be discussed later in the section
dealing with time histories. For instance, the PPV
in the VME study decays more rapidly and has a
higher PPVs at the same scaled distance as the
other studies. The nearly seconds-long delay
between velocity peaks indicates that the blasts
may have involved tunnel delays rather than the
more standard millisecond delays used in modern
pipeline blasting. PPVs in both the SwRI and
USBM studies at scaled distances below 5, fall
below the attenuation slope projected from larger
scaled distances. Many velocity ‘geophones’
‘pin-out’ at PPV around 4 ips. The Barr 2016
study has the lowest PPVs at the same scaled
distance and was detonated with NONEL initiation
with 350 ms downhole delays with 8 ms surface
delay sequences. None of the studies evaluated in
this paper included the use of newer electronic
delays.

4.2 Maximum strains

Despite the fall-off of the measured PPVs at small

scaled distances, Figure 4 shows that the strains in
all four comparative studies continue to increase
according to a power function. The y axis on these
graphs represents both PPV and maximum strain
Figure 4. Comparison of SD vs strain and PPV. and requires an explanation. The whole numbers
have different decimal values. The PPV whole
4 RESULTS numbers are in inches per second. The strains are
micro strains or 10-6 times the whole number.
4.1 Peak particle velocity Thus, shot 19 in the SwRI study produced a PPV
of 5 ips (125 mm/s) and a strain of some 180
Attenuation relations for the four studies are micro strain.
compared in Figure 4 beginning on the left For the most part, maximum strains at small
(lowest) with SwRI, and then moving right scaled distances continue to follow the power law
(upwards) USBM, VME, and Barr 2016 twice on determined from scaled distances greater than 10.
the right. This order will be maintained throughout There are only two shots (both in the SwRI study)
the discussion. As described earlier, the USBM that do not follow that observation in this suite of
blasts were those from surface coal mining and data. The two exceptions are shots 19 and 2. These
involved charges per delay of some 850 lbs (400 two shots and shot 20, which was detonated within

- 94 -
4 ft (1.25 m) of the pipeline and at a distance of methods and their use with these comparative data
one quarter of the 16 ft deep blast holes. As will be can be found in Dowding and Akkaya (2013) and
discussed later, this geometry for shot 20 places will not be discussed further because of space
the pipeline in the crater zone, and thus is not constraints.
representative of prudent design or actual practice.
There are two plots of attenuation of strain for
the Barr 2016 study on the right side of Figure 4.
These two different interpretations of the
measured strains are presented because of the
techniques employed to extract peak strains from
the recorded noisy time histories. They result in a
single (second to the right graph) and multiple
pulse strain time histories (furthest right graph).
Details of these extraction techniques are not
included in this paper due to space constraints.
They are available upon request from C. H.
Dowding or Y Gou.

4.3 Maximum Stresses

Blast induced stresses are compared in Figure 5 in

the same order as were strains. Comparison of
stresses calculated with the biaxial stress-strain
equation (black dots) with the single and multiple
pulse interpretations of the Barr 2016 study
(rightmost two graphs) and the three comparative
studies (left three graphs) are shown. Since only
deformation or strain can be measured, stresses
must be calculated regardless of the approach.
Some approaches to calculating stress are more
direct and more closely related to the strain, which
can be measured. The calculation most closely
related to the measured strain is that calculated by
the biaxial stress-strain relationship, which is
represented in the Barr 2016 study by the black
filled circles in the three comparative studies. The
equation shown in the key demonstrates the use of
the hoop and longitudinal strains (ϵhoop, ϵL) in
combination with the modulus of elasticity, E, and
Poisson’s ratio, ν. In the four comparative studies,
maximum strains, no matter the time of
occurrence, passage of the blast wave, or its sign
value (positive or negative) are added to calculate
the stress. Common time of the strain gauges in
the Barr 2016 study have allowed calculation of
stress hoop and longitudinal at the same ms time.
The other two methods of calculating
Figure 5. Comparison of SD vs stress.
maximum stress are both based upon some type of
scaled distance relationship. The SwRI method,
which is the most commonly used industry As was the case with the strain – scaled
approach due to simplicity, is represented by the distance plots, there are two plots of attenuation of
open circles in Figure 5. Stresses calculated stress for the Barr 2016 study on the right side of
through the compliant ground strain approach is Figure 5. These two different interpretations of the
represented by the filled squares. This method is calculated stresses results from the two different
explained in detail in Nyman, et al. (2008) and is techniques for extracting peak strains from the
abbreviated as NDO. Details of these other noisy strain time histories. Implication of the

- 95 -
differences in calculated stresses will be presented responses at similar absolute and scaled distances.
in the discussion section. Both techniques result in The single pulse pipe strain response, AN1 is
stresses that are small with respect to the specified shown in the top right of Figure 7. Particle
minimum yield stress of the pipes in the Barr velocities AN1 are shown on the bottom left. Pipe
study. strain response and particle velocity ground
motions at the pipe measured by VME from the
5 INTERPRETATION OF STRAIN TIME 10-hole, 3-delay blast is shown in the centre. As
HISTORIES shown in Figure 4 and Table 1, shot and pipe
geometry and charge weights per delay are similar
The interpretation of the Barr 2016 strain time to the VME study. Yet, filtering of only the high
histories shown in Figure 6 involves three different frequencies results in a single pulse response for
considerations: the high noise level of the strain AN1 while the VME measurements result in
readings, the unusual nature of the strain shape multiple pulse pipe responses. Furthermore, the
after high frequency filtering, and the offset of the particle velocity time histories for AN1 at the pipe
average during the recorded two seconds. Both show multiple responses as do the VME
longitudinal (top in Figure 6) and measurements. Comparison of single (top right)
hoop/circumferential (bottom) strain time histories and multiple (top left) interpretation of the Barr
are shown. As shown in Figure 4, one of the AN1 strain time histories with those from VME
highest pipe responses occurs is at location AN1 (centre) and the closest SwRI shot (4 ft, 1.25 m)
(maximum response to blast 1 at position N on within the crater zone (bottom right) demonstrate
pipe A) The raw signal on the left in Figure 6 the unusual nature of the single pulse
demonstrates the high ratio of noise to signal even interpretation of the Barr 2016 AN1.
for one of largest pipe strain responses. Comparisons of shot geometry and scaled
Figure 6 compares raw and filtered pipe distances with the other studies demonstrates that
response strain time histories in the Barr 2016 the single pulse interpretation of the AN1
study to blast 1, recorded at pipe A, position N distortion requires further discussion and
(AN1) showing effects of filtering of high verification. The only other single pulse strain
frequencies and filtering of high and low time history of full-scale field response was
frequencies. Filtering of the high frequency noise produced by a 13 hole shot recorded on an
shown in the centre of Figure 6 presents the inoperative pipe by SwRI shown the bottom right
resulting single pulse pipe response. of Figure 7. In this instance the pipe was closer
This single pulse for the high frequency filtered than one blast hole length as shown in the
AN1 response is unusual as shown by the comparisons of elevation views of the blast hole –
comparisons in Figure 7 with other measured pipe pipe geometry of the four studies in Figure 2. AN1

Figure 6. Comparison on raw and filtered strains in the Barr study.

- 96 -
 Figure 7. Comparison of single (top right) and multiple (top left) interpretation of the Barr AN1 strain time
histories with those from VME (centre) and the closest SwRI shot (bottom right) that demonstrate the unusual
nature of the single pulse interpretation of the Barr AN1.

was more than three blast hole lengths away. The present offsets of this nature except for shot 20
scaled distance for the close SwRI shot was 1 (shot from the SwRI study at a scaled distance of 1. The
20), whereas it is much greater (scaled distance of question then becomes, what is the significance of
17) for AN1 in the Barr 2016 study. these offsets? Long term measurements of crack
Comparisons of blast hole-pipe proximity and displacement in structures (Dowding 2008) show
scaled distances of AN1 with the other responses that long term temperature and humidity
indicate that filtering of the low frequency distortions are significant and overwhelm any
component of the single pulse AN1 in the upper short-term offsets shown in AN1. Interpretation of
right of Figure 6 should be considered seriously. the significance of the offsets requires knowledge
Time histories on the right side of Figure 6 show of the long-term thermal responses of the pipelines
the results of filtering both the high and low to changes in both environmental and operating
frequency components of the raw strain time conditions including pressure and installation
histories. In the left of Figure 7 one second of the stresses.
hoop strain time history after both high and low The USBM study does not discuss an issue
filtering (top left) is compared with the particle with a high frequency noise or offsets. Data
velocities at the pipe (bottom left). The length of logging instrumentation for the USBM study was
the velocity time histories is about 0.6 seconds. housed in a field shed to which grounded AC
Whereas the length of high and high and low power was provided. The USBM pipes were short
frequency filtered strain time histories is about 0.8 and were isolated sections, not part of an active
seconds. line. Power for the Barr 2016 project was provided
The strain remaining at the end of the two by a field generator which may not have been
seconds of recording in Figure 6 also requires sufficiently grounded. Additionally, pipelines
discussion. Filtering only the high frequencies monitored in the Barr study were the only
preserves the offset whereas filtering of both the operating pipelines and were subject to currents
high and low frequencies does not preserve the associated with cathodic protection, which induced
offset. The strain offsets are less than half of the noise. Thus, in future through additional
peaks associated with the high frequency filtering monitoring and analysis including baseline
interpretation. None of the other studies discuss or monitoring of the cathodic signature and effective

- 97 -
filtering, it is possible to believe that this noise can the ground surface is the present industry accepted
be eliminated from subsequent full-scale tests. approach for blasting near pipelines. The vibration
Ultimately, the effect of the noise of the cathodic limit set for each project varies widely across
protection on Barr 2016 active pipe is not fully owner, segment of pipe, and purpose (whether
understood. In either regard, effects of the strain they require blasting for their own new installation
offset were not noticeable in the pipe operation. or maintenance, or others are blasting for a project
Comparison of the attenuation of both peak near their facility). Technological advancement in
particle velocities and strains in Figure 4 of the both monitoring equipment and data processing
Barr study and the other three studies provides allow for faster interpretation of directly measured
another perspective of the two interpretations of pipe strain (and therefore stress) measurement.
the stain time histories. High vs. high & low Ultimately blasting near pipelines is constrained
frequency filtering or single vs. multiple pulses are by regulations limiting induced stress. Measuring
shown in the two rightmost attenuation plots. the induced strains (and calculating stresses)
Compared to the other three studies, the Barr 2016 provides a more direct means of calculating
study’s single pulse interpretations produce strains induced stresses than measurement of ground
that are unusually high for both 1) scaled distances vibration (and subsequent calculation of ground
at which they were measured as well and 2) peak strain and stress). Challenges in exposing the
particle velocities (PPV) with which they are operating pipe and installing direct-mount strain
associated. The multiple pulse interpretation of the gages will remain, but continued strain monitoring
strain time histories more closely matches strains while performing ground vibration monitoring
measured in the other studies at similar scaled near the pipe will allow for a stronger correlation
distances (vertical lines). They also better match of pipe strain and ground motion to be developed.
the strains associated with similar PPVs. This Reduction of noise in data collection can be
comparison is highlighted with the horizontal better achieved by better shielding data collection
(same strain) horizontal (same PPV) in Figure 4 lines, providing power from a utility rather than
and vertical lines (same scaled distance) in from a portable generator or improving generator
Figure 5. grounding. Studying the noise in the data that is
present in the pipeline from cathodic protection
6 CONCLUSIONS and other induced currents such as paralleling
overhead high voltage will aid in higher quality
The recent Barr 2016 study most closely field data and will further allow more refined
represented actual blasting conditions during signal processing. Finally, filtering the data to
pipeline construction for nearby blasting as the reduce the noise is necessary and determine what
work was conducted for an actual pipeline frequency ranges are appropriate to filter will
replacement project in field conditions under allow more effective analysis of the actual strain in
typical constraints and realities of a pipeline the pipe.
right-of-way. The scaled distances and offsets Stresses calculated from directly measured
were practical and effective at achieving the strains in the studies were low compared to the
desired excavation while limiting ground specified maximum yield stress of the pipes. No
vibrations. While pipe stresses varied across the reported damage occurred to the pipelines in the
four studies, the maximum strains at small scaled studies. As a result, the studies indicate that
distances continue to follow the power law blasting can be effectively performed in relatively
attenuation relationship determined from scaled close proximity to and at relatively low scaled
distances greater than 10. The only exception to distance to pipelines without damage provided
this consistency was the SwRI study which blasting is not conduced within the crater zone.
included two measurements within the crater zone.
Designing a pipeline trench or construction blast REFERENCES
where the pipe being monitored is within the crater
zone of the blast should be allowed only in Davidsavor, T.A. & Provost, D.A. 2020. Direct stress
extraordinary and highly controlled circumstances monitoring of pipelines and development of
(e.g. near high yield strain plastic pipe). Additional limitations – A Case Study. Proceedings 46th
studies are needed that focus on pipelines that are Annual Conference on Explosives and Blasting
either in service or are adequately constructed to Technique, Denver, CO.
reflect field conditions. Dowding, C.H. 2008. Micrometer crack response to
Establishing a limit of peak particle velocity vibration and weather. International Society of
and measuring ground vibrations near pipelines at Explosives Engineers, Cleveland, OH, 409 pgs.

- 98 -
Dowding, C.H. & Akkaya, U.G. 2013. Importance of
ground strain in predicting blast-induced strain
and stress in pipelines. Proceedings: RocDyn1
EFPL Lausanne, June.

Nymon, D.J., Dowding, C.H. & Oriard, L.L. 2008.

Evaluation of close-in blasting effects on welded
steel pipelines. Proceedings 7th International
Pipeline Conference, Calgary, Alberta, Canada.

- 99 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Building damage due to vibration from rock blasting –

instrumented test in Norway

K.M. Norén-Cosgriff & C. Madshus

Norwegian Geotechnical Institute, NGI, P.O. Box 3930 Ullevål Stadion, N-0806, Oslo, Norway

N. Ramstad
Multiconsult, P.O. Box 265 Skøyen, N-0213, Oslo, Norway

ABSTRACT: Building damage from blast vibrations is seldom observed in Norway. This may indicate
that the limit values are unnecessarily strict. In this study, two instrumented blast tests were performed in
a rock quarry. Two buildings made of Light Expanded Clay Aggregate (Leca) blocks, and one in
cast-in-place unreinforced concrete were erected. Two buildings were founded on rock and one on top of
an about 4 m thick filling. The test buildings were instrumented with geophones and strain sensors in
multiple positions. The blast tests were designed to give increasing vibration values starting at low values
and ending at vibration values more than five times the limit values according to the Norwegian standard.
Despite this, no visible damage was detected. The results indicate that today’s limit values in Norway
include a large safety margin for buildings on rock as well as on compacted stiff soil, when considering
damage to outer walls.

1 INTRODUCTION For this reason, two instrumented blast tests

were performed in Spulsåsen rock quarry in Våler
Construction activities such as blasting, piling, municipality in Hedmark, Norway, in November
compaction, excavation, and construction traffic 2018 and November 2020.
can produce vibrations of enough strength to cause
damage to neighbouring buildings and structures. 2 VIBRATION GUIDELINE LIMIT VALUES
Limit values for vibration from construction work IN NORWAY
are given in Norwegian Standards. However,
building damage assumed to originate from The Norwegian Standard NS 8141:2001 gives
vibrations is seldom observed. This may indicate guideline limits for vibration from ground work.
that today’s limit values are unnecessarily strict. These are values that buildings are supposed to
The determination of true limit values is very withstand through repeated exposures and are
important since they restrict the efficiency of the intended to prevent damage. The effect of the
blasting process, and hence, strict limit values can vibration frequency is indirectly considered by the
delay the progress and increase the costs. Little fact that the limit value depends on the distance,
new research has been done, and there is ground conditions and foundation method, in
particularly a lack of information about which role addition to the properties of the structure and
the frequency of the vibration plays. vibration source. For residential buildings the

- 100 -
guideline limit values for vibrations measured at
foundation level in vertical direction may vary
from a PPV of about 3 mm/s to about 80 mm/s.

3 DESCRIPTION OF TEST SITE AND

BUILDINGS

In the first blast test, two test buildings were

erected at the site, one in cast-in-place concrete
and one made of lightweight construction blocks.
Both were founded on an approximately
500 mm levelled and compacted layer of
gravel, over rock. The dimensions of both
buildings were 5 x 2 x 2.4 (l x w x h) metres.
The buildings had one door opening and one
window opening each.
The concrete building had 200 mm thick
concrete walls without reinforcement, on top of a
400 mm wide wall footing of reinforced concrete.
The walls and footing were cast-in-place with C30
grade concrete. The concrete was allowed to cure
for 30 days before the blast experiments were
performed. The lightweight construction block
building was constructed from 250 mm Leca
blocks (lightweight expanded clay aggregate) with
plastered outer surfaces. This building was
founded on top of a wall footing made from Figure 1. Test buildings and test areas. Upper)
330 mm wide Leca foundation blocks with first blast test. Lower) second blast test.
reinforcement. Reinforcement was also used above
the door and window opening.
In the second blast test, one building made of Before the first blast test a geological survey of
lightweight construction blocks (Leca) was the test site was performed by use of Lidar and
constructed on top of an approximately 4 m thick Orthophoto from drones. The geological survey
filling, established at the same location as the showed fine to medium-grained red granitic gneiss
buildings in the first test series. The test building containing lenses of amphibolite. The dominating
had one door opening and two window openings, direction of foliations is from the blasting area
and the dimensions were 7 x 3 x 2.4 (l x w x h) towards the buildings. In addition, the location of
metres. The fill was constructed of material from the buildings, sensors and boreholes were
the quarry with properties considered determined by GPS surveying.
approximately as moraines. The fill was
established in layers with careful compaction 3.1 Instrumentation
between each layer.
At the top of all building, joists were laid and In the first blast test, eight three-axial velocity
filled with crushed rock to simulate the mass and sensors (geophones) and eight strain sensors were
ground pressure from a typical detached house on mounted on each of the two buildings. In addition,
top of a basement. vertical vibration in three positions on ground
The calculated guideline limit values calculated and air blast pressure in two positions were
according to NS 8141:2001 were 50 mm/s for the measured.
buildings on rock and 16-23 mm/s for the building In the second blast test, 13 three-axial velocity
on filling. The buildings were visually inspected sensors (geophones), 16 strain sensors and one
between each blast round to detect and document microphone for air blast pressure were mounted on
any damage. In addition, the results from the strain the walls of the test building. In addition, two
measurements were reviewed correspondingly to three-axial geophones were mounted on top of the
detect any changes not visible to the naked eye. filling, two three-axial geophones were mounted
Figure 1 shows two views of the test buildings and on rock behind the filling and one vertical
the test area. geophone was mounted on rock under the filling.

- 101 -
 For the geophones and air blast microphone the first four rounds were all shot by single hole
AVATrace M80 measurement system was used initiation, with a delay between each hole in a row
with a 6000 Hz sampling frequency. Each three- of 10 ms, while round five was shot by two and
axial sensor was connected to its own autonomous two holes simultaneously starting in the centre of
four channel logger. The fourth channel of each the rows. The delay between the rows varied from
logger was used as a joint trigger channel. One 10 ms to 60 ms. The total amount of explosives
vertical geophone in front of each building were detonated in one blast round varied from 6.5 to
used as trigger geophones, ensuring that all 404 kg, and the explosives detonated per delay
measurement channels on each building were varied from 3.0 to 37.8 kg. The blasts were
mutually synchronised. designed to give equal dynamic loading on each of
The strain measurements were performed with the two test structures. The first blasting round had
a fibre-optic measurement system from Micron a minimum distance of 29 m from the test
Optics, using os3510 Fiber Bragg Grating Sensors structures, while the last had a minimum distance
(FBGS). The strain sensors were attached via rigid of 7 m.
brackets bolted to the structures. The strain sensors In the second blast test, four blasts rounds were
were mounted on the walls in a 45º angle on intact fired consisting of all together 220 charged holes.
walls, above door openings and above and below The number of holes detonated in one blast round
window openings. Dynamic strains were measured varied from three to 98 holes. The first round
over a gauge length of 110 mm. The sampling consisted of three single constricted charges,
frequency of the strain measurement system was which were shot with three seconds in between.
1000 Hz. This was considered enough since the The second and third blast rounds were shot by
frequency content of blast vibration usually are single hole initiation, with a delay between each
well below 500 Hz. This was also later confirmed. hole in a row of 5-10 ms in blast round two, and
2-10 ms in blast round three. In the fourth blast
4 EXECUTION OF BLAST TESTS round the last row was shot by three and three
holes simultaneously, while the delay between the
The blasts were designed to give increased holes in the other rows was 2-5 ms. The total
vibration strength, starting at a low value for the amount of explosives detonated in one blast round
longest distance and increasing progressively as varied from 12 to 1485 kg, and the explosives
the blasts came closer to the test structures. detonated per delay varied from 2.0 to 47.8 kg.
To gain full control, repeatability and The first blasting round had a minimum distance
traceability of the blasts, packaged emulsion and of 36 m from the test building, while the last blast
NG-based explosives together with electronic round had a minimum distance of 9 m.
detonators were used. The distance between the
rows (burden) was about 2.0 m, and the spacing 5 TEST RESULTS
between each hole was about 2.5 m. The
maximum borehole depth in each round was Table 1 and Table 2 show measured PPV, peak air
between 4.5 – 7 m. A 1.0 - 1.5 m crushed stone blast pressure and peak strain during the first and
stemming was used in all holes. second blast tests.
In the first blast test, five blasts rounds were 5.1 Vibration frequency
fired consisting of all together 143 charged holes.
The number of holes detonated in one blast round There are several methods to determine the
varied from two single holes up to 53 holes. The frequency content of a signal. In this study we

Table 1. Measured maximum PPV, strain and air pressure on wall from blast test 1.

Blast Leca building Concrete building

round Scaled PPV Strain Air Scaled PPV Strain
distance (mm/s) (strain) Pressure distance (mm/s) (strain)
(m/kg0.5) (Pa) (m/kg0.5)
1 11.5 32 75 234 11.2 22 17
2 14.5 52 72 0.8 12.9 30 15
3 5.4 89 159 339 5.7 53 24
4 3.2 133 334 425 3.3 101 40
5 1 >260 733 750 1.1 >260 >1750

- 102 -
 Table 2. Measured maximum PPV, strain and air cycle, as many instruments provide as the only
pressure on wall from blast test 2. output.

Blast Scaled PPV Strain Air

round distance (mm/s) (strain) Pressure
(m/kg0.5) (Pa)
1 14.4 23 - -
2 8.8 23 34 -
3 4.5 51 87 -
4 1.3 180 >3200 437

have determined the zero-crossing frequencies

around each peak in the time series. The zero-
crossing frequencies are determined by assuming
that the time between the zero-crossing before a
peak and the zero- crossing after a peak
corresponds to half a period of the dominant Figure 2b. Peak values vs frequency determined by
the zero-crossing method for sensor on rock vertical
frequency.
direction, b) Last blast round in 2020 with high
blast load.

Table 3 shows the zero-crossing frequencies

around the maximum peak determined for the
sensors on rock for all blast rounds.

Table 3. Zero crossing frequency around maximum

cycle for measurements in vertical direction on
rock.

Blast Blast Hor Frequency

test round dist (Hz)
(m)
1-2018 1 35 100
Figure 2a. Peak values vs frequency determined
2.1 33 65
by the zero-crossing method for sensor on rock
vertical direction, a) First blast round 2018 with 2.2 30 81
low blast load. 3 23 26
4 18 41
Figure 2 shows show the zero-crossing 5 14 21
frequencies for the sensors on rock in the first blast 2-2020 1.1 52 64
round in 2018, with low blast load, and in the last 1.2 46 93
blast round in 2020, with high blast load. For the 1.3 40 72
low blast load, the difference in zero-crossing 2 37 213
frequencies between the vibration cycles is small. 3 22 144
However, for the high blast load the zero-crossing 4 12 163
frequencies differ considerably between the
vibration cycles. In these circumstances, the use of
a single frequency value gives a poor description The results in Table 3 indicates a reduction of
of the frequency content. This indicates that if the dominant frequency with decreasing distance.
frequency dependent limit values are used, such as This is not in accordance with the assumptions
in the British Standard (BS 7385-2:1993) and U.S. behind the distance factor in the Norwegian
Code of Federal Regulations, all frequencies with Standard and elsewhere reported in literature.
corresponding amplitudes from the blast need to However, the blast tests were designed to produce
be determined and compared to the limit value increasing vibration values, which was achieved
curve, and not just the frequency of the maximum by both increasing the charge weights and

- 103 -
reducing the distance for each blast round. To available, from well controlled, well documented
obtain a clearer picture of the effect of distance rock blast rounds.
and charge weight, frequency and distance can be The first blast tests produced vibration values
scaled with the charge weight as described in above PPV = 260 mm/s and the second blast tests
Ambraseys (1968) and Westine (1991), using the a maximum PPV = 180 mm/s. These results are
following equations: well above the vibration limit values calculated
according to today’s Norwegian Standard, which
(1) are 50 mm/s for the test buildings on rock in the
first blast test and 16-23 mm/s for the test building
on filling in the second blast test. The maximum
(2) measured strain levels were > 1750 strain in the
first blast test and > 3200 strain in the second
blast test. This is above critical strain levels
Where:
reported in earlier studies. Despite this, no visible
dsc = scaled distance (m/kg0.5)
damage could be detected either on the walls or on
fsc = scaled frequency (Hz·kg0.5)
tiles in any of the test buildings.
d = distance (m)
The zero-crossing method results reveal a large
f = frequency (Hz)
spread in the frequency between the different
Q = maximum charge/delay (kg)
vibration cycles. This points to the difficulty of
using frequency dependent vibration limit values.
In Figure 3 scaled zero-crossing frequency
The results of the two blast tests indicate that
around maximum peak is plotted against scaled
limit values according to the current Norwegian
distance for the sensors on rock for all blast
Standard includes a large safety margin for
rounds. A clear tendency towards reduced scaled
buildings on rock and well compacted stiff soil,
frequency with increased scaled distance can be
when considering damages to outer walls, which
seen. This agrees with what is reported in Liu
this study was designed to investigate.
(2019).
7 ACKNOWLEDGEMENTS
This study was performed with support from the
research project Remedy (Risk Reduction of
Groundwork Damage), funded by the Research
Council of Norway, Grant Agreement 267674.
The field test was financed by NPRA Directorate
of Public Roads (Statens vegvesen
Vegdirektoratet), National Railroad Authority
(Bane NOR), Norwegian Association of Heavy
Equipment Contractors (MEF), The Norwegian
Defence Estates Agency (Forsvarsbygg) and
Norwegian Contractors Association (EBA). Their
support is highly appreciated. The field tests were
conducted in cooperation with the upper secondary
Figure 3. Scaled zero-crossing frequency versus school Solør VGS. We would like to thank the
scaled distance for sensors on rock. responsible teacher, Øystein Johansen, and the
students for excellent cooperation.

6 CONCLUSIONS REFERENCES
Ambraseys, N. & Hendron, A. 1968. Dynamic
These field blast experiments have contributed to behaviour of rock masses. In K. Stagg, & O.
increased understanding of vibration generation Zienkiewicz, Rock mechanics in engineering
and propagation from bench blasting in rock, practice (pp. 203-236). London: John Wiley &
response of buildings to ground vibration and Sons.
vibration damage mechanisms for concrete and
light weight aggregate masonry buildings. Two BS 7385-2:1993 Evaluation and measurement for
extensive sets of high-quality synchronised vibration in buildings – Part 2: Guide to damage
vibration and strain measurement data are made levels from ground-borne vibration.

- 104 -
Liu, D., Lu, W., Liu, Y., Chen, M., Yan, P. & Sun P.
2019. Analysis of the main factors influencing
the dominant frequency of blast vibration.
Hindawi shock and vibration, Article ID
8480905.

NS 8141:2001 Vibration and shock. Measurement of

vibration velocity and calculation of guideline
limit values in order to avoid damage on
constructions. (In Norwegian).

U.S. Code of Federal Regulations (CFR) 30 §715.19,

§816.67 and §817.67.

Westine, P.S., Dodge, F.T. & Baker W.E. 1991.

Similarity methods in engineering dynamics –
theory and practice in scale modelling – revised
edition. Elsevir.

- 105 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Effects on buildings from blast induced vibrations –

a numerical pre-study

P.D. Folkow, M. Johansson, J. Leppänen, G. Edefors & A. Naseri

Chalmers University of Technology, Göteborg, Sweden

ABSTRACT: It is well known that blasting in rock may cause structural damage to nearby buildings.
There are differences among most countries regarding legislation with respect to blast induced vibrations.
Criticism has been raised against this unclear situation, and it is of importance to develop more reliable
predictions on ground and structural vibrations. The present work has its origin based on the features
within the Swedish Standard. These guidelines are related to distance and underground conditions,
contrary to many other national standards which are based on frequency measures. There is a desire to
update the Swedish Standard; either based on distance, frequency or some other measures. This pre-study
considers numerical studies (2D FEM) for modelling blasting effects. The aim is to illustrate how various
parameters (material, geometry, load) influence ground vibration phenomena as well as structural effects
above-ground.

1 BACKGROUND There is a desire in the industry to modernise the

standard by better understanding the theoretical
Blasting causes ground vibrations that may cause background of how different vibration parameters
damages to nearby buildings and other structures. affect the damage in a building. It is considered
The Swedish Standard – Vibration and Shock (SIS poorly known how and when damage occurs, and
2011) used for ground borne vibrations states more detailed studies based on numerical
guidelines which are based on empirical data from modeling have been particularly highlighted
the 1960s and depends on parameters such as (Dowding 2000, Norén-Cosgriff et al. 2020).
distance and ground conditions. The scientific The aim of this work was to study how
support for the guidelines may be considered different phenomena affect the ground vibrations
weak, and criticism has been raised of its validity. and the risk of damage to buildings. Two different
In the Swedish standard, the permitted vibrations models were used, one for the load response and
are limited only to the vertical velocity component one for the structural response.
of the ground surface. Further, in most other In the first model, blast induced vibrations at
countries guidelines are used that are conceptually the ground surface were studied without any
different than in Sweden, and these guidelines are presence of buildings. The following parameters
based on frequency measures from the blast were investigated:
induced vibration (Gjödvad & Jern 2019, NGI − Blasting parameters (distance to charge,
Remedy 2019). The various national guidelines angle of incidence)
have their shortcomings and are mainly based on − Ground conditions (stiffness, density,
rules of thumb with a lack of scientific basis. geometry)

- 106 -
 In the second model, the structural response of where ρ is the density, E is the modulus of
a concrete wall with openings was studied using a elasticity and ν is the Poisson’s ratio. The ground
prescribed vertical movement of the ground wave speed c does not depend on the frequency
surface below the wall sole. The wall geometry content of the wave, and the relationship between
was fixed but the ground stiffness was varied to the wavelength λ and the frequency f can be
simulate various ground conditions. Further, the determined as:
prescribed vertical vibrations were applied in the
ground using combinations of various load c (3)
λ=
frequency and ground wave speed. f
The work presented here is a summary of a
numerical pre study (Folkow et al. 2021), and a At the ground surface, in addition to wave
series of subsequent works are planned to obtain a transformation at interfaces, various forms of
deeper physical understanding on the effect on surface waves will be created, and the most
buildings from blast induced vibrations. The long- dominant surface wave is the Rayleigh wave
term aim is to contribute knowledge to revise the (R-wave), which propagates slower than the
current Swedish standard. S-wave, and its speed is commonly approximated
as:
2 WAVE PROPAGATION PHENOMENA
0.862 + 1.14 ⋅ν (4)
= cR ⋅ cS
In this study, blast induced vibrations were studied 1 +ν
by numerical analyses using two dimensional (2D)
linear finite element (FE) analyses, both in time However, the R-wave has much lower damping
and frequency domain. First, the blast induced than both the P- and S-waves, and in addition to
vibrations, both in horizontal and vertical the vertical component in the displacement field, a
directions, were studied by varying the load significant component in the horizontal direction is
(amplitude, frequency content, duration, distance created, which is approximately 50% of the
to charge and angle of incidence). Furthermore, vertical component.
the wave propagation phenomena (wave speed,
damping, frequency content) were studied and 3 LOAD RESPONSE
interaction between the ground materials rock and
clay, by varying the layer thickness of the clay 3.1 Geometry, material and load
above rock, and the boundary effects at the ground
surface (reflection, change of wave type, Numerical calculations have been performed
displacement field). Thereafter, the structural using the general finite element programme
response was studied by using a prescribed COMSOL Multiphysics® (COMSOL 2019)
vibration with a given frequency. using two-dimensional (2D) plain strain models
In the model used for studying the load assuming linear elastic response. The use
response, the effects in the near field from the of 2D models for these studies is a simplification
blast were not included. Only the far field of reality, and in a conceptual pre-study as
wave propagation was studied by assuming a this, it is deemed to be good balance between
symmetrical pressure wave, which propagated in a accuracy and ‘cost’ regarding calculation time.
radial direction from the blast, resulting in a linear The ground vibration obtained are a direct
wave propagation with a pressure wave (P-wave) effect of the applied load and the purpose
propagating in radial direction, which may of the analyses performed has never been to
transform to a Shear wave (S-wave), for instance fully seek to imitate an actual blasting sequence
at the interface between the rock and the clay. The in rock. Hence, it is also not of interest to focus
P-wave propagates faster than the S-wave, and in on individual vibration values in the presented
an isotropic material the wave propagation results; instead, focus should be on how the
velocities can be determined as: resulting vibrations conceptually vary with
changed parameters.
E (1 −ν ) The geometry, boundary conditions and
cP = (1) application of blasting are shown in Figure 1.
ρ (1 +ν )(1 − 2ν )
The material parameters for rock and clay are
listed in Table 1 and the blast load properties are
E
cS = (2) shown in Figure 2.
ρ ⋅ 2 (1 +ν )

- 107 -
 Result point Free boundary (ground surface)

Clay

Rock Charge

P(t)

Charge

x
Low-reflecting boundaries (ground material)

Figure 2b. Blast load properties: (b) frequency

Figure 1. Geometry and boundary conditions of FE content.
model. The blast load was applied as a pressure P(t)
located inside a circle (diameter 32 mm) according
to the enhanced picture.

Figure 3a. Schematic illustration of (a) geometric

parameters varied in case studies.

Figure 2a. Blast load properties: (a) pressure-time

relation P(t).

Table 1. Material properties for rock and clay, wave

speeds are rounded values based on properties used.

Type E ν ρ cP cS cR
[GPa] [-] [kg/m3] [m/s] [m/s] [m/s]
Rock 60 0.25 2,600 5,300 3,000 2,800
Clay 0.050 0.495 1,700 1,000 100 95
Figure 3b. Schematic illustration of (b) appearance
of P- and R-waves.

Several case studies were made varying the

geometric parameters schematically shown in 3.2 Results
Figure 3a:
In the case studies, the resulting vibrations in the
Distance to charge: r = [30, 50, 100] m horizontal (x) and vertical (y) directions were
Angle of incidence: α = [5, 37, 60, 90] determined. Special focus was put on the
Thickness of clay layer: hclay = [0, 2.5, 5, 10] m appearance of P- and R-waves, as schematically

- 108 -
illustrated in Figure 3b. Here, the P-wave
propagates from the centre of explosion at a speed
higher than that of the R-wave. Therefore, the
P-wave, regardless of charging location, always
reach the result point first. However, when the
P-wave reaches the free ground surface, S-waves
are also generated as a boundary effect and for a
charge located near the surface, interaction
between P- and S-waves will give rise to R-waves
that propagate along the surface.
Figure 4 shows examples of the vibration vx(t)
and vy(t) in the horizontal and vertical direction,
respectively, at various angles of incidence α; here
for a distance r = 50 m. The peak obtained at time
t ≈ 0.009 s is caused by the P-wave when it Figure 4b. Examples of vibration-time relationships
in (b) vertical direction vy(t) for distance r = 50 m
reaches the result point. However, for α = 5, a
and varying angle of incidence α. Theoretical
secondary peak is also obtained at t ≈ 0.018 s, i.e. arrival times for P- and R-waves are approximately
a time which coincides well with the theoretical 9.4 ms and 17.9 ms, respectively.
value for the arrival of an R-wave and hence it is
also deemed to be caused by such a wave.
However, no corresponding peak for the R-wave
can be observed when α = 37-90.

Figure 5a. Maximum velocity amplitude in (a)

horizontal (x), direction when hclay = 0 m. Solid
and dashed lines correspond to P- and R-wave,
respectively.
Figure 4a. Examples of vibration-time relationships
in (a) horizontal direction vx(t) for distance r = 50 m
and varying angle of incidence α. Theoretical
arrival times for P- and R-waves are approximately
9.4 ms and 17.9 ms, respectively.

In Figure 5 the maximum velocity amplitude in

the horizontal (x) and vertical (y) directions are
compared for various distances r and angle of
incidence α when hclay = 0 m. The results are
distinguished of whether being caused by a P- or
R-wave, where the latter is assumed to cause the
vibrations at times larger than the theoretical
arrival time of an R-wave. From this comparison it
can be noted that an increased angle of incidence
means that horizontal vibrations vx decrease while Figure 5b. Maximum velocity amplitude in (b)
vertical vibrations vy increase. Further, for the vertical (y) directions when hclay = 0 m. Solid and
horizontal vibrations it can be noted that similar dashed lines correspond to P- and R-wave,
respectively.
values are obtained for each distance r when α =

- 109 -
5-37, and for the vertical vibrations the same
observation can be made when α= 60-90. It is also
worth noting that the horizontal and vertical
vibrations are of similar magnitudes when α =
37-60, i.e. there must be a rather flack or steep
angle of incidence to obtain a significant
difference in amplitude of horizontal and vertical
vibrations.

Figure 6c. Maximum velocities in vertical (y)

direction when hclay = 0 – 10 m: (c) r = 100 m.

4 STRUCTURAL RESPONSE

4.1 Geometry, material and load

To study the influence of vertical vibrations in the

ground on the structural response, a simplified
Figure 6a. Maximum velocities in vertical (y)
wall structure with openings was simulated using
direction when hclay = 0 – 10 m: (a) r = 30 m. the general FE programme Abaqus/CAE (Dassault
2019). The ground vibration was simulated using a
vertical movement

v y ( t=
) vA ⋅ sin ω t (5)

which came in from the left side of a wall, moving

in the horizontal direction with a speed according
to the ground wave speed c, see Figure 7a for a
schematic illustration. The applied vibration was
given a duration

tvibr = 3 ⋅ T (6)

where T is the eigenperiod of the vibration

determined as
Figure 6b. Maximum velocities in vertical (y)
direction when hclay = 0 – 10 m: (b) r = 50 m.
1 2π (7)
T= =
f ω
Figure 6 compares the maximum velocities in
the vertical (y) direction for various combinations The time it takes the vibration to pass the building
of distance r and angle of incidence α when hclay = is
0 – 10 m. From this it can be observed that a large
depth of clay (hclay = 10 m), in most cases, leads to lwall
twall = (8)
reduced vibrations compared to a case with no c
clay (hclay = 0 m); i.e. the clay has a dampening
effect on the vibrations. However, for a small which means that, depending on the combination
depth of clay (hclay = 2.5 m) the vibrations of wall length lwall, wave speed c and load
increase; i.e. a thin layer of clay may cause frequency f, it is possible that the vibration load is
amplification effects through interference in the terminated at the front end of the wall before any
rock/clay layer and the free ground surface. vibration load has even reached its far end.

- 110 -
 Figure 7b shows the geometry of the wall with Load parameters:
openings that have been studied in this report. The vA = 100 mm/s
geometry was chosen to be the same as used f = [25, 50, 100, 200] Hz
in blasting tests carried out in Norway c = [500, 1000, 2000, 4000] m/s
(Norén-Cosgriff et al. 2020). In these tests, the
structure was founded on an approximately 0.5 m A ground stiffness of k = 0.04 GN/m2 was
levelled and compacted layer of gravel, over rock. estimated based on the conditions used in the tests
The simulations were made using linear elastic, made by Norén-Cosgriff et al. (2020). However,
first order 2D plane stress elements (square shape, an increased stiffness (k = 0.16 GN/m2) was
side 0.1 m) with reduced integration. To model the needed to reproduce the lowest eigenmode of the
ground, linear 2-node springs were used which wall observed in the tests, see Section 4.2. The
were attributed a constant stiffness to simulate higher ground stiffnesses, k = [1, 10] GN/m2, were
various ground conditions. The top nodes of the then included as part of the parametric study.
springs were fixed to the bottom of the wall sole The purpose in this study was never to study
while the bottom nodes were velocity controlled the wall response values in detail but rather to
using vy(t) according to equation (5). observe trends and principles. Therefore,
simplified load values were used, e.g., an
amplitude vA = 100 mm/s and to restrict vy(t) to
just include a single frequency. Further, the ground
wave speed was chosen to approximately represent
that of both P- and R-wave for various type of
ground conditions.

4.2 Results
Figure 7a. Concept for simulation. Figure 8 shows the first four eigenmodes and
corresponding eigenfrequencies for the wall
Material parameters in wall and spring structure for all cases studied.
stiffness: The first eigenfrequency for the wall in
Norén-Cosgriff et al. (2020) was observed to be f1
Ec = 30 GPa, ν = 0.2, ρc = 2500 kg/m3, mroof’ = ≈ 26 Hz; i.e. approximately corresponding to a
240 kg/m (extra mass on top of wall). ground stiffness of k = 0.16 GN/m2 in the
k = [0.04, 0.16, 1, 10] GN/m2 analyses, compare with Figure 8b in which f1 ≈

mroof’

0.4
wgable

1.0
0.2

wwall 1.0

wsole
0.2
1.5 0.7 1.1 0.7 1.0
[m]
Figure 7b. Geometry of wall structure with widths wwall = 0.2 m, wsole = 0.4 m, wgable = 1.0 m.

- 111 -
 (a)

f1 = 14 Hz f2 = 23 Hz f3 = 62 Hz f4 = 109 Hz

(b)

f1 = 25 Hz f2 = 46 Hz f3 = 67 Hz f4 = 113 Hz

(c)

f1 = 41 Hz f2 = 97 Hz f3 = 104 Hz f4 = 141 Hz

(d)

f1 = 53 Hz f2 = 144 Hz f3 = 175 Hz f4 = 222 Hz

Figure 8. First four eigenmodes and corresponding eigenfrequencies in the wall structure for various ground
stiffness: (a) k = 0.04 GN/m2, (b) k = 0.16 GN/m2, (b) k = 1 GN/m2, (d) k = 10 GN/m2.

25 Hz. When k = [0.04, 0.16] GN/m2 the first two stresses from those analyses are summarized in
eigenmodes are dominated by the low ground Table 2 and from this it can be concluded that
stiffness and approximately correspond to rigid there is no clear trend of which load frequency or
body motions of the wall structure. However, for ground wave speed that produce the most critical
all other combinations the eigenmodes are clearly tensile stresses in the wall structure. The structural
affected by the wall stiffness. response of the wall is also affected by its
Figure 9 shows some examples of the eigenfrequencies but comparing the frequencies of
maximum tensile stresses obtained in the wall the first eigenmodes shown in Figure 8 with the
structure due to various combinations of vibration critical load frequencies in Table 2 does not
load and ground stiffness and from this it is explain the combinations of f, c and k found
indicated that the critical stress stage may be critical.
obtained at various load frequencies, depending on To further investigate this, the maximum
e.g., the ground stiffness. Further analyses were tensile stresses obtained in the corner regions of
made in which the wave speed c and ground the wall’s two openings have been studied, see
stiffness k were varied for load frequencies f = [25, Figure 10 for location and denotation. Based on
50, 100, 200] Hz. The resulting maximum tensile the plots in Figure 9, this is where the maximum

- 112 -
 (a)

f = 25 Hz f = 50 Hz

(b)

f = 25 Hz f = 50 Hz
Figure 9. Maximum tensile stresses in wall structure subjected to a vibration load with f = [25, 50] Hz and c =
2000 m/s when (a) k = 0.04GN/m2 and (b) k = 0.16 GN/m2. Note that different scales are used.

Table 2. Maximum tensile stress in the wall structure for various combinations of load frequency f, wave speed c
and ground stiffness k.

k [GN/m2] 0.04 0.16 1 10

c [m/s] 500 2000 500 2000 500 2000 500 2000
f σmax σmax σmax σmax σmax σmax σmax σmax
[Hz] [MPa] [MPa] [MPa] [MPa] [MPa] [MPa] [MPa] [MPa]
25 - 4.2 11 3 3 1 2.5 0.5
50 - 0.7 5 7 32 12 5.5 2
100 - 0.6 6 3 10 11 10 10
200 - 0.1 < 0.5 0.5 2 4.5 8.5 15

tensile stresses appear and in Figure 11 it is shown

how the maximum tensile stresses varies in the 1 2 5 6

eight corner regions for various f and c when k =

[1, 10] GN/m2. Here, it can be seen that the
maximum tensile stress in the corner regions does
not always appear for the same combination of f 8 7
and c; when e.g., c = 2000 m/s and k = 1 GN/m2,
the maximum tensile stress is obtained at f = 50 Hz 4 3
in three corner regions and at f = 100 Hz in five
corner regions. In Figure 8 the first four mode Figure 10. Regions in which maximum tensile stress
shapes and eigenfrequencies for this case is in the corners of wall openings were determined.
shown: fi = [41, 97, 104, 141] Hz; i.e. two
eigenfrequencies close to 100 Hz. Hence, these Table 3 shows a summary of all the analyses of
frequencies indicate that a load frequency of 100 the wall structure in which it is shown (marked
Hz should be critical for the wall structure, but as with ●) at what load frequency f the maximum
can be seen in Figure 11, this is not necessarily the tensile stress is obtained for a given combination
case. of wave speed c and ground stiffness k. Further, in

- 113 -
 (a)

c = 500 m/s c = 2000 m/s

(b)

c = 500 m/s c = 2000 m/s

Figure 11. Maximum tensile stresses in corner 1 to 8, according to Figure 10, when the wall structure was
subjected to load frequency f = [25, 50, 100, 200] Hz at wave speed c = [500, 2000] m/s when (a) k = 1
GN/m2 and (b) k = 10 GN/m2.

Table 3. Summary of for what combinations of load frequency f, wave speed c and ground stiffness k that
maximum tensile stresses were obtained in the wall structure.

k = 0.04 GN/m2 k = 0.16 GN/m2 k = 1 GN/m2 k = 10 GN/m2

f c [m/s] c [m/s] c [m/s] c [m/s]
[Hz] 500 2000 500 2000 500 2000 500 2000
25 - ● ●
50 - ● ● ●
100 - ○ ● ○
200 - ●

● = Maximum stress in wall.

○ = Maximum stress in at least one corner region according to Figure 10.
- = Not studied.

those cases when at least one corner region clear that there is no clear trend of what load
reaches its maximum tensile stress at another load combination leads to the highest tensile stresses;
frequency than this, this is shown as well (marked the results obtained indicate that high stresses
with ○). Based on this compilation it can be may be obtained for both low and high load
concluded that it is difficult to predict in advance frequencies and for low and high wave speeds.
what combination f and c will result in the most The trend observed is that the critical load
critical case for a given structure. Further, based frequency f increases with an increased ground
on the maximum stresses listed in Table 2 it is also stiffness k.

- 114 -
5 CONCLUSIONS AND FUTURE WORK 6 ACKNOWLEDGEMENTS

This pre-study has focused on studying wave The research presented in this paper was
propagation and vibration effects from blasting financed by Swedish Transport Administration
from both the load response and structural (Trafikverket) and Rock Engineering Research
response perspectives. Below, the most important Foundation (BeFo).
observations are listed:
REFERENCES
− Load response (ground vibrations generated
by blasting): COMSOL 2019. COMSOL Multiphysics® v. 5.5,
COMSOL AB, Stockholm, Sweden, 2019.
− Ground vibrations at the surface have in
many cases a significant horizontal Dassault 2019. Abaqus/CAE, Dassault Systèmes
component, in addition to the vertical Simulia Corp, US, 2019.
component that is considered in the Swedish
Standard. Dowding, C.H. 2000. Construction vibrations,
− The ground vibrations are largely affected by Prentice Hall, Englewood Cliffs. NJ, 2000.
the position of the explosive charge (distance
and angle of incidence. Folkow, P., Johansson, M., Leppänen, J. 2021.
− With looser foundations (e.g., clay on rock), Skadeverkan mot byggnad av markvibrationer
the thickness of the clay layer may have a från sprängning (Structural damage on buildings
considerable effect on the ground vibrations due to blast induced ground vibrations. In
and depend on e.g., its resonance effects. Swedish), Chalmers University of Technology,
Report ACE 2021:1, 2021., Göteborg, Sweden.
Structural response (effects of ground vibration
Gjödvad, J.F. & Jern, M. 2019. Vibration monitoring
on a building):
standards connected to the use of explosives in
Europe, EFEE 10th world conference on
− Based on studies of a simplified wall explosives and blasting, Helsinki, Finland, 2019.
structure, it has been noted that it is complex
to predict how the magnitude of the NGI Remedy 2019. Vibration induced damage due to
resulting stresses is affected by different construction work – State of the art report, WP4.
parameters. Vibrations induced by construction activity”
− It is not obvious in advance that a given Deliverable 4.1, rev 0, NGI Remedy Project,
combination of load frequency f and ground Norway, 2019.
wave speed c will result in large tensile
stresses. Norén-Cosgriff, K., Ramstad, N., Neby, A. &
− Varying combinations of f and c can give Madshus, C. 2020. Building damage due
significant differences in stress, both up and vibration from rock blasting, Soil Dynamics and
down, without observing any clear context in Earthquake Engineering, 138, 2020.
the effect of various parameters. Resonance
SIS 2011. Vibration and shock - Guidance levels for
effects can only partially explain the observed blasting-induced vibration in buildings, Swedish
scattering in results. standard SS 460 48 66, Stockholm, Sweden,
2011.
Future works:

− Continued studies of how the ground

vibrations are affected by various ground
conditions and the location of the charge.
− Parametric studies of wall structures with
various geometry and ground conditions
(including less stiff ground).
− Comparison of numerical simulations of wall
structure with experimental results obtained
from full-scale tests in Norway. Test data are
available for various kind of wall structures
and ground conditions.

- 115 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Elevated pipeline response to shaft blasting

C.T. Aimone-Martin
Aimone-Martin Associates, LLC

J. Redyke
Dykon Blasting Corp.

B.M. Meins & O.C. Meins

Detecht, LLC

ABSTRACT: Two 9.1 m diameter shafts were excavated by blasting from elevations -24.7 to 38.4 m at
nominal distance of 12 m from a 1.6 m diameter welded steel water pipeline elevated on concrete piers.
Shafts will convey water through a new 1.83 m pipe via microtunnels crossing beneath the Canadian
River in Oklahoma. The 60-year-old pipeline was instrumented at the support pier and at the unsupported
midspan with circumferential and longitudinal strain gauges to compare with global shear and elongation
strains calculated from velocity measurements. Maximum midspan lateral and vertical displacements
were 0.67 and 0.55 mm producing computed global shear strains of 30.8 and 35.3 µε, respectively. The
highest hoop and elongation strain measurements at the midspan pipe crown were 14.1 and 6.3 µε when
the peak ground vibration was 17 mm/s. Blast-induced pipeline deflections and resulting strains were
considered to be highly safe and protective of the aged pipeline.

1 BACKGROUND plate through hard rock that required blasting.

Additional blasting was needed to complete two
Microtunneling beneath the Canadian River launch portals for the MTBM. The construction
Crossing was required to replace an existing site was within the adjacent river floodplain,
elevated water supply pipeline, constructed in requiring several meters of cobblestone fill to
1962, and add a second line to increase water serve as a working platform.
supply into Oklahoma City in the U.S. from Figure 2 shows details of the pipeline
outlying reservoirs. The existing 762 mm-diameter supporting elements, concrete pier bases, pinned
welded ductile iron pipeline is supported on connections to concrete piers and upper rings. The
concrete piles adjacent to the river crossing. pipeline properties are given in Table 1. As built
Access for the two replacement pipelines and drawings did not identify the pipeline grade and it
micro tunnel boring machine (MTBM) was assumed that the material is ductile iron with
construction required excavation of two, 10.4 m properties similar to a grade C steel available at
diameter shafts in close proximity to the pipeline, the time of construction. Concrete pier P13 height
shown in Figure 1, to a final depth of 38.1 m. The above the ground surface is 2 m and the total
upper 21.3 m comprised a secant pile liner within height to the pipe top is 6.7 m. Blast and
unconsolidated materials and lower 16.8 m a liner monitoring plans were submitted for review by the

- 116 -
 Figure 1. Arial view of shafts on either side of the elevated pipeline after completion of secant piles (left) and
distances from shaft perimeters to piers P12 and P13.

owner’s consultants. Plans included provisions to accomplished by controlled blasting methods and
limit and monitor differential displacements and appropriate time delays.
strains at critical locations within the supporting
structure. Of greatest importance were vertical and Table 1. Ductile iron pipeline properties.
lateral deflections at the unsupported pipe midspan PIPE PARAMETERS
and controlling amplification of structure motions Welded steel Grade C
relative to pier base excitations. Concerns arose Pipe OD (m) 1.6
during plan reviews that comprehensive soil Wall thickness (mm) 25.4
structure interaction (SSI) dynamic modelling of Unsupported length between piers (m) 24.84
the piers supports was not performed. The Total load over unsupported length
72.5
approach taken by the blast and monitoring team (pipe+water) (Mg)
was to limit dynamic (blasting) midspan Young's modulus (GPa) 200
deflections and resulting axial and hoop strains in Tensile strength (MPa) 413.7
accordance with AWWA M11 (2017) and meet Minimum yield strength (MPa) 206.8
simply-supported steel pipeline guidelines by Design Factor (%) 50
Bonds (2017). In the simplest form, the Allowable Pipe wall hoop stress (design
recommended allowable midspan deflection limit 103.4
strength) (MPa)
for pipes can be taken as 5% of the outside Allowable strain (µε ) 517.0
diameter or this case 0.16 m. This was

Figure 2. Concrete pier P13 supporting steel pipe bents with rocker and sliding connections and single upper
support ring (left and middle) and deploying strain gauges from elevated cage at unsupported midspan (right);
A through C denote locations of instrumentation.

- 117 -
2 BLAST DESIGN strain to compare with measured strains.
Wire strain gauges were affixed to the pipeline
A total of eight blasts were conducted by Dykon above P13 and at the midspan at the springline and
Blasting between December 2020 and July 2021. crown to record longitudinal and hoop strains in
Holes drilled 76.2 mm in diameter were loaded the pipeline wall. Strain gauges, manufactured by
with 50.8 mm by 406 mm cartridge emulsion with HBM, were series Y 90‐degree ‘T’ rosettes with a
a cartridge density of 1.23 g/cc, 0.34 and 0.454 kg 6.0 mm measurement grid length, model
PETN boosters, det cord for tunnel round pre-split K‐CXY3‐0060‐1‐350‐4‐050‐N and attached to the
holes and electronic detonators. Charge weights pipeline with HBM Z70 adhesive. Strain
per delay ranged 2.86 to 10.5 kg. Sinking shots measurements were recorded using an EDAQ Lite
with a centre pull utilised 8 to 22 ms delays manufactured by HBM with an 8‐channel
outward with perimeter holes fired 25 ms. Drill Wheatstone bridge layer. A sample rate of 5 K sec
hole spacings were 1.22 m and depths ranged 2.4 -1 was used. The EDAQ was powered by one 12-V

to 3.7 m with powder factors of 0.9 to 1.5 kg/m3. battery connected to a 20 W solar panel. Both the
Tunnel rounds were designed with a burn cut, 0.9 EDAQ and seismographs were connected to
m hole spacings, drill depths of 2.7 to 3.2 m and modems to allow remote downloading of data via
powder factors ranging 2.93 to 3.45 kg/m3. telemetry as needed.
Strain gauge data were filtered and time
3 VIBRATION AND STRAIN MONITORING histories analysed for crown and springline peak
hoop and longitudinal strains representing strain at
Geophones were deployed to measure ground, localised sites of mounting. Pipe-mounted velocity
pipeline, and pipeline supporting structure records were integrated to calculate displacement
vibrations to calculate global strains between time histories to determine peak global axial
locations identified in Figure 2 as the pipe top or strains between the connection at the pier top and
crown (B) and midspan (C). Strain gauges, used to midspan and bending strains at the midspan.
directly measure pipeline strains, were affixed to This project provided the opportunity to
the pipeline at the pier top (B) and midspan (C) compare midspan global elongation strains
shown in Figure 3 at both the pipe crown and computed using velocity sensors with strain gauge
springline. Geophones were affixed to structures measurements for an elevated pipeline. To the
using hot glue and strapping around the pipe and authors’ knowledge, this has never been
the in-ground geophone was buried 0.15 m at performed. For this comparison, a total axial strain
location (A). was computed with contributions from both pure
All recording systems using sample rates of axial elongation, derived from the geophone radial
either 16,384 or 4096 sec-1, were co-triggered with (R) component parallel with the pipe, and bending
the master seismograph at (A). In this manner, the in each the vertical and lateral directions.
vibratory responses of the pier and pipeline were Perpendicular vertical motions (V component)
recorded simultaneously to enable analysis of contributed to bending strains at the crown while
differential structure motions and calculate global perpendicular horizontal motions (T component)

Figure 3. Axial view of pipe showing locations of strain gauges at the crown (top) and springline (side) (left)
and springline gauge at the midspan (right).

- 118 -
contributed to bending strains at the springline. It only 0.279 and 0.262 mm in the vertical and
was not possible to compute hoop strains from transverse directions.
geophone measurements.
4.2 Pipeline vibration response - natural
4 RESULTS frequency and damping
4.1 Ground motions Figure 4 shows typical component velocity time
histories measured at the midspan crown for blast
Table 2. Ground vibrations in terms of velocity (V) 7 and the near harmonic blast response in the V
and displacements (D) measured at the base of P13. and T components. The fundamental frequencies
in the T component were a consistent 3.8 Hz with
RADIAL VERTICAL TRANSVERSE
Blast damped oscillations lasting up to 20 sec. The V
V D V D V D
component was initially 7.3 Hz reducing to 5.2 Hz
(mm/s) (mm) (mm/s) (mm) (mm/s) (mm) lasting well over 20 sec. The percentage of
1 6.1 0.036 12.8 0.148 7.1 0.091
2 16.3 0.229 34.5 0.279 29.2 0.262
critical damping for the T component (β) was
3 10.9 0.141 17.0 0.072 14.2 0.147 approximated for the T and R components
4 9.7 0.053 17.7 0.153 7.0 0.104 computed using the following relationship:
5 10.8 0.077 18.1 0.142 14.8 0.126
6 12.4 0.168 13.7 0.122 18.7 0.170
7 6.8 0.049 10.6 0.163 6.0 0.092 (1)
8 3.8 0.019 5.6 0.040 4.9 0.027
where P2 and P1 are successive peaks. In this case,
damping in the T component is 2.8% and is typical
A summary of ground motions in terms of velocity of the range of damping for all blasts.
and displacements are given in Table 2.
Component ground velocities at the closest pier 4.3 Strain Measurements
support base ranged 5 to 34.5 mm/s. Peak
frequencies for all components averaged 53 Hz. Table 3 presents a summary of pipe strain
As such, the highest ground displacements were measurements at the crown and springline atop of
displacement (mm)
Peak component

P1 P2

Time (sec)

Figure 4. Blast 7 component V and T displacement time histories (top) and Fourier analysis plots of dominant
component frequencies (below).

- 119 -
 Table 3. Maximum hoop and longitudinal pipe strains at the crown and springline of the midspan and at
the pier top for eight blasts.

MIDSPAN Measured strain (µε) PIER Measured strain (µε)

Blast Hoop (ε L) Longitudinal (ε H) Hoop (ε L) Longitudinal (ε H)
crown springline crown springline crown springline crown springline
1 2.69 6.72 7.50 3.22 2.82 2.07 4.56 1.53
2 6.93 13.22 19.9 10.0 8.49 4.15 10.20 5.18
3 8.72 nu 20.2 20.0 14.14 6.30 12.91 4.56
4 7.97 4.28 14.0 6.37 8.52 2.96 7.99 3.11
5 3.34 nu 4.22 5.19 3.67 2.54 3.69 2.44
6 3.37 nu 5.03 5.26 3.51 2.21 4.57 3.44
7 4.22 5.01 8.89 4.60 3.94 2.28 6.26 2.92
8 2.12 nu 2.23 nu nm
nu - not usuable; strain data within the noise
nm - not measured
the pier and at the midspan. Typical strain time hoop and elongation stresses generated during
histories are provided in Figure 5 for blast 3 over blasting based on measured strains at the crown.
the initial 0.5 sec recording time. The pipe crown Dynamic elongation limits are set by the minimum
longitudinal peaks at the pier support shows yield strength for grade C steel shown in Table 1
uniform 166 Hz oscillations while the hoop crown as 206.8 MPa while allowable hoop stress is set by
frequencies are less uniform, averaging of 132 Hz. a design factor applied to yield. For the aged pipe,
Time histories clearly show an underlying low the monitoring team applied a 50% reduction
frequency component that persists in the pipe in factor or 103.4 MPa. The pipe working pressure of
the vertical and lateral components in some cases 1.3 MPa in the absence of possible surge pressures
well over 20 sec. Interestingly, these pipe were low and not included in the stress
displacement oscillations are perfectly matched by calculations.
strain gauge records as demonstrated in Figure 6 at Biaxial stress-strain equations were used to
the crown and springline during blast 7. The strain compute longitudinal (axial) and hoop stress based
gauge and displacement time histories are on measured strains are:
compared over a 10-sec window and show the
same overall response characteristics. (2)

4.4 Blast-inducted pipeline wall stresses

(3)
Pipeline integrity engineers are responsible for the
performance of pipelines carrying liquids and
gases ensuring that pipeline operations meet Where σL and σH are longitudinal and hoop stress,
standards and requirements for safety and υ is Poisson’s ratio of 0.3, and E is the modulus of
regulations. It is often the case that pipeline elasticity (200 GPa).
engineers are not acquainted with the safety Blast 3 generated the highest strains. Midspan
precautions exercised by blasters and monitoring hoop and longitudinal stresses were 3.3 and 5.0
experts to ensure pipelines can be protected to a MPa resulting in FOS values of 31 (103.4/3.3) and
high safety factor when subjected to transient blast 41 (206.8/5.0), respectively. At the pier top, hoop
excitations at distances beyond permanent rock and longitudinal stresses were less than 4.0 and 3.8
displacement limits of the blast. MPa resulting is FOS greater than 26 and 54.
To illustrate the degree of safety to which shaft Clearly, shaft blasting generating very low strain
blasting was performed, factors of safety (FOS) amplitudes was highly protective of the elevated
were computed for the pipeline by considering pipeline.

- 120 -
 Figure 5. Pipeline crown hoop and longitudinal strain time histories recorded during blast 3 at the midspan and
at the pipe overlying pier P13 showing 10 Hz and 7.3 Hz underlying frequencies at the pier top and midspan,
respectively.

Table 4. Global strains compared with measured strains (all units are µε).

Computed Strains
Peak Time-correlated Measured axial strains
Blast Shear Peak Bending
Pure Axial Computed total axial strain
vertical (V) horizontal (T) radial (R) crown springline crown springline crown springline
1 34.0 19.2 8.5 3.3 1.9 11.0 8.8 7.5 3.2
2 not measured 19.9 10.0
3 35.3 46.9 16.1 1.3 1.8 17.0 16.4 20.2 20.0
4 35.3 36.6 8.7 3.4 3.5 11.1 10.1 14.0 6.4
5 12.6 23.0 3.1 1.2 2.2 4.1 4.2 4.2 5.2
6 12.9 26.6 4.1 1.3 2.6 4.4 5.3 5.0 5.3
7 44.2 30.8 9.9 4.3 3.0 13.0 10.9 8.9 4.6

- 121 -
 V displacement

Crown longitudinal strain

T displacement

Springline longitudinal strain

Figure 6. Blast 7 time history comparisons of integrated velocity measurements with strain measurements
at the midspan.

4.5 Computed axial global strains compared applied at the crown and springline, respectively,
with measured strains D is pipe outside radius, and L is the full length of
the unsupported pipe (Dowding 1985). Total axial
There is great interest in using velocity transducers strain is the maximum of time-correlated addition
to globally estimate blast-induced strains in of both the pure axial and the bending. Note the
pipelines based on the ease of field deployment total axial is not the sum of the peak axial and
and reliability of long-term measurements in peak bending as constructive and destructive strain
comparison to the difficulties strain gauges can addition occurred at different points in strain-time
pose. In this regard, global axial strains computed history.
from time correlated differential displacements
between the pier top and midspan measured using
velocity sensors at the pipe crown were compared
with measured axial elongation strains at the
crown and springline to establish correlations.
These data are summarized in Table 4.
Shear strain in the vertical, transverse
(perpendicular to the pipe axis), and pure axial
(elongation) strains in the radial are simply
∆δi/(L/2), where L/2 is the half span length
between supports and ∆δi the maximum
time-correlated differential deflection between the
pier and midspan displacement measurements
integrated from velocities where i = V, T, or R
component. The bending strain (εb) is computed Figure 7. Correlation of measured and computed
as: axial strains.

(4) Computed total axial strains are compared to

measured longitudinal strains in Figure 7 at the
where δmax is the time correlated differential midspan. Strong correlations exist for both the
displacement in either the V or T directions crown and springline. However, computed

- 122 -
springline strains overestimate the actual, local 4.6 Amplification of pipe response
strains, likely due to the location of the geophone
used for computation, which was only on the pipe Displacements time history plots over 5 sec are
crown. On average, the computed crown strain is given in Figure 8 for blast 4 comparing ground
nearly a perfect predictor for actual, local strain vibration measurements with the pipe overlying
with a relationship factor of 1.0 and correlation the pier (a) and with the midspan (b). Although the
coefficient of 0.94. main blasting ground excitation lasted 90

Figure 8. Blast 4 transverse and vertical displacement time histories comparing (a) ground vibrations at the pier
base and top of the pipe overlying the pier, (b) ground vibrations at the pier base with the midspan pipe top.

- 123 -
milli-seconds, low amplitude horizontal ground strains. Deploying triaxial velocity transducers
motions continued for well over 10 seconds at onto the pipeline surface is far easier and provides
predominant frequencies of 2.8 and 3.8 Hz. This a stable measurement system in comparison to the
long-term, residual ground motion is likely due to challenges affixing strain gauges onto the surface
the saturated ground around the pier bases within for long-term measurements.
the river floodplain. The residual ground motions
matched the natural frequency of the pipe midspan REFERENCES
which was 3.8 Hz, as shown in Figure 5, resulting
in high amplifications in the midspan and pier top American Water Works Association (AWWA). 2017.
lateral directions relative to transverse ground M11 Steel Pipe: A Guide for Design and
Installation, Fifth Edition.
excitations. Similar lateral amplification was
observed between the ground and pier top while
Bonds, R. W. 2017. Design of ductile iron pipe on
vertical motions were not amplified and somewhat supports. Ductile Iron Pipe Research Association
constrained by the ring supports. (DIPRA).
Maximum amplifications of ground motions at
the pier midspan crown for shallowest and deepest Dowding, C. H. 1985. Blast vibration monitoring and
shaft blasts and one tunnel heading round are control. Prentice-Hall, Inc.
shown in Table 5 in terms of displacements.
Time-correlated values are computed as maximum
differential displacement between the pipe and
ground divided by the ground. In general, vertical
amplifications decreased with blast depth while
transverse amplifications increased with increased
confinement. The tunnel heading round with
horizontally drilled holes generated the highest
and lowest transverse and vertical amplifications,
respectively. Despite the amplifications, the phase
of oscillations match in most comparisons
resulting in the observed low strains.

Table 5. Amplification of crown midspan relative to

ground motions by component.

Blast Vertical Transverse

1 shaft 46.8 13.3

5 tunnel 2.8 50.4
7 shaft 10.7 38.6

5 CONCLUSIONS

Ground vibrations from shaft blasting within a

river floodplain generated low frequency, in-phase
response of an elevated pipeline. Measured hoop
and axial pipe strains were deemed highly
protective of the pipe integrity and far below
allowable yield stresses for grade C steel resulting
in a minimum FOS of 26. Using velocity
transducers affixed to the pipe crown to compute
time-correlated, global axial strains with
contributions from pure axial elongation and
perpendicular bending appears to be an acceptable
approximation of measured axial (longitudinal)

- 124 -
- 125 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Magnitude 2.8 Earthquake Caused by 90 Pounds (41 kg) Per Delay!

Not So Fast

T.A. Davidsavor, P.E. & J. Aiken

Barr Engineering Co., Duluth & Minneapolis, Minnesota, USA

R.E. Burnham, & J.K. Ratliff

Reminger Co., L.P.A., Fort Mitchell, Kentucky, USA

I.G. Wong
Lettis Consultants International, Concord, California, USA

C.T. Aimone-Martin
Aimone-Martin Associates, LLC, Lemitar, New Mexico, USA

ABSTRACT: A small quarry blast was conducted and shortly afterwards, area residents sensed ground
motion and building shaking indicative of an earthquake. In addition, seismometers up to 250 miles (400
km) away recorded a magnitude 2.8 earthquake located in the area. Positioned in a seismically inactive
area, southern Minnesota is not known for earthquakes. The earthquake was originally described by the
USGS to be a surface blast at the quarry. Analysis showed that the earthquake occurred shortly
(approximately seven seconds) after the quarry blast some distance away and well below the surface.
This paper describes the evaluation of the ‘seismic events’ including the blast, the following earthquake,
initial governmental response, and provides the framework for assessing the relationship between the
blast and the earthquake. The paper also discusses management of large volume claims from a complex
event and presents a claims management strategy with best practices.

1 INTRODUCTION decades and is in the middle of the city. The

project location is shown in Figure 1.
On April 25, 2017 at approximately 11:00 am CDT Shortly after the blast, some Mankato area
a routine controlled quarry blast was conducted at residents reported perceptions of ground motion,
the Jefferson Quarry located in Mankato, shaking of objects, as well as sensations of
Minnesota, Unites States of America. The blast building movement in the downtown Mankato
was a regularly performed activity in a region of area. In addition, seismometers up to 400 km
the United States well known for production of (~250) miles away recorded a 2.8 magnitude
high-quality dimension stone and proppant used in earthquake in the area. The earthquake was
the fracking process of hydrocarbon production. originally described by the United States
Mankato and its sister city, North Mankato, Geological Survey (USGS) to be a surface blast at
comprise a city of approximately 60,000 people. the Jefferson Quarry. Early media reports
The Jefferson Quarry had been active for several attributed the reported ground vibration as a

- 126 -
 Figure 1. Site location – Mankato, Minnesota, USA.

weather-related phenomenon associated with the information about blasting to the general public as
quarry blast. Ground motion was noted by well as persistent myths about blasting, ground
residents throughout the area and media coverage vibrations, and their connection to effects on
of the rare seismic event was extensive. Social structures. The law firm of Reminger was assisted
media aided in documenting some of the initial by Dr. Aimone-Martin and focused on establishing
responses. After the earthquake, which was unlike a process for claimants, assessing damage claims,
any observed seismic activity in the past decades and providing claim resolution on behalf of their
of blasting, numerous damage claims were made. client, the insurer, and the contract blaster.
The quarry owners and contractors acted quickly
to offer evaluation of property conditions to those 2 BACKGROUND
who claimed damage from the earthquake. In most
To provide context on the project, background
cases, the claimed damage was found to obviously
related to the blasting practices, geology, and
have occurred before the seismic event.
seismic records are provided.
How was such a ground motion possible from a
normal blast pattern which utilised a maximum of 2.1 Geologic background
41 kg (90 lb) of bulk emulsion per delay?
Immediately after the earthquake, teams focused The Jefferson Quarry is located where the Early
on assessing how the blast may have caused the Ordovician Prairie du Chien Group is exposed in
earthquake, how to evaluate and resolve damage the bluffs of the Minnesota River. Early mining at
claims, and how to permit the mine to return to the property an in the general area focused on
production as the blasting was not substantially production of high quality, thick-bedded Oneota
different than any other blast performed at the Dolomite, which was used extensively as building
property. In assessing the seismic event, Barr’s stone and high-quality aggregate products. The
blasting engineers, mining engineers, and underlying Jordan Sandstone is Upper Cambrian
geologists, with the assistance of seismologists at and was found historically to be an ideal source of
Lettis Consultants, adopted a multiple working silica for glass making and more recently, as a
hypothesis approach that included collecting proppant. The Jordan Sandstone is known to be
available data, evaluating reasonable explanations, well-sorted and consist of well-rounded sediments
and attempted to find the best fit to explain the which are highly durable. As a result of the size
observations. The best fit explanation for the and mechanical properties of the sand grains, the
observed effects was that the timing of the quarry Jordan Formation is mined extensively as a
blast with the earthquake that occurred proppant for hydraulic fracturing. The sand is
approximately 7 seconds later was simply processed and added to fracking fluids for the
coincidence. As a result of the Barr Engineering purpose of keeping induced hydraulic fractures
report, the quarry’s permit to operate was open. The consistency of the Jordan Sandstone
reinstated. However, the public was not entirely combined with the thickness of the unit, its
satisfied with the study revealing several relatively poor cementation, and its proximity to
interesting gaps in our ability to communicate major waterways and roadways has caused the

- 127 -
sandstone to be heavily mined in the area and Blasting was conducted in a typical bench
shipped across the Unites States. Blasting for the blasting operation (with or without free faces)
development of the sandstone is performed to depending on the area and the local circumstances.
allow excavation. The Jefferson Quarry is located near a residential
area with homes in relatively close proximity to
2.2 Blasting background the operation, so there was the potential
for ground vibrations and/or airblasts from
Blasting in the sandstone utilised 101 mm (4-inch) the operation to be perceived by nearby
diameter holes on a 9.1 m (30 ft) bench that are residential homeowners and these issues
drilled to a depth of approximately 10.1 m (33 ft) were considered in design and past blast
below grade to allow for sloughing of material in execution.
the holes. The design depth for the explosive A review of the past blasting activities found a
column was on average 8.5 m (28 ft) with no total of 55 blasts had occurred at the site in the
subdrill. The holes were typically water-filled due preceding year with an average charge
to operations occurring in the base of the quarry. weight/delay for all blasts in the past year at 31 kg
The drill pattern was based on a 3.7 m (12 ft) (68 lb) and the charge weight/delay had varied
burden and a 4.3 m (14 ft) spacing and each blast from 12 to 59 kg (27 to 130 lb). The charge
generally contained 30 to 65 holes in a staggered weight/delay for the blast on April 25, 2017, was
pattern. Stemming heights have varied from 1.8 m 41 kg (90 lb) per delay. Based upon standard rules
to 4.9 m (6 to 16 ft) with 9.5 mm (¾ inch) crushed of attenuation, the amplitude of the ground
stone used as the stemming material. The vibration at 300 m (1000 ft) from a blast using a
explosive used in the sandstone was an charge weight of 31 kg (68 lb) would be 4.3 mm/s
emulsion/ANFO blend consisting of 70% (0.17 in/sec), whereas a charge weight of 41 kg
emulsion and 30% ANFO. The density of this (90 lb) at the same distance would yield a ground
product is 1.28 to 1.30 g/cc. The emulsion blend is vibration amplitude of 5.3 mm/s (0.21 in/sec),
pumped from the bottom of the hole. The holes which is not able to be differentiated by the
were primed with one 500 gram (1 lb) cast average human.
booster. Non-electric detonators were used to The average powder factor for all blasts the
initiate each blast hole. The blast used a timing of previous year had been 0.32 lbs./ton and the
42 ms between holes in a row and 67 ms between powder factor ranged from 0.12 kg/ton to .21
rows. kg/ton (0.24 to 0.42 lbs./ton). The powder factor

Figure 2. Sandstone upon blasting/seismic event.

- 128 -
for the blast on April 25, 2017, was 0.17 kg/ton 2.3 Blasting seismic data
(0.23 lbs./ton), the lowest powder factor over the
past year. The Jordan Sandstone is a consolidated Ground vibrations were recorded by Vibra-Tech
formation but is only lightly cemented and does from the blast on April 25, 2017 and the peak
not require much energy to loosen the material for particle velocities (PPV) at several locations were
excavation, so the lower powder factor for the very similar to the past blasting activities with the
blast on April 25, 2017 was not significantly low maximum PPV recorded on the mine property at a
enough to be a concern. Figure 2 shows the blast distance of 74 m (243 ft) at 53.6 mm/s (2.11 in/s)
area after the April 25, 2017 blast indicating the at a frequency of 50 hertz and at a nearby
blast performed normally with no evidence of residence at a distance of 195 m (640 ft) within the
unusual performance. US Burau of Mines recommended limits set forth

Figures 3. Summary of PPV from previous blasts.

- 129 -
 Figure 4. Waveform from April 6 and April 25, 2017 blasts.

in Report of Investigation 8507 at 17 mm/s (0.67 activity. Since the blast appeared unremarkable,
in/s) at 51 hertz. A seismograph at a home at a seismologists were engaged.
distance of 254 m (834 ft) recorded a PPV of 12
mm/s (0.49 in/s) at a frequency of 62.5 hertz. A 3 UNEXPECTED OUTCOME
similar PPV of 6.7 mm/s (0.265 in/s) was recorded
at a distance of 277 m (907 ft) at a similar Earthquakes in Minnesota are rare and according
frequency of 35.7 hertz. A seismograph located at to Chandler (2020), “Minnesota has one of the
the city wastewater treatment plant at a distance of lowest occurrence levels of earthquakes in the
(1083 ft) recorded an initial PPV of (0.093 in/s) United States, but a total of 20 small to moderate
with general frequencies above 10 hertz, much earthquakes have been documented since 1860.”
higher than another separate impulse with the These earthquake locations and the location of
same magnitude, but with frequencies of about 3 documented regional faults are shown in Figure 5.
hertz. Figure 3 provides the waveform recorded at The most intense documented earthquake was
the nearby wastewater treatment plant on April 6, the Long Prairie earthquake (location 1) in 1860
2017, and of the blast that occurred on April 25, and was estimated to be a magnitude 5. Others
2017. The second event (the earthquake) with including a magnitude 4.7 (New Prague, location
significantly lower frequency is clearly visible in 2, 1860), a magnitude 4.6 to 4.8 (Morris, location
the record. Figure 4 provides a summary of the 11, 1975), and a magnitude 4.1 (Dumont, location
PPV from all blast events recorded over the past 18, 1993) in addition to an additional 13 greater
year. The data in Figure 4 plots the recorded than magnitude 2.8 have been document. The
PPV and frequency of the peak, clearly showing nearest previous earthquake to Mankato was
the April 25, 2017 blast behaved very centred near New Ulm (40 km, 25 miles northwest
much in line with the previous blasting at the of Mankato) in 1881 and was estimated to between
property. magnitude 3.0 and 4.0.
From the summary of the recently blasting The USGS located the event at the Jefferson
data, Figure 4 shows that the PPV from the April Quarry despite an apparent discrepancy between
25, 2017 blast falls within the bounds of the the relatively large magnitude of the event and that
historical data for blasts in the Jordan Sandstone expected from a quarry blast. This was possible
formation. Vibra-Tech aided the team greatly in due to the relatively large uncertainty associated
seismic data summary and analysis to evaluate the with automated location of the event using sparse
blast in question in comparison to past blasting seismograph coverage. Their automated solutions

- 130 -
 Figure 5. Seismicity in Minnesota from 1860 to 2020 (Chandler 2014).

were based only on P-wave arrival times and away, but given the uncertainty caused by
resulted in locations at 6 km northwest of the relatively few seismic monitors in area, they
Jefferson Quarry, with depth at 8 km +/- 10 km. settled on the quarry as a reasonable location.
Given the range of uncertainty, and media reports
circulating about the earthquake occurring at about 3.1 Media and Social Media response
the same time as the quarry blast, they located it at
the Jefferson Quarry without a thorough review of It seems clear that media reports, social media
available data. The USGS was not able to refine response to those reports, as well as web-based
the location of the epicentre accurately using crowd sourced information played a significant
standard automated modelling and initially located role in identifying the Jefferson Quarry as the
the epicentre near another quarry approximately source of the earthquake. Perhaps the biggest role
10 km (6 mi) north of the Jefferson Quarry. Other was that of the gap in understanding between the
modelled locations were further to the southwest. energy levels involved in a typical quarry blast as
Ultimately, the USGS attempted to find the best fit compared to the energy required to cause an
for the data and settled on the Jefferson Quarry in earthquake that is felt hundreds of kilometres from
Mankato. According to the USGS, there were the source area. Had the operators of the Jefferson
questions about the energy levels possible to cause Quarry known at the time that the earthquake
a response on a seismometer over 200 km (125 mi) occurred several seconds after the blast, they may

- 131 -
 Figure 6. USGS DYFI reports.

have made different statements to the media. preliminary locations that did not identify the
However, the proximity in time resulted in the Jefferson Quarry.
assumption that the two events were correlated. It’s a fair assumption that most scientists do not
While the traditional media and social media regard social media as a reliable or even useful
reports were putting out information linking the means to conduct scientific inquiry. However, in
Jefferson Quarry to the earthquake, the USGS was this case, there was information available from a
running automated software that put the epicentre large crowd sourced data platform hosted by
somewhere near downtown Mankato. It is not USGS called ‘Did you Feel it.’ Most responses
known why the USGS decided to attribute the were from the downtown Mankato area which is
earthquake to a quarry blast. However, it is known not unusual given the greater population density
that the USGS made their determination and than more rural areas. However, it was notable that
selected the Jefferson Quarry after assigning despite the comparable geology and population

- 132 -
density between the downtown area and the additional solutions, one with the depth fixed at
area near the Jefferson Quarry, there were the surface and one in which the depth was
proportionately more subjective reports of shaking allowed to vary along with the other hypocentral
to the south of the Jefferson Quarry that fit the parameters. The NEIC location in which the focal
modelled locations by LCI. Figure 6 shows the depth was fixed at the surface and the epicentral
frequency and locations of the DYFI reports. location was allowed to vary, locates 4 km
The social media reports combined with the northwest of the Jefferson Quarry. The NEIC
location model added confidence to an epicentre solution in which all the parameters were allowed
location that was significantly different from the to vary is located 6 km north of the Jefferson
location initially identified by the USGS. Quarry. The focal depth for this solution was 10 ±
8 km. Therefore, the NEIC locations have
3.2 Felt reports considerable uncertainties associated with them.
In February of 2022, the National Earthquake
To add an additional constraint to the assessment Information Centre of the USGS acknowledged
of the location of this earthquake, data collected by that the event was an earthquake and updated the
the USGS on felt intensities is shown in Figure 6. event to include the following text: "Based on
These data are earthquake effects reported by news media reports, low regional seismicity, and
citizens aggregated in geocoded squares 1 km in proximity to a quarry, this event was originally
size. The traditional Modified Mercalli (MM) felt labelled a quarry blast. Subsequent independent
intensities are converted to Community Decimal analysis of nearby data shows the signals observed
Intensities (CDI) to give an average assessment of on the regional seismograms are most likely from
felt effects in the box. Although the CDI maps are an earthquake. The observed event is most likely
prone to bias (population density, type of an earthquake that occurred several seconds after a
structures, interest in reporting, etc.) it can provide quarry blast."
useful, if anecdotal, information on the location of
the earthquake epicentre. In Figure 6 the locations 5 SEISMIC ANALYSIS
of the geocoded boxes are identified by squares
and are labelled with the CDI for that box. The It was apparent that the energy released by the 41
palest shading indicates a CDI or 1 where the kg/delay (90 lb/delay) quarry blast was not capable
darkest shading indicates a CDI of 5 and the of producing ground motions that were felt as far
number in each block indicates the reports on the away as 235 km (146 mi), and a separate
DYFI and does not include local calls. Most earthquake was the concern. Barr teamed with
responses are in the centre of Mankato with CDI Ivan Wong, of Lettis Consultants International
corresponding the MMI values of III-IV. (LCI), to evaluate the seismic data, locate the
Comparing the relocated epicentres for this study hypocentre of the earthquake and to opine on the
and the CDI data suggests that the earthquake comparative energy and seismic signature of the
epicentre is located south of the Jefferson Quarry quarry blast. LCI found that although a precise
and not at the locations determined by the NEIC. location was difficult to calculate due to the
paucity of nearby seismographs, it could be
4 USGS RESPONSE concluded that the quarry was most certainly not
the hypocentre of the quake. Using standard
In an email exchange with the Director of the seismological techniques and selection of an
National Earthquake Information Centre (NEIC) appropriate crustal model, the hypocentre was
of the USGS, we learned that the earthquake likely several miles south of the quarry, and at
located by the NEIC had its location fixed at the least 1 km below the surface.
Jefferson Quarry because their free location was The earthquake epicentre was assessed using
sufficiently close to the Jefferson Quarry that they the program HYPOINVERSE-2000 (version 1.40,
could not distinguish it from the Jefferson Quarry Klein 2014) to verify the NEIC locations and to
given the location uncertainties. The USGS test the sensitivity of these locations to variations
believed the timing of the event coincided with the in crustal velocity structure. HYPOINVERSE is an
blast, but they did question the origin of the event earthquake location algorithm developed by the
because it seemed too large for a blast. The event USGS that has become the standard for locating
was initially located by the NEIC in response to earthquakes and other seismic events throughout
news reports of strong ground shaking in the the U.S. Its purpose is to use the arrival times of
Mankato area. The NEIC fixed the location at the seismic P- and S-waves records at seismograph
Jefferson Quarry and indicated that there are two stations and calculate the event location.

- 133 -
It requires as input a velocity model of the crust, a Only solutions that had an RMS error of less
list of station locations, and a file of seismic phase than 0.30 sec were considered. The focal depths
arrival times. The relocation procedure focused on range from very shallow (0.4 km) to mid-crustal
improving the selection of P-wave arrival times, depths (10 km). The horizontal errors are on the
adding S-wave arrival times, and using several order of 1 km and the vertical errors range from 1
different regional velocity models to evaluate the to 2 km. Since the location statistics are very
sensitivity of the solutions to variations in crustal similar for each of these model runs, it is difficult
velocity. to select the best solution. This is often the case for
The seismograms from the regional earthquakes that are located using data from
seismograph stations were obtained from the regional seismograph stations. Unless there is a
Incorporated Research Institutions for Seismology station within about 10 km of the epicentre,
(IRIS) website (download date 19 May 2017). locations are typically not well constrained.
Records of the event were from three-component
broadband seismometers ranging in distance from 5.1 Can a blast cause an earthquake?
50 to 800 km away. The initial NEIC location only
used the P-wave arrival times. These arrivals were The assessment simply concluded that there was
reviewed and in some cases the P-wave arrival no evidence of a causal relationship due to the
time was repicked. The event was relocated using relatively small energy of a blast compared with
a simple three-layer velocity model and the travel the much higher levels associated with ground
time residuals examined to assess the accuracy of movement felt hundreds of kilometres away. One
the pick. Then the S-wave phase arrival times were possible theory is the ‘hair-trigger’ hypothesis
picked. The S-wave phase is more prominent on which states that tectonic stresses are so high that a
the records than the P-wave arrivals, which were small stress perturbation is sufficient to set off a
often very emergent. Adding S-wave phase data critically stressed fault. The problem with this
to the location procedure is very useful in argument is plausibility. Every week there are
constraining the hypocentral location. The event hundreds of quarry blasts within the roughly 200
was relocated to check the accuracy of S arrivals. km range of the seismographs that reported this
Once confidence was established in the selected earthquake, including dozens of blasts that had
phase arrival times, the event was relocated occurred before this event at this same quarry as
several times using different velocity models. well as at least a half dozen other quarries in the
These models were taken from the literature and vicinity of Mankato. If the stress build-up on a
were selected based on how well they represented fault was so close to rupture that a relatively small
the average crustal velocity of the North American blast and/or accompanying rock movement could
Midcontinent. Because most of the stations are at set it off, why hadn’t the earthquake occurred
least 100 km from Mankato, the details of velocity during larger blasts that had previously occurred at
variation near the surface is unimportant. At these this or nearby quarries?
distances, most of the seismic energy travels Although most blast experts would discount the
through the mid to lower crust or is refracted along potential for a quarry blast to result in an
the Moho. The closest station, N4I37B, was about earthquake, many in the public and potentially the
50 km from the Jefferson Quarry and several USGS, believe that the quarry blast was the
relocations were made using a model relevant to earthquake. This would require one to believe that
the crustal structure in the Mankato region. This the response felt 200 km away was a quarry blast.
was the Gambler High model (Braile 1989), which A USGS geophysicist was quoted in the Mankato
is tied to the midcontinental gravity high that Free Press as stating that the raw seismological
extends from Lake Superior southwest to about the data indicated that the explosion caused the
latitude of Mankato. While the use of a specific shaking and was quoted as stating “I would say it
velocity model for station N4I37B resulted in looks fairly blasty.” Of course, the obvious
reasonably good hypocentral solutions, it was problem with this position is that one would have
not materially different from those solutions to explain why hundreds of other blasts within the
using just a single model for all stations. region would not have triggered a similarly distant
HYPOINVERSE was run numerous times using seismological response and the significant
several velocity models and different weighting difference in energy between a magnitude 2.8
schemes for the P and S arrival times to arrive earthquake and a quarry blast that generated low
at a suite of hypocentral solutions with acceptable ground vibration. In addition, the onsite monitors
error statistics. showed a clear reduction in the blast energy within

- 134 -
hundreds of metres, not hundreds of kilometres. In 6.1 Damage evaluation
addition, the seismic signature of the quarry blast
was shown by LCI to be demonstrably different Of the approximately 50 letters initially received,
(much higher frequency) than the signature about 35 responses from residents identified
observed by USGS. damage they believed was from the earthquake
and were seeking compensation. Upon receipt of
6 CLAIMS AND RESOLUTION FROM the responses, each homeowner was contacted to
LEGAL PERSPECTIVE inform them that their claim would be evaluated
by a qualified engineer, at no cost to them, and to
Within minutes of the earthquake, social media set up appointments to inspect each property. This
accounts were flooded with reports of buildings process resulted in a few more residents dropping
shaking and widespread property damage. As a their claims. At the time of inspections, 27 damage
result, city officials set up a phone number to claims remained. Most claims were clustered
report damage concerns and maintained a list of within 0.5 kilometres (1,500 ft) of the quarry, with
each resident who called, regardless of the nature one outlier 1.5 km (1 mi) away. Within 120 days
of the complaint. By this time, overall suspicion of the earthquake, the inspections were complete.
and the majority of initial reports focused on the Counsel also held interviews with the local
blast event at the quarry as the likely cause of the authorities, who were considering criminal charges
earthquake. against the blaster who controlled the blasting
City officials suspended blasting operations and permit for the quarry. Ultimately, no criminal
the blasting company immediately contacted its charges were filed, and the blasting permit was
insurance company. Within 24 hours, the reissued.
insurance company reached out to counsel for Of the 27 homes inspected, all but three
advice. Working with the blasting company, showed signs of damage that obviously existed
Reminger immediately contacted city officials and prior to the earthquake, which was another
gained access to the list of the more than 50 advantage of a quick response and investigation.
residents who had reported damage concerns. Recent structural damage is relatively easy to
Assuming an average of just $20,000 per identify, and often includes jagged edges to cracks,
claim, we were facing an aggregate claim of the lack of foreign material inside the crack, and
potentially more than $1,000,000. Additionally, dust and other material from inside the crack
at least one of the homeowners alleged the surrounding the area. By asking the homeowners
earthquake triggered a landslide behind his for specific information during the inspections,
home, with more than $100,000 in estimated including when they last painted areas where
repairs. issues were present, evaluators were able to
Counsel took a proactive approach to eliminate most damage complaints, since most of
investigating and responding to all claims. Our the cracks had paint in or over them. However, a
first step was to convince city officials to allow us delay of six months or more would have made it
to actively manage the damage investigation. We much more difficult to distinguish between a
did not want City engineers, or others without pre-existing and new crack. If the inspections
experience in identifying blast damage issuing occurred after the homeowners had performed
damage assessments or causation opinions, since repairs or repainted, a substantial portion of
they likely lacked objectivity and the necessary evidence would have been lost.
experience. City officials were eager to turn the Based on the inspections, only three homes
claims management and investigation over to appeared to have recent damage. These homes all
experts, along with the list of residents who had suffered full or partial chimney collapses. From an
expressed damage concerns. As legal counsel for engineering perspective, this type of damage
the blasting company, our next step was sending makes sense relative to the ground motion of an
letters to each resident asking for a written earthquake. Low frequency ground vibrations tend
description of the alleged damage, along with to result in resonance of low frequency structures
photographs, and other proof of their claims. In such as chimneys, with upper structures more
our experience, when asked to take affirmative likely to suffer movement and resulting damage.
action to assert a claim, a number of people will The chimneys were likely already in poor
decide not to pursue it further. This approach also condition and the average age of the houses which
separated residents who called because they were experienced chimney damage was at least 80
anxious, but who had not actually identified years. Figure 7 provides a view of one of the
cosmetic or structural damage. chimneys with loose brick visible on the shingles.

- 135 -
 Figure 7. Chimney damage.

6.2 Claim settlement conclusion that the earthquake was unrelated to the
blast event, the motivation for settling the three
The investigations were followed with written chimney claims was to appease the local
reports from the structural engineer to each authorities with the goal of maintaining the blast
homeowner, utilising photographs from the permit for the quarry. As a result of the proactive
inspections to explain why the damage was not investigation and response, all claims were settled
seismic-related. Having established how each quickly for a nominal amount, and the Quarry was
claim was not consistent with vibration damage, able to maintain its blasting permit and eventually
the information from the engineer and the data resume operation.
from the seismology report concluded no causal
relationship existed between the blast on April 25, 7 SIMILAR EVENTS
2017, and the earthquake that occurred the same
day, but seconds later. Additionally, published In research of this event, similar events have
USGS data that showed a magnitude 2.8 seismic occurred where earthquakes were noted within
event was unlikely to cause structural damage. close timing to a quarry blast.
Without admitting fault, the claims of the three 7.1 Mccook, Illinois - 2013
homeowners who alleged damaged chimneys we
were able to be resolved for less than $15,000 US One identified was relatively recent in 2013 at
Dollars. Although counsel agreed with the the Federal Quarry in McCook, Illinois, USA.

- 136 -
A summary of the findings from the Illinois 7.2 Drummond island, Michigan – 2016
Department of Natural Resources Quarry
Inspection summarises: In Michigan, USA, in 2016, an earthquake was
The Department initially responded to the associated with a quarry blast at a limestone
reported seismic event on 11/4/2013 and operation in a remote area of the Upper Peninsula
conducted a blasting records and site inspection near the US-Canada border. A report from the
the following day. This inspection included a radio station 100.5 The River – Grand Rapids
comprehensive review of all blasting and states:
seismograph records, seismograph locations and a It's been a couple of years since Michigan
physical inspection of the quarry and blasting experienced an earthquake, but a mining explosion
sites. This inspection revealed that all blasting on Drummond Island came close. The explosion
activity at this site was conducted in accordance registered 2.7 on the Richter scale. The explosion
with the ground vibration requirements set forth at a dolomitic limestone mine on Drummond
within 62 Ill Admin. Code 300.225. Island took place on June 28, 2016. The
The operator monitors its blasting activities Drummond Island mining operation is run by
with four (4) permanently installed seismographs Carmeuse Lime & Stone. I spoke with Ray
surrounding the quarry and supplements LeClair, Carmeuse Lime & Stone's Director of
monitoring with an additional portable Operations. He oversees the operations in the
seismograph which it positions at the nearest Great Lakes region including the Drummond
protected structure to each blast. On 11/4/2013, Island operation. LeClair said the mining site on
these seismographs were positioned between Drummond Island has been open since the late
809-ft and 1,257-ft from the surface blast, and 1940s and after checking the report from June 28,
recorded two separate events. The first event 2016, they found "nothing abnormal about the
occurred at approximately 12:39 P.M. and is report" and he was "not quite sure why it
attributed to the surface blast conducted at the registered." LeClair said the blast was similar to
quarry. The subsequent event, recorded others that they do on the island about ten times
approximately seven (7) seconds later, is of each year.
unknown origin. Again, analysis of the The 2.7 registered by the blast is smaller than
records and seismograph information indicates the earthquakes felt in 2015 when two earthquakes
that the surface blast on 11/4/2013 was were recorded in lower Michigan. An earthquake
conducted in accordance with the applicable registered 4.2 in Galesburg on May 2 and a 3.3
standards. was recorded near Battle Creek on June 30. It is
This earthquake was also indicated by the not entirely unusual for a mining explosion in
USGS to be a quarry blast as summarised from a Michigan to set off USGS monitors. On February
report from the Chicago Tribune that day that 16, 2015 a 2.5 was recorded from a mining
read: explosion near Alpena. On January 12, 2015 a 2.3
The blasting was strong enough to register 3.2 was recorded near Negaunee in the Upper
on seismographs. But the U.S. Geological Survey Peninsula.
said the wave patterns do not match those from a
quake. "Based on what they've looked at, we're 7.3 Clintonville, Wisconsin – unrelated to
pretty sure it's from a blast," said Paul Caruso, a blasting – 2012
geophysicist with the Survey. "It is not an
earthquake." The Survey initially reported the In Clintonville, Wisconsin, USA, complaints of
magnitude as 3.7 and added it on a list of earthquakes over a five-day period in an area that
earthquakes. But it later reduced the magnitude to is seismically inactive and not near a quarry were
3.2 and added: "Quarry blast." noted for several days in March of 2012. Over 300
Again, the ability to generate a magnitude complaints were made over the five-day period.
3.2 earthquake from a relatively small Although much rarer than in seismically active
quarry blast is not possible from an energy areas, earthquakes do occur in seismically inactive
standpoint. Very similar to the earthquake areas. The occurrences tend to be obviously more
that occurred in Mankato, the lower frequency, infrequent and of lower average magnitude. With
higher energy earthquake occurred about mining operations spread around the country, and
8 seconds after the quarry blast with blasting happening at hundreds of properties, the
the quarry blast showing unremarkable PPV for simple chance of an earthquake happening in close
the blast. timing to a small blast cannot be simply ignored.

- 137 -
8 RECOMMENDATIONS close-in seismic stations but it appears that it
occurred to the south of the quarry by a few
A prompt and thorough investigation was key to kilometres at a depth typical of earthquakes in the
resolving the damage claims. Certainly, given the central U.S. Examination of the records of the
hysteria over the event, if we had delayed even six earthquake and felt reports indicate that it was the
months, the outcome would have been drastically earthquake that was felt by the local population
different. Another key effort was convincing and not the blast. As recorded by seismographs
city officials to allow us to take the lead in around the Jefferson Quarry, the blast was too
investigating the damage claims, rather than small to be felt at distances outside the immediate
allowing a city housing inspector or someone with vicinity of the quarry. This earthquake occurred at
no damage investigation experience, much less a similar magnitude and occurred at a similar time
blasting experience, to conduct the investigations. to the earthquake that was observed in McCook,
Individuals, even engineers, without blasting Illinois in 2013.
experience tend to be swayed by the general belief The relationship between the blast and the
that blasting vibrations are highly likely to damage earthquake is unknown. If the earthquake was
structures. We avoided that prejudice by using located at depth and a few kilometres south of the
structural engineers with strong blasting Jefferson Quarry, we believe it is unlikely that
experience. Additionally, by forcing concerned there is a cause-and-effect relationship. In our
homeowners to actively identify and provide proof judgement, the size of the blast would be too small
of their claims, we eliminated a significant number to trigger an earthquake even on a critically
of complaints. The early response enabled us to stressed fault.
mollify the hysteria by providing a source to direct Should a similar situation arise where an
homeowner concerns and establish trust, which earthquake occurs within close timing of a small
enabled completion all the inspections within four blast, Important takeaways include being proactive
months of the event. in addressing concerns, claims, and complaints
Ultimately, the key to the successful resolution and being transparent with respect to what is
of this matter was to have an experienced team in known and unknown regarding cause and effect.
place which allowed an immediately response. One of the unfortunate gaps in the data was that
From the onset, the blasting company, insurance the on-site seismometers were set to shut off 5
company, counsel and experts aggressively took seconds after blasting. Had they run only 20
the position that blast event did not cause the seconds longer, we would have had a much more
earthquake. But experience told us we needed to complete picture of the differences between the
adopt an initially neutral stand and investigate quarry blast and potentially a more accurate
each claim on its merits. Once the team was location of the epicentre with clear separation of
assembled, it was simply a matter of completing the quarry blast from the earthquake.
the investigation and finalising the reports.
In the case of the blast at the Jefferson Quarry REFERENCES
in Mankato and the Federal Quarry in McCook,
the operations would have benefitted from Mankato Earthquake Investigation Report - Regional
monitoring vibrations for a longer period of time Seismic Event on April 25, 2017, June 14, 2017,
after the blast occurred to clearly demonstrate the Barr Engineering Company.
ground motions from the blast and the separate
Braile, L.W., Hinze, W.J., von Frese, R.R.B. &
earthquake. Modern seismographs allow for
Randy Keller,G. 1989. Seismic properties of the
recording times of relatively long duration, and it
crust and uppermost mantle of the conterminous
would benefit from recording at least 20 to 30 United States and adjacent Canada. Geophysical
seconds after each blast to document any Framework of the Continental United States,
subsequent seismic events. Chapter 28., edited by L. C. Pakiser; Walter D.
Mooney, Geol. Soc. Of Am., Boulder, Colo.
9 CONCLUSIONS
Chandler, V.W. 1991. Aeromagnetic anomaly map of
Our analyses strongly indicate that the seismic Minnesota, State Map Series S-17, University of
event recorded and located by the USGS on 25 Minnesota, Minnesota Geological Survey.
April 2017 at 11:02 CDT was an earthquake and
not a blast at the Jefferson Quarry that occurred Chandler, V.W. 2020 & 2014. Minnesota at a Glance
about 7 sec earlier. The precise location of the Earthquakes in Minnesota, University of
earthquake is uncertain because of the lack of Minnesota, Minnesota Geological Survey.

- 138 -
 Persistent link at: Minnesota at a Glance
Earthquakes in Minnesota (umn.edu).

Klein, F. 2014. User’s guide to HYPOINVERSE –

2000, a Fortran program to solve for earthquake
locations and magnitudes, version 1.40, U.S.
Geological Survey Open-File Report 02-171
revised June 2014.

Mooney, H.M. 1979. Earthquake history of

Minnesota, RI-13, University of Minnesota, 20 p.

Mooney, H.M. 1979. Earthquake history of

Minnesota, Minnesota Geological Survey Report
of Investigations 23, 20 p.

Runkel, A.C., Meyer, G.N. & Lusardi, B.A. 2011.

C-26 Geologic Atlas of Blue Earth County,
Minnesota [Part A]. University of Minnesota,
Minnesota Geological Survey.

USGS (June 12, 2017), M 2-8 Quarry Blast – 2 km N

of Mankato, Minnesota. Webpage & Data
Archive, including community reports located at:
“https://earthquake.usgs.gov/earthquakes/eventp
age/us10008kr5#executive”.

https://www.chicagotribune.com/news/breaking/chi-
police-tremors-in-western-suburbs-from-quarry-
blast-20131104-story.html

Report: Earthquake, not blasting to blame for April

shake | Local News | mankatofreepress.com

https://abc7chicago.com/archive/9313036/

https://www.reuters.com/article/us-booms-
wisconsin/wisconsin-town-haunted-by-booms-
traces-noise-to-small-quakes-
idUSBRE82M11J20120323

https://rivergrandrapids.com/mining-explosion-on-
drummond-island-registers-2-7-on-richter-scale/

https://cse.umn.edu/mgs/earthquakes

- 139 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

New approaches for evaluating building response to blast vibrations

M.I. Álvarez-Fernández, M.B. Prendes-Gero, L. Conde-Fernández

& C. González-Nicieza
Rock Dynamics Research Group, University of Oviedo, Asturias, Spain

ABSTRACT: Artificial intelligence techniques and numerical simulations are employed to analyse the
response of a structure to the dynamic effects of a blast. As artificial intelligence techniques, a neural
network is used because it is one of the simplest techniques to apply. As numerical models, the finite
differences are employed. Both techniques are applied in a real case, a scale building built in the vicinity
of an open-pit mine, with the end of comparing the advantages and disadvantages of the techniques.
During the tests, the real response of the building is monitored using accelerometers located in different
elements of its structure. The results show that the neural network allows to obtain a reliable and simple
predictive tool capable of characterising the behaviour of the building. The numerical simulation,
although more complex, allows to analyse the behaviour of the structure and determine its weak points.

1 INTRODUCTION resonate if the ground vibration frequency matches

the natural frequency of the structure and, because
The increasing development of open pit mining, of this, the amplitude of the vibration can grow
driven by the growing demand for minerals, has and become larger than the inducing vibration.
contributed to an increase in the use of blasting Frequency (F) and peak particle velocity (PPV)
explosives. To this day, explosives are the most are the two most commonly used parameters for
efficient source of energy for breaking and measuring ground vibration. Some authors
excavating rock. When an explosive detonates highlight the importance of frequency because the
inside a blasthole, it instantly releases a large structural response depends on the frequency of
amount of energy in the form of pressure and ground vibration. This vibration is influenced by a
temperature. However, only a small portion of this large number of parameters such as the physical
energy is used to break the rock, so the remaining and mechanical properties of the rock mass, the
explosive energy is wasted in producing unwanted characteristics of the explosive used and the blast
effects such as vibrations, noise, subsequent design itself. Furthermore, this vibration is a
breakage, airwave propagation, etc. complex combination of different wave types
Ground vibration is literally a wave motion that (Brinkmann 1987). In general, variables such as
propagates from the point of detonation in the acceleration and velocity of a point over time are
same way as the wave created by a stone falling used to describe it, together with the frequency and
into a pond. When the vibration wave passes propagation velocity of the wave (Bender 2006).
through structures on the ground surface, it However, once the wave reaches a structure,
transmits the vibration to those ones. These not only the intensity, frequency or duration of the
vibrations can cause the affected structures to wave must be taken into account to establish its

- 140 -
safety, but also other variables directly related to phenomena can be difficult to model, because of
it, such as its damping coefficient or its natural the large number of variations in parameters which
frequency. describe material models, finite element types and
Although there are numerous studies related to sizes, and boundary conditions and explosive
dynamic behaviour and damage in structures loading, numerical models can be used to predict
(Faramarzi et al. 2014, Dogan et al. 2013, structural behaviour with fairly good accuracy in
Dowding 1996, González-Nicieza et al. 2014), the comparison to experimental tests (Ivančo et al.
major drawback is that each country has its own 2019, Draganić & Varevac 2018).
standards without consensus between them. In In recent years, a rapidly emerging approach in
general, it can be said that in most standards the this area focuses on using neural networks,
PPV or maximum velocity at which a particle machine learning algorithms, and neural fuzzy to
moves due to the effect of a wave is the parameter predict, simulate, and optimise blasts. The results
that reflects the intensity of the wave, although of these studies show the artificial intelligence as a
some consider its maximum value and others the powerful tool that can solve problems in diverse
modulus of the velocity vector. Other standards and effective ways. Applications of ANNs include
consider the frequency of vibration as a predicting blast-induced ground vibrations
determining factor, although in a simplistic way, (Khandelwal & Singh 2009, Shahri & Asheghi
without taking into account that the different 2018, Das et al. 2018, Álvarez-Vigil et al. 2012),
elements of a structure have different natural quantification of blast loading on a building
frequencies (Dowding 1980, Medearis 1996). behind a blast wall (Remennikov & Rose 2007) or
There is a second damage criterion based on along simple city streets (Remennikov & Mendis
energy, either taking into account the acceleration 2006). More recently, neural networks have been
peak and the frequency at that peak (Crandell applied to do prediction of blast loading in internal
1949), or expressed in terms of velocity (Nicholls environment (Dennis, et al. 2021). However, the
et al. 1971). More recently, it has become limitations of predictive models developed by the
necessary to take into account the energy carried ANN technique lie in the narrow scale of
by the wave at the characteristic frequency of the application since it is designed under the
structure, which can be calculated either from the conditions and properties of a specific area and
energy spectrum (Guosheng et al. 2011), or from cannot apply widely in other sites (Al-Bakri &
the waveform captured with geophones (Chandar Sazid 2021).
& Sastry 2015). In this work full-scale field tests have been
In recent years, the research has focused on done with the end of developing a neuronal
analysing the effect that an explosion can have on network as a predictive tool and carrying out
both civilian and military buildings. To this end, numerical simulations to analyse the behaviour of
equipment capable of simulating explosions has the structure and determine its weak points. The
been designed, such as the so-called shock tubes end is to know the variation of the values of the
(Schleyer et al. 2007, Haris et al. 2018, Poulin different parameters that characterise one wave,
2018, Lacroix & Doudak 2018), blast pendulum once this wave has gone through the structure
systems (McDonald et al. 2018, Jing et al. 2014) (going into it by the foundations and going out by
or blast simulator (Stewart et al. 2014, the beams of the roof). The analysed variables
Rodriguez-Nikl et al. 2011) among others. have been the acceleration, velocity, frequency
However, the most realistic representation of blast and duration in the three axes (X, Y, Z) and the
loading can be obtained only with full-scale field energy.
tests which best mimic real-life situations
(Draganić et al. 2018). Although, the blast 2 LOCATION AND CASE OF STUDY
experimentation is usually conducted on scaled
samples what reduces the need for large explosive An operational open-pit mine is chosen for the
quantities and consequently lowers the overall tests. In this mine, blasting is carried out
danger of injury. constantly and, therefore, it is possible to carry out
However, these tests are limited due to security a continuous study over time. The mine is located
restrictions and a lack of the considerable in Santa Marina of Piedramuelle near the city of
resources required. Therefore, numerical Oviedo, capital of the Principality of Asturias in
modelling and simulation have been proven to be a Northern Spain. In its square a 2 m high concrete
valuable tool in simulating the behaviour of structure with a base of 2.5 x 2.5 m was built on a
structures under blast loading. Although blast spread foundation (Figure 1).

- 141 -
 Figure 1. Process of construction of the real-scale structure.

3 METHODOLOGY 3.1 Definition of the elements

Once the structure is built: Measurements are made on two structural

elements at the bottom of the structure where the
− The structural elements to be measured are input parameters are recorded and on two
defined structural elements at the top where the output
− Accelerometers type Syscom, Etna, 9043, parameters are recorded:
5033, are installed and the maximum
acceleration, is recorded − NLB: North lower beam
− The neural network is developed − NUB: North upper beam
− The numerical simulation is implemented − SUB: South upper beam

Figure 2. Structure and nomenclature of the beams.

- 142 -
 Table 1. Maximum acceleration collected for each axis.

Ax Ay Az Ax Ay Az
Date Ele. Ele.
(mm/s²) (mm/s²) (mm/s²) (mm/s²) (mm/s²) (mm/s²)
20/03/19 BA 14.41 20.09 12.83 NUB 49.63 74.95 74.9
26/03/19 BA 16.03 29.08 17.85 SUB 48.73 92.52 41.97
16/04/19 BA 37.7 67.26 113.9 NUB 67.73 52.73 29.85
06/05/19 BA 169.49 191.49 422.18 NUB 71.85 54.94 35.28
15/05/19 BA 46.01 60.26 25.19 SUB 137.68 214.21 78.79

Table 2. FFT for each axis.

FFTx FFTy FFTz FFTx FFTy FFTz

Date Ele. Ele.
(Hz) (Hz) (Hz) (Hz) (Hz) (Hz)
20/03/19 BA 11.71 11.42 12.3 NUB 19.53 20.5 93.99
26/03/19 BA 42.72 28.02 20.01 SUB 19.53 20.26 93.5
16/04/19 BA 7.81 16.79 30.85 NUB 20.38 0.24 0.73
06/05/19 BA 8.05 8.05 23.92 NUB 20.01 0.97 0.49
15/05/19 BA 16.3 16.45 32.27 SUB 0.48 0.48 0.49

− BA: Base or structure slab objectives are:

3.2 Installation of the accelerometers − Determine the FFT and compare it with the
one obtained in MATLAB verifying the
Once the elements are defined, and the implemented algorithm
accelerometers are placed on them, controlled − Determine the event duration (Table 3) by the
blasting is carried out. At this point, it should be significant duration or interval in seconds,
noted that due to the pandemic, blasting is stopped between 5% and 95% of the Husid function,
for a period of 1 year. where most of the energy is concentrated
With the exception of the accelerometer (Trifunac & Brady 1975, Husid 1973)
SYSCOM 3 that records the velocity, the rest of
them record for each axis and each element the 3.3 Development of the neural network
maximum acceleration (Table 1), while the
velocity is obtained indirectly through the It is used a feedforward backpropagation (FFBP)
integration of acceleration, and the energy is network as the most suitable model for resolving
calculated as the sum of the squares of the velocity an identification problem. In a neural network, the
components. numbers of neurons and layers can be the
To obtain the Fast Fourier Transform (FFT) difference between success and failure. To choose
(Table 2), a MATLAB algorithm is used to the number of hidden neurons the geometric
calculate the dominant frequency from the pyramid rule is employed (Masters 1993), where
recorded waveform. In addition, the value the number of neurons follows a pyramid shape.
obtained is compared with the one obtained with First, one must set the number of neurons in the
the SismoSignal software that treats seismic input layer i and the number of neurons in the
signals coming from earthquake-type events. The output layer o. Table 4 lists several equations

Table 3. Event duration for each axis.

Date Ele. tx (s) ty (s) tz (s) Ele. tx (s) ty (s) tz (s)

20/03/19 BA 1.48 1.2 1.73 NUB 1.37 1.75 1.54
26/03/19 BA 0.19 0.18 0.13 SUB 1.12 1.51 0.97
16/04/19 BA 0.38 0.14 0.07 NUB 2.15 2.74 2.10
06/05/19 BA 0.33 0.27 0.14 NUB 1.91 2.82 2.69
15/05/19 BA 0.66 0.57 0.60 SUB 1.98 2.19 2.37

- 143 -
associated with the calculus of the hidden layer vibration wave corresponding to a blast is applied
neurons for one hidden layer h1 and two hidden to its base and the theoretical response is recorded
layers h2, while Table 5 summarises the in the different elements (foundations, beams,
architecture of the neural network, initially columns, etc.) to compare them with the real
employed. measured records.

Table 4. Computation of the number of hidden layer Table 5. Neural network architecture.
neurons using the geometric pyramid rule.
No. of input neurons 13
Hidden Geometric pyramid rule layers
No. of output neurons 13
1 𝑟𝑟 = √𝑖𝑖 × 𝑜𝑜 ℎ1 = 𝑟𝑟 No. of hidden layers 1
No. of hidden neurons 13
2 𝑟𝑟 = 3�𝑖𝑖/𝑜𝑜 ℎ1 = 𝑜𝑜 × 𝑟𝑟 2 ℎ2 = 𝑜𝑜 × 𝑟𝑟
No. of training cycles 100
Training data items 70%
3.4 Implementation of the numerical simulation Validation items 15%
Test items 15%
In addition to the neural network as a predictive
tool, the application of numerical computation
techniques has been explored, using FLAC3D, Figure 4 shows the vertical component of the
v.7.00, from Itasca Group Consulting. With this acceleration measured at the base and used as
end, a geometric model of the analysed structure input to the model, with a maximum value of 2.52
has been developed and available measurement cm/s2, together with the response recorded at one
campaigns have been used to calibrate the of the beams, whose maximum value is 6.4 cm/s2
parameters of this model (Figure 3). in the simulation, very similar to the 7.8 cm/s2
Once the structure has been modelled, a real obtained in reality.

Figure 3. Model geometry.

- 144 -
Figure 4. Real acceleration waves at the base of the model and response in the beam.

Figure 5a. Regressions obtained by considering all input and output data simultaneously, one hidden layer.

- 145 -
 Figure 5b. Regressions obtained by considering all input and output data simultaneously. (b) two hidden layers.

4 DISCUSSION networks for each axis with one and two hidden
layers of 5 neurons in them were analysed.
4.1 Neural network On the X-axis, there is hardly any difference
in the results obtained when one or two hidden
The calculation of the neural network with all layers are used, so as it is less computationally
input and output data, i.e. considering all axes at expensive, it is decided to work with the simplest
the same time and one hidden layer, show network with regressions close to 0.9 (Figure 6)
regressions over 0.94 for the training data items and a value of 0.06 for the mean square error.
but they drop to 0.65 for the test items (Figure 5a). On the Y-axis, the regression coefficient
To improve these results, it was considered to values, for a neural network with one hidden layer,
work with two hidden layers with the same are very high (Figure 7), and the mean square error
number of neurons in them and equal to 13. In this is equal to 0.09.
case, the results of the regression improve (Figure On the Z axis, both the value of the regression
5b), but the mean squared error presents very high coefficient and the value of the mean squared error
values, so the calculation of different neural is lower with two hidden layers decreasing from

- 146 -
 Figure 6. Regressions obtained for the X-axis with neural network with a hidden layer.

0.16 for one hidden layer to 0.05 for two hidden 9a) and the deformations (figure 9b) in different
layers. In any case the value of the regression is phases of the calculation with a deformation
very low for the validation data items and test data increased 100 times.
items (Figure 8).
5 CONCLUSIONS
4.2 Numerical simulation
The study shows that:
The analysis of the models shows the influence
of wave superposition and amplification − One net should be used for each axis.
of ground vibration, detecting situations − The results of the neural network are not as
of damping or amplification depending on the expected, but the low volume of data
characteristics of the structure and the different available has to be taken into account.
elements. Therefore, a large improvement of the results
On the other hand, the occurrence of can be expected when the training dataset is
unexpected natural frequencies in the structure can extended. Besides, a larger volume of data
be observed and their predominance varies from will allow better filtering of the data, enabling
one element to another. Thus, the dominant the elimination of outliers. This is expected
frequency for the spread foundation would to be done when the current situation
be 14.4 Hz, while for the beams it would reach stabilises.
48.6 Hz. − The numerical simulation has shown different
Figure 9 represents, on the deformed geometry situations that the neural network has not
of the structure, the vertical displacements (Figure been detected as the damping or amplification

- 147 -
 Figure 7. Regressions obtained for the Y-axis with neural network with one hidden layer.

of the vibration due to the superposition and Álvarez-Vigil, A.E., González-Nicieza, C., López-
amplification of ground vibration, and the Gayarre, F. & Álvarez-Fernández, M.I. 2012.
unexpected natural frequencies of the Predicting blasting propagation velocity and
elements. vibration frequency using artificial neural
network. International Journal of Rock
6 ACKNOWLEDGMENTS Mechanics and Mining Sciences, 55, 108-116.

The authors acknowledge the financial support Bender, W.L. 2006. Understanding blast vibration
provided by the Spanish Ministry of Economy and and airblast, their causes, and their damage
Competitiveness through the following ‘Explora potential. Spring 2006 and Fall 2007 workshops
Project’ BIA2015-72928-EXP. of the Golden West Chapter of the International
Society of Explosives Engineers.
REFERENCES
Brinkmann, J.R. 1987. The control of ground
Al-Bakri, A.Y. & Sazid, M. 2021. Application of vibration from colliery blasting during the
artificial neural network (ANN) for prediction undermining of residential areas. Journal of the
and optimisation of blast-induced impacts. South African Institute of Mining and Metallurgy,
Mining, 1, 315-324. 87(2), 53-61.

- 148 -
Figure 8. Regressions obtained for the Z-axis with neural network with two hidden layers.

Figure 9a. Results of the numerical simulations - Figure 9b. Results of the numerical simulations –
vertical displacement. deformations.

- 149 -
Chandar, K.R. & Sastry, V.R. 2015. A new method of wavelet packet energy spectra. Mining Science
estimating wave energy from ground vibrations. and Technology (China), 21, 35-40.
Geomaterials, 5, 45-55.
Haris, A., Lee, H.P. & Tan, V.B.C. 2018. An
Crandell, F.J. 1949. Ground vibration due to blasting experimental study on shock wave mitigation
and its effect upon structures. Journal Boston capability of polyurea and shear thickening fluid
Soc. Civ. Engrs, 222-245. based suspension pads. Defence Technology,
14(1), 12–18.
Das, A., Sinha, S. & Ganguly, S. 2018. Development
of a blast-induced vibration prediction model Husid, R. 1973. Terremotos. Andrés Bello, Santiago
using an artificial neural network. The Journal of de Chile.
the Southen African Institute of Mining and
Metallurgy, 119. 187-200. Ivančo, M., Erdélyiová, R. & Figuli, L. 2019.
Simulation of detonation and blast waves
Dennis, A.A., Pannell, J.J., Smyl, D.J. & Rigby, S.E. propagation. Transportation Research Procedia,
2021. Prediction of blast loading in an internal 40, 1356-1363.
environment using artificial neural networks.
International Journal of Protective Structures, Jing, L., Wang, Z., Shim, V. & Zhao, L. 2014. An
12(3), 287-314. experimental study of the dynamic response of
cylindrical sandwich shells with metallic foam
Dogan, O., Anil, O., Akbas, S.O., Kantar, E. & cores subjected to blast loading. International
Erdem, R.T. 2013. Evaluation of blast induced Journal of Impact Engineering, 71, 60–72.
ground vibration effects in a new residential
zone. Soil Dynamic Earthquake Engineering, 50, Khandelwal, M. & Singh, T.N. 2009. Prediction of
168–81. blast-induced ground vibration using artificial
neural network. International Journal of Rock
Dowding, C.H., Beck, W.K. & Atmatzidis, D.K. Mechanics & Mining Sciences, 46, 1214-1222.
1980. Blast vibration implications of cyclic shear
behaviour of model plaster panels. Geotechnical Lacroix, D.N. & Doudak, G. 2018. Experimental and
Testing Journal, 3(2). analytical investigation of FRP retrofitted
glued-laminated beams subjected to simulated
Dowding, C.H. 1996. Construction vibrations. blast loading. Journal of Structural Engineering,
Prentice-Hall, Englewood Cliffs. 144(7).
Draganić, H., Varevac, D. & Lukić, S. 2018. An
Masters, T. 1993. Practical neural network recipes in
overview of methods for blast load testing and
C++, Morgan Kaufmann: Massachusetts, MA,
devices for pressure measurement. Advances in
USA.
Civil Engineering, 3780482.

Draganić, H. & Varevac, D. 2018. Analysis of blast McDonald, B., Bornstein, H., Langdon, G.S., Curry,
wave parameters depending on air mesh size. R., Daliri, A. & Orifici, A.C. 2018. Experimental
Shock and Vibration, 3157457. response of high strength steels to localised blast
loading. International Journal of Impact
Faramarzi, F., Ebrahimi, F., Mohammad, A. & Engineering, 115, 106–119.
Mansouri, H. 2014. Simultaneous investigation of
blast induced ground vibration and airblast Medearis, K. 1996. The development of rational
effects on safety level of structures and human in damage criteria for low-rise structures subjected
surface blasting. International Journal of Mining to blasting vibrations. Kenneth Medearis
Science and Technology, 24, 663–669. Associates, Final Rept. to National Crushed Stone
Assn., August 1976, Washington, DC. p. 93.
González-Nicieza, C., Álvarez-Fernandez, M.I.,
Alvarez-Vigil, A.E., Arias-Prieto, D., López- Nicholls, H.R., Johnson, C.F. & Duvall, W. 1971.
Gayarre, F. & Ramos-Lopez. F.L. 2014. Blasting vibrations and their effects on structures.
Influence of depth and geological structure on the V.S. Bureau of Mines, Bulletin, 656.
transmission of blast vibrations. Bulletin of
Engineering Geology and Environmental, 73(4), Poulin, M., Viau, C., Lacroix, D.N. & Doudak, G.
1211-1223. 2018. Experimental and analytical investigation
of cross-laminated timber panels subjected to
Guosheng, Z., Jiang, L. & Kui, Z. 2011. Structural out-of-plane blast loads. Journal of Structural
safety criteria for blasting vibration based on Engineering, 144(2).

- 150 -
Remennikov, A.M. & Mendis, P.A. 2006. Prediction
of airblast loads in complex environments using
artificial neural networks. WIT Transactions on
the Built Environment, 87, 269–278.

Remennikov, A.M. & Rose, T.A. 2007. Predicting the

effectiveness of blast wall barriers using neural
networks. International Journal of Impact
Engineering, 34(12), 1907–1923.

Rodriguez-Nikl, T., Hegemier, G. & Seible, F.

2011. Blast simulator testing of structures:
methodology and validation. Shock and
vibration, 18(4), 579–592.

Schleyer, G., Lowak, M., Polcyn, M. & Langdon,

G. 2007. Experimental investigation of blast
wall panels under shock pressure loading.
International Journal of Impact Engineering,
34(6), 1095–1118.

Shahri, A.A. & Asheghi, R. 2018. Optimised

developed artificial neural network-based models
to predict the blast-induced ground vibration.
Innovative Infrastructure Solutions, 3(1).

Stewart, L.K., Freidenberg, A. & Rodriguez-Nikl,

T. 2014. Methodology and validation for blast
and shock testing of structures using high-speed
hydraulic actuators. Engineering Structures, 70,
168–180.

Trifunac, M.D. & Brady, A.G. 1975. A study on the

duration of strong earthquake ground motion.
Bulletin of the Seismological Society of America,
65 (3), 581-626.

- 151 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Proposed methodology to reduce microfractures in working stopes

utilising vibration modelling

N. Valencia
Ecole Centrale Nantes, Nantes, France

ABSTRACT: This article proposes a method that avoids the generation of microfractures caused by
vibrations in working stopes in granodiorite rock with the purpose of predicting the damages generated by
blasting operation at Consorcio Minero Horizonte mine in Parcoy, Peru. In order to evaluate the
microfractures impact at the working stopes, ramps, etc. two methodologies were followed: Firstly, the
algorithm based on finite element method in Matlab language for time discretisation and nodal variables
(modelling of ultrasonic waves) and Secondly, the interpolation analysis to determine vibration levels (far
and near field Devine’s model). Final results: Shotcrete consumption reduction from 174.117 to 157 m3
/month in July and September 2017 respectively; reduction of length microfractures from 25 to 23 mm
per month on excavation of 15.6 to 12% and reduction of the dilution from 25% to 12%.

1 INTRODUCTION by blasting aims to establish the behaviour of the

vibration produced by blasting (attenuation law)
A usual problem presented in the mining industry for a type of rock and by the BDI (Blast Damage
is to recognise the orientation and density of Index) and thus estimate the damage, taking as a
fractures generated by blasting in the mineral parameter the fracturing critical velocity of rock in
deposits: from the knowledge of this parameter working stopes.
one can identify the level of damages generated by
the explosives (density of the explosive, VOD 2 PROCESS OF THE PROPOSED
and drilling diameter) in the contours of the METHODOLOGY
excavation. In addition, it is possible to evaluate
and avoid the risks of accidents by rock falling and The present study was developed in the mining
establishing parameters to optimise the design of company Auriferous that has its mining activities
excavation, improve the granulometry and the in an area of 25,000 hectares (61,776.345 ac). This
damages control on the rock mass. operation is carried out entirely within the Pataz
From this operational problem, the Batholith, District of Parcoy, La Libertad Region,
microfractures after the blasting, has led us to Peru. It is estimated more than 80,000 m
carry out a study of vibrations to set up the (262,467.19 ft) of mining works carried out
influence of the damages on the last limit of the between old and modern, both horizontal and
excavation by deviation and overload, applying vertical.
Devine Far Field modelling. This study of vibrations initially consists in the
Two teams were used Minimate Blaster and characterisation and classification of rock (quality
Minimate Plus seismographs with current of bad rock, type IV-B, with an RMR = 21 - 30) to
calibration. The vibration measurement program prove the intrinsic properties such as compressive

- 152 -
strength, wave propagation speed and modulus of Where: D = 0→ No damage, D=0.5→
elasticity of rock mainly. Then the vibration Moderate damage, D=1.0→ Severe damage.
monitoring was performed at varied distances For calculation purposes, ‘D’ was considered
from the trigger point considering equal operating with a value of 0.
loads to interpret the curve of the attenuation
behaviour of the signal between the maximum 3.2 Determination of wave propagation velocity
vibration velocity (VPP) and the scaled distance (Vp)
(DR) using the point cloud wave vibration for our
rock mass, with this characteristic attenuation law We have (Equation):
we predict the damage influence halos of each
detonated hole and thus determine permissible Vp = 3500 + 1000Log(𝑄𝑄) = ⋯
operating charge of explosives for each rock 3,500 + 1000Log (0.12) = 2,583 m/s (1)
quality which were carried out from April 01
– 2017 to April 30 – 2017 with a total of 38 Where Q= Barton’s Q rating.
seismographs as a result of the lifting of 48
blasting’s between working stopes and full 3.3 Particle Critical velocity VPPc (VCRIT)
– face advance mining works. The charging of the
Level at which the altered casing rock presents
blast holes was made using dynamite cartridges
incipient fracture formation.
dynamite ‘A’ 45% of 1 1/8" X 12" (2.85 cm x
30.48 cm) and dynamite ‘B’ 45% of 1 1/8" X 8" 𝜎𝜎𝑡𝑡 ×𝑉𝑉𝑝𝑝 2,34×2583
(2.85 cm x 20.32 cm). (Technical specifications of 𝑉𝑉𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 =
𝐸𝐸𝑖𝑖 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷
=
19,8
= 304,5𝑚𝑚𝑚𝑚/𝑠𝑠 (2)
dynamite ‘A’, ‘B’ in Annex1).
3 VIBRATION ANALYSIS AND 3.4 Devine´s model, 1962
MODELLING
The DEVINE criterion (far field) was used for a
The conducted work includes the study of multiple regression analysis based on scalar
predictive vibrations and control, the objective is distance and the peak velocities recorded in each
to obtain seismic information of the ‘explosive- monitoring.
rock’ interrelation by providing a tool to
−𝛼𝛼
approximate and approach the real conditions of 𝑑𝑑
the blast. Data collection was obtained through 𝑉𝑉𝒑𝒑𝒑𝒑 = 𝐾𝐾 × � 1 � (3)
𝑄𝑄 3
field monitoring surveys (monitoring).
For each explosive at d=0.6 m from the centre Here:
of gravity of the shot with a permissible vibration,
V = particle velocity (mm/s)
limit of 304.5 mm/s at a drilling length of 1.829 m.
d = Distance to blasting point (m)
In order to analyse the vibration wave model K = Velocity factor
we have to calculate the following parameters:
Q = maximum charge per delay (kg)
3.1 Hoek & Diederich Equation α = slope of the best-fit line of the V vs. D/Q^1/3
plot on a log-log scale

Table 1. Input Data for Hoek & Diederich equation. From (3) we obtain:
Nomenclature Value 3
Bieniawski Rock Mass 𝑉𝑉
𝑄𝑄 = � × 𝑑𝑑𝛼𝛼 �
𝛼𝛼
(4)
RMR89 25
Classification 𝐾𝐾
GSI geological strength index 25
D disturbance factor 0 The geologist provided the parameters of the
modulus of deformation rock (table 2) corresponding to each of the 2
Em 1050.4 explosives used.
(MPa)
Em/Ei modulus relation 0.06
Table 2. Constants of the Rock Massif of the
deformation modulus of Dynamites at CMHSA.
Ei 17.5
the intact rock (GPa)
dynamic deformation Explosive K α
Ei Dynamite A 230.36 1.12
modulus of the intact 19.8
dynamic Dynamite B 852.05 1.81
rock (GPa)

- 153 -
 Figure 1. Model of the peak particle velocity attenuation with distance.

Table 3. Damage criteria.

Cameron McKenzie damage criteria
4V_crit’ → Intense Fracturing 4V_crit’ → Intense Fracturing → 1218mm/s
1V_crit’ → Creation of new invoices 1V_crit’ → Intense Fracturing → 304,5mm/s
1⁄4 V_crit’ → Existing fracture extension 1/4V_crit’ → Intense Fracturing → 76,13mm/s

Peak Particle Velocity Vpp is the maximum have 304.5 mm / s (11.98 in/sec) at 426.3 mm / s
value of the speed of the particles during a (16.78 in/s).
vibration: it can be easily measured by
seismographs. Knowing the distance from the
source, the PPV at that point and the parameters of
the rock, the evolution of the Peak particle velocity
can be calculated using equation (3).
Seismic waves represent a transfer of kinetic
energy: the energy is radiated in 3 dimensions
from the source, so the energy flux decreases with
the distance from the explosion. Moreover, due to
non-elastic phenomena in the rock matrix, the
wave dissipates energy. These factors induce a
reduction of the wave amplitude and of the
Figure 2: Attenuation Law, Dinamite A.
peak particle velocity with distance: this is
accounted by the attenuation factor α of equation
number 3.
According to the critical particle velocity,
different damage levels can be identified.
Using MATLAB, a simplified model has been
produced, reported in figure (2).

3.5 Damage criteria: damage generation range

for casing rock

Reference Table 3.
The maximum tolerable range for our case will
Figure 3. Attenuation Law, Dinamite B.
be considered from 1 to 1.4 times the Vcrit so we

- 154 -
3.6 Attenuation Law of Blasting wave 6.25 cartridges per delayed period (one blast
hole at a time) to 0.45 kg (0.992 lb) as maximum
Each explosive has a certain release of energy to charge per hole and an influence from the centre of
which a certain vibration level corresponds, so the 0.6 m (1.97 ft) blast is (Table 4):
vibrographic data was taken from each type of
explosive to determine its behaviour in the rock.
Table 4. Type of damage.
3.7 Maximum charge per delay Damage
𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 − 𝐴𝐴 → 𝑄𝑄 Moderate
For each explosive at 0.6 m from the centre of
gravity of the shot with a permissible limit = 0,45𝑘𝑘𝑘𝑘 → 𝑉𝑉𝑉𝑉𝑉𝑉
vibration of 304.5 mm/s in a drilling length of 6 = 455,44𝑚𝑚𝑚𝑚/𝑠𝑠
feet. 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 − 𝐵𝐵 → 𝑄𝑄 Smooth
= 0,45𝑘𝑘𝑘𝑘 → 𝑉𝑉𝑉𝑉𝑉𝑉
𝑑𝑑 = 302,91𝑚𝑚𝑚𝑚/𝑠𝑠
𝐷𝐷 = 3 (5)
�𝑄𝑄

Where: From the geomechanical characteristics of the

d = Distance to blasting point (m) rock we have the following summary table in
K = Velocity factor which it is worth noting that the average wave
Q = maximum charge per delay (kg) propagation velocity (Vs) is 2,969 m / s (9,740.81
D: Scaled distance m/kg3, for cylindrical loads. ft/s) and the average critical rock fracturing
velocity (VVPcrit) is 376.9 mm / s (14.84 in/s).
If d=0.6 m and replacing Eq. 2 in Eq. 4
maximum charge per delay is: 3.8 Vibration analysis for the influence of
damage
1.977
304.5×0.61.5176
DYNAMITE A → 𝑄𝑄max = � � Clearing the formula of Devine (Equation 3) the
314.18
=0.203 Kg (2, 6 cartridges per delay period) (6) distance ‘d’ and calculating the new formula with
the constants ‘K’ and ‘α’ by type of explosive used
304.5×0.61.1191
2.68 in each blast hole we determine approximately the
DYNAMITE B → 𝑄𝑄max = � � influence of damage by each one of the detonated
230.36
=0.45 Kg (7) blast hole sequentially as shown in the Table 6.

Table 5: Geomechanical parameters of rock.

CALCULATION OF SPEED PEAK CRITICAL PARTICLE

OF THE ROCK

CM HSA
Field data:
Underground mining : Lourdes Type of rock : altered granodiorite
Work : Cx2805 Squeeze of strata : sub horizontal frac turing to the exc av.
Analysis : Vibration analysis Material to blast : c learanc e
Date : 03/04/17 Filling of joints : Qz. Closed, alteration to the c eiling
Zone : North Spacing of joints : 0.13 - 0.18 c m
State of rock : deepening ramp

Geomechanical data: Max Min Prom

λ : Total number of frac tures per c ubic me 10 7 8.5
RMR89 : Roc k Mass Rating (in dry c onditions) 50 46 48
RMR : roc k Mass rating (tight) 35 31 33
δ : Density of the roc k (Ton/m3 ) 3 2.8 2.9 Estimation of
D
D : Damage index 0 0 0 Damage Index
GSI : Geologic al resistanc e index 30 26 28 Without damage 0
Em : Modulus of elastic ity of the roc k mass 1.65 1.15 1.4 Intermediate 0.5
EmΣi : Elasticity module relationship 0.08 0.06 0.07 Severe 1
Ei : Modulus of elastic ity of intac t roc k (Gp 20.2 18.1 19.2
Eid : Modulus of dynamic elastic ity (Gpa) 22.8 20.5 21.6
σ0 : Uniaxial c ompressive strength, (Mpa) 30 25 27.5

Output data: Min Max Prom

Calculation of RQD: RQD (%) = 44.00 70.00 57.00
Calculation of tunnel quality index: Q = 0.37 0.24 0.30
Calculation of the sonic velocity of the rock mass: Vs (m/s) = 3,066.00 2,873.00 2,969.00
Vs (ft/s) = 10,059.06 9,425.85 9,740.81
Calculation of critical particle peak velocity: Vppc (mm/s) = 402.70 351.10 376.90
Vppc (mm/s) = (0.1 x σc x Vs) / E Vppc (ft/s) = 1.32 1.15 1.24

- 155 -
 DISTANCE FROM BLASTING DAMAGE (ENVELOPE OF DAMAGE)
CMHSA
Field data:
Underground mining : CMHSA Type of explosive : Pulverulent dynamite
Work : Cx2805 Drill diameter : 45 mm
Analysis : Blasting face Type of blasting : Face
Date : 02/04/17 Seismograph : Minimate plus
Zone : North Material to fly : Waste
Filling of joints : Quartz , alteration to the ceiling

Monitoring data: Plus Blaster Plus Blaster Critical speed of rock fracturing
Operational load of the blast :8 :8 kg 17.637 17.6 lb Vppc 377 mm/s 15 in/s
Monitoring distance (d) : 35.3 : 35.3 m 1389.8 1390 in
Vertical : 54.0 : 9.14 mm/s 2.126 0.36 in/s Rock constants
Table 6. Calculation of the influence of damage.

Cross : 87.0 : 9.52 mm/s 3.4252 0.37 in/s α 1.81 1.12

Longitudinal : 17.7 : 13.8 mm/s 0.6969 0.54 in/s K 852.05 230.36
Vectorial : 201.0 : 15.0 mm/s 7.9134 0.59 in/s
PLUS BLASTER
Distances Distances

- 156 -
W W W2 W3 W3 W (kg) W (lb) Vppc Vppc Distances "d2 " Distances Distances Distances Distances Distances Distances Distances
Blast Hole W 2 (lb) "d" (m) "d" (in) Observations
(kg) (lb) (kg) (kg) (lb) PROM PROM (mm/s) (in/s) (m) "d2 " (in) "d3 " (m) "d3 " (in) "d2 " (m) "d2 " (in) "d3 " (m) "d3 " (in)
PROM PROM
1 0.62 1.37 0.78 1.72 0.92 2.03 0.85 1.87 377 15 1.23 48 1.45 57.09 0.51 20.08 0.59 23.23 0.87 34.25 The micro fissures reach up to 0.86 m (33.85 in)
2 0.62 1.37 0.78 1.72 0.92 2.03 0.85 1.87 377 15 1.23 48 1.45 57.09 0.51 20.08 0.59 23.23 0.87 34.25 The micro fissures reach up to 0.86 m (33.85 in)
3 0.99 2.18 0.99 2.18 1.00 2.20 1.00 2.20 377 15 1.56 61 1.57 61.81 0.64 25.20 0.64 25.20 1.1 43.31 The micro fissures reach up to 1.09 m (42.91 in)
4 1.03 2.27 1.01 2.23 1.00 2.20 1.01 2.23 377 15 1.59 63 1.58 62.20 0.65 25.59 0.65 25.59 1.12 44.09 The micro fissures reach up to 1.11 m (43.7 in)
5 1.44 3.17 1.20 2.65 1.06 2.34 1.13 2.49 377 15 1.88 74 1.67 65.75 0.77 30.31 0.68 26.77 1.33 52.36 The micro fissures reach up to 1.31 m (51.57 in)
6 1.03 2.27 1.01 2.23 1.00 2.20 1.01 2.23 377 15 1.59 63 1.58 62.20 0.65 25.59 0.65 25.59 1.12 44.09 The micro fissures reach up to 1.11 m (43.7 in)
7 1.46 3.22 1.21 2.67 1.06 2.34 1.14 2.51 377 15 1.89 74 1.67 65.75 0.78 30.71 0.69 27.17 1.34 52.76 The micro fissures reach up to 1.31 m (51.57 in)
8 1.21 2.67 1.10 1.10 1.03 2.27 1.06 2.34 377 15 1.72 68 1.62 63.78 0.71 27.95 0.66 25.98 1.22 48.03 The micro fissures reach up to 1.2 m (47.24 in)
9 0.62 1.37 0.78 1.72 0.92 2.03 0.85 1.87 377 15 1.23 48 1.45 57.09 0.51 20.08 0.59 23.23 0.87 34.25 The micro fissures reach up to 0.86 m (33.85 in)
10 0.96 2.12 0.98 2.16 0.99 2.18 0.98 2.16 377 15 1.53 60 1.56 61.42 0.63 24.80 0.64 25.20 1.08 42.52 The micro fissures reach up to 1.07 m (42.12 in)
 Table 7. Maximum operating loads per type of explosive.
Explosive Damage distance Q average Cartridges explosive/hole
Dynamite B 7/8” X7 0.2 m 0.06 kg 1.02
Dynamite A 1 1/8” X12 0.6 m 1.6 kg 7.98

Figure 4. Overbreak by 2017.

Figure 5. Overbreak by 2016.

4 RESULTS determined that the factor of overbreak is 12%

from January to August of 2017; while, for 2016 it
4.1 Overbreak in full-face advance was 15.6%. With the control and fulfilment of
Sampling was done in different works of these changes it is expected to reduce the over
exploration face or cross section, preparation and excavation of 9.2% by the end of 2017.
development of the mine, with different qualities Performing the analysis of vibrations as the
of rock mass measuring the real section and main object involves reducing the overbreak rate,
comparing with the design section it was which directly involves extracting and hauling less

- 157 -
 Table 8. Support cost 2016.
Waste Rocks Cost $ 15.6% $/TMS Cost $ 9.2% Savings $
308,121 TMS 1,468,995 4.8 1,333,847 98658

Table 9. Transportation cost 2016.

Waste Rocks Cost $ 15.6% $/TMS Cost $ 9.2% Savings $
308,121 TMS 352,967 1.15 320,494 32,473

Table 10. Underground extraction cost 2016.

Waste Rocks Cost $ 15.6% $/TMS Cost $ 9.2% Savings $
308,121 TMS 152,009 0.49 138,024 13,985

Table 11. Materials cost 2016.

Waste Rocks Total rock break Total $ $/TMS Cost $ 15.6% Cost $ 9.2% Savings $
308,121 TMS 822,458 2,317,336 2.82 868,154 788,154 58,305

waste, reducing the consumption of materials for extraction of waste rock in 27% less, this
maintenance, less energy consumption considered summarises the increase in mineral contribution
the most significant expenses in linear work. from 1800.0 t/day (1632.93 T/day ) to 2000.0 t/day
Below are attached 2 figures that represent the (1814.37 T/day) and achieve with the monthly
Overbreak during 2016 and 2017. The permissible progress from 75% to 96%, because of the
limit of Overbreak is 15% for the quality of the reduction of operational problems on blasting.
rock mass. (60% with an RMR of 21-30). Maintaining and improving these historical data of
The following costs analysis was performed efficiencies in the operation will be because of the
with data from 2016 and costs at 15.6% and 9.2% constant monitoring and improvement on these
in the forecast of overrun – see Tables 8 to 11. works.
In 2016, the cost of sustaining progress was US The implementation of pre-cut blasting
$1,468,995 - see Table 8. technology and decoupling of peripheral charges
The cost average for waste rock transporting with low-explosive (explosive used only in) have
for the year 2016 was 1.15 ($ / TMS), reducing the also made an important contribution.
overbreaking in 9.2% would have $ 32,473
annually. The waste rock is transported in dump 4.1.1 Tests of Hess relied on dynamite as confir-
trucks to a waste dam located 5 km (3.11 mi.) mation of low power explosives
from Balcón mine entrance in Curaubamba – See
The dynamite ‘B’ 45% of 7/8" X 7" (2.22 cm x
Table 9.
The cost average of extraction of waste rock for 17.78 cm) and 1 1/8 "X 8" (2.85 cm x 20.32 cm)
the year 2016 was 0.49 ($ / TMS), reducing the and dynamite ‘A’ 45% of 7/8" X 7" (2.22 cm x
over excavation by 9.2% would have $ 13,985 of 17.78 cm) and 1 1/8 X 12" (2.85 cm x 30.48 cm)
savings – see Table 10. used in the work of production and advance have
The cost for materials concept for 2016 was positive results regarding the control of the
US $ / 2,317,336. This includes materials for excavation. It is adapted to the poor quality of our
rehabilitation work and sustaining liabilities – see work rock. With a VOD that fluctuates between
Table 11. 2,200 m/s and 3,800 m/s, Hess 7 mm and low
In summary, there would be a projection in levels of vibration, (Tests carried out in, Figure 8,
savings of 9.2% of $ 203,421 annually compared 9 and 2,3 & 10, Annex 2).
to 2016. 4.2 Fragmentation analysis
As a result of these changes we have reduced
overbreaking from 18.5% to 12%, which directly The most common method to evaluate
reduces the consumption of shotcrete from 175 to fragmentation after blasting is a simple visual
110 m3/day, mucking times up to 25 minutes, estimation of the rocks on the pile of the broken

- 158 -
 Figure 6. Granulometry of the rock after rock blasting.

material. While this is a way of detecting the most − Materials 2.82 ($ / TMS)
superficial problems, it is a too subjective way to
carry out a program of deep evaluation of blasting. These items make a total of US$ 9.26 / TMS, it
The fragmentation analysis is presented below, means that in 2016 were moved 308,121.0 TMS
using the WipFrag software, which performs this with a total cost of US$ 4,476,998.13. That year
operation through photographic analysis of the the overbreak was estimated at 15.6% (Annex 2)
explosive charge. in linear works. For 2006 the break was
265,019.54 TMS with a total of $ 3,103,378.81
4.2.1 Fragmentation Analysis in a production with 14% overbreak.
work: Tajo 2926 S Mina Lourdes Split Considering the design adjustment in the
Blasting by vibration modelling, there would have
For this working stope, a 15 cm (5.90 in.) scale been savings of 9.2% over break, of $ 286,527.88
was used to estimate the P80. The result obtained for 2016 and $ 148,962.183. With an annual
after this analysis indicates a P80 of 7 cm (2.76 projection of US$ 203,421.00 of savings without
in.). considering the specific consumption of
explosives.
5 CONCLUSIONS The characterisation of the rock mass has a
great importance to carry out the damage analysis
The operating costs (2,016 were as detailed): according to the particle vibration limit levels that
it offers, being able to establish a rock
− Reinforcement 4.80 ($ / TMS) classification according to the speed of wave
− Transport 1.15 ($ / TMS) propagation according to the particularity of each
− Extraction 0.49 ($ / TMS) mineral deposit for the measurement and control

Figure 7. Granulometry results using the Photo analysis systems.

- 159 -
of blasting damage.
The near-field modelling of Devine requires a
greater collection of data (monitoring) to get closer
to the real ground conditions that allow us to
adjust the initial mathematical models and
establish design parameters such as operating
charges and intensity of damage according to real
conditions.
The use of low VOD dynamites and breaking
power such as the dynamite ‘B’ allows the charged
column to be lengthened without having to
increase the operating charge used as a continuous
charge, generating a smooth cut in the contours of
the excavation that avoid maintenance problems,
considering that the biggest problem of blasting is
not the fragmentation but the stability of the
mining labors because of the poor quality mass
rock of the mineral deposit which goes from very
fractured to intensely fractured, making difficult
the mining works.
The delay time is also very important as a
reductive factor of vibration that was already
established, effects of destructive couplings to 25
ms that are counted with the triggering of a drilling
at the same time, avoiding exceeding 0.45 Kg
(0.992lb) / challenged as operating load pattern.
According to the results obtained in the
fragmentation analysis, the P80 in the pit 2926 S
Lourdes was 7 cm (2.76 in.).
The additional cost of advancing to 25% of
over-excavation is $ 189/m ($57.61/ft), so the
forecast by the end of the year with the application
of these controls is expected to have
over-excavation of 9.2% that would be $ 65 /m
($19.82/ft) cost reduction of 65%.

REFERENCES

Sandoval, O. & Guillen H. 2011. Determination of

the width of mining with mechanised support.
Perumin 30th Mining Convention (pp. 2). Arequipa:
Perumin.

Instantel. (2013). Minimate Plus Operator’s Manual.

Ottawa, Ontario, Canada.

Practical blasting manual 5th edition, EXSA.

APPENDICES

Appear on the following pages.

- 160 -
APPENDIX 1. Technical specifications of dynamite ‘B’ and ‘A’.

APPENDIX 2. Tests performed at Dynamite ‘A’ and ‘B’.

Table 12: Results of quality controls, detonation speed, Hess, water resistance and sympathy Dynamite A.

Dynamite Detonation velocity Hess Water Sympathy

specifications 3,600 – 4,000 17 - 20 resistance 2Φ
Dynamite A 45% 3,805 m/s 19.52 mm 2 hr 40 min OK
1 1/8”X12”
Dynamite A 45% 3,917 m/s 19.36 mm 2 hr 40 min OK
1 1/8”X12”

Table 13: Results of quality controls, detonation speed, Hess, water resistance and sympathy Dynamite B.

Dynamite Detonation velocity Hess Water Sympathy

specifications 2,600 – 3,000 13 - 16 resistance 2Φ
Dynamite B 45% 2,821 m/s 13.88 mm Null OK
1 1/8”X12”
Dynamite B 45% 2,756 m/s 13.89 mm Null OK
1 1/8”X7”

Figure 8. Results of Dynamite "A" 45% Detonation Rate measurement (VOD).

- 161 -
 Figure 9. Results of Dynamite "B" 45% Detonation Rate measurement (VOD).

Figure 10. Results of the Hess after the detonation Dynamite 45% "A" and "B".

APPENDIX 3. Nomenclature

− CMHSA: Mining Consorcio Minero

Horizonte SA
− P80: This value means that 80% (by mass) of
the solid matter has a smaller grain size P80.
− Q: Operant charge weight of the explosive
charge that detonates by delayed period.
− Dynamite A: Semexsa 45%
− Dynamite B: Exsablock 45%

- 162 -
- 163 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Rock blast induced pore pressures in quick clay

J. Johansson, S. Gjengedal, J. Park, K.H. Andersen, C. Madshus, & S. Rønning

Norwegian Geotechnical Institute, Oslo and Trondheim, Norway

ABSTRACT: In 2009 rock blasting for a road construction pushed a block of rock into a quick clay area
resulting in a large landslide in Namsos, Norway. This slide also triggered the question of whether
vibrations other than earthquakes can cause quick clay landslides. A first study in 2012 resulted in a
Norwegian standard (NS 8141-3) for a vibration limit and measurement procedure for safer blasting near
quick clay. Recent projects have shown the need to better understand how blast vibrations affect quick
clay and how to deal with it during road and railroad construction. In connection with a road construction,
we measured blast vibrations both in and on rock and on ground surface and at 2 depths in clay as well as
pore pressure in the clay at 5 and 10 m depths. Blast vibration peak value of about 90 mm/s (vector sum)
caused a modest pore pressure increase of less than 3 kPa in the quick clay, which is smaller than the
natural variation over a season. Nevertheless, since pore pressure increase reduces the quick clay strength
it is recommended to stay below set limit of 25 mm/s in the standard if not a more detailed site-specific
analysis show higher vibration values can be allowed.

1 INTRODUCTION for some 12 blasts. Some key observations and

analysis are described below.
Rock blast induced quick clay landslides are
fortunately fairly uncommon in Scandinavia 2 EFFECT OF BLAST VIBRATIONS ON
(Bouchard et al. 2018). However, in 2009 blasting QUICK CLAY SLOPES
triggered an initial slide by pushing a block of rock
into the quick clay resulting in a large landslide There is still a lack of knowledge about the
(Saygl et al. 2017). This slide also re-initiated the mechanisms for how blast vibrations can trigger
discussion of whether blast vibrations can trigger landslides. For some landslides it is clear large
landslides in sensitive soils. In a follow up vibrations contributed to the slide, e.g. the Fröland
research study in 2010-2012 a vibration limit was (Bjurström 1982) slide in Uddevalla, Sweden, the
suggested and later implemented in the NS 8141- more recent La Romaine slide in Canada
3:2013. (Bouchard et al. 2018) where the slide occurred
However, the current standard does not cover right after the blast. For other landslides where
tunnel blasting nearby slopes. Therefore, to gather vibrations may have been a contributing factor,
more field data to understand better how blasting the slides have occurred some hours after
may cause strength reduction in quick clay due to the blast. A recent example is the landslide
pore pressure build up, an extended monitoring in April 2021 at Steinvika in Tana in
program was executed in connection with a road Eastern Finnmark, Northern Norway, where
construction project. Vibrations were recorded for the slide occurred some 10-15 hours after a
over 20 blasts, and pore pressure was monitored large blast.

- 164 -
 Contrary to, for example, earthquake shaking, Scandinavia, there are several studies on Canadian
blast vibrations decrease rapidly with the distance quick clays which provide useful insights to their
from the blasting area and therefore it is assumed cyclic behaviour (Abdellaziz 2021, Quinn et al.
that the pore pressure build-up and thereby 2012, and Lefebvre 1989). However, transferring
strength reduction occurs locally near the contact experiences has to be done carefully since quick
between rock and clay, closest to the blasting. If clay properties vary from site to site, and even
blast-induced pore pressure cannot dissipate more so between countries.
rapidly enough to rock fractures and/or more
permeable soil layer, often present between rock 3 SITE DESCRIPTION AND
and clay in Norway, a possible mechanism for INSTRUMENTATION
causing a failure may be that pore pressure
propagates to surrounding areas with lower For the construction of a new highway stretch of
stability contributing to a reduction in material the E6 North of Trondheim in central Norway, a
strength and resulting in an accelerating creep small hilly rock outcrop, some 15 m high has been
failure in the slope and possibly to a progressive trenched down with a total of 23 bench blasts as
failure. shown in Figure 1, in April-June 2021. An
Quick clay often has embedded layers of loose extensive blast monitoring program was executed,
silt of varying thickness. It is not understood measuring vibrations in both rock and clay, and
whether a failure starts in the silt layers and pore pore pressures at several locations in the clay.
pressure propagates into the clay, reducing its The ground conditions are relatively well
strength, or if it is the clay between the silt characterised with several rock depth soundings, a
layers which fails due to blast vibrations, or a few soil (CPTU) soundings, and samples were also
combination of both. taken up for laboratory index and shear stress tests
Another issue is the blast induced cyclic strain on the quick clay. The site is an old quartery
rates in the soil are much higher than, for example, infilled bay between a county road to the south and
earthquake vibrations. The effect of these high the hill trenched down to the North. The depth to
strain rates on pore pressure generation and gneiss bedrock increases from a few metres in the
accompanying effects on the stiffness and strength East to some 13-20 metres in the West.
of Scandinavian quick clays are not well Figure 2 shows a plan view of the vibration and
understood due to very few investigations with pore pressure sensor locations. Triaxial geophones
cyclic tests (Johansson 2013) on such clays. Since (Infra V12 and V12 b prod. Sigicom) were used to
earthquakes are more common in Canada than monitor vibrations. Three geophones were

Figure 1. Location of blasts (dark markers) and triaxial vibration sensors (MAxx on ground surface and outcrop,
M0xx in soil or in rock, and MPxx on houses). Numbers (1-15) in rectangular boxes next to the blast symbols
show the sequential blast number.

- 165 -
 Figure 2. Vertical section view of location of vibration and pore pressure sensors.

installed on the concrete blocks placed on line seem to agree with the time delay between
with 3 m and 9 m in between and pushed with an charges. The geometry and mechanical properties
excavator down into the top soil in the ground of the rock and soil do also affect the frequencies
surface, two geophones were installed in clay at in the soil, and further evaluation of this is
3.5 m and 5 m depth from ground surface (below underway. The number of stress cycles in the soil,
dry crust), one on outcrop rock and one in rock at which is important for strength reduction in soil
7.5 m depth. 4 pore pressure meters from Geotech due to pore pressure build up, appears closely
were installed at 2 locations at 5 and 10 m depths. related to the number of charge holes in the blast
The geophones were installed in clay at the desired design.
depth in clay as follows. The geophone was The largest vibrations in rock and soil were
attached to a plastic pipe, inserted in a bore hole recorded for blast number 6. Most of cycles in the
with casing, then pushed half a metre into the clay rock vibrations are around 50 mm/s with one pulse
below the bore hole and finally released from the of 100 mm/s (Figure 3) and dominating frequency
plastic pipe. Similarly, the geophone in rock, was of about 75 Hz, which is within 5% of the time
inserted beneath the clay through a cased hole and delay of 14 ms. The vibrations in clay at 5 m depth
fixed with swelling bentonite pellets. High signal (not shown) are overall similar to those in rock,
coherency between the sensors on outcrop and in only with slightly larger amplitudes of 60 mm/s.
rock gave confidence to the installation method In the clay at 3.5 m depth (Figure 4) the
and the measured vibrations. vibrations have larger vertical amplitudes with
several cycles above 100 mm/s and one pulse
4 BLAST INDUCED VIBRATIONS
over to 200 mm/s with vibration frequency of
The measured blast vibration duration ranges about 75 Hz.
between 0.2 – 1.0 seconds, depending on time Vibrations on the ground surface are to some
delay between charges, number holes, and decks in extent affected by the stiffness and thickness of the
the blast. The vibrations in rock tend to be layer with softer top soil above the clay's dry crust,
dominated by 1-3 frequencies between10 to 150 and show a distinct frequency around 25-30 Hz.
Hz. These frequencies appear to correspond to the Therefore, it appears better to measure vibrations
time delays (10-20 ms) between charges and down in the clay instead of on the soil surface, to
blast rows (70 ms). Vibrations in soil are also identify amplitudes, frequency content and
dominated by a few or sometimes only one duration for correlation with the pore pressure
frequency. One of the dominant frequencies, often measurements.

- 166 -
Figure 3. Vibrations in rock beneath clay at 8 m depth for blast 6.

Figure 4. Vibrations in clay at 3.5 m depth for blast 6.

- 167 -
 7
10
4 ?
pp19035
blast 20:02:50
6.95

6.9

6.85
Pore pressure (Pa)

6.8

6.75

6.7

6.65

6.6
19:30 19:45 20:00 20:15 20:30 20:45 21:00
Date & time Apr 29, 2021

Figure 5. A pore pressure increase of about 2.5 kPa measured in clay at 10 m depth (some 5 m above bedrock)
for blast number 9.

The vibrations measured during the blasts, are pressures in the subsequent blasts. The largest
in general stronger in the soil than they are down pore-pressure increase of 2.5 kPa (shown in
in the rock. More analysis is needed to better Figure 5) was measured for blast number 9 with
understand how the vibrations are transmitted the corresponding vibration values (vector sum
from the rock on the side of and beneath to the soil of top values) in clay of 87 mm/s at 5 m depth
and possible wave amplification in the soil. and 52 mm/s at 3.5 m depth. The largest
Measurement data should be further interpreted vibrations were measured for blast 6 (Figure 4),
accounting for the different vibration directions unfortunately the piezometers were not installed
and frequency content, distance and direction to until blast 8, during which about 1 kPa increase
blast, and geometry and properties of the rock and was observed for vibration level of about
soil. 30 mm/s.
5 BLAST INDUCED PORE PRESSURES Maximum pore pressure occurred some few
minutes after blasting and was reduced to
In general, the larger vibration amplitude and the background pore pressure within 1-2 hours. Due to
more load cycles, the higher pore pressure is built low sampling frequency of the pore pressure its
up in soil. Therefor the blasts closest to the border peak value may have been larger as indicated by
between the boundary between the rock and clay the dotted curve in Figure 5. The time delay
are of most interest for observing pore pressure between the blast and the pore pressure peak hint
increase since they give the largest vibrations in at the pore pressure possibly migrates from an area
the soil. The first 15 blasts caused vibration levels where larger pore pressures are induced by large
above 20 mm/s (vector sum) at the sensor in clay shear strains / stresses in the clay. Numerical
at 5 m depth. For this vibration level some pore calculations show that the largest cyclic shear
pressure is expected to have been built up. Some strains are induced in the zone in the clay up to the
small increases in pore pressure of 0.5-2.5 kPa rock wall while the pore pressure was measured in
have been measured after 4 of the blasts. For the clay at a distance of 5-10 m from the transition
another 3 blasts pore pressure may have increased, between rock and clay (an example is shown in
but the natural variation in pore pressure makes Figure 6). Due to the relatively rapid build-up and
interpretation difficult. Furthermore, some of diffusion of pore pressure, more frequent sampling
the early blasts may have contributed to soil of pore pressure, at least every minute or faster, is
densification making it less prone to build up pore recommended.

- 168 -
 The time it takes to reduce blast-induced pore 6 COMPUTATION OF VIBRATIONS AND
pressure is much faster than what lab tests PORE PRESSURES
indicate. It is likely that the field permeability is
much higher than measured in the laboratory. In general, larger vibration amplitude and more
For example, it is often the case that horizontal load cycles cause larger pore pressure increase in
permeability is higher than vertical permeability. It the soil. To correlate measured vibrations and pore
is possible that coarser material (moraines) that pressures, equivalent linear numerical analyses
has been observed between the quick clay and the have been performed to compute cyclic shear
rock, and cracks in the rock as well, contribute to strain and stresses in the soil with the same
the relatively rapid reduction of pore pressure. approach as in Johansson et al. (2020). The
The pore pressures measured so far are not of numerical analyses justify the simple approach of
concern with respect to slope stability since they estimating cyclic shear strain in the clay by taking
have been for stable slopes with small ground the ratio of the particle velocity and the secant
inclination. However, it is important both shear wave speed. Shear wave speeds were
vibrations and pore pressure are monitored estimated with empirical formulas (L'Heureux
when blasting near less stable slopes, since larger 2018) based on sounding (CPT) values and
pore pressures may be generated by vibrations undrained shear strength values in the clay. The
due to higher static shear stresses mobilised in the nonlinearity of the clay was accounted for by
soil. NGI’s in-house cyclic database (Andersen 2015).
We have assumed here that vibrations Back calculation of pore pressure was performed,
transmitted from rock to clay is the cause of the based on contour diagrams, similarly to the
measured pore pressure increase. It is also possible approach for computing permanent shear strains
that blast gas pressures, propagating through (Johansson et al. 2013). The analyses indicate that
cracks in the rock out into the clay, may contribute moderate pore pressures should be expected to be
to the pore pressure build up. Further evaluation of built up for 10 vibration cycles of 25 and 50 mm/s.
this mechanism should be conducted. However, for vibrations of 100 mm/s, the

Blast hole

Rock
Clay

Figure 6. Maximum shear strain plot from dynamic FE simulations of a rock blast in rock close to soft clay.
A zone with large shear strains marked with a white dashed line delineates a zone where excess pore pres-
sure could be generated.

- 169 -
substantial amount of pore pressure increases Even though the measured vibrations
should be expected, as Table 1 illustrates. superseded the NS 8141-3 vibration limit of 25
The computed pore pressures seem to be of the mm/s with a factor of 4 and more, only small or
correct order of magnitude. However, for a better moderate pore pressure increase of some 3 kPa
comparison, the computed pore pressures with the have been observed in soil some 2-5 m above the
measured ones further numerical analysis of pore bedrock. The pore pressures increase with the
pressure generation and spreading in the soil is cyclic shear strain amplitude in the soil and the
necessary. If the pore pressure is generated at the number of cycles. Cyclic laboratory tests indicate
boundary between the rock and the clay, and then a strong inverse dependence on the vibration
propagate out to the surrounding soil, the pressures frequency. Increasing the loading frequency in the
should likely decrease rapidly with distance from laboratory from 0.1 s to 1 reduces the pore
the generation zone. Thus, the piezometer location pressure build up by a factor of about 2 or even
in relation to this zone may have a large effect on more for the same load amplitude and number of
the observed pore pressures. load cycles. The high frequency of the blast
The computed pore pressures are based on vibrations may explain the relatively low excess
material data from laboratory experiments pore pressures observed.
performed at a load frequency of 1 Hz. The blast The measured pore pressures increase nearly
induced vibrations are in general around 10 Hz or instantaneously due to the blasting and then
higher. Thus, the strain rates are at least an order dissipate relatively quickly to the previous level
of magnitude larger due to blasting compared to within 1-2 hours after the blast. The pore pressures
lab tests. Increasing the lab test frequency from 0.1 so far measured from the blasts are lower than
Hz to 1 Hz reduces pore pressure generation with seasonal pore pressure variations in the area and
at least a factor of 2. It is therefore expected for smaller than measured pore pressures due to
blast induced vibrations at frequencies of 10 Hz pre-injection grouting of the rock ahead of the
and above, the pore pressure generation should tunnelling. The blast vibration induced pore
be even lower. Further exploration with high pressures are therefore of no concern to slope
frequency lab test may further clarify the issue of stability at the locations where it has now been
strain rate effects on pore pressure build up. measured as reported herein. When blasting near
less stable slopes, larger pore pressures may be
Table 1. Computed pore pressures based on cyclic generated by the vibrations due to higher static
loading contour diagrams for different vibration shear stresses mobilised in the soil. The
amplitudes. propagation of pore pressure may lead to an
undesirable strength reduction in parts of a slope
Vibration amplitude which is highly mobilised (high shear stresses in
25 50 100
(mm/s) relation to strength). It is therefore important to
Pore pressure monitor both vibrations and pore pressures closely
increase for 10 2 5 30 at critical locations. If induced pore pressures are
stress cycles (kPa) high, they should be allowed to dissipate before
the next blast is set off. Vibration sensors placed
inside the clay at some depth intervals give a better
7 SUMMARY AND CONCLUSIONS indication of the vibration loads than sensors
A rock blast induced landslide in 2009 in Norway placed on the ground surface. Sensors placed on
re-ignited the question of whether blast vibrations houses may give a factor 3 or lower vibrations
can cause instability in slopes with quick clay. than in the clay, and are not recommended for this
A follow-up research project resulted in type of evaluation.
recommendation to keep blast vibration under a The blast induced pore pressures are most
limit of 25 mm/s, under such conditions. This was likely generated in a boundary zone between rock
implemented in the Norwegian standard NS 8141- and clay closest to the blast area. The pore
3. However, the present issue of the standard only pressures then propagate away from this zone out
covers limited slope and blast geometries, and not, to the surrounding soil. Recent measurements not
for instance, tunnel blasting beneath quick clay described here indicate the propagation speed is of
slopes. Therefore, in 2021 in connection with road the order of 1 m/minute in a typically quick clay,
construction, vibrations and pore pressures in possibly controlled by the soil's hydraulic
quick clay has been measured during both surface conductivity and stiffness. Further investigations
and tunnel blasting. are needed to understand the mechanism for pore

- 170 -
pressure generation and propagation. The clay. Canadian Geotechnical Journal, 26:
hypothesis so far is that pore pressure is caused by 122–131. Canadian Science Publishing.
vibrations in the clay, in a similar way as for doi:10.1139/t89-013.
earthquake soil liquefaction. However, it may also
be possible that blast gas pressures may contribute L’Heureux, J.S. & Long, M. 2017. Relationship
to the pore pressures near tunnel blast sites. between shear-wave velocity and geotechnical
The data set of vibrations and pore pressures parameters for Norwegian Clays. Journal of
Geotechnical and Geoenvironmental
described in this paper and a corresponding set
Engineering, 143(6), 4017013.
from another recent tunnel blasting project will
https://doi.org/10.1061/(asce)gt.1943-
allow for further and more detailed studies to 5606.0001645
better understand how rock blasting affects quick
clay slopes. Johansson, J., Løvholt, F., Andersen, K., Madshus,
C. & Aabøe, R. 2013. Impact of blast vibrations
8 ACKNOWLEDGMENT on the release of quick clay slides. In
Proceedings of 18th international conference on
The permission to publish this material was soil mechanics and geotechnical engineering,
granted by Nye Veier (Norwegian public road ICSMGE, Paris.
company) and is highly appreciated. Contributions
to the work from the contractor Hæhre, from Johansson, J., Park, J., Madshus, C., & Wersäll,
rock engineering specialists at Sweco and C. 2020. Reducing impact of vibrations from
specialist providers of vibration and pore pressure compaction on slope stability. In EURODYN
measurements at Nitroconsult and Geotech are 2020: Proceedings of the XI International
gratefully acknowledged. Conference on Structural Dynamics.

REFERENCES Quinn, T.A.C., Robinson, S. & Brown, M.J. 2012.

High strain rate characterisation of Kaolin and its
Abdellaziz, M., Karray, M., Chekired, M., Delisle, application to Statnamic pile testing. In 9th
M.C., Locat, P., Ledoux, C. & Mompin, R. 2021. International Conference on Testing & Design
Shear modulus and hysteretic damping of Methods for Deep Foundations. Kanazawa,
sensitive eastern Canada clays. Canadian Japan.
Geotechnical Journal, 58: 1118–1134. Canadian
Science Publishing. doi:10.1139/cgj-2020-0254. Saygl, G., Madshus, C., Nordal, S. & Engin, H.K.
2017. Numerical modelling of a blast-induced
Andersen, K.H. 2015. Cyclic soil parameters for landslide. In Geotechnical Frontiers 2017.
offshore foundation design. The 3rd ISSMGE American Society of Civil Engineers.
McClelland Lecture. Frontiers in Offshore https://doi.org/10.1061/9780784480458.040
Geotechnics III, ISFOG'2015, Meyer (Ed).
Taylor & Francis Group, London, ISBN: 978-1- Yang, J.H., Lu, W.B., Jiang, Q.H., Yao, C. & Zhou,
138-02848-7. Proc., 5-82. Revised version in: C.B. 2016. Frequency comparison of
http://www.issmge.org/committees/technical- blast-induced vibration per delay for the full-face
committees/applications/offshore and click on millisecond delay blasting in underground
‘Additional Information’. opening excavation. Tunnelling and
Underground Space Technology, 51: 189–201.
Bjurstrom, G. 1982. The landslide at Froland, June 5, Elsevier BV. doi:10.1016/j.tust.2015.10.036.
1973. In Symposium on Slopes on Soft Clays,
Swedish Geotechnical Institute Report No. 17,
Linkoping, 113-126.

Bouchard, S., L’Heureux, J.S., Johansson,

J., Leroueil, S. & LeBoeuf, D. 2018. Blasting
induced landslides in sensitive clays. In
Landslides and Engineered Slopes. Experience,
Theory and Practice. edited by S. Aversa,
L. Cascini, L. Picarelli, and C. Scavia. CRC
Press. pp. 497–504.

Lefebvre, G., LeBoeuf, D. & Demers, B. 1989.

Stability threshold for cyclic loading of saturated

- 171 -
 4. Construction, mining and
quarrying (blasting)

- 172 -
- 173 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

A study of the fragmentation of rock masses by explosives -

Application to four Belgian case studies

J-P. Tshibangu & N. Gonze

Mining Engineering Department, Faculty of Engineering, UMONS-University of Mons, Mons, Belgium

P. Van Landschoot & V. Vanoverbeke

Sagrex, Braine-l’Alleud, Belgium

L. Rosini
Eurovia, Anderlecht, Belgium

ABSTRACT: In Belgium, many quarries are actively producing lime, cement and various crushed rocks.
This study will focus on two types of deposits; these are limestone and porphyry. We are interested in
optimising the blasting operations with respect to the geological material. We characterised the rock mass
mainly in terms of the Geological Strength Index or GSI (Hoek & Brown 1997). The areas under study
were included in 3D models of quarries. We determined the Powder Factor of the analysed shootings and
took images of the mucking piles prior to and during the loading operations. Among the interesting results
we got, the most important is the linear relationships between the GSI and the powder factor, especially
for limestone. We consider that this is due to the variability in mechanical properties because of
weathering process. Porphyry deposits are more massive and less affected by this phenomenon.

1 CONTEXT AND GEOLOGY them, limestone and porphyry are very

representative in terms of volumes. These
Belgium, and especially Wallonia, has a long geological materials are both hard rocks in
history of mining activities that made the country a general that need explosive for fragmentation;
major player in the coal and steel industry in the but they exhibit different geomechanical
past decades (Hasquin H. 1999), mainly during the characteristics. Porphyry is the hardest with a
18th and 19th centuries (1774 – 1847). In fact, very brittle failure and a strong abrasiveness as it
about 5000 years ago, started the exploitation contains free silica particles. Limestone on the
of flints in the country; the Spiennes other side, exhibits a bit less brittleness and
archeological site in the neighbourhood of Mons especially a very low abrasiveness (Descamps et
was exploited between 4,350 and 2,300 years al. 2013).
before our era (https://www.minesdespiennes.org). The deposits of those resources are composed
The development of the coal industry as well as of the rock masses and are characterised by sets of
the one of steel are both documented from the 12th discontinuities like joints, faults and bedding
century (Auquière 1977). planes. It is then for us worthwhile to compare the
Nowadays, most of the extractive activities deal behaviour of those materials with respect to the
with the exploitation of quarries and, among use of explosive.

- 174 -
 Figure 1. Location of the four sites under study in Belgium: 1 = Lemay; 2 = Lessines;
3 = Quenast; 4 = Rochefort.

For limestone, this study will focus mainly on For porphyry, on the other hand, most of
the Lemay’s quarry located in Tournai (see Figure measurements are performed in the Quenast
1). The Tournai’s basin is the most important quarry (see Figure 1) but we will also show
outcrop of limestone in Belgium, as it produces some results gathered in the Lessines’s quarry
about 20 million tons per annum for cement and for the sake of comparison. The porphyry deposits
crushed rocks. According to the regional geology are of intrusive type for which outcrops occur
(Hennebert & Doremus 1997), the Tournai’s in Belgium in the forms of sill layers (Region
limestone deposit is composed of six members that of Lessines) or massive intrusion (Quenast).
belong to the formations of Tournai and Antoing; The deposit mined out in Quenast for instance
these are Calonne, Vaulx and Chercq, Pont-a-Rieu, dates from 433 million years ago and it is a
Providence, Allain and Crampon. Those members micro-diorite composed mainly with quartz,
are composed of about 80 % of calcium carbonate, hornblende, feldspars and chlorite. The main use
argillaceous materials for about 20% and a for this material in Belgium is in the upper layer
proportion of silica. This makes the composition of road structures or the support material for
of this natural material almost suitable for cement rail tracks.
production without any additive. However, some
members can contain high proportion of silica that2 2 ROCK MASS MODELLING
induce some issues for the milling equipments; in
this case, the material is more suitable for crushed 2.1 Quarry model and location of experiments
rock. The main challenge in producing crushed
rocks is the process of weathering typical of Managers of quarries collect and organise various
limestones. In fact, the caving process (karst) information relating to the knowledge of the
affects the shallow geological formations and this deposit and its evolution during ‘mining’
could affect the quality of the final product. operations:

- 175 -
 Figure 2. Surface models of the Lemay quarry: (a) Location of blasted zones; (b) an example of pre and
post 3D scan of bench faces.

- Drillholes to characterise the chemical Lambert 72 coordinate system and we imported

composition if necessary and physical them in the project to cut the volumetric
properties of the materials geological model of the quarries. The construction
- Production data in the course of time of these 3D models can have several objectives:
- Evolving topography of the open pit and managing the evolution of the quarry,
related maps. identifying different geotechnical zones
(see further considerations about GSI) or locating
Based on this information and on the regional as accurately as possible the position of
interpretation of the geological map, a data base is blasted zones.
created in relation to the most important features Dealing with blast zones, the position of
of the material mined out. This is achieved by use each blasthole is measured with a Differential
of a dedicated ‘design and mining planning’ Global Positioning System (DGPS). Moreover,
software: GEOVIA GEMS. From the database, 3D laser surveys of the bench faces are
3D geological models of the deposits are built; carried out prior to each blasting and after
this allows for instance the assessment of the the complete mucking operations (Figure 2).
volume and tonnage corresponding to specific These survey measurements provide accurate
geological formations in the boundaries of information on the blasting patterns and
the quarrying permit. We recovered topographical allows the calculation of the total volume
digital files from the companies in the Belgian mined out.

Figure 3. An example of a bench face image processing for rock mass characterisation.

- 176 -
 Figure 4. The Geological Strength Index (Hoek & Brown 1997).

2.2 Rock mass quality depicts a typical interpreted image in the limestone
quarry.
The rock mass qualification was carried out The station position coordinates for pictures
in-field by surveying, taking pictures and are defined by use of a DGPS device; this allowed
describing the geometry of the benches. Figure 3 identifying the positions of joints and other

- 177 -
 Figure 5. Creation of GSI virtual drillholes from the blast planning.

features. As a result, for this case, we identified face to blast. First, the position of the blastholes
two main sets of sub-vertical fractures that we collars are surveyed, then, virtual drillholes tracks
named S1 and S2, to which a third set S3 are created in terms of GSI. This means that this
representing the stratification is added. The parameter can vary along the height of the bench
characteristics of identified sets of joints are as (Figure 5a). Figure 5b depicts one example of GSI
follows: N30 °E for S1 and N100-110 °E for S2 virtual-drillholes for one face to be blasted.
with a mean dipping of about 90°. These additional data added to the database of
Besides the identification of sets of the whole 3D model of the quarry.
discontinuities, the processing of data led to GSI drillholes are stored in the database in
deriving quality indexes. In this paper, we mainly terms of points information along a line. From
focus on the Geological Strength Index or GSI as that, a block model is created to assess the
described on Figure 4 (Hoek & Brown 1997). variability of the mechanical quality of the rock
Surveying the whole quarry and using GSI can material in the rock mass for blasting. One
lead to division into constant quality parts, or fundamental parameter for the block modelling is
geotechnical zones that can be managed the size of blocks. The choice we made for the
accordingly for blasting (Tshibangu et al. 2007). Quenast quarry was 2 m x 2 m x 2 m, while in the
This study will focus mainly on the blasted zones Lemay’s quarry it was 5 m × 5 m × 5 m. The
of concern. The working methodology is to try interpolation was performed by use of isotropic
reproducing as many details as possible on the (porphyry) or anisotropic (limestone) reverse

- 178 -
 Figure 6. Block model of the 15 March 2015 blast in the Quenast quarry.

distance method on the whole quarry model. A quarries. For each of them, we collected main
typical example of block modelling is shown in parameters, these are:
Figure 6 for the Quenast quarry.
- Number of blastholes and their position
3 BLASTING OUTPUT AND MUCKING - Drilling pattern
PILE - Length, diameter, and inclination of the
blasthole
3.1 Blast parameters - Preliminary observations of the shot:
presence of water, faults, cavities and other
Various shots were analysed in the targeted anomalies

Figure 7. Comparison between blasted tonnages estimated based on shooting pattern and the one calculated
based on 3D surveys.

- 179 -
- Quality and quantity of explosives and their The big challenge with this type of software is
distribution in the hole to collect relevant and good quality pictures of the
- Presence and type of intermediate stemming mucking piles. For all blasts analysed, several
- Type of initiation system (detonating cord, pictures were taken from different viewpoints and,
Nonel tubes, electronic detonato). when possible, at different loading periods. We
placed a scaling object on the pile for each photo.
We need this information for the calculation of As this was an exploratory study, different objects
the Powder Factor that we use to characterise the were used: a helmet, a graduated staff, and a
strength of explosives for a given blast. The 1-meter square frame.
amount of explosive is expressed in terms of This methodology allowed collecting pictures
Equivalent TNT, while the volume or tonnage of of the superficial part and the centre of the muck
blasted rocks is assessed by surveying the face piles (Figure 8). However, it was impossible to
prior and after the shot. take pictures continuously during all the mucking
We compare on Figure 7 the tonnage estimated process because of security and time reasons.
by two approaches: on one hand it is calculated We achieved the image processing by the Split
from the blasting pattern, and on the other hand Desktop software in three steps:
the one computed from the 3D survey. We noticed - The first step consists of an automated
in general that the discrepancy was significant delineation of the rock fragments by the
when residual debris of rocks were still present software. During this step, the software relies
during the post-survey of the face; in this case, we on the contrast differences in the image to
use the value calculated from the shooting pattern. delineate the different blocks (Figure 8 (b)).
However, this automatic delineation is not of
3.2 Mucking pile and block size sufficient quality to directly calculate the size
distribution curve. Indeed, many blocks are
Two types of results can describe the main output over-segmented (Figure 8 (b), point 2) and
of blast operations: the shape of the mucking pile shadows present on some blocks are
and the block size. The shape of the pile will lead considered as contour areas composed of
the choice of loading machineries; in Belgium fines (Figure 8 (b), point 1). Finally, the
most quarries use front wheel loaders that need a software analyses the whole image even if the
low-profile muck pile. In this study we focus on background is made of the benches and the
the block distribution that characterises the sky (Figure 8 (b), point 3).
fragmentation process. - The second step is a manual correction of the
Implementing fragmentation analysis is automatic delineation. During this step, the
complex and delicate in the context of mining user must define the area of interest of the
operations. Several methods can be used: picture to consider only the muck pile. It is
empirical methods, screening, image analysis. This then necessary to eliminate any inadequate
study used an image processing software (Split contours and to create missing ones. The
Desktop) in its freeware version. The methodology result of this step is presented in Figure 8 (c).
mixed automatic and manual treatments to analyse - Finally, the third step consists of defining the
fragment sizes. 'reference scale' from the known object on the

Figure 8. Split desktop image processing: (a) imported picture, (b) result given by the automated process [1 = shadow
problem; 2 = over delineation of a block; 3 = delineation of benches], (c) result from the manual processing.

- 180 -
 Figure 9. Pre and post manual processing cumulative curves.

picture. The software can compute the representativeness of the results with respect to
fragment sizes and establish the cumulative the pile. For this purpose, we analysed a part
size curve with this scale (Figure 9). Figure 9 of a pile and compared the granulometry obtained
shows the fragment size curves obtained with the one of the whole pile. Figure 10 (a)
before and after manual processing. This compares the two curves obtained. The difference
figure highlights the importance of manual is significant for the fine particles. Nevertheless,
corrections because the mean size is the two curves are comparable in terms of
underestimated when using the automated ‘medium’ size.
procedure. The second important result is the comparison
between photos taken before and during loading
Thanks to multiple pictures collected for each and their cumulative size curve. The curves in
muck pile, we analysed some parameters of the Figure 10 (b) indicate that the difference between
shooting. Indeed, we addressed the issue of the the two analyses is negligible.

Figure 10. Fragment size curves comparison: (a) analysis of the whole pile and part of it (b) analysis of the fresh
pile and its centre photographed during loading.

- 181 -
Figure 11. Relationship between GSI and powder factor (a) in limestone quarries and (b) in porphyry quarries.

- 182 -
4 RESULTS AND DISCUSSION weathering process. Porphyry deposits are more
massive and less affected by the variability in
The main objective of the study was to try linking mechanical properties.
the quality of rock mass and the use of explosives Such relationships can allow a more accurate
for its fragmentation. Based on the experience of assessment of the explosive charge that complies
miners, we consider that the fragment size with the rock mass for a given objective in terms
obtained is satisfactory for the current practice of granulometry or fragments size. In fact, the
(empirical approach). We then analysed the general trend is the decrease of charge with quality
possible correlation between the GSI of the rock mass; an approach that deserves to
characterising the rock mass and the powder repeat in more cases to improve the model.
factor. The graphs in Figure 11 present the
relationships between these two parameters. REFERENCES
Figure 11(a) shows the trend in the Lemay’s
limestone quarry on which the result from the Auquière G. 1977. Les industries extractives en
Rochefort quarry was superimposed. Figure 11(b) Belgique : leur passé, leur présent et leur avenir.
compares the results from the two porphyric
quarries. Descamps F., Da Silva N., Regnard S. & Tshibangu
For all results gathered, the general trend is the J-P. 2013. A database to manage rock mechanics
properties and assess formation abrasiveness for
decrease of the charge with the quality of the rock
drilling. Eurock 2013, Wroclaw, Poland.
mass. Interestingly, this trend appears to be
stronger in the case of limestone quarries. Indeed, Hasquin, H. 1999. La Wallonie son histoire, Luc Pire,
the correlation coefficient of the linear regressions Bruxelles, p. 283.
for those quarries are much higher than for the
porphyry quarries. For this last type of material, Hennebert, M. & Doremus, P. 1997. Carte géologique
the scattering is very high. Remember when de Wallonie-Feuille Antoing-Leuze (37/7-8),
analysing the quality of the rock masses that we Ministère de la Région Wallonne.
emphasised the higher variability in limestone
quarry because of weathering process. In porphyry Hoek, E. & Brown, E.T. 1997. Practical estimates of
deposits are more massive and less affected by the rock mass strength. International Journal Rock
variability in mechanical properties. Mechanics Mining Science, 34, 1165-1186.

5 CONCLUSION

To understand the blasting process in some

Belgian quarries we carried out this study in two
types of applications in relation to the mined
material: limestone and porphyry. The approach
involved the collection of blasting parameters
prior to the shooting; this led to the assessment of
the explosive energy expressed in equivalent
TNT. Connecting this energy to the amount of
fragmented rocks gave the Powder Factor.
After blasting, we took some pictures of the
muck pile and processed the images in a hybrid
way: automatic processing by a freeware software,
and manual correction to get the shape of the size
distribution curve. The desired approach was to
capture many images for each blast to show the
evolution of the mucking pile; however, this was
not always possible because of safety reasons and
lack of time.
Among the interesting results we got, the most
important is the linear relationships between the
GSI and the powder factor, especially for
limestone. In fact, this material shows higher
variability in mechanical properties because of

- 183 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Rock mass characterisation and ore grade assessment through drill

monitoring for open pit applications

J. Navarro & P. Couceiro

MAXAM

J.A. Sanchidrián & P. Segarra

Universidad Politécnica de Madrid

T. Seidl & P. Hartlieb

Montanuniveristaet Leoben

P. Schimek
VA Erzberg GmbHAbstract

ABSTRACT: This paper describes a new methodology to assess rock mass quality from drill-monitoring
data to guide blasting in open pit operations. Two rock description indexes are derived directly from
‘Measurement While Drilling’ (MWD) data. Principal component analysis has been used to combine
MWD information from two drill rigs; corrections of the MWD parameters to minimise external
influences other than the rock mass have been applied. The first index is a structural factor that classifies
the rock mass condition in three classes (massive, fractured and heavily fractured). From it, a ‘Structural
Block model’ has been developed to simplify the recognition of rock classes. Video recording of the inner
wall of 207 blastholes has been used to calibrate the results obtained. The second index is a ‘Strength-
Grade factor’, based on the combination of MWD parameters, that has been assessed from the analysis of
the rock type description and strength properties from geology reports, assaying of drilling chips
(ore/waste identification) and 3D unmanned aerial vehicle (UAV) reconstructions of the post-blast bench
face. Data from 302 blastholes, comprised in 26 blasts, have been used for this analysis. From the results,
four categories have been identified: soft-waste, hard-waste, transition zone and hard-ore. The model
determines zones of soft and hard waste rock (schisted sandstone and limestone, respectively), and hard
ore zones (siderite rock type). The model has been validated with new data from 53 blasts (647
blastholes). Finally, the Structural Block model has been combined with the Strength-Grade factor to
create the ‘X-Rock model’. This model, exclusively obtained from drill monitoring data, can provide an
automatic assessment of rock structure, strength, and waste/ore identification.

1 INTRODUCTION intersected by discontinuities evaluated with

indexes such as Geological Strength Index (GSI,
Rock mass is a combination of materials with Hoek 1994), Rock Quality Designation (RQD),
different mechanical properties (strength, (Deere & Miler 1966), Rock Mass Rating (RMR),
deformation, and stability characteristics), (Bieniawski 1995) and/or Q-Barton (Barton et al.

- 184 -
1974). The in-situ structural condition modifies the been reported. However, none of these methods
strength of the intact rock; the existence of or systems cover the combined analysis and
cavities, discontinuities, or fractured zones representation of both mechanical and structural
changes the response of the rock mass upon properties together. In addition, to the authors’
blasting. In blasting, rock density, uniaxial knowledge, few of them have been applied to blast
compressive strength, Young’s Modulus, design and validated in production environments.
orientation of discontinuities with respect to the This paper develops a new methodology for a
free face of the blast and the spacing between them sound rock characterisation based on two new
have been consistently employed to rate the indexes from MWD records: the first one classifies
blastability of the rock. Lilly’s blastability index the structural condition of the rock from the
(Lilly 1986, 1992) has been used to predict the variability of the MWD parameters and a
median size through the modified Kuznetsov discontinuity factor determined using Principal
equation in the Kuz-Ram model (Kuznetsov 1973, Components Analysis (PCA); the second
Cunningham 1987, 2005), and has also been used represents a combination of the strength properties
recently to describe rock mass characteristics by and iron content of the rock, based on the
Sanchidrián and Ouchterlony (2017) in the combination of MWD parameters. Before being
xp-fragmentation distribution-free model. A used for the analysis, a thorough correction of the
detailed rock mass characterisation, based upon MWD parameters has been applied to minimise
the combination of both mechanical, structural and external influences other than the rock mass.
chemical properties, is thus one of the most The analysis is applied to data from the Erzberg
important requirements to control the ore to be Mine, a large iron open pit mine in Austria. The
mined and to improve blasting results by adapting Structural factor classifies the rock condition in
the explosive energy distribution to the rock three classes: massive, fractured and heavily
condition so as to optimise the performance of fractured. This classification has been validated
downstream stages such as digging, hauling and and calibrated from video recording of the inner
comminution. wall of 207 blastholes. From it, a Structural
Since rock characteristics have an important Block model has been developed to simplify the
influence on the drilling response, technologies automatic recognition of rock trends.
based on monitoring the performance of drill rigs The Strength-Grade factor has been assessed
by measuring drill parameters (commonly known from a combined analysis of the rock type
as Measurement While Drilling, MWD) should be description and UCS from geology reports, assays
able to assess changes in the rock mass with high of the drill cuttings (ore/waste identification)
resolution (Navarro 2019). Substantial research and 3D unmanned aerial vehicle (UAV)
efforts aimed to correlate the MWD technology reconstructions of the post-blast bench face. The
with the geological/mechanical interpretation of factor ranks the rock into four categories:
the rock mass (Teale 1965, Scoble et al.1989; soft-waste, hard-waste, transition zone and
Peck 1989, Schunnesson et al. 2012, Leung & hard-ore. MWD signals from 302 holes in 26
Scheding 2015, Kahraman et al. 2016), on blasts, assays of the drill cuttings for 207 of the
one side, and with the structural condition 302 blastholes and geology reports of each blast,
(Schunnesson 1997, Peng et al. 2005, Tang 2006, have been used for this analysis. The Strength-
Ghosh et al. 2018, Navarro et al. 2019, van Eldert Grade factor has been validated with 53 new blasts
et al. 2020), on the other, (among others) have comparing the assays of 647 new holes with the

Table 1. Data information.

Assays of
In-hole
No. No. of MWD UAV drill
No. videos
Data set blasts/drill holes (No. (No. cuttings Rock type
blast (No.
rig drilled holes) photos) (No. of
holes)
holes)
Model & 15(D1)/11( Sch Sand, Lim &
26 321 302 207 4599 207
calibration D2) Sid
36(D1)/17( Sch Sand, Lim &
Validation 53 723 647 - - 647
D2) Sid

* Sch Sand= schisted sandstone; Lim=limestone; Sid=Siderite

- 185 -
associated Strength-Grade factor value. the overburden by a partly schisted sandstone.
Finally, the Structural Block model has been Average UCS values assigned from geological
combined with the Strength-Grade factor to create reports, and the ore/waste classification,
the X-Rock model, an overall novel rock factor, respectively, are siderite (125 MPa, ore), limestone
that, exclusively based on drill monitoring data, (waste, 125 MPa), and schisted sandstone (waste,
can provide a blastability assessment based on 40 MPa).
the rock structure, strength and waste/ore The rock mass is excavated by bench blasting
identification. using one or two rows of inclined holes with a
length of 15 to 30 m (49.21 to 98.42 ft). Two
2 DATA OVERVIEW SmartROC D65 DTH fully mechanised drill rigs
manufactured by Epiroc are used for the drilling.
The Erzberg mine is the largest open pit iron ore Positioning of the holes is controlled by the
mine in middle Europe and the largest deposit of operator but drilling and logging of drill
siderite in the world. It is located in the central- parameters is done automatically. Drilling is done
western part of Austria. The main iron mineral is a using extension tubes of 6 m (19.69 ft) length and
magnesium rich variety of siderite with an iron a bit diameter of 6 in (152 mm). A 4.16 revision
content between 20 and 42 %. The main waste control system was installed in the rigs and a
material within the ore bearing formation are sample interval of 0.05 m (0.164 ft) was set to
different types of limestone. The floor of the monitor the information. This makes up to 460
deposit is formed by a volcanic silicate rock and sample points per blasthole considering an average

Figure 1. Structural conditions identification from borehole video record.

- 186 -
blasthole length of 23 m (75.46 ft). sample materials. This determines an average
The drill parameters recorded while drilling measurement per hole of Fe content and other
are the following (acronyms and units for each components such as SiO2, CaO, Al2O3, MgO,
parameter in brackets, (Navarro et al. 2021): feed Mn, Na2O, K2O, S and P. In general, a rock with
pressure (FP, bar), percussive pressure (PP, bar), Fe content above 20% is considered ore and
rotation pressure (RP, bar), penetration rate (PR, below waste.
m/min), air pressure (AP, bar), hole depth
(HD, m). 3 DATA ANALYSIS
Table 1 describes the available data for MWD,
borehole videos, UAV flights, assaying, and the One general problem with MWD data analysis is
rock type/s identified from the geological reports that the logged response is affected by the rock
used to develop the Structural model and the mass conditions, the drill rig control system and
Strength-Grade factor models and the data set used external influences different than the rock such as
to validate the later. the calibration of the monitoring sensors, the
In-borehole video recording was carried out by hole length and/or the drill rig performance
a digital camera manufactured by Kummert, (Schunnesson 1998). Together, these add
composed of a front lens, illuminated by led lights. uncertainty to the data that must be corrected to
The data acquisition system was connected to the highlight changes in the parameters that depend on
camera by an insulated cable. This was rolled in a the rock properties. This process comprises:
cable wheel with a distance counter, so the image
in the video is always linked to a depth. Video 3.1 Filtering of unrealistic values
recordings of the borehole wall in front of the
The empirical probability distribution of each
probe was done while lowering it. The extensive
MWD parameter is built from the complete data
hole video recording program conducted in field
set values (26 blasts, comprising 302 blastholes
tests allows to identify different structural rock
sampled every 5 cm, see Table 1). Values within
conditions represented in Figure 1. Normally, a
the 99% confidence interval are retained for the
weakness area (Figure 1d) is a combination of
analysis (Navarro et al. 2018, 2019, 2021).
several small/medium size discontinues and/or
small cavities. It may also be related to lithology 3.2 Removal of systematic peaks related to the
intrusions of softer rock. A heavily broken area parameters response when a new rod is
(Figure 1g) is a low competent zone that includes a added
combination of medium/large size discontinues
and/or medium/large cavities. When the drill reaches the end of the feeder, a new
UAV images correspond to the post-blast bench rod must be added to continue the drilling. During
face after removing the muck-pile material. This this procedure, percussive pressure, feed pressure
face of the bench is closer to the blastholes so it and rotation pressure parameters are shut down
can give a more accurate representation of the rock (Navarro et al. 2019, 2021). The logging system
drilled. Photos were taken from a DJI Matrice 200 starts recording values again when the bit exceeds
drone. Image processing have been done through the last length measured. This generates systematic
Pix4D (Pix4D 2020). Post-processing of the point drops (see drops in Figure 2a left graph, at every 6
cloud and 3D representation have been done with m, 19.69 ft)) that do not reflect any information of
Matlab (Matab, 2020). the rock conditions, and that must be filtered out in
Assaying is done over drill cuts gathered a post analysis. Figure 2a (right graph) shows the
automatically while drilling. For that, cuttings are filtered signal from Figure 2a (left graph), in
flushed out of the borehole with compressed air which all peaks associated to rod changes have
and transported away with a suction device. The been removed. Noted that the peak at about 3 m
collar of the blasthole is closed with a suction (9.84 ft) is not removed by this filtering, since it
hood and the discharged cuttings are introduced in does not correspond to rod additions, but to a
a cyclone. Samples are cyclically collected from possible rock disturbance.
the cyclone at a preset depth interval to cover
the entire blasthole length. A sample mass of 3.3 Correction of the hole depth influence:
approximately 1 - 2 kg (2.2 – 4.4 lb) is collected in
a container sealed to the ejection point of the Systematic variations generated by the drilling
cyclone. X-ray Fluorescence Spectrometry was system and external influences other than the rock
used to identify the different components of the can be recognised and removed by averaging, for a

- 187 -
 Figure 2. a) Filtering of systematic peaks due to the addition of a new rod: Raw signal (left); Filtered signal
(right). b) Average signals for Av.PR, Av.PP, Av.FP and Av.RP with depth. The shaded colours correspond to
the average signals for drill rigs D1 and D2, respectively. HD is hole depth.

large amount of data, the logged response for more details). The variability of penetration
(Schunnesson 1998, Hjelme 2010, Ghosh et al. rate and rotation pressure have been combined in a
2018, Navarro et al. 2018, 2019, 2021). Figure 2b discontinuity index (DI) as:
shows the average signals of penetration rate
(Av.PR), percussive pressure (Av.PP), feed ��������
1 𝑃𝑃𝑃𝑃𝑣𝑣𝑣𝑣𝑣𝑣 ,𝑖𝑖 − 𝑃𝑃𝑃𝑃
2
1 𝑅𝑅𝑅𝑅𝑣𝑣𝑣𝑣𝑣𝑣 ,𝑖𝑖 − �������
𝑅𝑅𝑅𝑅𝑣𝑣𝑣𝑣𝑣𝑣
2
𝑣𝑣𝑣𝑣𝑣𝑣
pressure (Av.FP) and rotation pressure (Av.RP) 𝐷𝐷𝐷𝐷𝑖𝑖 = � � � + � � ,
2 𝑠𝑠𝑠𝑠𝑠𝑠(𝑃𝑃𝑃𝑃𝑣𝑣𝑣𝑣𝑣𝑣 ) 2 𝑠𝑠𝑠𝑠𝑠𝑠(𝑅𝑅𝑅𝑅𝑣𝑣𝑣𝑣𝑣𝑣 )
calculated at every 0.5 m hole length for the 302
blastholes analysed. The calculation is shown for with 𝑖𝑖 = 1,2, … , 𝐿𝐿 (1)
the two drill rigs separately (represented with
different shades, to see whether there is any other where, PRvar,i and RPvar,i are penetration rate
systematic variation caused by the calibration or variability and rotation pressure variability,
set-up of the rig sensors. From the four graphs, respectively, at each sample number i, L is the
rotation pressure (Av.RP) shows a well-defined number of sample points of the PRvar and RPvar
increase with hole depth (HD) for the two rigs. It signals and , , and
is noticeable that significant jumps occur in the
signal of this parameter after adding a new rod, at are mean and standard deviation of
6, 12, 18 and 24 m (19.68, 39.37, 59.05 and 78.74 the penetration rate variability and rotation
ft), see dashed lines, bottom right graph, Figure pressure variability.
2b. This can be caused by energy loses in the Principal Component Analysis tools (PCA)
couplings after installing a new rod or by the have been used to correlate PR, PP, FP and
additional difficulty to rotate the rod as the weight normalised RP (RPNorm) parameters, plus the
increases down the hole. The normalisation of the calculated PRvar, RPvar and the DI. Figure 3a
hole length influence in the rotation pressure represents the loading plot (influence of each
parameter is calculated for each rig separately as parameter in the analysis) for the first and second
there are two different trends with different ranges principal components (PC1 and PC2,
of values. respectively). The plane generated by the two
components explains the 66.02 % (47.47 %,
4 STRUCTURAL MODEL component 1 and 18.55 %, component 2) of the
total variance. The first component explains a
Schunnesson (1997) and Navarro et al. (2021) for larger variability in the data and the relation
rotary-percussive drilling and Ghosh et al. (2018) between parameters explains the drilling response
and Navarro et al. (2019) for Wassara ITH to the rock. According to Figure 3a, the first
hammer, demonstrated an increase in the component is dominated by the fracturing
fluctuation of the penetration rate and rotation parameters (PRvar, PRvar and DI) to the right and
pressure when drilling through a fractured zone. the percussive pressure and feed pressure to the
These variations are highlighted and calculated as left, showing a negative correlation between them.
the sum of the residuals over a defined interval This behaviour of the parameters is in line with the
along the borehole (see Navarro et al. 2019, 2021 physical drilling response explained in the

- 188 -
 Figure 3. a) Loading plot of PC1 and PC2 for the Structural factor calculation; b) Loading plot of PC1 and PC2
for the Strength-Grade factor calculation. PP (percussive pressure), FP (feed pressure), RPNorm (normalised
rotation pressure), PRvar (penetration rate variability), RPvar (rotation rate variability).

literature (Schunnesson 1998, Peng et al. 2005, (Figure 1c), medium size discontinuities
Ghosh et al. 2018, Navarro et al. 2018, Navarro et (Figure 1e) and/or a small size cavity
al. 2021). (Figure 1f).
Values obtained from PC1 have been compared - Heavily-Fractured zone: zone of heavily
with video records of the inner wall of the broken rock mass (Figure 1g), large size
blastholes, aiming to associate the structural rock discontinuities (Figure 1h) and/or medium or
classes described in Figure 1 to the different large cavities (Figure 1i).
values of PC1. Based on the videos, three
structural classes have been defined: To rank the three structural classes in terms of
the PC1, the probability distribution functions for
- Massive zone: zone composed by massive each structural class have been calculated. The
rock (Figure 1a), small and/or isolated point where two adjacent curves cross indicates
discontinuities and/or fractures filled with the threshold between two structural zones. Figure
rock of similar properties (Figure 1b). 4 represent an example of the Structural factor
- Fractured zone: blocky zone composed by a where images are located at the corresponding
weakness area (Figure 1d), changes in the depths, with the definition of the specific structural
lithology by intrusions of softer material class identified. It is apparent that the Structural

Figure 4. Example of the Structural factor from video records of the inner wall and division of the Structural
factor in zones based on abrupt changes in mean value of the signal for Blast B14 (Row 1, Hole 3).

- 189 -
factor accurately identifies the different structural index since they are normally caused by the local
classes. structural condition of the rock. The same
It is often difficult to interpret Structural factor procedure has been followed to divide the signal
graphs because of the existence of many small based on abrupt changes of the mean value;
intervals of different rock classes, that complicates however, the purpose here is to find dominant
the automatic recognition of rock condition trends zones in the signal that may reflect strength trends
(Navarro et al. 2019, 2021). To solve this, the from the MWD. For that, boreholes are divided in
results of the Structural factor have been divided zones characterised by relatively constant mean
in zones based on abrupt changes of the mean values of the PC1.
value in the signal to obtain a Structural Block PC1 has been plotted against the iron content
model. The procedure divides the signal in available for the data set blasts used to develop the
sections and estimates the mean value for each model. Figure 5a represents, on one side, the
one. These sections are varied iteratively until the correlation between the mean value of the longest
total squared error attains a minimum (Killick et section of PC1 and the chemical analysis (dots and
al. 2012, Lavielle 2005). The minimum of diamonds markers, left axis); markers rare shaded
the residual squared error is obtained by using according to the lithology of the blast where the
Gaussian log-likelihood. Figure 4 also shows the info was collected (as from geology reports, see
result of this procedure; vertical black dashed lines Table 1). Data from rigs D1 and D2 are also
indicate the edges of the sections and thick differentiated (dots and diamonds, respectively).
horizontal lines represent the average values In general, values from both rigs are grouped
within these sections; they are coloured according together and follow the same trend, which indicate
to their structural rock class (see legend). The the consistency of the methodology used to
analysis of the Structural Block model is simpler standardise data from different drill rigs (see,
now as minor fluctuations are filtered out. Figure 2b). Figure 5a also shows the histograms
PC1 classified into waste/ore according to the iron
5 STRENGTH-GRADE FACTOR content (amount of data in the histogram bars
indicated in the right axis).
The goal is to identify strength properties in the Rock areas associated with schisted sandstone
rock from the MWD data and to associate them and limestone rock types (shaded markers, see
with predominant lithologies and the iron content Figure 5a) are considered waste since their iron
(per hole) from assaying of the drilling chips content is in general lower than the 20%. These
(ore/waste identification). In line with section 4, a markers are normally related to high PC1 values.
PCA has been carried out on PR, normalised RP As described in section 2, schisted sandstone and
(RPNorm) parameters, PP and FP. The analysis has limestone have UCS values of 40 MPa and 125
carried out for 26 blasts (see Table 1), comprising MPa, respectively. Since the PR parameter
302 blastholes. Figure 3b represents the loading dominates the positive side of the PCA1 and this is
plot (influence of each parameter in the analysis) commonly used for hardness properties (the softer
for PC1 and PC2. The plane generated by the two rock the higher its value), it is reasonable a higher
components explains the 94.22 % (88.61 %, PC1 for schisted sandstone than for limestone. On
component 1 and 5.61 %, component 2) of the the other hand, blasts carried out in areas of
total variation among all parameters. According to siderite rock type (darker markers), which
Figure 3b, the first component is highly dominated constitute the ore deposit with an iron content
by the penetration rate on the positive side. This equal to or higher than 20%, are associated to low
parameter is commonly used for hardness PC1 values. Although limestone and siderite have
properties recognition (Schunnesson 1998), so its a similar UCS value of around 125 MPa (see Table
high influence is reasonable. Percussive pressure 1), PC1 can differentiate the two rock types in
and feed pressure parameters dominate on the most cases; for this reason, we have called PC1 a
negative side, showing a negative correlation with ‘Strength-Grade’ factor. However, there are
the penetration rate. Since component 1 covers situations where the drilling responds similarly to
nearly 90 % of the total variation, it can be said limestone and siderite, shown in the overlap zone
that it carries most of the drill system’s response to of the histograms in Figure 5a. Strength-Grade
the geomechanical properties of the rock mass; factor values on the left side of the histogram for
thus, the following analysis is based on it. ore and on the right side for waste clearly
Contrary to the Structural Block model, local represent potential identification of siderite and
variations in the signal are not relevant for this limestone lithologies, respectively. Values in the

- 190 -
 Figure 5. Correlation study between the MWD-calculated PC1 (Strength-Grade factor) and iron content associated
to each hole. a) Data used for the model. b) Data set used for validation (627 holes from 53 new blasts).

right tail of the waste histogram are associated to shows the histograms classified into waste/ore
schisted sandstone. according to the iron content of the data
From the results shown in Figure 5a, four represented. From the results, the model identifies
Strength-Grade categories of rock have been areas soft-waste (and waste material) in the 100%
defined combining the rock type description and of the holes investigated, areas of hard-waste
the iron content associated to the Strength-Grade (and waste material) in the 91% of the holes
factor values. These are indicated in Figure 5a, investigated, transition zones with a 67%/33%
separated by vertical dashed lines. (waste/ore) probability of the holes investigated,
and areas of hard-ore (and ore material) in the 90%
- Soft-Waste: Strength-Grade factor values of the holes investigated. This is in line with the
associated to waste areas of schisted results obtained in Figure 5a, supporting the
sandstone in 100% of the holes investigated robustness of the model.
- Hard-Waste: Strength-Grade factor values The Strength-Grade factor has been applied to
associated to waste areas of limestone in the map the whole mine area where the data set of the
90 % of the holes investigated model calibration was collected (Table 1). The
- Transition Zone: Strength-Grade factor result is represented in Figure 6. The model
values related to a range of difficult identifies areas of hard-ore and some transition
recognition between limestone (waste) and zones in the centre of the deposit (marked with
siderite (ore). 39 % of the holes falling into two dark dashed lines), indicating potential areas
this category are drilled in waste and 61% of ore material in this zone. To the left side of the
in ore graph, areas of soft-waste, hard-waste and
- Hard-Ore: Strength-Grade factor values transition zones appear, representing potential
associated to ore areas of siderite in 87 % intrusions of limestone and schisted sandstone
of the holes investigated. rock types in the deposit. In general, the categories
identified by the Strength-Grade factor are in line
The validation of the model has been carried with the different lithologies observed in the 3D
out with 53 new blasts (627 blastholes). For that, reconstructions.
the Strength-Grade factor has been applied to the 6 X-ROCK MODEL
MWD data of these blasts and compared with the
respective assaying values. In line to Figure 5a, the The Structural Block model has been combined
results of the analysis are plotted in Figure 5b, with the Strength-Grade factor in an overall rock
where markers are shaded according to the factor, exclusively obtained from drill monitoring
lithology of the blast as from internal reports and data, to provide an assessment of the structural
data from rigs D1 and D2 are also differentiated condition, strength properties and waste/ore
(dots and diamonds, respectively); Figure 5b also identification of the rock blasted. This factor has

- 191 -
 Figure 6. Application of the Strength-Grade factor to the whole mine area where the data were collected.

been named by the authors as X-Rock model. For combination of both Structural Bock model and
each Strength-Grade factor category, there are Strength-Grade factor into the X-Rock model in a
three structural classes, making twelve different production blasts (B16, marked with a white oval
X-Rock categories. Figure 7 represents the in Figure 6). A complex situation with three

Figure 7. Representation of the Structural Block model (left-upper graph), Strength-Grade factor (right-upper
graph), X-Rock model (left-bottom graph) and assaying analysis (right-bottom graph) for a production blast (B16).

- 192 -
lithologies can be distinguished (as from the 3D The second index is a Strength-Grade factor,
reconstruction of the post-blast bench face), based on the combination of MWD parameters in
corresponding to schisted sandstone, limestone principal components analysis, that has been
and siderite, each one associated with a type (see assessed from the analysis of the rock type
legend). The Structural Block model (left-upper description and UCS from geology reports, assays
graph, Figure 7) shows an area of heavily broken of drilling chips (ore/waste identification) and 3D
material in the first 5-7 m for all the holes (top UAV reconstructions of the post-blast bench face.
white line), identified by the model as fractured Data from 302 blastholes from 26 blasts have been
and heavily fractured categories, and two small used for this analysis. The Strength-Grade factor
fractured and heavily fractured zones in the ranks the rock into four categories: soft-waste,
schisted sandstone and the limestone (lower white hard-waste, transition zone and hard-ore. The
lines on the left) that are associated to areas of a Strength-Grade factor has been validated for a new
more intensely fractured rock. The siderite side data set of 53 blasts (647 blastholes) comparing
(left part) is more massive, with possible the values of the model with the iron content from
discontinuities across holes (dashed white lines). the assays of each hole.
The Strength-Grade factor (right-upper graph, Finally, the Structural Block model has been
Figure 7) identifies these three variations in the combined with the Strength-Grade factor to create
rock and associates the schisted sandstone area the X-Rock model, an overall rock factor
with soft-waste category (first three blastholes, exclusively obtained from drill monitoring data, to
from left to right), the limestone area with assess the structural condition, strength properties
hard-waste category (next two blastholes) and the and waste/ore identification of the rock.
siderite area with hard-ore and some transition Assessments for different conditions would require
zone categories (for the remaining ones). The a re-calibration of the model for the new site and
results are in line with the ore/waste identification drill rigs, following the methodology described.
from the assays (right-bottom graph, Figure 7); the A detailed rock mass characterisation per hole
transition zone is related to ore. Finally, the in an overall Rock Factor exclusively obtained
X-Rock model (left-bottom graph, Figure 7) from MWD data, provides an early information of
combines the characteristics identified by the two the rock to be mined, both in terms of strength and
previous indexes. jointing, and in terms of ore grade. This factor
allows improving blast design, hence better
7 CONCLUSIONS controlling blast results (e.g. rock fragmentation
and wall stability), by adapting blast design
Two rock description indexes are derived directly
parameters to the rock condition. In addition, the
from MWD data: one to classify the structural
potential ore/waste identification from MWD data
condition of the rock, and the other to assess the
may help to an early and fast ore/waste rock
strength properties and iron content of the rock. In
classification, helping to prevent ore waste and
order to use the drilling parameters, corrections are
dilution from blasting. The amount of assaying
carried out to minimise external influences other
required may also be reduced resulting in cost
than the rock mass. These comprise: (i) filtering
savings.
unrealistic values, (ii) removing systematic peaks
due to the addition of a new rod, and (iii)
8 ACKNOWLEDGEMENTS
correction of the hole depth influence. The
corrected MWD data have been combined using This work has been conducted within project
principal components analysis. ‘SLIM’ funded by the European Union’s Horizon
The first index is a Structural factor that 2020 research and innovation program under grant
classifies the rock mass in three classes (massive, agreement no. 730294. The authors would like to
fractured and heavily fractured) from the acknowledge VA Erzberg GmbH for their
variability of the MWD parameters and a valuable input into the project.
discontinuity factor determined using principal
components analysis. Video records of the inner REFERENCES
wall of 207 blastholes have been used to calibrate
the results. A Structural Block model has been Barton, N., Lien, R., Lunde, J. 1974. Engineering
developed as an improvement of this factor to classification of rock masses for the design of
simplify the automatic recognition of structural tunnel support”, Rock Mechanics and Rock
rock trends. Engineering, vol.6, no.4, pp. 189-236.

- 193 -
Bieniawski, Z.T. 1995. Classification of rock masses Leung, R. & Scheding, S. 2015. Automated coal
for engineering: the RMR system and future seam detection using a modulated specific energy
trends. Rock Testing and Site Characterisation, measure in a monitor-while-drilling context.
pp. 553-573. International Journal of Rock Mechanics and
Mining Sciences, vol. 75, pp. 196–209.
Cunningham, C.V.B. 1987. Fragmentation
estimations and the Kuz-Ram model-four years Lilly, P.A. 1986. An empirical method of assessing
on. Proceedings of the 2nd International rock mass blastability. Proccedings of the Large
Symposium on Rock Fragmentation by Blasting, open pit mining conference, Newman, pp. 89–93.
August, pp. 475-487.
Lilly, P.A. 1992. The use of the blastability index
Cunningham, C.V.B. 2005. The Kuz-Ram in the design of blasts for open pit mines.
fragmentation model–20 years on. Proceedings of Proceedings of the Western Australian
the European Federation of Explosives conference on mining geomechanics, Kalgoorlie,
Engineers, Brighton, England, pp.201-210. West Australia, pp. 8-9.

Deere, D.U. & Miller, R.P. 1966. Engineering Matlab, 2017. Natick, MA: The MathWorks Inc.
classification and index properties for intact rock.
Illinois University at Urbana, Department of Navarro, J., Sanchidrián, J.A., Segarra, P., Castedo,
Civil Engineering. R., Costamagna, E. & López, L.M. 2018.
Detection of potential zones in tunnel blasting
Ghosh, R. 2017. Assessment of rock mass quality and from MWD data. Tunneling and Underground
its effects on charge ability using drill monitoring Space Technology, vol. 82, pp. 504-516.
technique. Doctoral dissertation, Luleå Tekniska
Universitet. Navarro, J. 2019. The use of measure while drilling
for rock mass characterisation and damage
Ghosh, R., Gustafson, A. & Schunnesson, H. 2018. assessment in blasting. Ph.D. thesis, Dept. of.
Development of a geo-mechanical model for Geological and Mining Engineering. Universidad
chargeability assessment of borehole using drill Politécnica de Madrid.
monitoring technique. International Journal of
Rock Mechanics and Mining Sciences, vol. 109, Navarro, J., Schunnesson, H., Ghosh, R., Segarra,
pp. 9-18. P., Johansson, D. & Sanchidrián, J.A. 2019.
Application of drill-monitoring for chargeability
Hjelme, J.G. 2010. Drill parameter analysis in the assessment in sublevel caving. International
Loren tunnel. M.Sc. thesis in Geosciences, Journal of Rock Mechanics and Mining Sciences,
University of Oslo, Department of Geosciences. vol. 119, pp. 180-192.

Hoek, E. 1994. Strengths of rock and rock masses. Navarro, J., Seidl, T., Hartlieb, P., Sanchidrián,
ISRM News Journal vol. 2, no. 2, pp. 4–16. J.A., Segarra P., Couceiro, P., Schimek, P. &
Godoy, C. 2021. Blastability and ore grade
Kahraman, S., Rostami, J. & Naeimipour, A. 2016. assessment from drill monitoring for open pit
Review of ground characterisation by using applications. Rock Mechanics and Rock
instrumented drills for underground Mining and Engineering, vol. 54, pp. 3209–3228.
construction. Rock Mechanics and Rock
Engineering, vol. 49, no. 2, pp. 585-602. Peck, J. 1989. Performance monitoring of rotary blast
hole drills. Doctoral Thesis, McGill University,
Killick, R., Fearnhead, P. & Eckley, I.A. 2012. Montreal, Canada.
Optimal detection of changepoints with a linear
computational cost. Journal of the American Peng, S.S., Tang, D., Sasaoka, T., Luo, Y., Finfinger,
Statistical Association, vol. 107, no. 500, pp. G. & Wilson, G. 2005. A method for quantitative
1590–1598. void/fracture detection and estimation of rock
strength for underground mine roof. Proceedings
Kuznetsov, V.M. 1973. The mean diameter of the of the 24th International Conference on Ground
fragments formed by blasting rock. Soviet Control in Mining, pp. 187-195.
Mininig, vol. 9, no.2, pp. 144-148.
PIX4D, 2020, ‘www.pix4d.com.’
Lavielle, M. 2005. Using penalized contrasts for the
change-point problem. Digital Signal Processing, Sanchidrián, J.A. & Ouchterlony, F. 2017. A
vol. 85, pp. 1501–1510. distribution-free description of fragmentation by

- 194 -
 blasting based on dimensional analysis. Rock
Mechanics and Rock Engineering, vol. 50, no. 4,
pp. 781-806.

Schunnesson, H. 1997. Drill process monitoring in

percussive drilling for location of structural
features, lithological boundaries and rock
properties, and for drill productivity evaluation.
Ph.D. thesis, Dept. of Environmental Planning
and Design Division of Applied Geology, Luleå
University of technology, Luleå.

Schunnesson, H. 1998. Rock characterisation using

percussive drilling. Int Journal of Rock
Mechanics and Mining Sciences, vol. 35, pp. 711-
725.

Schunnesson, H., Elsrud, R. & Rai, P. 2011. Drill

monitoring for ground characterisation in
tunnelling operations. Proceedings of the 20th
International Symposium on Mine Planning and
Equipment Selection, 12-14 October, Almaty,
Kazakhstan.

Schunnesson, H., Poulopoulos, V., Bastis,

K., Pettersen, N. & Shetty, A. 2012. Application
of computerized drill jumbos at the
Chenani-Nashri tunnelling site in
Jammu-Kashmir, India. Proceedings of the 21st
International Symposium on Mine Planning and
Equipment Selection, 28-30 November, New
Delhi, India, pp. 729-751.

Scoble, M.J., Peck, J. & Hendricks, C. 1989.

Correlation between rotary drill performance
parameters and borehole geophysical logging.
Journal of Mining Science Technology, vol. 8,
no.3, pp. 301-312.

Tang, X. 2006. Development of real time roof

geology detection system using drilling
parameters during roof bolting operation. PhD
Thesis, Department of Mining Engineering,
University of West Virginia.

Teale, R. 1965. The concept of specific energy in

rock drilling. International Journal of Rock
Mechanics and Mining Sciences, vol. 2, pp.
57–73.

Van Eldert, J., Schunnesson, H., Johansson, D. &

Saiang, D. 2020. Application of measurement
while drilling technology to predict rock mass
quality and rock support for tunnelling. Rock
Mechanics and Rock Engineering, vol. 53, wp.
1349-1358.

- 195 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Studying blast induced rock movement in open-pit mining with

small benches

I. Koprev, I. Garkov, D. Kaykov, & D. Terziyski

University of Mining and Geology “t. Ivan Rilski” Sofia, Bulgaria

ABSTRACT: Blast movement monitoring is one of the leading ways to establish a good knowledge about
post-blast ore locations, as well as to optimise blast design parameters. In this paper, qualitative
parameters are considered in hypothesis testing to establish a better understanding of the key factors,
influencing blast movement and blast direction. Tests are performed to establish the significant difference
between sample medians. In addition, median differences are compared to the average rock size for
practical reasons. To gain a better understanding on how blast design affects movement, a correlation
matrix is analysed. Conclusions drawn from this analysis show that utilising a packaged watergel
explosive provides lower movement results, as well as less variation of movement. In addition, results
show that operations with small benches utilising a low powder factor can achieve relatively consistent
blast movement results which are close to the assumptions implemented by blast movement modelling
algorithms.

1 INTRODUCTION 2 METHODOLOGY
Blast movement is a phenomenon which is 2.1 General information
actively researched around the world for the last
40 years. Certain typical patterns in blast induced A 2-year-long study was conducted in the ‘Ada
rock movement are recognised, which complement Tepe’ gold project, which is an open-pit mine
fundamental theory of rock blasting. However, located in South-eastern Bulgaria. All blasting
many researchers base their conclusions purely on operations are conducted twice a week on 5 m
practical experience and a particular set of benches with vertical blast holes with the same
observations. Rarely robust mathematical or diameter of 105 mm. Two types of explosives are
statistical methods are used for proof of concept, utilised in the mining operation: ANFO and
when in fact mathematical models can lead to packaged watergel. Depending on the rock
superb modelling accuracy of blast movement (Yu hardness two types of drilling patterns are utilised:
et al. 2021). Therefore, this paper aims to provide 3 x 3.5 m for medium-hard rocks and 3.2 x 3.7 m
a robust study of blast induced movement, as well for softer rock types. The powder factor varies
as a template for operations utilising small benches between 0.34 – 0.46 kg/m³. Average rock size
and a low powder factor in softer rock after blasting is approximately 250 mm. All
environment. drilling patterns are made in a staggered

- 196 -
configuration and are initiated in an echelon 2.2 Parameters of interest
pattern with NONEL 42 ms delays within rows
and 17 ms between rows. Each blasthole uses Movement data was generated as a report from
bottom initiation. During the study two types of BMT’s software – BMM Explorer v 3.6
blasting were utilised – buffered and free face, (https://www.blastmovement.com). Each 3D
depending on whether there is unexcavated movement vector is described with a combination
material in front of the front face of the blast of three independent variables – horizontal
panel. In addition, blasting operations are movement, vertical movement and direction of
conducted on panels which have one or two movement. The parameter denoting the angle of
additional surfaces to the two regular free surfaces deviation for horizontal movement (αH) was not
in bench blasting. After blasting, excavators are initially present in the report and was calculated as
used for the selective mining of each bench in two the difference between the orientation angle of the
flitches due to the nature of the mined ore. perpendicular to the isochrones in front of the
Blast movement is observed by utilising BMM and the orientation of the movement
blast movement monitors (BMMs), produced direction. For this study the isochrones from firing
by Blast Movement Technologies (BMT) patterns were generated by the O-Pitblast software
(https://blastmovement.com/). Depending on the (https://www.o-pitblast.com/). Thus, the horizontal
number of ore polygons in each flitch for a single movement towards the free face (MH-ff) and the
shot, the number of monitors used can vary horizontal movement deviation (MH-dev) are
between 2 and 8 per flitch. Two monitors are used introduced as the components of the horizontal
for each flitch of the 5 m benches. During the movement vector. Their values can be calculated
study period over 1000 blast movement monitors via the equations:
were utilised for grade control during blasting and
for navigating shovelling operations. Each monitor 𝑀𝑀𝐻𝐻−𝑓𝑓𝑓𝑓 = 𝑀𝑀𝐻𝐻 . cos 𝛼𝛼𝐻𝐻 (1)
is placed either in the same or a separate 𝑀𝑀𝐻𝐻−𝑑𝑑𝑑𝑑𝑑𝑑 = 𝑀𝑀𝐻𝐻 . sin 𝛼𝛼𝐻𝐻 (2)
monitoring drill-hole, depending on whether there
is an ore polygon within that location of the Therefore, to describe the random nature of
respective flitch. If there are two monitors for a blast induced rock movement one of the following
single drill-hole, stemming is placed on top of the combinations can be utilised for each 3D
bottom BMM. After the last monitor is placed for movement vector: (MH, MV, αH), (MH-ff, MH-dev,
each drill-hole, it is filled with stemming. After MV) or (M3D, αV, αH), where αV is the inclination
blasting a staff member locates each BMM with a of the movement vector. In this current research all
specialised detector. 7 movement variables are utilised to identify how

Table 1. Nominal values used for studying blast induced rock movement.

Parameter Label Values

Flitch level FL 1 – Top flitch
2 – Bottom flitch
Explosive type ET 1 – ANFO
2 – Packaged watergel
Buffer presence BP 0 – Free face blasting
1 – Buffered blasting
Season type ST 1 – Spring
Season type can also be interpreted as 2 – Summer
weather conditions as the parameter is 3 – Autumn
nominal 4 – Winter
Voids proximity VP 0 – Within 5 m proximity
Radius based on highest horizontal 1 – Within 5 m proximity
movement
Rock type RT 1 – Metamorphic clasts and clays
2 – Carbonite breccia and Breccia conglomerates
3 – Siltstones with high content of quartz

- 197 -
each one is affected by blasting. In addition, the each BMM is placed. The decision of introducing
heave effect (He) is also introduced to this analysis these parameters to the analysis is that by utilising
to indirectly measure rock swelling in different the usual blast design parameters, there are a lot of
conditions. The parameter is measured as a repeating variables which cannot explain the
difference in elevation levels of the pre- and variance of blast movement. Hence, the
post-blast surfaces. The main groups of factors introduction of these ‘local parameters’ is a way to
which are known to affect blast movement are achieve a better evaluation of the specific
blast design parameters, rock properties as well as conditions in which each BMM is placed. Part of
the position of BMM within the blast panel. Many the overall blasting parameters (ID and NFF), in
factors within these groups are studied thoroughly addition to the mentioned above local parameters
by previous authors. However, not much are put in a correlation matrix to evaluate their
information is given on how these factors affect influence on the movement variables.
blasting operations with small benches and how
are they affected by weather conditions or by the 2.3 Statistical analysis
presence of different geological features.
To identify the key factors, certain nominal Due to the non-normal nature of the data, non-
parameters had to be assumed for the purpose of parametric tests were conducted for detailed
hypothesis testing. Table 1 shows the introduced studying of the blast induced rock movement. The
parameters and their respective values. Kolmogorov-Smirnov test (K-S test) was utilised
Typically blast design parameters used in a in this study for performing normality distribution
correlation matrix with movement parameters tests for each of the parameters in the samples.
include Burden (B), Spacing (S), Stemming length The Wilcoxon signed-ranks test (Wilcoxon 1945,
(T), Hole Length (HL), Explosive Column Length Siegel 1956) is utilised for comparing top and
(EL), Powder factor (PF), Number of free faces bottom flitch movement due to their dependence
(NFF) and Installation Depth (ID) (Yu et al. on the same monitoring hole location. The
2021). The current correlation analysis is based on Mann-Whitney U-test (M-W U-test) (Mann et al.
local parameters such as – Local Burden (LB), 1947, Fay et al. 2010) is performed to identify
Local Spacing (LS), Local average subdrilling whether the medians of each two compared
length (LJ), Local average stemming length (LT), samples are the same, while the K-S test for two
Local powder factor (LPF), Local gas products independent samples is used as a more robust
per m³ of blasted rock (LGP), Local energy factor method to identify any underlying differences
(LEF). These parameters represent the effect the between the distributions of the two samples,
three closest blastholes have in the zone where based on their cumulative functions. In addition,

Table 2. Hypothesis tests for establishing the key factors leading to variation in blast movement.

Null hypothesis (H0) Purpose Performed test

Studied parameters are normally Normality check for further hypothesis Kolmogorov-Smirnov
distributed testing
Movement is identical for both Identify whether BMM installation Wilcoxon signed
flitches depth is a key factor ranks
Movement is identical for both Identify whether explosive type is a Mann-Whitney U
explosive types key factor Kolmogorov-Smirnov
Movement is identical for buffered Identify whether buffer type is a key Mann-Whitney U
and free face blasting factor
Movement is identical for all Identify whether seasonal weather Kruskal-Wallis H-test
seasons conditions are a key factor Kolmogorov-Smirnov
Movement is identical for locations Identify whether void presence is a Mann-Whitney U
in and out of voids proximity key factor Kolmogorov-Smirnov
Movement is identical for larger and Representativity check Kolmogorov-Smirnov
smaller samples
Movement is identical for all rock Normality check for further hypothesis Kolmogorov-Smirnov
types testing Kruskal-Wallis H-test
Identify whether rock type is a key
factor

- 198 -
 Figure 1. Q-Q Plots for the studied parameters of movement (Normality test). M3D, MH, MH-ff, MH-dev
(upper row in that order), αH, MV, αV and He (lower row in that order).

the M-W U-test is affected by heteroscedasticity sample containing all observations during the
(Kasuya 2001). Therefore, this makes the M-W period (over 1000). Due to the non-normal
U-test the less reliable one, when differences in distribution of the data, a Spearman’s rho was used
variances between the analysed subgroups is for the correlation analysis, which was based on
present. In addition, before each comparison, a the representative sample. All statistical analyses
Levene’s test of equality of variation is performed were performed using IBM’s software SPSS
(Levene 1960). This was applied to check whether Statistics.
the results from each non-parametric test are valid
(Brown et al. 1974). The Kruskal-Wallis H-test3 3 RESULTS
(Kruskal & Wallis 1952, Corder et al. 2009) is
similar in its applicability with the Mann-Whitney 3.1 Normality check
U-test, however it can be used for comparing two
The results from the Kolmogorov-Smirnov test
or more independent samples. The assumed
show that none of the movement parameters
significance level of the tests is 0.05 with a
follow a Normal distribution. Therefore, a
confidence level of 0.95. Table 2 shows the
nonparametric test should be utilised for
conducted hypothesis testing, the purpose of the
comparing movement parameters further. Figure 1
hypothesis test, the types and the respective
represents each parameter’s Q-Q Plot.
method used.
Figure 1 graph serves only for illustrative
Due to a number of missing values in the
purposes of the general variance and skewness of
analysed sample, a smaller sample of 150
the movement parameters. A Normality test was
observations was drawn out from the total number
also used for each sample in the following
of observations during the period of the case study.
comparisons to justify the use of nonparametric
The sample was tested for representativity by
methods.
applying the K-S test to compare their distribution
functions. The smaller sample contained a more 3.2 Movement in both flitches
detailed information on rock types, obtained by a
previous geophysical study (Grigorova 2020). It A typical way of placing BMMs is on each flitch
was also used for obtaining the additional local level. This is done so that movement can be
parameters, which were not initially available. The monitored distinctly along the bench height for the
analysis on how ‘rock type’ influences blast respective ore polygons in the top and bottom
movement is based on the representative sample, flitch. The results for the hypothesis testing are
while all other hypotheses are tested using the represented in Table 3 and Figure 2.

Table 3. Results for installation depth hypothesis test.

Parameter M3D MH MH-ff MH-dev αH MV αV

Measure M m m m ° m °
P-value <0.001 <0.001 <0.001 0.425
Invalidated due to heteroscedasticity
Results Reject H0 Reject H0 Reject H0 Retain H0

- 199 -
 Figure 2. Box and whiskers plots of the movement parameters depending on the BMM installation depth.
M3D, MH, MH-ff, MH-dev (upper row in that order), αH, MV and αV (lower row in that order).

The K-S test was used for identifying any simplicity of the muckpile geometry. In addition,
significant differences between the distributions of movement can vary significantly between both
αH, MV, αV. Results show that there αH and MV flitch levels which reinforces the idea of
differ on both flitches from a statistical standpoint monitoring both flitches. In terms of horizontal
(α<0.001). The angle deviation of movement does movement, the difference between the two
not differ significantly on both flitches (α=0.067). medians is 0.5 m which is relatively low,
Therefore, no significant difference can be compared to previous works, but nonetheless
established in the horizontal movement deviation. impactful for shoveling operations.
This relation can be drawn from the fact that
at a small bench height (5 m) the gases generated 3.3 Movement in buffered and free face blasting
by the blast do not have sufficient energy at
the lower flitch level to cause it to move further. Buffered blasting was less likely to occur in the
Therefore, the upper flitch’s movement is bigger mining operation, however, it should be accounted
due to the low resistance the gases meet towards for as significant difference in movement
the front and upper part of the bench. The parameters can occur according to previous works
conclusion which can be drawn from this (Thornton, 2009). In many cases a buffer in front
comparison is that in open pit mines with small of the free face of the blast panel is unintentional
benches and a low powder factor, the D-shaped due to the presence of leftover blasted rock. Figure
curve of blast movement cannot be distinctly 3 represents the percentage of free face and
observed as in cases with higher benches. buffered blasts which were observed, as well as a
Therefore, for future modelling problems the general idea of how buffered blasting is
D-shape effect can be neglected for achieving performed.

Figure 3. Blast movement cases observed in buffered or free face blasting (left). Illustration of buffered
blasting (right) – source: Mala gіrnicha enciklopedіja 2004.

- 200 -
 Table 4. Results for buffer presence hypothesis test.

Parameter M3D MH MH-ff, MH-dev αH MV αV He

Measure m m m m ° m m °
M-W U-
P- 0.088 0.016 0.015 0.168 0.191 0.068 0.002 <0,001
test
value
K-S test 0.277 0.031 0.039 0.126 0.314 0.205 0.003 0.001
Retain Reject Reject Retain Retain Retain Reject Reject
Results
H0 H0 H0 H0 H0 H0 H0 H0

Figure 4. Box and whiskers plots of the movement parameters depending on buffer presence. M3D, MH, MH-ff,
MH-dev (upper row in that order), αH, MV, αV and He (lower row in that order).

The M-W U-test and K-S were both conducted uncover whether there are cases where explosive
for this analysis as the two samples are types affect vertical movement in buffered
independent and occurred on different occasions. blasting scenarios or whether there is a difference
The results from the two tests are represented in in vertical movement in only one of both flitches.
Table 4 and Figure 4.
Researchers point out that utilising buffered 3.4 Movement for different explosive types
blasting can lead to the prolonged work of gases
generated by the blast, which leads to improved ANFO, as the cheaper of the two utilised blasting
rock size (Anachkov et al. 1974). On the other agents, was mainly used for the operation. Figure
hand, this leads to a more chaotic blast movement 5 represents the percentage of blast movement
and an unpredictable movement in a vertical observations in conditions of both explosive types,
direction (Thornton 2009). It is not surprising that as well as a visual representation of them.
the medians for MH and MH-ff are slightly less for However, the reader should be informed that this
buffered blasting. In addition, the heave effect in is not the actual watergel explosive used. The
buffered blasting is bigger as buffered blasting photo serves only as an illustration. Results are
affects the direction where gases push fragmented shown in Table 5 for both tests used.
rocks upwards. However, in this study no Results show that there is a statistically
significant difference of vertical movement in both significant difference in movement between
cases was established. Movement, represented by utilising ANFO and a packaged watergel
M3D does not exhibit an overall difference, due to explosive. Horizontal movement is approximately
the lack of statistically significant difference in 0.5 m bigger in the case of ANFO. This can be
vertical movement. Nevertheless, this should be attributed to higher amount of gas products
investigated thoroughly in a future study to released by ANFO which led to pushing rock

- 201 -
 Figure 5. Blast movement cases observed in blasting with different blasting agents (left). Types of blasting
agents used – ANFO (centre), Packaged watergel (right).

Table 5. Results for explosive type hypothesis test.

Parameter M3D MH MH-ff, MH-dev αH MV αV He

Measure m m m M ° m M °
M-W U- <0,001 <0,001 <0,001
0.717 Invalidated due to heteroscedasticity
P-value test
K-S test <0,001 <0,001 <0,001 0.558 <0,001 <0,001 <0,001 <0,001
Reject Reject Reject Retain Reject Reject
Result Reject H0 Reject H0
H0 H0 H0 H0 H0 H0

further compared to watergel explosives. In A conclusion which can be drawn is that the
comparison, the Packaged watergel explosive has packaged watergel explosive proves to give
a higher relative strength and velocity of consistently better results in terms of movement
detonation (VOD). This causes improved rock compared to ANFO and therefore, the feasibility
fracturing resulting in the lower resistance in the of its application should be investigated in future
flow of gases between the induced fractures. This works.
leads to the less overall movement. Although one
can notice a significant difference in the 3.5 Movement during different seasons
distribution of the angle of horizontal movement
deviation, the resultant vector of horizontal This analysis was conducted only on observations
movement deviation had no statistically significant where ANFO is used because of its more common
difference in both subsamples. Overall results can use, compared to the Packaged watergel explosive
be seen in Figure 6. (Figure 7). ANFO was used during the winter and

Figure 6. Box and whiskers plots of the movement parameters depending on the explosive type. M3D, MH,
MH-ff, MH-dev (upper row in that order), αH, MV, αV and He (lower row in that order).

- 202 -
 Figure 7. Blast movement cases observed in blasting during different seasons (left). Annual regional temperature
and rainfall for the studied operation (right).

Table 6. Results for season type hypothesis test.

Parameter M3D MH MH-ff MH-dev αH MV αV He

Measure M m m m ° m ° m
P-value Invalidated due to 0,003 0,003 0,393 0,908 0,003
Result heteroscedasticity Reject H0 Reject H Retain H0 Retain H0 Reject H0

Table 7. Comparison between movement parameters depending on season type.

Parameter MH-dev, αH He
Sample ST=1 ST=2 ST=3 ST=4 ST=1 ST=2 ST=3 ST=4
ST=1 - -
ST=2 Retain - Retain -
ST=3 Retain Retain - Retain Retain -
ST=4 Reject Retain Retain - Retain Reject Reject -
1 – Spring; 2 – Summer; 3 – Autumn; 4 – Winter

spring as 2/3 of each month for both seasons had unequal variances (Welch 1947). In both cases no
no rainfall or snowfall. During the summer period significant difference between season subsamples
only 5 days per month had rainfall which led to the can be found. The same result can be achieved
unavoidable use of the packaged watergel in these when applying a paired K-S test. Vertical
conditions. movement and inclination of movement exhibit no
Unsurprisingly, the magnitude of blast significant difference. In terms of the magnitude
movement is not affected by different and variance of M3D, MH and MH-ff from an
temperatures and humidity during seasons, engineering perspective the median difference
however movement can vary significantly between between the seasons is below 0.1 m, which
seasons. Part of the results were invalidated after compared to the average rock size after blasting
performing a Levene’s test of variance. As it can can be considered as negligible. Results from
be seen by the results in Table 6, M3D, MH and comparing movement parameters are done by
MH-ff are heteroscedastic and therefore, they were using a post-hoc test with Bonferroni corrections.
excluded from the analysis. A detailed comparison between seasons is shown
In statistical terms comparing, M3D, MH and in Table 7.
MH-ff can be done either by violating the A more detailed picture on movement variation
requirement of equal variances for performing a between seasons can be found in Figure 8.
K-W H-test, or a normal distribution must be Previous studies prove that temperature is a
assumed in order to perform a Welch t-test for factor which can influence the VOD of explosives

- 203 -
 Figure 8. Box and whiskers plots of the movement parameters depending on the season of observation.
M3D, MH, MH-ff, MH-dev (upper row in that order), αH, MV, αV and He (lower row in that order).

(Mertuszka et al. 2019, Dobrilović et al. 2014). M. studies to claim whether this is coincidental or not.
Dobrilović et al. studied how temperatures close to As a conclusion, seasonal conditions do not seem
meteorological conditions can influence the VOD to impact the magnitude of the resultant blast
of ANFO following a linear relationship. During movement vector, however they can affect the
winter due to the coldest monthly temperatures for deviation of blast movement in different season
the year, the VOD of ANFO do drop to certain conditions. Nonetheless, the observed difference
degree, which potentially leads to the smaller between movement parameters may be a
crushing zone radius. Therefore, gases working statistically significant, but their absolute
in a prolonged time in a confined space difference is not of significance for mining
hypothetically can lead to the observed bigger operations.
heave effect during winter, compared to hotter
conditions during summer. Again, the observed 3.6 Movement in proximity of void zones
difference is low as an absolute value which is
approximately the same as average size of material During the exploration of the deposit certain zones
(0.15 m). During spring and autumn the deviation were identified as potential voids which were
form movement towards the free face exhibits a mapped and the overall shape of their zones was
statistically significant level of difference, which interpolated. However, at certain times in order to
cannot be explained in a straightforward manner. monitor ore polygons closely, a BMM should be
However, this must be investigated in future placed in proximity or inside the supposed

Figure 9. Blast movement cases observed in proximity to cavern and void proximity (left). An exemplary blast
panel with BMM locations near assumed boundary of the void zone (right).

- 204 -
 Table 8. Results for void presence hypothesis test.

Parameter M3D MH MH-ff MH-dev He αH MV αV

Measure m m m m ° ° m M
M-W U-
P- Invalidated due to heteroscedasticity 0.559 0.383 0.039
test
value
K-S test <0,001 0.001 0.001 0.475 0.035 0.191 0.537 0.080
Reject Reject Reject Retain Reject Retain Retain Reject
Result
H0 H0 H0 H0 H0 H0 H0 H0

Figure 10. Box and whiskers plots for the movement parameters depending on the void proximity.
M3D, MH, MH-ff, MH-dev (upper row in that order), αH, MV, αV and He (lower row in that order).

locations of these void zones as shown in Figure 9. 3.7 Movement in various rock types
This test aims to identify whether movement
variation, especially the values of αH and αv seem Rock type data was not initially available for each
to differ significantly due to changes in the BMM location in the desired detail. Usually
direction of blast wave propagation in the presence mining operations attribute similar values or the
of voids. The results from the hypothesis tests are same value for rock type parameters for a series of
shown on Table 8 and Figure 10. blasts, which was the case here. In order to make a
No significant differences can be observed in better distinction of rock types in each blast panel,
terms of changes of movement direction, but data collected by previous geophysical
interestingly the analysis shows that more outliers measurements was used for this paper. Electrical
in terms of αH and αv are present at the zone out of Resistivity Tomography (ERT) was used in a
the voids’ proximity. This can be explained by the previous study to characterise layers by exploiting
fact that the areas, which are supposed to represent resistivity contrasts between different geological
the existence of voids may not be entirely precise. features. The specific electrical resistance of
Therefore, part of the observations with an conglomerates and clays is lower due to their
attribute of VP=1 could be ‘false alarms’ and vice saturation with pore water. The silicified
versa – part of the observations absent of voids sedimentary rocks are pointed out as one of the
may actually be in their presence. Therefore, this highest electrical resistance rocks. Thus, the
analysis cannot be performed in a robust manner. specific electrical resistance of the breccia
However, this shows that a different methodology conglomerate with quartz inclusions is distinctly
should be used in the future for detecting the higher than the one of metamorphic clasts and
correct shapes of void zones. clays (Grigorova 2020). Figure 11 represents a

- 205 -
 Figure 11. Blast movement cases observed in different rock types (left). Rock types data used from Electrical
Resistivity Tomography (right). (Source: Grigorova 2020).

Table 9. Results for rock type hypothesis test.

MH-
Parameter M3D MH MH-dev αH MV αV He
ff

Measure m m m m ° M ° m

Invalidated due to
P-value Discarded from 0.299 0.181 0.821 0.679 heteroscedasticity
analysis due to
heteroscedasticity and Retain Retain Retain Retain
Result normality H0 H0 H0 H0

section of a blast panel, where each BMM was Results are shown in Table 9.
placed. Due to heteroscedasticity, the results from the
This hypothesis aims to identify the potential K-W test were discarded for M3D, MH, MH-ff
difference in blast induced movement in different and He.
geological features while applying a similar A better understanding of the movement
powder factor for the considered blasting panels. variation can be gained from Figure 12.

Figure 12. Box and whiskers plots for the movement parameters depending on rock type of placement zone.
M3D, MH, MH-ff, MH-dev (upper row in that order), αH, MV, αV and He (lower row in that order).

- 206 -
 Table 10. Non-parametric tests comparison between movement parameters depending on rock type.

Parameter M3D, MH, MH-ff He

Sample RT=1 RT=2 RT=3 RT=1 RT=2 RT=3
RT=1 - -
RT=2 Reject H0 - Reject H0 -
RT=3 Retain Reject H0 - Reject H0 Reject H0 -

Interestingly the discarded M3D, MH and MH-ff difference, however, of the heave effect in both
for all three groups exhibit a value of α=0.2 cases can be attributed to similar swelling factor
when applying a K-S test for normality check. for both rock types. Nevertheless, from a practical
Therefore, they are assumed to be normally standpoint the absolute difference between the
distributed. A Welch t-test was performed for all medians in both cases is lower than the average
three subsamples of rock types to investigate mean rock size. When comparing Zone 2 and Zone 3 it
differences between M3D, MH and MH-ff. A was also established that M3D, MH and MH-ff are
Games-Howell post-hoc test was then applied for significantly different from a statistical standpoint.
identifying the difference and the level of their The difference between the two groups of medians
significance between each group. The He is approximately 0.3 m. From a practical
parameter exhibits a distribution, which cannot be standpoint this distance is higher than the average
assumed to be normal. Applying a K-S test rock size in this case study, which can lead to
remains the only valid way for testing whether certain levels of loss and dilution if neglected. A
there is an underlying difference between the bigger heave effect in Zone 2 was also observed
empirical cumulative functions between the heave compared to the one in Zone 3, which can be
effect for all three rock types without violating attributed the better breaking of the breccias
normality assumptions or assumptions on the from Zone 2 compared to the poorly fractured
sample variances. Results are shown in Table 10. siltstone.
The results show that no significant difference
can be observed for rock movement in Zones 1 3.8 Correlation analysis
and 3. This similarity can be drawn from the fact
that wet clays can negate blast movement due to For this test the Spearman’s rho was used as the
their plasticity and lower velocities of propagation distribution of the variables is non-normal.
of stress waves. As the ERT study suggest, the The overall high variation in the data, as well as
Zone 1 had the lowest resistivity which can be the imbalanced sample all lead to lower levels of
attributed to their saturation with pore water correlation. Nonetheless, blast design parameters
(Grigorova 2020). Lower movement in Zone 3 can are key for the magnitude of movement as shown
be explained due to the insufficient level of the on Table 11.
powder factor for proper rock fragmentation. In Out of the correlation matrix it can be observed
addition, the two zones display a statistically that, the magnitude of the 3D vector is highly
significant difference in their heave effect (He) affected by the horizontal component, in particular
distributions. This also can be attributed due the the vector, perpendicular to the isochrones.
insufficient blasting energy in Zone 3, which can Therefore, it is affected mainly by the blast design
lead to a bigger percent of oversize and therefore parameters as well as the amount of gas products
less heave. However, this difference as an absolute each explosive type releases during detonation.
value is negligible in practical terms but should be This can be explained by the pushing force of the
kept in mind in blast movement modeling. expansion of explosive gases inside the confined
Hypotheses for M3D, MH and MH-ff were rejected space of the drill-hole. Installation depth also
due to the presence of a statistically significant seems to affect movement, it has a significant
difference in movements when comparing Zone 1 effect on vertical movement. The local energy
and Zone 2. This can be explained by the fact that factor has a relatively low correlation with all
clays from Zone 1 propagate stress waves induced movement parameters. Based on previous
by blasting slower compared to sedimentary rocks. research, this can be attributed to the unknown
This results in less horizontal movement in Zone 1 coefficient of the blasting chemical reaction’s
compared to the easily breakable breccia. The little efficiency, as well as the exact percentage of

- 207 -
energy distributed strictly for work in the physical operations with small benches. In terms of
sense of the word (Mitkov 2020). For future movement, horizontal angle deviation and
research different sampling strategies could be inclination of movement seem to vary a lot, but the
applied given that a bigger number of observations low magnitude of blast movement negates their
is obtained, which could lead to a more balanced high level of variance. In conclusion, this analysis
sample for the sole purpose of analysing how proves that operations with small benches utilising
movement is affected by blast parameters. a low powder factor achieve blast movement
Nonetheless, the information this correlation results which are very close to the assumptions, on
matrix provides is essential for planning future which blast movement algorithms work due to
operations, as it is compliant with the suitability of high correlations between horizontal movement
each drilling pattern used in the current geological perpendicular to the isochrones and the 3D
conditions, which is essential for the mining of movement vector. In addition, this study can be
future pushbacks. used as a starting point for future modelling of
blast movement, as well as for operations which
4 CONCLUSIONS aim to reduce their powder factor to minimise
movement.
The main conclusion which can be drawn from
this study is that blast movement is indeed REFERENCES
highly variable in terms of the distribution of the
parameters of movement. However, when Anachkov, A., Trendafilov, S. & Hristov, S. 1974.
comparing the medians of movement variables as Otkrit dobiv na polezni izkopaemi. Sofia, 1974
an absolute value, a small difference in practical (transliterated from Bulgarian).
terms can be observed. On one hand, this result is
Brown, M.B. & Forsythe, A.B. 1974. Robust tests for
satisfactory, as this lowers the number of factors
the equality of variances. Journal of the
having a significant effect on blast movement. On
American Statistical Association, 69, pp. 364-367
the other hand, this leads to an unexplained 1974.
variance in terms of movement. This can seem
paradoxical at first, but blast movement is very Corder, G.W. & Foreman, D.I. 2009. Nonparametric
much affected by all qualitative parameters above statistics for non-statisticians. Hoboken: John
– BMM installation depth, the presence of a buffer Wiley & Sons. pp. 99–105. ISBN
in front of the free face, the type of explosive used, 9780470454619.
the weather conditions, the geological conditions
in terms of voids presence and rock types. At the Dobrilović, M., Bohanek, V. & Žganec, S. 2014.
same time, operations utilising small bench Influence of explosive charge temperature on the
blasting, the influence of these factors is velocity of detonation of ANFO explosives.
downscaled, especially when utilising a low Central European Journal of Energetic
powder factor. Even though significant differences Materials, 2014, Vol. 11, no. 2, p. 191-197.
do occur in terms of the statistical distributions of ISSN 1733-7178.
movement parameters, the difference between the
examined medians points out that in certain cases Fay, M.P. & Proschan, M.A. 2010. Wilcoxon–Mann–
Whitney or t-test? On assumptions for hypothesis
some factors can be neglected as they are not
tests and multiple interpretations of decision
crucial for the end-result in mining operations. In
rules. Statistics Surveys. 4: 1–39. doi:10.1214/09-
most cases a significant difference in vertical SS051.
movement was not observed, which makes vertical
movement, and therefore, internal dilution an Grigorova, M. 2020. Geophysical investigations in
unavoidable effect. Although there is a small open pit ‘Ada Tepe’ Gold Mine in southeastern
absolute difference between the medians of the top Bulgaria. International Scientific Conference
and bottom flitch movement, both flitches should GEOBALCANICA 2020, p. 47-54. DOI:
be monitored in terms of better grade control. In http://dx.doi.org/10.18509/GBP.2020.05.
addition, blasting with packaged slurry provides
consistently better results in terms of horizontal Kasuya, E. 2001. Mann–Whitney U test when
and vertical movement compared to ANFO, which variances are unequal. Animal Behaviour 61
makes it a potential substitute. Therefore, a (2001): 1247-1249.
suitable explosive type and blast designs
compliant to geological conditions are Kruskal & Wallis 1952. Use of ranks in one-criterion
fundamental for controlling blast movement in variance analysis. Journal of the American

- 208 -
 Statistical Association. 47 (260): 583–621. https://www.o-pitblast.com/ (last accessed on
doi:10.1080/01621459.1952.10483441. 28.01.2022).

Levene, H. 1960. In contributions to probability and https://www.meteoblue.com/bg (last accessed on

statistics: Essays in honour of Harold Hotelling, 28.01.2022).
I. Olkin et al. eds., Stanford University Press, pp.
278-292.

Mala gіrnicha enciklopedіja : u 3 t. / za red. V. S.

Bіlec'kogo. — D. : Donbas, 2004. — T. 1 :
A — K. — 640 s. — ISBN 966-7804-14-3.
(Transliterated from Ukranian).

Mann, H.B. & Whitney, D.R. 1947. On a test of

whether one of two random variables is
stochastically larger than the other. Annals of
Mathematical Statistics. 18 (1): 50–60.
doi:10.1214/aoms/1177730491.

Mertuszka, V., Fuławka, K., Pytlik, M. & Szastok,

M. 2020 The influence of temperature on the
detonation velocity of selected emulsion
explosives, Journal of Energetic Materials, 38:3,
p. 336-347, DOI: 10.1080/07370652.2019
.1702739.

Mitkov, V. 2020. Vzrivna tehnika i tehnologiya,

Sofia, 2020, ISBN 978-954-353-430-2.
(transliterated from Bulgarian).

Siegel, S. 1956. Non-parametric statistics for the

behavioral sciences. New York: McGraw-Hill.
pp. 75–83. ISBN 9780070573482.

Thornton, D. 2009. The implications of blast induced

movement to grade control. Proceedings of the
Seventh International Mining Geology
Conference, Perth WA, Australia, pp. 287-300,
2009.

Welch, B.L. 1947. The generalization of ‘Student’s’

problem when several different population
variances are involved. Biometrika. 34 (1–2):
28–35. doi:10.1093/biomet/34.1-2.28.

Wilcoxon, F. 1945. Individual comparisons by

ranking methods (PDF). Biometrics Bulletin.
1 (6): 80–83.

Yu, Z., Shi, X. & Zhou, J. et al. Feasibility of the

indirect determination of blast-induced rock
movement based on three new hybrid intelligent
models. Engineering with Computers 37,
991–1006, 2021.

https://www.blastmovement.com/ (last accessed on

28.01.2022).

- 209 -
 5. Demolition blasting

- 210 -
- 211 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Analysis of air blast overpressures within a negative

pressure boundary

F.B. Kuhnow

ABSTRACT: Harmonic energetic delamination of a T2 building is being considered for demolition at a

formerly known atomic lab. The T2 building is within a containment structure that is under a negative
pressure of -49 Pa (-0.2 inches of WC) so that no remaining radioactive material is leaked out. The T2
containment enclosure is a portal frame-type system that spans east/west over 35.05 m (114’-11 7/8”) and
spaced 2.9 m (9’-6”) apart. The west side of the portal frames is supported by a 64 cm (2’-1”) thick
concrete wall which is backfilled to the west and open to the T2 basement to the east. The east side of the
portal frames is supported by a 61 cm (2’-0”) thick concrete wall that is backfilled to the east and open to
the tank farm vaults to the west. A concrete slab-type footing supports both the east and west walls.
Transverse walls provide lateral support for the walls as part of the original design. The west wall has an
elevated slab that provides further stiffening of the concrete wall between the transverse walls.
Previously, delamination had been considered for the neighbouring S2 building, but plans to do so were
aborted. During the review period for S2 several issues were brought up by authorities, most notably Air
Overpressure (AOP). Most typical demolition projects using blasting methods are performed in open air,
under atmospheric pressure. Using these methods within the containment enclosure led to failure to prove
success to the level of certainty authorities were expecting. At T2, the AOP concern also exists, although
there are portions of T2 that are more favourable to gaining client’s approval. Unfortunately, the most
challenging of the T2 delamination involves a 2.4 m (8’) concrete wall. This challenge is founded on the
fact that it is approximately 7.6 m (25’) from the containment boundary. A preliminary calculation
suggests that AOP at this distance will exceed the operational pressure transient that the containment is
designed to absorb. The production of expansive gases during detonation events produces increases in the
base air pressure values due to compression of the existing volume of air within the facilities. Effects of
this phenomenon are not generally considered in open air AOP estimates and were considered in the
demolition of the T2 building given that delamination activities involving detonation of explosives will
take place within a fixed volume, within the negative pressure containment. A theory was proposed to
phase the demolition such that the blast burden and AOP relieves to the pipe tunnel side. In addition, a
sound pressure wave survey was conducted inside the T2 building to estimate site specific attenuation
curves under negative pressure. These results were used to develop site specific attenuation curves that
led to a blast design that ensured pressure levels would not exceed the allowable 49 Pa (0.2 inches of WC)
at the containment boundary, making demolition of the T2 building by explosives possible.

1 AIR BLAST OVERPRESSURES time-histories of these air overpressures. AOPs

travel at the speed of sound. Hence, at sea level,
Air overpressures (AOP) can be simply defined as the velocity of sound in air is approximately 335
additional pressure generated from a blast above m/sec (1,100 ft/sec) at 7oC (45oF) with no wind.
normal atmospheric pressure. Air vibrations are As the temperature and wind velocity increase, the

- 212 -
sonic velocity increases. Therefore, the sonic graph and the AOP is determined by statistical
velocity affects the arrival time of air methods as in ground vibration. Since air blast is
overpressures. measured in decibels, pressures can be plotted in a
AOPs are pressure waves that create a log-log scale graph versus cube-root scaled
compression or positive pressure (push) followed distance and the AOP can be monitored to
by a dilatation or negative pressure (pull) effect. determine full waveform as with ground
The amplitudes are measured in pascals (Pa), vibrations. The cube root scaled distance SD3 is
millibars (mb) or pounds/inch2 (psi) above the used to evaluate air pressure attenuation using
ambient pressure. The pressures are reported as Equation 2:
time histories. Air blast overpressures may also be
reported as the sound equivalent in decibels (dB) SD3 = R/W1/3 (2)
and are calculated from measured pressures by
Equation 1: The best fit line to estimate the AOP from
scaled distance, as defined for ground vibration, is
𝑃𝑃 determined by Equation 3:
𝑃𝑃𝑠𝑠 = 20 𝑥𝑥 𝑙𝑙𝑙𝑙𝑙𝑙 � � (1)
𝑃𝑃𝑜𝑜
P = A (SD) – B (3)
Where: Ps = pressure level, dB (psi); P = measured
pressure, pascals (psi); Po = reference pressure, Where: P = air overpressure, dB (psi); SD3 = cube
typically 2x10-9 (psi) root scaled distance, m/kg 1/3 (ft/lb1/3); A = site
The methodology to estimate air blast constant (intercept of line at SD3 value of 1);
overpressures and ground vibration is similar. B = slope of the line (note that the slope is
Both present the same limitations. AOP is plotted negative)
against the cube root scale distance using a log-log Typical AOP data from similar operations were

Table 1. Air overpressure from open air operations at normal atmospheric pressure.

Distance
from
Intercept of Blast to
the line at a Monitor Charge
Air Over- Air Over- Air Over- Air Over- SD3 Value Slope Point Weight
pressure pressure pressure pressure of 1 of the (Feet) (lbs) SD
(psf) (in H2O) (dB) (psi) (A) Line (B) (R) (W) (SD3)
5988.56 1152.28 203.13 41.58723 19.57 -1.122 1 7.5 0.511
1375.74 264.71 190.36 9.55376 9.79 -1.122 2 7.5 1.022
581.93 111.97 182.88 4.04118 6.52 -1.122 3 7.5 1.533
316.05 60.81 177.58 2.19477 4.89 -1.122 4 7.5 2.043
196.84 37.87 173.47 1.36693 3.91 -1.122 5 7.5 2.554
133.69 25.72 170.11 0.92837 3.26 -1.122 6 7.5 3.065
96.39 18.55 167.27 0.66936 2.8 -1.122 7 7.5 3.576
72.6 13.97 164.8 0.5042 2.45 -1.122 8 7.5 4.087
56.55 10.88 162.63 0.3927 2.17 -1.122 9 7.5 4.598
46.87 9.02 161 0.32549 1.96 -1.1 10 7.5 5.109
11.17 2.15 148.55 0.07758 1 -1.1 20 7.5 10.217
7.15 1.38 144.67 0.04966 1 -1.1 30 7.5 15.326
5.21 1 141.92 0.03619 1 -1.1 40 7.5 20.435
4.08 0.78 139.79 0.02831 1 -1.1 50 7.5 25.544
3.34 0.64 138.05 0.02317 1 -1.1 60 7.5 30.652
2.82 0.54 136.58 0.01955 1 -1.1 70 7.5 35.761
2.43 0.47 135.3 0.01688 1 -1.1 80 7.5 40.87
2.14 0.41 134.18 0.01483 1 -1.1 90 7.5 45.979
1.9 0.37 133.17 0.01321 1 -1.1 100 7.5 51.087

- 213 -
obtained from the demolition contractor. The data of the project demolition. The engineering group
corresponds to open air operations for an initial has proposed the use of harmonic delamination as
charge load of 7.5 lbs. The data results are shown the primary excavation method of the wall,
in Table 1. allowing for a faster completion of the final stages
The interpreted attenuation curve representing of the overall demolition given that time
the data from Table 1 in units of decibels is constraints of the project. The harmonic
described by Equation 4 and plotted in Figure 1. delamination plan consists of making sure the blast
burden and AOP relieves to the pipe tunnel side,
AOP= 191 SD-.090(dB) with a 95% level of so that air overpressures are redirected to the
confidence (4) larger volume area, therefore decaying in strength
as they approach the enclosure. The challenge was
AOP = 327 SD-1.45 (in H2O) with a 95% the unknown rate of decay that AOPs would
level of confidence (5) experience given that the inside enclosure is under
the negative pressure, as opposed to atmospheric
2 FORMER ATOMIC POWER pressure, and the pressure the AOP exert on the
LABORATORY DEMOLITION CASE enclosure boundary. Several methods for
STUDY estimating pressure decays were researched, and
little to none were found that would estimate how
A theory was proposed to phase the demolition of the AOP waves would decay when subjected to a
an 8-foot structural wall to expediate completion pressure other than the atmospheric pressure. It

Figure 1. AOP attenuation curve - open air.

- 214 -
was determined that the AOP cannot be estimated − Measurements were taken at three locations:
by AOP attenuation curves given that the known vault monitor EHK030033 (17.25’ from the
AOP attenuation curves as those would only apply source), monitor EHK090091 (31.75’ from
to blasts conducted in the open, under atmospheric the source), and the pipe tunnel west wall
pressure, which was not the case for the referent monitor (52’ for the source).
case study. Therefore, the team worked developing
a site-specific plan to measure AOPs, and The source of the sound pressure was Skidsteer
modelled AOPs as sounds pressures since these Hydraulic Hammer and measurements were
would best describe the transient pressure under recorded during a two-shift day period. The
negative pressure. It was known that AOP monitoring equipment include 3M Edge q4
propagate through the atmosphere like dosimeter set as Fast Response C- weighted scale.
compression waves (p-waves) on the ground. It
was also known that no shear waves (S waves) 3 EFFECT OF DEMOLITION ACTIVITIES
exist in the atmosphere, since air has no shear ON THE CONTAINMENT ENCLOSURE
strength.
The plan layout (Figure 2) included the The current configuration of the T2 building
following steps: consists of levels 322 and below (Figure 3)
enclosed on a containment structure that was
− A sound wave survey was conducted inside placed to avoid leakage of radioactive material to
the T2 building to estimate site specific sound the environment. During demolition activities,
wave attenuation curves within the enclosure. maintaining the strength and preventing the
− The allowable pressure was determined as the addition of load to the east and west walls
pressure ‘just’ smaller than the magnitude of supporting the portal frame are necessary for the
the ventilation fan head pressure, i.e. +44 Pa continued integrity of the containment enclosure.
(+0.18 inches of WC), so that a pressure The wall supporting the east side of the portal
differential would be just below zero at the frame was maintained in its current configuration
containment boundary in order to maintain and no loads were added. Upon completion of
the pressure draft. This pressure in units of decontamination activities, and removal of the
dB is equivalent to 127 dB. containment enclosure, the vaults were removed.

Figure 2. Sound wave survey monitor locations.

- 215 -
 Figure 3. T2 building configuration.

Decontamination efforts will have a de minimis of reinforced concrete and, except for a small
effect on the wall strength. portion near S2, is buried below grade. A
ventilation partition was erected approximately
4 S2/T2 TUNNEL midway along the tunnel as part of the operation.
The S2/T2 tunnel connects the S2 and T2 Prior to open air demolition of T2, the portion of
buildings for the purpose of conveyance of the tunnel south of the partition was off limits to
material and personnel. No process equipment was personnel due to industrial hazards. Open air
housed in the tunnel. The tunnel structure consists demolition activities at S2 will not affect the north

Figure 4. T2 building, typical section,

- 216 -
 Figure 5. Proposed drilling plan for demolition of 8’ wall.

portion of the tunnel. Should personnel access enclosure can be used to our advantage.
south of the partition become necessary after open
demolition commences, it may be done with 6 DELAMINATION STAGING PLAN
adequate planning (i.e., suspend heavy equipment
activities and inspect for collateral damage). To manage the air blast overpressures (AOPs)
generated by the blast, the charge was placed in a
5 DEMOLITION OF 2.4 M (8-FT) WALL way such that the blast wave could be skewed
(Figure 4). The explosives were placed away from
The 2.4 m (8 ft) wall extends from north to south the containment boundary side of the wall such
38.4 m (126 ft) in length and is enclosed by the that the burden and AOP were on the pipe tunnel
304 and 322 level slabs. Demolition of this wall side. For instance, drilling holes and placing
presents some challenges. However, the structural explosives so that they are 91 cm (3 ft) from the

Figure 6. 8-ft Wall demolition staging section.

- 217 -
pipe tunnel side of the 2.4 m (8 ft) thick wall, Explosive Load: W := 3kg(7 lb)
leaving 1.5 m (5 ft) from the vault side of the wall
(containment boundary ~7.6 m (25 ft) from the Dynamax Pro: n := 32 mol
kg
vault side of the 2.4 m (8 ft) wall), leaving a 1.5 m
(5 ft) thick wall and 91 cm (3 ft) of fractured Grelease := n ⋅ W = 96.00 mol
material on the pipe tunnel side (Figure 5 and 2
Gas Constant: m
Figure 6). Then the next phase of delamination R := 8.31kg ⋅
2
would leave 91 cm (3 ft) of wall on the vault side sec mol ⋅ K
and 60 cm (2 ft) of fractured material on the pipe Temperature: T := 278K
tunnel side, then 30 cm (1 ft) of fractured material Tunnel Vol.: VT = 1,430 m3 (51,000 ft3)
on the pipe tunnel side and 91 cm (2 ft) on the T (3.2 psf)
vault side. Beyond that demolition by mechanical Pressure := n ⋅ W ⋅ R ⋅
VT
= 155.09 Pa
means would start.
Volume at P atm: T 3
Grelease ⋅ R ⋅ = 2.19 m
7 GAS NUISANCE 101300Pa
T
At this point, the AOP generated by the detonation PressureE := n ⋅ W ⋅ R ⋅ = 11.19 ⋅ Pa (0.24 psf)
of the charge is unknown. The production of VE
expansive gases during detonation events increases Ventilation Rate: Q := 32000cfm
the base air pressure values due to compression of 3
Equiv. Radius: VT (25 ft)
the existing volume of air within facilities. r := 3⋅ = 6.99 ⋅ m
Channelling or funnelling of these gases through 4 ⋅π
T
existing corridors, conduits and slab openings may Time to Exhaust: Grelease ⋅ R ⋅
101300Pa
focus these rapid pressure rises in localised areas Time :=
Q
= 0.14 s
of the structure. Effects of this phenomenon are
not generally considered in open air AOP As shown above, the pressure within the pipe
estimates and were considered in the demolition of tunnel due to the expansion of gases was 155 Pa
the T2 building given that delamination activities (3.43 psf) at the equivalent radius. When the
involving detonation of explosives will take place expanding gases encounter the open volume of the
within a fixed volume, under a negative pressure. enclosure (tunnel/enclosure), the volume increases
significantly (from 51,000 ft3 to ~700,000 ft3).
7.1 Pressures Due to Expanding Gases
Based on this volume increase, the pressure
Based on the specifics of the explosive product, decreased from 155 Pa (3.2) psf to 11.2 Pa (0.24
the gas volume produced by a single detonation is psf). In addition, it is expected that the ventilation
32 mol/kg of explosive. The dimensions of the system will take approximately 0.14 sec (or 140
tunnel are 35 m x 6.1 m x 6.7 m (115’ x 20’ x 22’) milliseconds) to dissipate the expanding gases
with a volume of approximately 1,430 m3 (51,000 from the explosive detonation.
ft3), and an equivalent radius of 7 m (23 ft).
9 PRESSURES DUE TO DETONATION OF
Using the ideal gas equation PV= nRT,
EXPLOSIVES USING AOP EQUATION
assuming 1 charge load of 3.4 kg (7.5 lbs) per
delay, the explosion will generate approximately Based on the specifics of the explosive product
155 Pa (3.24 psf) of pressure within the pipe and using the AOP attenuation curve obtained
tunnel due to the expanding gases. The target is to from similar open-air operations (Figure 7), the
maintain a pressure no greater than 49 Pa (0.2 AOP generated by the detonation of the charge
inches of WC) or (1.03 psf) at the containment would be in the order of 8.5 in H2O, which will
boundary. significantly exceed the allowable 0.2 in H2O.
8 ASSUMPTIONS AOP = 327 SD-1.45 (in H2O) = 8.5 in H2O
− AOPs behave as ideal gas
− The process of expansion of explosive gases However, the fact that the attenuation curve is
is adiabatic expected to be different when under negative
Given the following: pressure, this estimate was ruled out.
Ideal Gas: PV=nRT 10 SOUND PRESSURE SURVEY RESULTS
Volume of Enclosure:
3 To estimate the unknown AOP attenuation curves
V := 19821m (700,000 ft )
3
E for transient pressures under the negative pressure,

- 218 -
 Figure 7. Dyno Nobel Dynamax Pro specification.

the survey was conducted within the T2 building attenuation curves obtained from open air
containment structure which was under the -49 Pa operations.
(-0.2 inches of WC or -1.03 psf) of pressure. The resulting attenuation curve within the
Several recording stations were placed within the containment of the T2 building was described by
containment, and sound pressure at different the following expression (Equation 5), and results
distances, during several intervals were recorded. were plotted in logarithmic scale:
Sound pressures due to the hydraulic hammer
wall scabbling operation in T2 vault area were P = 184.73 SD-0.213 (5)
recorded. Site limitations force the work to cease
before 9:00 PM each day. The data between 6:00 When comparing attenuation curves obtained
pm and 8:30 pm segment presents the best source from conventional open air operations and the
of comparison across the subject wall boundary attenuation curves within the T2 containment
(Figure 8). Consequently, an attenuation curve was (Figure 9), it is clear from the figure the slope of
developed, and the results were compared to the AOP (inside T2) curve is about 2.5 times

- 219 -
 Figure 8. Sound pressure survey within T2 Building @ -49 Pa (-0.2 inches of WC).

steeper than that of the AOP (open air) given that To answer this question, the attenuation curve
the slope in the exponential equation is negative, developed using the results obtained from the
and that the ratio is 1.81 (0.21/.084), the higher sound wave survey (equation 5) was the basis to
(negative) the faster the dissipation. This result estimate the approximate transient pressure at a
makes sense as one would expect that the given scale distance.
ventilation system would ‘accelerate’ the transient
pressure dissipation. Consequently, one could 11 DISCUSSION
conclude that the transient pressure attenuation
would vent out faster. The question still is, what At this point, an estimate of attenuation curves for
would the pressure be when the air blast wave open air operations, attenuation of sound pressure
reaches the boundary? (@ 25 ft from the blast). within the containment, and pressures due to

Figure 9. AOP attenuation curve comparison.

- 220 -
 Figure 10. Comparison of AOPs at different conditions.

explosive’s expanding gases have been calculated. Given the following:

As previously mentioned, the goal was to maintain
a pressure no greater than 49 Pa (1.03 psf) at an d := 2 in
equivalent SD of 41 ft/lb 0.33. In terms of dB, this SGex := 1.45
pressure was equivalent to 128 dB (Figure 10).
It was clear from Figure 10 that the allowable γ conc := 150 lb
3
pressure of 128 dB would be exceeded when ft
the AOP wave reaches the boundary if the
open-air attenuation curve was used. However,   SG  .33 .33
 ex 160  
it was determined from the sound wave k := 30 ⋅  ⋅   = 31.00
survey that the AOP pressure would fall   1.4   γ conc  
well below the allowable pressure at the (Ash’s Equation)
same scale distance given that the sound wave
attenuation curve has a much ‘steeper’ slope. Burden, B := k ⋅ d = 5.17 ⋅ ft
Therefore, one can conclude that AOP within Spacing, S := B = 5.17 ⋅ ft
the negative would decay much faster than Stemming, S := 2 ⋅ d = 4.00 ft
in open air, and that the pressure level at the t
boundary would be far below the target. The Wall Height, W := 14 ft
H
attenuation curve (equation 5) was then used to Hole height, H := W − 2 = 12.00 ft
develop a blast design. H
Note: It is worth noting that the rate of decay is Column Height, C := H − S = 8.00 ft
t
independent of the charge load. For instance:
A charge load of 5 lb, 7 lb, or any other charge Length of the building floor L := 126 ft
load would decay at the same rate given that the No. of Holes Required: L = 24.38
decay rate is described by the negative component S
of the equation (slope). Explosive Load per hole:
W’:= 0.341 ⋅ SG ⋅ d2 ⋅ C = 15.82 lbs
12 BLAST DESIGN ex
W'
Based on the findings of this analysis, a blast P F := = 1.17 lbex
design was developed for the demolition of the γ conc ton
8-ft wall following specifications included in the B' ⋅ S' ⋅ H ⋅
ISEE Blasters’ Handbook, 18th Edition. 2000

- 221 -
13 BLAST PATTERN ADJUSTMENT It was concluded that attenuation curves from
‘open air’ cannot be used as reference when
Given that the purpose of this operation was blasting under pressure conditions other than
mainly to make the structure just fail in tension atmospheric pressure. This was confirmed by the
(not complete breakage) while maintaining the sharp difference in rate of decays (slope) on both
integrity of the enclosure, it was decided to conditions. Detonation of low charge explosives
decrease charge load from 7.2 kg (15.82 lbs) to 3.4 within the containment of the building was
kg (7.5 lbs) and strategically placed the blast holes therefore possible.
in a way that would allow skewing the The use of electronic detonators is
shockwaves towards the pipe tunnel and away recommended to accurately achieve the required
from the enclosure. As a result, the new blast delay time between charge loads.
design reduced the burden from 1.5 m (5 ft) to Disclaimer - The findings and conclusions in
0.91 m (3 ft) and the spacing to 0.91 m (3 ft). This this paper are those of the author and do not
reduction in the blast pattern resulted in an necessarily represent the views of the government
increase in the number of holes required from 24 or private entities. Mention of company names and
to 42, and therefore, resulted in a higher powder products does not constitute endorsement by the
factor. author.

13.1 New Blast Design: REFERENCES

B' := 3.00 ft Defense Special Weapons Agency 1997. DAHS CWE

Manual for the design and analysis of hardened
S' := B' = 3.00 structures under conventional weapons effects
W' := 7.5 lb (Vols. 1-4). Los Angeles, CA.
W'
P F := = 0.59 lbex Hustrulid, W.A. 1999. Blasting principles for open
γ conc ton pit mining: Theoretical Foundations (1st ed.).
B' ⋅ S' ⋅ H ⋅ Boca Raton, FL: Taylor & Francis.
2000
International Society of Explosive Engineers [ISEE].
No of Holes No := L = 42.00
(2012). Blasters’ Handbook (18th ed.).
S'
Cleveland, OH: International Society of
Explosive Engineers, Inc.
14 CONCLUSIONS AND
RECOMMENDATIONS Morhard, R.C. 1987. Explosives and rock blasting.
Dallas, TX: Field Technical Operations, Atlas
Based on the preceding analysis, measurement of Powder Company.
air overpressures under negative pressure is
challenging. However, the analysis suggests that Sigma Engineering Solutions 2016. Report on
harmonic energetic delamination of the 2.4 m (8 estimation of air overpressure values for the
ft) wall within the enclosure was possible without energetic delamination of concrete walls at
compromising the integrity of the enclosure. The building G-2 Knolls atomic power laboratory.
pressure levels were maintained below the target Las Vegas, NV.
level of -49 Pa (-1.03 psf, or 128 dB). The
maximum charge of 3.4 kg (7.5 lbs) was used. The Siskind, D.E., Stagg, M.S., Kopp, J.W. & Dowding,
charge was detonated in a scattered sequence so C.H. 1980. Structure response and damage
that there was enough time to vent out the gases. produced by ground vibration from surface mine
This delay time was in the order of 140 blasting. Pittsburgh, PA: United States Bureau of
milliseconds given the rate of the ventilation Mines.
system was 906 m3/min (32,000 cfm). Since the
volume of the containment was significantly larger
than that of the tunnel, the pressure at the
boundary dropped below the target value of 49 Pa
(1.03 psf or 128 dB) at the arrival. The pressure at
the boundary using the ideal gas equation was
estimated as 12.7 Pa (0.26 psf) which was well
below the target value of 49 Pa (1.03 psf).

- 222 -
- 223 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Demolition and reuse of a 152 m chimney at Amager Resource

Center, Copenhagen

C.A. Andersen & S. Qvist

NIRAS A/S, Allerød, Denmark

P. Brask
I/S Amager Resource Center, Copenhagen, Denmark

E.C. Thyrring
J. Jensen Nedrivning A/S, Lynge, Denmark

J. Schneider
DABT ApS/Dexploc A/S, Ballerup, Denmark

ABSTRACT: An old incineration plant in Copenhagen, Denmark has previously been demolished and the
last remaining structure to demolish was a 152 m tall concrete chimney. In the immediate vicinity of the
chimney there was new incinerator plant and a powerplant providing both electricity and district heating.
Due to the limited space the fall of the chimney had to be shortened. The considerations and process for
tendering the chimney for blasting, getting permission from authorities, dialogue with neighbours and
other stakeholders, selection of the demolition company, and the idea of recycling the concrete for new
structures will be described. The environmental impact had a very high priority and comparisons on
forecasts and measuring was recorded. The blast planning was done with the assistance of experts from
Germany and differences in culture and traditions could be observed as a curiosity. The practical work
was challenged by heavy wind up to the day before blasting. On Sunday 31st October 2021 in the
morning the chimney was successfully blasted. The environmental impacts in the form of dust, noise,
vibration, air overpressure and ejected fragments were monitored, and only negligible impacts were
observed.

1 INTRODUCTION The demolition was part of a plan to renew the

waste management in Copenhagen. From 2014 to
On 31 October 2021 the 152 m tall chimney of the 2017 a new state of the art waste incineration plant
former Amagerforbrændingen waste incinerator ‘Amager Bakke’ was built north of the existing
plant in Copenhagen was successfully demolished. plant. From 2017 to 2020 the existing plant was
Amagerforbrændingen was built in 1970 to handle demolished, leaving only the chimney. After the
waste from the city of Copenhagen and parts of the demolition of the chimney, a waste sorting plant is
suburbs. planned on the site.

- 224 -
 Figure 1. Chimney and surrounding area. Chimney (1), new incineration plant (2), ammonia tanks (3), transformer
building (4), powerplant (5), municipality workshops (6). Insert: Location near central Copenhagen.

The demolition was interesting for several The demolition had not been announced by the
reasons: Very few structures of this size have media until the time of demolition. The main
previously been demolished in Denmark, the reason for this was to avoid the gathering of large
chimney was located in an urban area with more crowds during the COVID-19 pandemic.
than 1 million people living within view of the
chimney, a number of important utility buildings 1.1 The chimney
and infrastructure were located in the immediate
proximity and finally the reuse of the concrete for The chimney consisted of a relatively thin
environmental reasons. reinforced concrete shell, 152.3 m tall from the

Figure 2. Left - the chimney, seen from north east before the demolition. The removal of outside paint is in
progress. To the right side of the frame, the new incineration plant is visible. Right - Original drawing of the
chimney showing the shape and intermediate decks among other things.

- 225 -
 Figure 3. Area surrounding the chimney. The approximate location of the main power lines and main
district heating pipes is denoted by the black line (7). Other numbers as per caption to Figure 1.

foundation to the top (148.7 m from ground level could potentially be affected, if the demolition
to the top). would not go as planned.
The chimney’s foundation was placed 3.6 m To the north, the new incinerator plant was
below ground level. From the foundation to a located about 105 m from the chimney (2). 150 m
height of 51.4 m the chimney was shaped as a to the northeast was a storage facility for ammonia
parabolic ‘trumpet’ and the top part was shaped (3). The nearest building to the northeast, at a
like a frustum of a cone. At the base the chimney distance of 200 m, contained transformers
had a diameter of 14 m and at the top the diameter supplying electricity to Copenhagen (4) and
was 5.4 m. The thickness was 33 cm at the base, further northeast was the power plant
gradually decreasing to a constant thickness of 18 Amagerværket (5) at 280 m. A large tank farm
cm from a height of 73 m to the top. area was located 450 m to the east (outside the
Inside the chimney, intermediate floors were figure). Along the road to the south were the main
placed at 18 m, 58.7 m, 102.7 m and 144 m above cables supplying electricity and district heating to
ground. Copenhagen (7). And 45 m to the south and
The structure was reinforced using traditional west were buildings used by the Copenhagen
reinforcement. A 7.9 m tall portal was placed at Municipality for technical operations and
ground level on the eastern side. The estimated maintenance (6).
total mass of the concrete shell of the chimney was Any damage to any of the named neighbouring
2,100 metric tons. facilities had the potential to severely disrupt the
The chimney was fitted with an internal rack supply of services to the city of Copenhagen.
and pinon elevator, an internal emergency ladder,
aviation obstruction lights and guardrails at the 2 ORGANISATION
top. Prior to the demolition, the smoke pipe had
been removed and the external paint had been I/S Amager Resource Center (ARC) hired the
cleaned off. engineering consultancy NIRAS A/S (NIRAS) to
conduct preliminary studies of the demolition of
1.2 The surroundings the chimney. The preliminary studies showed that
the demolition of the chimney using explosives
Near the chimney, several other buildings and was feasible and NIRAS was asked to draft tender
utilities were located within a range where they materials and assist during the tendering process,

- 226 -
 Figure 4. Organisation diagram of the demolition project.

and to conduct construction management and The meetings gave the authorities advance notice
supervision during the execution. that the project was underway, enabling any
Three bids were received and the bid from J. questions and requirements to be handled early
Jensen Nedrivning A/S (J. Jensen) was accepted. and included in the tender material to ensure that
One of the requirements in the tender material was the contractors application for permission to use
that the contractor should have international explosives would be quickly accepted. During the
experience from the demolition of similar planning phase, a joint inspection visit was
structures. Therefore J. Jensen hired Dr. Ing. conducted with participants from the police and
Rainer Melzer from Planungsbüro für fire and rescue service.
Bauwerksabbruch in Dresden, Germany to help In parallel with the meetings with the
draft the demolition plan and Michael Schneider authorities, meetings were also held with
from RL Liesegang GmbH in Cologne, Germany neighbours. The meetings gave ARC the
to provide quality assurance of the execution. opportunity to explain exactly what was being
Furthermore, the Danish company DABT ApS planned and the neighbours were able to voice
was hired to design actual blasts and draft the their concerns so any arising issues could be
drilling, charging and initiation plans. handled.
Because the neighbours mainly consisted of
3 COMMUNICATION WITH AUTHORITIES, utility installations, any damage to neighbours
NEIGHBOURS AND OTHER could potentially provide significant disruption to
STAKEHOLDERS the services provided to the city of Copenhagen.
The dialogue with the neighbours therefore
One of the reasons the demolition project was provided important input to the amount of
successfully completed, was due to the early protection measures implemented.
involvement and good and positive dialogue with
the authorities and neighbours. 4 DEMOLITION PLAN
The authorities included the Copenhagen
Municipality, the Danish National Police, the 4.1 General method
Copenhagen Police, and the Greater Copenhagen
Fire and Rescue Service. In the preliminary studies carried out by NIRAS
Regular meetings were held from five possible methods for the demolition were
approximately 6 months prior to the demolition, identified. The first method was the traditional
before the tendering process was initiated. method for demolition of tall structures, where the

- 227 -
 Figure 5. Possible demolition methods. Single cut (top left), two successive cuts (top right), zigzag (bottom
left) and vertical implosion (bottom right).

chimney is demolished using a single cut at the north-easterly direction.

bottom, causing it to fall in a single piece in all its In the second method, the chimney would be
length. To avoid hitting any nearby buildings or demolished in two stages, first the top half, and
utilities, the direction of demolition should be in a then the lower half a few days later. Both parts

Figure 6. Dynamic simulation of the collapse mechanism made during the tender phase with the upper cut in
75 m height and a delay of 2 seconds before initiation of the upper section. In the end, the height of the upper
section was raised to 90 m and the delay before initiation of the upper section was increased to 3 seconds.

- 228 -
would fall in a north-easterly direction. A dynamic finite element simulation of the
This method would reduce the size of the collapse was carried out to confirm that the
impact area, but also introduce the requirement for method would work and to help decide how the
some more complicated assessments about the demolition should take place. Based on a series of
stability of the lower part. dynamical simulations, it was in the end decided to
The third method would again see the chimney raise the location of the upper cut and introduce a
demolished in two parts, but with the two cuts delay before initiating the upper cut, to ensure that
facing in opposite directions and taking place at the top of the chimney would not end up too far to
the same time, causing the chimney to collapse in the southwest.
a zigzag shaped manner. This method would also
reduce the impact area, but in case of a misfire in 4.2 Demolition plan
one of the cuts, a large portion of the chimney
could potentially fall outside the planned impact A demolition plan was made by Planungsbüro für
area. Bauwerksabbruch by Dr.-Ing. Rainer Melzer who
The fourth method was a vertical implosion of has a long reference list and great knowledge
the chimney, where the chimney would be crushed within this field.
under its own weight. This method had the A requirement was that the top of the
smallest impact area, but if the concrete strength chimney must not land too far behind the
was unevenly distributed, a part of the chimney chimney as this would cause a problem for
could potentially fall in an uncontrolled direction. trucks who shall pass on the road behind the
The fifth method was mechanical demolition chimney. The upper cut was placed 90 m above
from the top using a movable platform and a small ground level.
excavator. The specifications of the concrete from the
A risk analysis of the different methods was drawings showed σmin ≈ 27 MN/m². Tests showed
performed, and the third method (zigzag) was 44.5 MN/m² and the concrete was in excellent
recommended. condition and much better than expected.

Figure 7. Plan of the area where the chimney is expected to land, including some protection measures.
The sand cushion in the impact area was later changed to a series of trenches and berms (left) and demolition
plan showing the location of the blasted sections (right).

- 229 -
 Based on the demolition plan the blasting was A substantial number of holes were to be
approved and detailed planning could start. drilled, and a hydraulic drilling machine was used.
Drill holes Ø 38 mm were drilled in a 0.35 x 0.35
4.3 Blast plan m pattern with a hole depth of 25 cm. In total 464
holes was made in the lower cut.
4.3.1 Blasting of the bottom section A specific charge of Lspec ~1 kg/m³ was chosen
and test blasting documented that this was OK.
A plan for drilling, charging and firing of the Eurodyn 2000 Ø 25 mm was selected as the
lower cut was made. blasting agent and clay was used for stemming.
A test blast on the chimney verified the It was specified that a redundant firing system
calculations and the planned principle for the should be used. This meant that 464 detonators
covering. were to be used in each side of the chimney and

Figure 8. Initiation system on the northern side of the lower section. Two EXEL MS detonators are used in
each hole, one ignited by a detonation-cord placed above and one ignited by a detonation cord placed below.

- 230 -
they would all be ignited independent of each The number of holes was reduced to in total
other. 106 holes.
As charging would take days to be done it was Because the upper cut was 90 m up in the air,
decided to go for EXEL MS detonators. and because buildings were located approximately
EXEL MS step 1-20 was used and was ignited 45 m behind the chimney, heavy covering was
by a detonating cord 5.3 g/m. required. The covering consisted of a layer of
The charging was completed without any geotextile fabric, a layer of chain link fence, a
problems, but it took time to keep everything nice layer of overlapping heavy blasting mats and
and easy to overview. finally a second layer of geotextile fabric.
Geotextile fabric and chain link fence was brought
4.3.2 Blasting of the upper section all the way around the structure, to catch any
debris from the breaking of the concrete on the
Thickness of the chimney at the upper section was rear face of the chimney.
18 cm.
The plan from Dr.-Ing. Rainer Melzer indicated 5 EXECUTION
that a small pre-weakening in each side would be
sufficient to form the hinge. The concrete in The work at the chimney was carried out by a
between would be drilled and blasted. small but effective team. The tools used were
As the concrete was very hard and drilling was mainly handheld power tools for cutting and
difficult, it was decided to increase the drilling, excavators for digging and low-level
pre-weakening in each side and to pre-weaken the lifting and mobile cranes for high level lifting of
front as well. both materials and a basket as work platform.
New calculations were done by Niras which A drill rig was used to drill the boreholes for the
showed that the top of the chimney would not lower cut.
become unstable and the windspeed could be
storm or even hurricane without becoming a risk. 5.1 Challenges
Due to the very small thickness of the concrete,
the drill holes were drilled at a 45° angle and the The pre-weakening, drilling, charging, and
drill pattern was intensified to 0.25 x 0.25 m. It covering 90 m above the ground showed to be
was decided to use NSP711, a PETN based plastic much more challenging as everything was
explosive approximately 20% stronger than supposed to be done by hand and working from a
Eurodyn 2000. The specific charge was increased free hanging basket as this was the only type of
to 2 kg/m³ as the stemming would be very short. crane available. During the work at the upper cut,
Redundant ignition was a requirement. high winds turned out to be a big problem. At

Figure 9. Left - work in progress charging the southern side of the upper section at a height of 90 m. The cut-
out of the central pre-weakening can be seen on the left side, and part of the heavy covering can be seen on the
far left. Two cranes were used when working on the covering. Right - the location of the rubble after the
demolition. The photo is taken from the top of the new incineration plant, at a height of approximately 75 m.

- 231 -
wind speeds above 4 m/s, it proved impossible to measures at his own expense, knowing that he
work from the basket. would be liable for any damages to neighbouring
This was very challenging, both during the properties. The minimum protection measures
drilling and charging, but especially during the included earthworks, container barriers, fragment
covering work for the upper cut. protection and dust protection.
This was remedied by a redesign of the Earthworks of sand and crushed concrete to
pre-weakening and more concrete was taken out. cushion the fall of the chimney. The original plan
In a very good dialogue, it was documented that called for a single large sand cushion for the
the pre-weakening in the upper cut was not an chimney to land on. This was changed into a series
issue, and it saved a lot of hard work for the of alternating dikes and ditches to save material.
contractor. On the road behind the chimney, the sand was
The heavy blast mats were hung up on hooks placed on top of steel plates to ease the following
drilled into the structure above the blasting area. clean up. The road is the main access for garbage
The blasting mats had to be fixed to the chimney trucks to the incineration plant, so it was important
by use of heavy load straps as the mats was not to be able to use it again as soon as possible after
supposed to move when hit by gusts of wind. the demolition.
The challenges presented by high winds Further earthworks and container barriers were
delayed the work by one week. placed to prevent fragments hitting nearby
buildings and storage tanks.
5.2 Cultural differences Wooden plates were placed covering
particularly vulnerable façades facing the
A number of cultural differences were observed chimney. This included the large glass panes in the
between how structures are blasted in Denmark site access building, and part of the transformer
and in Germany: building where the façade consisted of thin metal
panels. Also, a number of ventilation inlets at the
- Normally in Denmark we remove quite a bit transformer building and the powerplant were
of the structure and check that the remaining covered with plastic sheets and duct tape in order
columns to be blasted are strong enough to to prevent dust intrusion. Lastly, the skylights
withstand static loads and wind until blasting of the municipality workshop, which was the
is carried out, whereas in Germany it is more nearest building, were covered using heavy
common to do less pre-weakening and blast a plastic sheets.
larger proportion of the structure. From a few hours before the blasting and until
- At the upper cut, where the concrete was the chimney was successfully demolished, an
relatively thin, the boreholes were drilled at exclusion zone was maintained, and guards were
an angle pointing 45° downwards, in order to set out in order to prevent members of the public
obtain holes that were long enough to contain from entering the area. Guards on land were
larger charges (i.e. fewer holes were needed) provided by J. Jensen and a boat to prevent access
and long enough to contain both the charge from the harbour was provided by the Port of
and stemming material. In Germany the holes Copenhagen.
would traditionally have been drilled The combination of the covering and the
horizontally. protection measures worked very well. The only
- Heavy blasting mats as covering at the upper damages recorded were a few small holes in a light
cut would not normally have been used in corrugated plastic roof over a temporary workshop
Germany. In Denmark there is very little close to the chimney.
experience with blasting at a height of 90 m.
At least for this project, heavy blasting mats 6 ENVIRONMENTAL IMPACTS
were successfully used and no flying
fragments from the upper cut were recorded. The environmental impacts in the form of
vibrations, noise, air overpressure and dust
5.3 Protection measures from the demolition was monitored using a
comprehensive monitoring program.
To prevent damage to the surrounding buildings
and utilities, several measures were put in place. A 6.1 Vibrations
list of protection measures required as a minimum
was given in the tender material. The contractor A forecast of the expected vibration levels was
had the freedom to deploy further protection made by NIRAS during the initial studies.

- 232 -
 Figure 10. Measuring points for vibrations, noise and air overpressure.

The forecast was based on the method stated in 6.2 Noise and air overpressure
DIN 4150-1. The forecast and measured maximum
values are presented in Table 1. The noise level (Lmax) on the technical building for
The actual measurements were generally low district heating was measured as 114.3 dB(A),
compared to the estimates. The highest value of which is adjusted to 108.3 dB(A) to account for
7.8 mm/s at the municipality workshop the microphone being mounted on a wall. The
corresponds to 39 % of the guideline value noise level near the residential area to the west was
according to DIN 4150-3. measured as 96.7 dB(A).

Table 1. Forecast and measurements of vibrations. The range is taken from the centre of the chimney foundation.

Max. vibration
Location Range [m] Forecast [mm/s]
level[mm/s]
New incineration plant 105 9 4.40
Ammonia tanks 152 8 3.80
District heating main pipe (NE) 170 8 -
District heating technical building 105 11 6.10
Transformer building 220 7 2.75
Power plant 280 5 1.30
Site access building 61 11 -
Municipality workshop 48 12 7.80
Municipality storage 58 10 -
Citycontainer 167 7 2.45
Margretheholm (dwellings) 307 4 1.35

- 233 -
 Figure 11. Amount of dust collected at various locations.

When listening to the recorded sound, both the collection was done in accordance with the
lower and upper blast and the impact of the Norwegian standard NS 4852:2010.
chimney against the ground are clearly audible, The video footage of the demolition shows that
but interestingly the small detonators of the the visible dust cloud was airborne for about 3-4
initiation system for the upper cut are also clearly minutes after the blast and that it was moving
audible. slowly in an easterly direction. As the cloud
45 m from the chimney the air overpressure drifted out over the water, no complaints were
was 229.7 Pa from the lower blast and 44.0 Pa received with regards to dust.
from the upper blast. The air overpressure from the
collapse of the chimney was 374 Pa. 6.4 Reuse of concrete
In the area near the dwellings, the air
overpressure was 8.2 Pa from the lower blast and The recycling of concrete rubble e.g. as base
19.2 Pa from the upper blast. The air overpressure material for road construction has been done
from the collapse after the impact of the chimney for many years, however the direct recycling
with the ground was 47 Pa. into new structures has proved more difficult.
The maximum blast induced air overpressure The trouble is, that it is difficult to obtain
measured (229.7 Pa) corresponds to 92 % of the concrete that is good enough to pass the
guideline value for buildings according to SS 02 quality requirements for new structures.
52 10. Therefore, concrete has generally been
downcycled into products of lower value than
6.3 Dust the original structure.
Recent tests in Denmark have been
The demolition of a concrete structure of this size successful in reusing concrete as aggregate
will naturally give off a lot of dust. Water cannons in new concrete and obtaining a result with
were used to reduce the amount of airborne dust, sufficient quality to be usable in new structures.
but these were only partially successful. In 2018 a new waste sorting facility was built
Several dust collection bins were set up at in Copenhagen using concrete with recycled
strategic locations all around the chimney. Dust aggregate.

- 234 -
 At the time of writing, it is expected that the
high quality concrete from the chimney will be
used as aggregate in new concrete.

7 CONCLUSIONS

The demolition project has now successfully been

concluded. The chimney fell exactly as planned
and landed in the designated area. There was no
damage to neighbours and the environmental
impacts were well below the allowed levels.
Some of the main pointers are:

- Early dialogue with authorities and

neighbours proved fruitful
- The collapse mechanism was successfully
verified using a numerical simulation, which
provided a higher level of confidence in the
plan
- The use of heavy covering was cumbersome
but was worth it
- Working from a hanging basket in high winds
proved more difficult than anticipated
- Some interesting differences between normal
practice in Denmark and Germany were
noted
- No damages were recorded
- Environmental impacts were well below the
allowed levels

REFERENCES

Environmental Protection Department. 2016.

Construction and Building Regulation in
Copenhagen. The Technical and Environmental
Administration, City of Copenhagen,
Copenhagen.

DIN 4150-1:2001-6. Vibrations in buildings - Part 1:

Prediction of vibration parameters.

DIN 4150-3:2016-12. Vibrations in buildings - Part 3:

Effects on structures.

SS 02 52 10:1996. Vibration och stöt -

sprängningsinducerade luftstötvågor -riktvärden
för byggnader.

NS 4852:2010 Luftundersøkelser, Uteluft, Måling av

Støvnedfall.

- 235 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Technical and safety aspects for the demolition of a chimney with a

height of 150 m in a difficult location

R. Laszlo, E. Ghicioi & C. Jitea

National Institute for Research and Development in Mine Safety and Protection to Explosion –
INSEMEX Petrosani, Romania

A. Nicola
University of Petrosani, Romania

L. Miklos
Industrial Demolition Company, Dîmbău, Romania

ABSTRACT: The paper describes the practical way and security conditions of realising the demolition of
an industrial chimney located in an oil refinery, having in its immediate vicinity sensitive equipment and
constructions. The chimney has a height of 150 m and consists of the outer tubular structure of reinforced
concrete and an inner chimney made of refractory brick masonry. In order to establish the possible
secondary effects generated by the blasting works, a risk assessment was made with reference to seismic
wave, air shock wave, noise, dust and the level of shock generated by the impact of chimney construction
with the soil. The demolition work was carried out successfully and due to protection and safety measures
no unwanted events were recorded, and the measured values of the seismic wave generated by the impact
with the ground of the chimney were non-dangerous, being below the values estimated in the risk
assessment.

1 INTRODUCTION efficiency, quality and security. The basic idea

when performing a building demolition is that the
The restructuring of industrial activities has as a destructive effect on the neighbouring structures to
consequence the conservation or decommissioning be protected has to be negligible, the number of
of constructions or areas that can later be the elements destroyed by the blast has to be as small
object of the development of new projects. As a as possible, as well as like the explosive quantities
result, the demolition process using blasting works that are blast at once.
have a challenging application from the technical The paper describes the practical way for the
point of view but as well as due to the reduced demolition of an industrial chimney located in an
time consumption, labour and cost. A large oil refinery, the technical and safety solutions
number of demolitions works by using explosives adopted for its successful overturn in the intended
and characterised by a high degree of difficulty, direction, in very sensitive location conditions
have shown that the use of the blasting technique regarding the constructions and installations in its
is a proper alternative from the point of view of immediate neighbouring. The aim for cleaning the

- 236 -
 Figure 1. Oil refinery platform.

area it is the construction of a new cogeneration successfully and due to protection and safety
plant for providing the oil refinery with electricity measures no unwanted events were recorded, and
and technological steam. the measured values of the seismic wave generated
In order to establish the possible secondary by the impact with the ground of the chimney were
effects generated by the execution of the chimney non-dangerous, being below the values estimated
demolition by blasting works, a risk assessment in the risk assessment.
was made with reference to seismic wave, air
shock wave, noise, dust and the level of shock 2 INDUSTRIAL CHIMNEY CONSTRUCTION
generated by the impact chimney construction DESCRIPTION
with the soil.
The demolition work was carried out The industrial chimney was located inside a large

Figure 2. Chimney and neighbouring objectives.

- 237 -
 Figure 3. Free space between chimney walls.

Romanian oil refinery as is shown in Figure 1, with a radius of 4.65 m. The thickness of the
and served a thermoelectric plant in order to reinforced concrete wall is variable, from 62 cm in
provide water for thermal heating and domestic the lower area to 21 cm at the top, while the
hot water for the inhabitants of the city. wall thickness of the brick chimney is constant,
In the immediate vicinity of the chimney that is 20 cm.
being demolished, there are a large number of The foundation of the concrete chimney is of
structures and pipeline networks, the most reinforced concrete screed type (Figure 3) with a
sensitive and important of which are (Figure 2): diameter of 31.50 m and a thickness of 1.80 m,
which rests at the lower level of - 4.40 m on 45
− to the north-west, at 120 m, 6 KV power reinforced concrete piles with a diameter of 1500
station mm, which have a length of approx. 36 m and is
− to the west, at 200 m, cooling tower embedded in the layer of sand, limestone and
− to the south-west, at 70 m, trestle with pipes sandstone. The walls of the chimney made of brick
− to south-west, at 120 m, chemical station masonry are supported on a foundation with
− to south-east, at 40 m, boiler building its own structure of reinforced concrete pillars
− to the south-east, at 90 m, trestle with pipes of 50 x 50 cm up to a height of +14.50 and on
− to east, at 72 m, gas and boiler operation consoles formed by the sliding wall and ring
building beams in the form of a wheel with spokes
− to north-east, at 135 m, fuel depot. at each 8.00 m between elevations +14.50 and
+150.00 m.
The chimney has a height of 150 m and Between the reinforced concrete wall and the
consists of the outer tubular structure of reinforced brick chimney wall, there is a free space used for
concrete, having a radius from the bottom side the internal control of the structure and which in
(level + 16 m) of 6.85 m, respectively and 5.43 m the area of elevation 16 - 25 m is about 1.6 m
at the top side (level +150 m), as well as from the (Figure 3). Related structures - flue gas ducts, are
inner chimney made of refractory brick masonry made of solid brick masonry and reinforced

- 238 -
concrete and are developed up to elevation + 17.5 − avoiding for a long time the presence of dust,
m (Figure 4). noise, vibrations, etc.

3 CHIMNEY BLAST DEMOLITION The choice of a demolishing method by

blasting variants is conditioned by the physical
3.1 Considerations on demolition by blasting
state of the construction, by the existence of
works
structures in the vicinity of the demolition site, by
Carrying out a large number of demolition works the possible effects of the demolition on these
with the help of explosives, characterised by a objectives. Each construction creates a special
high degree of difficulty, showed that the use of case, separately, the calculation of the blasting
the blasting technique is a safe alternative and parameters adapting to each situation. The
often the most appropriate in terms of efficiency explosive load required for the dismantling of a
and safety. Compared to conventional demolition certain constructive part, is dependent on the type
procedures, the use of blasting works offers the of explosive used, the material being shot, the type
following advantages in terms of hazards, of construction being demolished and the
environmental protection and work: geometry of the location of the blasting holes.
The demolition process chosen must meet the
− blocking for a short period of time some following requirements:
neighbouring parts and traffic areas - periods
of time limited only to the duration of the - directing the fall to protect the nearby
actual execution of the blasting works - active structures and maintain the
compared to the usually long restrictions on production/activity process
the use of these spaces and areas, to - protecting buildings near the target, against
demolition by conventional means seismic action, shock wave, and throwing
− rapid removal of the state of danger, concrete or metal pieces under the effect of
especially where buildings are in an advanced the explosion
stage of deterioration - destroying the integrity of the construction,
− avoiding working at height for long periods so that the dismantled elements can be
of time transportable or loaded mechanically.

Figure 4. Chimney foundation and related constructions.

- 239 -
 Figure 5. Chimney overturning direction and cut location.

3.2 The chimney demolition process elevation +14.5 m, at intervals of 8.00 m, between
the sliding wall and the brick masonry wall, there
The complete tubular development of the chimney are consoles formed by annular beams in the shape
(without the existence of technological gaps such of a wheel with spokes. Over these reinforced
as the chimney flues) is made above the screed, concrete brackets and spokes are placed nets to
starting from the elevation +17.5 m. Above facilitate staff access to inspect the condition of

Figure 6. Cut preparation at elevation + 23 m.

- 240 -
 Figure 7. Images of two windows at the limits of the detachment cut.

the chimney construction. The access to these - realization of the detachment cut for which
platforms is made by a metal staircase arranged purpose will be drilled in the concrete wall,
inside, in the free space of 1.6 - 1.0 m. When from inside the chimney, starting from the
demolishing the chimney, the overturning elevation +23.5 m, from bottom to top, 6
procedure will be applied in a given direction. The rows of holes placed in the square network
existence at the elevation +23.0 m of the first visit (Figure 6 and 8), as follows:
/ work platform inside the chimney, determined • row 1 of holes, starting from the
the choice of the area from this elevation for the elevation +23.5 m, on the opening of 1/4
execution of the chimney overturning cut (Figure of the inner perimeter of the reinforced
5).This platform allowed the execution of concrete chimney (9.8 m), with a
preparatory works - drilling of mine holes in the number of 23 holes. The holes will be
body of the concrete and brick chimney, as well as arranged symmetrically on either side of
those of loading with explosives and initiation the axis of the overturning direction
means in conditions of increased safety (Figure 6), • row 2 holes drilled above row 1 holes,
compared to the alternative in which these works on the opening of 1/2 of the inner
would have been executed outside the chimney perimeter of the reinforced concrete
from self-lifting platforms. chimney (19.6 m), with a number of 45
holes. The holes will be arranged
3.3 Preparatory stages for demolition of symmetrically on either side of the axis
chimney structure of the overturning direction
• rows 3,4,5 and 6 holes drilled above row
Given that in the immediate vicinity of the
2 holes, on the opening of 2/3 of the
chimney, at a distance of 150 - 200 m in the
inner perimeter of the reinforced
direction North, North - East, there are no
concrete chimney (26.2 m), with a
objectives to be protected, its demolition will be in
number of 4 rows x 57 holes per row.
this direction (Figure 5). In order to demolish the
The holes will be arranged
chimney structure, the following phases have been
symmetrically on either side of
completed:
the axis of the overturning direction.
- creation by blasting works of two openings - drilling two rows of holes in the wall of the
(windows) at the limits of the detachment cut, brick chimney inside the concrete chimney,
located at 2/3 of the inner perimeter, having a in order to make a section for loosening the
width of 1.2 m (corresponds to twice the brick masonry in the area of the detachment
thickness of the reinforced concrete wall) and cut. For this purpose, at elevation + 25 m, two
a height of 1.2 m (corresponds to the height rows of 66 holes per row will be drilled on
of the arrangement of four rows of holes 2/3 of the perimeter of the brick chimney
along the length of 2/3 of the inner perimeter (19.47 m). The holes will be arranged
of the reinforced concrete chimney), Figure 7. symmetrically on either side of the axis of
The realisation of the two openings also the overturning direction (Figure 9)
served to establish the way in which the - decommissioning in the area of the
reinforced concrete structure behaves under detachment cut of all additional elements
the action of explosive loads such as metal stairs and lightning conductor.

- 241 -
 Figure 8. Diagrams of the detachment cut in the concrete wall.

Figure 9. Section for loosening the brick masonry in the area of the detachment cut.

3.4 Blasting parameters - type of explosive and To make a loosening zone in the brick masonry
mode of initiation chimney, the following drilling & blasting
parameters were used (Figure 9):
The explosive charges were sized according to the
section of the construction elements to be blasted. - wall thickness, G p = 0.20 m
For the removal of the detachment cut in the - burden, W = 0.10 m
reinforced concrete chimney, the following - hole spacing, a = 0.30 m
drilling & blasting parameters were used - row distance, b = 0.25 m
(Figure 8): - number of rows, Nr1 = 2 on 2/3 of
perimeter x 66 holes/row
- wall thickness, G p = 0.60 m - hole length, L h = 0.13 m
- burden, W = 0.30 m - explosive charge per hole, Q h = 0.040 kg
- hole spacing, a = 0.45 m - total explosive charge, Q T = 6.0 kg.
- row distance, b = 0.40 m
- number of rows, Nr1 In order to collapse the chimney construction in
= 4 on 2/3 of perimeter the expected direction, all the explosive charges
x 56 holes/row; 1 on 1/2 of perimeter from the concrete and brick chimneys are
x 44 holes/row; 1 on 1/4 of perimeter detonated instantly.
x 24 holes/row Dynamite type explosive loads were placed in
- hole length, L h = 0.40 m holes and the explosive charges in each hole were
- explosive charge per hole, Q h = 0.20 kg initiated with detonating cord. At the cut edge, the
- total explosive charge, Q T = 60 kg. ends of these secondary lines are connected to a

- 242 -
 Figure 10. Initiation network.

detonating cord main line together with that one the boilers located at 40 m in a southerly direction
coming from brick chimney area. A non-electrical from the chimney and the gas and boiler operation
detonator was placed at the end of the detonating building located at 72 m in the east direction from
cord main line (Figure 10). the chimney. The two objectives being the closest
The total amount of explosive used to demolish to the chimney, were taken as a reference to assess
the chimney construction is as follows: the side effects of the demolition by blasting
works.
- Q Te = 73 kg dynamite The seismic effect due to the detonation of
- Q Tdc = 400 lm detonating cord explosive charges during the chimney demolition
- N d = 8 pcs of nonelectric detonators works is negligible due to the distribution of the
- N c = 10 pcs of nonelectric total charge on a number of small charges (0.040 -
connectors/starters. 0.200 kg.) placed in the structure in areas above
4 RISKS EVALUATION ground.
The risk assessment showed that the oscillation
The success of the demolition by blasting works is velocity generated by the explosion shock, even if
also conditioned by the solution of the protection the explosive charges would detonate in holes
of the civil and industrial constructions against the drilled in the ground, is 0.370 cm/s, a value that is
action of the side effects of the explosions carried below 2.1 cm/s. chosen as admissible.
out during the demolition. The impact of a collapsing structure on the
Blasting activities may include: ground can result in seismic effects, the magnitude
of which depends on the energy released on
- seismic effects generated by the detonated
impact. This energy depends on the mass of the
explosive charge
structure being demolished, the height of its center
- seismic effects produced by the impact of
of gravity and the characteristics of the ground on
falling on the ground of the structure that is
which it falls.
being demolished
- overpressure The risk assessment showed that at for the
- dust and gas objective closest to the chimney - the boiler room,
- throwing pieces of material under the action the value of the velocity of seismic waves possible
of the explosion. to be generated at the impact of a collapsing
structure of the chimney on the ground, is 1.032
The industrial structures in the immediate cm/s, value lower than 2.10 cm/s the one chosen as
vicinity of the chimney that was demolished are permissible.

- 243 -
 Figure 11. Damping bed arranged on the overturning direction.

In order to reduce the impact with the ground, a not dangerous for building or human safety. The
damping bed was arranged in the direction of fact that the initiation network made up of the
overturning the chimney according to those detonating cord (the detonating cord is the
presented in Figure 11. component that most favours the creation of the air
In order to verify the level of vibration induced shock wave) is located inside the chimney, also
in the ground by the impact when the chimney helps to reduce the value of the air shock wave
collapsed, three seismometers were placed at the propagated around the chimney.
objectives considered the most sensitive. The Environmental pollution occurs to a very small
results of the measurements are presented in Table extent and is very limited in time and space of
1 and they confirm that the seismic values induced action and is due to explosive gases and dust
by the collapse of the chimney are lower than the resulting from the partial crushing of concrete.
permissible ones. Specific to the blasting works and especially the
demolition works, is the fact that the gas
Table 1. Measured vibration levels at the chimney generating event is limited in time and number.
collapse. Thus, the duration of the emission source is in the
order of hundreds of milliseconds. To reduce dust
Measuring Distance from PPV pollution, the surface on which the building is to
location location to chimney [cm/ fall will be sprayed with water, both before, during
[m] s] and immediately after its collapse.
Boiler room 34 0.69 Throwing of small material under the effect of
Demineralisation 122 0.27 the explosion, is diminished by the installation at
station the level of the detachment cut, where the
Power station 164 0.62 explosive charges are located, of local means of
110 kv protection with materials made of welded wire
mesh with large mesh, flexible wire mesh with
small mesh and an external protection. of
The overpressure of the shock wave also has geo-textile cloth placed 0.5 m away from the
very low values, due to the distribution of the total construction. The external protection shall cover
explosive loads over a large number of holes with the area of the explosive charges, exceeding by at
small loads. least 1.0 m the lower and upper part of the
The risk assessment showed an estimated detachment wedge. Also, to protect the buildings
amount of overpressure value of 0.0238 Kgf/cm2, in the immediate vicinity of the chimney against
a value which, in terms of the effects generated, is large pieces of material thrown or detached during

- 244 -
 Figure 12. The sequence of the chimney collapse.

demolition, protective screens made of straw bales according to each specific situation. Blasting
were placed outside them. parameters and explosive charges are sized
according to the type of material and section of the
5 CONCLUSIONS construction elements to be blasted.
The restructuring of industrial activities has as a The most commonly used method of
consequence the conservation or decommissioning demolition is the collapse of the construction itself
of constructions or areas that can later be the or overturning in a given direction. For this
object of the development of new projects. As a purpose, numerous works of structural preparation,
result, the demolition process using blasting works removal or reduction of the section of constructive
have a challenging application from the technical elements are made beforehand.
point of view but as well as due to the reduced All these approaches are described in this
time consumption, labor and cost. A large number paper and based on the blasting concept and
of demolitions works by using explosives and risk assessment needed to be taken in
characterised by a high degree of difficulty, have consideration, it is presented the practical way
shown that the use of the blasting technique is a of realizing with success the demolition by
proper alternative from the point of view of blasting works (Figure 12 and 13) of an industrial
efficiency, quality and security. chimney with 150 m height located inside of an
Each construction creates a special case, the oil refinery and in the immediate vicinity of
calculation of the blasting parameters is adapted sensitive structures.

Figure 13. Chimney after demolition.

- 245 -
REFERENCES

Kovacs, A., Laszlo, R., Gheorghiosu, E., Ilici, S. &

Vasilescu, G. 2019. Proceedings of the 19th
International Multidisciplinary Scientific
Geoconference SGEM, Albena, Bulgaria, 2019.

Laszlo, R., Radeanu, C., Ilici, S., Grecea, M. &

Levente, M. 2019. Technical and safety aspects at
the demolition by blasting works of a cooling
tower with a high of 72.00 m. 29th Symposium on
International Occupational Health and Safety,
University of Petrosani, Romania, 2019.

Laszlo, R. & Fissgus, C. 2016. Technical

documentation for demolition by blasting works
combined with classical works of the cooling
tower, from Energoterm CET Resita, Petroşani,
Romania, 2016.

Laszlo, R., Fisguss, C. & Galvitu, M. 2014. Technical

documentation for demolition by blasting works
combined with classical works of furnace no.6
from the premises of Arcelor Mittal Galati steel
plant, Petroșani, Romania, 2014.

Laszlo, R. & Fisguss C. 2008. Technical project for

demolition by blasting works combined with
classical works of 2 cooling towers from the
premises of Petrom Ploiesti Oil refinery,
Petrosani, Romania, 2008.

Laszlo, R., Eckehardt, Z. & Ille, M.I. 2005.

Demolition of metal structures. Experience
gained in the demolition of five furnaces on the
premises of Siderurgica Hunedoara Company.
Proceedings of the third EFEE World Conference
on Explosives and Blasting, Brighton, U.K.,
2005.

Tat, S., Zaporojan, M.& Fisguss C. 1985. Explosives

and blasting techniques applied in industry,
Petrosani, Romania, 1985.

- 246 -
- 247 -
 6. EU directives and
harmonisation work

- 248 -
- 249 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Implementation of track & trace solutions in existing

production environments

F. Hirthammer
TTE-Europe GmbH, Dresden, Germany

ABSTRACT: A few years ago, the term ‘track & trace’ for explosives was only known in Europe and
Brazil at most; now, the introduction of this technology is experiencing an undreamed-of boom
worldwide. As a result, more and more countries are enacting legislation to introduce track & trace of
explosives. However, this development is associated with considerable expenditure by the explosives
manufacturers. New equipment for marking the explosives and digitally recording the data must be
integrated into the production process. This should be done as inexpensively as possible, without reducing
production capacity and speed, and with as little change to the procedures as possible. In this paper, based
on real customer experiences, the ways to implement a track & trace solution are highlighted.
Furthermore, lessons learned from the experience of European explosives manufacturers are presented.
Finally, risk analysis and concrete examples in the implementation of planning in production are
pointed out.

1 FIRST STEPS − Should the track & trace system be

introduced company-wide or only for specific
The first step towards introducing a track-and- production sites and products?
trace system always takes place in the decision- − What are the minimum legal requirements to
making bodies of the explosives manufacturer. be met in the company’s own country or
The reasons for thinking about such a far-reaching concerning export countries?
decision are very diverse. There can be issues of − Should the project be implemented by the
competitiveness as well as compliance. company’s own IT department or with the
All globally active explosives manufacturers help of external service providers?
use track & trace systems to varying degrees. For − Is there a willingness to invest considerable
example, the export of explosives to the European financial resources?
Union, Brazil, and other countries is only allowed − In what time frame should the
if traceability down to the unit level is guaranteed. implementation be carried out?
In addition, the system must meet well-defined
requirements. 2 THE PREPARATION PHASE
Once the decision has been made to introduce a
track & trace system, the following fundamental Once the basic questions have been clarified, the
questions must be clarified, which serve as a basis first steps must be taken to launch the project. The
for the further project steps: basis for further project steps is creating a scope of

- 250 -
work/business requirement specification − Presentation of the expected time
document. This can be done internally or by requirements in a timeline
external suppliers. − Cost analysis for the project with a maximum
The document’s content is the precise analysis deviation of +- 30%.
of the current state of the site planned for the MVP
project and the definition of the project goals. Furthermore, concrete proposals for designing
Furthermore, the time frame for implementation the required labels for single items, packaging, and
and a budget estimation should be part of it. If an pallets must be provided in the document based on
external service provider has already been the legal requirements.
selected, the requirements for this provider must
be defined precisely. 3 PROJECT TEAM
This results in one of the first concrete project
steps. Next, the external service provider or the Once the requirements have been defined, and the
internally commissioned department draws up the basic framework for implementation has been
specifications. Then, finally, an on-site evaluation established, implementation occurs at the actual
is required. The evaluation includes an assessment production site. The following procedures have
of the initial situation at the site and the resulting proven to be effective. The appointment of an
detailed proposals for the desired implementation experienced project manager with overall
of the track & trace solution. responsibility from the customer side is crucial for
Part of the document should be: the project’s success. This person coordinates the
cooperation of all parties involved, monitors the
− The concrete project description includes the project’s progress according to the timeline, is
project phases available as a contact person to clarify questions,
− The requirements for the IT infrastructure monitors compliance with the budget, and reports
− The description of the required changes in the to the stakeholders on the project’s progress.
production processes He/she is responsible for regular reporting to all
− The analysis of the impact on productivity departments involved and communicates any
and workload for production problems that arise to the entire project team. The
− Proposals regarding the type of labelling to be project manager is responsible for reacting
applied to the explosives (based on the immediately in problem situations and initiating
applicable legislation) appropriate measures to solve the problem. He/she
− Proposals relating to the marking and code takes over the risk management and defines
capture hardware to be implemented possible red flags in the course of the project.
− Description of the relevant additional work Ideally, the project is started with a kick-off
steps within the production process to capture meeting where all parties involved participate. It is
the data (this has to be provided in detail for essential to define the project team from the
each product line/explosive individually) responsible employees of the manufacturer and the
external service providers. The sole assignment of
− Description of any necessary structural
the responsibility for the project implementation to
changes in compliance with the safety
external parties inevitably leads to the project’s
regulations for potentially explosive
failure.
atmospheres
Depending on the client’s structure, the
− Listing of the production hardware to be
following departments should be involved at a
procured or provided, the required computer
minimum.
server hardware, and the software
From the client’s side:
components necessary
− Description of the software required for − Finance
recording the data of the explosives − Engineering
− Description of the track & trace software, − IT department
including the needed storage and shipping − Site Manager
processes − Site responsible for production, IT,
− Security and backup concept to prevent data distribution
loss
− Risk analysis of the entire project From the contractor’s side (in case of
− Training plan for users, administrators, and contracting to external service providers):
the support team

- 251 -
− Contractor’s project manager the marking of explosives. The marking can be
− Responsible persons for development, done differently by applying a label, direct inkjet
administration, support printing, or laser engraving. Each of these methods
has advantages and disadvantages and must
This approach ensures a universal always be considered in the context of the
understanding of the project by all parties explosives to be marked.
involved. The project team gets to know each Depending on the products (incomplete
other and identifies the key contacts. This can selection), there will be a number of
support a professional working relationship, considerations.
especially when the project is outsourced to
external service providers. 5.1 Bulk explosives

4 IMPLEMENTATION INTO PRODUCTION Preferred type of marking: labelling of bags with

pre-printed or on the fly printed labels.
The track & trace of explosives down to the Data acquisition: use of scanning stretches
individual item level requires labelling the above the conveyor belt.
produced explosives according to the legal Challenges: integration of the applicators and
requirements of the corresponding country or the scanners into the existing production flow.
country to which the products are to be exported. Notes: labelling can be done with the help of
Even though the specific provisions of the automated applicators or by hand, depending on
individual legal requirements differ in the the production process. Therefore, a slowdown of
implementing regulations, certain essential the production process is not expected.
elements can be found in all rules.
These are: 5.2 Cartridges – general

− The marking of the explosives with a unique Preferred type of marking: depending on the
serial number diameter and the type of cartridges (dry or wet),
− The provision of a code to identify the marking can be done by applying labels, using
manufacturer pre-printed packaging material, or printing by
using inkjet printers. The application can be
In addition to this basic information, depending manual, semi-automated, or automated.
on the country, information on the date of Data acquisition: placing one or more scanners
production, the minimum shelf life, approval to read the barcodes applied on the cartridges.
information of the specific product, and many Challenges: depending on the selected marking
other features are required. These requirements method, the integration of the application
must be considered when implementing the system technology represents the most significant
and possibly combined if several export countries challenge. This concerns the drying of the
with different regulations are to be served. To cartridges after the water bath and the handling of
ensure this, the information must be suitably a large number of explosives produced per minute
placed on the products. This occurs in the and small diameter. In addition, the necessary
production process and extends by marking sorting and arranging on the conveyor belt has to
the products and recording the individual product be mastered. The capture of the barcodes requires
data. the alignment of the cartridges in the reading range
The implementation of a track & trace system of the scanners. This requires individual
consists of two main areas. On the one hand, this technological solutions, especially for small
is the provision of the necessary components in diameters and a high number of individual items
the production environment and, on the other produced.
hand, the software for realising the track & trace Notes: the selection of the preferred marking
process. method depends strongly on the production
process and the number of cartridges produced per
5 PREPARATION OF THE PRODUCTION minute. This must be considered individually and
PROCESS does not allow a clear preference. Various
methods can be used for sorting and aligning the
In the evaluation process for implementing the cartridges. These range from rotating transport
project, it is necessary to pay special attention to systems to the use of pick & place robots.

- 252 -
5.3 Booster Data acquisition: data acquisition occurs via
permanently installed scanners or hand-held
Preferred type of marking: depending on the scanners, depending on the process sequence.
diameter and the type of booster, marking can be Challenges: depending on the production
done by applying labels, printing using inkjet process, damaged pieces may have to be cut out of
printers, or laser engraving. the detonating cord when it is rolled up. The cuts
Data acquisition: depending on the number and must be noted on the reel and the box labels, and
diameter of the boosters and the type of production specified rules must be followed. It must be
(manual, semi-automated, automated), the ensured that the composition of the boxes with the
barcodes are recorded by permanently installed or corresponding number of coils complies with the
manually operated scanners. rules concerning the maximum number of allowed
Challenges: especially in booster production, cuts.
the protection class in the production rooms must Notes: the application of labels is mainly made
be observed. Based on this, scanners and printers manually. The rules to apply labels to the cord
can only be used with appropriate protective varies according to the countries’ legal
devices. requirements.
Notes: regardless of their type, the markings are
applied before production to avoid problems with 6 RISK ANALYSIS
the increased protection class of the production
The following explanations of the risk analysis
areas.
represent only a small part of the required
5.4 Detonator – general considerations. However, they point out focal
points and must be considered in the respective
Preferred type of marking: applying pre-printed project on the concrete conditions.
labels or labels printed on the fly during the The risk analysis assumes that known methods
production process to the tube (wire). recognised as inappropriate are not repeated and,
Data acquisition: depending on the type of therefore, do not have to be considered again. In
production process, data is mainly captured by addition, the examination must be based on the
permanently installed scanners. In the case of project conditions and at the respective production
manual application, these must be positioned so sites.
that no additional process steps have to be There are several ways to perform a risk
integrated into the scanning process. On the other analysis. However, it should be noted that a
hand, automated data acquisition is carried out by complete assessment of all risks is not possible
setting up scanning lines between the individual and that a residual risk must always be expected.
production steps in the transport process. The potential risk is calculated as the product of
Challenges: depending on the production the impact and the probability of occurrence.
process (manual, semi-automated, automated), risk = impact x probability of occurrence.
high demands are placed on applying the labels. A widely used method of analysis is the
For example, the tags must be applied accurately monetary designation of the impact. The resilience
to prevent them from sticking together in the of the risk analysis depends on the comprehensible
packaging, and the sometimes high production definition of the risk criteria. Another point is the
speeds must be mastered. To ensure 100% barcode definition of impact classes, which can be directly
capture, measures must be taken, especially in linked to monetary impact and dependent on the
automated production, to align igniter tags project budget.
uniformly and control the complete data capture The following risk matrix can be created as an
from all detonators. example from assessing the possible impact and
Notes: labelling can differ for the different their probability of occurrence and the resulting
types of detonators and must always be considered impact classes.
product and process-dependent. Risk matrix to be filled for four impact classes.
A small, incomplete selection of risk criteria for
5.5 Detonating cord implementing a track and trace solution is
presented below.
Preferred type of marking: applying on-the-fly External Risks - arise from problems that
printed labels to the spools and box. Inkjet print on originate outside the company itself. They can be
the detonating cord sleeve is necessary for a larger influenced or not influenced. Examples are
diameter detonating cord. weather events, pandemics, politically motivated

- 253 -
 Figure 1. Risk matrix – example.

events (not influenceable), or external service 7.1 Project Planning

providers’ selection (influenceable).
Internal Risks - must be differentiated in The customer appointed a dedicated but
various areas. These can be general events such as inexperienced project manager. In the course of
the illness of essential employees, termination, the project, he lost the overview and control of the
commercial problems such as delays in the sub-projects. This resulted in unplanned financial
approval of financial resources, work-related expenses and generated a significant time delay.
risks in construction changes or software The project manager had to be replaced during the
implementation, and many more. project. The necessary training period for the new
It is not enough to consider the possible risks; it project manager caused further delays. The
is also necessary to identify ways out for the selection of the project manager is essential to
categories to minimise the impact when problems ensure the success of the project.
almost inevitably occur. During the project planning phase, certain
The topic of risk analysis is often constraints imposed by union activity were not
underestimated in project preparation. It is observed. Therefore, the involvement of the trade
important to consider risk assessment as an unions from the very beginning of the planning is
ongoing task throughout the entire project. It is necessary.
often the only possibility to take effective
countermeasures already at the initiation of 7.2 Procurement
problems.
In the procurement process, hardware components
7 EXAMPLES FROM DIFFERENT REAL were selected based mainly on the lowest price,
PROJECTS contrary to the advice of the specialist
departments. As a result, these components had to
The following describes some situations and be replaced at a significant financial expense in the
problems from an implemented project and how project. Therefore, close cooperation between the
they were overcome.

- 254 -
purchasing and specialist departments is required accompanied by additional costs. The cause was
to find the best purchasing strategy for success. the subcontractor’s ignorance of the working
conditions in the production of explosives. The
7.3 Information technology main contractor failed to provide the necessary
knowledge through appropriate measures (visit to
The quality of an existing ethernet network in the production, talks with operators ...).
production area was deemed sufficient without
verification. However, during test operation, 8 CONCLUSIONS
problems were found with data transmission, so
the reliable functionality of the acquisition devices The introduction of Track & Trace systems at
could not be guaranteed. In addition, the explosives manufacturers is no longer the
subsequent installation of supporting components exception but the rule. The labelling and
led to project delays. These could have been traceability of products is a legal requirement
avoided by a straightforward check of the network in more and more countries. Therefore,
in advance. manufacturers are dependent on implementing
The two-way data exchange between the ERP such systems. However, it has also been proven
system (Enterprise Resource Planning) and the that the available control of the entire supply chain
track & trace solution was part of a project. leads to increased safety.
Unfortunately, due to a lack of consultation However, the complexity and effort required to
between the departments of the group, the implement such a project should not be
upcoming update of the ERP system was not taken underestimated. In addition, considerable financial
into account during the program-technical and human resources are necessary to guarantee
implementation of this task. This led to the project’s success.
considerable additional costs in terms of time and Working with an experienced external service
money. Therefore, the involvement of all relevant provider for project implementation in this
departments in a group during such a project is business area is advisable. The specifics of the
indispensable. explosives business require extensive business-
specific knowledge that cannot be acquired within
7.4 Site management a short period.
Due to limitations in scope, this document can
In the user acceptance test phase, inexplicably only provide a brief, incomplete insight. Above
high reject rates were found. As a result, from the all, it is intended to call the necessary attention not
point of view of the project team, including those to underestimate possible upcoming projects.
responsible on-site, production process changes
were introduced, which were not intended to have
any impact on production. It was not considered
that additional training sessions would have been
necessary due to the insufficient education level of
the operators. The correct assessment of the
performance and the ability of the employees to
accept changes in the usual processes must be
taken into account in advance of the employee
training sessions. The training plans must be
adjusted accordingly.

7.5 Software vendor

The main external contractor commissioned the

manufacturer of a software component to design
and deliver the user interface. Despite extensive
briefing by the main contractor, the result did not
meet the requirements in any way. The unique
conditions of use of the software were not
taken into account and could not be used in
production. An extensive redesign was necessary,

- 255 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

PECCS – Pan-European Competency Certificate for Shot firers

/blast designers

T. Tuuna
EFEE Council Member, Spark & Stone Concept

ABSTRACT: Is there such a thing as a Pan-European shot-firer? No, if they are not the collectors of
shot-firer certificates from around Europe. Is that a problem? Often, it is. The world has become smaller,
companies are big and transnational, but workforce is often immobile due to the differences between
countries, even though the job description is mostly the same or at least very similar. EFEE has taken
serious steps to change this situation. One of the main outcomes is the EFEE Education System.

1 EFEE EDUCATION SYSTEM greatly welcomed, but it was also acknowledged

that there’s still a lot of work to do.
In 2004, EFEE started with creating a ‘Training It was decided to create a group of people who
manual for rock blasting technique’ and in 2006 would start working towards another project to
they also created the ‘Training manual for finalise the training means. Finally, in 2016, after
demolition blasting’. two years of unsuccessful applications to gather
These two manuals became the base for the another funding from the EC, the PECCS –
ESSEEM (European Shotfirer Standard Education Pan-European Competence Certificate for Shot-
for Enhanced Mobility) project, which was led by firers and blast designers project started.
Karl Kure and Aslak Ravlo from the NFF - Eventually in cooperation, the EFEE Education
Norwegian Tunnelling Association in cooperation System was created – this includes not only the
with EFEE. Thanks to active participation of PECCS materials, the Online systems of PECCS
different specialists, members of EFEE, the materials and the plan to continue with the
Shotfirer Committee and with funding from Certificate, but it now has the Shot-firer
European Commission, the project was carried Committee inside EFEE and an administrator
through from 2008 to 2010. working together with the materials in the future,
The ESEEM project ended with a result of 1200 to hold the quality high, to bring in modifications
slides of materials, collected from highly regularly as the industry develops and to be in
appreciated specialists, authors and teachers constant communication with the organisations
from Europe. To make this material more who use the PECCS materials.
comprehensible, NFF organised and funded the
work to reorganise the material. The results of all 2 PAN-EUROPEAN COMPETENCE
this work were presented in Zandwoort, CERTIFICATE FOR SHOT-FIRERS
Netherlands, in 2013 for the representatives from
23 countries. The idea behind a more universal PECCS material has been put together as a Pan-
training means to harmonise the levels of shot-firer European material, which means that it includes
knowledge and competencies in Europe was information, which may not be familiar in every

- 256 -
 Figure 1. Knowledge and skills according to EQF.

country or which might seem out of place. We to the every-day work of a shot-firer, the needed
assure you, the material has been reviewed many know-how and, how to become better at the job.
times by many specialists, it has been put together At this time, these materials are intended as
by active professionals, teachers, trainers, continuing education, on EQF levels 4-5
doctorates, and shot-firers who have been in the depending on the country (Figure 1). It means the
industry for years and have much experience. The person qualified to study with this material is
material will be reviewed and updated every 2 supposed to be already a working and experienced
years. (You can check the date on the bottom of (at least 2 years) shot-firer. The aim is to enhance
the slide.) and harmonise the levels in Europe into a
The training material is divided into comprehensible outcome for all shot-firers and
Powerpoint presentations and exercises (practice educators. As the material also includes an online
questions) and will be made available to each testing system, it could be verified, which
teacher / school individually through knowledge is already there and what needs more
shotfirer.eu/PECCS. The idea behind the focus yet.
information in this material is strongly connected This material does not include any kind of

Figure 2. PECCS chapters in suggested order.

- 257 -
 Figure 3. Enhancing the shot-firer education.

legislative information from any country or area. shot-firers, who can then turn to EFEE and apply
The legislation must be learned separately in each for the Pan-European Competence Certificate for
new working area. Shot-firers/blast designers. This certificate then
At this moment PECCS includes the following will allow them to work abroad without extra
chapters, see Figure 2, but as EFEE Education training, only the necessary know how of local
System is already built up and the aim is to always regulations.
move forward, we plan to create a demolition At this moment, many educational
chapter. organisations have already taken the possibility to
use the PECCS materials. But we still need more
3 SO HOW DO WE DO THIS? cooperation to reach further over Europe.

Our aim with the PECCS materials is to spread it 4 CONCLUSION AND FUTURE WORK
through Europe, make it accessible to all
shot-firers in various countries. At the same time So in order to gain access to the PECCS materials
the Certificate, based on the materials is already as a trainer, please write to Info@shotfirer.eu – we
acknowledged by many EFEE member countries, will have a basic Agreement for signing and after
and there will be more every day. This is where we that all you need to do is create a user on
need more cooperation between educational www.peccs.datel.ee webpage, and the PECCS
entities and EFEE. We want to give the materials administrator will give you access for
for use to all educational entities, who are training downloading the materials in PowerPoint and in
shot-firers. They will have the materials free of PDF format.
charge, with simple and basic agreement, signed Every educational entity will get a different set
by both parties, this includes exercises, examining of examination questions from the administration
questions and the certificate of attendance for and the updated materials in every two years.

- 258 -
Figure 4. Places where our PECCS material is being used. (this map will be modified before the conference
according to current statistics about PECCS materials. At this moment, there’s Sweden, Germany and France –
and more countries which universities and authorities have gained access to PECCS).

Figure 5. PECCS
collaborating partners
2016-2019.

- 259 -
Meanwhile, the Shot-firer committee, the PECCS
working Group, and all active members of EFEE
will do whatever is needed to move closer to the
original goals.
Here is the list of all educational entities, who
have PECCS materials for use and who have
agreed for me to share their information: For
example, Dresdner Sprengschule.

5 ACKNOWLEDGEMENTS

All the partners collaborating in this project with

creating the PECCS material have done a vast
amount of work, sometimes without sufficient pay.
Also, we are thankful for the members of the Shot
firer Committee in EFEE, with who’s support we
could create this project.

- 260 -
- 261 -
 7. Health, safety and environment

- 262 -
- 263 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Cold case investigation - Who blasted the transmission tower and

killed four people?

M. Ruspeckhofer
Naarn, Austria

ABSTRACT: In 1967, a transmission tower, holding a 220 Kilovolt powerline, located at the Italian-
Austrian border on a mountain pass (altitude 2,300 m) was blasted by terrorists. Approaching soldiers
stepped on hidden bombs or mines. Four people were killed, and one person was severely injured.
Suspects from Austria were arrested and put on trial in Austria and in Italy. After more than fifty years, a
historian found new evidence in old records. Because of his findings the blasting of the tower and the
bombs were reinvestigated. The results exculpate the three young men (two of them still living) and
suggest a completely different story.

1 PREFACE the south. In 1961, some farmers decided that

this could not continue and started to blast
In 1967 an electricity transmission tower located infrastructure objects, especially transmission
directly at the border of Austria and Italy was towers, which were considered symbols for the
blasted. Italian forces were alarmed and went to suppression. Also, other objects were blasted, e.g.
the location. Near the tower, a young soldier was an aluminium statue of the Duce Mussolini which
killed by a mine. Four specialists were brought in was (in the 1960s!) placed next to a power plant.
and investigated the case. They found electric Finally, the Italian government backed down
installation material at the foundation of the pylon. and started serious negotiations concerning an
When they returned, they triggered another mine autonomy. Some people did not agree with these
which killed three of them and injured one. negotiations. On the one side, a certain group
That is the official Italian version of the within the BAS ‘Befreiungsausschuß Südtirol’
incident. (Board for Freedom in Southern Tyrol) who
pushed for a referendum concerning the
2 HISTORICAL CONTEXT nationality of Southern Tyrol, and on the other
side, Italian fascists who considered this part of the
After WWI, the province of Southern Tyrol was country a colony and did not want to change
given to Italy. The 98 % German speaking anything.
population was not asked if they liked it or not. Italy assumed that this terrorist attack was
The fascist regime of Mussolini did everything to committed by Austrians and applied strong
make Italians out of Tyroleans. After WWII, Italy diplomatic pressure to get appropriate results.
industrialised the country. Mountain rivers Among other things, Italy threatened to block
produced electricity. Jobs were created in great negotiations of the non-EG member Austria with
numbers and apartments were built. Just: German the EG concerning trade tariffs. The whole
speaking Tyroleans got neither jobs nor Austrian economy was at stake. Then the Austrian
apartments. These were only for immigrants from police arrested 3 suspects.

- 264 -
 Figure 1. The twisted tower.

These three young men claimed that they were between Italy and Austria (altitude: 2,300 m).
asked to take over the transport of an injured They claimed that they did not reach the pass
member of the BAS to an Austrian hospital. The because they realised in time that something was
meeting point was a mountain pass next to Mt. wrong. Suspicious from the beginning (the letter
Porze (Cima Palombino), which forms the border which informed about this task did not contain

- 265 -
certain security signs), they turned and went back. investigate the blasting side of the event.
A train ticket later helped to reconstruct the
timing. For the lawsuit, two independent certified 3 AVAILABLE SOURCES OF
experts concluded that they had a maximum time INFORMATION
span of thirty minutes to commit the crimes. The
three Austrians were put on trial in Austria and - Book by the historian Dr. Speckner,
were acquitted. In Italy they were also put on trial containing numerous files, interviews etc.
– in absence and without even informing them – - Local witnesses
and were convicted. To calm the situation Austria - Technical literature (Vogel, Oriard etc.)
ordered the army to secure the border and prevent - Historic expertise by Massak from 1967
further attacks. (as presented in the book by Dr. Speckner)
Fifty years later, a historian from the Austrian - Expertise by Dr. Melzer, static engineer
Army investigated this operation and found specialist in construction blasting
documents which suggested a completely different - Blasting instructions for the BAS (as
story. Protocols of engineers who were at the described in the book by Herlinde Molling)
location one day after the bombing stated that they - Interviews with contemporary witnesses
did not find any evidence of bombing and killing. 4 ISSUES
Lots of other information (documents from the
archives) was worked into a book which was 4.1 Why did the tower fall to the northeast when
published in 2015. However, the historian did not the ropes pull towards southwest?
look closely enough at the details of the blast and
of the bombs. When I criticised, he assigned me to The transmission tower is on the highpoint of the

Table 1. Safety distance for stray current from high voltage power lines.

- 266 -
220 Kilovolt powerline from Italy to Austria. At pocket watches (for a time delayed ignition) and
this point it also changes direction. Dr. Melzer, batteries. It is quite doubtful that a person would
expert for structural engineering, concluded that risk working with these very sensible igniters of
the ropes exert a power of 300 kg to one side. But the safety class ‘A’ (Antistatic) directly on an
if the tower is blasted as the contemporary Massak electricity tower of a 220 KV high voltage power
expertise suggested, it would not topple but stay line. Also, how long would it take during night
upright because of its weight. Only if somebody hours without any artificial light (the place can be
would remove a part of the corner profiles on the monitored from a nearby Italian army base)?
northwest side it would topple to this side. The The electric igniters of that time had a bridge
northwest is uphill. The mast toppled slowly wire resistance of 1.5 Ohm and were safe up to a
against the pull of the ropes. The outrigger was not current of 0.18 amps. Current igniters are safe up
damaged at all. There was also no stress relief of to 0.45 amps. Blasting at altitudes of 2,300 m
the power lines, so the neighbouring towers were or in the vicinity of high voltage power lines is
not damaged. Had the mast fallen to the southwest, restricted to highly insensitive igniters which are
downhill, the situation would have been quite safe up to 4 amps.
different, resulting in strong damage of the tower In the 1980s the US-blasting consultant Lewis
itself, and also the neighbouring towers due to Oriard investigated the response of 0.18-amps-
stress relief of the cables. It is therefore not very igniters when used in blasting near 220 KV
likely that the tower was blasted by ‘terrorists’. powerlines. To his surprise these high-sensitive
igniters proved safe even in unfavourable settings
4.2 The materials, how were they used? within 8 m of the powerline. But despite these
results Mr. Oriard decided to use highly
Italy presented electric installation material. How insensitive 4 amps igniters from Schaffler & Co,
was it used? Was it possible to prepare a blast with Austria, for all blasting operations near these
these materials? power lines.
The presented material consisted of several However, blasting near power lines is different
cables, plugs, remains of electric igniters class ‘A’, than using these igniters directly at a tower which

Figure 2. Phase of the blasted electricity corner tower Nr. 119, grounded, which lead to the ground fault.

- 267 -
is, at that time of the night, moist from dew. of burning. This would again support the
Especially if there is an alternative: BAS-members assumption that somebody grounded the power
usually used detonating cord, which made things line before loading.
much faster, easier and safer to work. Especially at
night. 4.4 First victim: the radio operator Amando
The registration of the ground fault by the Piva
power plant and the perception of the blasting
noise by the nearest army base differ by 35 When the first soldiers approached the site, a
minutes. It is safe to assume that somebody hidden bomb went off and killed the radio
grounded the power line half an hour before it was operator. (Official Version.) Later the experts did
blasted. If several people are working, this is a not find any hint of a trigger mechanism.
reasonable time frame for preparing a blast, and The first information from Italy to Austria
therefore a reasonable assumption. stated that: “An electricity tower was blasted.
Of four charges only three went off.”
4.3 Ground fault We can assume that in the (first) blast, one
charge did not detonate. The size of the charge
When a 220 KV-cable causes a ground fault, you for the corner profile and of the bomb: 0.5 – 1 kg
may expect some sign of burning. However, high explosive (Source: contemporary experts).
Figure 2 shows nothing like this. The power plant When a radio comes close to an undetonated
in Soverzene tried to restart operation of the power charge with a highly sensitive igniter, it is possible
line about five minutes after the first ground fault, to set off the charge (see Table 2). The radio
without success. The records show that the second operator must have been very close to the
ground fault was caused by the same line as the charge because of his injuries. Also a few hundred
first. This tells us that Figure 2 should show the metres away was another very strong radio in
site of the original ground fault (if the blasting of a vehicle.
the tower was the reason for the ground fault).
The engineer who later repaired the tower also 4.5 Second mine
concluded that the tower tilted rather slowly
because there was no damage on the outrigger, After the radio operator was killed, a group of
which would benefit the formation of an electric specialists was brought in to investigate the case.
arc. But according to picture 2 there were no signs These were trained in finding and defusing bombs.

Table 2. Technical report: Calculated safe distance for radios according to frequency and strength of the signal.

- 268 -
 Table 3. Recommended distances for 1 ohm electronic detonators from RF sources such as fixed and mobile
transmitters including cellular telephone service, amateur radio and citizens’ band. Minimum distance in feet.

Data from: Institute of Makers of Explosives, December 2011.

These four specialists stepped on a contact device The contemporary expert, Alois Massak, also
and another bomb went off (size about 2 – 3 kg found remains of batteries in the bomb crater.
TNT - official version: 6 kg). Three of the Later they were identified as batteries of a type
specialists were killed, one life-threateningly which were sold in Austria. Trials show that
injured. This suggests that these four specialists batteries near a blast of two kg TNT will simply
were walking very closely together while disappear (evaporate). Mr. Massak therefore found
searching for bombs. Quite an unusual and evidence which simply could not have been there.
unlikely behaviour. The first bomb, killing the
radio operator, went off on the upper side of the 5 A MORE REASONABLE VERSION OF
road which leads to the tower. The other bomb THE STORY
went off on the lower side. The expert at the time,
Massak, considered this a very logical array. Looking at these findings, a different version of
Massak had plenty of experience from WWII. The the crime story is more likely. Two of the three
Italian experts seemingly did not expect a bomb at Austrian suspects were well known for their
this place. But why? Massak was led to the site for activities in Southern Tyrol. So, a letter was
investigation. He found the contact device. On the written to lure them to the mountain pass. At the
contact device was a message from the BAS: “You same time, a team of blasting specialists prepared
shall never have our land!” But when the news the site. They waited for the Austrians to reach the
published the incident, one picture showed the pass. As they did not show up, they decided to
device with the message. Almost identical to the blast the tower. First, they applied some device to
original but with one spelling difference. It seems ground the power line. The night before, the power
that there were two almost similar devices. One for plant switched off the line for a few hours
the Austrian expert, and another one for the press. (maintenance work). This time could have been
Two of the victims were found more than fifty used to prepare the grounding for the next night.
metres from the bomb crater. How did they come After electricity was switched off (the line
to this place? The average inclination is about 30 - grounded), they went to work. Unfortunately, a
35 degrees with the fall line leading to a different little mistake caused the failure of one charge. This
direction. The snowfield nearby did not show any is a quite common problem when blasting on
signs of blood. The explosion seems to have torn metal structures. Sometimes the current takes a
and burnt away all clothing, leaving strong signs shortcut through the metal structure. Somehow the
of burning. This would not happen with a buried radio operator came close to this charge and set it
charge of roughly 2 kg. Trial blasts did not show off. Later the tower was once more charged and
any signs of burnt clothes on dummies. blasted. This time three charges were placed on the

- 269 -
northeast side. That was the ‘safe side’, and it Speckner, H.: Zwischen Porze und Rosskarspitz.
determined the direction of the fall. The accident
was declared a bomb attack. At the same time, Speckner, H.: Von der ‘Feuernacht’ zur Porzescharte.
another accident (perhaps with explosives, or a
traffic accident, or an air crash) happened Speckner, H.: Unpublished materials.
elsewhere. The corpses were brought to the
mountain pass. There were rumours that there was
an accident during a mine-training lesson at the
army base at Kreuzbergpass (another mountain
pass nearby). So, two independent events or
accidents were combined to tell a new story. That
was not unusual at the time. The author can prove
that an alleged bombing of a custom guard
building was in fact a deflagration of heating gas
from the kitchen.

6 WHO WAS RESPONSIBLE FOR THIS

INCIDENT?

This question cannot be answered by a blasting

expert. But the historian Dr. Speckner offers
circumstantial evidence.
A fragment of a letter was found by the
engineers of the electricity company at the mast.
This fragment contained an address of a known
member of the radical Gladio organisation.
Dr. Speckner therefore assumes that the
blasting of the mast was performed by members of
some fascist organisation with the purpose to
hinder progress in autonomy negotiations.

REFERENCES

Beyling-Drekopf: Sprengstoffe und Zündmittel.

Bureau of Mines: A study of rf hazards at low and

medium frequencies to blasting in underground
coal mines. Vogel, G.: Zünden von
Sprengladungen.

Hasler, H. 1967. Wissenschaftliche Neubeurteilung

der Rechtssache Porzescharte 25. June 1967.

Institute of Makers of Explosives: Safety guide for

the prevention of radio frequency radiation
hazards in the use of commercial electric
detonators (blasting caps).

Leidel, S., Shimer: RF hazards to electric detonators

– The New IME SLP-20.

Molling, H.: So planten wir die Feuernacht.

Oriard, L.L.: Explosives engineering, construction

vibrations and geotechnology.

Schaffler et al.: Die elektrische Minenzündung.

- 270 -
- 271 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Key learning from Beirut ammonium nitrate accident - Evaluation

of warehouse safety management in Indonesia

A. Wibowo, & A. Sutoyo

PT Multi Nitrotama Kimia, Jakarta, Indonesia

ABSTRACT: The ammonium nitrate explosion in Beirut on August 4th, 2020, was a wake-up call for
ammonium nitrate stakeholders in the mining and construction sectors, particularly regarding the deadly
risks associated with explosives raw material management. The explosion destroyed most of the city of
Beirut, and was caused by the unsafe, prolonged storage of 2,750 tons of ammonium nitrate. The
explosion resulted in over 100 of fatalities, destroyed surrounding buildings, and incurred losses
amounting to billions of dollars. This does not include the thousands of people injured, and lengthy
socio-economic impacts on the country, made worse by the COVID-19 pandemic. In Indonesia, there are
nearly 100 explosives warehouses managed by warehousing service companies, mining companies, and
explosives providers. They are mostly located within the mining facilities, or in specialised warehousing
areas. This study was conducted to analyse the potential, risks, and probability of a similar incident
occurring in Indonesia.

1 INTRODUCTION detonation through impact or friction. Pure

ammonium nitrate is difficult to detonate under
The incident in Lebanon is a cautionary case ambient conditions of temperature and pressure.
regarding the deadly risks associated with Flame, spark, rough handling, impact or friction
explosives management. On August 4th, 2020, are not known to cause a propagated detonation.
an explosion shocked the city of Beirut, An explosion of ammonium nitrate can be initiated
predominantly caused by about 2,750 tons of with high explosives under ambient conditions
ammonium nitrate, which was stored unsafely and (Western Australia - Code of Practice Storage
for a prolonged period, in the same facility as Solid Ammonium Nitrate 2021).
other, incompatible chemicals (Seddon & Shiotani In Indonesia, there are almost 100 commercial
2020). According to World Bank Group (2020), explosives warehouses managed by mining
nearly 200 people lost their lives, and damages companies and licensed explosives companies.
were estimated to be between $3.8 billion to $4.6 These warehouses are located in both mining and
billion, excluding the losses incurred. The blast industrial areas, and have various capacities,
damaged structures up to 6.5 miles away from the ranging from 50 tons to 7,000 tons of ammonium
centre, and nearly 40,000 buildings in total were nitrate and with varying environmental conditions.
affected. With so many of these high-capacity facilities, it
Ammonium nitrate (chemical formula can be said that there is a possibility of a similar
NH-4NO3) is commonly used as an explosive in accident occurs in Indonesia.
mining industry, and as a fertiliser in the Ammonium nitrate in Indonesia is mostly used
agricultural industry. On its own, it is a very stable in the mining industry, particularly in coal mining,
compound and quite insensitive to accidental where the consumption of ammonium nitrate is

- 272 -
 Figure 1. Methodology of study.

around 400,000 tons per year. This number the Western Australian Code of Practice had been
remained steady in the last two years and is used as a reference (Step 2). Evaluation of
expected to remain that way for the next five warehouses at risk (Step 3) is carried out on
years. According to the Indonesia Department of warehouses that have indications or potential to
Energy and Resources (2021), Indonesia produced have large-scale (major) incidents. In order for the
up to 600 million tons of coal in 2021, and study to remain within the scope, the risk being
approximately 75% of the production is on the assessed is limited to the impact on the
island of Kalimantan. surrounding environment, particularly regarding
The scope of this study is limited to the sample the risks of fire and explosion. There are several
of ammonium nitrate warehouses in Indonesia key terms to be analysed, such as:
with capacities of 200 tons up to 5,000 tons; both
end-user and transit warehouses. This study − Possibility of an activity that triggers a fire
excludes warehouses belonging to ammonium − Possibility of incompatible materials stored in
nitrate production facilities in Indonesia. the warehouse or nearby
− Possibility of warehouse fires escalating into
2 METHODOLOGY widespread fires (local emergency plan)
Using the Beirut incident as a learning material, − Possibility of explosion of the warehouse
this study analyses the best practice of storing − The impact of the explosion on the
ammonium nitrate as one of the important factors surrounding environment
that caused the explosion in Beirut. Moreover, one − Emergency response.
of the main indications of the explosion in Beirut
was the prolonged storage of AN and indications The analysis used in this study is best practice
of incompatible materials being stored in the same by analysing the worst case scenario of each
warehouse. warehouse, especially warehousing that is outside
This study was carried out by conducting a the mining lease.
warehousing survey, inspection of things in the According to Western Australia Code of
warehouse and arrangement of materials in the Practice Storage of Solid Ammonium Nitrate 4th
warehouse, then compared with International edition (2021), there is a hazard in relation of
Standards (from other countries), in this case using explosion (in term of Ammonium Nitrate storage).
standard from Australia. This evaluation is mainly When heated by an external fire, the explosion
carried out for several warehouses located in sensitivity of the decomposing melt increases
Kalimantan, and Karimun Island. dangerously with increasing temperature and is
The flowchart of this study is described in further increased:
Figure 1.
To strengthen the analysis, the standards used − if the AN is impure, or the melt mixes with
are not only those currently enforced in Indonesia; contaminants and/or fuels; and/or

- 273 -
− if molten AN becomes confined in drains, inspection of Indonesian warehouses, located in
pipes, plant or machinery and the both industrial areas and mining lease. The
decomposition gases cannot escape freely. National Police Chief No. 17/2017 which
Decomposing AN can also explode if regulates Permit, Security, Monitoring and
mechanical shock impacts the hot melt. Control of Commercial Explosives regulation is
Fires involving AN have caused explosions, used for the permit and control of explosives in
but there have been more fires involving AN general. In the mining areas, the explosives
without explosions, depending on the warehouse management system is regulated by the
circumstances. Ministry of Mining and Energy’s Decree No
209/2018 Regarding Explosives Safety and
The risk of an explosion is decreased by Blasting Practice for Indonesia Mining Industry,
reducing the potential for the AN to be: which states that explosives warehouses must be at
an adequate distance away from surrounding
− heated, such as in a fire communities and other storage facilities; the
− contaminated regulation also regulates the overall management
− confined. system, and analysis of potential impacts on the
surrounding environment. These two regulations
Given the nature of modern formulations of are inter-correlated.
AN, explosions of solid AN (excluding those Safety management of AN warehouse from
initiated by explosives) without prior fire is these regulations are then compared to the
unlikely. If all potential sources of fuel can be Australian Standard, in this case the Western
eliminated, the chance of an accidental explosion Australia Department Mines, Industry Regulation
is remote. and Safety; Code of Practice – Safe Storage of
Solid Ammonium Nitrate 4th edition (2021) being
3 STUDY ANALYSIS referred to.
The data obtained are listed in Table 1. The
The study was conducted by survey and physical data is a sampling of several mining areas with

Table 1. Initial data for ammonium nitrate storage in warehouses.

Distance of
Other & ammonium
Ware- Ammonium type of Capacity Distance to nitrate
Type of
# house Province nitrate dangerous other DG Community warehouse to
Location
Name volume (kg.) goods (kg.) (m) another DG
nearby warehouse
(m)
1 W1 Coal Mine East Kalimantan 120,000 DG Class 1 10,000 7,000 20
2 W2 Coal Mine East Kalimantan 4,000,000 DG Class 1 200,000 2,000 120
3 W3 Coal Mine East Kalimantan 500,000 DG Class 1 200,000 5,000 100
4 W4 Coal Mine East Kalimantan 350,000 DG Class 1 5,000 5,000 50
5 W5 Coal Mine East Kalimantan 2,500,000 DG Class 1 50,000 10,000 250
6 W6 Coal Mine Cent. Kalimantan 2,000,000 DG Class 1 100,000 6,000 700
7 W7 Coal Mine Cent. Kalimantan 1,000,000 DG Class 1 20,000 10,000 150
8 W8 Coal Mine South Kalimantan 500,000 DG Class 1 5,000 10,000 30
9 W9 Coal Mine Cent. Kalimantan 1,000,000 DG Class 1 15,000 10,000 50
10 W10 Coal Mine South Kalimantan 550,000 DG Class 1 30,000 4,000 50
11 W11 Quarry Riau Island 300,000 DG Class 1 75,000 2,000 100
12 W12 Coal Mine South Kalimantan 260,000 DG Class 1 3,000 20,000 20
13 W13 Coal Mine Cent. Kalimantan 200,000 DG Class 1 1,000 16,000 20
Standalone
14 W14 complex South Kalimantan 3,000,000 DG Class 1 200,000 600 50
Standalone
15 W15 complex South Kalimantan 1,000,000 - 2,000 0
Standalone
16 W16 complex East Kalimantan 5,000,000 - 2,000 0
Stand lone
17 W17 complex East Kalimantan 2,000,000 - 650 0

- 274 -
 Figure 2. Ammonium nitrate warehouse.

general warehouses representing the overall warehouse surveyed and evaluated. The
condition of explosives warehousing in Indonesia warehouses are commonly made from steel
and stand-alone warehouses. There was total 17 structure, have a concrete floor, natural lighting

Figure 3. Ammonium nitrate stacking in the warehouse.

- 275 -
and ventilation. No electricals are allowed in the Apart from the safe distance criteria tabulated
warehouse. Ammonium nitrate stored mostly low above, there are other standards according to
density type in in a bulk bag of 1250 kg, 1000 kg Western Australia Code of Practice Safe Storage
or 25 kg. Most commonly used is 1250 kg bag. of solid ammonium nitrate – (4th edition) and the
Ammonium nitrate warehouses in mining areas National Police Chief Decree No 17/2017 that has
are located in a specific complex, where other been compared to that provided below (but are not
explosive materials are stored in other buildings. limited to):
Separation distance is maintained according to
Indonesia regulations. Typical explosive materials − Warehouse buildings are to be made from
(DG Class 1) stored include pentolite boosters, non-combustible material and with adequate
emulsion cartridges, and detonators. All of those ventilation
materials are to be kept in separate buildings. − Floors shall be made of concrete
Whilst ammonium nitrate warehouses located − The warehouse design must have adequate
outside the mining areas are stored in specialised natural lighting
and dedicated locations. Most of ammonium − Dedicated ammonium nitrate - warehouses
nitrate warehouse in Indonesia is designed as are to only be used for ammonium nitrate
stand- alone dedicate steel and concrete building (no other material allowed within the same
structure (Figure 2 & Figure 3). building)
All warehouses within the mining area are at an − Equipped with a lightning protection
adequately safe distance from the surrounding system
communities rather than warehouses outside the − The area is clear of vegetation or materials
mining lease areas (Table 2). that can cause fire (at a distance of 10 m)

Table 2. Actual distance vs distance as per regulation and standard.

# Warehouse Ammonium Actual distance Distance to Distance to Distance to

name nitrate to community community community Other DG
volume (kg) (m) according to cccording to Class 1
Australia standard Indonesia warehouse
(m) *) regulation (m) **) s (m)
1 W1 120,000 7,000 560 509 20
2 W2 4,000,000 2,000 1,805***) 700 120
3 W3 500,000 5,000 890 700 100
4 W4 350,000 5,000 790 647 50
5 W5 2,500,000 10,000 1,546***) 700 250
6 W6 2,000,000 6,000 1,436***) 700 700
7 W7 1,000,000 10,000 1,143***) 700 150
8 W8 500,000 10,000 890 700 30
9 W9 1,000,000 10,000 1,143***) 700 50
10 W10 550,000 4,000 938***) 700 50
11 W11 300,000 2,000 750 597 100
12 W12 260,000 20,000 750 581 20
13 W13 200,000 16,000 660 557 20
14 W14 3,000,000 600 1,642***) 700 50
15 W15 1,000,000 2,000 1,143***) 700 0
16 W16 5,000,000 2,000 1,943***) 700 0
17 W17 2,000,000 650 1,436***) 700 0
*) Distance according to Department of Mines, Industry Regulation and Safety, 2021, Safe Storage of solid
ammonium nitrate – code of practice (4th edition) reissued: Department of Mines, Industry Regulation and
Safety Western Australia, 35 pp
**) Maximum Safety distance according to Indonesia Regulation, SK Dirjen ESDM No 209.K/2018 " Safety
Explosives and Blasting Practice for Indonesia Mining Industry. (page 30-34)
***) distance of warehouse over 500 ton of ammonium nitrate is calculated based on “D = 17.8 Q1/3 for resi-
dential buildings including hotels, motels and other accommodation. Where Q is NEQ of explosives (0.3*total
AN stored)

- 276 -
 Table 3. Identified warehouses having potential safety issues.

# Ware- Ammonium Distance to Community Distance to other DG Note

house nitrate warehouses
volume Actual Distance Distance Actual Distance Distance
(kg.) distance according according to distance according according
(m) to Indonesia (m) to to
Australia Regulation Australia Indonesia
Standard (m) Standard Regulatio
(m) (m) *) n (m) **)

Issue in regard
distance AN
1 W14 3,000,000 600 1,642 700 50 35 21
warehouse to
community
Issue in regard
distance AN
2 W17 2,000,000 650 1,436 700 - - -
warehouse to
community

*) Maximum Safety distance according to Indonesia Regulation, SK Dirjen ESDM No 209.K/2018 " Safety
Explosives and Blasting Practice for Indonesia Mining Industry. (page 30-34)
**) Distance according to Department of Mines, Industry Regulation and Safety, 2021, Safe Storage of solid
ammonium nitrate – code of practice (4th edition) reissued: Department of Mines, Industry Regulation and
Safety Western Australia, 35 pp

− No smoking, no naked flames and other 4 RESULTS

activities that cause fire. Any hot work
activities shall be managed according The results of the evaluation of the sample
to SOP. warehouses against both Indonesian Regulations
− There are fire extinguishers available outside and Australian Standards are presented in Table 3.
the warehouse building and security post, as
well as adequate fire hydrants − Inadequately distanced warehouse W14 from
− There are three keys for each warehouse. community
The keys are kept by the Warehouse − Inadequately distanced warehouse W17 from
Keeper(s), Security Officer(s), and Police community
Officer(s) - specifically Indonesia National
Police Regulation The next step is identification any control
measures to ensure risk of being in fire situation
− No electrical systems (equipment, wiring and
and explosion is reduced. This means that there is
lighting) allowed (specifically Indonesia
a risk happen to community, especially quantities
National Police Regulation)
of ammonium nitrate stored in W14 & W17 are
− There is a security post, and CCTV cameras quite significant (3,000 tons and 2,000 tons,
installed in and around the warehouse - respectively), and the fact that these warehouses
specifically Indonesia National Police are relatively nearby to villages, roads, and other
Regulation - specifically Indonesia National public facilities. At least 20,000 people live within
Police Regulation a 10 km radius of both W14 and W17.
− Warehouses must have adequate Apart from the distances, both W14 and W17,
telecommunications equipment (to as well as other warehouses in Indonesia have
communicate to emergency services and local adapted good explosives warehousing practices,
authorities) - specifically Indonesia National because the warehouse keepers have complied
Police Regulation with the National Police Chief Regulation (2017)
− Warehouse Keeper to be trained and certified and have also maintained 24-hour warehouse
in Managing Explosives Warehouses (by the supervision. In the case of these warehouses, it is
Police Force) - specifically Indonesia important to make sure that current controls (Table
National Police Regulation. 4) are maintained.

- 277 -
 Table 4. Risk assessment result, current control and further action.

Warehouse Major Risk Control in Place Action

W14 – issue Small Fire Supervised by security officer for 24 hours; Reducing quantity of AN to reduce
with safety knockdown to any possible fire will be quickly identified. possibility of explosion (less than 25
distance to the building SOP and warehouse management are in % of current capacity).
community and storage place. Review Regulatory Standards;
Fire Emergency System in place, but bigger consider 'good practice’ or
fires may require Local Emergency International Standards.
Services. Emergency drill with Local
DG nearby is considered a safe distance Emergency Services to be conducted
away (with safety mounding in place); the on a regular basis.
possibility of knock down effect is Regular monitoring
unlikely.
W17 – Issue Small Fire Supervised by security officer for 24 hours Reducing quantity of AN to reduce
with safety knockdown to hence any possible file will be identified in possibility of explosion (less than 25
distance to the building timely manner. % of current capacity) Review.
community and storage; No incompatible material stored nearby. Regulatory Standards; consider 'good
escalated into SOP and warehouse management are in practice’ or International Standards.
bigger fire and place. Emergency drill with Local
explosion Fire Emergency System in place, however Emergency Services to be conducted
bigger fire requires Local Emergency on a regular basis.
Services. Regular monitoring.

The fact that all warehouse keepers are trained − High standard regulations regarding the
and certified in Managing Explosives Warehouse storage of explosives shall be maintained, it
(by the police force) is a good start. should cover consideration of the knockdown
However, to ensure safety in both warehouses effect caused by fire or incompatible
in question, it is recommended to reduce the materials that are able to initiate ammonium
quantity of ammonium nitrate stored, in order to nitrate material.
meet both Indonesian and International Standards. − Availability of fire emergency system and fire
Regulation or Standard review must also be brigade that are able to reduce the potential of
carried out in order for reduce explosion risks and small fires and uncontrolled fires.
knock down effect to community.
REFERENCES
5 CONCLUSION
Department of Mines, Industry Regulation and
There are two (2) warehouses identified as critical Safety, 2021, Safe storage of solid ammonium
based on this study, due to significant amounts nitrate – code of practice (4th edition) reissued:
of ammonium nitrate stored and inadequate safety Department of Mines, Industry Regulation and
distance. Also, in this perimeter there are some Safety Western Australia, (pp 35).
schools, markets and other public facilities
at risk (if there is a failure of current control Indonesia National Police. Peraturan Kapolri
systems). Republik Indonesia No 17 (2017). Chief National
There are some takeaways to consider, to avoid Police Regulation No 17/2017 regarding Permit,
a similar explosion from occurring in Indonesia. Security, Monitoring and Control of Commercial
They are as follows: Explosives. (pp.65-70).

International Chemical Safety Card. International

− Management of ammonium nitrate warehouse
Labor Organization (ILO). Ammonium
shall also refer to international standard as
Nitrate Properties (ISCS:0216).
additional best practice. https://www.ilo.org/dyn/icsc/showcard.display?p
− SOP to ensure safety requirement shall be in _lang=en&p_card_id=0216&p_version=2.
place. This covers ownership of ammonium
nitrate products which are to be made clear, Ministry of Energy and Mineral Resources Republic
and owners must be held responsible for its Indonesia. Surat Keputusan Direktur Jendral
safekeeping (including warehouse security, Energi dan Sumber Daya Mineral. Mining and
control, and monitoring). Energy's Decree No 209 (2018). Explosives

- 278 -
 Safety and Blasting Practice for Indonesia
Mining Industry. (pp. 30-34).

Seddon. B & Shiotani. H. 2020. The Beirut Port

explosion: understanding its impact and how to
reduce risks from explosive precursors. Geneva,
Switzerland: UNIDIR. (pp. 4-24).

Wibowo, A. 2022. Ammonium nitrate explosion at

Beirut and key leaning to improve supervision
and management of explosives managerial in
Indonesia. ISEE 2022.

World Bank Group (2020). Beirut rapid damage and

needs assessment. Washington DC. (pp. 23).

- 279 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

The effects of dynamic pressure in blasting

G. Cavanough & A. Torrance

QMR, Brisbane Australia, Kilmorie Consulting, Cooks Hill, Australia

ABSTRACT: Effective blasting requires delays between the firing of blast holes to allow movement of
the burden and good fragmentation. Detonating holes will induce a pressure in adjacent undetonated
holes. This dynamic pressure may be large enough to either crush the detonator and lead to a failure of the
internal firing chain, either pyrotechnic or electronic, or be sufficiently high to cause detonators to fire
sympathetically and initiate the explosive column at the incorrect time. Failure of the detonator will result
in a misfire. The pressure may also interfere with the explosive consistency and can sufficiently
desensitise the product to cause poor detonation performance, misfire and potentially a fume event.
Tensile dynamic pressure events may also cause back-break. This paper presents field measurements into
the potential contribution of dynamic pressure to the misfire of detonators, back-break and the generation
of post blast fume.

1 BACKGROUND Breakage of ground occurs at much lower

levels of tensile force as compared to compressive
Misfires are an unwanted blasting event but are force. Hence tensile dynamic pressure can
still a regular occurrence in a wide range of mining contribute to back-break.
and quarrying applications. They are more likely This paper presents research into the potential
to occur in wet formations or if clay seams are contribution of dynamic pressure to the misfire of
present. This paper reports on field measurements detonators, back-break and the generation of post
of dynamic pressures in large scale blasting blast fume.
activities in open cut mining and quarry
operations. In addition to the pressure level, the 2 PRODUCT ‘IN HOLE’ SENSITIVITY
velocity of propagation of the pressure between
holes is quantified. This provides a method to Most emulsion explosive products use micro
understand ways to limit the risk of misfire by balloons or chemical gassing to provide
using the appropriate inter-hole timing and primer sensitisation. Detonation performance and the
position in the explosive column. ability of bulk emulsion explosive products to be
Post blast fume may be observed when using initiated is influenced by the instantaneous density
high density emulsion products in soft ground of the product. The density is in effect a measure
conditions. Generally, no fume is observed if of the sensitivity. The lower the density the more
ANFO is used in place of emulsion in this rock heat generated by the compressive shock wave
type, provided it is not damaged by water ingress. propagating along the explosive column which in
The propensity of higher density and VOD turn ensures complete and sustained detonation of
products to generate fume in soft ground is the explosive column. Normal porous ammonium
attributed to lack of confinement. ANFO is nitrate used in ANFO has a bulk density of around
relatively unaffected by dynamic pressure whereas 0.75 g/cc. Higher, non-porous Prill with a density
emulsion will be desensitised under pressure. of up to 1.10 g/cc may also be used. The lower

- 280 -
density ANFO is capable of absorbing fuel sensitisers may be added to emulsion in some
throughout the Prill and is easier to initiate than specific examples to provide a constant density
the high density Prill for which the fuel sits on the over the entire length of the explosive column. The
surface. The low density allows ANFO to have addition of micro balloons increases the cost and
a high degree of sensitivity and is constant also the risk of having to transport a sensitised
throughout the borehole. product.
In contrast, ammonium nitrate based emulsions When gassed emulsions are placed under
that are sensitised by chemical gassing have a pressure, the density increases as the gas bubbles
varying density along the length of a borehole. compress and eventually dissolve in the emulsion,
There is an increased hydrostatic pressure at the in a similar manner to CO2 gas in water to make
bottom of the long explosive columns that soda water. When the pressure is released, the
increases the density of the explosive in that emulsions will return to the original density.
location and can cause a decrease in the likelihood Therefore, to desensitise a gassed emulsion
of successful detonation. A theoretical density of requires the detonation to occur when the emulsion
these products can be calculated from first is under pressure. This is readily observed in the
principles, taking into account the density of the field when detonation cord is used to initiate
product, its compressibility and frictional effects emulsion products. Poor overall product
on the walls of the blasthole, but this is rarely performance and post blast fume is often the result
tested in the field. Compression testing in due to deflagration. Deflagration occurs instead of
laboratories shows that the resulting emulsion is detonation due to the much higher shock wave
always denser, meaning a loss in gas volume velocity of the detonation cord pressurising the
within the product causing the product to emulsion at the time of detonation.
desensitise. In the case of micro balloon sensitisation, the
Glass or plastic micro-balloons mechanical application of pressure can break the micro

Figure 1. Examples of back break in open cut and underground mines.

- 281 -
balloons and desensitise the product. explosive column. It is necessary to locate the
sensor close to the detonator to measure the
3 DETONATOR FAILURE pressure at this critical position. Figure 2 shows a
QMR Dynamic Pressure Sensor located above the
Detonators general sympathetically detonate at primer.
high pressure (>350MPa) due to high pressure and Electrical disturbances were used to determine
physical deformation of the detonator shell detonator firing times to allow calculation of the
initiating the base charge. Crushing and shrink propagation velocity of the dynamic pressure,
wrapping may occur at a lower pressure initiation of the primer and products and
(~190MPa) and prevent the detonator from firing identification of sympathetic detonations.
due to damage to the internal firing chain.
6 DYNAMIC PRESSURE
4 BACK-BREAK
Dynamic pressure is induced in a blast hole by the
Back-break induced from blasting adversely detonation of adjacent holes. The velocity of
affects the pit walls in open cut mining and causes propagation and the peak pressure level decreases
dilution and instability in underground mining. with an increase in ground hardness. Figures 3 and
Most importantly, the safety conditions of a back- 4 show a schematic of dynamic pressure events
broken region can be significantly compromised measured in a large open cut iron ore mine. A hole
and it must be addressed to prevent injury to is initiated by top and bottom primers with
personnel and/or damage to equipment. This identical delay times. The detonation induces a
occasionally results in the necessity to alter mine dynamic pressure in an adjacent hole with a
plans to prevent an incident from occurring in a propagation velocity of 400 msec-1. The duration
back-broken region increasing costs. Figure 1 of the dynamic pressure event is the same as
shows examples of back-break in open cut and duration of the detonation. The dynamic pressure
underground mines. was 100 MPa at the bottom sensor and 20 MPa at
the top sensor indicating top priming is a method
5 INSTRUMENTATION to reduce the risk of detonator failure due to
dynamic pressure. Pressure is readily transmitted
QMR Dynamic Pressure Sensors are placed in the
by water and clay seams with dynamic pressure
explosive column. This provides a true measure of
levels higher in soft ground conditions.
the pressure level at that particular location in the
In wet clay formations the propagation velocity
can be less that 500 m sec-1 at pressure greater
than 150 MPa.
Sustained dynamic pressure events were also
measured. Figure 4 is the same data as in Figure 2
with a second event after the pressure event
described in Figure 3. This event had a maximum
pressure of 110Mpa and was the same level by
both sensors.
Figure 5 shows dynamic pressure
measurements taken in a hard rock quarry. The
rock structure contained near vertical clay seams.
Detonation of the holes on the bottom bench
induced a sustained dynamic pressure in the top
bench. Propagation velocity was calculated from
the start time of the events at 350 msec-1.
Figure 6 is a dynamic pressure event at a large
open cut coal mine. The product used plastic
micro-balloons for sensitisation. The detonation of
the adjacent holes induces a typical dynamic
pressure pulse with a 55 MPa maximum pressure
at the sensor. There was no detonation pressure
signal from the sensor indicating the product did
Figure 2. QMR Dynamic Pressure Sensor taped to not initiate. Analysis of the associated data
downline just above the primer.
confirmed the bottom primer detonated but failed

- 282 -
Figure 3. Typical dynamic pressure induced by detonation of adjacent hole. Note, pressure is lower at the top of
the explosive column (i.e. 100 MPa at bottom and 20 MPa at top).

Figure 4. The short duration event as per Figure 2 were followed by a longer duration event that was the same
pressure (maximum 110 MPa) at both top and bottom sensors.

- 283 -
Figure 5. Detonation of middle section and resulting dynamic pressure experienced at top section with
identical time slices.

Figure 6. Dynamic pressure caused desensitisation of the product in the toe of the hole resulting in only top
primer initiating the explosive column and creating a significant fume event.

- 284 -
 Figure 7. Electrical disturbance and dynamic pressures measured in the back row hole of ground with
significant back-break.

to initiate the explosive column. The dynamic the bottom primer and deflagration induced by
sensor cable was cut at the precise delay time of detonation of the top of the explosive column.
the top primer which confirmed that the top primer Figure 7 is a dynamic pressure event at a large
initiated the explosive column. This blast resulted open cut coal mine in a hole in the backrow of a
in significant post blast fume. It is likely the blast that resulted in significant back-break. The
dynamic pressure crushed the micro-balloons firing times of adjacent holes was identified and
preventing initiation of the explosive column by allowed calculation of the propagation velocity of

- 285 -
the dynamic pressure as approximately 950 msec-1.
The dynamic pressure has a tensile pressure of
16.8 MPa which is sufficient to break ground of
hardness of up to 100 MPa compressional strength
and hence is the likely cause of the back-break.

7 CONCLUSION

Dynamic pressure is induced in a blast hole by the

detonation of earlier fired holes. The velocity of
propagation and the peak pressure level decrease
with an increase in ground hardness. Pressure is
readily transmitted by water and clay seams with
key features of the pressure pulse being:

− The pulses are the same duration as the

duration of detonation of the hole that is
creating the pressure
− Is at higher pressure at the bottom of the
explosive column
− Arrival time at the top of the explosive
column is later than at the bottom detonation
of the explosive column.

Large ground movement can result in long

duration events that are the same pressure level
over the entire explosive column.
Dynamic pressure can result in sympathetic
detonation, misfire, back-break and fume events
and should be measured and used to develop delay
timing of blast patterns to prevent these events.

REFERENCES

Cavanough, G., Onederra, I. & Torrance, A. 2011. In

situ measurement of detonation pressure and
temperature of commercial explosives. European
Federation of Explosive Engineers Lisbon
Conference 2011, 19-20 September 2011, pp
391-400.

Torrance, A., Cavanough, G.L., Lusk, B. & Seo, M.

2019. Large scale measurement of velocity of
detonation. International Society of Explosives
Engineers 45th Conference, January 27-30, 2019,
Nashville USA.

- 286 -
- 287 -
 8. New applications and training

- 288 -
- 289 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Blasting activities with fast-combusting energetic compositions in

dimension stone quarries and for demolition of structural elements
in civil engineering

P. Shishkov, N. Stoycheva & V. Penev

University of mining and geology St. Ivan Rilski, Sofia, Bulgaria

ABSTRACT: In some cases of resources extraction, as well as in construction at industrial and urbanised
territories, commercial explosives are not safe enough for the surroundings with regard to the generated
fly-rocks, air-blast, toxic fumes, seismic waves and vibrations. The main reasons for these harmful
impacts of explosion are the velocity and the mechanism of the chemical reaction of explosive
decomposition. The authors shifted the focus of their research from detonating explosives to high-speed
combusting energetic materials. The major focus of the study is the production of low explosive
non-detonating mixtures from long-term stored smokeless gunpowders and ammonium nitrate prills in
different configurations, and popular pyrolant compositions. The samples of different cartridge casings,
filled with non-detonating propellant mixtures or pyrotechnic compositions were examined with two
methods – measurements of the velocity of propagation and field tests. The application of waste
single-base-powders and double-base-powders for creation of non-detonating explosive cartridges,
suitable in dimension stone mining, as well as in blasting activities in areas with increased requirements
was examined. Keywords: non-detonating blasting cartridges, propellants, cautious blasting, dimension
stone extraction, blasting demolition

1 INTRODUCTION explosives range between 2500-7500 m/sec.

Detonation pressure is the pressure in the reaction
1.1 About explosive chemical decomposition zone when explosive molecules break down. The
latter is an important indicator of the explosive’s
The velocity of detonation (VoD) is the rate at ability to induce a good fragmentation (Mitkov
which the detonation wave travels through the 2010 (A))
explosive charge. The higher this velocity, the Deflagration is a subsonic reaction of chemical
greater the ‘force’ or the crushing effect of the decomposition of the explosives. It is typical for
explosive. Higher velocity explosives are better propellants and pyrolants, which could act on the
suited for blasting in hard rocks, and low-velocity solid medium by the pressure generated by
explosives are appropriate for work in soft and gaseous products from chemical reaction. There is
cracked rocks. In general, ‘low-speed’ explosives practically no shock energy generation. In reality,
tend to release gaseous products for a relatively such an effect occurs when a charge of blasting
longer period, and therefore exhibit better heaving gunpowder is ignited in an appropriately tamped
effect. Detonation rates of different industrial blast hole (Shishkov & Stoycheva 2018).

- 290 -
1.2 About pyrotechnic compositions additives like dibutylphtalate or dimethylacrylate
or camphor (phlegmatisers), diphenylamine
In accordance with definition by (Boychev & (stabiliser), KNO3 or K2SO4 (pore-forming salt),
Asenov 2020 B) pyrotechnic compositions are a graphite, remaining alcohol-ether solvent, etc.
substance or mixture of substances intended to Double-based propellants (DBP) are mixtures of
produce special effects - heat, light, sound, smoke, nitrocellulose (NC), 2-8 % additives (similar
gas, or a combination thereof, as a result of to these in SBP) and nitro-esters, usually
spontaneous exothermic red-ox reactions that do nitroglycerin (NG). The content of NG is about 10
not require oxygen from external sources. The - 38 %. The boiling point of NG is 50 oC. DBP
main ingredients are oxidiser and fuel (reducer). might have higher energy content than SBP. Their
Various additional modifiers could be present in caloric content varies depending on their types
the mixtures, regarding needed effect. Depending between 3800 and 5200 kJ/kg. The grains of DBP
on the speed of the chemical reaction, several are also with different sizes and shapes. They
different forms of chemical conversion could could be compact or porous regarding the type.
occur – thermal decomposition, burning mode, The prices of waste SBP and DBP in Bulgaria are
combustion, deflagration, atypical detonation and around 0.10 EUR/kg (Shishkov & Stoycheva
detonation. The factors that determine which form 2018).
would develop are: The authors focused their researches to
laboratory experiments for combustion rates and
− chemical nature of the components; field tests for explosive performances of
− degree of dispersion; high-speed combusting energetic materials on the
− homogeneity of the mixtures; base of pyrotechnic compositions and secondary
− presence of impurities, especially moisture; propellants.
− type and power of the initiating pulse;
2 EXPERIMENTAL SETUP
− location of application of the initiating pulse
in the volume of the charge;
− quantity of the bulk mixture;
− density of the composition;
− degree of the confinement of the mixture;
− construction of the pyrotechnical device
(Boychev & Asenov 2020 B)

High-performance propellants, imitation

mixtures, photo mixtures, and so called "’lash
powders’ are types of pyrotechnic compositions
which are suitable for achieving a gas-generating
high-temperature effect in a time range of
milliseconds.
In 1992, Kuwahara and Ochiai introduced the
term ‘pyrolants’. These are materials, typically
including metallic or non-metallic fuels (Al, Mg,
Ti, Si, B, S) and inorganic [Ba (NO3)2, NaNO3,
KClO4] and organic [C2Cl6, (C2F4)n] oxidants, in
the combustion of which emit bright light and
condensed products (hot gases). They are
characterised by a high enthalpy of combustion
(1-30 kJ/g) and a density in the range of 2-10
g/cm3.

1.3 About waste smokeless gunpowders

Figure 1. Top - general image of DBP and SBP
It is well known, that single based propellants type. Below - after grinding in the mill.
(SBP) generally are pelletised or extruded porous
grains in different sizes and shapes, which usually The propellants, investigated in this study are
contains 93 – 97 % pyroxylin and 3 – 6 % nitroglycerine gunpowder (further referred to as

- 291 -
DBP) with the brand name NDT-3 18/1 and the professional pyrotechnic purposes, manufactured
18/1-branded pyroxylin SBP. The test specimens by ‘META_PYRO’ s.r.o., Czech Rep.
are in form of long tube-shaped bodies, and are The velocity of deflagration was measured by
further processed by grinding after their extraction two different devices:
from decommissioned ammunitions (Fig. 1). The
size of propellant grains should be similar to the − with apparatus Trio Chronos, manufactured
size of the porous prills of ammonium nitrate as by TRIO Electronics Ltd., Rep. of Serbia,
prevention against stratification of the ingredients. using optic fibre sensors.
Three different compositions were designed
and mixed for further research at the laboratory
testing field, at a stone quarry, and for demolition
of hardened concrete:

− Mixture #1: flash-powder composition 68%

KNO3 + 32% Al (dark) with Oxygen Balance
= -1.55%
− Mixture #2: 50% grinded SBP + 50%
NH4NO3 prills with Oxygen Balance =
-11.00%
− Mixture #3: 40% grinded DBP + 55%
NH4NO3 + 5% Al (dark) with Oxygen
Balance = -8.65%
Figure 3. apparatus for measurement of VoD
Ready compositions were loaded in aluminium optical sensors.
testing tubes, plugged at both sides (Figure 2) with
the following dimensions:
- with apparatus CNT-66 Pendulum,
− 320 mm / ф10 mm inner diameter/ 1 mm wall manufactured by BRL Test Inc., USA, using
thickness contact wire impact sensor.
− 320 mm / ф20 mm inner diameter/ 1 mm wall
thickness

Each pipe has 3 drill holes: the first one is for

electric ignition, the second one is for sensor №1
(located 3 cm from the first hole) and third for
sensor №2 (located 25 cm from the second hole).
The ignition of the samples was made using
regular commercial electric bridge-wire igniters
with smooth burning fuse-head for fireworks and

Figure 2. Test samples - loaded and plugged Figure 4. Apparatus for measurement of VOD
aluminium pipes with electric igniter. and sample with impact sensor.

- 292 -
 Sensors and the electric starter were connected and detonation of that sensitive flash-compositions
to the testing equipment. All laboratory are higher density, bigger diameters, and larger
experiments were done at the laboratory testing volumes of the charges. The authors decided to use
area of Minproekt - Dragichevo. petards with a smaller diameter than the blast-hole.
The existing air gaps between the decoupled
3 RESULTS AND DISCUSSION charges and the walls of the bore-hole prevent
increasing the pressure to dangerous rates for
The velocity of explosive decomposition for the transition from combustion to detonation. Very
different samples are given in Table 1. low concentrations of an explosive charge in the
drill-holes between 0.10 kg/m3 and 0.15 kg/m3 has
Table 1. Experimental results for velocity of to be achieved for split-blasting without undesired
explosive reaction of different samples. crack propagation. This is the reason, in the case
of bore holes longer than 1.0 m, multi-deck
Sample Inner diameter Mixture Velocity charges are a better choice for smooth blasting.
number of the test type of reaction
tube (mm) (m/s)
1 10 #1 522.31
2 10 #1 10.88
3 10 #1 224.46
4 10 #2 202.35
5 10 #2 289.15
6 10 #2 250.82
7 10 #3 348.71
8 10 #3 284.26
9 10 #3 304.62
10 20 #1 1625.32
11 20 #1 747.39
12 20 #1 698.75
13 20 #2 491.71
14 20 #2 410.34
15 20 #2 463.16
16 20 #3 598.21
17 20 #3 482.76
18 20 #3 554.16

The results show that pyrolant Mixture #1

(flash-powder) provides to higher velocities of
explosive decomposition and in bigger diameters
of the charges is inclined to pass from combustion
- to detonation, which is not acceptable to
industrial requirements for low-explosive charges.
Propellant containing compositions Mixture #2 Figure 5. Splitting of a boulder with one charge
(with SBP) and Mixture #3 (with DBP) are of 0.05 kg. Mixture #1 in single blasthole with
length 0.60 m.
increasing their velocities of deflagration with
enlarging the diameter of the charge, but samples
containing DBP and Aluminium are releasing All petards were equipped with traditional
more energy, as expected. fuse-head electric igniters, usually applied in
After laboratory experiments, some field tests professional fireworks events. Each device was
at the stone quarry were conducted. Few boulders protected with water-resistant polymer foil. These
with similar shapes and sizes were selected for electrically controlled low-explosive cartridges
testing of blast-splitting capabilities. Regarding were used like bottom charges in single or double
the risk of transition from deflagration to the parallel blast holes, drilled in stones with volumes
detonation of ‘flash powder composition’ (Mixture between 4 - 6 m3. The selected rock boulder
#1), small charges of 0.050 kg was poured samples were free standing (with free surfaces)
without compaction in well-plugged paper tubes and similar shapes. Proper stemming has a leading
with inner diameter 25 mm and 150 mm length. role in the quality of the performance of the non-
The conditions for possible shock wave generation detonating explosives in the blast-splitting

- 293 -
 Figure 6. Propellant based cartridges with Mixture #3 in plastic foil bag.

activities. When we prefer to deprive from the pyrotechnical boosters. 3-4 g of pyrolant Mixture
advantage of the supersonic pressure wave, which #1 and electric igniter was used for the preparation
causes a sharp impact on the rock, we have to of each petard, which was intended to provide the
ensure long enough retention of the explosive boosting energy for massive inflammation of our
gases pressure in the drill hole. foil-cartridges with Mixtures #2 and #3. As it is
Very good results were achieved after treating described by Boychev & Asenov (2020 A) during
the boulders with a single charge of the deflagration of flash-mixtures, fine metal particles
aforementioned petards in a short borehole with a of the fuel are evaporating and due to the high
diameter 45 mm and length 0.60 m. A smooth- temperatures and pressure, they transition into a
walled cleft was obtained throughout the volume plasma state. Driven by the high pressure of the
of the stone. No additional cracks were observed hot gasses, the sparks and plasma are penetrating
in the area of the blast hole (Figure 5). deep through the gaps between the bigger grains of
After the experiments with flash powder the propellant. In this way, several layers are
composition, further tests with receipts containing ignited simultaneously in the depth of the charge.
secondary smokeless powder were conducted. For The combustion of propellant mixtures starts with
the field tests at the stone quarry Mixtures #2 and a higher initial rate and pressure, which in the
#3 were used for the preparation of 0.050 kg and conditions of tamped drill-hole contributes to a
0.100 kg charges in thin plastic bags fitted to the shorter period of increase of the velocity of the
diameter of the drilled holes (Figure 6). As it is explosive reaction. In practice, a higher
known, smokeless gunpowders have so-called deflagration rate means a larger amount of hot
progressive combustion, the speed of which gases, emitted per unit of time. This affords better
depends on the pressure in the burning area, the performance of the low-explosive.
power of the ignition pulse and the surface of New experiments were performed with charges
initial ignition. For this reason, the authors of Mixture #2 (in plastic bags) on boulders.
preferred to ignite those propellant charges not Pyrotechnical booster was loaded first on the
directly by electric igniters, but with small bottom of the 0.60 m deep drill-hole. Plastic

Figure 7. Splitting of a boulder with one charge of 50 g Mixture #2 in single blasthole with length 0.60 m.

- 294 -
 Figure 8. Splitting of rock body with two parallel boreholes. Effects from preloaded bottom charge of 300 g.
Mixture #3 and decked second charge of 150 g. Mixture #3 in every single blasthole with length 1.80 m.

cartridge with 0.050 kg of the tested composition in the blast-holes. The total explosive quantity of
with a diameter close to the hole size was tightly each blast-hole was 0.300 kg propellant mixture.
inserted over the booster. Then, the blasthole was Reliable inert stemming between charges and
filled with reliable stemming. There was an especially over the second charge to the mouth of
excellent clean splitting of the stone after a the blast-hole assured so desirable condition for
low-noise explosion with a slight displacement of enough pressure of gaseous products. That
the two parts (Figure 7). guarantees as better as possible crack formation
The last experiment was done for the extraction for splitting.
of a stone block with a volume of almost 8 m3 by Although all measures have been taken to avoid
splitting from a rock body with three free faces. detonation when using propellant composition, the
Taking into account more complex conditions for combination of high-speed deflagration with too
explosive influence, the authors preferred to use large volume of gases (generated by preloaded
two longer bore-holes and cartridges from packed bottom charges) caused an unnecessary throwing
in a thin polymer bag Mixture #3. The length of effect and several additional cracks in unexpected
the blast holes was 1.80 m. For better distribution directions (Fig. 8). This was a sign that a more
of the energy of compressed gases, the explosive precise calculation of the parameters of drilling
charge was separated into two parts with inert and blasting was needed.
stemming between them and simultaneous After field tests at the stone quarry, an in-situ
ignition. Two handmade cartridges weighing 0.100 experiment for fragmentation of hardened concrete
kg each with one electrically ignited pyro-booster inside the mixing barrel of the concrete truck was
were placed for bottom charge in every hole. They made (Figure 9). The processes for crushing of
were well sealed with 30 cm semidry sandy-clay consolidated cement mixtures and non-reinforced
stemming. One pyro-booster, equipped with concrete are identical to those in the fragmentation
electric igniter and single cartridge of 0.100 kg of rocks. In them, intense cracking can be caused
Mixture #3 was used for the second decked charge even only under the influence of rapidly expanding

Figure 9. Concrete mixer – exterior and interior with consolidated concrete inside.

- 295 -
 Figure 10. Loading of the charges inside the blast holes.

gaseous products. Experiments for blasting of For the design of the blasting-plan, the concrete
consolidated concrete by non-detonating charges, body was considered similar to a dug foundation
without allowing deformations of the metal of non-reinforced concrete. In order to achieve a
housing, are the next step in the study of the more gentle blasting effect with the successive
potential of low-speed energetic compositions for opening of free surfaces, a model of the bench
delicate blasting not only in the extraction of blasting by the method of small charges was
dimension stones, but also for special blasting provided. Considering the delicate technical
works. conditions of the site, an opportunity was sought
The available hardened concrete mixture was for a choice between explosives that do not
distributed along the entire length of the barrel, detonate. In this particular case, a pyrolant
filling 37.2% of its total geometric volume. explosive was chosen. Flash powder composition
The remaining free volume of 62.8% was (Mixture #1) provides a relatively large volume of
inconvenient, but sufficient to carry out gaseous products per kilogram of explosive and a
preparation for special blasting works. Given the sufficiently short period of rate increase to
specific conditions of the site and the irregular deflagration. Small charges of 0.050 kg was
shape of the concrete body, which was located in poured without compaction in well-plugged paper
the lower part of the barrel, the blasting for tubes with inner diameter 25 mm and 150 mm
swelling the concrete was carried out in stages in length. Blastholes with diameter of 30 mm and
several sectors to preserve the structural integrity lengths from 0.35 to 0.60 m were used. Charge
of the mixer. construction – column type.

Figure 11. Fragmentation of the concrete after blasting.

- 296 -
 In order to achieve sufficient cracking of the some rupture damages or over-breaks around the
concrete body, the relative distance between the charged area.
individual blast holes should have been the same. In-situ experiments for application of fast
A staggered layout of the blasting pattern was combusting pyrotechnic composition for
chosen. Due to the fact, that non-detonating blast-swelling of concrete in the volume of the
pyrotechnic charges were used, the closer spacing mixer have shown satisfactory results – good
of 0.30 m was determined. Electric igniters for fragmentation and absence of damages on the
professional pyrotechnic purposes, connected in a mixing barrel. There was no fumes emission,
series circuit, were preferred as the most suitable air-blast or fly-rocks after explosion of pyrolant
means for inflammation. Their number was composition.
determined depending on the blasting rounds
parameters. The main goal was to avoid 5 ACKNOWLEDGEMENT
overloading the structure of the concrete-mixer
The authors would like to express their sincere
caused by instantaneous expansion of the solid
gratitude to the management and employees of the
material during blasting. Manual loading with an
laboratory of MINPROEKT – Dragichevo,
individual approach to the blast holes was
Bulgaria; to RIOLIT B – Plovdiv; and to
preferred. Charges were placed to the bottom of
OGNENA HRIZANTEMA LTD - Panagiyurishte
the blasthole with the help of a tamping rod
for provided sites, personnel and equipment during
(Figure 10); for shorter boreholes were used single
laboratory and field experiments.
charges of 0.050 kg. For longer holes double
charges were applied. A reliable inert stemming REFERENCES
was made.
Excellent swelling fragmentation of treated Boychev, Y. & Asenov, S. 2020 (A). Non-lethal
concrete was performed after blasting with devices with multi-sensory action,
pyrolant charges (Figure 11). Proceedings of the annual scientific
conference of Military university Vasil Levski,
4 CONCLUSIONS vol.10, 28-29.05.2020, Veliko Tarnovo,
Bulgaria, ISSN 1314-1937, p. 224 – 230 (in
Researches for application of waste SBP and DBP Bulgarian with English abstract).
after utilisation of decommissioned ammunitions
for obtaining of non-detonating explosive Boychev, Y. 2020 (B). Pyrotechnic compositions
cartridges, suitable for dimension stone mining, as for non-lethal noise flash devices, Proceedings
well as for blasting activities at challenging and of conference “Technics. Technologies.
complex conditions. The velocities of propagation Education. Safety. 2020”, vol. 3, year IV,
of the reaction of chemical destruction of tested Borovets, Bulgaria, ISSN: 2535-0315 (Print),
3 different high-energetic compositions were p.178-180 (in Bulgarian with English
between 202 and 747 m/s depending on the abstract).
diameter of the charges and ingredients.
Kuwahara, T. & Ochiai, T. 1992. Burning rate of
Propellant-based samples do not show any
magnesium/TF pyrolants. Kogyo Kayaku, 53
tendency for transition from combustion to
(6), p.301-306.
detonation in case of ignition with soft burning
electric fuse-head. Mitkov, V. 2007. Manufacturing of explosives for
Pyrolant Mixture #1 shows higher velocities of civilian purposes. Publishing house “Sveti
deflagration and in bigger diameters of the charges Ivan Rilski”, Sofia, ISBN 978-954-353-049-6
is inclined to pass from combustion to detonation, (in Bulgarian).
which makes it useful mainly for decoupled
charges, mini-pyro-boosters or multi-deck charges Mitkov, V. 2010 (A). Safety manufacturing and
from chained petards with small diameter, usage of explosives (Bezopasnost pri
separated with air gaps. proizvodstvo i upotreba na eksplozivi).
In case of splitting with multiple blast-holes in Publisher: Izdatelska Kashta “Sveti Ivan
a row, for achieving more smooth and equable Rilski” - 343 p., Sofia, ISBN 978-954-353-
cracks in necessary cut-planes, the spacing and 131-8 (in Bulgarian).
collateral between drill-holes should be précised.
Despite the fact, that explosive is not Mitkov, V. 2010 (B). Assessment and risk
detonating, a preloaded blasthole could cause management of malicious acts aimed at

- 297 -
 potentially hazardous Hydrotechnical
constructions. Conference proceedings from
the International conference of Blasting
techniques, Stara Lesna, Slovak Republic, pp.
212-220, ISBN 978-80-970265-2-3, (in
English).

Mitkov, V. 2011. Development and research of

new charges from utilized ammunition
(Razrabotvane i izsledvane na novi zariadi ot
utilizirani boepripasi), Geology and mineral
resourses Nr. 4, p. 30-33, ISSN 1310-2265 (in
Bulgarian).

Shishkov, P. & Stoycheva, N., (ed.10-11), 2018.

Advanced blasting techniques for extraction of
dimension stones, Mining and geology
magazine (MDG Magazine) ISSN 0861-5713
(in Bulgarian with English abstract), circle
10-11, pp.42-49.

- 298 -
- 299 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Optimising blast hole loading with MWD and 3D image analysis

B. Gyngell, T. Buschjost, T. Worsey & G. Diehr

Strayos Inc, Sydney, New South Wales, Australia

ABSTRACT: In recent years, Smart Drills have enabled precise GPS hole navigation along with the
generation of rich Measure While Drilling (MWD) data to provide a new perspective on subsurface
conditions (Sandvik 2021). Over the same period, drones have enabled the collection of a visual,
geometric, and hyperspectral geological data. Integrating these new data types into the blast design
process allows blasters to optimise fragmentation and prevent flyrock by tailoring each segment of a blast
hole to its specific conditions (Epiroc Rock Drills AB 2019). However, the application of these new
technologies has not yet gained widespread adoption in day-to-day practices, largely due to the
operational complexity introduced. To address this issue, this research sought to establish a new workflow
for optimising hole loading using MWD and drone data that reduced operational complexity rather than
increasing it. The objective was to identify a practical and robust process that could be incorporated into
everyday operations rather than just special projects.

1 INTRODUCTION techniques are described for data capture, analysis,

blast design, and blast performance measurement.
The purpose of this research was to establish a This represents a collaborative effort between
new practical and robust process for incorporating the site leadership, the blasting contractor, the
Measure While Drilling (MWD) data into drilling OEM and the blasting software provider to
everyday blast design processes. In particular, create an efficient process for achieving measured
the team focused on evaluating a method for improvements in blasting outcomes.
identifying seams by combining MWD data from In closing, the paper discusses how these
multiple holes with 3D contextual data. This techniques can be applied to facilitate more
research lays the groundwork for enabling custom streamlined implementation of variable energy
hole loading by seam as well as AI-powered seam loading. It also presents the future opportunities
detection. that machine learning will create for automating
Firstly, this paper provides an interpretation seam detection and charging design tailored to
framework for MWD data and an understanding of rock conditions.
how MWD can be used to gain unique insight into
rock mass properties. A list of common MWD 2 MWD INTERPRETATION FRAMEWORK
parameters is presented with definitions.
Next, the paper outlines a real case study Measure While Drilling (MWD) data refers to the
demonstrating a new workflow for using MWD sensor data collected from production drill rigs
data to identify different strata bands and apply during operation (Schunnesson 1990). This is
this to make better loading decisions. Practical generally comprised of geospatial data (e.g.

- 300 -
 Table 1. Most common data fields for MWD data.

Data field Data type Description

The vector distance from the hole collar position to the
Geospatial
Hole depth current drill bit position measured in ft or m. Note: This is
(MWD)
not just the elevation component
Geospatial The velocity of the drill bit as it moves through the rock
Penetration rate
(MWD) during drilling measured in ft/s or m/s.
Pressure The piston pressure used to power the hammer in Top
Percussion pressure
(MWD) Hammer or DTH drills
Pressure The hydraulic pressure exerted on the drill stem to move
Feed pressure
(MWD) the bit in the direction of drilling
The compressed air pressure exerted to push the crushed
Pressure
Flush pressure rock from the drill hole up the outer annulus of the drill
(MWD)
pipe and out of the hole
Pressure The hydraulic pressure applied to create torque on the drill
Rotation pressure
(MWD) stem and rotate it as it moves through the rock
An estimate of the rock hardness calculated through a
Calculation
Rock hardness proprietary algorithm based on the other Data Types
(MWD)
recorded. This may not have units.
Hole collar position determined by known GPS receiver
Hole collar latitude,
Geospatial location mounted on the drill and known displacement of
longitude and
(HNS) the drill bit from the receiver based on control system
elevation
information.
Hole toe latitude, Hole toe position calculated from the known Azimuth and
Geospatial
longitude and Inclination of the drill bit at the known Hole Collar
(HNS)
elevation Position.

penetration rate), pressure data (e.g. feed pressure) HNS offering out of the box. Synchronising
and calculation data (e.g. rock hardness) (Scoble this system with site survey control can
Peck & Hendricks 1989). It is logged at regular require an additional service.
intervals (~every 1” or 2 cm) down the hole as the - After-market MWD
drill operates (Epiroc Rock Drills AB 2019). - Older drills can be retrofitted with MWD
Additionally, MWD-enabled drills are also often sensors by dedicated service and hardware
equipped with a Hole Navigation System (HNS) providers
which captures GPS information on the hole collar - After-market HNS
position and projected path (Sandvik 2021). - Most drills can be retrofitted with a HNS by
A summary of the most common output data the major GPS system providers
fields can be found in Table 1.
This machine information is collected by a 3 METHODOLOGY
range of different hardware devices and control
systems; however, the output structure and format Our research methodology followed six main
are relatively standardised across all providers steps, detailed herein.
(Mining Editor 2021). The main options for
collecting MWD data are listed below. 3.1 3D photogrammetric data collection

OEM MWD The equipment used comprised of:

- Most new drills from major OEMs come
equipped with the sensors to capture MWD - Off-the-shelf drone with internal RTK GPS
data. Activating the data collection from these system (RTK not used for this project).
sensors require be an additional service. - GPS Rover connected with mine survey
control system.
OEM HNS
- Most major OEMs partner with GPS system First, a series of five Ground Control Points
providers to offer the option for an integrated were marked out on the bench surface and floor

- 301 -
 Figure 1. Drone photo locations and ground control points.

below. These points were surveyed with the GPS The design was then exported directly from the
Rover to provide reference points for the 3D cloud-based blast design platform in the OEM’s
model photogrammetry process. IREDES file format. This was then uploaded
Next, the drone was flown over the bench to wirelessly to the Smart Drill through the web
collect a set of 57 overlapping images to be used portal provided for the OEM’s communication
to generate the 3D survey model. These comprised service.
of two flight modes: an autopilot flight mission
taking seven passes of nadir photos of the bench 3.3 Measure while drilling data collection
surface and floor; and a manual flight mission
taking two passes of oblique photos looking at the The equipment used for capturing MWD data was
bench face. Figure 1 shows the locations of each a Smart Drill fitted out with OEM drill sensors and
photo taken (white circles) and Ground Control GPS Hole Navigation System (HNS). It was
Points (crosses). synchronised with the mine site survey control
system to match the data used for the 3D model
3.2 3D photogrammetric processing and drilling Ground Control Points.
design The holes were drilled using the HNS system
to ensure accurate placement on the bench.
The images and Ground Control Point data were As drilled HNS hole locations were compared
uploaded to a cloud-based photogrammetry and with the design locations in the blast
blast design platform to generate a survey grade design platform to ensure drilling quality
3D photo model of the bench. control.
The team then used this 3D photo model to The MWD data fields captured were Time Tag,
design the drill pattern with the tools built into the Depth Tag, Antijamming State, Feed Pressure,
same software platform. The team designed an 8 ft Flushing Pressure, Penetration Rate, Percussion
× 9 ft (2.4 m × 2.7 m) three row staggered pattern Pressure, Rock Detect, Rotation Pressure, Rotation
with 29 holes as shown in Figure 2. Front row hole Speed, Stabilator Pressure, Flushing Flow State,
locations were optimised by analysing 3D burdens. Engine rev/min, Automatic Drilling, Flushing

- 302 -
Figure 2. Drill pattern layout.

Figures 3. MWD data (penetration rate) in context of 3D photo model.

- 303 -
 Figure 4. Blast loading design performed in context of 3D burdens and MWD data.

Level, Feed Level, Rock Contact Level, Tools the blasting software which aggregate and analyse
Load Level. data from across all holes in the shot.
Averages were taken across planes at each
3.4 Measure while drilling seam analysis depth interval in the shot to identify trends in
the bench geology.
The MWD and HNS data were pulled wirelessly
from the drill back into the cloud-based blast 3.5 Contextual blast loading design
design platform for analysis.
The platform processed the MWD data and The final step of the process was to integrate the
automatically combined it with the 3D photo MWD and 3D survey model insights into the blast
model using the HNS information. Figure 3 loading design process.
shows the Penetration Rate extracted from This was achieved by visualising the MWD
MWD data visualised in the context of the parameters as colour-coded bars beside the Burden
3D photo model. Profile in the loading design module of the cloud-
Seam detection was also supported by based blast design platform as shown in Figure 4.
quantitative analysis of the metrics at each depth For each hole, the explosive products were
interval. This was conducted using tools built into adjusted based on the data collected.

Figure 5. 3D photogrammetric model of muck pile with fragmentation analysis.

- 304 -
 Table 2. Blast hole loading determined by burden and MWD analysis.

Hole 1 Poured ANFO to 31’ 3” Packaged Emulsion to 14’ 2.5” Packaged Emulsion to 10’, Stem
Hole 2 Poured ANFO to 28’ 3” Packaged Emulsion to 12’ 2.5” to 10’, Stem
Hole 3 Poured ANFO to 28’ 3” Packaged Emulsion to 12’ 2.5” to 10’, Stem
Hole 4 Poured ANFO to 10’, Stem
Hole 5 Poured ANFO to 25’ 3” Packaged Emulsion to 20’, poured ANFO to 15’ 3” Packaged
Emulsion to 10’, Stem
Hole 6 Poured ANFO to 18’ 3” Packaged Emulsion to 10’, Stem
Hole 7 Poured ANFO to 16’ 3” Packaged Emulsion to 12’ 2.5” Packaged Emulsion to 10’, Stem
Hole 8 Poured ANFO to 10’, Stem
Hole 9 Poured ANFO to 28’ 3” Packaged Emulsion to 18’ 2.5” Packaged Emulsion to 10’, Stem
Hole 10 Poured ANFO to 31’ 3” Packaged Emulsion to 26’, Poured ANFO to 20’ 3” Packaged
Emulsion to 12’ 2.5” Packaged Emulsion to 10’, Stem

3.6 Artificial intelligence post-blast analysis 4 RESULTS

After the blast was fired, the muck pile was The seam analysis identified two distinct seams
immediately flown with the drone to capture within the bench geology:
post-blast images for blast performance analysis.
These images were uploaded into the same cloud- - A soft seam from 9.5–15 ft (2.9–4.6 m)
based blast design platform to create a post-blast - A hard seam from 25–29.5 ft (7.6–9.0 m).
photogrammetric 3D model with automatically
generated Fragmentation and Muck pile All front row holes were custom loaded
Movement analyses as shown in Figure 5. based on burden and MWD data as shown in
In this case, Ground Control Points were not Table 2.
required, and the 3D model was created using the Tailoring the blast loading to the MWD data
drone’s onboard GPS. To enable accurate Muck and 3D Burden data resulted in high quality blast
pile Movement analysis, the elevation of the outcomes. Fragmentation and Muck pile
post-blast model was calibrated to the pre-blast Movement analysis results are shown in Figures 6
model using a common visible marker. and 7 as well as Table 3.

Figure 6. Results of automated fragmentation analysis.

- 305 -
 Figure 7. Example cross-section generated by automated muck pile movement analysis.

been increased due to soft ground or voids and

Table 3. Results of automated muck pile movement parts where it could be decreased because of hard
analysis.
ground.
This concept of a ‘variable minimum burden’ at
Pre-blast volume 3782 yd³
different depths within a hole, based on rock
Post-blast volume 4749 yd³
properties and explosives options, has the
Swell factor 1.26
potential to bring significant safety benefits
Average centre of mass 37.8 ft to the blasting industry in preventing
movement flyrock and should be further explored in
Maximum throw 51.2 ft future research.
Average throw 164.4 ft The approach established in this research also
Average trough drop 19.8 ft represents a key enabler for more widespread use
of differential energy loading. If variable density
gassed emulsion product was available, the team
5 ANALYSIS could have designed the explosives column in
decks of different densities side-by-side with the
The results indicate that incorporating MWD for MWD output as in Figure 4. This could have
seam identification and blast loading optimisation allowed for a more streamlined charging operation
can be done in a practical way for everyday in the field by avoiding the need to switch back
drilling and blasting operations. and forth between Poured ANFO and packaged
This workflow produced improvements in blast decks within each hole.
outcomes and increased efficiency in drilling
operations by eliminating the requirement to 6 CONCLUSIONS
manually measure and lay out the holes.
During this research, the team identified that This research achieved its purpose of establishing
the usage of Post-Processed Kinematic (PPK) a practical and robust method for applying MWD
drones could further streamline the field data to optimise blast hole loading in everyday
operations by removing the need to mark out operations. This lays the groundwork for a wide
Ground Control Points. scope of future work optimising efficiency and
The key enablers identified throughout the safety of blasting operations.
research process for implementing this approach New research should build on this by
were as follows: investigating the impact that a ‘variable minimum
burden’ based on rock properties has on flyrock
- Strong collaboration between site teams, and front row fragmentation.
drilling contractors and blasting contractors. Additionally, future research should look to
- Focus on quality over speed while getting establish AI algorithms for automatic seam
processes up and running. detection from MWD data. This could use a
- Minimising data touch points and automating combination of pattern recognition and clustering
processes where possible. methods to automate that process step.
All these steps lead toward a future where
Interestingly, tailoring the blast loading with highly tailored and optimised blasting is
the visualisation of MWD and 3D burden data accessible across more sites. Using technology
highlighted several situations where a loading to unlock new workflows that are practical
decision made purely based on burden would not as well as powerful will be the key to
have been optimal. There were parts of holes lifting the industry to the next level of safety
where the minimum burden threshold should have and efficiency.

- 306 -
7 ACKNOWLEDGEMENTS

We would like to sincerely thank Sam Howrigan

and Shelburne Limestone Corporation for making
their site and data available to conduct this
research.
We would also like to thank Strayos, Inc. for
making their cloud-based photogrammetry and
blasting software platform available for processing
the data and performing the analysis/blast design.
We would also like to express thanks to
Sandvik for providing Smart Drill hardware and
expert input on the details of MWD interpretation.
Finally, we would like to thank Nathan Rouse,
Dyno Consult and Mike Allen, Martin Marietta for
their feedback and guidance for MWD data
applications.

REFERENCES

Epiroc Rock Drills AB, 2019. Reference Book:

Drilling in Surface Mining, Quarrying and
Construction, Örebro, Sweden.

Mining Editor, 2021. Sandvik evolves drill rig to

support fully autonomous operation, Australasian
Mine Safety Journal. Retrieved August 2021
from https://www.amsj.com.au/sandvik-evolves-
drill-rig-to-support-fully-autonomous-operation/.

Sandvik, 2021. Choosing the right rig for the job.

Retrieved August 2021 from
https://go.rocktechnology.sandvik/l/490131/2020
–05–01/5ryhdb.

Schunnesson, H, 1990. Drill process monitoring in

percussive drilling. Luleå University of
Technology, Division of Mining Equipment
Engineering.

Scoble, M. J., Peck, J. and Hendricks, C. 1989.

Correlation between rotary drill performance
parameters and borehole geophysical logging,
Mining Science and Technology, 8(3), 301–312.

- 307 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

The art of detonation – energetic creation

P. Bosch
Member of the Dutch Association of Explosives Engineers

ABSTRACT: I am a visual artist from Belgium. Since the start of my practice, I have a great interest in
working with energetic materials. In the beginning, this mainly resulted in art projects in which the
visitors interacted with light and fire. Then I started to use pyrotechnics, and my activities evolved into
happenings with explosions of paint/ice I initiate the transformation caused by energetic material and tried
to visualise this in an aesthetic way. I use energetic material as an art medium to create sculptures,
installations, performances and videos. Recently I realised there was still unexplored territory in the art
landscape and wondered what was needed to get there. In order to be able to deal with high energetic
material knowingly and skilfully, I took courses in working safely with explosive substances. I received
my diplomas in performance art, pyrotechnics, firework, civil safety and blasting engineering. According
to civil law, I operate as a blasting engineer and allowed to blast small objects such as rocks, tree stumps,
bridges or structures up to 6 metres. Up until now, I have mainly been focused on discovering, defending
and enabling the energetic material as art medium. My interest is the yet to be discovered creation
possibilities of high-energy material. This research, the use of energetic materials as an art medium, is
supported by the department of Arts and Culture.

1 IMPRESSIONS from a project in which the useless but beautiful

fragmentation qualities of ice are displayed. For
1.1 Canvas this work (Figure 2) pyrotechnical electric squibs
were ignited with 9 v firing system.
‘Canvas’ is an installation with a traditionally
looking set-up of painting material that falls apart
1.3 Black powder
before the painter could start. This installation
questions the use of the canvas and cherishes the
‘Black powder’ was a live performance, that
moment when it comes to an end, and tries to
allowed 100 protected participants to experience a
show this in a way typical to the material. For this
controlled pyrotechnical deflagration of black
work (Figure 1) pyrotechnical electric squib was
powder fuse. These performances safely showed
ignited with a 10 cap blasting machine.
the almost surreal and typical properties of
1.2 Ice experiments energetic material, and made it possible to
experience them in a positive way.
‘Ice experiments’ is a video in which naturally red For this work (Figure 3) pyrotechnical electric
coloured ice blocks in different arrangements matches and 3000 metre rapid fuse based on black
are being pulverised. The video shows a selection powder were used.

- 308 -
Figure 1. See the ‘Canvas’ installation video by scanning the QR code.

Figure 2. See the ‘Ice experiments’ installation video by scanning the QR code.

- 309 -
 Figure 3. See the ‘Black powder’ installation video by scanning the QR code.

2 ART OF DETONATION The places where I will actually be able to

create this kind of work I have yet to discover and
2.1 Methodology are probably military blasting sites, private test
facilities or quarries. For all the experiments I will
I am developing a series of 100 low tech use a similar protocol that renders the phases
experiments with high energetic materials as an art visible: before, during and after initiation and
medium to examine the full range of the creation impact on the surroundings. Related concepts
aspects. The application within the segment ‘art associated with the use of this medium are;
industry’ is a niche market, nevertheless I see initiation, detonation, discharge, launch, gravity,
enormous potential. The use of shock tubes, perimeter, light, shockwave, speed, time and
detonation cord, detonators and a range of space. Illuminating the concept of explosives from
different charges could be applied to regular art ‘100 different angles’ and thus being able to
media like clay, stone, metal, paint and wood. expand the rather limited perception and
Expanding the terminology associated with high connotation surrounding it, to high-energy
energetic materials by for example study the work moments. I would like to exhibit the experiments
and life of Alfred Nobel as inventor of dynamite to a wider audience once they are finished.
and the founder of the Nobel prize I consider as a
part of this path to deal with the duality of high 2.3 Experiments
energetic material and the perception around it.
Also more scientific, technical support could lead First test cases will be on view during this world
to a successful initiation of the segment ‘art of conference at the booth with the red flag.
detonation’. Being already trained and allowed to A series of further experiments that allow to
work with it, brings me every time into state of visualise the concept of explosion with energetic
being with servitude and hyper focus. material:

2.2 Laboratory − examine imprint of the shape and impact of

the explosion in different carriers in metal,
I have a mobile atelier to work with non-energetic wood, stone, soil
material where needed and allowed. − make shock wave visible with the smallest
I will further develop this into a ‘high energetic possible amount of energetic material and
unit’ to actually perform these 100 experiments regular medium
on site. − use pressure wave of detonation or

- 310 -
 Figure 4. Ideas for future experiments.

deflagration to test playing wind instruments is the reason why I want to make them visible,
such as flute, trumpet tangible and insightful. They are materials that, as
− initiate detonation and reveal the light, sound it were, contain the life to which I can relate.
and pressure waves with their velocities up to My aim is to develop the pilot project ‘the art
8000 m/second of blasting’ in 2022.
− develop ascent/launch principles starting with Publication and exhibit the experiments; a
spatial figures such as sphere, cone, cylinder, methodology for work with a high energetic
beam, cube medium in the arts.
− deconstructions based on gravity and I hope to meet and will look for people during
everyday discarded objects based on their the conference who are interested in this niche
own construction segment.
− further experiments as - explosion symphony, For more information, questions, support
explosion garden, 1 kg variation, craters, or discussion, please contact
inside a blast info@pietervandenbosch.be or find me at the
− subjects from left to right/top to bottom: booth with the red flag. Have a blast.
breach the door to success, last-ladder, 100
gr, blasting paint, cone.

3 CONCLUSIONS

The explosion itself is the ultimate form of

creation and inspiration for me. I barely
understand it but when I'm working on explosions
it feels logical and makes sense. When an
explosion occurs, time and space come together
for me in a unique energetic way. The concept is
inextricably linked to this medium that literally
explodes into countless particles. The energy
released, the discharge has something
fundamental, a universe of options. The intangible

- 311 -
Figure 5. Pieter Van den Bosch.

- 312 -
- 313 -
 9. Technical development

- 314 -
- 315 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Automatic 3D Fragmentation Analysis from Drone Imagery

A. Gaich, M. Pötsch, & M. Baumgartner

3GSM GmbH, Graz, Austria

ABSTRACT: The paper presents a new method for determining the particle size distribution of muck
piles. It uses drone imagery, automatic 3D model generation routines and a combined 3D - 2D approach
for particle delineation and sizing. The presented method combines shape analysis and image processing
and shows improved robustness against ambiguities related to shadows, rock structure, or different light
conditions. Additionally, it inherently provides real-world scale without ground reference.
Using the presented method entire muck piles are analysed in a single run providing hundreds of
thousands of single particle measurements. Thus, the results are statistically robust against single
mismeasurements. The method has been audited using laboratory scale tests. Current applications
include quarries and mines but also related applications in underground operations and riprap dams.
Similar to other remote sensing approaches this method only analyses the top layer of the muck pile.
However, it automatically generates a result for the entire muck pile automatically without extensive
post processing and it is an economic approach for an objective assessment of the blasting works.

1 INTRODUCTION Existing methods usually apply 2D image

processing algorithms on digital imagery. Due to
The quantification of blasted rock is an objective the nature of the images, there are shadows and
way to rate the quality of the underlying blast. occlusions which make automatic 2D analyses
Such blast may be quantified by several prone to errors when it comes to delineating single
parameters, and by certain volumes (e.g. blasted particles. In doing so, there is a tendency to
volume vs. bank volume) or the resulting combine smaller particles to larger clusters and
fragmentation of the muck pile. The quantitative erroneously divide larger blocks into smaller
assessment of the fragmentation of broken rock at fragments at edges that are due to shadows or rock
larger scale is an elaborate task. For a complete structures.
description of a muck pile, only sieving would Furthermore, existing 2D analysis software
deliver the true fragment size distribution but the usually rely on some user intervention and
required efforts make it impractical for regular therefore apply just on (representative) parts of the
application. muck pile rather than the entire one. What if the
A reasonable compromise between high-effort chosen parts are not representative?
and no-quantification at all is using contact-free In contrast, three-dimensional models describe
measurement techniques. Although these the surface of a muck pile in a comprehensive,
techniques analyse just the visible part of a muck detailed way. This data forms a proper basis for a
pile, they allow doing that at relatively low efforts more complete analysis of the visible fragments of
but deliver objective comparisons between a muck pile. A new algorithm uses such detailed
different blasts. 3D models and applies a combined 3D - 2D

- 316 -
analysis. It detects larger particles based on their needs to be detected (Xmin). The presented
shape and uses image processing for refinement method requires the following relationship for the
and the delineation of smaller particles which in image resolution:
principle already describes the presented method.
The general procedure includes the following GSD = 1/3 * Xmin (1)
steps: (i) (aerial) imaging of the area of interest
(the muck pile), (ii) 3D model generation, and (iii) Where, GSD = ground sample distance (mm/px).
application of the new automatic fragmentation And Xmin = size of the smallest particle to be
analysis algorithm. These steps are outlined in the detected (mm).
following sections. So imaging/flight planning shall be in a way
that the average ground sample distance is three
2 DATA ACQUISITION - DRONE FLIGHT times smaller than the smallest particle to be
detected.
Image data is the basis for all subsequent analyses,
so obviously their quality is of prime importance. 2.2 Drone altitude
This applies both to the image quality itself
(contrast, sharpness) as well as to the geometric In general, 3D model generation shows improved
arrangement of the photos. Some basic rules to performance if GSD is consistent in the photos.
pay attention on the geometric arrangement This means that the camera/drone has the same
include: distance to the object surface. In case of objects
with significant shape variations it is required to
- Provide a definition for the smallest adapt the flight path accordingly (see Figure 1
fragments to be detected as it will mainly left). However, in case of drone flights it might be
drive the flight planning resp. the required sufficient to fly the drone at a single altitude see
ground sample distance (GSD) which then Figure 1 right) and keep the camera looking
links to the flying altitude vertically downwards (aka nadir flight).
- Take images with similar GSDs and view The following equation is valid for nadir
angles flights:
- Ensure sufficient image overlap.
D = f * GSD / (PS * 1000) (2)
2.1 Ground sample distance (GSD)
Where
The ground sample distance is the size of a pixel D = distance to the object’s surface (m)
on the ground, or in other words the resolution of f = focal length (mm)
the images. It is the principal parameter for the GSD..= ground sample distance (mm)
flight planning, i.e. which geometric conditions PS = pixel size (mm)
shall be compiled when flying the drone. As for
fragmentation analysis the ground sample distance 2.3 Image overlap
directly links to the fines threshold, i.e. the size of
particles which can be detected as single distinct The overlap between images provides the required
elements. Thus, it is necessary to define to which redundancy that enables comprehensive 3D model
level of detail the analysis shall be done, i.e. generation. Experience showed that each surface
what is the size of the smallest particle that point shall be visible in at least 5 photographs.

Figure 1. Objects with large variations in depth require adapted flight paths (left); in case of a muck pile
simplified flights at constant altitude may be possible (right); D is the distance between the object’s surface
and the camera; note that D equals the flying altitude only in case of ‘nadir’ flights, i.e. with the camera
looking vertically downwards.

- 317 -
 Figure 2. Suggested distance between subsequent photos: 80% in flight direction, 75%-80% laterally;
Dp = interphoto distance; Dt = intertrack distance.

This means that subsequent photos overlap by at 2.4 Registration

least 80%. In case of muck piles it is often
convenient to fly the area in a grid path (see Figure The images for 3D model generation from drones
2). If the overlap between photos in flight direction need to have additional information for scaling and
(interphoto distance Dp) is 80% of the field of positioning of the result. This information can be
view (image area) then the overlap in lateral either visible one in the images in form of targets
direction (intertrack distance Dt) might be on the ground (ground control points GCP) or by
75%-80%. storing positional data from the drone while taking
In order to estimate the distance between the photo.
subsequent photos the following relationships may Modern drones comprise on-board global
be used: navigation satellite systems (GNSS). Co-ordinates
are determined at the time of photo taking and they
Dp < 0.2 * #pixelsY * GSD (3) are subsequently stored in the image files. Even
though such co-ordinates do not show surveying-
Where
grade accuracy, experience showed that this
Dp = interphoto distance (m)
information is good enough to put 3D models into
#pixelsY = number of pixels vertically (1)
scale. The achievable scale accuracy is about
GSD = ground sample distance (mm)
1% provided that photo taking followed the
Dt < 0.2 * #pixelsX * GSD (4) recommendations above. Please note that this
method is not suggested for any other means of 3D
Dt = intertrack distance (m) model survey where positional accuracy or model
#pixelsX = number of pixels horizontally (1) orientation is critical.
GSD = ground sample distance (mm) Using drone GNSS it is not necessary to
establish a scale reference on the muck pile for
Practical values from a quarry example using a fragmentation analysis. This significantly eases the
drone with 20 MPix and f = 8.8 mm (rounded). application of the method in quarries, since the
mining process is disentangled from the surveying
Xmin = 3 cm smallest particle to be detected process. Data for fragmentation analysis can thus
GSD = 1 cm ground sample distance be quickly and reproducibly gathered in an active
D = 40 m flying distance to the muck pile mining area.
(altitude)
Dp = 6 minterphoto distance 3 3D MODEL GENERATION
Dt = 14 m intertrack distance
Photogrammetry enables the generation of
On many drones, the time between subsequent three-dimensional models from a series of
photos can be set to an automatic interval. overlapping photographs. The introduction of the
Complementary to the example: if the photo- Structure from Motion concept (Szeliski 2011) as
interval is set to 3 s, then the drone needs a travel well as the broad availability of drones brought
speed of 2 m/s to get a photo every 6 m. For a intense application of this technology in the
higher level of automatic drone flights these mining sector.
parameters can be used to set-up a flight plan Structure from Motion includes a series of
using dedicated flight planning tools. processing steps that allows computing a

- 318 -
 Figure 3a. Flight pattern for an interactive flight of a muck pile.

comprehensive set of 3D surface points that used to generate 3D models at sufficient high
combine to a surface description in a accuracy.
photo-realistic style. Due to highly redundant
information, geometric deviations of the used 4 FRAGMENTATION ANALYSIS
camera (lens distortions) are determined and The new algorithm allows for minimised user
compensated while generating the 3D model. This intervention. It uses a procedure that combines 3D
auto-calibration ability makes modern shape analysis with 2D image processing and
photogrammetry algorithms capable to produce includes the following main steps:
accurate 3D models even from low-grade cameras,
so even low priced off-the-shelf drones can be - Detection of particles through a joint

Figure 3b. 129 images taken in four paths, resulting in the above 3D model, it shows an average GSD of 0.7 cm.

- 319 -
 Figure 4. Left: 3D model with delineated particles ‘coloured’ according to size; right: detailed view.

depth-map-texture analysis (3D) If size is determined by sieving then the

- Application of 2D detection in remaining smallest cross section of the fragment may be
regions dominant which underestimates elongated
- Synthesis of step 1 and 2 particles. However, for larger fragments up to
- Particle sizing (3D and 2D) oversizes only the largest diameter of an object
- Data collection. may be the determining value in order to decide if
loadability/crushability is compromised and
4.1 3D detection secondary breakage is required.
Concluding, size determination is not an
An iterative coarse-to-fine strategy applies that unambiguous procedure.
uses the actual 3D shape of the surface. For this The suggested algorithm implements the
part of the analysis not only the ground sample intermediate axis of an ellipsoid that is fitted to a
distance is essential but also the spacing between detected fragment thus follows mainly the rule that
the points that describe the surface of the muck applies for sieving.
pile. Another decision that can be made at this
stage is on the likeliness to be a single particle or a 4.4 Data collection
composition.
Once sizes are determined, it is a usual way to
4.2 2D detection collect all values and describe fragmentation in the
form of a cumulative particle size distribution
Those areas that do not clearly reveal its curve. It describes the amount of elements in
composition from shape analysis are treated with percent that passes a certain size of a mesh.
common image processing algorithms. Several discrete values are taken from the particle
size distribution curve in order to allow a straight-
4.3 Sizing forward quantitative comparison of different
muck piles.
Obviously, the assignment of physical size to
particles is crucial. An important role plays the 5 APPLICATION EXAMPLES
treatment of the shape of the fragments. Very often
ratios between largest and smallest dimension 5.1 Lab scale data
(length vs width) or a ratio between volume and
surface of the object (or in 2D: circumference vs. A laboratory scale model has been established by
projected area) is taken (Schubert 1988). Montanuniversität Leoben within the scope of the
The shape of fragments may be determined by SLIM research project (sustainable low impact
its extensions in 3 dimensions. Spherical or mining, Sanchidrián 2018), an EU funded research
cubical fragments will have similar values for all initiative to improve the economic and ecological
dimensions whereas laminated/flatty particles impacts of surface blasting. The model provides a
(2 large, 1 small dimension) or elongated particles muck pile scaled by 1:25 following ideas from
(1 large, 2 small dimensions) differ from that. So, Ouchterlony (Ouchterlony et al., 2007). Screened
depending on the definition of sizing different rock was configured in different mixtures as a
results may be obtained for non-cubic, synthetic muck pile serving as ground truth. Also
non-spherical fragments. image taking was scaled down by 25 in order to

- 320 -
 Figure 5. A synthetic muck pile composed of different size mixtures served as ground truth (left) –
photo: courtesy of MUL; 3D model and detected particles, ‘colourization’ according to size (right).

emulate real conditions (Tscharf et al. 2018). the visible surface. The analysis of the results
Figure 5 shows a picture of the installation. included the determination of the precision of
The following analyses were performed with the method (its repeatability) and the accuracy
the mixture ‘uniform’ (see Figure 6 left) in order (deviation from the ground truth). Figure 6 shows
to have a comparison with common blast results in some resulting particle size distributions of
quarries. different synthetic muck pile installations and the
The synthetic muck pile was reconfigured ground truth from sieving.
several times by tilting the construction back and The following tables shows some results from
forth. This way the same rocks are arranged the data analysis. Three different installations of
differently including that different particles form the same muck pile mixture (“uniform”) are

Figure 6. Particle size distribution for lab scale tests; left: ground truth for four different muck pile mixtures,
bold line is ‘Uniform’ mixture; right: comparison of ground truth with differently arranged muck piles of the
same mixture M1: ‘Uniform’.

- 321 -
 Table 1. Comparison of ground truth data with three different arrangements of the synthetic muck pile for
estimating accuracy.

Accuracy
mm M1-W1-H10 M1-W2-H10 M1-W3H10 Average GroundTruth Delta
x20 9 10 9 9,4 9 -0,4
x50 18 19 19 18,5 20 1,5
x80 31 34 35 33,1 40 6,9
# Fragments 11045 10588 12012
Mean Absolute Error 2,9

Table 2. Variation of the measurements for determining precision; MAE stands for Mean Absolute Error.

Precision
mm Average M1-W1-H10 M1-W2-H10 M1-W3H10
x20 9,4 -0,4 0,4 0,1
x50 18,5 -0,8 0,3 0,5
x80 33,1 -2,0 0,5 1,5
MAE 1,1 0,4 0,7
Mean 0,7

independently analyzed and compared with the regardless whether it is own or subcontracted
ground truth. Table 1 provides an average of the work.
three installations for x20, x50, and x80 and shows The regular analysis of blasts inside an
the mean absolute error to be 2.9 mm. operation increases the understanding of the link
The tables also provide the mean absolute error between blast design and blast result. It reveals
(MAE) as this value correctly treats all sample quantitatively if blasts deviate from the target and
measurements with equal weight whereas the Root allows to adapt for the next blast in case of the
Mean Square (RMS) does not as it gives more latter.
weight to larger deviations. The definition of the smallest particles to be
Table 2 compares the average with the single detected and the used camera/lens determine the
analysis and gives an indication of repeatability flight altitude. Typical values for quarry blasts lie
resp. precision. The result showed to be 0.7 mm. within 25 and 40 m above ground. Depending on
Comments - even if 2.9 mm is a good value for the extent of the muck pile a typical data set
accuracy, one needs to keep in mind that the comprises 150 – 250 photographs. The
visible part of the synthetic muck pile differs for computation of a 3D model therefore takes about
every pass, which means there is always a kind of 40 – 60 minutes on a laptop computer and
variation to be expected. fragmentation analysis might take another 10
The small value for repeated measurements minutes. However, the steps perform fully
(precision) shows that the method is inherently automatic.
robust. It indicates also that the method sensitive Figure 7 shows a snapshot of an automatic
in detecting general changes in the particle size analysis from a quarry blast. Particles are
distribution, which helps for visualizing trends in ‘coloured’ according to size which instantly
the size distribution. provides a perception of the size distribution and
highlights oversizes.
5.2 Quarry blasts
5.3 Other
The determination of the fragmentation
distribution is of vital interest in quarry blasts. The The algorithm has been applied to data from
information gained is useful for quality control underground mining as well as mapping the

- 322 -
 Figure 7. Left: 3D model of a surface blast with automatically determined fragments ‘coloured’ according to
size, red blocks are larger than 1.2 m, yellow range from 0.4 to 0.8 m. Right: particle distribution curve built
from over 300,000 measurements; the curve comes completely from measurements, not from curve fitting.

Figure 8. Automatic fragmentation analysis from passing LHDs.

Figure 9. Automatic size determination for determining the condition of dam slopes.

condition of dam slopes (see Figure 8 and 3D-2D approach. The spatial information eases
Figure 9). correct separation/delineation of single fragments
from image data and supports the process of
6 CONCLUSIONS assigning physical size to the detected particles.
Particle size distribution is determined from The method has been applied on lab scale with
photogrammetric 3D models utilising a combined ground truth data from screening. The high

- 323 -
repeatability (precision) showed that the method is
very sensitive to detect changes in the fragment
size distribution and also provides reasonably
accurate estimates of absolute particles sizes.
The new approach has also been used on quarry
data. It automatically processes the entire surface
of a muck pile in a single run. The resulting
particle size distribution curves contain several
hundreds of thousands of individual
measurements, which means that the resultant
passing values (x20, x50, x80) are rather robust if
single mismeasurements should occur.
The approach has been in use on a large
number of quarries in the meantime. Besides
quality control it has also been applied to support
the calibration of fragmentation prediction
algorithms.
Other applications of the methodology include
underground mining and dam inspection.

REFERENCES

Ouchterlony, F., Nyberg, U., Bergman, P. & Esen, S.

2007. Monitoring the blast fragmentation at
Boliden Mineral’s Aitik copper mine, In:
Proceedings 4th EFEE World Conference on
Explosives & Blasting, pp 47-62 (European
Federation of Explosives Engineers: Vienna).

Sanchidrián J.A. 2018. SLIM. Technology for

blasting to improve mining, In: Rock
Fragmentation by Blasting: Proceedings of the
12th Int. Symp. on Rock Fragmentation by
Blasting - Fragblast 12, June 2018, pp. 783-793.

Schubert, H. 1988. Aufbereitung fester mineralischer

Rohstoffe, Band 1. 4. stark überarbeitete Auflage,
Leipzig: VEB Deutscher Verlag für
Grundstoffindustrie.

Szeliski, R. 2011. Computer Vision: algorithms and

applications. Springer, London, New York.

Tscharf, A., Schenk, F., Mayer, G., Fraundorfer, F. &

Bischof, H. 2018. UAV-Anwendungen im
Bergbau – Modellversuche als Beitrag zur
photogrammetrischen Forschung und
Entwicklung. In: J. Benndorf (Ed.), 19.
Geokinematischer Tag (1 ed., Vol. 2018, pp.
30-43).

- 324 -
- 325 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Automatic blast design based on an autonomous drone

T. Bernard
TBT/DNA-Blast Software

ABSTRACT: The use of drone technology to optimize blasting has become very common over the last 3
years especially in quarry operations where controlling the face profile is key to preventing fly rock and
providing good fragmentation. Nevertheless, this process requires an operator on site to fly the drone
when the bench is ready (face cleaned and weather ok). It then takes some time to download and process
the data. The technology described in this paper shows how all the processes can be automated from
drone flight to blast design, including the calculation of the 3D cloud of points, generating the critical
face profiles, and positioning holes automatically according to burden specification. The blast engineer
will only have to review the generated blast design based on the given parameters and adjust it if
necessary. It will also allow multiple sites to be managed remotely, saving travelling time and money.

1 INTRODUCTION taking into account for example the minimum

width between the latter and the free face.
Digitalisation has invaded mining over the last few With the evolution of technology, the 2D
years. It all began, as in many other sectors, with profiler became 3D enabling us to obtain a cloud
the use of drones to democratise topography. The of points on the free face. Consequently, this
so-called 3D setting up of a blast has almost enabled us to produce a so-called critical profile,
become commonplace in quarries. i.e. a profile at the hole, not in a plane, but
Let's step back in time a little to understand the zigzagging on the face capturing the shortest
challenge that technology provides. To obtain a distance at each elevation. This digital processing
good blast result (good fragmentation, with little enables us to visualise the smallest and largest
fly rock) we need to efficiently control the burden burdens on the height of the bench which allows
of the first row (width between the hole and the us to adapt the loading to guarantee the safety with
free face). Indeed, for the following rows (behind) regard to fly rock and good fragmentation.
the distance compared with the row in front is The increasing use of drones for less than five
obtained by positioning the holes when drilling. years has enabled the safety of this process to be
Historically, the width of the burden was firstly simplified and improved by providing a set of
measured via the so-called ‘fishing rod’ technique, images when flying over the area and applying a
i.e. by using a rod with a rope attached to one end photogrammetry algorithm.
and a weight that was positioned at the foot of the Although technology has improved
bench, which provided the best position for the considerably since the use of the ‘fishing rod’, all
hole with regard to the foot of the bench. these technologies have a point in common, man
Then the 2D laser profiler came along which remains at the centre of the process, handling the
enabled us to obtain a bench profile in a vertical various tools on site. There is nothing wrong with
plane, thus allowing the positioning of the hole, that apart from the fact that you need a specialised

- 326 -
- 327 -
 Figure 1. A day in tomorrow’s blast engineer’s life.

operator (a measurement technician or a remote technologies available at the moment to optimise

pilot for the drone) available at the right moment, blasting in a quarry or mine. This optimisation
i.e. when the free face is clean. Now on an consists of designing the blast with the best
operational site (mine or quarry) that moment is parameters in order to guarantee the results within
not easy to anticipate and depends on numerous the best time frame. The process is supported by
operating parameters, possible equipment failures, technology that automates the most fastidious
and the weather. tasks as much as possible but which leaves man at
The aim of the technology presented below is the centre of the decisions and checks. An overall
to enable the free face logging to be virtually description is provided with Figure 1.
independent of the operator, his availability and
the conditions on site. 3 THE FUTURE OR REALITY?

2 A DAY IN TOMORROW’S BLAST Some of you could think that this short story is just
ENGINEER’S LIFE science-fiction and that the operations described
are far from the daily reality of blasters on site. Of
Let's look at the main stages of a blast engineer's course, it is true, as presented, that this process is
day in the near future. The latter uses all the only used on a few operations in the world and

- 328 -
 Figure 2. Automatic drone station.

often partially like electric cars 10 years ago or way and a lot less than 10 years will be necessary
mobile telephones 20 years ago. Evolution is on its for it to become the standard in our industry, being

Figure 3. Workflow.

- 329 -
Table 1. Stages of the automatic location of a blast.

Stage 1.

Definition of the flight area by the

user with the help of a polygon on
the present background.

Stage 2.

The server automatically calculates

a flight plan, taking into account
the real topography of the terrain.

The 3D view of the present site’s

cloud illustrates the photography
path for the photogrammetry.

Stage 3.

Automatic take-off of the drone

for its task from a fixed station
enabling it to recharge its
batteries and unload the data to a
server for future processing.

The operation is fully automated

and requires no remote pilot.

Stage 4.

Photogrammetry flight, flying

over the Ground Control Points.

Here, we can see a target with a

specific pattern enabling it to be
automatically identified during
future processing.

- 330 -
 Table 1. Stages of the automatic location of a blast. Continued.

Stage 5.

Automatic transfer of photos to the

server and automatic calculation of
the cloud of points with the help of
RTK georeferenced photos or
ground control points marked by
targets which are automatically
detected by photogrammetry
processing.

Stage 6.

Automatic update of the overall

site cloud keeping the most recent
points compared with the initial
cloud.

Stage 7.

The automatic location of the holes is

carried out based on the critical
profiles of the bench calculated with
the cloud of points, then the min. and
max. burden rules, with regard to an
average profile of the front at each
hole. The same principle is applied
for the following rows, using the
holes of the first row as a reference.

Stage 8.

Transmission of the blast design to a

smart drill connected with the help of
the IREDES protocol.

given that the development speed of technology is from desires to reality. The photo in Figure 2
exponential. illustrates the type of drone which, installed in a
While you read this publication, the quarry in France, enables all the tasks mentioned
technological and regulatory evolution has already above to be carried out automatically
enabled us to transform automatic drone tasks (photogrammetry, filming the blast, site

- 331 -
Table 2. Stages of the automatic measurement after a blast.

Stage 9.

Upon the order issued by the person

in charge of the operation (via mobile
telephone) automatic take-off of the
drone from its fixed station for the
blast quality control task.

The operation is fully automated and

requires no remote pilot.

Stage 10.

The drone automatically films the

blast from its predefined position.

Stage 11.

Totally autonomous, photogrammetry

flight over the muck pile, and flight
over the Ground Control Points if no
RTK is available in order to create a
3D cloud of points of the muck pile
and an orthophoto.

Stage 12.

Automatic analysis of the muck pile

fragmentation based on the
orthophoto.

Here you can see a display of the % of

fragments passing a 500 mm screen
with a colour code on a grid
superimposed on the muck pile.

- 332 -
inspection). (Warning: This technology requires and is used throughout the whole world without
specific authorisations which vary from one restraint.
country to another). Therefore, history has taught us that the
Regarding the other technology for the benefits of technology end up by triumphing
automatic design of blasts based on clouds of over the hesitancy of the first users who were
points obtained by photogrammetry, it is available confronted with it and who often display resistance
via specialised software which controls the to change which is one of man’s characteristics.
automation of the process. Regarding the loss of competency and
know-how that could result from the
3.1 So, how does it actually work? implementation of such technology, this is also a
non-issue for three reasons:
The principle is based on a cloud of points which
describes the topography of the site in question - Firstly, the development of algorithms and
and its automatic update (Figure 3), in order to be their maintenance require excellent
able to have updated data available to design the competency for them to be as relevant as
blast. possible
The overall automation procedure is described - Secondly, although the process is automated,
in Table 1 for the stages (1 - 8) before the the verification of the coherence of the results
blast, and Table 2 for the stages (9 – 12) after the obtained remains under the supervision of
blast. competent staff; automation only applies to
fastidious and repetitive operations that might
4 BUT WHO WOULD DARE MANAGE THE be carried out in a less than perfect manner by
DESIGN OF A BLAST BY A MACHINE? humans
- The last reason is connected with the
I have heard quite a few remarks on the fact that implementation of an industrial process,
such a tricky and risky task as blasting could not based on equipment and gear inescapably
be entrusted to a machine or an algorithm, and that subject to failures or breakdowns that require
with time, the expertise and job of blasters would us to have a downgraded operating mode
disappear, etc. solution, hence inevitably replaced by
These are the traditional comments about any humans.
technological evolution. There are numerous
similar examples everywhere. Let's quote the ABS Competency will not diminish but will partly
(System preventing the wheels of a car becoming be transferred to the staff designing the systems,
locked when braking) example which, when it maintenance staff and process supervisors.
appeared, provoked lively reactions from some
drivers who thought that this system would reduce 5 CONCLUSION AND FUTURE
the competency of drivers without offsetting the DEVELOPMENTS
enormous safety advantage, enabling cars to brake
in a straight line, without skidding and over a Today, the automatic design of the drilling plan
shorter distance. Who would want a car without enables us to optimise the results of a blast
ABS today? It should be remembered that this (fragmentation, casting) by fine-tuning the latter
system was installed on aircraft as early as 1930 according to the geometry of the free face almost
before being used for cars! automatically. This is under the supervision of a
Another example, which is closer to our job as blast engineer who validates the stages of the
blasters, is the introduction of the initiation of process and is still the guarantor.
blasts with a wireless system, chiefly using radio The process can be launched remotely when the
waves. It was introduced by Delta Caps configuration of the operation and the weather are
International in 2000. The technology was firstly suitable (mucked and cleaned bench, moderate
rejected by almost everyone on the pretext that it rain and wind). It also enables you to supervise
would never be possible to guarantee the safety of several operations simultaneously whilst
a blast with such technology, yet essentially this guaranteeing the best results for the operation with
technology was to enable the master blaster in the help of centralised competency.
charge of the initiation to position himself in a To move further ahead, full automation of the
place that would enable him to guarantee a much blast design is at an experimental stage with the
better visual safety of the blast area. Twenty years automatic calculation of loading and the initiation
later, the technology has become commonplace sequence according to the aims defined.

- 333 -
6 ACKNOWLEDGMENTS

TBT would like to thank VICAT for having

allowed the implementation of the autonomous
drone on its Courbaisse site, without which there
could have been no experiments with this
automation process.

- 334 -
- 335 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Density monitoring of bulk emulsion explosives in

confined boreholes

F.F. Sitonio
Sr. Mining Engineer, Vale

J.C. Koppe
Universidade Federal do Rio Grande do Sul

ABSTRACT: Parameter control of explosives used in blasting operations have become more relevant
over the years, bringing a better understanding of detonation performance. Giving into consideration that
bulk emulsion density’s variation reflects directly on blasting results due to its importance in product
sensitivity and velocity of detonation (VOD), blast designs are conducted assuming that cup density
measurements are reliable and explosives gassing agents will perform as per specifications.
Nevertheless, confinement conditions in boreholes like temperature, presence of water and surrounding
circumstances of rock mass are often neglected on predictive modelling and off-site measurements. This
paper describes the assembly of a device equipped with pressure sensors and its utilisation in confined
boreholes of an open pit mining operation, aiming to obtain in-situ monitoring of density variation in
several explosive columns. After we developed a software to get and process data of a hardware
prototype and the device functionality was validated in water, oil and quarry mining experimental trials,
7 of 291 boreholes in a gold mine operation were chosen to be monitored with two probes each, placed
at the top and at the bottom of the explosive column profile. The results obtained enabled a comparison
between the bulk emulsion density variation when acting in different environments, since we selected
boreholes with diameter of 101.6 mm (4 in) and 139.7 mm (5.5 in) and some with presence of water.
Besides the increase of explosive density with depth related to hydrostatic pressure increase, we found
an overall error of 6% in cup density measurements when compared with sensors placed at the bottom of
boreholes and 2% when compared with sensors placed at the top of explosive column. These findings
provide detailed insights into how bulk emulsion explosives behave in different circumstances including
depth and borehole diameter, giving the opportunity to re-evaluate blast design parameters such as its
geometry, sequence pattern, charging profile and explosive composition.

1 INTRODUCTION incompatible blast designs, especially because

gassing expansion is not well known along
Successful blasting operations with bulk emulsions blastholes columns when they are found in
are led by controlling some factors of influence, confined situation. Besides environmental hazards,
including explosive composition and its initial inadequate blasting results often leads to poor
charging density. Although bulk emulsion efficiency in subsequent mining operations
explosives have high reliability and some such as loading, hauling, crushing and grinding
properties are previously known, there is a lack of cycles, increasing overall cost of mineral
fidelity in cup density measurements, leading to exploitation.

- 336 -
 Egly & Neckar (1964) developed and patented different rates and temperatures. The study
explosive emulsions. With some adding observed a linear increase in explosive’s VOD
modifications to be considered over the past 50 with densities from 0.95 to 1.15 g/cc, with
years, bulk emulsion explosive matrix contains detonation failure above 1.27 g/cc. Nevertheless,
oxidisers dissolved in water as droplets surrounded the study took in consideration density
by an immiscible fuel and is not capable of measurements without confinement, like cup
detonation on its own. To achieve detonating measurements, commonly used in open pit
behaviour, it requires sensitiser in fixed amount. operations. Besides that, hydrostatic pressure
Due to its rheology, emulsion explosive also needs phenomena were neglected, even it has an
a certain reaction time inside the blasthole to important role in real density value. As pointed
expand its volume and obtain ideal density before by Cavanough & Onederra (2011), in-situ
stemmed. Medina (2014) mentions that hydrostatic measurements of VOD, temperature and pressure
pressure affects final explosive product. Thus, considering confinement and rock mass properties
higher densities are found at the bottom of the hole around blastholes can lead to a better
since bubble sizes are lower. understanding of the explosive charge.
Cudzilo et al. (2002) studied five explosive In accordance with pioneer projects developed
performances with glass micro-balloons and in the past two decades as Zeman & Trzciński
ammonium nitrate solution as matrix, mixed with (2002) and Canto (2018), and at the same time
specific nitrates. They observed a reduction in aiming to sponsor an evolution in rock blasting
explosive performance when adding metallic operations control, the present work is based in
nitrite to aqueous solution, besides a great trial tests with an electronic device built to
influence in emulsion sensitivity with borehole monitoring the density of commercial explosives
diameter variation. In contrast, Cheng et al. (2017) in open pit mining operations. Since reducing
observed a higher detonating power with VOD is intrinsically bonded to bulk emulsion
increasing titanium hydride rate as sensitising. explosive’s final density, and results in lower
Mishra et al. (2017) related VOD with gassing detonating pressures through unavailability of
agent’s variation and bulk emulsion densities. shock-wave energy (Yunoshev et al. 2012),
Sodium nitrite were used as gassing agent with experiments highlighting its real value are

Figure 1. Output vs. pressure differential of the MPX5700AP sensor.

- 337 -
 Figure 2. Hardware plate, equipped with two MPX5700AP sensors and Arduino Nano V3.0 board.

necessary to evaluate blasting quality. continuous formation when sensitised and then
charged in boreholes. In consequence, the density
2 SENSOR ASSEMBLY of product decreases as gassing reaction proceeds.
At the same time, reaction time is controlled
To monitor real-time pressure variation in by gasifier proportion, emulsion formula and
confined boreholes and relating to the density of temperature.
bulk emulsion explosives, we assembled a device
equipped with two MPX5700 pressure sensors,
following its recommended power supply
decoupling and output filtering schema. Ideal for
microcontroller-based systems, MPX5700AP
sensor series delivers a 2.5% maximum error
result of absolute pressure measurements, and it is
capable of reading proper values in a 0 °C to 85 °C
(32 °F to 185 °F) environment situation. Figure 1
shows the sensor output signal relative to pressure
input.
Responsible for interfacing hardware and
software in this project, the Arduino Nano V3.0 is
shown in Figure 2, attached to the soldered board
with electronic components such as condensers,
necessary for MPX5700AP functionality.
In addition to assembling all electronic circuit,
there was a necessity of creating test probes with
capacity to hold high levels of air pressure, as
can be seen in Figure 3. Bicycle tube, clamps,
hydraulic hose and several connection types were
some of the elements used to build them. The
probes were capable to hold internal pressures
above 100 kPa (14.5 psi), as observed in field
tests. Figure 3. Inflated test probe, developed to work
in confined situation at high pressure levels.
3 THEORETICAL FOUNDATION
The weight that supports each point across the
A given amount of explosive emulsion tends to width of explosive column maintains a distinct
increase its initial volume because of bubble’s pressure gradient up to borehole end. In that case,

- 338 -
the amount of mass at a given point varies prototype was then tested in a laboratory
depending on the depth, because density also environment, using a polyvinyl chloride tube
varies in direct relation to the compression representing a borehole, filled with water and
gradient. Nitrogen bursts, submitted to different kitchen oil to check its accuracy in various
degrees of pressure thus presenting different density measurements. An air pump equipment
volumes, and these spaces left by the pressed and hose connections were also used to
bubble are occupied by emulsion, increasing mass, control pressure flow in the probe, according to
so density increase as well. Figure 4.
Equation 1 shows the basis of density calculus The column weight presses the probe when
through absolute pressure measurements, in the inflated. Immediately, the signal is transmitted by
software developed to receive and process the hydraulic hose to the MPX5700AP sensor, that
MPX5700AP output data. It is important to converts the signal into electric potential
mention that, to evaluate the density at a given difference. The voltage is transformed in pressure
depth in this project, cup density measurement is units. This pressure level result is sent to USB port
also needed. where the Arduino board is connected.
When the inflated probe is slowly deflated,
𝜌𝜌(𝑖𝑖) = 1 / (1/𝐷𝐷b𝑔𝑔 – 1/𝜌𝜌(𝑖𝑖 − 1)) ∗ (𝑃𝑃𝑃𝑃(𝑖𝑖 – 1)/ surrounded in a liquid or emulsion, it will be
𝑃𝑃𝑃𝑃 (𝑖𝑖) + 1/𝐷𝐷bg)) Equation (1) observed that the air flow that leaves the probe
detains momentarily when the pressure inside of it
Where, is equal to the surrounding pressure. This happens
𝜌𝜌(i) = density at a given depth, taking atmosphere because the flow rate is proportional to the
as reference, i.e., i=0 (kg/m³) (lb/yd³) difference of pressure between internal and
𝜌𝜌(i-1) = density at past distance reference (kg/m³) external environment, so the force unrolled by the
(lb/yd³) elastic walls of the probe are balanced. Figure 5
Dbg = bulk emulsion density before gassing, shows pressure measurements made in field tests
measured by cup test (kg/m³) (lb/yd³) captured in software, highlighting the moment
Pt(i) = hydrostatic pressure at a given depth plus where it holds for a higher time, indicating that
atmospheric pressure (kPa) (psi) internal pressure is equal to emulsion explosive
Pt(i-1) = total pressure at past depth reference. environment.
(kPa) (psi) The probe was always first inflated at higher
pressure rates than external, and slowly deflated to
4 DENSITY TEST PROCEDURE get the exact instant where external pressure is
equal to internal. The process should be repeated
In order to evaluate the testing procedure and the three to five times to confirm previous sampling
equipment reaction to the process, the hardware data.

Figure 4. Laboratory test methodology.

- 339 -
 Figure 5. Data acquisition in the software developed for pressure monitoring. The dashed line highlights the
instant where pressure holds in a longer interval when probe is deflated, representing the emulsion pressure in
probe walls.

5 FIELD TESTS mining process was not negatively impacted, and

the data acquisition could be realised with
With the procedure successfully evaluated in precision.
off-site situation, small tests in a quarry mining In October 2020, we conducted a major test in a
blast operation located in Northeast Brazil gold ore mining located in North Brazil (Figure 6).
confirmed its effectiveness in full-scale operations. 7 of 291 boreholes were then monitored with both
The methodology was conducted taking in bottom and top probes placed at the column.
consideration the operational flow time, so the Aiming to get various data results, we selected

Figure 6. Blasting operation of the gold ore mining at North Brazil.

- 340 -
 Figure 7. Data acquisition in a gold ore mining, located in North Brazil.

holes with diameters of 101.6 mm (4 in) and 139.7 depth) were then observed, once probes located at
mm (5.5 in), since ore and waste had different the bottom of the blasted holes had higher density
diameter specifications. Boreholes with water values if compared with the ones located at the top
were also chosen. Figure 7 shows the data of the column at the same hole. Additionally, it is
acquiring procedure. important to mention that the hole with 249 ID for
example, which had a higher amount of water in
6 DATA PROCESSING emulsion mixture due to hose blown while
charging (in order to unblock its flow), presented
Once all data were collected, they were processed minimum density results. This inadequate
and then interpreted, generating an analysis of operation procedure is often made and should
explosive charging process in the gold ore mining always be avoided, as test results shows its
operation. For each borehole and its explosive influence in final density, and probably in the
final density along the detonation column, one explosive’s VOD.
monitoring report is produced as shown in
Figure 8. 8 CONCLUSION AND FUTURE WORK

7 ANALYSIS AND RESULTS DISCUSSION Explosive manufacturers use models to predict

change. Nevertheless, results prove the necessity
The following results shown in Table 1 had to make in-situ measurements to have better
conformity with theorical basis. As quoted above, accuracy in charge factor and emulsion
hydrostatic pressure effects (that increases with performance. Density variation is not uniform

- 341 -
 Figure 8. Borehole density monitoring report.

Table 1. Bulk emulsion monitoring results.

A0 A2 h Ø T Dbg Dag 𝜌𝜌A0 𝜌𝜌A2
Hole ID W
(m) (yd) (m) (yd) (m) (yd) (mm) (in) (m) (yd) (g/cc) (g/cc) (g/cc) (g/cc)
54 1.7 1.86 6.2 6.78 8.4 9.19 101.6 4.0 1,6 1.75 Y 1.207 1.154
125 1.8 1.97 5.8 6.34 7.8 8.53 139.7 5.5 1.6 1.75 Y 1.219 1.154
1.32 1.13
176 1.8 1.97 6.5 7.11 8.5 9.30 101.6 4.0 0.9 0.98 N 1.219 1.165
249 2.3 2.51 6.0 6.56 8.2 8.97 101.6 4.0 1.7 1.86 N* 1.189 1.141
82 1.7 1.86 6.6 7.22 8.6 9.41 139.7 5.5 1.8 1.97 Y 1.213 1.167
117 1.1 1.20 5.4 5.91 7.3 7.98 139.7 5.5 1.4 1.53 Y 1.31 1.15 1.201 1.155
194 2.0 2.19 6.5 7.11 7.9 8.64 101.6 4.0 1.0 1.09 N 1.212 1.157

Where:
A0 = probe height, placed at the bottom of the hole (m) (yd) Dag = final cup density measure, 30 minutes
A2 = probe height, placed at the top of the hole (m) (yd) after gassing expansion (g/cc)
h = hole length (m) (yd) 𝜌𝜌A0 = explosive emulsion density, gauged by
Ø = hole diameter (mm) (in) probe A0 (g/cc)
T = stemming height (m) (yd) 𝜌𝜌A2 = explosive emulsion density, gauged by
W = presence of water probe A2 (g/cc).
Dbg = initial cup density measure, before gassing expansion (g/cc)

- 342 -
along the borehole column, once real diameter Propellants Explosives Pyrotechnics, Anhui,
may change, and geological surrounding can affect China, 42(6).
it as well. Thus, several conditions and ideas can Cudzilo, S., Kohlicek, P., Trzcinski, V.A. & Zeman,
be used as future work. The next steps using the S. 2002. Performance of emulsion explosives.
sensor built for this project will evaluate the ideal Combustion, Explosion and Shock Waves, Vol.
stemming time, measuring several variations from 38, No. 4, pp. 463-469.
0 to 1 hour of gassing reaction before borehole is
Egly, R.S. & Neckar, A.E. 1964. Water-resistant
confined, to see and how stemming really affects
sensitizers for blasting agents. US Patent,
final density of explosive emulsion. Additionally,
(31)615, New York, USA.
a temperature sensor will be added to the device,
aiming to control one more changing parameter Hansson, H. 2009. Determination of properties for
along the borehole column, giving more precise emulsion explosives using cylinder expansion
measurements. tests and FEM simulation. Thesis – Swedish
Blasting Research Centre (SWEBREC), Luleå,
9 ACKNOWLEDGEMENTS Sweden.
We wish to thank Delion Science and Engineering Medina, R. 2014. Evaluación técnico-econonómica-
(represented by Victor Hugo Paredes Canto) for its ecológica de los resultados de las pruebas
consulting in software development, Great Panther realizadas usando emulsiones gasificadas em
Mining (Tucano Gold Mine) for providing the best Cuajone – Southern Perú", Thesis – Universidad
environment to run the project. Our special thanks Nacional de Ingienária, Lima, Peru.
go to Fundação Luiz Englert (FLE) for all
financial and academic support. Miranda, V., Leite, F.S., Brito, P., Frank, G. &
Soeiro, J. 2019. Borehole deviation control using
REFERENCES electronics: An Euler’s approach. International
Society of Explosives Engineers – 45th Annual
Agrawal, H. & Mishra, A.K. 2018. A study on Conference on Explosives & Blasting Technique
influence of density and viscosity of emulsion – Nashville, Tennessee, USA.
explosive on its detonation velocity. Model Meas
Control, C78(3), pp. 316-336. Mishra, A.V., Rout, M., Singh, D.R. & Jana, S.P.
2017. Influence of gassing agent and density on
Bhandari, S. 1997. Engineering rock blasting detonation velocity of bulk emulsion explosive.
operations. Rotterdam Brookfield: A.A.Balkema. Geotechnical and Geological Engineering,
36(4-5).
Campbell, A. 2018. Effect of explosive density and
ring burden of recovery on fragmentation and ore Wilches, W. & Scovira, D. S. 2014. Implementation
recovery in sublevel cave mines. Caving of high density ammonium nitrate prill in
Conference, Queensland. emulsion blends in an open pit coal mining
application. International Society of Explosives
Canto, V.H. 2018. Medicíon de la densidad de Engineers – 40th Annual Conference on
explosivos gasificables en el fondo de taladro a Explosives & Blasting Technique – Denver,
través de un dispositivo de monitoreo de presíon Colorado, USA.
hidrostática em minería superficial, Universidad
Nacional de Trujillo, Peru. Yunoshev, A.S., Plastinin, A.V. & Silvestrov,
V.V. 2012. Effect of the density of an emulsion
Cavanaugh, G. & Onederra, I. 2011. Development of explosive on the reaction zone width.
pressure and temperature gauges to monitor in Combustion, Explosion and Shock Waves,
situ performance of commercial explosives. Vol. 48, No. 3, pp. 319-327.
Transactions of the Institution of Mining and
Metallurgy, Section A: Mining Technology
120(2), Brisbane, Australia, pp. 74-79.

Cheng, Y., Meng, X., Feng, C., Wang, Q., Wu, S.,
Ma, H. & Shen, Z. 2017. The effect of hydrogen
containing material TiH2 on the detonation
characteristics of emulsion explosives.

- 343 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Fundamental study of reactive ground chemistry

G. Conroy & K. Robinson

Orica Ltd, R&I Group, Kurri Kurri, Australia

ABSTRACT: Reactive ground is the spontaneous reaction between ammonium nitrate and metal
sulphides, particularly pyrite, FeS2. It is a significant safety risk in mines where sulphides are present.
Past identification of the reaction mechanism has been made difficult due to the naturally occurring
mineral’s complex chemical matrix. By using chemical grade FeS2, a fundamental study of the reaction
was conducted using electrochemistry, SEM, XRD, XPS and gas analysis as part of a PhD conducted at
the University of Newcastle, sponsored by Orica Ltd, a leading explosives manufacturer. The study
showed that electrochemistry could be used to identify reaction peaks associated with the reactive
ground mechanism, and that changes in electric potential could also be used to identify the onset of
reaction in a small reactor vessel. The study showed that the reactive ground mechanism involved a
series of reactions, with a slow reaction generating oxidation products such as Fe(II) and Fe(III) sulphate
on the FeS2 surface. These iron oxide products react exothermically with nitrates and could occur on
samples oxidised in moist air, and therefore different levels of oxidation could explain differences in
reactivity between FeS2 samples. Reaction rates were found to be highly temperature dependent, and
thus the transition from a slow reacting system to thermal runaway was governed by heat and mass
transfer to and from the reaction zone. A reaction mechanism for this process has been proposed that
differs from that put forward by other authors as it does not rely upon a build-up of reactants or
intermediate species to transition from a slow reacting, inert system to thermal runaway.

1 BACKGROUND TO REACTIVE GROUND violent that it results in premature or unplanned

AND THE CURRENTLY ACCEPTED initiation of explosive (AEISG 2017)’.
MECHANISM It is known that one of the key minerals present
in reactive ground is pyrite, FeS2, which can
Ammonium nitrate (NH4NO3) is the most undergo a complicated decomposition reaction
commonly employed bulk explosive in the mining mechanism in the presence of water and oxygen,
industry. It may be used either as a blend of solid with both the Fe and S species undergoing
ammonium nitrate prill and a hydrocarbon such oxidation and reduction reactions (Dos Santos et
as a fuel oil (ANFO) or as an emulsion of al. 2016, Garrels et al. 1960).
supersaturated ammonium nitrate in fuel oil. The reaction mechanism for reactive ground is
Reactive ground is defined as ‘ground that characterised by a long induction period, where no
undergoes a spontaneous exothermic reaction after reaction can be physically observed, followed by
it comes into contact with nitrates. The reaction of an initiation phase, where the temperature of the
concern involves the chemical oxidation of system increases rapidly. During this phase, NOx
sulphides (usually of iron or copper) by nitrates, gases are commonly evolved. The rapid rise in
and the liberation of potentially large amounts of temperature can then lead to the decomposition of
heat. The process is unpredictable and can be so the NH4NO3.

- 344 -
 The current reaction mechanism is based upon 2 ELECTROCHEMICAL ANALYSIS
the Rumball Mechanism, which was proposed in
1991 (Rumball). This mechanism states that 2.1 Experimental setup
during the initiation stage, ferrous (Fe(II)) and
ferric (Fe(III)) iron will react with nitrate ions to Technical grade FeS2 was obtained from Fischer
produce NO gas and nitrous acid, and that the Scientific, with a purity of >95%. Surface oxides
transition from the induction to the initiation were removed by washing with hot, concentrated
phases is triggered by the build-up of these hydrochloric acid (HCl 37 wt%) until the wash
intermediates. liquid ran clear. The washed and unwashed
Many studies on reactive ground have focussed samples were examined using scanning electron
on characterisation of reactive ground samples, microscopy (SEM) with backscatter.
with subsequent variations in samples affecting Electrochemical cells were prepared by taking
results. This study has instead used technical grade 0.100 g of FeS2 and mixing with 0.12 g of KS6-L
FeS2 as a laboratory substitute for reactive ground, graphite (Timcal) and 0.05 g of electrolyte. A filter
thereby reducing sample variation and allowing a paper spacer was inserted and the mixture was
more fundamental study to be performed. pressed into the base of the cell at 200 psi (1420

Figure 1. Diagram of cyclic voltammetry cell as used in this study showing (1) SCE reference electrode;
(2) Counter electrode; (3) Teflon spacer; (4) Electrolyte; (5) Pressed sample pellet; (6) Perforated spacer
disc; (7) Working electrode; (8) Holding bolts; (9) Perspex holder; (10) Working electrode contact.

- 345 -
 Figure 2. Voltammogram of FeS2 in 1M NH4NO3, showing (a) voltammogram current, (b) current for
platinum wire at 0 V in electrolyte and (c) electrolyte potential.

kPa) for 30 s. A 3 mm Perspex spacer, drilled with FeS2 surface being reduced on the sensing
50 1 mm holes was placed above the pressed disc platinum wire set at 0 V. As the cycle is reversed,
and 10 mL electrolyte added to complete the cell the potential decreases at potentials of 0.4, -0.2
as shown in Figure 1. and -0.6 V. This indicates that reduction reactions
Voltammograms were obtained by measuring are occurring at these potentials that reduce the
cell current as the potential was varied. Unless concentration of ionic species at the surface. It
otherwise stated, the scans began at the open should be noted that the oxidation peak at -0.1 V is
circuit (rest) potential and were swept in the not associated with a change in potential at the
negative direction, cycling between -0.8 and 0.85 electrolyte/surface interface, indicating that it this
V versus SCE at a rate of 0.1 mV/s. reaction does not change the concentration of
species in the electrolyte.
2.2 Development of the reaction through cycling To validate the experimental methodology and
and oxidation the use of technical grade FeS2, the experiment
was repeated using a sample of reactive ground.
Cyclic voltammetry involves the measurement of The reactive ground sample had been shown to be
current as the potential of the electrochemical cell reactive using the AEISG test and analysed by
is varied. The peaks in a voltammogram represent Bureau Veritas Analytical Laboratories and found
an electrochemical reaction, with the height and to be 37.5% sulfide, 13.6% sulfate and 0.15%
shape of the peaks governed by the nature of the elemental sulfur. Figure 3 shows that the
redox process, as well as more fundamental voltammogram for this sample in 1 M NH4NO3
variables such as electroactive species show the same oxidation and reduction peaks as
concentration and diffusion coefficient. the FeS2 sample in Figure 2.
Figure 2 shows the results from a sensor
electrode experiment of FeS2 in 1 M NH4NO3, 2.3 Identification of reactive peaks
with (a) the cell current, (b) the current response
from the sensor electrode set at 0 V versus SCE, To identify whether it was the NH4+ or NO3- ions
and (c) the open circuit potential or the electrolyte participating in the reactions observed in Figure 2,
adjacent to the FeS2 surface. The oxidation of cyclic voltammetry experiments were conducted
Fe(II) to Fe(III) is expected to occur at 0.45 V with FeS2 in electrolytes in which one or both of
versus SCE (Davison et al.). A reversible peak can these species were absent. Figure 4 compares the
clearly be seen at this potential. As the Fe(II) is voltammograms for FeS2 in 1 M NH4NO3 and 1 M
oxidised, the potential at the surface, shown in plot potassium sulphate (K2SO4). As can be seen, both
(c) increases. Trace (b) shows that a negative plots show an oxidation and reduction peak around
current is caused by ionic species produced on the 0.45 V versus SCE, corresponding to the

- 346 -
Figure 3. Voltammogram of reactive ground sample in 1M NH4NO3 showing (a) first and (b) fifth cycles.

Figure 4. Voltammograms of FeS2 in (a) 1 M NH4NO3 and (b) 0.5 M K2SO4.

- 347 -
 Figure 5. Voltammogram of FeS2 in (a) 0.5 M (NH4)SO4 and (b) 1 M NaNO3.

Fe(II)/Fe(III) redox pair. Above 0.5 V, the remained inactive. Reducing the minimum cycle
NH4NO3 voltammogram shows a reduction peak potential to -0.8 V did not increase the activity of
that is absent in K2SO4 plot. Figure 5 shows the the system, while increasing the maximum
voltammograms for FeS2 in 0.5 M (NH4)2SO4 and potential to 0.85 V increased reactivity. This
1 M NaNO3. The absence of this reduction peak in indicated that it was the reactions at high
the ammonium sulfate ((NH4)2SO4) electrolyte potentials that increase the reactivity of the
shows that the reduction is associated with the system.
nitrate species. Increasing the concentration of
NH4NO3 in the electrolyte to 2 M and above 2.5 Identification of the surface vs electrolyte
increases the size of this peak. These results are
significant as they show that the NO3- ion plays a While the above experiments determined that the
direct role in the reaction with FeS2, reacting on system became active with the application of high
the surface and not with Fe2+ in the electrolyte. potential, they could not discriminate between
changes in system activity was due to changes in
2.4 Development of the reaction peaks the concentration of species in the electrolyte or in
changes to the FeS2 surface. To determine this, a
The voltammograms of the first and fifth cycles of cell containing FeS2 and 1 M NH4NO3 was cycled
FeS2 in 1 M NH4NO3 are presented in Figure 6. until reaction peaks were observed. The electrolyte
This figure shows that the first cycle exhibited was then removed from the cell and the cell
very little activity, while the second and flushed and filled with fresh NH4NO3 solution.
subsequent cycles showed numerous oxidation and This cell was then cycled and the resulting
reduction peaks, indicating that redox reactions are voltammogram presented in Figure 8. This figure
taking place. This indicates that the system is shows that the first cycle of a spent cell with fresh
initially inert, but then becomes reactive after electrolyte had reaction peaks present and was that
being exposed to high cell potential. of an active system. The electrolyte that had been
The voltammograms of FeS2 cycled between removed from the cycled cell was rust-brown in
different potential ranges are presented in Figure colour, indicating the presence of iron oxides. This
7. When cycled between -0.4 to 0.5 V, the system electrolyte was placed in a second cell with fresh,

- 348 -
Figure 6. Voltammogram of reactive ground sample in 1M NH4NO3 showing (a) first and (b) fifth cycles.

Figure 7. Voltammograms of FeS2 in 1M NH4NO3 cycled with potential ranges versus SCE of (a) -0.4 to 0.5
V; (b) -0.8 to 0.5 V; (c) -0.2 to 0.85 V and (d) -0.8 to 0.85 V.

- 349 -
Figure 8. Voltammogram of FeS2 from a cycled cell with 1M NH4NO3 electrolyte removed and replaced
with fresh electrolyte.

Figure 9. Voltammogram of FeS2 with electrolyte from a previously cycled voltametric cell.

- 350 -
 Figure 10. Temperature time plot for FeS2 sample in NH4NO3 using the AEISG test method.

uncycled FeS2, with the voltammogram for this an FeS2 sample which had undergone reaction. To
cell presented in Figure 9. The voltammogram for obtain this reacted sample, 18 g of FeS2 was
this cell showed that the first cycle was not active, reacted with NH4NO3 in accordance with the
and that the system did not become active until AEISG procedure for reactive ground (AEISG
high potentials were applied. This indicates that 2017). The time versus temperature plot for this
the activation of the cell was due to changes in the sample is shown in Figure 10 and indicates that the
surface of the FeS2 itself, not due to the increase in sample was reacted to completion.
concentration of observable iron species, such as The solid residue from this test was collected
Fe2+ or Fe3+, or nitrogen species such as NO or and placed within a voltametric cell and cycled.
nitrite (NO2-) within the electrolyte. This is Figure 11 shows that unlike the unreacted FeS2
consistent with observations of the AEISG test, samples, the test residue sample showed activity
which show that while weathering solution, a on the first cycle and on subsequent cycles,
mixture of Fe2(SO4)3 and FeSO4 will reduce the even when the maximum potential was limited
induction time and therefore accelerate, an to 0.6 V. This result shows that, during
induction period is still required. This result is, the AEISG test, FeS2 goes from an inert
however, inconsistent with the Rumball surface to an active surface that displays similar
mechanism, which proposes that the transition reactivity to the FeS2 surface exposed to high
from an inert to an active system is due to the potentials.
increase in concentration of NO or nitrous acid, The presence of active FeS2 material at the
both of which would be within a recovered completion of the AEISG test indicates that
electrolyte. the completion of the reaction is due to the
consumption of available NH4NO3, not due to
2.6 Examination of a spent sample
complete dissolution of the FeS2. This has
To determine whether the changes in reactivity implications for the use of Urea as a reactive
observed in an electrolytic cell were consistent ground inhibitor, as this chemical which is
with changes in FeS2 observed during the reactive consumed in the initiation reactions. Once all Urea
ground reaction, voltammetry was performed on has been consumed, activated FeS2 will still be

- 351 -
 Figure 11. Voltammogram of FeS2 residue from the AEISG test in 1M NH4NO3.

present, leading to the potential for subsequent 3.1.1 Sample 1. acid washed FeS2
reaction with NH4NO3.
All samples were reactive on the first cycle, Technical grade FeS2 was obtained from Fischer
indicating that the removal of surface residues, Scientific, with a purity of >95%. Surface oxides
such as nitrate salts or soluble Fe(II) or Fe(III) were removed by washing with hot, concentrated
species or exposure to air at elevated temperatures hydrochloric acid (HCl 37 wt%) until the wash
does not reduce surface activity. Soaking the liquid ran clear.
residue in 1 M HNO3 for 24 hours did produce an
inert surface, indicating that active sites had been 3.1.2 Sample 2. oxidised FeS2
removed with the acid treatment, which may also
dissolve some of the FeS2 itself to reveal a fresh, 20 g of Sample 1 material was mixed with 5 g
unreacted surface. With the application of high water to form a slurry and placed in an oven at a
potentials, however, a reactive surface could be temperature of 100oC for 24 h. Thus, the sample
regenerated on this fresh FeS2 surface. was exposed to water and air at elevated
temperature.
3 SURFACE CHANGES
3.1.3 Sample 3. spent FeS2
The above results indicate that FeS2 is oxidised
from an inert to a reactive state in a voltametric 18 g of Sample 1 material was mixed with 18 g
cell, and that this reactive state is also found in the NH4NO3and weathering solution, a solution
solid residue of a reacted FeS2 sample. To further of FeSO4 and Fe2(SO4)3 at 55oC, and reacted
investigate surface changes of FeS2, oxidised and to completion, following the AEISG test
reacted samples were compared using SEM, XRD method. Solid material from this test
and XPS techniques. was recovered and washed with 1M
HCl and then filtered with demineralised
3.1 Sample preparation water in a Buchner vacuum filter until
the liquid ran clear. The sample was then
The following samples were prepared: allowed to air dry.

- 352 -
 Figure 12. SEM images of acid washed FeS2.

3.2 Surface results considerable SO4- on the surface layer.

3.2.1 Acid washed FeS2 Table 1. SEM backscatter analysis results for acid
washed FeS2.
The SEM images of the unreacted, acid washed
FeS2 sample at different magnifications are shown Element At % 1 Sigma
in Figure 12. This shows a framboidal structure, Oxygen 2.5 0.4
such that each grain consists of rhombic crystals Sulphur 59.6 1.8
aggregated together. Acid washing, which Iron 37.7 1.4
removed surface impurities, has etched the surface
of each crystal. Backscatter analysis of this
material shows that this sample is 59.6 at % 3.2.2 Oxidised Sample
sulphur, 37.7 at % iron and 2.52 at % oxygen. The
sulphur to iron ratio of 1.58 is interesting, as it The SEM images of the water-oxidised sample, at
shows that the surface is sulphur depleted. different magnifications, are given in Figure 13.
The XRD analysis of this sample did not This shows two distinct structures. The darker are-
identify any oxidised species. If it is assumed that as show plate-like inclusions, while the lighter ar-
all of the oxygen detected is due to sulphate eas show rhombic crystals, in which some of the
formation, then it may be calculated that 1% of the darker material is also present. Results of the
surface sulphide has been oxidised. This result was backscatter analysis of each of these regions is
confirmed with XPS analysis which showed given in Table 2.

Figure 13. SEM images of oxidised FeS2 samples showing (a) light material and (b) dark plate material.

- 353 -
 Table 2. SEM backscatter analysis of light and dark shows less distinct structures. The backscatter
regions of oxidised FeS2 samples. analysis of this material shows that the sample is
heavily oxidised, containing SO42-. The presence
Dark Plate of nitrogen, however, shows that ammonium
Material Light Area compounds are also present. This is consistent
At 1 At 1 with the XRD analysis presented above, which
Element % Sigma % Sigma showed the presence of (NH4)2Fe(SO4)2·6(H2O) an
ammonium Fe(II) sulphate and ammoniojarosite,
Oxygen 64.8 4.3 16.8 1.1 (NH4)Fe3+3(SO4)2(OH)6.
Sulfur 19.9 0.8 54.4 1.7
Iron 13.3 0.8 27.7 1.3 Table 3. SEM backscatter analysis of spent FeS2
sample.

The backscatter analysis shows that the dark Element [norm. at %] Error in wt.%
plate material is much higher in oxygen. The ratio (1 Sigma)
of oxygen, sulphur and iron is, given the margin of Nitrogen 4.5 0.5
error, is consistent with FeSO4.H2O, the species
identified by XRD analysis. Oxygen 61.1 4.1
The morphology of the lighter areas is Sulphur 16.6 0.7
consistent of the structure of unreacted FeS2. The
Iron 13.1 0.8
backscatter analysis of this sample shows that it is
higher in oxygen than the non-oxidised sample. If
it is assumed that the oxygen is in the form of
FeSO4.H2O, then the ratio of the remaining The above analysis indicates that oxidation of
Sulphur to Iron, is 2:1. This analysis indicates that FeS2 involves the formation of FeSO4. Reacted
the lighter material is FeS2 on which the dangling material appears to also contain FeSO4, but also
surface S- bonds have been oxidised to sulphate. contains ammonium compounds.
The XRD analysis of the dry material showed
that the surface contained 10.2% Szomolnokite, 4 EFFECT OF REACTION TEMPERATURE
FeSO4.H2O, indicating that a fraction of the
sulphide within the FeS2 has been oxidised to The SEM image of the water-washed solid residue
sulphate. No Fe(III) species were detected. from the reaction of 6 g FeS2, 6 g NH4NO3 and 1.2
g weathering solution at 55oC (Sample 6) is shown
3.2.3 Washed spent sample in Figure 15. This sample was visibly different to
the acid washed spent sample (Sample 3) obtained
The SEM images of the washed spent material, at from reacting 18 g FeS2, shown in Figure 12,
different magnifications, is shown in Figure 14. which did not show the distinct crystal structure on
This structure does not contain the distinct the surface. Backscatter analysis was used to
framboids of the unreacted material, but instead obtain the S:Fe:O ratio for this substance as

Figure 14. SEM image of spent FeS2 sample at low and high magnification.

- 354 -
 Figure 15. SEM image of sample of FeS2 after reaction with NH4NO3 and weathering solution at 55oC.

17.5:13.5:52.5. This ratio was found to be 24% Fe2(SO4)3, which is similar to that of Sample
consistent with a surface containing 58% FeSO4 3, given in Table 3 confirming that the differences
and 26% Fe2(SO4)3 with the remainder unreacted in surface were due to temperature.
FeS2.
It was speculated that differences between the 5 REACTION OF FES2 AND NH4NO3 WITH
residue from the reaction of 6 g FeS2 and 18 g UREA
FeS2 may be due to heat transfer, with the larger 5.1 Effect of temperature and urea
surface area to volume ratio of the smaller reactor concentration
vessel resulting in greater heat transfer, and thus
the reaction being maintained at 55oC for the Urea is a known inhibitor of the reactive ground
entire reaction. mechanism (Forshey Ruhe et al. 1968, AEISG
When the temperature of the water bath was 2017, Djerdjev Priyananda et al. 2018). To
heated to 75oC, there was a visible change in the understand the effect of Urea on the surface
reaction, with more vigorous gas evolution and a chemistry of FeS2, urea was added to the
more distinct colour change to red/brown. In FeS2/NH4NO3 weathering solution system.
addition, there was visible boiling of the water on Reaction was said to have occurred once either a
the outside of the reaction vessel, indicating that colour change or bubble formation could be
the internal temperature was greater than 75oC. observed. The time taken to react varied with the
The SEM image of the solid residue is given in mass fraction of urea mixed with NH4NO3. A plot
Figure 16. This surface had an obviously different of time to reaction vs urea mass fraction at 55oC is
surface structure to that shown in Figure 14, with given in Figure 17.
the absence of surface crystal structures and Inhibition by Urea is also known to be affected
increased pitting and cavitation. Backscatter by temperature, with increased temperature
analysis indicated that the S:Fe:O ratio was reducing the time to reaction (Rumball 1991,
10:13:60 and that the surface was 76% FeS2 and Djerdjev Priyananda et al. 2018). Urea was found

- 355 -
Figure 16. SEM image of sample of FeS2 after reaction with NH4NO3 and weathering solution after
reaction at >75oC.

Figure 17. Time to reaction vs mass fraction Urea at 55oC.

- 356 -
 Figure 18. SEM image of sample of FeS2 in NH4NO3, 1% Urea and weathering solution at 55oC, prior to
visible gas evolution and colour change.

to inhibit reaction between 6 g FeS2, ammonium 5.2 Examination of the surface

nitrate and 1.3 g weathering solution at a
concentration of 1% Urea at up to 90oC. At 0.75% A mixture of 1.3 g weathering solution, 6 g acid
Urea, reaction occurred at 75oC after 42 minutes, washed FeS2 and 6 g NH4NO3 containing 1%
and for 0.625% Urea, after 49 minutes at 55oC. Urea, was heated to 55oC. Figure 18 shows the
Reaction proceeded with bubble evolution and SEM image of the solid sample taken at 30
colour change. minutes. Gaseous reaction and colour change was

Figure 19. SEM image of FeS2 surface after 50 minutes in ammonium nitrate containing 0.75% Urea at 55oC,
prior to visible reaction after temperature increased to 75oC.

- 357 -
 Figure 20. SEM image of FeS2 surface after reaction at 75oC with ammonium nitrate containing 1% Urea.

not observed after 95 minutes at this temperature. temperature was increased to 75oC, whereupon
These images show that while no gas evolution or reaction occurred within 2 minutes. The SEM
electrolyte colour change was observed, a layer of image of the surface after reaction is presented in
oxidised material had begun to form on the surface Figure 20. Like the SEM images of reactions at
of the acid washed FeS2. this temperature in the absence of Urea, presented
A mixture of 1.3 g Weathering Solution, 6 g in Figure 20, the FeSO4 in the surface was
acid washed FeS2 and 6 g NH4NO3 containing removed by reaction.
0.75% Urea, was heated to 55oC. SEM images of
the solid material taken after 50 minutes, prior to 6 REACTIVITY
visible reaction, are presented in Figure 19. This
shows that the FeSO4 crystals are larger and cover Reactivity of samples was assessed by performing
a greater area of the FeS2 surface than those the AEISG test at 1/3 scale, using a water bath to
presented earlier. After sampling, the reactor maintain sample temperature. Initial results for

Table 4. Results for reaction between 6 g NH4NO3 and 6 g FeS2 samples with 1.3 g weathering solution at room
temperature, 40oC and 55oC.

Sample Description Temperature Peak potential Comments

1 FeS2 in NH4NO3, with 55oC 0.7 Rapid Reaction at 14 minutes.
weathering solution
2 FeS2 in NH4NO3, with 40oC (1 hr) 0.67 No reaction after 1 hr at 40°C,
weathering solution then 55oC rapid reaction once increased
to 55°C
3 FeS2 in NH4NO3, with Room temp 0.44 No Reaction observed over 4
weathering solution (22oC) days.

- 358 -
Figure 21. Plot of potential and temperature vs time for 6 g FeS2 in 6 g NH4NO3 and 1.3 g water at 70oC,
showing (a) sample temperature, (b) water bath temperature and (c) potential.

Figure 22. Plot of potential vs time for 1.3 g weathering solution and 6 g sodium nitrate placed in a
water bath at 70oC.

- 359 -
 Figure 23. Plot of Fe(II) concentration vs time for 0.56 g FeSO4 and 1.26 g Fe2(SO4)3 in 1.5 g water in (a) 6 g
water, (b) 6 g NH4NO3, or (c) 6 g NH4NO3 mixed with 6 g FeS2, as measured by KMnO4 titration.

FeS2 are given in Table 4. For the mixture of weathering solution and
While a reaction could clearly be observed, NH4NO3, there is a decrease in the Fe(II)
through visible colour change and bubble concentration, which corresponds to the visual
evolution, an exotherm could not be detected. observation of bubble generation and colour
To measure the onset of reaction, the cell change. The Fe(II) concentration then falls to
potential was measured using a calomel electrode. around zero and reaction ceases. For the mixture
The plot of temperature versus time for reaction that also contains FeS2, the Fe(II) concentration
conducted at 70oC is given in Figure 21. It remains above zero for the entire test period,
clearly shows a change in potential at the onset despite a more vigorous reaction being observed.
of reaction. The above results indicate that while Fe(II) is
The change in cell potential indicated that the reacting with NO3-, oxidising Fe(II) to produce
ratio of Fe(II) to Fe(III) was changing in the Fe(III), the reaction is more vigorous and is
electrolyte. Figure 22 shows that this change in significantly more exothermic in the presence of
cell potential was also observed when weathering FeS2. If the concentration of Fe(II) is not falling to
solution (a mixture of Fe(II) and Fe(III) sulphates) zero in the presence of FeS2, it implies that
was mixed with ammonium nitrate in the absence additional Fe(II) is being formed in the reaction
of FeS2. While a colour change and minor gas process. Authors such as Rumball (1991) have
evolution was observed, the reaction was far less stated that Fe(III) will oxidise FeS2 by the
vigorous than when FeS2 was present. following reaction:
The concentration of Fe(II) in solution could be
determined through titration with KMnO4. Figure Reaction 3 FeS2 + 14Fe3+ + 8H2O => 15Fe2+ +
23 shows that for all three reaction systems of 2SO42- + 16 H+ (R.3)
Fe(II) in water, NH4NO3 or NH4NO3 and FeS2, the
Fe(II) concentration initially increased. This may Rates of reaction are also temperature
be due to an initial change in the equilibrium dependent, with time to reaction onset decreasing
between species in solution, with hydration and with increased temperatures. While no reaction
pH change affecting solubility. In the absence of exotherm was investigated for water bath
NH4NO3, the Fe(II) concentration for weathering temperatures less than 70 degrees Celsius,
solution was found to decrease, and then return to gas and steam evolution was observed
this equilibrium value. when the bath temperature was greater than 75

- 360 -
degrees, indicating that thermal runaway had reactions (R.22), (R.28) and (R.29). Thus, the
occurred. reaction will not become extinguished with the
consumption of Fe2+, and so the exothermic
7 THE REACTION MECHANISM reaction will be able to continue to accelerate,
leading to greater gas generation and a
The above reactions show that during the initial
significantly larger exotherm.
reaction phases, FeS2 is oxidised to form FeSO4 on
In an insulated system, the heat of reaction
the surface, and that this FeSO4 reacted directly
would heat the reactants, increasing the reaction
with nitrate at the onset of reaction. While the
temperature and reaction rates, and thus creating a
presence of FeSO4 will make the sample more
positive feedback loop that would lead to thermal
reactive, the presence of an oxidised layer, or the
runaway. In a non-insulated system, however,
addition of weathering solution, did not promote
heat losses result in only a fraction of the heat
immediate reaction unless the temperature was
generated by reaction contributing to a
increased to greater than 75oC.
temperature increase. The greater the temperature
For reactions between ammonium nitrate and
difference between a heated system and its
FeSO4 in the absence of FeS2, initial reactions
surroundings, the greater the heat loss, as given by
must be either:
the equation:
3 Fe2+ + NO3- + 4H+ ⇋ 3 Fe3+ + NO(g) + 2H2O Q = UAdT (1)
(R.22)
At steady state, the temperature of the reactor
2 Fe2+ + NO3- + 3H+ ⇋ 2 Fe3+ + HNO2 + 2H2O will be that temperature at which the reaction
(R.28) produces heat at a rate that matches the heat flux.
The Arrhenius equation shows that reaction rate
The build-up of a FeSO4 on the sample surface
constant, and therefore the heat generated by the
shows that, initially, these reactions that consume
reaction is exponential with temperature, while
Fe2+ must be slower than the surface oxidation
Equation 1 shows that heat flux is linear. Figure 24
reactions. Once reaction (R.28) had produced
shows the exponential curve for heat generation
sufficient HNO2, then reaction (R.29) could also
and the linear curve for heat flux in a hypothetical
proceed:
reaction scenario. At low temperatures, while the
Fe2+ + HNO2 + H+ → Fe3+ + NO(g) + H2O reaction rates are low, there is only a small
(R.29) difference between the exponential curve for heat
of reaction and the linear plot, and so the
Each of these reactions are exothermic, and so temperature rise would be small. At higher
as reactions proceed, the reaction system will heat temperatures, however, the exponential nature of
up and this increase in the reactor temperature will the curve means that reaction rates increase rapidly
increase the reaction rate. Thus, in the absence of and the difference between heat generation and
FeS2, reaction (R.29) will be limited by the heat flux increases, and so thermal runaway would
availability of HNO2 and all reactions will be expected.
accelerate once the temperature is increased by In some systems heat loss from the sample is
reaction. As has been described, however, the sufficient to maintain reactor temperature,
temperature increase is too small to be detected in preventing reaction rates from increasing further
the AEISG test and the reaction ceases once all and preventing thermal runaway. This was
Fe2+ has been oxidised to Fe3+. observed in small reactors where, below 75oC, the
For a reaction system containing FeS2, the reactor would not go above the water bath
description of reaction is similar except that temperature even when reaction was observed. It
reactions (R.3) and (R.30) are possible: was also observed that when a reaction was
cooled, reaction rates could be slowed sufficiently
FeS2 + 14Fe3+ + 8H2O → 15Fe2+ + 2SO42- + 16 H+ for visible gas evolution to cease. As water was
(R.3) lost through the reaction process, the remaining
ammonium nitrate solution would crystallise,
FeS2 + 5NO3- + 6H+ → Fe3+ + 2HSO4- + 5NO(g) + leading a solid reaction product.
2H2O (R.30) It should be noted that during this initial
reaction time, the bulk of the reactor may not
With these reactions, Fe2+ will continue to be change in temperature by a measurable amount.
generated, replacing the Fe2+ consumed in Crystal defects in the FeS2 will lower the

- 361 -
 Figure 24. Plot of (a) heat flux and (b) heat generation from reaction vs temperature for a hypothetical reaction
system in an uninsulated reactor.

activation energy and thus have a higher reaction vigorous reaction, however, mass transfer may
rate. This may result in only one or two reactive limit a system and result in unreacted material
sites that begin to react and create local ‘hot remaining in the reactor vessel after reaction has
spots’, which may only constitute a small fraction ceased.
of the material in the reactor, or may even be
single active sites on which Fe(III) will oxidise the 8 CONCLUSIONS
FeS2 crystal surface. In the absence of FeS2,
localised reaction has also been observed on the Commercially obtained FeS2 could be used as a
crystal surface of undissolved ammonium nitrate. substitute for reactive ground samples to study the
Once reaction has begun in these spots, then the reactive ground mechanisms. Electrochemistry
temperature will increase locally, and reaction may could be used to identify reaction peaks associated
accelerate quickly even if the bulk temperature is with the reactive ground mechanism, and that
largely unchanged. It is only once reaction in these changes in electric potential could also be used to
hot spots becomes sufficiently vigorous that bulk identify the onset of reaction in a system where
temperature rise may become measurable, and the temperature rise could not be determined.
temperature rise is sufficient for reaction Cyclic voltammetry showed that acid washed
throughout the sample. samples of FeS2 are initially inert, and that cycling
There are also mass transfer considerations. In to high potentials created a reactive surface.
order for a reaction to take place, there need to be Oxidised or reacted samples showed reactivity
sufficient reactants present. While initially, without exposure to high potentials.
NH4NO3 will be in excess at the FeS2 surface, SEM imagery showed that oxidation products
oxidation of the surface will deplete the NO3- are formed on the surface of FeS2, and that this
concentration, and reaction may even be surface was maintained for reaction at 55 C, but
insufficiently fast to generate the exotherms for the that the oxide surface was removed with higher
reaction to be sustained. An oxidised surface, temperature reactions. These iron oxide products
however, will have FeSO4 available for reaction react exothermically with nitrates, and so different
without depleting NO3-, and thus will appear levels of oxidation could therefore explain the
‘more reactive’ than a non-oxidised sample. Once differences in reactivity between FeS2 samples.
reaction has begun, and gas evolution is occurring, Reaction rates were found to be highly
mixing will aid mass transfer. Even with a temperature dependent, and thus the transition

- 362 -
 Figure 25. Photo of spent reactor vessel, showing reaction zone (light), Fe3+ rich material (top) and a dark layer
of unreacted FeS2 in vessel base (bottom).

from a slow reacting system to thermal runaway Davison, W., Buffle, J. & DeVitre, R, 1998.
was governed by heat and mass transfer to and Voltammetric characterisation of a dissolved iron
from the reaction zone. Urea was observed to sulphide species by laboratory and field studies.
inhibit reaction, but an oxide layer was still formed Analytica Chimica Acta, 1998. 377(2-3):
on FeS2 in the presence of Urea. p. 193-203.
The Reactive Ground Mechanism is a series of
reactions, whose rates are governed by heat and Dos Santos, E.C., Silva, J.C.M. & Duarte, H.A. 2016.
Pyrite oxidation mechanism by oxygen in
mass-transfer. This novel mechanism differs from
aqueous medium. Journal of Physical Chemistry
that proposed by other authors as it does not rely
C, 2016. 120(5): p. 2760-2768.
upon a build-up of reactants or intermediate
species to transition from a slow reacting, inert Garrels, R.M. & Thompson, M.E. 1960. Oxidation of
system to thermal runaway. pyrite by iron sulfate solutions. American Journal
of Science, 1960. 258: p. 57-67.
REFERENCES
Rumball, J.A. 1991. The interaction of partially
AEISG, Code of Practice Elevated Temperature and weathered sulphides in the Mt McRae shale
Reactive Ground. 2017. formation with ammonium nitrate. Department
of Chemical Engineering. 1991, Murdoch
Chandra, A.P. & Gerson, A.R. 2010. The University.
mechanisms of pyrite oxidation and leaching:
A fundamental perspective. Surface Science
Reports, 2010. 65(9): p. 293-315.

- 363 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Ground improvement blasting - further development of the BDR

classification scheme

T.N. Little
Director and Principal Consultant, Blasting Geomechanics Pty Ltd, Perth, Western Australia

ABSTRACT: This paper reports on the further development of the Blast Design Requirements (BDR)
classification system developed by the author, mainly extended to include ground improvement blasting
techniques. The full updated 2022 version of the BDR classification system is provided. The body of the paper
explores eight ground improvement blasting applications. Two case studies are used to demonstrate the BDR
classification system approach in the context of void treatment blasting. Finally, based on this work a number
of observations and conclusions are formulated. The eight ground improvement blasting applications discussed
in the paper are: Compaction blasting which aims to increase the in-situ density of the blasted material;
Destress blasting can be used to redistribute local stresses by transferring them to a non-fractured zone ahead
of the destressing zone and benefits mining personnel by creating a safe low stressed ‘cushion’ of rock at the
immediate mining face; Disruption blasting aims to increase the shear strength of weak layers or zones and
thereby improving ground stability; Hazard release blasting involves eliminating mass movement (gravity
driven) hazards; Permeability enhancement blasting aims to increase the permeability (p) of a volume of
ground for a beneficial purpose; Restoration blasting can be used on quarried rock slopes to construct a
landform sequence similar to that in the surrounding natural landscape; Reinforcement blasting aims to
reinforce the target volume weak ground which is below water; and Void treatment blasts that are used to
collapse. Two case studies using the BDR classification system are discussed, the first relates solely to
collapsing and filling an underground void network using a void treatment blast (Blast A). The second case
study relates to a more complex blast that includes void areas, ore and waste materials, and is adjacent to pit
wall on one side. The second case study blast, is termed a complex modified production blast (Blast B).
Based on this current work five important observations and conclusions are made: (i) The use of the BDR
classification system makes the blast design process easier, reduces the requirements risk, provides design
priorities and makes them more explicit. (ii) The range of ground improvement applications using explosive is
large and new applications are expected to emerge in the future. Blasting using explosives can be used to:
densify soils, destress rocks, strengthen weak rock zones, restore landscapes, enhance drainage pathways,
reinforce weak earthen materials and eliminate dangerous near surface cavities. (iii) The range of ground
improvement blasting applications demonstrates that blasting with explosives is very versatile as it can be
adapted to many different functions or activities. (iv) Blasting operations are unique in four important ways:
They are high hazard, strongly regulated and controlled leading a low residual risk; they have rapid feedback
cycles; they can be used to leverage downstream processes; and are very versatile (they can be adapted to
many different functions or activities). (v) The requirements approach presented here, if adopted, should
improve the quality of blast designs and hence blast outcomes leading to practitioner and stakeholder
satisfaction. It is also expected to assist with lateral thinking and innovative blast design strategies that will
benefit society in the areas of safety, economics, aesthetics, environment and engineering.

1 INTRODUCTION hazard but low residual risk), they have rapid

design and implementation feedback cycles (need
Blasting operations are unique in three important cautious start and then accelerated learning), and
ways: they are potentially high consequence (high they occur early in the value chain of extraction

- 364 -
operations (can be used to leverage downstream ‘Blast implementation management process’
processes). Because of these three unique and involves three time-sequenced sub-elements:
important features a blast management framework pre-blast, during blast, and post blast.
consisting of design management and The BDR Classification systems was developed
implementation management is indicated. The in part to management a blasting operation’s
process to achieve sustainable blasting operations requirements risk. Requirements risk is the
is illustrated in Figure 1 and termed the ‘Blast potential for losses due to a project’s requirements
themselves or requirements management process.
Such risks are closely tied to the quality of
requirements analysis or the absence of such an
analysis. The absence, or low-quality requirement
analysis represent a risk to a project as garbage-in-
garbage-out, meaning that projects that are based on
faulty requirements are likely to face issues and
potentially fail. Figure 2 presents the managerial
and technical blasting objectives required to
achieve sustainable blasting operations. These are
now discussed in the following paragraphs.

1.1 Managerial blasting objectives

Figure 2 indicates that all blasts need to satisfy the

managerial blasting objectives. That is, all blasts
Figure 1. Components of a Blast Management
Framework (Little 2018). must be safe, legal, add value, manage any special
conditions and work within site constraints.
It should be noted that, sustainable blasting
Management Framework’. This process has two operations will be those that consistently achieve
major components: ‘Blast Design Management both the managerial blasting objectives and
Process’ and ‘Blast Implementation Management the primary technical blasting objectives.
Process’. Figure 1 indicates that the ‘Blast design This starts with blast design which is the
management process’ involves two sub-elements, fundamental engineering control to obtain
these are: blast design requirements, and blast consistent output that achieves all relevant
design and analysis. The second element is the blasting objectives.

Figure 2. Managerial and technical blasting objectives (Little 2019).

- 365 -
 Table 1. Technical blasting objectives (Little 2019).

Technical Blasting Objective Interpretation in soil and rock blasting context

G Grade /Ore control Reducing unplanned ore loss and unplanned dilution.
F Fragmentation control Obtaining the desired fragment distribution.
D Damage control Maintain the strength of the remaining rock mass. Reduce
damage to valuable minerals. (fines or cracks)
P Profile control Reducing over-break, underbreak and bridging.
M Muckpile control Obtaining required muckpile and ore flow characteristics.
E Environmental control Reducing unwanted blasting emissions e.g. vibration, airblast,
fumes, dust, flyrock, water pollution and asset damage.
I Ground Improvement control Improving the physical, structural, mechanical or hydraulic
properties of the ground.

geological materials in a blast volume, the

1.2 Technical blasting objectives orientation of the free face relative to the
blastholes, the number and type of primary
The seven technical blasting objectives shown in blasting objectives. The first three classification
Figure 2 are briefly summarised in Table 1. elements are objective physical properties and the
fourth relates to technical blast design objectives.
2 THE 2022 UPDATED BDR
The BDR classification scheme has the following
CLASSIFICATION SCHEME
format X_XG_XF_XXXXXXXO with possible
This section presents the 2022 update, to the Blast value of X explained in Figure 3.
Design Requirements (BDR) classification scheme 2.1 X-Blast location relative to the earth’s
(Little, 2011, 2015, 2017, 2019). The BDR surface
classification scheme uses four classification
elements: the blast location relative to the earth’s This element relates to the blast location relative to
surface, the number of different grades of the earth’s surface. Only three options have been

Figure 3. The BDR classification scheme.

- 366 -
proposed. The X in the X_XG_XF_XXXXXXXO Environmental control objectives sub-divisions:
can take the values of S for a blast on the Surface,
U for an Underground blast and W for an - Ea – Airblast sensitive blast
underWater blast. - Ed - Dust sensitive blast
- Ef – Flyrock (wild) sensitive blast
2.2 XG-Number of different grade geological - Eg – Fume sensitive blast
materials in a blast - Ev – Vibration sensitive blast (heritage and
infrastructure protection)
In this paper the term bulk blast is used when two - Ep – Water pollution sensitive blast
or more blocks of materials with different grade - Er – Land pollution sensitive blast (from
categories (e.g. high grade, low grade, waste) are explosives residue).
shot in the same blast. In contrast, the term
selective blast is used for a blast containing only Ground improvement sub-divisions:
one grade category (e.g. high grade ore).
The X in the XG can be S or M with the - Ic – Compaction blast
following meanings: - Is – Destress blast
- Id – Disruption blast
SG - Single ore blast (selective ore blast) - Ip – Permeability enhancement blast
MG - More than one grade of material in the blast - Ir – Restoration blast
(non-selective bulk blast) - Ih – Hazard release blast (scaling etc)
WG - Single waste blast (waste only). - If – Reinforcement blast
- Iv – Void treatment blast
2.3 XF-Orientation of the free face
2.5 Some basic blasting objective rules
A distinction is also made as to the orientation of
the closest free face (F), the three general classes - All managerial blasting objectives must be
are: horizontal (e.g. paddock blast), vertical (e.g. met for every blast, e.g. safe, legal, value
bench blast) and combination (e.g. sub-vertical and adding
horizontal within the same blast). - All blasts will have one or more technical
The X in the XF can be H, V, O or A with the blasting objective(s) and these are called
following meanings: primary blasting objectives and need to be
listed in priority order
HF – Horizontal free face with blasthole - Primary blasting objective implies that it
perpendicular to the free face needs to be explicitly addressed during the
VF – Vertical (or sub-vertical) free face with blast design process and will also require
blasthole parallel to the free face performance targets to be set
OF – Horizontal or vertical free face (form of - Blasting objectives that are relevant to the
short hand) blast context but not considered to be primary
AF – Horizontal and vertical free face (complex for a particular application are termed
free face conditions). secondary blasting objectives
- All secondary blasting objectives need to be
2.4 XXXXXXXO-Number and type of primary kept within acceptable levels
blasting objective in priority order - If any of the primary blasting objectives are
not met the blast would be considered to be
The BDS classification system allows up to seven
less than adequate and design refinements
primary blasting objectives but it is generally
will be required for similar future blasts.
limited to three or four primary objectives. It
should be clear that the more primary blasting 3 TYPES AND CLASSIFICATION OF
objectives the more complex the design trade-off GROUND IMPROVEMENT BLASTS
become.
The X’s in the XXXXXXXO component can Ground Improvement blasting (I) was added to the
be any one of the following, but repetition is not BDR classification scheme (Little 2019) without
permitted: G – Grade control, F – Fragmentation any explanations. That addition is considered
control, M – Muckpile control, D – Damage valid, as the overall aim is for the BDR
control, E – Environmental control (see classification scheme to be able to accommodate
sub-categories), P – Profile control and I – Ground all ground blasting applications. This section of
Improvement (see sub-categories). the paper aims to provide sufficient detail and

- 367 -
Table 2. Ground improvement blast classified using the BDR classification system.

Blast type Classification Primary blasting objective

Compaction blasting S_SG_HF_IcO To compact a volume of ground. Use Ic for this sub-
objective.
Destress blasting S_SG_OF_IsO To reducing the stress (destress) in the target volume of
ground.Use Is for this sub-objective.
Disruption blasting S_SG_HF_IdO To increase the shear strength by disruption of a
volume of ground containing weak layers. Use Id for
this sub-objective.
Hazard release blasting S_SG_OF_IhO To eliminating mass movement (gravity driven)
hazards using blasting. Use Ih for this sub-objective.
Permeability S_SG_HF_IpO To enhance the permeability of a volume of ground.
enhancement blasting Use Ip for this sub-objective.
Reinforcement blasting W_SG_HF_IfO To reinforce the target volume of ground which is
below water. Use If for this sub-objective.
Restoration blasting S_SG_VF_IrO To achieve landform modification for engineering or
environmental purposes. Use Ir for this sub-objective.
Void treatment blasting S_SG_OF_IvO To collapse and fill voids (natural or man-made) for
safety and production purposes. Use Iv for this
improvement sub-objective.

Table 3. Ground improvement specification and example acceptance criteria.

Blast type Specification Example criteria

Compaction blast (Ic) Density requirements - Cone Penetration Test (CPT)
Surface settlement - Amount of settlement (mm)
Destress blast (Is) Defined relaxed zone - Local stress reduction
Local stress transfer mechanism - % reduction in seismic events
Disruption blast (Id) Shear strength (C and 𝛷𝛷) - Area disrupted (improved)
Increased stable angle - Increased stable angles
(degrees)
Hazard release blast (Ih) Number of identified rockfall hazards - # reduced hazards per km
Critical hazards removed - # of hazards made safe
Cost effective risk reduction - Cost-benefit ratio
Permeability Well defined volume - Water pressure (kPa) reduction
enhancement blast (Ip) Particle size distribution requirement upgradient of blasts
Swell (and void) distribution - Increased flow rates (m3/d)
Permeability enhancement target - Increased hydraulic
Fines limitations conductivity(m/s)

Restoration blast (Ir) Landform design specification - Stakeholder acceptance

Readiness for subsequent activity - Compliance to specification
Reinforcement blast (If) Replace underwater silt with crushed - Rate of advance and
rockfill repair/rework
Void treatment blast (Iv) Defined void shape based on old plans - # voids collapsed and filled
Delineated by probe drilling - % Danger zones on pit floor

- 368 -
 Figure 4. Ground improvement classification and objectives.

Figure 5. Geotechnical aspects of compaction blasting (Shakeran et al. 2016) and relevant photograph.

explanation for organisations that use ground charge weight in each hole, (ii) depth of charges in
improvement blasting and for blast designers to the hole, (iii) scattering pattern of charges in
understand the various concepts. In addition to height, (iv) distance between the blast holes, (v)
the primary blast objectives discussed in Table 2, phasing and the number of blast stages, and (vi)
all ground improvement blast will need to develop sequence of blast holes and explosive scheduling
or externally imposed quantitative or qualitative in each phase. Determination of these design
specifications and acceptance criteria (KPIs). Table parameters should be based on the final optimum
3 provides measurement techniques and example results. Figure 6 illustrates a real plan and profile
criteria for each of the eight ground improvement of blast holes arrangement along with the charging
categories. locations and sequences of explosion.

3.1 Compaction blasting (Ic) 3.2 Destress blasting (Is)

This method is also called explosive compaction. Destress blasting (sometimes called precondition-
Unconsolidated materials (soil, sand, silt, mud, ing blasting) is a pro-active control measure used
tails) are blasted in such a way that they contain underground which guards against strain bursting
less voids and water and hence become denser. or similar seismic phenomena. The three main
Effective parameters in design of compact mechanism classes are: Strain bursts/face bursts;
blasts fall in six categories are as follows: (i) Pillar bursts; and Fault slip bursts (Yao et al. 2016,

- 369 -
 Figure 6. Plan of blast pattern and sequence with typical blasthole charge design (Shakeran et al 2016).

Mitri 2001). In hard rock mines, destress blasting characteristics of rock mass and hence change
is thought to be a means of maintaining and its failure mechanism from brittle elastic to
extending natural fracturing around an excavation. yielding type. Successive destress blasting
It should be noted that the mine services and applications develop aprogressive yielding to
infrastructure need to be protected prior to the maintain the cushion as the face advances.
de- stress blasts. It is perceived that inducing Furthermore, continued yielding dissipates
fractures by blasting in the rock ahead of the face continued loading of the destressed zone.
can produce the following effects: Therefore, the destressed zone yields
gradually rather than fails suddenly and
- Local stress transfer by fracturing: Destress violently.
blasting induces a set of fractures in the rock Figure 7(a) shows that the aim of destress
mass. This fracturing results in local blasting, when developing in a highly stressed
readjustment of stresses by transferring them environment, is to create a fractured destressed
to a non-fractured zone ahead of the zone and hence push the peak stress away from the
destressing zone. The benefit to the mining drift face. Figure 7(b) illustrates a simple destress
personnel is that a low stressed ‘cushion’ of blast design to achieve such stress and ground
rock is produced in the immediate face. conditions.
- Modifying rock mass properties: Fracturing
by destress blasting reduces the load-carrying 3.3 Disruption blasting (Id)
ability of the rock apparently by reducing the
confined rock mass strength and modulus of The aim of this method of blasting is to disrupt the
elasticity. The collective results of this change continuity of weak rock seams or zones where the
in the properties is stress decrease in the blasted material will largely be left in place, but
fractured zone. stronger, to form the batter or slope face. Charges
- Modifying rock mass failure mechanism: above or below (or both) the weak zone can be
Rockburst is normally associated with brittle used to move stronger materials along a potential
elastic rock failure. Fractures induced by failure path. This method is not new, the earliest
destress blasting provoke yielding example known was in civil works in Tennessee

- 370 -
 Figure 7a. Destress blasting concepts, (a) pre-blast and post-blast stress conditions and fractured
destressing zone (Roux et al. 1957).

‘shot-in-place buttress’ in the late 1960s (Moore technique is generally applied where the seam dip
1986). Two methods of disruption blasting are lies between 5° and 20°. To achieve best results,
discussed in this section. They are: Targeted buffer the blast must be strong enough to disturb the
blasting and Bench buffer blasting (Softwall ground, but not completely pulverise it. The
blasting). disruption blast disturbs the continuity of the
Targeted buffer blast strip blasting (Latilla et bedding plan shears (or weak zones) and results in
al. 2016) is designed to disrupt identified plane(s) an increased cohesion and friction angle, and
of weakness, disturbing them in order to increase consequently a higher factor of safety or slope
cohesion and friction angle. The explosive charge reliability.
weight per hole is generally significantly less than Softwall blasting (Kelso 2011) also called a
that used for a production hole of the same depth. bench buffer blast (Latilla et al. 2016) is used
Once exposed, the batter or slope will appear less where the entire batter face and the bench behind it
damaged than it would in a normal buffer (or is assessed as being so structurally disturbed that it
softwall) blast. A secondary advantage of targeted is better to blast it and disrupt the continuity of all
buffer blasting is improved drainage, which lowers structure, as far as is practical. The batter and
the phreatic surface. It should be noted the bench are blasted with a similar charge weight as a
scheduling of a disruption blast must be such that normal production blast and the blasted material is
the buffer strip is blasted before mining of the coal battered back to between 40.0° and 45.0°.
or other strata down dip from it. In other words, This method would generally be used where the
while the planned buffer blast toe is still buttressed majority of structures are dipping at over 20.0°.
by solid rock or coal. Targeted buffer blast strip is Kelso (2011) discussed a series of highwall
utilised where a target zone, usually a coal seam instability events developed into a 700 m-long
containing bedding plane shears, has been tension crack located 80 m from the crest, resulting
identified. The intention of the targeted buffer in a large block of the highwall creeping towards
blast is to disrupt the bedding plane shears at seam the active open pit workings. The risk of further
level and then displace the rest of the overlying highwall instability was high with the potential for
strata without completely fragmenting it. This premature closure of the pit and loss of coal

- 371 -
 Figure 7b. Destress blasting concepts (b) a simple six hole destress blast for horizontal development
(Carr et al. 1999).

Figure 8. (A) Toe position with reference to targeted buffer blast strip after failure (B) example of bedding
plane shear (Latilla et al. 2016).

reserves at a time of high global demand for are too large for removal using scaling bars
quality coking coal. The softwall method consists and jacks. It can also be used to remove portions
of a highwall slope design where the rock mass is of slopes that could act as features to project
blasted beyond the pit limit to disrupt rock defects. falling rocks from the face towards a sensitive
The softwall design was successfully adopted as structure.
the preferred geotechnical slope design to manage Example 1. Elimination of potentially unstable
potential highwall instability in the final mining rock hazards: Trim blasting is an effective method
strip for the pit. of removing potentially unstable rock masses.
3.4 Hazard release blasting (Ih) Effective trims are undertaken by experienced
blasters who are in contact with the Engineer on
Rock slope engineering generally addresses the site as well as the Owner. Other key considerations
formation of new slopes, but it can be employed to in trimming include analysing the geological and
reduce hazards by remediating existing cut or geotechnical properties and noting nearby
natural slopes (Example 1). A second application sensitive structures or other nearby features that
is an innovative approach used at a coal mine in may be affected by blasting. When estimating PPV
Australia (Example 2). A third example relates to at a sensitive structure the Engineer should be
avalanche control blasting is presented as Figure cognizant of the effects of a blast shadow on PPV.
9(d). In rock slope remediation, hazard reduction The blast plan and post blast record are also
blasting (sometimes called trim blasting) is used important documents that should be provided by
to remove unstable rock masses on slopes that the blaster and reviewed by involved parties.

- 372 -
 Figure 9. Overview of elimination of structure controlled highwall failures (a) Risk elimination trim blast
drilling controls. (b) Photo of the trim blast wedge for the third strip. (c) Survey scans for all mined and
mining strips so far showing successful risk elimination outcomes. (d) Photo illustration of avalanche
control blasting.

Every borehole should be considered an when drilling and loading of blasts is being
investigation hole, and the blast should be conducted. This way there is a deeper
designed to the conditions of the rock mass rather understanding of the rock mass and construction
than the conditions in the blast plan. It is the issues that may arise. It can also be verified that
Authors’ recommendation to be present on site the blast plan is being followed, including how the

Table 4. Four permeability enhancement blasting application.

Application Issue/Opportunity Possible blasting

1 Developing Enhance permeability for mine See design and implementation discussion
drainage pathways dewatering or depressurisation. that follows.
2 Improving coalbedLow gas permeability restricts the Deep-hole permeability improvement
gas extraction improvement of coalbed gas technique of blast fracturing has been
extraction efficiency. applied to improve coalbed gas extraction
rates, with favourable results.
3 In-situ retort Oil shale contains a solid In a true in-situ (TIS) process, a deposit is
hydrocarbon called kerogen that first fractured by explosives and then
when heated (retorted) yields retorted underground. In modified in-situ
combustible gases, crude shale oil, (MIS) processing, a portion of the deposit
and a solid residue called spent, is mined and the rest is shattered
retorted, or processed shale. (rubbleised) by explosives and retorted
U/G.
4 In place leaching A modified in-situ leaching method Permeability enhancement achieved by
for extracting gold using low blasting. A major attraction is recovery of
environment impacts. metals without need open pit excavations,
tailings dams or surface waste dumps.

- 373 -
boreholes are delayed and where delays are being 3.5.1 Opening up drainage pathways
placed. Trim blasting is an effective way to
remove potentially unstable blocks of rock that This involves using explosives in carefully placed
may be endangering public safety and critical drillhole to fracture the rockmass in strategic
infrastructure. locations to open up drainage pathways. This
Example 2. Elimination of structure controlled changes the natural drainage pathways by
highwall failures at an open cut coal mine: increasing the fracture permeability in the targeted
The geotechnical recommendations were to blast location. The benefits of such an approach may be:
the unstable ground first (called trim blasting),
and followed by dragline removal and scaling. - Increase the rate of reduction in the East wall
The full case study is available from the ACG operational water table
(Li 2020). Figure 9(a) shows a plan of a typical - Can be used to reduce the number of
blast pattern. Figure 9(b) and (c) show the dewatering wells required
on-ground result and a plan of the final series of - The cost of such a technique is relatively low.
results respectively (After Li, 2020). Figure 9(d)
provides one example of avalanche control blast In some cases, it is possible to use controlled
(from the web). blasting to increase the permeability of tight
ground and open up drainage pathways. An
3.5 Permeability enhancement blasting (Ip) example is shown in Figure 10, where a
low-permeability clay gouge zone was impairing
This ground improvement technique has wide drainage of rocks on the footwall side of the
applicable for engineering and environment structure which, in turn, was causing elevated pore
projects and involves hydraulic modification. Time pressures in the pit slope above the bench. Three
only allows for four applications to be introduced, lines of blast holes to a depth of 45 m were
see Table 4. installed across the 'tight' zone. The lines of holes

Figure 10. Opening up drainage pathways by blasting (Beale et al. 2013).

- 374 -
 Figure 11. Three practical restoration approaches (a) Bench crest removal, (b) Reduction in slope angle,
and (c) Selective landscaping. (Modified from Coppin & Bradshaw 1982.).

were about 20 m apart and the aim was to create quarried rock slopes to construct a landform
three drainage pathways from the saturated rocks sequence similar to that in the surrounding natural
in the footwall into the dewatered rocks on the landscape. The natural reclamation process of
hanging wall side of the structure. The shot was abandoned quarries can be accelerated through a
progressed from the hanging wall to the footwall process called landform replication. Through
with long delays. The resulting drainage caused carefully designed restoration blasts, talus slopes,
pore pressures in the pit wall above the bench to buttresses, and headwalls of rock quarries and
reduce. surface excavations can be created that can be
revegetated to produce landform and plant
3.6 Restoration blasting (Ir) assemblages similar to those that occur on natural
valley sides. Figure 11 illustrates three practical
‘Landform replication’ can involve the use of restoration approaches, all of which are relatively
controlled ‘restoration blasting’ techniques on straight forward shot firing applications.

Figure 12. Diagram of reinforcing soft ground by blasting. (a) Explosive charge A is fired, an expansion
space is formed, and a compressive wave is caused; (b) stones and silt in the crater are thrown upward by
the explosion; (c) the thrown stones as well as some silt fall down, the silt in the crater is squeezed out, and
the backfilling due to blasting at A is completed. (After Zhang 2016).

- 375 -
 Table 5. Blasts design requirement for Blast A.

Blast type BDR Classification Primary technical blasting objectives

Void S_SG_HF_IvMFO The design for this type of blast is very site specific, and is
treatment influenced by: the unique shape of each 3D void; surrounding
blast rock properties; and concurrent and subsequent mining
processes. The muckpile target is to fill the void, and the
fragmentation target is to
avoid or reduce the need for subsequent secondary blasting.

Table 6. Technical objectives and design approaches for Void Treatment Blasts.

Technical blasting General design approaches

objectives
Iv Voids treatment - Employ strategically located and extended length blastholes.
- The design is based on probe drilling results or similar site investigation.
- Use appropriate drillhole orientation, length, spacing, charge length, and
timing.
M Muckpile control - Use timing to provide the required swell and to collapse and fill the
targeted void volume.
F Fragmentation - Use spacing, timing and charge length to achieve acceptable
control fragmentation.

Table 7. An example of a BDR classification and blast design requirements statement.

BLAST DESIGN REQUIREMENTS STATEMENT

PIT CUTBACK PIT
BLAST DOMAIN Felsic volcanics and sediments.
BLAST NUMBER Blast A
BLAST TYPE Void Treatment Blast
SPECIAL CONDITIONS Probe holes are sometimes loaded with explosives and used as
blastholes if they terminate above the void and will reduce need
for additional drilling.
SITE CONSTRAINTS No additional restriction or constraints.
BLAST DESIGN S_SG_HF_IvMFO
REQUIREMENTS
CLASSIFICATION
S – surface blast.
SG – single grade material.
HF- horizontal free face perpendicular to drillholes.
Iv – Void treatment is highest priority primary blasting objective
M – Muckpile control is the next highest priority primary blasting objective
F - Fragmentation control is the lowest priority primary blasting objective but needs to be
acceptable. O - Objectives, or in full, technical primary blasting objectives.

The blasting sequence is described in Figure 12.

3.7 Reinforcement blasting (If)
The advantages of this technique are: low cost,
This novel technique involves squeezing the silt stability of dams is good since blasting reinforces
out by blast throw and the space which was the backfilled part of dams, and construction speed
originally occupied by the silt is immediately filled is high since stone backfilling is carried out from
by falling stones during blasting (Zhang 2016). the land. The disadvantages of the technique are

- 376 -
 Figure 13. Void treatment blasting (Little & Blair 2020) a) Void treatment design; b) during blast image.

the vibrations caused by blasting and the irregular the work activity. An example for Blast A is given
sections of dams that need to be repaired. The in Table 7.
mechanism and operation parameters of this Figure 13 shows the blasthole layout and delay
technique still need to be studied. sequence and a during blast image for the void
treatment Blast A.
3.8 Void treatment blasting (Iv)
4.2 Modified production blast (Blast B)
The technique of void treatment blasting is the
subject of Section 4. This is a complex production blast which requires
the designer to cater for underground openings
4 GROUND IMPROVEMENT BLAST (voids), ore and waste control and wall control
APPLICATIONS blasting. It is termed a modified production blast
because it has a trim component consisting of a
4.1 Void treatment blasting (Iv) batter and buffer row and larger diameter
production fired in the same blast, see Figure 15.
Sometimes surface mining take place in areas of A presplit row was also fired a day earlier but is
historic underground mining or natural cavities. If not critical to the current discussion. Table 8 The
underground workings are encountered in or use of electronic detonators and multiple point
below the open pit they are sometimes treated initiation enables such a complex blast to be
(collapsed and filled) using specially designed contemplated and designed.
blasts, called “Void Treatment Blasts”. The mine The BDR approach suggested in this paper and
can choose to blast separately isolated voids or to previously, is that a Blast Design Requirements
treat voids as they occur within a production Statement isissued to the blast designer prior to the
or waste blast. Table 5 illustrate the BDR work activity. An example for Blast B is given in
classification for void treatment blast and Table 6 Table 9.
give the general design approaches for the same. The complexity of the blast is well illustrated
The BDR approach suggested in this paper and by the SW corner details (Figure 15) in Blast B.
previously is that a Blast Design Requirements The void treatment rows are shown with some the
Statement is issued to the blast designer prior to 165 mm diameter holes reaching 31.7 m in depth.

- 377 -
 Table 8. Technical objectives and design approaches for this complex Modified Production Blast.

Technical blasting General blast design approaches Approach used for Blast B
objectives
Iv Ground - Strategically located and extended - Employ strategically located and
Improvement (voids) length blastholes (based on probe extended length blastholes.
drilling results or similar). - The design is based on probe
- Use appropriate drillhole orientation drilling results or similar.
and length, spacing, charge length, - Use appropriate drillhole
and timing. orientation and length, spacing,
charge length, and timing.
G Grade/Ore control - Selective blasting (ore only) and - Bulk blasting and selective mining
non- selective digging - Fire along strike NS.
- Bulk blasting and selective digging - Visual ore spotting
- Blast movement monitoring
D Damage control - Employ appropriately designed - Use separate presplit blasts.
presplit, batter row and buffer row - Batter and buffer rows as part of
prior to large diameter waste modified production blast.
blastholes.
F Fragmentation - Use appropriate blasthole diameter, As per general design approach
control burden, spacing, subdrill, charge
length, and explosive type.
M Muckpile - Use timing to provide the required Use 100 ms delay on the control row
throw and heave, and throw to suit
the excavation equipment.

Table 9. An example of a blast Design Requirements Statement using the BDR classification.

BLAST DESIGN REQUIREMENTS STATEMENT

PIT Cutback Pit
BLAST DOMAIN Felsic volcanics and sediments.
BLAST NUMBER Blast B
BLAST TYPE Modified Production Blast
SPECIAL CONDITIONS Previous underground mining requiring void management
SITE CONSTRAINTS No additional restriction or constraints.
BDR CLASSIFICATION S_MG_VF_IvGDFMO
S – surface blast.
MG – ore and waste in the production blast area of the blast. VF- vertical or sub-vertical free
face parallel to drillholes.
Iv - Void treatment is highest priority primary blasting objective (site perspective)
G – Grade/ore control is the second highest primary blasting objective in this case bulk blasting
of ore and waste will be undertaken and selective ore digging is required (ore spotting is
undertaken).
D - Damage control is first priority blast objective in the wall region (trim blast design required).
F - Fragmentation control is the second lowest priority primary blasting objective but needs to
be acceptable. M – Muckpile control is lowest priority primary blasting objective but needs to be
acceptable
O - objectives, or in full, technical primary blasting objectives.

Note that probe drilling holes are used to delineate and the two-part blast pattern (void in between) are
the extent of the underground openings. mapped on Figure 16(a). Figures 16 (b) and (c)
The two monitoring locations used for Blast B present the charge mass per hole and the delay

- 378 -
 Figure 14. Blast B area showing drill hole pattern and visible vertical opening (Little & Blair 2020).

Figure 15. SW corner detail of Blast B, showing void management zones & void treatment design.

time contours respectively. be classified by its wave arrival time, t, at the

Figure 16(d) deserves a more extensive monitor as well as its distance, d, to that monitor.
explanation. Each blasthole in any blast pattern can The arrival time will depend upon the blasthole

- 379 -
 Figure 16. Pre and Post details for Blast B (After Little T N and Blair D P, 2020).

initiation time as well as the wave travel time blasthole in an extended pattern. This association
through the ground from blasthole to monitor. This of a value of V for each blasthole is termed
information can then be used to associate each vibration impact mapping. It should be noted that
blasthole with a specific time on the vector this approach can also be used for pre-design
vibration waveform (for either acceleration or analysis to eliminate expected vibration peaks.
velocity). The value of the vector vibration, V, at Figure 16(e) is the Vector PPV record by the
this specific time can then be associated with the in-pit monitor approximately 67 m from closest
particular blasthole. The main advantage is that the blasthole (flyrock did reach the monitoring site).
method uses only measured vibration data as well Finally, a post-blast photo, see Figure 16(f), has
as blast design information and can give an been included to demonstrate the collapsed void
estimation of the vibration impact, V, from each areas and the general fragmentation distribution.

- 380 -
5 DISCUSSION AND CONCLUSIONS and innovative blast design strategies that
benefit society, in the areas of safety,
Based on this work five important observations engineering, economics, aesthetics, and
and conclusions are made: environment.

- The blast management framework outlined 6 ACKNOWLEDGEMENTS

consists of a blast design management
process and blast implementation Terry Coleman, Tanika Little and Dr Dane Blair
management process. The first step in the are thanked for proof reading this paper. Blasting
blast design management process is to clearly Geomechanics Pty Ltd are also acknowledged for
determine the blast design requirements, and allowing this paper to be produced, published and
the use of the BDR classification scheme presented. The European Federation of Explosives
makes this process easier. This step also Engineers also acknowledged for accepting the
reduces the requirements risk, provides abstract, reviewing the paper for publication and
design priorities and makes them more for hosting such worthwhile events.
explicit. Details have also been given
regarding the latest addition to the BDR REFERENCES
classification scheme for ground Coppin, N.J. & Bradshaw, A.D. 1982. Quarry
improvement blasting. reclamation. Mining Journal Books Ltd.
- The range of ground improvement
applications using explosives, is large and Carr, C., Rankin, D. & Fuykschot, J. 1999.
new applications are expected in the future. Development of advanced blasting practices at
This paper has demonstrated that blasting can Forrestania Nickel Mines. Proceedings of EXPLO
be used to: densify soils; destress rocks; '99, AusIMM, pp. 239-246.
strengthen weak zones; restore landscapes;
enhance drainage pathways; eliminate gravity Beale, G., De Souza, J., Smith, R. & St Louis, R.
hazards; reinforce weak earthen materials 2013. Guidelines for evaluating water in pit slope
underwater; and finally, eliminate hazards due stability. Ch 5 Implementation of slope
to cavities. All these techniques involve depressurisation systems. Beale and Read
improving one or more of the following: (Editors). CSIRO Publishing and CRC
Press/Balkema.
physical, structural, mechanical or hydraulic
properties of the ground. Kelso, I.J. 2011. Highwall slope stabilisation by the
- The current work has shown that ground softwall method, Proc. Slope Stability 2011:
improvement blasting using explosives can be Int. Symp. On Rock Slope Stability in Open Pit
adapted to many different situations and Mining and Civil Eng, Vancouver, Canada,
applications. Furthermore, the use of blasting September.
methods to improve ground conditions
requires considerable expertise and should be Latilla, J. & Tumur-Ochir, B. 2016. Buffer blasting to
supported by all the components of the Blast control movement along bedding plane shears.
Management Framework. 2016 Coal Operators’ Conference. The
- Based on this paper, blasting operations have University of Wollongong.
been shown to a be unique in four (rather than
three) important ways. Specially, (i) they are Li, J. 2020. Elimination of structure controlled
potentially high consequence activities (high highwall failures at an open cut coal mine. Slope
hazard but low residual risk); (ii) they have Stability 2020 Dight P M (ed.) Australian Centre
for Geomechanics, Perth, ISBN 978-0-9876389-
rapid design and implementation feedback
7-7.
cycles (need cautious start and then
accelerated learning); (iii) they occur early in Little, T.N. & Blair, D.P. 2011. The influence of
the value chain of extraction operations (can geology on blasting hazards and results. In 8th
be used to leverage downstream processes) International Mining Geology Conference
and (iv) are very versatile. Queenstown NZ, 22-24 August 2011.
- The requirements approach presented are (Australasian Institute of Mining and Metallurgy:
expected to improve the quality of blast Melbourne).
designs and hence blast outcomes leading to
practitioner and stakeholder satisfaction. It is Little T.N. 2011. Geological controls on drilling and
also expected to assist with lateral thinking blasting operations. In EXPLO 2011 Conference

- 381 -
 8 - 9 November 2011 Melbourne, Australia, Zhang, Z.X. 2016. Rock fracture and blasting - theory
(Australasian Institute of Mining and Metallurgy: and applications. Butterworth-Heinemann is an
Melbourne). imprint of Elsevier.

Little T.N. 2015. Classification and development in

grade control blasting for surface mines. In
proceeding of 11th International Symposium on
Rock Fragmentation by Blasting. 24-26 August
2015, Sydney, Australia.

Little, T.N. 2017. Blast design requirements

classification system for blast purposes. In
proceeding of EFEE 9th World Conference on
Explosive and Blasting, Stockholm, Sweden.

Little, T.N. & Lovitt, M.B. 2018. Role of blasting in

assisting with value-based ore control in
underground mines. In proceeding of 12th Int.
Symp. on Rock Fragmentation by Blasting. Luleå
Sweden 11-13 June 2018.

Little, T.N. 2019. Wall control blasting – Design and

Analysis. In proceeding of EFEE 10th World
Conference on Explosive and Blasting, Helsinki,
Finland 16-17 Sept 2019.

Little ,T.N. & Blair, D.P. 2020. Blasting review,

monitoring and analysis report. Confidential
report to client. Mitri, H.S., 2001. Practitioner’s
guide to destress blasting in hard rock mines.
McGill University.

Moore, H. 1986. The construction of a shot-in-place

rock buttress for landslide stabilisation. 37th
Annual Highway Geology Symposium, Helena,
Montana, August, pp 137-157.

Roux, A., Leeman, E. & Denkhaus, H. 1957.

De-Stressing: a means of ameliorating rockburst
conditions. Part 1 - The Concept of de-stressing
and the results obtained from its application.
South African Institute of Mining and Metallurgy,
pp. 101-119.

Shakeran, M., Eslami, A. & Ahmadpour, M. 2016.

Geotechnical aspects of explosive compaction.
Hindawi Publishing Corporation Shock and
Vibration Volume 2016, Article ID 6719271, 14
pages.

Yao, M., Forsythe A, and Chinnasane, D.R. 2016.

De-stress blasting strategy for mining in highly
stressed sill pillars at Vale’s Sudbury mines –
Two Case Studies. Workplace Safety North’s
Mining Health and Safety Conference, Sudbury,
Canada – April 5-7, 2016

- 382 -
- 383 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Implementation of Technology to Fully Digitalize the Drill & Blast

Process in a Remote Setting

A. Bihlar, L. Bermingham & R. Farnfield

EPC Sverige, Eskilstuna, Sweden., EPC United Kingdom, Alfreton, Derbyshire, United Kingdom

ABSTRACT: The development, innovation and implementation of new drone and digital technologies
now provides the opportunity to fully digitise the drill and blast process. These new technologies allow
for Blasting Engineers to be a part of the entire process remotely whilst still maintaining involvement at
every interaction point; from the pre-blast preparation through to the post-blast analysis and to be able to
share key data seamlessly with all key stakeholders in the process.
Tools such as drones, in-field drilling and explosive loading data capture, automatic borehole probes
and a bespoke cloud-based data storage applications provide the opportunity to: improve the safety of
surveying practices and accuracy of borehole positioning, optimise the explosive energy distribution
whilst also providing a platform to capture and analyse key performance indicators, allowing for a
process of review and continuous improvement.
Our studies show implementing the above technologies and innovations results in a significant impact
on blast and economic related KPI`s.

1 INTRODUCTION Each of the steps are detailed in this paper and a

case study is presented which shows the benefits
The development, innovation and implementation that employing digital technology and analysis can
of new drone and digital technologies provides the have on a large-scale quarrying operation.
opportunity to fully digitise the drill and blast
process. The development of key digital tools such 2 USE OF DRONE TECHNOLOGY
as Expertir blast design software, the Explore
cloud-based storage database and Expertab, the Developments in drone technology in recent years
data logging application allow for Blasting has led to its adoption across a range of industries
Engineers to be a part of the entire process around the world to perform numerous functions.
remotely and involved at every interaction point; For the mineral extraction industries, drones have
from the pre-blast preparation, through to the post become a crucial tool in performing highly
blast analysis, whilst sharing everything accurate surveys for determining precise
seemlessly with all key stakeholders in the blast hole locations whilst offering improved
process. safe working practices, in contrast to traditional
Figure 1 shows the operation workflow of the laser scanner techniques due to their ability
entire process where each of the processes feed to be operated remotely from face extraction
through the cloud based database. operations.

- 384 -
 Figure 1. Operational workflow.

Implementation of photogrammetry techniques activities within the quarrying/mining operation

provides the ability to utilise drones in capturing and perform the survey from one location,
overlapping photographs of a blast site and tie providing line of sight to the drone can be
these photographs together whilst using maintained. In comparison to using a laser
the in-built GPS module, creating a highly scanner, the surveyor would typically need to be
accurate georeferenced 3D model (Adderley et al. positioned in front of the face, usually on the
2017). quarry floor, where there is an increased likelihood
The process of drone surveying with in-built of interaction with other working activities, which
Real Time Kinematic (RTK) module enables for a could possibly cause delays to face excavation
simpler surveying practice. With the RTK module, work, as well as working in closer proximity to
the drone receives real time corrections to its mobile plant. It may also be necessary to scan
onboard GPS from a local Global Navigation faces form multiple positions, thereby further
Satellite System (GNSS) station, enabling each of increasing the risk to the surveyor and time taken
the photographs to be geo-referenced, removing to conduct the survey.
the need to place ground control points (GCP) and There is a significant improvement in accuracy
perform an additional RTK survey to within the survey also. Typical errors within the
geo-reference the resulting 3D model. This survey range from 1-3cm.
reduces the necessary training requirement for Geological conditions which are visible in the
staff to perform a survey and then upload the face can now be captured within the photographs
resulting photographs to an online cloud-based and overlaid onto the resulting 3D model. This
platform. Once uploaded, the Blast Engineer is allows for the blast designer to identify and
then able to access the photographs and create the examine in detail areas of weakness in the face
3D model of the surveyed area and import into the which may extend further into the rock mass
blast design software. within the blast area, therefore allowing detailed,
The use of drones offers a number of data led control measures to be employed within
advantages when used for surveying and face the blast design to mitigate any potential safety
profiling compared to laser scanner methods. issues.
Generally, the survey can be conducted in a more Drones are capable of obtaining measurements
controlled environment. The surveyor can be on the face which would normally be obscured to a
positioned on the bench, isolated from other laser scanner by features such as rock traps. This

- 385 -
 Figure 2. RTK-drone in action.

ensures that holes are positioned accurately and them into dedicated photogrammetry software.
safely at the toe of the face. The software aligns the position of the photos
In addition to this, a typical drone survey can based on the geo-tag assigned to each photo,
be completed in 10 to 15 minutes making it a more which has been acquired by the RTK module
efficient process over laser profiling methods. within the drone. The images are tied together
based on the software identifying identical single
3 CREATION OF THE 3D SURVEY MODEL points within multiple images. These points are
known as tie points and link the images together.
The Blast Engineer can download the survey Figure 3 shows a screenshot of the software
photographs from an online database and import creating a 3D model with the level of

1.9cm accuracy of model

Figure 3. Complete photogrammetry model done with RTK-drone.

- 386 -
 Figure 4. 3D model imported into blasting software – Expertir.

accuracy within the survey identified. design software, which assigns holes within the
The resulting, georeferenced model can then be blast dependent on the specified burdens and
uploaded into blast design software from which spacings and designed angles, whilst also taking
the face profile scan can be processed as shown in into consideration the angle of the face(s). The
figure 4. position of the holes can be the adapted to
maximise the efficient distribution of explosive
4 ADAPTIVE HOLE LAYOUT energy, based on the designed hole loading. This
process offers precise placement of holes and
The location of blast holes can be determined explosive charges and removes the need for a
using an automatic layout function within the blast Blast Engineer to visit site and physically measure

Figure 5. Layout of drill holes created from automatic parameters.

- 387 -
 Figure 6. Drill rig equipped with HNS in Sweden.

and mark out hole locations on the bench - a drill rig, which allows the GPS controlled rig to
process in which will often result in less efficient automatically position at the designed hole
positioning of holes. Blast Engineers can perform locations and drill in adherence to the designed
the blast design process at a central location and parameters. This eliminates the need to pre-mark
upload the designed drill parameters onto a the drill holes, allowing for the driller to focus on
cloud-based storage system via the integrated blast quality whilst drilling at the precise location
design software, Expertir. determined by the Blast Engineer.
Whilst drilling, it is now possible for the entire
5 DIGITALISED DRILLING PROCESS
drilling process to be measured and recorded
Once the preliminary design has been completed, digitally, by use of an online database such as
the Blast Engineer is able to upload the blast Explore. This gives the driller the ability to
geometry to an online platform, from where the download the drill plan from the Explore platform
drilling data can be downloaded by the driller and directly to his/her tablet (or other electronic
uploaded to the drill rig equipped with Hole device) via the Expertab application. Upon
Navigation System (HNS). This enables the driller opening the file, the driller is presented with a hole
to upload the hole co-ordinates directly into the plan of the blast and by selecting an individual

- 388 -
 Figure 7. Screenshot of the drilling function of Expertab.

Figure 8. Logging of borehole geology within Expertab.

hole, the relevant drilling parameters are download and import the drilling data directly into
displayed; hole depth, inclination and azimuth, as the blast design software and therefore fully
shown in the figure 7. digitalizing the drilling process and reducing the
Whilst each hole is being drilled, the driller can resources required on site.
record the ground condition encountered within
the application and adjust the hole depth if 6 AUTOPROBE
required as shown in figure 8. Once completed for
the entire set of holes, the recorded drill data is The Autoprobe tool has been developed for drillers
then able to be uploaded on to the online platform to probe each hole after it has been drilled. Once
via the application. set up, it is capable to automatically deploy down
Whereas with common working practices, the the hole to measure the depth, and then
Blast Engineer would have to be on site to automatically retract to specified intervals to
generate the drill log and return to site to collect perform its measurements which are then captured
the recorded data, it is now possible to eliminate in a digital format. By pre-loading the drill plan
this requirement. The Blast Engineer can into the accompanying application, the driller can

- 389 -
 Figure 9. Autoprobe in action – probing the holes automatically.

measure the hole depth, angle and azimuth and Figure 10 shows the Autroprobe application
analyse the drilled measurements against the measuring slight misalignment in the azimuth
designed parameters. This provides the ability for from designed (left hand side) and slight
the driller to measure and identify any deviations deviation towards the bottom of the hole (right
directly after they have been drilled, giving the hand side).
opportunity to re-drill, if necessary, before the drill In addition to improved resource efficiency,
rig moves to a new location. Identifying the need digital drilling data allows for all information to be
to re-drill a hole will ensure that holes are drilled utilised in the blast design process, allowing for
in positions to maximise the energy distribution any geological anomalies to be visualised in a 3D
within a blast. This in effect acts as a quality format in combination with other digital survey
assurance process for the Blasting Engineer. The data, to support optimised design of the explosive
recorded data can then be sent to the Blast loading of each blast hole. This reduces the
Engineer to incorporate in the final blast design, likelihood of misinterpreting any safety
therefore removing the need for an Engineer to information provided by the driller. An example of
attend site. this is shown in figure 11.

- 390 -
Figure 10. View in Autoprobe application.

Figure 11. Integration of drilling data within the design process.

- 391 -
 Figure 12. Hole loading design in Expertab.

This example shows that importing the digital platform. This provides both a means to remove
drilling information into the blast design software the requirement of paperwork on a blast and
provides precise location of geological anomalies enables the entire drill and blast process to be
within each hole. The blast designer can then measured and recorded digitally. The process also
modify the explosive loading within the holes in provides the ability for an engineer to design a
which these anomalies are present to ensure the blast and send the relevant information required to
correct control measures are in place to prevent charge and initiate that blast reducing the travel
any dangerous occurrences, such as flyrock or requirements between sites and time spent
misfire. working within an active quarrying/open pit
mining environment. This therefore streamlines
7 DIGITISED HOLE LOADING PROCESS the blasting process and reduces the hazards which
are present with working within an active
Once completed, the hole loading design can then operation.
be uploaded onto the Explore platform. On the day
of the blast, the design can be downloaded onto 8 CLOUD BASED STORAGE DATABASE
Expertab so that the MEMU operator has all the
necessary data for every hole within the blast on a Availability of a central cloud-based storage
mobile device. The information includes a plan database is essential for streamlining the drill and
layout of the blast with each hole identifiable via blast process as outlined above and enables Blast
designated identifiers from the blast design Engineers to perform a variety of design work for
process. Upon selecting a hole, the MEMU sites from a central location. This in effect acts as
operator is presented with the explosive loading an information highway between blast designers,
configuration of that hole, including planned drillers, shotfirers and MEMU operators.
charge(s), accessories and their locations, In addition to this, a database such as the
stemming and inter-deck stemming heights. An Explore platform becomes an effective analysis
example of this is shown in figure 12. tool. A range of blasting data and key performance
The application is connected to the onboard indicators (KPIs) can be extracted from the
controls of the MEMU via Wifi. This allows for uploaded blast designs for analysis.
the entire hole loading process to be controlled Accurately measuring and recording specific
directly from the mobile device. As a result, all operational site performance KPIs provides the
changes made to the designed loading during the ability to analyse the performance of the blast
charging of a blast are recorded. All recorded data relative to the operation’s requirements. This
can then be uploaded back to the online database performance data can then be cross correlated with

- 392 -
 Figure 13. Online Database system – Explore.

nominal and actual blast parameters. A central reduce the amount of fine material by improving
database provides the perfect platform to perform the energy distribution within each blast.
this task. Through this process, the effects of However, this cannot only be monitored in
specific parameters can be established and isolation as a decrease in fines may not be entirely
therefore blasts can be optimised to meet the indicative of improved blast performance, as the
operation’s requirements, and drive a process of balance of energy may shift to the opposite side
continuous improvement for the entire drill and and generate more oversize. Therefore, it is
blast process, which is bespoke for every important to also to monitor the power
operation. consumption of the primary crusher. Further to
this a study of oversize produced from the blast
9 CASE STUDY was conducted using fragmentation analysis to
confirm no increase. Consistent improvement in
A project termed ‘The Digital Quarry’ project was overall targeted fragmentation will result in lower
undertaken at a large-scale limestone quarry in the power consumption of the primary crushing
UK in 2018. The objective of the project was to process, increasing the energy efficiency and
implement new technologies and techniques and to reducing production of low value or waste
capture their impact to the entire quarrying materials whilst controlling secondary breaking
operation. costs.
The project comprised of 2 phases. Phase 1 was In phase 2, new technologies and techniques were
the baseline phase where no changes were made to introduced into the drill and blast process. These
the drill and blast process and site KPIs indicative included:
of blast performance were measured and recorded.
The aim of this phase was to provide a baseline
1. Full implementation of drone technology
data set from which the impact of future changes
in performing face profile surveys
made to the drill and blast process could be
determined and quantified. This period extended 2. Introduction of an HNS drill rig
throughout 2018. 3. Adaptive hole layout
Two KPIs chosen to be monitored were the 4. Geometric delay times
percentage of fines material which was scalped off
at the primary crusher, and power consumption of This phase spanned from 2019 to 2020. Site
the primary crusher. The theory is that by KPIs were monitored throughout both phases and
monitoring the volume of fine material, it will be the impact that phase 2 has had on the operation
possible to determine whether any changes will can be seen in figure 14.

- 393 -
 Figure 14. Processing plant data showing reduction in fines and power consumption.

Figure 15. Data showing increase in percentage primary split and throughput.

The data shows that from 2019, the percentage crusher increase, that overall tonnage tipped at the
of blasted material which was scalped off at the crusher also increased.
primary crusher, reduced from an average of 26% The table summarizes the results from the
in 2018 to 20.56% in 2019 and 14.84% in 2020. project.
That is an overall reduction of 11.17% over the
project. Table 1. Impact of the Digital Quarry Project.
The average power consumption of the primary
crusher was also reduced during the project. An
average power consumption was calculated based
on the volume of material crushed and was
measured as kWhr/T. During 2018, modification
work was carried out on the processing plant
and therefore power consumption data after
this point will only be considered in order
to truly reflect the impact of the project. Overall, the project has resulted in consistently
From 2019 to 2020, the average power improved blast fragmentation which has led to a
consumption had decreased by 0.13kWhr/T, tightening of the overall blast particle distribution
(a 30% reduction). curve. This has led to a higher primary crusher
Figure 15 shows the primary crusher throughput and a lower percentage of fine
throughput and the percentage split of material material. This has demonstrated the impact of
being fed through the crusher and material scalped implementing some of the new technologies and
off. techniques to the drill and blast process, and
The results show that not only did the highlights the importance of monitoring site KPIs
percentage of material fed through the primary which reflect blast performance, in order to fully

- 394 -
quantify the true cost benefit of improved blasting
technology and techniques.

10 CONCLUSION

Implementation of the technology and tools

detailed in this paper shows that it is now possible
for a Blasting Engineer to work remotely and still
maintain involvement throughout the entire blast
design process, irrespective of location and
distance.
Experience with this “concept” through the past
year has demonstrated that engineers can now
focus on the blast design and are capable of
providing support to a greater number of sites, by
significantly reducing the time travelling to
operations situated in remote locations. A
cloud-based database which acts as a central point
for data transfer is vital for this process.
The Digital Quarry case study has clearly
demonstrated that by harnessing the latest
technology, a significant positive impact can be
made to a quarrying/mining operation.
Overall, it can be concluded that the
implementation of these digital technologies is
practical, and achievable, for all types of location
no matter the size. It is the operational throughput
that eventually defines the size of the benefit gains
which can then be sustained and improved upon to
the advantage of all stakeholders.

REFERENCES

Adderley G, Bolitho S, Farnfield R, Keverne B &

Wetherlt A. 2017. A Comparison of Blast Face
Surveying Techniques. Proceedings of the 9th
EFEE World Conference on Explosives and
Blasting, Stockholm, Sweden 10 – 12 Sep 2017.

- 395 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Multiple blasthole dynamic pressure (MBDP) model to alleviate

dynamic pressure in blastholes

R. Yang
Orica USA Inc. Watkins, Colorado, USA

ABSTRACT: With mine-to-mill projects, many mines want to use blasts with high powder factors to
generate fine rock fragmentation or increased fines from blasts for increased mill throughputs. However,
reduced blasthole spacing with high powder factors or using short delays between blastholes can often
generate excessive dynamic pressure among blastholes. Excessive dynamic pressure a later firing charge
receives from earlier firing charges can cause the charge desensitisation and detonate with low velocity
of detonation (VoD) and generate excessive post blast fumes. Such dynamic pressure may also damage
the initiation systems causing misfires that are dangerous incidents. When the dynamic pressure exceeds
the threshold of the initiation system, sympathetic detonation can occur, which is a significant safety
hazard and disrupts the mine productivity. Recently, a Multiple Blasthole Dynamic Pressure (MBDP)
model was developed. The MBDP model simulates the dynamic pressure generated from earlier firing
explosive charges received by a later firing charge before it is initiated. The model simulates all blast
design parameters. It can be used to explore various design scenarios for alleviating the dynamic
pressure within a blast pattern. The MBDP model can be applied along with other blast prediction
models to improve blast designs for desirable blast results. The model is suitable for open pit, tunnelling
and underground blasting.

1 INTRODUCTION explosive charge can increase. If it passes the

optimum density (e.g. 1.30 g/cc), the VoD can
Using high powder factors or fast timing may not drop and the detonation can produce fumes and
get the expected benefits from the blast. Reduced result in lower energy release.
blasthole spacing for high powder factors or fast Figure 2 shows bubble energy reduction due to
timing can often generate excessive dynamic pre-shock by dynamic pressure. The results were
pressure among blastholes. It is important to measured from underwater test blasts (Yang &
alleviate the dynamic pressure when designing a Pelletier 1998). An underwater blast can be
blast. Consequently, a dynamic pressure prediction measured for shock and bubble energies, which
model is required to examine the dynamic pressure add up as total underwater blast energy. The shock
and alter the design if necessary to alleviate the energy is directly related to VoD. The higher the
dynamic pressure within a blast pattern before the VoD the higher the shock energy. The bubble
blast is implemented in the field. energy is related to the gas volume and its
All void-sensitised explosives have a general temperature generated from the underwater
density - (VoD) relationship, as illustrated in explosion. The bigger the volume and the higher
Figure 1. In the figure, some typical densities, and the temperature the larger the bubble energy.
a failure density of 1.36 g/cc are used as examples. Pre-shock dynamic pressure upon an explosive
Under dynamic or static pressure, the density of an charge can cause lower VoD and lower bubble

- 396 -
 VoD 1.25g/cc

Smaller Dead
Safety pressured
0.85g/cc margin

Low-order
Larger detonation
safety
margin
No fume

Density

1.3 g/cc 1.36g/cc

Figure 1. General density-VOD relationship for void-sensitised explosives.

Figure 2. Explosive bubble energy reduction due to pre-shock by dynamic pressure.

- 397 -
energy from the explosion of the charge. Figure 2 contribution from multiple charges of a blast
shows that at the dynamic pressure of 1500 psi the according to their location and delay timing. PPV
50 mm diameter cartridge explosive starts to calculation in the MBDP model is largely the same
reduce energy release. For large diameter bulk as that in the Multiple Blasthole Fragmentation
emulsion, similarly it will reduce energy release (MBF) model (Yang 2018), except the MBDP
under dynamic pressure, although the starting model only account for the PPV contribution
pressure may be higher than 1500 psi (such as before a receiving charge is initiated. The PPV
4000 psi). calculation in the MBDP model enables major
Furthermore, if the dynamic pressure a charge contributions of dynamic pressure to be simulated.
receives before its initiation is high enough, it can The model is built based on quantities that can be
cause the detonator in the charge to fail to initiate. easily measured in the field and on
If the dynamic pressure is even higher, it can lead well-established assumptions. Each donor charge
blastholes to sympathetically detonate, which can is computationally divided into spherical charge
be a serious safety issue to mine property or elements (the length equals to the diameter) that
personal safety. has an initiation time, as shown in Figure 3.

2 MBDP MODEL 2.1.1 Dominant charge element for a calculation

The model assumes that rock at the vicinity of point
blastholes can be approximated as shock state, at
A dominant charge element for a calculation point
which the dynamic pressure and particle velocity
is defined as the charge element having the
are governed by the rock Hugoniot relationship.
smallest scaled distance among all charge
Therefore, if the peak particle velocity (PPV) of
elements in the blast that are initiated prior to the
the rock is known at a point in the rock, the
initiation of the charge that contains the
dynamic pressure can be estimated. The dynamic
calculation point according to the design. The time
pressure at a point in an explosive charge received
of the shock wave arrival to the calculation point
before its initiation is different from but closely
from the dominant charge element is defined as
related to the pressure received from the same
the dynamic pressure time (tf) at the point.
donor charges at the same point if the rock would
Contribution from a charge element is determined
occupy the same space.
according to its scaled distance to the point and
2.1 PPV calculation at a point the difference between its shock wave arrival
time (tx) and the dynamic pressure time (tf)
PPV calculation at a point in rock is to reflect the at the point.

Receiver charge Donor charge

A charge element x
Dynamic pressure arrival time: d
tx=d/c + ti+L/VoD
A primer in a neighboring charge,
to - designated initiation time,
c - ground sonic velocity.

Pressure contribution if tx < to

One receiver charge can receive pressure from
ti – primer initiation time
multiple donor charges

Figure 3. Each charge is considered an as assembly of spherical elements with an associated initiation time.

- 398 -
 Figure 4. Complete PPV charge weight scaling law.

2.1.2 Complete Scaling Law to model VoD, distance sdc, where it is assumed that the blast
nonlinearity, and scale up vibration can be safely recorded as the nearest
distance. The blast vibration PPV at a scaled
A near-field signature hole PPV versus scaled distance greater than sdc are measured in the field
distance is established from the site from the signature hole blast vibration monitoring.
characterisation using the near-field signature hole The PPV at a scaled distance smaller than sdc is
vibration monitoring for each site (Yang 2018). interpolated by a power law joining the point
For each charge element, a complete (dual) charge (sdb,ub) and (sdc,PPVc). are the scaled
weight scaled distance law is established in the distance and the peak particle velocity (PPV) at
model - two scaling laws joined at the scaled the borehole wall. Figure 4 display a complete

ρ 0VOD 2
pb =
2r 8
n
pb vbn = pc v c
pb
2n
r 
ub pb = pc  c 
 rb 
w0

Figure 5. Sketch of pressure and particle velocity at the blasthole wall containing an explosive charge.

- 399 -
 calculation point,
ts peak pressure time tf

d d d
l l l

Figure 6. Sketch the increase of the contributing charge length with distance from a charge.

PPV charge weight scaling law for the best fit and PPV (Yang 2018) to be modelled with Monte
the 97.5% upper bound. Since the large range of Carlo simulations.
PPV from several blasthole spacings to the At the close vicinity of a borehole wall, only
borehole wall, the log scale is used for the PPV the charge elements of which the shock wave
axis. arrival time within the contributing time window
The borehole wall PPV ( and the (Figure 7) contribute to the pressure rather than the
corresponding scaled distance are obtained. whole charge.
When the distance from a charge element to a
is obtained from the borehole pressure that is
calculation point is large, the time window width
determined from the explosive property, loading associated with the element is large and the
coupling, and the rock Hugoniot parameters. The contributing charge length becomes longer. When
is calculated by assuming the contributing a calculation point is sufficiently far from a
charge segment has the length of the diameter of charge, the whole charge length could become the
the charge. The distance from the charge element contributing charge and the actual contributing
to the borehole wall is the blasthole radius, as charge weight could approach to the effective
shown in Figure 5, where are charge weight, although each charge element
borehole pressure, borehole wall particle velocity, contributing a negligible PPV and the whole
and weigh of the charge element. contribution from the charge is small due to the
The complete (dual) charge weight scaling law large distance. The PPV calculation approach
for a charge element is intended to model the presented here models the PPV contribution at the
contributions to dynamic pressure or peak particle vicinity of blastholes. The approach can simulate
velocity at the vicinity of the explosive charges multiple charge contributions with delay time
from their detonation. The properties of each considered in different blastholes. The
explosive charge such as density, VoD, blasthole contributing charge length also increases relative
diameter, and nonlinearity of rock in the complete to the VoD of the explosive charge. The higher the
PPV charge weight scaling law are modelled VoD, the longer the contributing charge length for
explicitly. The decoupled charges can also be a given distance to a charge. Figure 6 demonstrates
modelled as shown in Figure 5. Both the best-fit the increase of the contributing charge length with
(average) and the 97.5% upper bound regressions distance from a charge. It contrasts with the
for the charge weight scaling law also enables the method proposed by Persson et al. 1994 that is
geological random effects causing the variation of more suitable for far-field PPV calculations.

1
f ( dt )

tx tf
t x - 0.5Tx tx + 0.5Tx

Figure 7. Time window associated with a calculation point and a charge element.

- 400 -
 Figure 8. Duration estimates from measured pressure waveforms at different distances to donor charges.

2.1.3 Variable time window of a charge element time tf, the weighting function
- nonlinear charge weight integration equals to 1.
A variable width of a time window is used for ∆𝑡𝑡 𝑥𝑥
−𝜇𝜇 � �
each charge element to compute its contribution at 𝑓𝑓(∆𝑡𝑡𝑥𝑥 ) = 𝑒𝑒 0.5𝑇𝑇 𝑥𝑥 (1)
different distances to a calculation point. Figure 7
shows a time window associated with a charge The waveform duration (Equation 2) is used as
element at a calculation point. If the shock the width of the time window (Tx) and calculated
pressure arrival time from the dominant charge based on distance from the charge element to the
element (the dynamic pressure time tf) is within calculation point. The waveform duration of
the time window, the charge element has a dynamic pressure increases with distance due to
contribution weighted by the assumed exponential high frequency attenuation. The relationship
weighting function (1). If the shock wave arrival between the waveform duration and the distance
time at the point tx equals the dynamic pressure can be established from the measurement of the

Figure 9. Dynamic pressure duration with distance (1 m=3.2808 ft).

- 401 -
dynamic pressure waveform duration change with 2.1.4 Non-linear charge weight scaling for
distance. Figure 8 demonstrates the duration superposing contributing charges
estimates from measured dynamic pressure
waveforms at different distances from the donor To model the non-linear superposing of the
charges. A parabolic fitting of the waveform explosive detonation effects at the vicinity of
duration with distance can more reasonably explosive charges, non-linear charge weight
simulate the pressure waveform duration increase superposing is used. Figure 10 shows a calculation
with distance than the previously used linear point ‘O’, its dominant charge element, and one of
relationship (Yang 2018). Near a blasthole, the the contributing charge elements within the blast
dynamic pressure is shock wave or nonlinear and ( .
the duration increases quickly with distance. The charge weight of the element is , which
Further away from the blasthole the dynamic
is scaled to an effective charge weight ( ) by
pressure becomes elastic wave and the duration
time increase almost linearly with distance. The assuming to be at the same distance as the
data in Figure 8 was measured in a mine site with dominant charge element to the point “O”, and
relatively hard rock and the fitting equation generate the same PPV contribution as the element
(Figure 9) has the form (Equation 2): at its original location:

𝑇𝑇𝑥𝑥 = 𝑎𝑎 ∙ (𝑑𝑑𝑥𝑥 )𝑏𝑏 (2)

𝛽𝛽
𝑑𝑑
where a and b are fitting parameters. The statistical 𝑤𝑤𝑒𝑒𝑒𝑒 = 𝑤𝑤𝑥𝑥 � �
𝛾𝛾
(3)
𝑑𝑑 𝑥𝑥
fitting considers the charge element detonation
wave duration over the charge element 2r at
When a charge element is located at the same
the borehole wall. distance as the dominant charge element, it yields
If the dynamic pressure time is within the an effective charge weight ( ) close to its
time window tx-0.5Tx <tf<tx+0.5Tx, then the original charge ( ) weight, since
charge element is a contributing element to the β
d γ
PPV at the point. Where tx is the dynamic pressure when d x → d , there is :   → 1, therefore, wex → wx
arrival time from the charge element to the point,  dx 
Tx is the time window width calculated from When a charge element is located much farther
Equation (2). A nonlinear charge weight than the dominant charge element to the point ‘O’:
integration with a weighting function modelling
, , the effective
delay timing was used to account for the
accumulative effect of contributing charge charge weight becomes negligible to contribute to
elements from the blast (Yang & Scovira 2008). PPV. The non-linear charge weight superposition

a charge element within the blast

dx wx

Dominant charge
element for Point O

Figure 10. A calculation point (O), its dominant charge element, and another charge element in the blast.

- 402 -
Equation (3) accounts for the non-linearity in the The PPV is calculated as:
near field and is modelled in the complete charge
weight scaling law described in Figure 4. PPV = α ⋅ sd e
−β
In addition, the MBDP model accounts for the (6)
confinement reduction and blast wave screening
from earlier firing adjacent charges. This is where and β are determined from the charge
achieved by a screening factor ( ) modifying weight scaling law in Figure 4 described above.
the distance ( in Equation (3) from a charge It is worth noting that Equation (5) is different
element to the point in the rock. If the rock in the from the conventional charge weight scaled
path of the charge element to the point is damaged attenuation equation. The former includes not only
from earlier firing blastholes, the effective distance the site-specific dynamic pressure attenuation
(modified) is increased by the screening factor. characteristics. It also includes the blast design
The method was reported in a previous paper on timing, initiation point in a charge, VOD, distance
the Multiple Seed Waveform (MSW) blast from a contributing charge, loading distribution
vibration model (Yang & Scott 2008, Yang & Kay and explosive types in each blast hole.
2011).
2.2 Dynamic pressure at a point
2.1.5 Accumulated effect charge weight –
After the PPV is calculated, the dynamic pressure
The sum of all effective contributing charge at a point can be calculated as:
elements using the weight function
(Equation (1)) can be calculated as: 𝑝𝑝𝑏𝑏 = 𝜌𝜌𝑟𝑟 ∙ 𝑝𝑝𝑝𝑝𝑝𝑝 ∙ 𝐷𝐷 = 𝜌𝜌𝑟𝑟 ∙ 𝑝𝑝𝑝𝑝𝑝𝑝 ∙ (𝑐𝑐 + 𝑠𝑠 ∙ 𝑝𝑝𝑝𝑝𝑝𝑝)
= 𝜌𝜌𝑟𝑟 ∙ 𝑝𝑝𝑝𝑝𝑝𝑝 ∙ 𝑐𝑐 + 𝑠𝑠 ∙ 𝜌𝜌𝑟𝑟 ∙ 𝑝𝑝𝑝𝑝𝑝𝑝 2 (7)
𝑤𝑤𝑐𝑐𝑐𝑐 = ∑𝑘𝑘𝑥𝑥=1 𝑓𝑓(∆𝑡𝑡𝑥𝑥 )𝑤𝑤𝑒𝑒𝑒𝑒 (4)
3 MBDP MODEL CALIBRATION
where k is the total number of contributing charge
elements including the dominant charge element, The input to the MBDP model is the PPV
is the xth scaled contributing effective charge attenuation versus scaled distance of near field
weight (refer to Equation (3)), and is the signature hole blast vibration, ground sonic
shock wave arrival time difference of the xth velocity, and the detailed design information of
contributing charge ( ) from that ) of the blastholes. The model predicts the dynamic
pressure each charge receives before its initiation
dominant charge element.
in a blast pattern.
In Equation 4, the accumulated effective charge
If the dynamic pressure was measured at
weight wce at a calculation point is dependent on
enough blastholes, the predicted pressure may be
the charge weight per hole in the blast, the VOD,
compared with the measured to determine the
and the delay time among the blast holes. The
value of the model calibration parameters.
reduction of dynamic pressure can be achieved by
However, the measurement of the dynamic
decreasing wce through decrease of the charge
pressure can have significantly scattering, due to
weight per hole, using a lower VOD, or lower
the random nature of the geology. Therefore, a
energy explosive product, increase blasthole
sufficient number, (e.g. >30) of measurements
spacing, and altering the delay time among blast
may be required to obtain representative dynamic
holes and charges.
pressure within a blast pattern. Since the cost of
2.1.6 PPV at a calculation point the dynamic pressure measurement is generally
high, it may not be practical to make extensive
From the accumulative effective charge weight number of measurements to calibrate the model. In
the effective charge weight scaled distance such cases, a holistic view of the blast evaluation
must be applied. The model may be calibrated
can be calculated at a point in rock using the
based on observations such as intensity of post
distance (d) from the dominant charge element to
blast fumes or any misfire occurrences, etc. -
the point (Figure 10). The effective charge weight
calibrating the model for predictions within the
scaled distance is:
ballpark range and using it for comparing trends of
d different scenarios of blast design.
sd e =
β
( wce )γ The scatter of the dynamic pressure can be
(5)
excessive or occurs in a random nature mainly due

- 403 -
 Figure 11. High dynamic pressure at the buffer row from the pre-split blastholes (Pmax=760 bar).

to the random nature of geological discontinuities 4 MODELLING EXAMPLES

that could randomly connect blastholes.
Nevertheless, alleviating dynamic pressure peaks After the MBDP model is calibrated at a site it can
by increasing blasthole spacings or reducing be used to model blast design scenarios to alleviate
charge weights or changing delay timing among dynamic pressure amplitude with a blast pattern.
blastholes using dynamic pressure modelling can
reduce the probability of excessive pressure peaks 4.1 Example 1
to reduce misfire or explosive pre-compression.
Increasing blasthole spacings or reducing the A modelling exercise was conducted at an open pit
charge weight can reduce the probability of weak mine where the rock was fractured and soft. The
geological materials connecting blastholes blasts malfunctioned with undetonated detonators
together. found in the muckpiles. Excessive dynamic

New design without pre-shear blastholes

Figure 12. Dynamic pressure (Pmax=520 bar) from new design was reduced from the previous design (Pmax=760 bar).

- 404 -
 3ms 3ms

Figure 13. MBDP model predicted dynamic pressure a charge receives before its designated initiation.

Figure 14. MBDP model predicted dynamic pressure reduction with all lifters fired instantaneously.

pressure within blast patterns was suspected. The 4.2 Example 2

MBDP modelling determined that presplit
(pre-shear) blastholes donated high dynamic The MBDP model was also applied at a tunnelling
pressure at the buffer row as shown in Figure 11. site in Europe where misfires were observed with
The pressure peaks are at blastholes between the occasionally finding of damaged unfired
instantaneous firing of the presplit blastholes and detonators (possible locations marked with
the buffer row blastholes. The blastholes circles). Figure 13 shows that the MBDP model
malfunctioning occurred in buffer rows. predicted dynamic pressure a charge receives
According to the modelling result, a new blast before its designated initiation time. The high
design was recommended that the presplit row was intensity of the dynamic pressure was identified in
removed as well as the blasthole spacing was the lifter blastholes which were initiated at the
increased. Figure 12 shows that the predicted centre of the row and fired with 3 ms delay
dynamic pressure from the new design was successively outwards. The lifter blastholes are
significantly reduced from 760 bar to 520 bar. fired last of all other blastholes in the round. Since
Since then, the blasts with the new design have this tunnelling is in a rural area, the blast vibration
performed without misfires. The model input for is not a concern. Therefore, a design scenario with
the rock condition and near-field blast vibration all lifter blastholes fired instantaneously to
attenuation are taken from another site. The eliminate the dynamic pressure received by a
absolute pressure value may not be true. However, neighbouring blasthole. Figure 14 shows the
the trend seems to be reliable. MBDP model prediction of the new design with

- 405 -
all lift blastholes fired instantaneously. The time window broadening, and delay time
dynamic pressure peak from the altered design is modelling. Blasting and Fragmentation, Vol. 2,
reduced from 580 bar to 405 bar (by 30%). No.2, 91-116.

5 CONCLUSIONS Yang, R. & Kay, D.B. 2011. Multiple seed blast

vibration modelling for tunnel blasting in urban
Excessive dynamic pressure a later firing charge environments, Blasting and Fragmentation, Vol.
received from earlier firing charges can cause the 5, No. 2, PP. 109-122.
charge desensitisation to a low VoD, low energy
release, and generate excessive post blast fumes. Yang, R. 2018. Modeling VoD and initiation effects
Such dynamic pressure may also damage the on rock fragmentation within multiple blasthole
initiation systems causing misfires. When the fragmentation (MBF) model. Blasting and
Fragmentation Vol. 11, No. 2, 2018.
dynamic pressure exceeds the threshold of the
initiation of the primer, sympathetic detonation
can occur, which is a significant safety hazard and
can disrupts the mine productivity. Recently
developed MBDP model was shown to be a useful
tool to assist alleviating blast dynamic pressure in
blast patterns to reduce the probability of blasthole
misfires or to increase explosive performance
in the field. The MBDP model is based on
fundamental physics as well as empirical field
measurement. It shares the same input with the
MBF model, which makes it easy to apply in the
field. The MBDP model can model the effects of
all blast design parameters on the dynamic
pressure a charge receives before its designated
detonation. The MBDP model has been applied to
tunnelling and open pit blasts and found to be
useful. The MBDP model can be applied along
with other blast prediction models to improve blast
designs for desirable blast results. The model is
suitable for open pit, tunnelling and underground
blasting.

6 ACKNOWLEDGEMENT

Mr G.C. Zhao, E. Hugaas, P. Reinders, L, Pratt, &

R. Proulx from Orica are greatly acknowledged for
their assistance in the field work and discussions.
The author also acknowledges the Orica
management’s permission to publish this paper.

REFERENCES

Persson, P., Holmberg, R. & Lee J. 1994. Rock

blasting and explosives engineering, CRC Press
L.L.C.

Yang, R. & Pelletier, L. 1998. Evaluation of the

resistance to pre-compression of high energy
magnums in field and underwater blasts. Orica
Internal Report, 1998.

Yang, R. & Scovira, D.S. 2008. A model of peak

amplitude prediction for near field blast vibration
based on non-linear charge weight superposition,

- 406 -
- 407 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Optimisation of surface blasting by photogrammetric 3D models

from drone imagery

A. Gaich, M. Pötsch & M. Baumgartner

3GSM GmbH, Graz, Austria

ABSTRACT: 3D models from drone imagery are now state-of the-art. On the one hand, they are useful
for doing blast designs in a reproducible manner and the other hand, to get objective analyses of the blast
results. The connection of blast design and blast analysis forms a closed loop that enables the
optimisation of blast sites in a traceable and documented way. Optimisation bases on the tracking of key
performance indicators (KPIs) that compute automatically from design and blast result. This enables a
stepwise, reproducible improvement of surface blasts by including specific characteristics of the site.
The contribution provides a description of the control loop and required components for it.

1 PROBLEM STATEMENT 2 METHOD OVERVIEW

Negative impacts of sub-optimal blasting in The suggested method reproduces what blasters
quarries include economic and ecologic, as well as usually perform in their daily job: they use their
safety aspects. Sub-optimal blast designs lead to experience for designing a blast based on
non-uniform, unwanted fragmentation and knowledge they gained from previous blasts. This
consequently to increased efforts in loading and applies for example by adapting the design based
hauling as well as efforts in primary (and on local site characteristics (e.g. bench face
secondary) crushing. It leads to irregular bench geometry and geologic situation).
faces and uneven floors that influence the next The algorithmic approach to such method
excavation round negatively. And there is less requires two things: (i) a blast design that is
usable material. reproducible, i.e. all determinant parameters of the
The ecologic aspect includes higher CO2 blast site are recorded and (ii) an objective analysis
emissions at the same production volume, as well of the blast result, e.g. by measurement of several
as increased energy consumption for loading, characteristic properties of the muck pile. From
hauling, and crushing. both steps representative values are determined
Related to safety and environmental impacts, and stored. These parameters or key performance
better blasts provide better walls and floors and indicators (KPIs) provide an independent way to
reduce impacts in the vicinity of a quarry by compare blasts with each other.
omitting fly-rock and reducing vibrations. Obviously, changes in the design lead to
So altogether, there are many reasons to make changes in the result. If certain KPIs from design
good blasts. and KPIs from result are recorded for each blast,

- 408 -
then blast design and blast analysis are linked of drones gave photogrammetry new
together forming a control loop (see Figure 1). momentum.
This control loop is executed until specified target Structure from Motion (Snavely et al. 2008,
values from production are met. Depending on Szeliski 2011) enables the generation of detailed,
actual site conditions different KPIs may be taken complex, and highly accurate 3D models from a
into account, for example: set of unorganised photos. Prerequisite is sufficient
redundancy between the images, i.e. every spot of
- Fragmentation parameters (x50, x80, N) the surface of interest must be visible in several
- Muck pile volume and shape, swell factor photographs.
- Burden, Spacing, S/B ratio, total drilling One important characteristic of Structure from
length Motion is its ability for auto-calibration.
Geometric distortions of the lens systems (always
present) are compensated while generating the 3D
model. In consequence, no pre-calibration is
necessarily required, thus cameras and drones that
are not specifically made for surveying tasks are
now eligible for surveying tasks.
Another characteristic feature of Structure from
Motion is that it allows for a highly automated
procedure that – under certain circumstances –
enables even fully automatic 3D model generation.
Depending on the size of the region of interest
and the required level of detail, 3D models of blast
sites or muck piles may contain 50–300 photos
(3D models of entire operations may include some
1.000s of images). Models are generated on site
and the positional accuracy of about 1-2 cm is
achieved without special efforts.
Positional accuracy comes from registering the
Figure 1. Reproducible blast design and objective 3D model in a superior co-ordinate system, aka
blast analysis form a control loop that enables referencing. There are several possibilities for
stepwise improvement of the blast layout referencing a 3D model, such as using surveyed
according to given operational target values. ground control points (GCPs) or by using precise
The connection bases on tracking of key
performance indicators.
positioning information from the drone itself.
Figure 2 shows a 3D model that computed in
about 30 minutes from 150 photos on a laptop
Using suitable software, KPIs are determined computer. It contains 2 million surface points and
directly from the 3D models. They are collected in has a ground sample distance (GSD) of 1 cm/px.
an appropriate data base and visualised over time. Referencing bases on 4 GCPs, positional accuracy
This documents the current performance in an is 1 cm.
operation or shows the way how the optimisation
procedure performs. 4 REPRODUCIBLE BLAST DESIGN
The next sections describe important
components of the system. Reproducible blast design requires a permanent
record of all geometric relevant parameters of
3 THE DATA BASIS – 3D MODELS FROM a blast site. This includes the entire area to blast
DRONE IMAGES including the bench face as well as the drill
pattern.
Photogrammetry enables to generate three- A detailed 3D model captures the area to blast
dimensional models from a series of overlapping (Figure 3). A digital drill pattern contains at least
photographs. Its beginning dates back into positional data on the collars of the boreholes,
the 1850s with a broad application in the their lengths, inclinations, and azimuths. Ideally, it
early 20th century. The introduction of new also includes the position as drilled as well as the
algorithms from ‘Computer Vision’ in the late actual course of the hole and the actual length.
1990s and the ‘Structure from Motion’ concept Both together (3D model and digital drill
in the 2000s as well the broad availability pattern) form a comprehensive documentation of

- 409 -
 Figure 2. 3D model of a blast site; length was about 115 m; the muck pile covers an area of out 4,500 m2.
The model contains 150 photos taken with an interactively flown, off-the-shelf drone.

Figure 3. Left: 3D model of a blast site computed from 100 drone images; right: drill pattern in digital form.

the blast site that enables to review all relevant do the survey second, all planning is reactive and
information also later in time – which we therefore the possibilities for optimisation are reduced.
call reproducible. Figure 4 shows the result of a proactive blast
Besides comprehensive documentation, it is design. Borehole positions and angles are adjusted
important to do blast design in a proactive way. – in particular for the front row – to account for
This means to plan the blast according to the the geometric and geologic conditions of the bench
actual geometric and geologic situation and faces. Special focus is put on the hole burdens to
includes the adaptation of the drill pattern meet a target corridor which, on the one hand, aids
according to the actual conditions of the bench to reduce fly-rock incidents and, on the other hand,
face. Consequently, this requires the 3D model to forms the basis for a balanced energy distribution.
be generated before doing the design. In contrast, On the face, lighter shading highlights areas at
when drilling a regular standard pattern first and which the target burden could be complied while

- 410 -
 Figure 4. Program interface for proactive blast design; the software shows the minimum burden over the entire
bench face as well as the distances between the boreholes using distinct colourisation.

darker shading highlights areas with excess burden result, an objective analysis uses quantitative
due to the bench irregularity. Light burden areas parameters, i.e. measurements. The starting point
could be avoided. is obviously the muck pile, its position, shape, and
The resulting borehole plan exports directly to volume but, very important, also the particle size
modern drill rigs and the current drill locations distribution. Modern approaches now allow
read back into the optimisation software. A performing the latter from drone images in an
consistent data round-trip from design as planned automatic manner.
to execution and back as drilled could be
established. A common interface for that follows 5.1 Automatic 3D particle size determination
the IREDES initiative (International Rock
Excavation Data Exchange Standard, IREDES Contact-free particle size distribution include two
2010). major steps: (i) the separation of single particles
and (ii) assigning size to the separated fragments.
5 OBJECTIVE BLAST ANALYSIS Classically, such approaches use 2D image
analysis methods to separate individual fragments.
In contrast to a subjective assessment of a blast Depending on the light situation, this process can

Figure 5. Strong shadows mislead 2D image processing algorithms (left) while using the 3D shape makes
fragment boundaries easier to detect (right).

- 411 -
 Figure 6. Automatic determination of the particle size distribution from a 3D model with particles coloured
according to size; the combined 3D-2D analysis allows the analysis of the entire muck pile surface, so that the
distribution curve consists of several 100,000 individual measurements.

be ambiguous, since prominent edges appear in the highly reduced.

images that are not originating from fragment Since the 3D model provides a detailed
boundaries but rather from rock structures or description of the entire muck pile surface it can
shadows from the shape of the fragment (see be evaluated in a single run. Consequently, the
Figure 5). resulting size distribution curve builds from
If the separation of particles is incorrect, all measurements over the entire area (instead sample
subsequent steps such as sizing are also. locations). Therefore, the number of measurements
It is more robust to do such analysis on a quickly may reach some hundreds of thousands.
detailed 3D model and to include 3D shape So even if single mismatches should occur, they
processing into the particle separation procedure. do no influence the course of the distribution curve
It eases particle delineation and sizing and significantly.
misdetections caused by shadows or structures are Figure 6 shows a 3D model with automatic

Figure 7. The comparison of 3D models determines the volume of the muck pile and the bank volume.
Their ratio leads to the swell factor.

- 412 -
 Figure 8. KPI tracking connects blast design with blast analysis and closes the control loop.

particle size analysis and the resulting distribution Obvious examples are the position, shape and
curve. The passing values, e.g. at 50% and volume of the muck pile. However, fragmentation
80%, or the non-uniformity N are parameters or the particle size distribution is probably one of
that can be determined directly from the the most important indicators for assessing the
distribution curve. quality of the blasting work.
In the actual case, particles are coloured These values from the blast analysis are
according to their size, which provides an instant collected and linked to other KPIs, e.g. from the
view on the size distribution and oversizes as well blast design. Examples for KPIs from design
as fines highlight immediately. include burden, spacing, their ratio S/B, or drill
pattern irregularities.
5.2 Volume measurements By connecting the KPIs from the blast
design with the KPIs from blast analysis,
Volumes are determined from 3D models by
a control loop is created (see Figure 8),
comparing surfaces. The prerequisite is that both
which is used to determine the actual quality
3D models are available in the same coordinate
of blasts and records the effect of any changes
system. It is also important that the drone flights
for optimising the blasting. Depending on
are done at the right time in order to acquire the
economic or ecologic target values from the
correct, unobstructed surfaces of interest. For
operation, different KPIs from larger set are
example, the precise volume of the muck pile
selected, subsequently recorded, and their
requires a survey before any mucking and after the
progression visualised.
entire muck pile has been cleared.
KPIs are usually displayed over time or as
Figure 7 shows two examples for volume
correlations with each other. For example, Figure
measurements: (i) the bank volume is determined
9 shows the value x50 of the particle size
from the cleared bench faces before and after
distribution for 30 blasts in an operation and the
blasting, while (ii) the muck pile volume is
course of the non-uniformity factor N is given.
determined by comparing the 3D models before
Figure 10 shows the resulting muck pile volumes
and after mucking.
and the swell factors while Figure 11 plots the
6 KPI TRACKING relationships between individual parameters, e.g.
between the total drilling length and the resulting
An objectified comparison of different blasts resp. muck pile volume or another way of showing
(sic) an objective evaluation is carried out by the uniformity of the muck pile by correlating
determining key performance indicators (KPIs). x80 and x50.

- 413 -
 Figure 9. Tracking of x50 (left) and the non-uniformity N (right).

Figure 10. Tracking of the muck pile volume (left) and swell factor (right).

Figure 11. Correlation of parameters: muck pile volume vs total drilling length (left); x80 vs x50 (right).

The system used for KPI tracking permits a free form a comprehensive data basis for efficient
configuration of various parameters, as well as the planning, documentation, and quantification of
import of externally determined values, (e.g. blasts. A multitude of drones is suitable and
energy consumption at the crusher, specific modern photogrammetry software creates 3D
explosives consumption, costs per drilling meter, models automatically and accurately in reasonable
etc.). This way, a generic correlation of KPIs time and on site.
becomes available. The resulting 3D models are used for
reproducible blast design and objective blast
7 CONCLUSIONS
analysis. Blast design shall be done proactively by
Photogrammetric 3D models from drone images adapting the layout according to the specifics of

- 414 -
the actual rock mass and bench face conditions.
Blast analysis include the determination of the
particle size distribution that is performed fully
automatically using a combined 3D-2D approach,
as well the determination of several volumes.
Blast design and blast analysis are linked by
tracking key performance indicators (KPIs). Any
changes in the design lead to corresponding
changes in the KPIs (of design and analysis),
which offers the opportunity to adapt the design to
site specifics, (e.g. face irregularities and geologic
situation) until the result meets given performance
parameters of the production (e.g. fragmentation).
Continuous KPI tracking reveals deviations
immediately and allows applying countermeasures
at the earliest stage hence ensuring to keep the
desired quality of the blasting result.
In addition to the economic advantages of the
method, CO2 emissions are sustainably reduced
and the safety aspect (fly-rock) is constantly taken
into account.

REFERENCES

IREDES, 2010. Drill Rigs Equipment Profile V 1.3.

IREDES Initiative Ladbergen.

Snavely, N., Seitz, S.M. & Szeliski, R. 2008.

Modelling the world from internet photo
collections. International Journal of Computer
Vision: 80(2):189-210.

Szeliski, R. 2011. Structure from motion. In:

Computer Vision. Texts in Computer Science.
Springer, London. https://doi.org/10.1007/978-1-
84882-935-0_7.

- 415 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Thermo-mechanical behaviour of high strength concrete containing

polypropylene under thermal loading

Z. Kurdo & L.G. Clough

School of Civil Engineering & Surveying, University of Portsmouth, UK

ABSTRACT: This paper discusses an experimental study investigating the thermo-mechanical

behaviour of high strength reinforced concrete (HSC) containing polypropylene and the development of
a combined thermal, compression and impact rig for testing the HSC members in equivalent fire and
blast conditions. The reduction in concrete compressive strength at high temperatures and after cooling is
one of the most critical processes affecting the load bearing capacity of reinforced concrete structures.
Previous investigations have shown that explosive spalling in HSC members exposed to fire can be
mitigated through addition of polypropylene or steel fibres to the concrete mix. When concrete
containing polypropylene fibres is exposed to fire, the fibres will melt at a relatively low temperature of
162– 170 C, forming randomly oriented pores inside the concrete microstructure. These pores facilitate
dissipation of high vapour pressure generated within a concrete member. Concrete cubes were subjected
to temperature increments of 20 C per minute for 25 minutes, up to 500 C. The temperature was
maintained at 500 C for 30 minutes before cooling. Compressive strength tests were carried out
according to BS EN 12390-3-2009 on specimens after cooling. It was found that the addition of steel and
polypropylene fibres does not influence the compressive strength. However, the use of polypropylene
fibres in concrete can reduce or eliminate explosive spalling when exposed to high temperature,
providing channels for pore pressure to escape. The mix proportions adopted in this study was based on
the work by Khalifa et al. (2001) and Varona et al. (2018) who carried out extensive studies to the
understanding of the way in which fibres act and to optimising the fibre dosage. This paper also details
the development of an experimental rig to observe the dynamic response of reinforced concrete
structures with polypropylene fibres, to combined thermal and impact (akin to blast) loads. In order to
replicate thermal loads from an explosion event, a bespoke thermal rig was developed using ceramic
heating elements. To replicate imparted energy similar to a blast load, the bespoke rig was developed
with drop weights above the heated specimens. A modified hydraulic press was adopted within the rig to
apply axial load to the concrete specimens. The thermo-mechanical behaviour of a numerical model was
compared to experimental samples using a finite element analysis (FEA). The results of experimental
work and the FE analysis revealed a reasonable agreement, with slight deviation due to concrete spalling
in some cases. Keywords: Concrete, Compressive strength, Thermal, Fire. Polypropylene fibre, blast.

1 INTRODUCTION resistance. However, prolonged fire conditions

have a wide range of consequences (Qin et al,
Reinforced concrete structures outperform other 2021). Cracking and spalling of the surface are
structural materials in terms of inherent fire signs of distress, induced by variable expansion of

- 416 -
subsequent layers of the concrete as it is subjected Concrete is a complex material whose
to heat and by internal pressure due to evaporation properties can dramatically change when subjected
of moisture embedded within the concrete. to high temperatures. The primary effects of fire
The consequences of high temperatures can be on concrete are loss of compressive strength and
seen in the form of surface cracks and spalling spalling. Even though much research has been
(Hertz & Sorensen 2005). When subjected to high undertaken regarding these physical processes,
and rapidly increasing temperatures, akin to fire there is still a need for further thorough studies of
situations, spalling occurs as layers of concrete the effects of thermal exposure (Fletcher et al.
separate from the surface. This separation can 2007). Reinforced concrete is an inhomogeneous
occur immediately after intense heating and is material, consisting of a mixture of cement paste,
followed by high energy explosions, it can also aggregate, and steel or fibre reinforcement. Each
occur later in the fire due to fracturing from of these components reacts differently to heat
increasing weakness, (Kodur 2014). The exposures, therefore the behaviour of the
repercussions are restricted because the damage is composite system in a fire is complex to
little, but excessive spalling can result in early loss characterise and model.
of structural support and stability. When concrete is heated, it initially expands
To mitigate fire induced spalling in RC until evaporation of the capillary moisture,
(reinforced concrete), researchers have suggested reaching temperatures of around 150 C. Further
the addition of different fibre combinations, such heating causes the adsorbed water to be removed
as polypropylene fibres, steel fibres and hybrid from the gel pores, resulting in shrinking up to 300
fibres. Polypropylene fibres melt at relatively low C (Hager 2013). The concrete mixture continues to
temperatures (162 C) thus leaving micro and shrink until most of the adsorbed water is removed
macro channels randomly oriented inside the and some of the hydrated compounds are
concrete that help diffusion of pore pressure destroyed up to around 500 C, at which point it
(Bilodeau et al. 2004) resulting in spalling begins to expand again (Sarshar 1989).
mitigation. Uncertainties about the effect of heat Aggregates make approximately 75% of the
on concrete members are among the reasons for volume of concrete and hence have a significant
having to introduce higher safety factors in impact on its thermal behaviour. For different
concrete design than in steel design works where forms of aggregate, however, the volume of the
the effect of heat on steel members is well aggregate grows as the temperature rises.
understood. (Khoury 2008). Consequently, a complicated process involving
A succession of large explosive incidents aggregate thermal expansion and cement matrix
combined with considerable thermal loading (fire hydro-thermal phase transformation occurs. The
or explosive) have occurred in recent years. cement matrix shrinkage is overwhelmed by
Recent explosion events including the Bangladesh aggregate expansion. This indicates that the
mosque gas pipeline explosion (Aljazeera 2020), component materials have seen a major volume
Beirut warehouse explosions (BBC 2020) and shift, resulting in severe internal pressures that
explosion at China chemical plant (The New York might lead to concrete explosive spalling and the
Times 2020) have highlighted a growing need to structure collapsing (Naus 2006).
understand the response of structures to equivalent The design and results of a number of
high explosive blast and thermal events. experimental trials to investigate the response of
Therefore, a fire-resistant analysis of reinforced steel and polypropylene fibre reinforced concrete
concrete structural elements based on temperature cubes to compression and long duration thermal
was needed. loads are presented in this paper. A computational
Therefore, an investigation into the blast model using finite element analysis (FEA) was
resistance of RC structures at high temperatures is developed for verification of the experimental
needed. In terms of loss of strength, research has investigation into the thermo-mechanical response
shown that at intermediate temperatures ranging of concrete structures at high temperatures.
from 100 C to 400 C (Arano et al. 2021), the To replicate the in-service loading conditions
compressive strength of HSC can be reduced by on the RC columns, it was necessary to develop
nearly 40%, compared to room temperature and construct a test rig. The rig was designed to
strength. This strength reduction is approximately allow exposure of the concrete column surface for
20 to 30% higher than normal strength concrete the impact load to be applied successfully. To
(NSC) exposed to the same temperatures facilitate the constant compressive load a 200 kN
(Phan 2002). (max) heavy duty steel plate hydraulic press with

- 417 -
 Figure 1. Mixing and casting concrete cubes.

pump and manometer was adopted to provide an extensive studies regarding the optimisation of
equivalent in-service load. A drop weight impact polypropylene fibre dosage in HSC.
rig was designed and constructed with free-falling
guided mass to laterally impact the column at its 2.1 Specimen preparation
midpoint. The rig will provide experimental data
to validate numerical simulations of combined Each material component was accurately weighed
thermal, compression and impact loads on in accordance with Table 1 mix proportions. In a
concrete columns. The development process horizontal pan mixer, a dry mix of coarse
included the design and fabrication of the aggregate, fine aggregate, sand, cement, and silica
thermally insulated steel box, housing four fume was added and mixed for 5 minutes. By
ceramic heating elements, to heat the reinforced mixing the dry particles before adding water the
concrete columns. In order to maintain high liquid superplasticisers reduced the potential for
temperatures, heat resistant and insulating blanket the particles to clump together (Maca et al. 2013).
was used on the inside of the steel box. The The mixing process with steel fibres was extended
dimensions of the steel box were 500 mm long by for another 4-5 minutes to ensure that the fibres
570 wide and 270 mm high. were evenly distributed throughout the mixture
(Bekaret 2022). The moulds were prepared by
2 EXPERIMENTAL WORK spraying a demoulding agent to the interior
surfaces prior to casting. Figure 1 show the
This paper studies the behaviour of compressive process of mixing and casting of HSC. The
strength of concrete cubes that are subjected concrete mix was poured into the 100 mm cube
to long duration high temperature at 7, 14, moulds on a vibrating table, which vibrated for
and 28 days. The experiments were conducted 60-90 seconds ensuring removal of air bubbles
as part of wider research study into the response from the wet concrete.
of RC columns to combined thermal & A total of nine samples for each mix design
blast loading. Specimens were subjected to were cast. The cubes were cast using the mix
heat load then compression loading (after
cooling), control samples were also tested Table 1. Mix proportions of the concrete mixtures
under compression load alone. Specimens were per cubic metre.
subjected to temperature increments of 20 C
per minute for 25 minutes, up to 500 C. Material NSC HSC HSFRC
The temperature was maintained at 500 C for Cement (kg/m3) 733 415 415
30 minutes before the cooling process. After Water 220 139 139
30 minutes of cooling the specimens were Superplasticiser - 13.75 13.75
tested under compressive loading. Compressive (kg/m3)
strength tests were carried out according Silica fumes (kg/m3) - 45 45
to BS EN 12390-3-2009, it was found that the Silica sand (kg/m3) - 432 432
addition of steel fibres and polypropylene does Sharp sand (kg/m3) 220 439 439
not influence the compressive strength. The mix Fine aggregate (kg/m3) 557 488 488
proportions adopted in this study (Table 1) Coarse aggregate 20 736 561 561
was based on the work by Khalifa et al. mm
(2001) and Varona et al. (2018) who undertook

- 418 -
 Table 2. Summary of the different mixes and different tests.

Material Property Test Curing durations Total No. of test

Control mix containing: - Compression 7 days 27
2.0 Kg of PP test 14 days
2.5 kg of PP 28 days
3.0 kg of PP
Mix containing 2.5 kg PP Heating Compression 7 days 9
test 14 days
28 days
Mix containing steel fibre Heating Compression 7 days 9
and 2.5 kg PP test 14 days
28 days

proportions as shown in Table 1. The silica fume with a specific gravity of 0.91 and a melting point
was not used as a replacement, but as an addition 162 C. Table 3 shows polypropylene fibres
to the cement. While the preservation of silica characteristics.
sand improves HSC homogeneously, the particle Two different high strength concrete mixes
size distribution is still not optimal, leaving voids (one with 2.5 kg/m3 of polyproline and one with
in the matrix that are filled with silica fumes. 2% steel fibre) were subjected to temperature up to
Published data indicate that silica fume concrete, 500 C, then cooled to room temperature and
when exposed to temperatures up to 300 C, subsequently loaded to failure in uniaxial
exhibits more spalling than non-silica fume compression. 13 mm long, high yield strength steel
concrete (Noumowe 2005). fibres with a diameter of 0.16 mm were used in the
Moulds were covered with a damp cloth and concrete mix. A 2% fibre content was chosen to
plastic sheet (Figure 1), then kept at room match typical steel fibre construction mixes which
temperature for 24 hours until the concrete was balance improved mechanical characteristics and
set. Cubes were de-moulded after 24 hours then cost, (Barnett 2007).
stored in the water tanks for curing. The cubes
were cured for a range of days: 7, 14, and 28 days. 2.2 Heating method
Table 2 shows a summary of the different mixes
and different tests are undertaken. After curing, the A temperature regulative furnace (Carbolite Gero,
cubes were removed from the curing tank and RWF1100 capable of reaching 500 in five
placed in the drying oven, set to a temperature of minutes) was used to simulate the effect of fire
110 C for 24 hours, until the cubes were dry. (through heat) on the concrete (Figure 2)
(Carbolite 2022). As the heating furnace housed
Table 3. Characteristics of polypropylene fibres three specimens at a time, each test was conducted
(Sika 2022). with three cubes, the average of the three is
reported in this work.
Density at solid state (𝑘𝑘𝑘𝑘/𝑚𝑚3 ) 930 The specimens were placed in the enclosed
Density at liquid state (𝑘𝑘𝑘𝑘/𝑚𝑚3 ) 850 furnace to minimise temperature variation between
Thickness (µm) 50 specimens. The specimens were heated to 500 C at
Width (µm) 150 a rate of 25 C /min. Thermocouple inserted
Length (mm) 19 & compared to the PTC 12/20 s displayed
Melting temperature (C) 162 temperature. The furnace was built with PID
Temperature at vaporisation (C) 341 temperature control and separate temperature
Burning temperature (C) 460 display Once the furnace reached 500 C the
temperature was maintained at a constant 500 C
for 30 minutes. To avoid thermal shock, the
In all mixes, 2.5 kg/m (0.25% by volume), samples were cooled in the furnace to 180 C
FIBERMESH (150-12 mm) type polypropylene before they were removed. Water was allowed to
fibres, were added. The 20 mm long escape during heating since the specimens were
polypropylene fibres were a non-absorbent type not sealed.

- 419 -
Figure 2. Electrical furnace used to heat the cubes, left) prior to heating, right) post heating.

Figure 3. Compression test setup.

- 420 -
2.3 Compressive testing The temperature time history specified by BS
EN 1992-1-2 Concrete General Rules- Structural
Compression tests were carried according to BS fire 2004 (Corri 2008) was applied uniformly onto
EN 12390-3-2009 out on the 100 mm cube the exposed surfaces of the specimens. Thermal
specimens to determine compressive strength and radiation and thermal convection transferred heat
estimate the elastic modulus (Figure 3). The into the specimens through the concrete surface.
specimens were tested at a rate of 6 KN/s in an The percentage change in length of a specimen per
ADR Touch Control PRO 2000 Auto BS EN degree of temperature rise is known as the
compressive machine (ELE International 2022) coefficient of thermal expansion. Table 4
until failure to determine the compressive strength. summarises the main material thermal properties
Because the effect of the steel fibres on for concrete which were used as input parameters
compressive strength is typically very low in the transient thermal analyses.
(Fehling et al. 2015), these tests were undertaken
to identify variation within the concrete matrix Table 4. Thermal and mechanical property of
across samples. The density of the specimens was Concrete.
also evaluated in order to establish the relative
consolidation of the samples due to vibration or Material Property Value Unit
self-compaction. Poisson’s ratio 0.3 n/a
Coefficient of thermal 12.0E-6 C-1
3 NUMERICAL MODELLING expansion
Thermal Conductivity 1.5 W/m. C
Using the finite element analysis (FEA) software
Specific heat 2.65 mJ/mm3.
LUSAS, coupled (thermal and structural),
coefficient* C
non-linear, transient analyses were performed on
the concrete cubes from the experimental tests.
*Specific Heat Coefficient = Specific Heat Capacity
Solid elements (HX20) were used to model the x Density
cubes. The HX20 is a conventional high order
iso-parametric element with nodes on the mid-side An implicit nonlinear analysis was used. For
and linear stress fluctuations (Lusas 2019). The large models, inverting the matrix is highly
element mesh of the developed model is shown in expensive and requires advanced iterative solvers
figure 4. An initial model created with thermal (over standard direct solvers). These solutions are
load applied (applied temperature and thermal unconditionally stable and facilitate larger time
flux) was examined within the FEA modelling steps. Despite this advantage, the implicit methods
framework. A two-point Gauss integration system can be extremely time-consuming when solving
was utilised to define the material mechanical nonlinear problems (Clough 2017). An implicit
characteristics in the long direction. was preferred over an explicit analysis because
explicit can only be used with consistent time
steps, Rayleigh damping parameters are not
supported, and stress variations across an explicit
element are not supported.

4 RESULTS AND DISCUSSION

Experimental tests have demonstrated the

effectiveness of fibres as a means of reducing the
water vapour pressure in concrete. Internal
pressures can be efficiently reduced by using PP
fibres at an amount of around 2.5 kg/m3. The
addition of fibres reduces the maximum pore
pressures and decreases the temperature of the
pressure peaks. Furthermore, it has also been
found that when the fibre concentration exceeds
2.5 kg/m3, the impact of fibres does not
significantly increase.
Figure 4. Mesh of the finite element model. Figure 5 compares the compressive strength
. (post cooling) of concrete specimens to the

- 421 -
 Figure 5. Cube compressive strength containing 2.5 kg/m3 PP.

Figure 6. Micro cracks at the surface of the specimens after heating.

compressive strength prior to heating as a function external of the cubes. Figure 6, also presents the
of curing time. The values are provided as a ratio appearances and observed cracks of the specimens
to the strength at 24 C (room temperature). Each tested in under compression conditions post fire
point on the graph indicates the average residual exposure.
compressive strength of one of three specimens. The polypropylene regulates moisture
Figure 6 shows micro cracks were visible in all expulsion from the concrete at both normal and
specimens exposed to fire. The radial cracks on the high temperatures. As a result, the pore structure at
outer edge of the cubes were mainly caused by the high temperatures may have a significant impact
heating process that is directly applied to the on the spalling behaviour of high strength

- 422 -
 Figure 7. Shows specimens exploded under compression test post heating.

polypropylene fibre reinforced concrete (Bilodeau phase.

et al. 2004). The distribution and amount of moisture in
The benefits of PP fibres in enhancing the concrete is closely related to its performance when
permeability of high-performance concretes, exposed to high temperatures. The structural
minimising the likelihood of spalling behaviour in changes that occur during curing impact the curing
real fires, were supported by the findings. Melting process and modify the mechanical and physical
polypropylene fibres improves the thermal characteristics of concrete (Dauti et al. 2018).
behaviour of steel fibre high strength concrete. In High moisture content confined inside the concrete
the event of prolonged high-temperature exposure, specimens is one of several key factors
not all water is removed quickly enough from the contributing to spalling at high temperatures.
high-strength concrete with no PP. This will result Figure 7 shows concrete specimens exploded
in vaporisation at higher temperatures and the under compression test post heat exposure from
formation of high pressures within the mixture control mix for 7, 14 and 28 days.
(Arano et al. 2021). The extra porosity and narrow According to Noumowe, both concrete
channels formed by the melting of polypropylene compression and tension strengths decrease after
fibre minimises the internal vapour pressures in the temperature reaches 350 C (Noumowe 2005).
the concrete, reducing the potential of spalling. This may be due to discrepancies in thermal
Dimensions and quantity of fibre may have an expansion coefficients between particles and
impact on microstructural behaviour. Under the calcium hydroxide breakdown causing
thermal treatment at 200 C, the permeability of the microcracks to form. Previous research has
fibre reinforced high-performance concrete observed a decrease in strength over 300 C,
increases (Ozawa & Morimoto 2014). This is due typically occurring between 300 and 400 C. He
to the fibre melting at 170 C during the heating also emphasised on the overall stiffening of the

Figure 8. Ceramic heating apparatus.

- 423 -
 Figure 9. Impact load rig.

cement gel or the increase in surface forces between 200 and 350 C. There is strong agreement
between gel particles owing to the elimination of between the work carried out by Noumowe and
absorbed moisture caused a recovery of strength experimental work in this study. Particularly, in

- 424 -
 Figure 10. Compressive load rig.

terms of regaining compressive strength post 400 To apply axial load to the concrete specimens,
C heat exposure. the rig used a modified hydraulic press. The
According to this experimental test series, a hydraulic press is capable of exerting 200 kN and
sufficient amount of polypropylene fibres can 188 mm ram travel. Figure 10 shows an in-situ
lower the probability of spalling. Despite the fact compressive load rig, while it is being calibrated
that even when thermal expansion is unconstrained with an external loading machine.
at the fire-exposed surface. However, it was noted
that a number of control samples experience 5 CONCLUSIONS
spalling.
As part of a wider study investigating the The purpose of this study was to obtain data on the
response of RC columns to combined thermal impacts of high temperatures up to 500 C on the
and blast loads loading a number of experimental characteristics of two concretes. A high-strength
rigs was developed. In order to replicate concrete with polypropylene fibres was compared
thermal loads from an explosion event, a to a high-strength concrete without fibres.
bespoke thermal rig was developed using All HSC specimens containing silica fumes
ceramic heating elements. Figure 8 shows but no PP, have shown higher compressive
RC beam being heated with ceramic elements strength compared to specimens with PP
up to 460 C Thermocouple of J type attached fibres. However, explosive spalling occurred
to the face exposed to the heating element to due to high pore pressure build-up. This suggests
monitor the temperature. that the internal pressure may be the primary
To replicate imparted energy similar to a cause of explosive spalling. Significant reduction
blast load, the bespoke rig was developed in pore pressure in HSC can be achieved with
with drop weights above the heated specimens. the additions of PP fibres in the mixture.
Figure 9, illustrates the bespoke rig was Reduction in pore pressure appeared to increase
created with drop weights above the heated with increasing dosages up to 2.5 kg/m3. The extra
specimens to simulate imparted energy equivalent porosity and narrow channels formed by the
to a blast load. melting of polypropylene fibre may minimise

- 425 -
internal vapour pressures in the concrete, Analytical Methods in Geomechanics. 42.
eliminating the potential of spalling. 10.1002/nag.2802.

6 ACKNOWLEDGMENT Ele International, 2022: https://www.ele.com/product/

adr-touch-control-pro-2000-auto-bs-en-
The authors would like to express gratitude to the compression-machine-
University of Portsmouth for funding this research.
Fehling, E., Schmidt, M., Walraven, J. & Ãhlich,
REFERENCES S. 2015. Ultra-high performance concrete UHPC
: fundamentals, design, examples.
Aljazeera News, ‘Bangladesh mosque gas pipeline Beton-Kalender Ser. Wilhelm Ernst & Sohn
explosion’ Verlag fur Architektur und Technische, Berlin,
https://www.aljazeera.com/news/2020/09/killed- Germany.
bangladesh-mosque-gas-pipeline-blast-
200905072009767.html, 2020. Fletcher, I., Welch, S., Torero, J., Carvel, R, &
Usmani, A. 2007. Behaviour of concrete
Arano et al. 2021. Material characterisation approach structures in fire. Thermal Science - THERM SCI.
for modelling high-strength concrete after 11. 37-52. 10.2298/TSCI0702037F.
cooling from elevated temperatures.
Hager, I. 2013. Behaviour of cement concrete at high
Barnett, S.J. et al. 2007b. Ultra-high performance temperature. Bulletin of the Polish Academy of
fibre reinforced concrete for explosion resistant Sciences, Technical Sciences. 61. 10.2478/bpasts-
structures. In: Russell, M.I., Basheer, P.A.M. 2013-0013.
(Eds.), Concrete Platform. Queen's University of
Belfast, Belfast, UK, pp. 565-573. Hertz, K.D. & Sorensen, L.S. 2005. Test method for
spalling of fire exposed concrete. Fire Safety
BBC News, ‘Beirut explosion: What we know so far’ Journal 40, 466–476.
https://www.bbc.co.uk/news/world-middle-east-
53668493, 2020. Khalifa et al. 2001. High-temperature behaviour of
HPC with polypropylene fibres From spalling to
Bekaert 2022. ‘https://www.bekaert.com/en/product- microstructure.
catalog/construction/underground/steel-fiber-
reinforced-concrete-faq#3’. Khoury, G.A. 2008. Concrete spalling assessment
methodologies and polypropylene fibre toxicity
Bilodeau, A., Kodur, V.R. and Hoff, G.C. 2004. analysis in tunnel fires. Structural concrete, 9(1),
Optimization of the type and amount of 11-18.
polypropylene fibres for preventing the spalling
of lightweight concrete subjected to hydrocarbon Kodur, V. 2014. Properties of concrete at elevated
fire. Cement and Concrete Composites, 26(2), temperatures. ISRN Civil Engineering. Vol 2014,
163-174. Article ID 468510.

Bratina, S., Cas, B., Saje, M. & Planinc, I. 2005. LUSAS 2019. User Reference Manual. Surrey, UK.
Numerical modelling of behaviour of reinforced Ozawa, M. & Morimoto, H. 2014. Effects of
concrete columns in fire and comparison with various fibres on high-temperature spalling in
Eurocode 2, Int. J. of Solids & Structures, 42 high-performance concrete, Constr. Build. Mater.
(2005), 21-22, pp. 5715-5733. 71 (2014) 83–92.

BSI, 2009. Testing hardened concrete. Making and Naus, D. (2006). The effect of elevated temperatures
curing specimens for strength tests. BSI. on concrete materials and structures—A
Literature Review at ORNL. 10.2172/974590.
Butterworth, J.W., Lee, J.H., & Davidson, B.J. 2002.
Experimental determination of modal damping Noumowe, A. 2005. Mechanical properties and
from full scale testing. microstructure of high strength concrete
containing polypropylene fibres exposed to
Dauti, D. et al. 2018. Modelling concrete exposed to temperatures up to 200 C." Cement and Concrete
high temperature: Impact of dehydration and Research, 35, 2192-2198.
retention curves on moisture migration.
International Journal for Numerical and Phan, L. 2002. High-strength concrete at high
temperature - an overview, Utilisation of high

- 426 -
 strength/high performance concrete. International
Symposium 6th Freunde der Bavingenieur,
Leipzig, GE.

Qin et al. 2021. A comprehensive review on fire

damage assessment of reinforced concrete
structures.The Authors. Published by Elsevier
Ltd.

Sarshar, R. 1989. Effect of elevated temperatures on

the strength of different cement pastes and
concretes. Imperial College of Science
Technology and Medicine University of London.

Sika, 2022.
https://gbr.sika.com/dms/getdocument.get/7f4699
51-b306-41d8-961b-
09de104f93c3/sika_fibermesh_-150.pdf

The New York Times, 2019. ‘Explosion at China

chemical plant’ https://www.nytimes.com/2019
/03/22/world/asia/china-chemical-plant-
explosion.html.

Varona et al. 2018. Influence of the nature of

aggregates on the behaviour of concrete subjected
to elevated temperature.

Wang, H-F. & Zhang, R-L. 2019. A method for

determining Rayleigh damping parameters of
complex field. Stavební obzor - Civil
Engineering Journal. 28. 420-434.
10.14311/CEJ.2019.03.0034.

- 427 -
 Maastricht Conference Proceedings 2022, R. Holmberg et al
© 2022 European Federation of Explosives Engineers, ISBN 978-0-9550290-7-3

Author index

Aiken J. 126 Johansson J. 164 Sanchidrián J.A. 184

Aiken P.E. 126 Johansson M. 106 Saraiva B. 238
Aimone-Martin C.T. 116, Scovira D.S 62
126 Kala V. 22, 50 Segarra P. 184
Álvarez-Fernández M.I. 140 Kaykov D. 196 Seidl T. 184
Andersen C.A. 224 Koppe J.C. 336 Schimek P. 184
Andersen K.H. 164 Koprev I. 196 Schneider J. 224
Kouvonen T. 28 Shishkov P. 290
Barnett S. 80 Kreivi J.A. 80 Sitonio F.F. 336
Baumgartner M. 316, 408 Kubat E. 22, 50 Steyl Q. 62
Bermingham L. 384 Kuhnow F.B. 212 Stoycheva N. 290
Bernard T. 326 Kurdo Z. 416 Sutoyo A. 272
Bihlar A. 384
Bosch P. 308 Laszlo R. 236 Tahernejad Javazm P. 50
Brask P. 224 Leite F. 28 Terziyski D. 196
Breeds C. 72 Little T.N. 364 Thyrring E.C. 224
Burnham R.E. 126 Leppänen J. 106 Torrance A. 280
Buschjost T. 300 Tshibangu J-P 174
Madshus C. 100, 164 Tuuna T. 256
Cavanough G. 280 Marijnissen J.S.C. 80
Clough L.G. 416 Meins B.M. 116 Valencia N. 152
Conde-Fernández L. 140 Meins O.C. 116 Yang R. 396
Conroy G. 344 Miklos L. 236 Van Landscoot P. 174
Couceiro P. 184 Miranda V. 28 Vanoverbeke V. 174
Victor B. 62
Davidsavor T.A. 90, 126 Naseri A. 106
Diehr G. 300 Navarro J. 184 Walker C. 80
Dowding C. 90 Nicola A. 236 Wallace J. 8, 72
Nikkhah A. 50 Wibowo A. 272
Edefors G. 106 Norén-Cosgriff K.M. 100 Wong I.G. 126
Worsey T. 300
Farnfield R. 384 Oliveira A. 28
Folkow P.D. 106
Park J. 164
Gaich A. 316, 408 Penev V. 290
Garkov I. 196 Pötsch M. 316, 408
Gates W. 8 Prendes-Gero M.B. 140
Ghicioi E. 236 Provost D.A. 90
Gjengedal S. 164
Gonzalez-Nicieza C. 140 Qvist S. 224
Gonze N. 174
Gou Y. 90 Ramstad N. 100
Gyngell B. 300 Ratliff J.K. 126
Redyke J. 116
Hakimi J. 50 Robinson K. 344
Hartlieb P. 184 Rønning S. 164
Hirthammer F. 250 Rosini L. 174
Ruspeckhofer M. 264
Jabalbarzai G.R. 50
Jitea C. 236 Saeedi G.R. 50
- 428 -