Pl Drevland Jakobsen

Estimation of soft ground tool

life in TBM tunnelling

Thesis for the degree of Philosophiae Doctor

Trondheim, January 2014
Norwegian University of Science and Technology
Faculty of Engineering Science and Technology
Department of Civil and Transport Engineering

NTNU
Norwegian University of Science and Technology
Thesis for the degree of Philosophiae Doctor
Faculty of Engineering Science and Technology
Department of Civil and Transport Engineering
Pl Drevland Jakobsen
ISBN 978-82-471-4973-7 (printed ver.)
ISBN 978-82-471-4974-4 (electronic ver.)
ISSN 1503-8181
Doctoral theses at NTNU, 2014:30
Printed by NTNU-trykk

Abstract
Increased urbanisation constantly demands more infrastructure, often requiring the
construction of tunnels and facilities underground. The complexity of Tunnel Boring Machine
(TBM) function and the complicated nature of soft ground and soil working environments
make the estimation of wear a challenging issue.
The following tests and procedures are included in the original contribution to this PhD study;
1) an overview and presentation of various laboratory methods designed to estimate and
assess soil abrasivity in connection with soft ground TBM tunnelling, 2) the development of
models, based on simplified laboratory tests, for the estimation of TBM tool life when
excavating soil and soft ground, 3) the development of the Soft Ground Abrasion Tester
(SGAT), designed to increase the validity of simplified estimation tools.
The simplified laboratory tests incorporate the Soil Abrasion Test (SAT), quartz content
measurements, and the geotechnical uniformity index C u . Test results have been correlated
with, and validated against, TBM tool life and performance data from 16 TBM projects.
Laboratory test results obtained from the SAT provide a good estimation of soft ground
excavation tool life. Furthermore, an empirical formula obtained by means of multiple
regression analysis using SAT and C u values as variables, and soft ground tool life as the
dependent variable, has been derived, and provides a good estimation of soft ground
excavation tool life. In addition, the validity of the formula is evaluated against an on-going
European TBM tunnelling project.
The SGAT has been developed in order to study how variation in geotechnical parameters
such as soil compaction and density, water content, groundwater pressure and soil
conditioning additives influence the abrasivity of soils. Furthermore, the SGAT provides
opportunities to measure the influence of abrasive wear by varying TBM parameters such as
thrust, rpm and torque. The main results obtained from the SGAT are that the moisture
content of a soil sample, and thus its compactibility, influences soil abrasivity by as much as
250%. There is a clear correlation between measured weight loss and torque requirement,
and a reduction of torque by approximately 40% is achievable with proper soil conditioning.
In addition, measured wear can be as low as 20% of that for an unconditioned sample.
Research results have been achieved mainly by using the following four approaches, 1)
literature studies, 2) laboratory tests, 3) field research, and 4) discussions and experience
sharing with individuals, users and experts in the tunnelling industry. These methods were
chosen since they offer a variety of approaches to the complex problem of soil abrasivity in
soft ground TBM tunnelling, and to avoid an exclusive focus on any single source such as
ideal laboratory tests, published literature or field experience.

Preface and acknowledgements

This PhD study has provided an extraordinary opportunity to develop, systematise and publish
hypotheses and results linked to soft ground TBM excavation tool wear. The problem
formulation has been relevant, and the results of the study have been the subject of requests
from project owners, contractors and TBM manufacturers. This has provided a source of extra
motivation during the study.
The use of soft ground TBM tunnelling technology is not widespread in Norway, and research
thus necessitated numerous field trips abroad to gather information. The field trips have
provided much inspiration since contractors have communicated a great deal of information.
The opportunity to experience new countries and cultures has also provided much personal
motivation.
The collection of real TBM production data has been the most time-consuming activity. Field
trips have also been a challenge for my small family. During the first year of the study, my
wife and mother-in-law took care of my son Theodor. Having my mother-in-law living under
the same roof in a 70m2 apartment provided a perfect excuse for numerous getaways. I am
very grateful to my lovely Senobia, my mother-in-law Leonarda, my son Theodor, and my
parents. Without their support, the travelling and completion of the doctorate would have been
very difficult.
All field data have been provided by helpful contractors, clients and TBM manufacturers such
as Tim E. E. Becker (previously at Lemme Tiefbau Berlin, and now at Becker Engineering
and Consultants), Impregillo S.p.A., Mannfred Khler (bb Austria), and Lutz Zur Linde
(Herrenknecht). The results presented in this thesis would not have been achieved without
feedback from the tunnelling industry.
The laboratory results and developments presented in this thesis and other papers would not
have been achieved without the help of Filip Dahl (SINTEF Rock Engineering). Filips
interest, assistance and feedback have provided me with much motivation during the four
years of the doctorate study. The development of the new apparatus Soft Ground Abrasion
Tester has been achieved by means of a collaboration between Torkjell Breivik (NTNU
Department of Rock Engineering and Geology), and Lars Langmaack and Herbert Egli
(BASF Construction Chemicals). The numerous frustrations experienced during this process
have finally resulted in a useful outcome. For tribology explanations and guidance as well as
discussions on laboratory set-ups, associate Professor Nuria Espallargas have been very
helpful.
Combining field and laboratory research approaches has been an expensive process, and the
financial support provided by the Norwegian Tunnel Society (NFF, www.tunnel.no) has been
invaluable.
2

As regards the acquisition of field and laboratory data, several students have written their
theses on relevant subjects taken from the PhD work. I am very grateful to Anders Palm (who
obtained data and samples from a project in the Middle East), Wojtek Smolen and Marek
Multan (who were following up an ongoing soft ground TBM project in Europe), and Ivar
Sletta, Andreas Hauso, Ole Fredrik Brattberg and Leon Eide for their contributions during
initial testing of the Soft Ground Abrasion Tester.
Jardar Lohne has made considerable contributions in terms of discussion and the structuring
of published papers. Fellow PhD student Wolfgang Kampel has been a great source of
information in relation to the interpretation of statistics and use of the SPSS software.
Recently qualified PhD students Vegard Olsen and Yangkyun Kim have provided excellent
advice on structuring the thesis and have helped in revealing inconsistencies in earlier drafts.
Last, but not least, I would like to show my appreciation to Professor Amund Bruland, who
has supervised this work. Without his quick decisions and assistance in obtaining funds to
finance field work, laboratory tests and other activities, there would be nothing to show for
this work but desktop and literature studies.

Contents
1

Introduction ........................................................................................................................ 6
1.1

Background.................................................................................................................. 6

1.2

Goals and Objectives of the Thesis ............................................................................. 7

1.3

Limitations and constraints on the research ................................................................ 8

1.4

Structure of the thesis .................................................................................................. 9

1.5

Published papers ........................................................................................................ 10

Materials framework ........................................................................................................ 13

2.1

General....................................................................................................................... 13

2.2

Materials understanding ............................................................................................ 14

2.2.1

Soil, soft ground and rock .................................................................................. 14

2.2.2

Tribological framework...................................................................................... 15

2.2.3

Excavation tools in soft ground TBM tunnelling............................................... 20

2.2.4

Influence of soil conditioning additives and bentonite ...................................... 22

2.3

Soft ground excavation mechanics ............................................................................ 23

2.3.1

Excavation of mixed face and boulders ............................................................. 25

2.4 State-of-the-art testing and estimation of abrasive wear for soft ground and soil TBM
applications........................................................................................................................... 28

2.4.1

LCPC Abrasivemeter ......................................................................................... 28

2.4.2

Mill tests............................................................................................................. 30

2.4.3

Dorry Abrasion Test........................................................................................... 34

2.4.4

Miller Slurry Test and reciprocating ball-on-plate test ...................................... 35

2.4.5

Rubber Wheel Test............................................................................................. 37

2.4.6

Predictive method developed by the Japanese Tunnelling Society.................... 37

2.4.7

Penn State Soil Abrasion Testing System .......................................................... 38

2.4.8

Newly-Developed Abrasion Test (NDAT)..................................................... 39

2.4.9

Turin Test........................................................................................................ 40

2.4.10

Soil Abrasion Test .......................................................................................... 41

2.4.11

Soft Ground Abrasion Tester ............................................................................. 41

2.4.12

Discussion and summary.................................................................................... 41

Research methodology and design ................................................................................... 43

3.1

Literature study.......................................................................................................... 43

3.2

Field research............................................................................................................. 44

3.2.1

General ............................................................................................................... 44

3.2.2

Tender documents and pre-investigation reports ............................................... 47

4

3.2.3

TBM data logger system .................................................................................... 47

3.2.4

Monthly and final reports ................................................................................... 47

3.2.5

Records of tool replacements ............................................................................. 47

3.2.6

Site visits ............................................................................................................ 47

3.2.7

Geographical distribution of field data............................................................... 48

3.3

3.3.1

General ............................................................................................................... 49

3.3.2

Soil Abrasion Test (SAT)............................................................................... 51

3.3.3

Soft Ground Abrasion Tester (SGAT) ............................................................... 53

3.3.4

Relating laboratory research to tool life recorded in the field............................ 55

3.4

Discussion the experiences and opinions of individuals and experts ..................... 56

3.5

Research methodology related to the published papers............................................. 57

Results and analyses......................................................................................................... 60

4.1

Field research............................................................................................................. 60

4.2

Soil Abrasion Test ................................................................................................. 63

4.3

Soft Ground Abrasion Tester..................................................................................... 68

4.4

Other estimators......................................................................................................... 71

4.4.1

Geotechnical Uniformity Index, C u ................................................................... 71

4.4.2

TBM diameter .................................................................................................... 72

4.4.3

Overburden......................................................................................................... 73

4.4.4

Mineralogical content......................................................................................... 74

4.5
5

Laboratory research ................................................................................................... 49

Comments and analyses............................................................................................. 76

Discussion ........................................................................................................................ 79
5.1

General....................................................................................................................... 79

5.2

Inconsistencies observed during research.................................................................. 81

5.3

Literature ................................................................................................................... 82

5.4

Estimation of soft ground TBM tool life ................................................................... 82

Conclusions and recommendations .................................................................................. 86

6.1

Comparison between thesis findings and the PhD plan ............................................ 86

6.2

Conclusions ............................................................................................................... 87

6.3

Further work and future perspectives ........................................................................ 88

References ........................................................................................................................ 90

Published papers............................................................................................................... 99

Introduction

1 Introduction
1.1 Background
The Norwegian University of Science and Technology (NTNU) has, together with SINTEF,
established a long tradition in the testing of engineering rock properties as a basis for
evaluating tunnelling performance.
The development of the NTNU estimation model for hard rock TBM performance and drill
and blast (D&B) tunnelling performance has been active for several decades. The first model
to address hard rock TBM tunnelling applications was published in 1976, and was followed
by updates in 1986, 1993, 1994 and 1998 (Bruland 2000). NTNU has also developed
estimation models for rock quarrying (Olsen 2009), and for drill and blast (D&B) tunnelling
costs and capacity (Zare 2007).
The current hard rock drillability database, incorporating parameters such as Drilling Rate
Index, Bit Wear Index and Cutter Life Index, contains data from more than 3000
unique rock samples from 50 countries (Dahl et al. 2012). The NTNU estimation model for
hard rock TBM tunnelling performance, time and costs (Bruland 1998b) is currently being
revised by PhD student F. Javier Macias as part of a project funded by the Research Council
of Norway (NFR) and the industry-financed research project Future Advanced Steel
Technology for Tunnelling (FAST-Tunn), in co-operation with NTNU and SINTEF.
In addition to development of the drillability indices, NTNU and SINTEF have conducted
several hard rock abrasiveness measurements using the Cerchar Abrasivity Index (CAI),
mineralogical analyses by Differential Thermal Analyse (DTA), x-ray diffraction (XRD) and
thin section analyses.
In 2004, a request to evaluate soil abrasivity properties from an on-going tunnelling project
was received by the GEMINI Centre for Underground Technology. The request was
submitted by a contractor experiencing short tool life, resulting in low TBM productivity and
increased costs. The outcome of this request was a transition to abrasivity tests on soil, based
on the existing Abrasion Value Cutter Steel (AVS) test, originally developed for hard rock
(Nilsen et al. 2006a). Following the initial publications by (Nilsen et al. (2006a); Nilsen et al.
(2006b); Nilsen et al. (2006c)), two Masters theses on this topic, (Jakobsen 2007) and
(Klemetsrud 2008), were completed at NTNU, as well as a commercial test project using 34
soil samples from the Brightwater Conveyance tunnel. Testing was initiated by Jacobs
Associates, a design company and consultancy preparing the geological baseline reports
(GBR) for the Brightwater project. In order to provide information and estimations of soil
abrasivity they wanted to include AVS test results. As the number of soil samples increased,
the name AVS test was changed into Soil Abrasion Test (SAT). The actual measurement
of abrasivity is the same for both test procedures with the exception of the sample
preparations (Nilsen et al. 2007; Jakobsen et al. 2013a).
6

Introduction

Both initial testing and the Masters theses focused on the Soil Abrasion Test (SAT) and its
ability to measure soil abrasivity. Measurement of the benefits of soil conditioning additives
in hard rock and soft ground using the Ball Mill Test has continued in parallel with SAT
testing. In the period 2006-2010 testing was carried out in the absence of any particular plan
to obtain an estimation model of tool life in soft ground TBM tunnelling. However, as the
volume of commercial test results built up over the years, the need for a more systemised
approach resulted in the present thesis. This PhD thesis on soft ground and soil abrasivity is a
result of an initiative from the GEMINI Centre of Underground Technology 1.

1.2 Goals and Objectives of the Thesis

The initial main goals of this study addressing abrasive wear on TBMs excavating soil and
soft ground were prepared as part of the 2010 PhD research plan (Jakobsen 2010). The main
goals were:
1) To find a reliable and versatile methodology for determining the potential of soil and
soft ground to cause abrasive wear on TBM tools.
2) To propose an index or simplified model for estimating tool life in connection with
soil and soft ground tunnelling projects.
In order to achieve this, three underlying objectives and sub-tasks were defined and planned:
x

x
x

To obtain TBM field data and soil samples for laboratory testing in sufficient amounts
to generate a statistical model. The term sufficient at this initial stage is taken to mean
data and corresponding samples from 5-10 projects carried out in varying ground
conditions (e.g. clay, silt and sand).
To evaluate the results of current laboratory methods against observed abrasive wear,
based on correlations between laboratory measurements and field observations.
To propose a new laboratory method which enables the testing of abrasive wear
resulting from reconstructed in-situ soil and soft ground (involving the properties soil
density, pressure, larger grain size distribution range, and the use of soil conditioning
additives). To evaluate if the proposed new laboratory tests provide a better estimate
than current (2010) procedures.

The goals proposed in 2010 have been addressed using the following problem formulations
and research questions:
1) Is the Soil Abrasion Test (SAT, after Nilsen et al. 2007) adequate as an estimator of
tool life in soft ground and soil TBM tunnelling?
a. If it is applicable, what is the extent to which it can be used or,
1
A formalized co-operation between NTNUs Department of Civil and Transport Engineering, NTNUs
Department of Geology and Rock Engineering, and SINTEF Rock Engineering.

Introduction

b. Is it necessary to adapt the SAT using other geotechnical properties in order

to achieve an estimator?
2) Is a new soil abrasivity test apparatus capable of measuring the abrasivity of in-situ
soils, and is this an advance in terms of estimating tool life for soft ground and soil
TBMs?

1.3 Limitations and constraints on the research

The following limitations and constraints were placed on the work involved in this thesis:
x
x

x
x
x
x

Work and data acquisition is limited to tunnel excavation using Earth Pressure
Balanced Shield (EPB), slurry shield and pipe-jacking machines.
TBM project data recorded in this thesis originates primarily from homogenous soil
and soft ground, without mixed face and boulders. The absence of mixed face and
boulders will most probably result in better correlations and empirical relations
between the laboratory tests and field observations presented in Chapter 4.
The systematic evaluation of the benefits of soil conditioning additives on abrasive
wear shall be limited solely to laboratory trials.
Recorded tool lives used in this thesis shall include both ripper and scraper tools. Disc
cutter life is not included here, but is included in part in Jakobsen et al. (2013a).
Tool life estimates does not take TBM operation and TBM design into account.
Discussions concerning the various soft ground and soil Tunnel Boring Machines
(TBMs), as they relate to variations in ground conditions, shall not be included in the
thesis.
There is insufficient data and samples to develop an estimator providing 100%
certainty. The study shall present estimation trends and indications, and their
associated uncertainties.
There is insufficient data and samples to distinguish between different soft ground
tools respective tool lives.

The proposal in the PhD plan (Jakobsen 2010) was to conduct field experiments using the
same TBM, and with the same contractor, in order to evaluate tool life using equipment from
different manufacturers under the same ground conditions. These experiments were not
successful for two main reasons:
x The tool manufacturers were reluctant to participate in such experiments. This
outcome was recognised in the PhD plan as a potential show-stopper.
x The contractor intended for the trial could not offer any large operative TBMs (> 4 m
diameter) during the relevant time period. This meant that there were too few tools on
the cutter head, such that positioning would most likely influence the outcome of the
experiment to a greater extent than tool quality.

Introduction

1.4 Structure of the thesis

The thesis is comprised of six chapters, together with an appendix containing the published
papers.
Chapter 1 contains background information addressing the objectives of the study and the
limitations and constraints on research. The chapter is linked to Papers 1, 2 and 3 listed in
Table 1.
Chapter 2 presents theoretical considerations supporting the research, taken from literature
searches and discussions with individuals in the tunnelling industry. More specifically, the
chapter introduces the terminology and definitions linked to the fields of soft ground, soil and
hard rock, wear theories and soil conditioning, as well as reviewing the work carried out by
other researchers and research groups to estimate abrasive wear in soft ground and soil in
connection with TBM tunnelling. The chapter is linked primarily to Papers 6 and 10, although
information concerning research obtained from literature is included in all the published
papers.
Chapter 3 discusses the research methodology, and includes general information about the
literature search, the tunnel projects included in this thesis, laboratory research and analysis
methodology. Also included are the reasons for selecting the research methodologies and a
discussion of the various benefits and shortcomings linked to the methods in question.
Furthermore, links are included to published papers dealing with the methodologies. The
chapter is linked to Papers 1, 3, 6, 7 and 9 listed in Table 1.
Chapter 4 presents the results and statistics obtained from the field and laboratory research
conducted during this study, and attempts to compare findings made in the field to those in the
laboratory. The chapter is linked to Papers 2, 3, 4, 5 6, 8 and 9.
Chapter 5 discusses the findings from both this study and the published papers, including
inconsistencies and weaknesses encountered during the research, as well as the reliability and
validity of the results. It also includes a record of checks made of the validity of a variety of
estimations against a current TBM excavation project. The chapter is linked to Papers 3, 6, 9,
11 and 12.
Chapter 6 presents the main findings of this thesis and the papers, and makes
recommendations for further work.
The published papers referred to in this thesis are presented in the Appendix. A list of these
papers is provided in Table 1.

Introduction

1.5 Published papers

This PhD work has included the publication of 12 papers. The content of these papers varies
somewhat, and the following is a brief summary of their relevance. The aim of their
publication was to promote the acquisition of additional data and samples for use in the PhD
study.
Paper 1: Classifications of properties influencing the drillability of rocks, based on the
NTNU/SINTEF test method includes a summary of the results of tests on hard rock samples
carried out at the NTNU/SINTEF drillability laboratory. The relevance of this paper is limited
to research methodology. It also demonstrates how laboratory test results organised in
databases can also provide valuable statistics for soft ground and soil abrasivity.
Paper 2: Soil Abrasion in TBM tunnelling represents a state-of-the-art statement on the
research subject at the beginning of this present PhD work. The paper also presents some
early results taken from a few sites concerning the relationships between SAT values and
recorded tool life. However, these early correlations are affected by poor field data quality.
They are based on considering several ripper tools on one cutter head spoke as a single tool,
and for this reason should not be compared directly to the correlations presented by Jakobsen
et al. (2013a) and Figure 45 and Figure 46 in this thesis. The usefulness of Paper 2 is that it
provides early notification of the research plan and subject, and serves to promote the study.
Paper 3: Review and assessment of the NTNU/SINTEF Soil Abrasion Test (SAT) for
determination of abrasiveness of soil and soft ground presents an evaluation of the SAT
test against variables such as soil mineralogy, grain shape, and recorded tool life from 9
different TBM projects. The paper also provides general presentations and discussions of the
consequences of tunnelling in abrasive soil and soft ground.
Paper 4: Tunnelling in abrasive soils review of a tunnel project in Germany presents pipe
jacking performance data from 8 projects as well as a demonstration of how SAT values
correlate with tool life data from these projects. This paper is also the first to demonstrate how
the geotechnical uniformity index (C u ) influences tool life. The paper is the direct outcome of
a joint project with a former contractor facing severe wear problems, and the results have
been very valuable for providing hands-on dissemination of the SAT test, as well as
promoting development of the SGAT test.
Paper 5: Overview of pipe-jacking performance review of tunnel projects presents the same
findings and discussions as in Paper 4, in German. It does not contain any additional
information but has served to further disseminate research in this topic.
Paper 6: Challenges of Methods and Approaches for Estimating Soil Abrasivity in Soft
Ground TBM tunnelling is a result of a paper (Paper 7) prepared for the NORDTRIB
symposium, held in Trondheim in 2012. It presents and discusses various methods and
approaches used for estimating soil abrasivity in soft ground TBM applications, and it
attempts to compare available test methods with tribological literature and industry
experience.
Paper 7: Overview of Methods and Approaches used at NTNU/SINTEF to Estimate Soil
Abrasivity in TBM tunnelling was prepared for, and presented at, the NORTRIB symposium.
The main aim of this paper was to promote feedback from mechanical engineers, tribologists,
10

Introduction

and materials scientists on the approaches used in this PhD study for estimating soil and soft
ground abrasivity.
Paper 8: Predicting the abrasivity of in-situ like soils presents some early results from, and
ideas developed by, NTNU, SINTEF and BASF which later resulted in the SGAT study
presented in Paper 9. Unfortunately, Paper 8 was not widely disseminated.
Paper 9: Development of the Soft Ground Abrasion Tester (SGAT) to predict TBM tool wear,
torque and thrust summarises how the SGAT apparatus was developed, together with
SINTEF and BASF, as a part of this PhD work. The paper also presents results and lessons
learned from observing the capabilities of the apparatus.
Paper 10: Anti-wear and anti-dust solutions for hard rock TBMs focuses mainly on hard rock
TBM tunnelling. However, the early results regarding the reduction of abrasive wear using
conditioning foam, as presented in this paper, have been applied in the soil and soft ground
studies.
Paper 11: Influence of corrosion on abrasion of steels used in TBM tunnelling represents an
outcome of participation with the NTNU/SINTEF Gemini Centre for Tribology. The paper
demonstrates how a synergetic combination of abrasion and corrosion can accelerate wear
rates on cutter tools. The paper is yet to be published, but is currently close to finalisation. An
outline of the manuscript is attached to the thesis as a part of the collection of papers in the
Appendix.
Paper 12: TBM Cutter Steel a challenge for Norwegian steel suppliers summarises the
activities which form part of the FAST-Tunn research project currently being carried out at
NTNU and SINTEF. The paper demonstrates that the research groups (Gemini Centres for
Underground Construction and Tribology) are working on topics such as steel development
for cutter tools, the improvement of existing empirical methods, and the development of
numerical methods to assess and estimate rock breaking and tool wear.

11

Introduction

Table 1 List of published papers linked to this PhD work, sorted by subject 2.
Paper
no.

Year

Title

Authors

Journal

Peer review

2012

Dahl, F., Bruland, A.,

Jakobsen, P. D.,
Nilsen, B. and Grv,
E.

Tunnelling and
Underground Space
Technology

Peer-reviewed by
two external
reviewers

2010

Classifications of properties
influencing the drillability of rocks,
based on the NTNU/SINTEF test
method
Soil Abrasion in TBM Tunnelling

Jakobsen, P. D. and
Dahl, F.

Peer-reviewed by
the organisation
committee

2013

Jakobsen, P.D.,
Bruland, A. and Dahl,
F.

2012

Review and assessment of the

NTNU/SINTEF Soil Abrasion Test
(SAT) for determination of
abrasiveness of soil and soft ground
Tunnelling in abrasive soils
review of a tunnel project in
Germany

Korean Tunnelling
Association,
Mechanised
Tunnelling
Symposium
Tunnelling and
Underground Space
Technology

Peer-reviewed by
the organisation
committee

2013

Becker, T. and
Jakobsen, P. D.

2013

Jakobsen, P. D. and
Lohne, J.

WEAR

Peer-reviewed by
to external
reviewers

2012

Jakobsen, P.D.

NORDTRIB
Proceedings,
Trondheim

2012

Overview of pipe-jacking
performance review of tunnel
projects
Challenges of Methods and
Approaches for Estimating Soil
Abrasivity in Soft Ground TBM
Tunnelling
Overview of Methods and
Approaches used at NTNU/SINTEF
to Estimate Soil Abrasivity in TBM
Tunnelling
Predicting the abrasivity of in-situ
like soils

Korean Tunnelling
Association,
Mechanised
Tunnelling
Symposium
Presented at NO-DIG
Berlin

2013

Jakobsen, P.D.,
Langmaack, L., Dahl,
F. and Breivik, T.
Jakobsen, P.D.,
Langmaack, L., Dahl,
F. and Breivik, T.

Tunnels and
Tunnelling
International
Tunnelling and
Underground Space
Technology

Peer-reviewed by
technical
conference
committee
Accepted by
magazine editor

10

2010

Langmaack, L.
Grothen, B. Jakobsen,
P.D.

World Tunnelling
Congress, Vancouver

11

NA

Influence of corrosion on abrasion

of steels used in TBM tunnelling

Under review in Rock

Mechanics and Rock
Engineering

12

2013

TBM Cutter Steel a challenge for

Norwegian steel suppliers

Espallargas, N.,
Jakobsen, P. D.,
Macias, F. J.,
Langmaack, L.
Grv, E., Jakobsen,
P.D., Kane, A.,
Hoang, H., Smading,
S., Sagen, T.B.

Peer-reviewed by
two external
reviewers
Peer-reviewed by
conference
committee
Issued to the
journal 11th of
November 2013

Tunnelling Journal

No peer-review

Development of the Soft Ground

Abrasion Tester (SGAT) to predict
TBM tool wear, torque and thrust
Anti-wear and anti-dust solutions for
hard rock TBMs

Jakobsen, P. D. and
Becker, T.

Peer-reviewed by
two external
reviewers

No peer-review

Paper 1 is related to background material, providing a description of the status of the NTNU/SINTEF drillability laboratory.
Papers 2-6 discuss tunnelling in abrasive soil and soft ground conditions, as well results from the Soil Abrasion Test.
Paper 7 summarises work carried out by other researchers on this topic.
Papers 8 and 9 present the development of the new Soft Ground Abrasion Tester developed during this Ph.D study.
Paper 10 provides background information on the use of polymer-enriched foam in TBM tunnelling.
Paper 11 provides an introduction to tribo-corrosive wear in TBM tunnelling.
Paper 12 summarises on-going research at NTNU/SINTEF into wear and tool life in TBM tunnelling.

12

Materials framework

2 Materials framework
2.1 General
The estimation and measurement of soil and soft ground abrasivity for TBM tunnelling
applications are relatively recent developments in the tunnelling industry. An initial literature
study identified various methods for predicting tool life in soft ground and soil. (Nilsen et al.
(2006b); Thuro et al. (2007); Gwildis et al. (2010)) present various methods that have been
used for measuring soft ground and soil abrasivity for TBM tunnelling applications. Initial
studies have also presented examples of TBM excavations involving exposure to abrasive
ground conditions and their impact on tunnelling performance (Nilsen et al. 2006a),
(Holzhuser and Nilsen 2006) and (Babendererde 2010). During this study, several papers on
the topic of soft ground and soil abrasivity have been published, including (Barzegari et al.
2013), (Gharahbagh et al. 2010; Gwildis et al. 2010; Rostami et al. 2012), (Khler et al. 2011)
and (Drucker 2011).
Similar studies identified during this work include the following; 1) Mr. Florian Kppl at
Herrenknecht is preparing a PhD thesis on soft ground tool life and the influence of boulders
in collaboration with the Technical University in Munich, 2) Mr. Eshan Alavi Gharahbagh has
conducted his PhD work on the subject of soil abrasivity and the identification of a reliable
soil abrasivity index, involving the development of a new test apparatus (Rostami et al. 2012),
3) Ms. Petra Drucker is conducting a PhD study on the estimation of soft ground and soil
abrasivity, also involving the development of a new testing device and 4) the Japanese Tunnel
Society has published a report concerning the estimation of tool life in connection with EPB
and slurry shield tunnelling (personal communication with Mr. Nakamura Toshiaki, Obayashi
Corporation, November 2010). Information about these studies has been communicated via
contacts made at conferences, and not as part of the literature study.
The main outcomes of the initial literature study and review of theories are as follows:
1) Definitions (e.g. soil, soft ground, abrasion, wear)
2) An overview of laboratory methods used to measure and estimate soft ground and soil
abrasivity for TBM tunnelling
3) An analysis of the limitations of existing laboratory methods for estimating soft
ground and soil abrasivity
4) The establishment of a framework for linking general tribological experience
regarding wear to applied theories developed in relation to tool life for TBMs
excavating soft ground and soil
5) An understanding of explanations and theories related to the mechanics of soft ground
and soil excavation
6) The benefits and influence of soil conditioning additives
7) An understanding of soft ground and soil excavation mechanics.

13

Materials framework

2.2 Materials understanding

2.2.1

Soil, soft ground and rock

In an engineering perspective, the terms soil, soft ground, soft rock and rock are poorly
defined. The ISRM (1978) defines the terms rock and soil on the basis of a materials uniaxial
compressive strength (UCS):
x Soil < 0.25 MPa UCS
x Soft rock <25MPa - ISRM definition: Extremely low strength, very low strength and
low strength rock
x Hard rock >25MPA - ISRM definition: Medium strength, high strength, very high
strength and extremely high strength rock
The term rock as defined by the NTNU/SINTEF Engineering Geology laboratory is a
material possessing a brittleness value of 20 < S 20 < 80. Soft rock may be defined as having
a brittleness value of 65 < S 20 <80, and hard rock 65 > S 20 .
In this thesis, as in Jakobsen et al. (2013a), the term soil is defined as a sample that can be
indented with a hand, finger or nail. Samples that cannot be indented, and which have a
brittleness value S 20 < 80 are defined as rock. Soft rock is defined as a material occupying a
transitional DUHDEHWZHHQVRLODQGURFNZKHUH8&6YDOXHV 10 MPa. The term soft
ground covers soil and soft rock, which is applicable for Soil Abrasion Test evaluation.
Figure 1 shows the relationship between UCS and Brittleness, and the corresponding
classifications developed by Dahl et al. (2012) and ISRM (1978).

14

Materials framework

Figure 1. Correlation between uniaxial compressive strength (UCS) and Brittleness (S 20 ). The
coloured boxes refer to the UCS classification ranges provided by the ISRM and corresponding ranges
in the S 20 classification (the green box includes extremely low strength, very low strength and low
strength materials according to the ISRM classification). From Dahl et al. (2012).

2.2.2

Tribological framework

Tribology is defined as the science and technology of interacting surfaces in relative motion,
and encompasses terms such as wear and abrasion. This chapter presents wear theories
relevant to steel tools interacting with soft ground and soil materials encountered during soft
ground TBM tunnelling. The theories originate from research work looking into the applied
subject TBM tunnelling, and are taken from publications addressing general wear laws
linked to tribology. The researchers involved define terms such as wear, abrasion, primary
wear, secondary wear, etc. in slightly different ways.
Frenzel and Babendererde (2011) define primary wear as the loss of material from the cutting
blade of the excavation tool. This is what is normally understood as wear in connection
with excavation tools. Secondary wear is defined as the effects on other excavation tool
components such as disc cutter hubs and bearings, and is a result of passive interaction
between the excavation tool and the tunnel face or muck (Frenzel and Babendererde 2011).
Finally, according to Frenzel and Babendererde (2011), wear on the cutter head structure is
referred to as subsequent wear.
Nilsen et al. (2006a) define primary wear as wear on excavation tools and surfaces such as
drag bits, disc cutters, scrapers and buckets/reamers. These components are designed to be
replaced at appropriate intervals. Secondary wear occurs when primary wear, as described
above, becomes excessive, leading to wear of the structures such as the cutter head, spokes

15

Materials framework

and cutter head mounting saddles, which are designed to hold the tools in place (Nilsen et al.
2006a; Nilsen et al. 2007).
Khler et al. (2011) defines the term abrasiveness as the capacity of a given type of ground to
remove material from tools, and refers to Plinninger (2007) who combines all geological
influences on tool wear into the term. However, according to Khler et al. (2011) and
Plinninger (2007), this definition is unsatisfactory, since the term does not include the
influence of non-geological factors.
In the tunnelling industry, the terms wear, abrasive wear and abrasion are commonly used to
provide a measure of tool life, even though the terms are not necessarily descriptive of wear
mechanisms. Bruland (2000) defines the term cutter life in hard rock TBM tunnelling, based
on the time the cutter tools are exposed to abrasion caused by the rock.
The tribological literature (Hutchings 1992; Stachowiak and Batchelor 2004) clearly defines
the terms abrasion and abrasive wear as forms of wear caused when a material is loaded
against particles with equal or greater hardness. In tribology, two abrasive wear models are
described; two-body abrasive wear and three-body abrasive wear.
Two-body abrasive wear occurs when particles harder than the tool, or firmly-held grits, act
like a cutting tool against solid material. Three-body abrasive wear occurs when the abrasive
particles are free to roll and slide over the surfaces of two solid materials.
In general, two-body abrasive wear causes greater wear and material removal rates than threebody abrasive wear (Hutchings 1992). The main reason for this is that in a two-body abrasion
system the interacting particles are more confined, thus increasing the contact forces between
them.
Scanning Electron Microscope (SEM) observations can be used to demonstrate whether a
worn surface has been exposed to two- or three-body abrasive wear. In two-body abrasive
wear, SEM photos commonly reveal parallel scratches, whereas three-body abrasive wear
most often produces grooves. In terms of TBM tunnelling, three-body abrasion is the result of
low confinement friction soils interacting with excavation tools, while two-body abrasive
wear occurs in hard rock tunnelling, where disc cutters interact with a hard rock face
containing fines. In the case of cohesive soils containing harder and coarser particles or mixed
face conditions, combinations of two- and three-body abrasive wear effects will be expected.
Table 2 lists the tribological terms and definitions frequently used in soft ground TBM
tunnelling.
Figure 2 shows worn-out ripper tools (primary wear) and worn-out hard facing as well as
structural damage to the cutter head structure (secondary wear). The wear on this pipe-jacking
machine lead to unscheduled maintenance, resulting in increased costs and low performance
(Jakobsen and Becker 2012). Figure 3 shows close-ups of the worn ripper tools (primary
16

Materials framework

wear), and Figure 4 shows secondary wear on the outer rim of the cutter head, causing the
actual TBM diameter to be reduced. This phenomena is also mentioned by Nilsen (Nilsen et
al. 2006a; Nilsen et al. 2007). Figure 5 shows worn out reamer tools designed to protect the
cutter head structure from wear along its rim.

Figure 2. Worn-out pipe-jacking slurry shield cutter head and tools from the Sammler Ost project in
Hamburg. The TBM diameter is 2.24 m. Red circles show fractures (secondary wear) in the cutter
head. The contractor failed to replace the excavation tools in time. Blue circles show worn-out ripper
tools resulting from abrasive wear (primary wear). Photo by Tim E. E. Becker.

Figure 3. Worn-out ripper tools abrasive wear. Photo by Tim E. E. Becker.

17

Materials framework

Figure 4. Reduced cutter head diameter due to abrasive wear to hard facing.
Photo by Tim E. E. Becker.

Figure 5. Worn-out carbide inserts and hard facing on a scraper/reamer tool due to abrasive wear.
Photo by Tim E. E. Becker.

18

Materials framework

Term

Table 2 Tribological terms relating to degradation mechanisms

identified in relation to soft ground TBM tunnelling.
Definition
Remarks
Reference

Adhesive wear

Material degradation due to high

temperatures and contact stresses

Abrasion

Degradation mechanism caused by

friction synonymous with
abrasive wear.
Ability of rock and soil to induce
abrasive wear on excavation tools
Quantitative indication of the
abrasive properties of a given
material. Equivalent term to
abrasivity
Degradation resulting from
abrasion. Tribological literature
distinguishes between two- and
three-body abrasive wear

Abrasivity
Abrasiveness

Abrasive wear

Fatigue and fretting

wear
Impact and erosive
wear 3

Primary wear
Secondary wear

Two-body abrasive
wear

Three-body abrasive
wear

Tribo-corrosion
Wear

Unlikely in soft ground

tunnelling due to low
contact stresses.

Results from minor oscillatory

movement between two solid
surfaces in motion
Material degradation due to particle
impact

Wear on TBM excavation tools

designed to be replaced
Wear on TBM parts not designed to
be replaced (cutter head structural
wear, wear on slurry lines, screw
conveyor, etc.).
Abrasive wear which occurs when
harder particles or firmly held grits
act like a cutting tool on a solid
material
Abrasive wear which occurs when
abrasive particles are free to roll
and slide over the surfaces of two
solid materials
Degradation consisting of abrasive
and corrosive wear
Degradation of materials in the
absence of a specific degradation
mechanism

(Lislerud 1997)

(Hutchings 1992)

(Lislerud 1997)
(Lislerud 1997)

Three-body abrasive wear

is likely to be the most
prevailing abrasion process
in soft ground TBM
tunnelling

(Lislerud 1997)

(Hutchings 1992)

May cause plastic

deformations (overload of
materials). Impact wear
explains the degradation of
soft ground tools
excavating boulder ground

(Hutchings 1992)

(Nilsen 2007)
(Nilsen 2007)

Likely to occur during

hard rock cutting

(Hutchings 1992)

Likely to occur in friction

soils

(Hutchings 1992)

(Espallargas et al. under

review)
(Hutchings 1992)

Impact and erosive wear and three-body abrasive wear mechanisms both involve loose particles (e.g. sand),
but the systems differ in terms of the origin of the forces acting between the particles and surfaces. In abrasive
wear the particles are pressed against the surface (Hutchings 1992).

19

Materials framework

2.2.3 Excavation tools in soft ground TBM tunnelling

A variety of excavation tools are used for tunnelling in soil and soft ground (Figure 2 to
Figure 5). Disc cutters are generally used if the rock mass exhibits a compressive strength in
excess of 20 MPa (Khaligi 2011). In mixed ground conditions, involving layers of varying
strength, and in soft ground conditions containing boulders, disc cutters are used to break the
hard faces into smaller fractions which can then be mucked out through slurry lines or the
EPB screw conveyor.
Soft ground TBMs are also fitted with disc cutters designed to penetrate concrete reception
shafts, since excavation of the concrete lining may be too hard for conventional soft ground
tools. If friction is too low, disc cutters are deployed in a fixed position which causes flat-edge
wear and blocked bearings.
All TBM parts in contact with the soil are exposed to wear. Figure 6 shows a typical slurry
shield TBM fitted with a variety of tools exposed to abrasive wear. Excavation tools such as
disc cutters, ripper and scraper tools are designed to be replaced when they become worn. The
cutterhead structure and the openings (buckets) are also exposed to wear, but these are not
replaceable TBM components.

Figure 6. Mixed condition cutter head fitted with scraper/ripper tools and disc cutters.
Photo; Herrenknecht.

20

Materials framework

Figure 7. Photos of a worn-out 17-inch disc cutter (left) and a new disc cutter (right).

Drag bits, teeth and picks are used in cohesive ground where the greater part of the excavated
material consists of clay and silt. Scraper tools are commonly used in sandy ground, and
ripper tools in coarse ground conditions, including gravels. Figure 8 shows the excavation
tools commonly used in soft ground tunnelling. In addition to these tools, reamers are
commonly used to protect the cutter head rim, and to enable an over cut.

Figure 8. Photos of a selection of soft ground tools as they relate to ground conditions (Babendererde
2010). The upper row of photos shows the different soil types involved.

The materials used in disc cutters typically consist of high-alloy tool steels and hardened and
tempered steels, with hardness values of up to 56-60 Rockwell C (HRC). Cutter tools are
selected according to the hardness of the rock, and the need for toughness and resistance to
wear. In soft ground applications, disc cutters are occasionally fitted with carbide buttons. The
21

Materials framework

main aim of these buttons is to achieve greater friction between the rolling disc and the face.
Tools used in soft ground conditions, such as drag bits, teeth and scrapers typically consist of
a steel body of high-alloy tool steel (42 Cr Mo, equal to AISI 4140) fitted with wolfram
carbides (WC) with HRC values above 70 (90% WC and 10% CO) (Smading 2013).
Materials selection guarantees an extremely hard and wear-resistant WC tool tip, and a
tougher steel body to hold the WC buttons.
2.2.4 Influence of soil conditioning additives and bentonite
In soft ground tunnelling soil conditioning additives or bentonite are used extensively in order
to achieve suitable soil rheology and sufficient face support pressure. In some cases involving
a self-stable face, soil conditioning is either not implemented, or is limited to the use of water.
Suitable soil rheology should be assessed in terms of the following; a) the mucking-out
operation (conveyor belt, train or muck pumps), b) the type of disposal area and machinery
used, c) the experience and preferences of the contractor, d) the TBM design (available
torque, length of the screw conveyor), and e) the type of soil (clay, silt, sand or gravel).
Figure 9 shows typical face support options as they relate to ground conditions. In general, to
ease the mucking-out operation, the consistency of the conditioned soil should be solid or
plastic (Langmaack 2009). Slurry face support is appropriate in highly permeable ground
conditions such as gravels, since the water and the fines in the bentonite combine to form a
filter cake between the TBM and the tunnel face. EPB is most appropriate in cohesive ground
conditions. Here, fines under pressure will contribute towards establishing a stable face.

Figure 9. Selection of face support under various ground conditions (Herrenknecht 2013).
By using appropriate soil conditioning, tunnelling using EPB face pressure can achieve higher
permeabilities (10-3).

22

Materials framework

In general, there are three types of soil conditioning additives used in connection with EPB
tunnelling; 1) foams, which fill the working chamber and reduce wear on the excavation tools
and cutterhead structure, 2) anti-clay additives which reduce clogging between the soil and
cutterhead and excavation tools, and which improve soil rheology prior to mucking-out, and
3) polymers which increase soil adhesion and make the soil less permeable (Langmaack
2009).
The use of appropriate soil conditioning additives enables EPB TBMs to excavate in nearly
all ground conditions where water pressures are less than 9 bars. In very coarse friction soils
such as gravels, a combination of foam, polymers and filler materials increase soil
permeability, thus making it possible to maintain earth pressure for face support. In sands and
silts, foams are widely used to provide adequate filling of the EPB working chamber, and clay
foam and anti-clay additives are used to reduce adhesion and clogging (Langmaack 2009).
In slurry shield TBM tunnelling, a bentonite suspension (water and bentonite) is used both to
lubricate the slurry lines for mucking out and to achieve proper face support. Bentonite is a
natural material consisting mainly of montmorillonite. Slurry shield TBM tunnelling is carried
out mainly in permeable soil and soft ground conditions, where the bentonite suspension is
used to establish a membrane called filter cake, which assists in maintaining face pressure
during excavation (Min et al. 2013).
The use of soil conditioning additives and a bentonite suspension is vital in order to make it
possible to excavate soft ground and soil. Both are injected at the TBM cutterhead, and their
use is thus very relevant to studies related to the estimation of excavation tool life during soft
ground TBM tunnelling (Peila et al. 2012). However, most research on soil conditioning
additives carried out so far has concentrated on soil rheology and face stabilisation, and water
inflow control (Langmaack 2002; Vinai et al. 2008; Thewes and Budach 2010). However,
studies of the benefits of soil conditioning in relation to TBM operation parameters, such as
torque requirements and abrasive wear reduction, are now emerging (Peila et al. 2012;
Gharahbagh et al. 2013; Jakobsen et al. 2013b).

2.3 Soft ground excavation mechanics

The mechanical cutting of geological materials, including drilling, involves the use of
indenters or drag bits. Indenters are more widely used than drag bits, even though theoretical
considerations would suggest that the former require more energy to excavate a given volume
of rock or soil (Hood and Roxborough 1992). Drag bits are exposed to shear loading which
causes shear stresses leading to tool deformation or bit breakage. It is this loading mechanism
which prevents the more widespread use of drag bits during rock excavation.
In the case hard rock, indenter bits break the rock by inducing a force acting normal to the
rock surface. A drag bit breaks the rock by inducing a force acting almost normal to the rock

23

Materials framework

surface (Hood and Roxborough 1992). Figure 10 and Figure 11 illustrate the basic principles
of rock breaking using indenters and drag bits.

Figure 10. Rock-breaking using a disc cutter (indentation bit). After Bruland (1998c).

Figure 11. Rock/soft ground excavation using a drag bit. After Hood and Roxborough (1992).

Several researchers have developed models to describe rock-breaking using indentation bits
(Hood and Roxborough 1992). Most of these models consider a situation in which a single
disc cutter is in contact with hard rock. However, this situation is not relevant to soft ground
excavation mechanisms, even though rock cutting today is mainly carried out using
indentation cutters (Hood and Roxborough 1992).
According to Verhoef (1997), theories of rock cutting using drag picks were developed by
Evans (1962) and Evans (1965). These early theories were based on observations during
experiments involving coal-breaking. The Evans (1962) and Evans (1965) model considers
the breaking of coal along a failure surface to be a highly tensile process. It demonstrates that
a crushing zone exists close to the steel tools contact with the coal, and fractures develop.
Chipping occurs between the open surface and the fracture. This mechanism agrees well with
that of Bruland (1998c), who explains how hard rock breaks using disc cutters.

24

Materials framework

Soil and soft ground materials behave more like plastic than rock, and as such the basic
theories of Evans and Bruland do not always apply to the excavation of soft ground materials.
Hard rock excavation involves a combination of compressive crushing to induce tensile
stresses in a relatively brittle rock mass. Soft ground excavation, on the other hand, involves
the ripping of a plastic or elastic material. In single-grade sand, the excavation process
involves controlling the flow of the excavated material rather than inducing high contact
stresses between the tools and the tunnel face. Thus, the process of crushing intact grains is
assumed to be less intense during soft ground and soil excavation, than for operations in hard
rock. In cases of compacted moraine materials, a combination of ripping and the induction of
tensile failure is believed to occur.
2.3.1 Excavation of mixed face and boulders
Mixed face conditions exist when the tunnel face contains sections exhibiting variable
properties (Bruland 1998e). According to Tth et al. (2013), an industry accepted standard
definition for a mixed face situation is where the uniaxial compressive strength ratio between
the weakest and strongest material in the face is less than 1/10. However, mixed face effects
such as high peak loads on excavation tools and high cutter head torque are experienced in
connection with tunnel faces exhibiting higher ratios. The Singapore Circle Line Project
encountered mixed face problems such as high peak loads, destructive wear, inconsistent
performance and face pressure. Here, the UCS ratio was greater than 1/10. As a result of cases
such as these, the term mixed face has been redefined (Tth et al. 2013) as follows;
Mixed face ground is the ground, where there are two or more geological materials
simultaneously present on the tunnel face with significant differences in material properties
that influence significantly, a) penetration rate of the TBM or b) operational parameters of
the TBM or c) support system installed behind the TBM."
Tth et al. (2013) also present an overview of recent publications documenting problems
encountered under mixed face conditions. The SMART tunnel in Kuala Lumpur is one
example where the face consisted of a limestone formation overlain by silt and sand. The
limestone was weathered into an irregular and inhomogeneous rock mass which caused a
number of difficulties for TBM operations (Klados and Yeoh 2006).
Mixed face conditions were also encountered in the Porto Metro line excavation. The geology
consisted of massive granite overlain by soil deposits with a set of variously weathering strata
in between (Tth et al. 2013). Figure 12 shows disc cutter consumption data for the various
weathering grades of the granite and soil conditioning schemes. However, it is difficult to
conclude a relationship between weathering grade and tool life from these data.

25

Materials framework

Figure 12. Disc cutter consumption during excavation in granites of various weathering states. From
the Porto Metro line S (Nilsen et al. 2007).

Jung et al. (2011) present tool life and TBM performance data from a 1.66 km tunnel section
along the Bundang subway line in Seoul, excavated under the Han river. The geology
consisted mainly of hard and soft rock with some transition zones where both the hard and
soft rock types are present. The soft rock exhibited RQD values from 20 to 30, Q values from
0.17 to 0.92 and an elastic wave velocity less than 2500 m/s. In terms of tool life, the TBM
manufacturer estimated a consumption of 237 disc cutters, while the NTNU hard rock TBM
prognosis model predicted 634. The actual consumption was 1263 cutters. According to Jung
et al. (2011), the high consumption was the result of the fact that a single set of cutters was
used for all rock types (from soft rock to hard rock), and the mud from the excavated soil
clogged the chamber and cutter tools causing uneven wear on the cutters. In terms of disc
cutter consumption and wear type, about 75% of the cutters were changed due to abrasive
wear, while the remaining 25% were unevenly worn, cracked (due to mechanical overload)
and dislocated. This example may indicate that high tool consumption represents a major
consequence of mixed face situations.
Steingrimsson et al. (2002) presented data from the hard rock Karanjukar project in Iceland,
which included sections of mixed face consisting of a variety of different basalts. Based on
experience from the Karanjukar project and NTNU drillability indices, a TBM penetration
rate estimation model was suggested, assuming that it is the hardest parts of the face which
control net penetration of the TBM. In order to estimate net penetration rate, that resulting
from the NTNU TBM prognosis model is multiplied by a correction factor K AB , which is
derived from the proportion of the hard layer divided by that of the soft layer. Tth et al.
(2013) compared the Steingrimsson model with the Singapore Circle Line Project, concluding
that this approach was not applicable to a rock/soil mixed face situation. In order to address
this, Tth et al. (2013) presented a linear multivariate data analysis approach.
Occasionally, boulders are encountered during soft ground tunnelling projects in glacial till
deposits. In some projects major problems have been encountered when boulders obstruct
26

Materials framework

tunnelling progress by causing wear and destroying cutter tools (Dowden and Robinson
2001). Having survived glacial or fluvial transport, such boulders are generally considered to
be harder than the indigenous rock (Tarkoy 2008).
Ozdemir (2008) presented initial test results obtained from a linear cutter test on the influence
of boulders on the excavation process. In an attempt to simulate a soft ground matrix
containing boulders, the test was performed on concrete blocks containing hard rock boulders.
The initial results showed how measured forces and peak forces varied according to boulder
density, although no results in terms of tool life estimation and impact wear were
forthcoming.
In open excavation mode, boulders encountered in a face can be accessed relatively easily,
broken into smaller fragments and excavated. In closed mode, the breaking of boulders is
more difficult. If the location of a boulder is anticipated, it is common to fit the cutterhead
with disc cutters specifically designed for breaking boulders. This procedure works if the
surrounding soil matrix has sufficient stiffness to lock the boulder in position during
excavation. If the boulder is relatively soft (UCS < 100 MPa), conventional soft ground tools
such as rippers will be sufficient to break it (Dowden and Robinson 2001).
In closed mode slurry shield excavation, crushers are sometimes installed in the cutter
chamber. Provided that the cutter head opening is sufficiently large, the boulder will pass
through and can be crushed into smaller fragments (< 15 cm), before being extracted through
the slurry line. However, tunnelling using EPB TBMs may encounter screw conveyor
problems. In general, fragments which are one-third of the screw conveyor diameter may be
extracted (Dowden and Robinson 2001).
If boulder fragmentation fails, contractors may have to resort to manual intervention by divers
equipped with hydraulic rock splitters. Boulders with diameters of between 0.5 to 1.5 metres
take between 10 to 90 minutes to remove. If the intervention is carried out in open mode,
grouting of the tunnel face may be required in order to achieve adequate stability.
There exists no recognised model for estimating the impact of boulders on tool life and TBM
performance. However, Dowden and Robinson (2001) suggest that geophysical methods be
applied to enhance boulder visualisation during TBM tunnelling. If the TBM operator is
aware that a boulder is located ahead of the TBM, parameters such as thrust, rotation and
performance can be reduced in order to minimise potential breakage of the TBM tools and
cutter head structure. Thorough pre-investigations using core drilling, probe drilling or
geophysical methods may be used to determine density and boulder size. If the project tender
documents provide estimates of the average amount of boulders and their estimated sizes,
contractors will be able to submit bids including appropriate TBM designs.
As mentioned previously, the influence of boulders and mixed face situations are not
considered in this study which focuses on estimating soft ground excavation tool life.
27

Materials framework

However, it is clear that both may reduce excavation tool life and net penetration considerably
due to the levels of impact wear and high peak loads they cause.

2.4 State-of-the-art testing and estimation of abrasive wear for soft ground
and soil TBM applications
The accurate simulation of wear is a recurrent problem in tribological engineering and
research (Stachowiak and Batchelor 2004). In the case of abrasive wear testing, the
tribological literature distinguishes between the following generic approaches (Hutchings
1992):
x Pin-on-disc test (e.g. Soil Abrasion Test, Dorry Abrasion Test)
x Pin on abrasive plate (e.g. Cerchar Abrasivity Index, reciprocating the pin-on-disc
and Miller slurry tests)
x Pin on abrasive drum
x Rubber wheel abrasion test
Mill tests have been used in order to measure abrasive wear on geological materials in which
the type of contact and environment is similar to those in rubber wheel abrasion tests
(Langmaack et al. 2010; Ojala et al. 2012; Rostami et al. 2012).
The selection of laboratory apparatus for the estimation of abrasive wear depends on the type
of contact and environment we wish to simulate. The following sections describe various
methods and apparatuses found during the literature search for the estimation of abrasive wear
on shield TBMs excavating soft ground and soil.
2.4.1 LCPC Abrasivemeter
The LCPC Abrasivemeter has been developed by the Laboratoires des Ponts et Chausses (the
French Laboratory for Bridges and Roads; (LCPC 1990); see Figure 13). The test apparatus
and procedure are based on a steel impeller rotating for 5 minutes in a 500g sample consisting
of crushed rock and natural or crushed soil of 4.0 6.3 mm fraction size. The impellers
dimensions are 25 mm x 50 mm x 5 mm and it is composed of a relative soft steel alloy with a
Rockwell B hardness value of between 60 and 75.
The impeller rotates at 4500 rpm. In the case of coarse soils, the 4.0 - 6.3 mm fraction can be
sieved out. The LCPC Abrasivemeter standard test procedure is not suitable for clay, silt and
sand samples (Jakobsen and Lohne 2013).

28

Materials framework

Figure 13. The LCPC test apparatus. 1) motor, 2) rotating impeller, 3) jar containing the abrasive,
4) funnel tube. (Ksling and Thuro 2010)

The impellers weight loss is measured after each test, and this value represents the abrasivity
parameter. The LCPC abrasivity coefficient (LAC) is calculated as
LAC = (m 0 m) / M
where (m 0 - m) is the weight loss of the impeller after a single test, and M is the soil or rock
materials weight (0.0005 t; (normalization 1990). The soils brittleness properties can also be
measured by the LCPC Abrasivemeter by comparing the sieve curves of the initial 4.0 - 6.3
mm sample fraction with the particle size distribution after the test.
In 2006, the Technical University of Munich started to conduct research into soil abrasivity
assessment using the LCPC abrasivemeter. The LCPC abrasivemeter is also designed to
measure hard rock abrasivity, in a manner similar to the AVS approach used at NTNU. Thuro
et al. (2006) demonstrate clearly that an increase in quartz content increases the LCPC
abrasivity coefficient, and that coarser particles (gravels) produce higher abrasivity
coefficients than finer particles (clay, silt and sand). Furthermore, Thuro et al. (2007)
compared the LCPC abrasivity coefficient with the Cerchar Abrasivity Index (CAI) in order
to utilise existing relationships between the CAI and tool life for the LCPC abrasivemeter.
Khler et al. (2011) studied the relationship between the LCPC abrasivity coefficient (LAC)
and equal quartz content (EQC). This correlation proved to be very poor, resulting in no clear
relationships in a data set consisting of 22 samples taken from the recently completed Inntal
railroad tunnel project in Austria (contract H3-4 and H8). It was also concluded based on
results from the same project that it is not possible to predict tool wear by using just one
parameter, such as the LCPC value, and that the greatest influences on wear are the grading
curve and compaction of the soil (Khler et al. 2012).
29

Materials framework

Figure 19 and Figure 44 show LCPC LAC / ABR values correlated against Los Angeles
Abrasion Values, Micro Deval values and SAT values respectively.
An evaluation of the LCPC abrasivemeter based on published data suggests that it provides a
measure of abrasive wear and impact/erosive wear resulting from the very high rotation speed
of the steel impeller.
2.4.2 Mill tests
The Nordic Ball Mill Test, the Los Angeles Abrasion Test and the Micro Deval Test are
similar in many ways. The test apparatuses and procedures consist of a rotating drum
containing a soil sample mixed with steel balls or pins. These tests have been developed to
determine road surface quality by measuring the degradation of geological materials
(Gudbjartsson and Iversen 2003).
These mill tests expose steel samples to a combination of impact and abrasive wear. However,
abrasive wear on these samples is likely to be less significant due to low contact stresses
between the steel and soil. Water and other additives can be introduced to mill tests in order to
evaluate their influence on abrasive wear (Langmaack et al. 2010) and (Gennari 2004).
Nordic Ball Mill Test
The Nordic Ball Mill Test has been used to determine the influence of soil conditioning
additives on the abrasivity properties of crushed rock and natural soil samples. The test used
in the NTNU/SINTEF laboratory is a modified version carried out without the use of steel
ribs, and with a rubber-lined drum designed to reduce wear caused by steel interacting with
steel. The test procedure is easy and straightforward. A 1500g sample, made up of grains less
than 16 mm in diameter, is exposed to 20 circular steel bits (composed of ordinary
construction steel, each 16 mm in diameter) for 5400 revolutions that are equivalent to a test
duration of 60 minutes. The rotation speed of the drum is 0.97 m/s. The weight loss among
the steel bits is measured after testing and represents the abrasivity value as defined in the
Ball Mill Test (Jakobsen and Lohne 2013) and (Klemetsrud 2008).

30

Materials framework

Figure 14. The Ball Mill Test apparatus can be used to assess the reduction in abrasivity resulting from
introducing water and/or soil conditioning additives. The test apparatus consists of a rotating drum
filled with soil (0 - 16 mm diameter) and 20 steel bits.

The addition of foam-enriched soil conditioners clearly reduce the weight loss of the steel
samples, as shown in Figure 16 (Klemetsrud 2008). As is shown in Figure 15, test results also
indicate that the abrasivity of most geological materials (rock and soil) increases up until a
certain level of moisture content is reached.

Figure 15. The influence of water content on steel wear using the Ball Mill Test for 4 different soil and
crushed rock samples (Klemetsrud 2008). The figure shows a clear tendency towards increased steel
wear as water content increases up to a certain level.

31

Materials framework

Figure 16. Reduction in abrasivity due to the introduction of a soil conditioning foam (Klemetsrud
2008). The Tonstad clay sample produced no measureable steel wear.

The contact forces between the steel and soil particles resulting from gravity and the tumbling
of the drum are relatively low. However, this low degree of contact is not representative of
real situations and cannot simulate the relatively high thrust values, torque ripping and
scraping of the soil which occurs during TBM excavation operations (Jakobsen and Lohne
2013).
Los Angeles Abrasion Test
The Los Angeles Test apparatus (Figure 17) consists of a cylinder with a rotation speed of
between 30-33 rpm. The test duration may be between 100 and 500 revolutions. The steel
samples comprise between six and twelve 47 mm diameter balls, each ball weighing between
390 and 420g (Ugur et al. 2010).
The quantity of soil or aggregates used during a single test is 5000g, with a sample size >
1.6mm diameter. In order to determine road aggregate properties, soil or aggregate
degradation indices are measured after first 100, and then again after 500, revolutions (Ugur
et al. 2010).

32

Materials framework

Figure 17. The Los Angeles Abrasion Testing apparatus (Ugur et al. 2010).
The rotating drum contains 4 paddles for lifting the soil and steel balls.

As an alternative to evaluating the degradation of soil and aggregates, the Los Angeles
Abrasion Test can also be used to measure weight loss incurred by the steel balls for soft
ground abrasivity applications (Nilsen et al. 2006b).
Micro Deval Test
The Micro Deval test (see Figure 18) is commonly used in Canada to determine abrasion
caused by aggregates. The test principle is to place an aggregate sample together with a fixed
volume of water in a jar mill. The jar mill contains steel balls similar to the Los Angeles
Abrasion Test (Fowler et al. 2006).
The aggregate sample, weighing 1500g, is soaked in two litres of water for one hour prior to
testing. Following preparation, the sample is placed in the Micro Deval jar mill together with
5000g of steel balls, each 9.5 mm in diameter. The drum is sealed and rotates at 100 rpm. The
test duration is dependent on the grain size curve of the aggregate, and varies between 95 and
120 minutes (Fowler et al. 2006).
Material degradation is measured by sieving the aggregates after testing. Figure 19 shows the
relationship between the LCPC abrasion value and the Micro Deval Value.

Figure 18. The Micro Deval Test apparatus (left) and schematic set-up (right)
(Serveal-Instruments 2013)

33

Materials framework

In addition to the degradation of soil and aggregates, the Micro Deval test can be used to
measure weight loss incurred by the steel balls for soft ground abrasivity applications (Nilsen
et al. 2006b).

Figure 19. Correlations between LCPC ABR values and corresponding Los Angeles Abrasion Values
and Micro Deval Values. The data are obtained from the LCPC (2006).

2.4.3 Dorry Abrasion Test

The Dorry Abrasion Test (Figure 20) employs the resistance of aggregates to surface wear by
abrasion induced by a rotating steel plate. The test measures the volumetric loss of the
aggregate and provides a value for the aggregates resistance to abrasion (Nilsen et al. 2006b).
The test is also referred to as the AAV test and is described in the UK Standard EN 1097-8
BS 912. In the test, two samples of the soil or crushed rock material material, are prepared
and placed in rectangular moulds. The steel samples are mounted against a circular rotating
steel wheel in diametrically opposing clamps (Klemetsrud 2008). The aggregate abrasion
value (AAV), is given by the percentage of weight loss incurred by the samples.

34

Materials framework

Figure 20. The Dorry Abrasion Test (Zeal-International 2013)

2.4.4

Miller Slurry Test and reciprocating ball-on-plate test

The development of the Miller test originates from vertical excavation operations carried out
in the petroleum industry (Nilsen et al. 2006b; Rostami et al. 2012) (ASTM 2001), which
have later been applied to soft ground tunnelling projects in order to estimate wear (Gwildis et
al. 2010).
The test apparatus (Figure 21) consists of a tray-shaped sample container filled with a test
slurry such as soil mixed with bentonite. A standard steel block moves back and forth with a
fixed normal force (22.24 N). The test duration is 6 hours, and the weight loss of the steel
block provides the so-called Miller Number. The test is also able to measure Slurry Abrasion
Response (SAR), which is determined by testing various types of steel block on the same
slurry suspension (Nilsen et al. 2006b).

Figure 21. The Miller Slurry Test Machine (Nilsen et al. 2006b)

35

Materials framework

Gwildis et al. (2010) collected TBM performance data, tool wear data and geotechnical data
from the Brightwater Conveyance project carried out in Seattle. Based on these data, a
Normalised Wear Parameter (NWP) is proposed, given by (108 * W) / L, where W represents
wear (in mm) and L is the travel length of the tool (also in mm).
The NWP is then correlated against SAT values, Miller Slurry Test Numbers and quartz
content, and compared with the average energy consumption of the TBM (MJ/m3). The
analysis indicated that the driving factors behind tool wear linked to soft ground TBMs are a
combination of high values for the abrasiveness descriptors (e.g. Miller Slurry Test Number,
SAT or quartz content) and cutter head energy consumption, thus the soil strength.
The Tribology Gemini Centre at NTNU and SINTEF has been conducting tribological tests
similar to the Miller Slurry Test, in order to determine abrasive and corrosive wear. The tests
have been run both on reciprocating ball-on-plate apparatus (Figure 22) and by using the
Rubber Wheel Test (Figure 23).
The reciprocating ball-on-plate test consists of a 6 mm diameter steel ball moving back and
forth with a stroke length of 10 mm, either across a rock sample or in a slurry environment.
The steel ball has a normal force of 5 N (Espallargas et al. under review).
The degradation of the steel ball is measured using an SEM microscope, and is used to
provide a qualitative evaluation of the wear mechanism (abrasive wear, corrosive wear or a
combination known as tribo-corrosion).

Figure 22. Reciprocating ball-on-plate test (Espallargas et al. under review).

Evaluations based on the Miller Slurry Test and the reciprocating ball-on-plate test indicate
that they are both able to measure abrasive wear on particles, and tribo-corrosive wear when
liquid and additives are introduced (Grdal et al. 2012).
36

Materials framework

2.4.5

Rubber Wheel Test

The Rubber Wheel Test consists of a container holding slurry (a chemical environment
including soil) and a rubber wheel which lifts the slurry and exposes it to a steel sample
applying a force of 220 N (Figure 23) (Espallargas et al. under review). The rubber wheel has
a linear speed of 2 m/s, equivalent to 200 rpm.
The degradation of the steel sample is measured using an SEM microscope in a similar
manner to that used for the reciprocating ball-on-plate test.

Figure 23. Schematic diagram of the Rubber Wheel Test (Espallargas et al. under review).

Evaluation of the rubber wheel test shows that it is able to measure abrasive wear particles
and tribo-corrosive wear on the introduction of liquid and additives (Grdal et al. 2012).
2.4.6

Predictive method developed by the Japanese Tunnelling Society

The Japanese Tunnelling Society has developed a formula for the estimation of ripper tool
wear (Nakamura 2011). Tool wear (LVexpressed by the following equation:
    

Equation 1

where
K is a coefficient of wear (mm/km)
D is the TBM diameter (mm)
N is the cutter head rpm
L is the tunnel length (km) and
V is the TBM performance (mm/min.)

37

Materials framework

The wear coefficient K is the most problematic factor included in the formula, and the Society
supplies no information about how the coefficient is measured other than by the application of
experience data (see Table 3). According to Nakamura (2011), Japanese contractors use their
own, empirically-derived, wear coefficients.
Table 3. Coefficient of wear prepared by the Japanese Tunnelling Society (Nakamura 2011)

Alluvial clay
Diluvial clay
Sand
Gravel sand
2.4.7

EPB TBM

Slurry shield TBM

3.0 3.5 * 10-3 mm/km

1.7 2.4 * 10-3 mm/km

8.0 15.9 * 10-3 mm/km

5.0 11.3 * 10-3 mm/km

-3

4.8 15.2 * 10-3 mm/km

-3

9.8 23.0 * 10-3 mm/km

10.6 19.7 * 10 mm/km

15.9 29.6 * 10 mm/km

Penn State Soil Abrasion Testing System

The Penn State University research group was the first to develop and describe a dedicated
abrasion test for in-situ (and similar) soils (The Penn State Soil Abrasion Testing System
(PSAI)) (Gharahbagh et al. 2010; Gharahbagh et al. 2011; Rostami et al. 2012; Gharahbagh et
al. 2013) (Figure 24). The apparatus consists of a rotating blade located at a fixed position
(depth) within the soil sample (Figure 25). It provides an opportunity to evaluate the influence
of water content variations and rotation speeds on a soil sample. The consolidation of the soil
is not controlled and the excavation tool does not penetrate fresh soil material during testing.
The PSAI testing system is capable to test soils consisting up to cobble dimensions, at 0 to 10
bars pressure.
The research conducted at Penn State University shows that overpressure shows no significant
influence on the rate of wear of the propeller. However, it clearly demonstrates that water
content, and thus the compactibility of the soil, influences the rate of wear, and that finergrained soils produce less wear than coarser particles.

38

Materials framework

Figure 24. Illustration of the Penn State Abrasion Testing System (Gharahbagh et al. 2011).

Figure 25. Examples of the Penn State Abrasion Testing Systems paddle/propeller
(Gharahbagh et al. 2011).

An evaluation of the literature concerning the Penn State Soil Abrasion System apparatus
suggests that it is able to measure abrasive wear, impact/erosive wear (when coarse particles
are introduced), and tribo-corrosive wear on the introduction of liquid and additives.
2.4.8

Newly-Developed Abrasion Test (NDAT)

Barzegari et al. (2013) have developed a test called the Newly-Developed Abrasion Test
(NDAT) (Figure 26). The test apparatus consists of a rotating steel plate exposed to samples
of soil or crushed rock. The apparatus can carry out tests under pressure and can also test the
influence of soil conditioning additives.

39

Materials framework

Figure 26. Illustration of the NDAT device (Barzegari et al. 2013).

An evaluation of the literature concerning the NDAT apparatus demonstrates that it can
measure abrasive wear particles and tribo-corrosive wear on the introduction of liquid and
additives. Due to the large contact area between the rotating plate and the soil, the influence
of impacts is expected to be less than that observed for tests such as the Turin Test (see
below) and the Penn State Soil Abrasion System.
2.4.9

Turin Test

The University of Turin (Politecnico Torino Tunnelling and Underground Space Center and
Laboratory) has collaborated with UTT Mapei to develop a laboratory apparatus to carry out
comparative wear tests on conditioned soils (Peila et al. 2012). The test comprises a tank
containing a soil sample and a circular metal disc exposed to wear (Figure 27). The soil
sample is compressed with 2 kPa confinement pressure both prior to and during testing. The
tool is maintained in a fixed position, and as such no penetration is involved.

40

Materials framework

Figure 27. Illustrations of the Turin Test device (Peila et al. 2012).

An evaluation of the literature concerning the Turin Test apparatus demonstrates that it can
measure abrasive wear, impact/erosive wear (when coarse particles are introduced) and tribocorrosive wear on the introduction of liquid and additives.
2.4.10 Soil Abrasion Test
For information about the NTNU/SINTEF Soil Abrasion Test, the reader is referred to
Section 4.2 and Papers 2, 3, 4, 5, 6 and 7 (Jakobsen and Dahl 2010; Jakobsen 2012; Jakobsen
and Becker 2012; Becker and Jakobsen 2013; Jakobsen and Lohne 2013; Jakobsen et al.
2013a).
2.4.11 Soft Ground Abrasion Tester
For information about the Soft Ground Abrasion Tester, the reader is referred to Section 4.3
and Papers 6, 8 and 9 (Jakobsen et al. 2012; Jakobsen and Lohne 2013; Jakobsen et al.
2013b).
2.4.12 Discussion and summary
Several of the laboratory procedures and methods used to estimate soil abrasivity are derived
from hard rock abrasivity procedures and road aggregate testing. In recent years (since 2010)
a change has occurred in the sense that researchers are now attempting to design new test
devices dedicated to the measurement of wear in connection with applications related to
complex soils and soft ground. At present, there are several tests available, and there are in
fact more tests and apparatuses than there are results and predictive models related to each
respective test. This corresponds with general tribological findings documented in (Meng and
Ludema 1995), who conclude that due to the complexities involved in estimating wear, small
differences in wear tests and resulting estimates, such as for materials life, often spawn new
tests and resulting empirical relationships.
41

Materials framework

The Cerchar Abrasivity Index (CAI) is one of the most common approaches used for
measuring rock abrasivity (Deketh 1995; Plinninger et al. 2003; Rostami et al. 2005;
Michalakopoulos et al. 2006; Alber 2008; Ksling and Thuro 2010)). The test apparatus is a
type of pin-on-plate device, consisting of a steel pin which scratches the rock sample surface
over a length of 1 cm (see Figure 28).

Figure 28. Diagram illustrating the Cerchar Abrasivity Index (CAI) apparatus (after West (1989)) .
1) load, 2) pin guidance, 3) steel pin, 4) rock sample, 5) vice sled, 6) hand crank.
After (Ksling and Thuro 2010).

Even though the Cerchar Abrasivity apparatus is one of the most commonly used approaches
for measuring rock abrasivity, the initial literature study failed to reveal any CAI results
relating to soft ground and soil fragments. The most likely reason for this is that the CAI
apparatus is designed to carry out abrasivity tests on a rock samples measuring typically 5 x 5
cm. It is not impossible to run tests on soil samples as small as this. However, it is possible
using this apparatus to carry out measurements on coarse soil particles such as gravel and
stone. At the same time, assessments of the abrasivity properties of gravel and cobbles are
also absent from most of the other approaches discussed in the literature study. An alternative
possibility would be to combine fine soil particles (silt and sand) with coarser particles
(gravel) in a Hoek cell as suggested by Alber (2008). This might provide an opportunity to
evaluate the soft ground abrasivity of an in-situ (or similar) material using the Cerchar
apparatus.

42

Research methodology and design

3 Research methodology and design

This chapter provides a description of the methodological approaches used to acquire data, as
well as general information about the data analysis methods used in this PhD work. The data
acquired consists of empirical findings from the literature review, field data and laboratory
data, as well as experience-sharing with contractors, clients and TBM manufacturers.
Throughout this Chapter the terms validity and reliability are used. Validity is an
expression of the extent to which an indicator, such as a value obtained in the laboratory,
represents an adequate measure of what it is intended to measure. Reliability is an expression
of the extent to which repeated use of the same indicator would produce identical results (i.e.,
a measure of reproducibility) (Samset 2008). Figure 29 is a diagram illustrating high validity
and reliability versus low validity and reliability.

Figure 29. Diagram showing high validity and low reliability (left) and low validity and high
reliability (right) (Samset 2008).

The general aim of this research is to find empirical relationships between laboratory test
results and actual tool life observations. Hopefully, the research and its proposed
methodology will continue as more data and experience accumulate.

3.1 Literature study

Initially as part of this PhD study, a literature search was carried out with the aim of
establishing a knowledge platform obtained from other researchers work (Jakobsen 2010). A
variety of search words and strings were input to several online databases in order to obtain
literature sources and background information. The process of searching and reading literature
in journals, books and magazines has been a continuous process throughout this PhD study.
References found during the literature search have also been used to track down additional
information about specific subjects. Most of the references used in this thesis are available on
the internet via NTNUs online library service, while other sources have been purchased.
43

Research methodology and design

Publications found during the literature study consist mainly of papers published in
international journals and scientific monographs which employ peer-review processes. The
information in these sources is thus quality-assured by journal reviewers and editors.
However, the tunnelling industry also shares its experience and knowledge by means of
channels such as conference proceedings and industry journals and magazines. Information
sources of this type have been used in this thesis, as they provide documentation of hands-on
experience, and may themselves also act as sources of further research and contacts. The
insights into hands-on experience obtained from such material have been assessed more
critically in cases where peer review quality assurance is lacking.
As the PhD study progressed, other studies on soft ground and soil abrasivity were being
carried out elsewhere. References to some of these studies were found entirely by accident.
This suggests that it is likely that some relevant sources remain undiscovered.
The literature has been evaluated in terms of; 1) its relevance to the research work, 2) validity
and 3) reliability. Criterion 1) relevance to my work has been subdivided into sub-criteria
such as general background information (facts about TBMs, soil conditioning, tribology,
etc.), laboratory methods for evaluating soil abrasivity, and case studies on soft ground TBM
tunnelling. This Chapter provides a summary of the findings derived from the literature
search.
The process of repeating such a literature research on the same subject would involve an
equivalent utilisation of the same search words and online databases as presented here. A
suggested modification would be to improve the organisation of the literature collected by
sorting it according to relevance, validity and reliability.

3.2 Field research

3.2.1 General
In the TBM tunnelling industry, data showing a contractors performance and consumption
are often treated confidentially. This has made data acquisition for this thesis a challenging
process. In order to obtain sufficient amounts of data, it was necessary to establish
confidentiality agreements with contractors. The field data acquired are stored at NTNU and
can be made available to other researchers under certain conditions. Table 4 provides a
summary of the tunnel sites included in this PhD study. The information has been simplified
in order to safeguard confidentiality issues, and reduce the risk of third parties linking a
project number to an actual site4.
The projects have been selected on the basis of the following;
x a variety of ground conditions
4
The normalised presentations of tool life from the various tunnels (Figure 39 - Figure 45) do not contain
approximations.

44

Research methodology and design

o soil lithology in order to include data from clay, silt, sand and gravel
o overburden
a variety of excavation methods (EPB and slurry face support)
variation in TBM diameter
a variety of contractors and clients.

x
x
x

The field data are derived mainly from projects involving the use of small diameter TBMs (<
4 m). Moreover, the ground conditions in these projects consist mainly of homogenous soils.
Ground involving mixed face and boulders are not included. Tunnels driven using smalldiameter TBMs are usually excavated by small contractors. These companies are easier to
approach in terms of data and knowledge sharing. Larger contractors, which excavate both
small and large tunnels, generally operate with a larger organisational structure, and obtaining
decisions regarding data sharing is more problematic.
The expansion of data sets by including a larger number of projects involving large crosssection tunnels (> 10 m), combined with the inclusion of mixed face and boulder condition
cases, would be possible if time was not a constraint on such work. Table 4 and Figure 30
provide summaries of the sources of data and general information related to the field work.
Table 4. Tunnel sites included in this PhD study.
Project number

Face
support

Approximate
diameter [m]

Approximate
tunnel length
included in the
study [m]

Region

1
2
3
4

Slurry
Slurry
Slurry
EPB

5.5
13
3.125
6.2

5500
5700
3300
3 x 3000

North America
Central Europe
Middle-East
Middle-East

Slurry

3.04

375

Central Europe

Slurry

2.2

1200

Central Europe

Slurry

2.2

900

Central Europe

Slurry

140

Central Europe

Slurry

200

Central Europe

10

Slurry

3.1

1200

Middle-East

11
12
13
14
15
16

EPB
EPB
EPB
EPB
EPB
EPB

6.2
6
6.5
9.5
6
2

7500
2 x 2500
2 x 3000
6000
7500
700

South Europe
North America
South Europe
South America
South Europe
South-east
Europe

Site visit

Permission to visit sites and collect tool life data was obtained via direct contact with
contractors and machine manufacturers. The actual collection of data was carried out by the

45

Research methodology and design

author, Masters students in the process of their theses (one project), and by the contractors
operating the TBMs.
Tool replacement rates are determined based on the individual contractors own criteria and
empirical experience in relation to tool life. Some contractors attempt to excavate using worn
out tools in order to reach a shaft, while others employ a systematic approach to the
replacement and inspection of tools involving scheduled hyperbaric interventions. The various
criteria adopted for replacing and inspecting tools influence the information quality of tool life
records.
The field data vary in terms of both quality and extent. These variations are the result of
inconsistencies in agreements and relationships between myself and the contractors. Some
contractors and clients are willing to share data only after the completion of projects, making
soil sampling very difficult. Such data have been accepted and used, because they are
valuable for ranking purposes (for parameters such as tool life or distribution of tool life).
Field work has not included an evaluation of the various tool types (design and material
quality). Nor does it encompass the influence of soil conditioning additives on abrasive wear.
The reason for not including these factors is the relatively long tool life associated with
tunnelling in soft ground and soil. This makes such tests both time-consuming and costly (see
Section 1.2). An evaluation of soil conditioning additives is included as part of the laboratory
work (see Section 3.3).
Detailed information about TBM performance, including sheets containing detailed data on
tool replacements, have been obtained for eight projects, together with soil samples and preinvestigation reports and results. For the remaining projects, soil samples and tool life data is
less detailed.
Excavation tool life consumption, as described in this thesis, has been recorded on the basis of
tool replacement logs obtained from the tunnel projects. Tool life consumption logs have been
collected both during site visits and from contractors who participated in this study. For
several of the tunnels listed in Table 4, tool changes were carried out only after the TBM had
finished the tunnel, or had entered an intermediate shaft. For projects 1, 2, 4, 8 and 10
replacements were carried out along the tunnel drive, and in these cases tool life has been
calculated and expressed as instantaneous tool consumption (Bruland 1998b). The
instantaneous tool life parameter includes varying tool life (e.g. sm3/pcs) for the various
sections along the tunnel in question. The most recent tool replacement carried out at a given
tool position is also taken into account when calculating instantaneous tool life, meaning that
the tool life at each position on the TBM cutter head is calculated.
Sections 3.2.2-3.2.6 provide detailed descriptions of various sources of information used
during the field research, while Section 3.2.7 discusses the geographical distribution of field
data locations.
46

Research methodology and design

3.2.2

Tender documents and pre-investigation reports

Pre-investigation results from tender documents are available for projects 4, 5, 6, 7, 8, 9, 10

and in part for project 11. The data typically consist of information suggested by Deutsche
Vereiningung fr Wasserwirtshcaft (2008) and include the following;
x Sieve curves of the soil
x Contaminated ground
x Longitudinal drawings and maps
x Shear strength on a limited number of samples and/or SPT number
x For project 11 the tender included abrasivity test results using the Soil Abrasion
Test
Pre-investigation results provide important information for building a database of the various
soil parameters influences on tool life. This information is valid, but it must be kept in mind
that disputes may arise in situations where contractors disagree with clients regarding the
information contained in pre-investigation reports.
3.2.3 TBM data logger system
Modern TBMs are equipped with data logger system that records a number of machine
parameters. The main parameters collected for this thesis are thrust, torque, rpm, power
consumption (A), station/chainage and penetration rate. Data log files are available for
projects 1, 5, 6, 7, 8 and 9.
3.2.4 Monthly and final reports
Monthly reports are generally compiled for tunnel projects as a means of informing clients
about progress, problems, and other issues that may impact on costs and time. Such reports
have been obtained for projects 1, 3, 5, 6, 7, 8, 9 and 16. These reports do not reveal any
details about tool life, but they do provide good indications as to overall production and time
consumption related to a variety of operations. They may also contain data on TBM downtime
due to interventions and the replacement of excavation tools.
3.2.5

Records of tool replacements

Contractors make records and log excavation tool consumption. The quality of such records is
dependent on contractor experience and tool consumption. In projects where tool
consumption is low (e.g. involving the replacement of only a few tools after finalising a
drive), such records do not exist. In contrast, for projects where consumption is high, tool
records are generally very detailed. The unit adopted for tool life in this study is excavated
in-situ solid cubic metres per tool (sm3/t).
3.2.6 Site visits
47

Research methodology and design

Site visits have varied from a few days to weeks in duration. Contractors operating the TBMs
at these sites have been welcoming, and have shared their data and experience. They have also
provided assistance in collecting soil samples for laboratory tests. In addition to quantitative
data, the site visits have provided an excellent source of personal feedback and an opportunity
for me to disseminate my research.
3.2.7 Geographical distribution of field data
The geographical variation exhibited by the field data involves 11 countries from NorthAmerica, Europe and Middle-East (Figure 30). The study has been unsuccessful in obtaining
data from projects in the booming Asian market. However, soil samples have been obtained
from a few projects in East-Asia, although TBM performance data and tool replacement
records are lacking for these projects. Agreement was reached with a Chinese contractor to
exchange data and results, but due to difficulties in shipping the soil samples to Norway, the
only data obtained were TBM performance figures. No tool life data were obtained. A
combination of linguistic and cultural difficulties, customs administration procedures and
geographical distance are the likely reasons for this. In Africa, mechanised tunnelling is used
mainly in mining projects. A recently completed southern African TBM project, which
included a section of clay, was approached during this study, but no data exchange took place.
The lack of relevant TBM field work results from Scandinavia is related to the predominance
of hard rock conditions in this region. Some tunnels in Malm and Copenhagen have been
excavated using EPB TBMs. However tunnelling in rocks such as limestone containing flint
is not considered relevant to this study. In summary, there are no indications that the data are
influenced by geographical location. Thus, regardless of location, all sources are valuable as
providers of more empirical data.

48

Research methodology and design

Figure 30. Countries with soft ground TBM tunnelling projects included in the field research part of
this study. (Excel template from Choropleth Maps 5).

3.3 Laboratory research

3.3.1

General

Several laboratory tests are included in this study, as follows;

x
x

x
x
x

Testing carried out by myself (mainly using the Soft Ground Abrasion Tester (SGAT)
and Soil Abrasion Test (SAT))
Testing carried out by Masters students employed as research assistants (SGAT). The
testing has been supervised by myself and laboratory personnel at SINTEF Geology
and Rock Engineering.
Testing carried out by Masters students in connection with their theses (SGAT and
SAT). The theses have been carried out under my supervision.
Purchased testing by which NTNU pay for laboratory services provided by SINTEF
(SAT, sieve curves, quartz content by DTA, and x-ray diffraction)
Ordinary commercial laboratory testing by which a SINTEF client purchases
laboratory test results, and makes them available for further research (mainly SAT
tests).

This approach has been quite useful because results have been obtained via several sources,
thus increasing the amount of test results available to this thesis. All of the SAT, SGAT and
XRD laboratory results have been obtained by means of close co-operation between myself,
5

http://www.clearlyandsimply.com/clearly_and_simply/2009/06/choropleth-maps-with-excel.html

49

Research methodology and design

students, and SINTEF laboratory personnel. In this way, the reliability of the results has been
assured. LCPC results obtained from other sources (see Table 5) were provided by a TBM
project owner who was paying for tests carried out at another laboratory. The results have
been found to be valid, since the laboratory belongs to a university known for its high quality
standards.
Table 5. Laboratory test results included in this study, and personnel who carried out the tests.
Source
SAT SGAT 6 XRD LCPC
20
3
17
PDJ
9
6
PDJ MSc
284
17
SINTEF
2
2
Employed students
2
2
MSc students
21
Other sources

Natural variations in soil materials influence laboratory test results. Figure 31 shows the
difference in grain size distribution between two batches of material taken from the same
natural soil deposit. Such variation has been a source of error, especially in the case of SGAT
testing which permits the testing of grain sizes from 0-10 mm. Such variations are less
influential in the case of SAT tests because these are carried out on a more limited range of
grain sizes (0-4 mm).

Figure 31. Variation in grain size distribution for two batches of a material used for calibration testing.

Each soil type has undergone to evaluate the influence of variations in water content, density and soil
conditioning additives.

50

Research methodology and design

Some of the samples obtained from TBM projects originate from excavated material and
contain soil conditioning additive residues. The influence of these residues is discussed in
Section 4.3.
3.3.2 Soil Abrasion Test (SAT)
At the start of this PhD study, the published description of the Soil Abrasion Test (SAT)
was accompanied by standard test procedure (Nilsen et al. 2007). The SAT procedure and
apparatus are derived directly from the Abrasion Value Cutter Steel (AVS) test (Nilsen et al.
2006c). The procedure has not been changed in any major way during this PhD study.
However, the influence of not using grains over 4 mm, thus the possible change of the
mineralogical content of the soil samples is discussed by Jakobsen et al. (2013). The main
sources of error linked to the SAT test are:
x Control of flow rate of abrasives (manually controlled)
x Alignment of the SAT steel piece on the rotating disc
x Re-grinding of SAT steel test pieces.
No quantification of sources of error has been carried out for the SAT test as part of this
thesis. However, this can be achieved by processes such as running SAT tests on the same
abrasive at various flow rates, or by making different operators run the same tests.
The SAT procedure provides reliable test results (see Figure 32). A comparison of the
validity, or robustness, of the test results is presented in Paper 3 and Section 4.2.

Figure 32. Reliability of the AVS/SAT test obtained by testing crushed limestone, quartzite and
trondjemite. The x-axis (1-3) refers to three different years when tests were carried out.

51

Research methodology and design

Various references such as (RETC 2007), (Langmaack 2011), (Impregilo 2011),

(Herrenknecht 2013) and (Robbins 2013), which discuss the SAT, have directed criticism at
the simplified test procedure. More specifically, the SAT tests the abrasive properties of
cohesionless loose and dry soil materials, as shown in Figure 33. In fact, the real interaction
between soft ground excavation tools and soil materials is more complex because factors such
as compaction, the influence of water, and the use of soil conditioning additives all impact on
abrasive wear potential (see Figure 34). In order to develop a test procedure with higher levels
of validity, and capable of testing in-situ (or similar) soils, the Soft Ground Abrasion Tester
(SGAT) was developed as part of this PhD study.

Figure 33. Interaction between dry and loose soil particles and the steel bit during an SAT test
(Photo by Filip Dahl).

Figure 34. A soft ground face consisting of soft claystone (UCS < 2 MPa)
and gypsum (UCS < 5 MPa).

52

Research methodology and design

3.3.3

Soft Ground Abrasion Tester (SGAT)

The development of the Soft Ground Abrasion Tester (SGAT) was carried out in order enable
abrasion testing on in-situ (and similar) soils, and to increase the validity of tool life testing
for soft ground and soils.
A detailed test procedure for commercial use of the SGAT is yet to be developed. The test
apparatus is designed to evaluate the influence of several variables on abrasive wear and
torque. For this reason, the test procedure must be decided prior to the testing of a new batch
of sample material, on the basis of what the results are intended to show. A generalised
preliminary test procedure is presented in (Jakobsen et al. 2013b). The main sources of error
linked to the SGAT test are as follows:
x Precise control of water content and density distribution along the sample
x Re-use of tools which are becoming deformed
x Preparation of steel tool prior to first use
x Variations in soil sample properties such as grain size distribution and mineralogy
x Inconsistencies between the data logger and real world data.
The following tests and measurements have been carried out in order to assess and quantify
the reliability of, and possible sources of error linked to, the SGAT tester;
x Tests on the same abrasive (soil sample) with different grain size distributions. This
measurement also includes re-use of the abrasive
x A comparison of results obtained from running tests on a new tool with those using a
used tool (after 5, 10 and 20 tests)
x Manual control of thrust and torque according to measurements using a scale and
torque wrench (see Figure 35 and Figure 36).
Reliability testing of the SGAT concluded that variations in the grain size distribution of the
abrasive influence measured wear. Re-use of the abrasive should not be carried out because
crushing of the sample material introduces more fines, which in turn promote increased
potential cohesion and wear. The findings shown in Figure 36 demonstrate that there are no
inconsistencies between the thrust and torque values measured in the SGAT apparatus. The
decrease in torque over time is due to reduced resistance in the gear mechanism as it reaches
approximately 40C. This normally occurs after 20-30 minutes of operation at room
temperature (see Figure 37).

53

Research methodology and design

Figure 35. Control of thrust and torque on the SGAT apparatus.

Figure 36. Relationships between data log values and measured values
for torque and thrust using the SGAT apparatus.

54

Research methodology and design

Figure 37. Development of torque plotted against time at various rotations

without soil samples using the SGAT apparatus.

3.3.4

Relating laboratory research to tool life recorded in the field

Tool life data from the TBM projects have been correlated with measured laboratory values in
order to establish a statistical relationship between actual tool life and laboratory data. Thus,
an important part of this thesis is to establish correlations between variables such as soft
ground TBM tool life recorded in the field and various geological and geotechnical
parameters. Tool life is expressed in the units solid cubic metre excavated soil or soft rock
per excavation tool (sm3/pcs). This unit is adopted mainly with the aim of including a variety
of TBM diameters in the same correlation. Bruland (1998b) starting point for the relationship
between tool life and the Cutter Life Index is based on TBM boring hours. However, the
boring hours parameter is not known for all projects used in this thesis (Table 4). As a result
this parameter has been rejected for the purposes of this thesis, since the resulting tool life
consumption data would be less based on less data. However, a correlation between the
parameters boring hour tool life and volume tool life is provided in Figure 38 for those
projects for which TBM operation and tool life data are available.
Correlation coefficient and statistical significance
The R2-value (correlation coefficient) is commonly used to demonstrate the relationship
between two variables. R2 provides an expression of the proportion of the total variation of
one of the variables that can be accounted for, or explained by a relationship with a random
value in the other variable (Walpole et al. 1998). An R2 value of 0.5 (50%) shows that the
total variation of values of variable 1 in a given sample is accounted for by a relationship to
values of variable 2. Thus, R2 demonstrates how well a correlation expresses the variation
between two variables, and does not include the variation among the variables.
Kim (2009) discussed a number of geoscience authors criteria for validity in relation to
calculated correlation coefficients. Geoscience-related regression values are often very low
because influences on geological and operational parameters are most commonly multivariate.
55

Research methodology and design

Cesano et al. (2000) states that geological or hydrogeological variables that are correlated at
0.5 (R2 = 0.25) often can be considered as a high correlation. Holmy (2008) refers to R2
values of DVproviding low to medium statistical support for several hypotheses.
Henriksen (2008), however, considers regression values of R2 < 0.09 as indicative of weak
correlations. Based on the literature examined in this study, values of R2 < 0.1 are not
regarded as significant correlations. Correlation coefficients greater than R2are assigned
the following qualitative expressions of validity:
x
x
x
x
x

R2 < 0.1
0.1 < R2 < 0.25
0.25 < R2 < 0.5
0.5 < R2 < 0.75
0.75 > R2

No correlation
Weak to medium correlation
Medium to strong correlation
Strong correlation
Very strong correlation

In addition to the correlation coefficient, statistical significance has been evaluated. Statistical
significance is an expression of how certain we are that a difference or relationship exists
among the variables under consideration, and of the extent to which the calculated probability
of a result is not due to coincidence. For bivariate correlations (correlation between two
variables), the Pearson Significance is used for normally distributed variables, and the
Spearman correlation for values not normally distributed (Helbk and Westgaard 2008). In
cases where one variable is normally distributed and the other not, the Spearman Significance
is used. Calculations of the correlations and the Pearson and Spearman Significance
parameters have been made using Microsoft Excel and IBM SPSS respectively. In order to
determine whether a data set is normally distributed or not, the one-sample KolmogorovSmirnov Test in SPSS has been used.
Qualitative and subjective assessment has also been used to evaluate the validity and
usefulness of the correlations. Multiple regressions (correlations between a dependent variable
and several independent variables) have been carried out using the software SPSS for tool life
estimates involving several variables.

3.4 Discussion the experiences and opinions of individuals and experts

Discussions and the sharing of ideas and experience with contractors, clients and TBM
manufacturers has been shown to be a very useful way of obtaining feedback on work in
progress, and a useful means of gaining access to more data (see also Section 3.3 concerning
the reason for developing the SGAT apparatus). The SGAT apparatus is a direct outcome of
discussions and experience-sharing with other individuals and experts, specifically those from
the BASF Construction Chemical Company (Langmaack 2011). The process of carrying out
research and development together with manufacturers and suppliers is helpful in the
following ways:
x

Quick and direct feedback from product and research end-users is obtained
56

Research methodology and design

Access to relevant experience and data is facilitated. In the case of development of the
SGAT, this involved access to real soil samples, soil conditioning additives and field
experience
Financial support can be obtained.

However, joint development and research involving manufacturers and suppliers may lead to
a monogamous situation, excluding other suppliers, thus experiences. Due to time constraints,
all tests involving development of the SGAT apparatus carried out to date (medio 2013) have
been performed jointly with BASF. However, there exists no prohibition on carrying out
testing using other suppliers products, and in autumn 2013 testing of the response of
bentonite on the SGAT apparatus was carried out. This will be repeated in winter 2014.
Discussions and experience-sharing have been carried out in connection with Projects 2, 4, 5,
6, 7, 8 and 9 listed in Table 4. This dialogue has provided a valuable source of feedback and
new ideas concerning what might be included in further research. In some cases, the
information obtained via discussions with experts can be biased, so in order to not to base all
data acquisition on the subjective opinions of contractors and site owners, a large proportion
of the soil samples and TBM data have been collected by the authors during site visits.
As a supplement to the projects listed in Table 4, consultancy work involving soil abrasivity
measurements has been carried out. The findings of this work are not included in this thesis
due to an on-going dispute between the contractor and the client, although it is hoped that
these findings can be published after the dispute has been settled.
It is possible that information obtained by means of discussions with individuals representing
the various parties to the dispute is selective, and that this will influence the soil samples they
send for testing. SINTEF was contracted to measure abrasion properties and drillability for
both the client and the contractor for both a soft ground project in the US, and a hard rock
project in Europe. The contractors tended to select abrasive samples for testing, while the
client selected less abrasive samples.
In addition to discussions with the tunnelling industry, a GEMINI Centre for Tribology has
been established at NTNU from which I have obtained tuition in tribology from experts at
NTNU and SINTEF.

3.5 Research methodology related to the published papers

Table 6 shows the research methodologies used in the preparation of the published papers
included in this thesis. As the table demonstrates, several of the papers include literature data,
field data and laboratory data. The reasons for this are as follows;

57

Research methodology and design

1. An increase in the volume of literature which became available on the subjects of soil
abrasivity and the estimation of tool life in soil and soft ground TBM tunnelling
during the PhD study.
2. The main approach adopted in this study is to find empirical relations between
laboratory and field.
Table 6 also reveals the lack of field data related to the recently proposed SGAT tester. The
reason for this is the delay in completion of the SGAT tester, which in turn resulted in delays
in establishing a systematic correlation study between SGAT results and tool life. However,
as already mentioned, an evaluation of the SGAT apparatus response to testing soil with
bentonite is currently on-going. In this project, SGAT results will be evaluated against actual
TBM performance and tool life data obtained from an ongoing slurry TBM project.

58

Research methodology and design

Table 6. Research methodologies used in the preparation of the

published papers included in this thesis.

Paper Title
1

2
3

4
5
6

8
9

10
11
12

Literature Field
Lab.
Dialogue
research research input

Classifications of properties influencing the

drillability of rocks, based on the
NTNU/SINTEF test method
Soil Abrasion in TBM Tunnelling
Review and assessment of the
NTNU/SINTEF Soil Abrasion Test (SAT)
for determination of abrasiveness of soil and
soft ground
Tunnelling in abrasive soils review of a
tunnel project in Germany
Overview of pipe-jacking performance
review of tunnel projects
Challenges of Methods and Approaches for
Estimating Soil Abrasivity in Soft Ground
TBM Tunnelling
Overview of Methods and Approaches used
at NTNU/SINTEF to Estimate Soil
Abrasivity in TBM Tunnelling
Predicting the abrasivity of in-situ like soils
Development of the Soft Ground Abrasion
Tester (SGAT) to predict TBM tool wear,
torque and thrust
Anti-wear and anti-dust solutions for hard
rock TBMs
Influence of corrosion on abrasion of steels
used in TBM tunnelling
TBM Cutter Steel a challenge for
Norwegian steel suppliers

59

X
X

X
X

X
X

X
X

X
X

X
X

X
X
X

Discussion

4 Results and analyses

The main results of this PhD study are set out in the research papers listed in Table 1. This
Chapter describes supplementary results, updates and as yet unpublished results such as the
following;
x
x
x
x
x
x

A statistical overview of recorded tool life

An update concerning the relationship between SAT values and recorded tool life
data obtained from soft ground TBMs
An update of the statistical overview of recorded results (SAT values and other
estimators such as quartz content and mineralogy)
A correlation between various abrasivity estimators such as quartz content and SAT
results
Current research work on the recently developed Soft Ground Abrasion Tester
(SGAT)
Correlations between parameters such as TBM diameter, overburden, quartz content,
soil uniformity index and recorded soft ground TBM tool life.

4.1 Field research

The field data and experience-sharing (collectively referred to here as Field research)
obtained as part of this study comprises approximately 500 replaced tools from a total of 49
downtimes during 14 separate projects. The reader is referred to Section 3.2 for more
information about the sites selected for this study.
Bruland (2000) records cutter tool life in hours, and back calculates cutter life using the units
m/h and sm3/h. The reason for expressing tool life in hours is to include the TBM operation in
the tool life estimate. However, excavation tool life data available to this study is expressed
largely in terms of sm3/h, due to a lack of TBM performance data (which demonstrates the
relationship between chainage and TBM machine hours). Figure 38 shows the relationships
between tool life in terms of h/tool and tool life in sm3/tool for projects where TBM
performance data were available. The black regression line and formula includes all data sets,
while the red line and formula excludes two outliers (data points 370 and 190). The outliers
are derived from data from the Bergedorf pipe-jacking project in Germany, where
performance was high (12 m/shift gross production, corresponding to 4 m/h net penetration
rate).

60

Discussion

Figure 38. Relationship between recorded tool life (in hours per tool) versus solid cubic metre per tool.
The black regression line and formula are derived from all available data, while the red regression line
and formula are derived from all data excluding two outliers obtained from the Bergedorf pipe-jacking
project.

Figure 39 shows the range of recorded soft ground tool life expressed in sm3/h. Recorded tool
life values range from approximately 50 sm3/h to 3500 sm3/h. The lowest value is derived
from a pipe-jacking project using a small diameter TBM in a well-graded soil with high
quartz content, while the highest values are derived from a soft ground project in weathered
limestone and claystone (UCS < 2 MPa comprising mainly calcite). The figure also illustrates
EPB and slurry shield tool life data. Due to the small number of data sets for EPB tool life, it
is inadvisable to carry out a direct comparison between slurry shield tool life and the EPB
tool.

61

Discussion

Figure 39. Cumulative distribution of recorded soft ground tool life.

Figure 40 is a box whisker diagram showing the distribution of recorded tool life (sm3/h) for
slurry and EPB face support approaches. According to SPSS, the data points on the right of
the slurry whisker are outliers 7. The figure shows the wide range covered by the few EPB tool
life data points available, and serves to emphasise that a direct comparison between slurry and
EPB tool life cannot be made based on available data.

Figure 40. A box whisker plot (SPSS) showing data distribution

for EPB and slurry soft ground tool life.

Outliers are defined as values greater than 1.5 interquartiles from the 25th or 75th percentiles. An interquartile
is 3rd quartile 1st quartile, and is represented by the width of the box in the box whisker plot.

62

Discussion

At the time of writing, a field study is in progress at a major slurry TBM project in Europe. At
present, the TBM has experienced one episode of downtime of about 1 weeks duration due to
the need to replace approximately 50% of the ripper and scraper tools after 300 metres of
tunnelling. Both the field data and laboratory measurements will be published in a paper
following completion of this PhD study.

4.2 Soil Abrasion Test

A total of 313 unique SAT samples are included in this study (see Table 5 for information
regarding the origin of the samples). Descriptions of the SAT test procedure and the
geological samples used during SAT testing can be found in Paper 3 (Jakobsen et al.
2013a).
Since publication by Jakobsen et al. (2013a), the number of SAT tests has increased from
254 to 313 such that the distribution of SAT values can now be updated. Figure 41 shows
how the cumulative distribution of SAT values has developed as more tests have been
conducted. Visual inspection of the figure would suggest that there has been no significant
change in the distribution in the period 2010 to 2013.

Figure 41. Historic development of the cumulative distribution of recorded SAT values.

The classification system as presented in Jakobsen et al. (2013a) is retained because the more
recent results did not change the data distribution pattern. The classifications of SAT values
according to Jakobsen et al. (2013a) are as follows;
-

6$7LVFODVVLILHGDVlow
7 < SAT < 22 is classified as medium
6$7LVFODVVLILHGDVhigh

63

Discussion

Similarly, the correlations between SAT values and quartz content have been updated in the
light of the more recent test results (Figure 42). The new data have resulted in no change to
the trend presented by Jakobsen et al. (2013a). The bivariate correlation between SAT
values and quartz content has been found to be statistically significant (Table 7).

Figure 42. Correlation between SAT values and quartz content. N=70.
Table 7. SPSS output data on the bivariate correlation between SAT values and quartz content.
SAT-QUARTZ
Pearson Correlation

.528

Sig.

.002

69

Figure 43 shows that the correlation between SAT values and the Vickers Hardness
Number Rock (VHNR) contains data supplementary to those presented in Jakobsen et al.
(2013a).This correlation has also been found to be statistically significant (Table 8).

64

Discussion

Figure 43. Correlation between SAT values and Vickers Hardness Number Rock (VHNR).
N=30. (from Jakobsen et al. 2013a).
Table 8. SPSS output data on the bivariate correlation between SAT values and VHNR.
VHNR-SAT
Pearson Correlation

.287

Sig.

.000

44

Some LCPC data have been obtained from a tunnel project owner following a recently
completed TBM project in Austria. Some SAT values have been evaluated against these
LCPC data (Figure 44). The figure shows a predominance of LCPC values of approximately
500, with SAT values vary from between 3 and 26. The variation in the SAT and LCPC
values requires further analysis when new data becomes available. The relationship between
SAT and LCPC values is classified as weak to medium according to the definitions
presented in Section 3.3, and the correlation is not found to be statistically significant (Table
9). However, the volume of data available is insufficient as a basis for determining whether
there is any form of correlation between the SAT and LCPC values.

65

Discussion

Figure 44. Correlation between LCPC and SAT values. N=8.

Table 9. SPSS output data on the bivariate correlation between SAT and LCPC values.
SAT-LCPC
Pearson Correlation

.317

Sig

.444

Since the publication of Jakobsen et al. (2013a), which includes a correlation between SAT
and soft ground tool life, new numerical and empirical data have been obtained (Figure 45
and Table 10). The main difference in terms of the correlation is the influence of data
obtained from a project with low SAT values and corresponding high tool life, which
suggests that a logarithmic correlation provides a better fit to the data than the earlier
exponential relationship. Table 11 has been prepared as a means of assessing the differences
between the correlation presented in Figure 45 and that presented in Jakobsen et al. (2013a).

66

Discussion

Figure 45. Correlation between SAT values and soft ground tool life.

Table 10. SPSS output data on the bivariate correlation between SAT values
and soft ground tool life.
SGlife-SAT
Pearson Correlation

-.404

Sig.

.003

47

Table 11. A sensitivity analysis between SAT tool life estimates referred to in this thesis.
SAT values
Estimate from
Updated estimate
Jakobsen et al.
2013a
0.5
5
10
15
25
35
50

1171
851
597
418
206
101
35

2349
1175
822
615
355
183
2

The difference between the previous estimate (Jakobsen et al. 2013a) and the updated version
is due to the introduction to the data set derived from a single project containing low SAT
values with corresponding high tool life. One conclusion to be drawn from this is that the
relationship between SAT values and soft ground tool life is not empirically saturated in the
67

Discussion

sense that the introduction of new data has had a major influence on the empirical
relationship.
The introduction of new data since the publication of Jakobsen et al. (2013a) has made it
possible to make rough comparisons of tool life data for EPB and slurry shield face support
methods. Figure 46 shows; a) the correlation between all available soft ground tool life data
and SAT values (black regression line and corresponding formula), b) for EPB soft ground
tool life and SAT values (green line) and c) for slurry soft ground tool life and SAT
values (red and purple lines). The red line includes all slurry data, while the purple line
excludes four outliers (see also Figure 40). These outliers are derived from slurry shield
tunnelling projects carried out in single-graded sands with high quartz contents and
correspondingly high SAT values. Due to their single-graded distributions, these soil types
did not generate high contact forces between the excavation tool and the tunnel face, hence
the reduced incidence of abrasive wear and longer tool life.

Figure 46. Correlation between SAT values for slurry and EPB tool life.

4.3 Soft Ground Abrasion Tester

The design and development of the Soft Ground Abrasion Tester (SGAT) is a direct outcome
of this PhD study. Almost all the results obtained to date are available and have been
published (Jakobsen et al. 2013b). They demonstrate the ability of the SGAT apparatus to
measure torque and thrust requirements in connection with small-scale drilling operations in
soft ground. The apparatus also enables an evaluation of how variation in parameters such as
compaction, water content, soil conditioning type and quantity influence thrust, torque and
tool life.
68

Discussion

Figure 47. The SGAT apparatus.

There is currently no field data against which to test the validity of the SGAT. However, one
of the samples tested by the SGAT was obtained from a formation known to have caused
severe wear problems during an unnamed tunnel project completed in the mid-2000s. The
SGAT values from this formation are high, ref. soil sample no. 3 described in Jakobsen et al.
(2013b). This qualitative observation indicates that the SGAT provides promising results, and
further research is currently being carried out (2013 - 2014).
The on-going research is focused on two areas; 1) the validity and usefulness of the SGAT for
estimating thrust, torque and wear on soft ground TBMs in different soft ground conditions,
and 2) the reliability of the test results (variation within the test results). Item 1 has already
been commenced and involves the previously mentioned investigations in connection with the
follow up at a major slurry TBM project in Europe. Figure 48 shows the grading curves for 3
soil samples obtained from this project, and Figure 49 SGAT wear measurements on 3
samples with varying water content, and with and without bentonite. The use of bentonite
reduces wear by 50% compared to an unconditioned sample. This finding corresponds with
those presented in Jakobsen et al. (2013b), which reported that rates of wear can be reduced to
20% of that observed for an unconditioned sample if the correct type and amount of soil
conditioning foam is used.

69

Discussion

Figure 48. Grading curves for some of the samples obtained during follow up at the ongoing slurry
project in Europe.

Figure 49. SGAT wear measurements on 3 soil samples an ongoing European slurry TBM project. The
legend indicates the tunnel length in metres and the presence (B) or absence (NB) of bentonite.

70

Discussion

4.4 Other estimators

During the collection of soft ground tool life data and soil samples for laboratory testing,
several other parameters were acquired. In this Chapter, the following factors are investigated:
x
x
x
x
x

The influence of soil mechanical strength as measured by the Index of Uniformity,

also referred to geotechnical Uniformity Index and C u on soft ground tool life
The influence of overburden on soft ground TBM tool life
The influence of TBM excavation diameter on soft ground TBM tool life
The influence of quartz content on soft ground TBM tool life
The influence of rock hardness as measured by the Vickers Hardness Number Rock
(VHNR) on soft ground TBM tool life.

These factors have been investigated using bivariate correlations against the variable
recorded soft ground tool life.
Overburden data have been investigated using the One-Sample-Kolmongorov-Smirnov Test
to see whether or not they exhibit normal distributions. The results showed that the data were
not normally distributed, so the relationship between soft ground tool life and overburden has
been evaluated using the Spearman significance test.
4.4.1

Geotechnical Uniformity Index, Cu

The Geotechnical Uniformity Index (C u ) is a measure of the range of grain sizes (uniformity
of grain size distribution) within a given soil sample (Emdal 2002). The uniformity of grain
size of a soil influences its mechanical properties such as compressibility and shear strength,
and is relatively easy to obtain using sieve tests. The index is calculated using the following
equation;

Equation 2

where D 60 and D 10 are the grain diameters of the 60% and 10% passing fractions,
respectively.
Figure 50 shows the correlation between C u and soft ground tool life, and the data in Table 12
indicates that the bivariate correlation is statistically significant. The correlation is biased
significantly by a single outlier C u value equal to 12. According to ASTM (2011), C u values
for well gravel are greater than 4, and for well graded sands greater than 6. Sieve curves
obtained from moraine deposits in Norway have produced C u values ranging from 22 to 180
(Emdal 2002). In order to achieve such well graded materials, rock fractions including fines
are required. A C u of 180 is thus regarded as an extreme value.
71

Discussion

The absence of C u values between 6 and 12 indicates that this trend is far from having
achieved empirical saturation.

Figure 50. Correlation between the Geotechnical Uniformity Index (C u ) and soft ground tool life.
Table 12 SPSS output data on the bivariate correlation between the
Geotechnical Uniformity Index (C u ) and soft ground tool life
Cu-SGlife
Spearman's rho

Correlation Coefficient
Sig. (2-tailed)

-.672
.000

29

4.4.2 TBM diameter

Experience from hard rock TBM tunnelling demonstrates that an increase in TBM diameter
results in longer cutter life (Bruland 2000). The main reasons for this are as follows;
x The ratio of face cutters to centre and gauge cutters increases with increasing TBM
diameter. The face cutters are exposed to more favourable working conditions than the
centre and gauge cutters.
x As TBM diameter increases, the average cutter on a cutter head has a less curved
rolling track which probably results in lower lateral forces on the cutter ring (Bruland
2000).
In order to check if this also applies to drag bits, ripper tools and scrapers in soft ground
tunnelling, TBM diameter has been correlated against soft ground tool life (Figure 51). The
72

Discussion

figure shows that, based on the collected data, TBM diameter must be disregarded as an
appropriate estimator independent of other variables. The correlation was also found not to be
statistically significant (Table 13). However, it is expected that excavation diameter may
influence relative tool life in connection with projects involving soft ground and soils with
similar abrasivity properties.

Figure 51. Correlation between TBM excavation diameter and soft ground tool life.
Table 13. SPSS output data on the bivariate correlation between TBM diameter
and soft ground tool life.
Diameter-SGLife
Pearson Correlation

.009

Sig. (2-tailed)

.693

41

4.4.3 Overburden
Khler et al. (2012) presented the relationship between tool wear (pieces per metre of tunnel)
and overburden thickness from the recently completed Inntal project in Austria, and a
meaningful correlation was achieved. This would be as expected since increasing overburden
thickness usually increases compaction and in-situ density, which in turn require higher
torque and thrust from the TBM, thus resulting in higher contact forces between the
excavation tools and the tunnel face. This reasoning has been investigated using a bivariate
correlation between recorded tool life and overburden thickness for all the projects in this
thesis (Figure 52).

73

Discussion

Figure 52. Correlation between overburden and soft ground tool life.

Table 14 shows the statistical output for the bivariate correlation between soft ground tool life
and overburden. It can be seen that based on the low correlation coefficient of 0.0004 and the
high Sig (0.413 > 0.05), no correlation emerges from the current data set. However, it is
expected that overburden may influence the tool life in connection with projects involving
soft ground and soils with similar abrasivity properties.
Table 14. SPSS output data on the bivariate correlation between overburden thickness
and soft ground tool life.
OverburdenSGlife
Spearman's rho

Correlation Coefficient

.000

Sig. (2-tailed)

.413

4.4.4

36

Mineralogical content

The mineralogical content of rock and soil is expected to influence soft ground tool life.
Several researchers have used abrasivity tests to demonstrate relationships between the
presence of quartz and other hard and abrasive minerals and tool life (Tamrock 1999; Nilsen
et al. 2006a; Deutsche Vereiningung fr Wasserwirtshcaft 2008; Frenzel et al. 2008; Dahl et
al. 2012). In order to validate the influence of various minerals on soft ground tool life,
correlations have been carried out using the parameters quartz content and Vickers Hardness
Number Rock 8 (VHNR).
Quartz content is found to provide a statistically significant, medium to good, correlation with
soft ground tool life (Figure 53 and Table 15). The correlation is influenced somewhat by four
data sets in which the quartz content was zero (0%), and which all recorded high soft ground
8

See paper 6 for explaination of Vickers Hardness Number Rock (VHNR).

74

Discussion

tool life values greater than 1500 sm3/tool. By changing these quartz content values from
0% to 0.5%, a logarithmic correlation would have been achieved, resulting in a correlation
coefficient of around 0.5.

Figure 53. Correlation between quartz content and soft ground tool life.
Table 15. SPSS output data on the bivariate correlation between quartz content
and soft ground tool life.
Correlations
Quartz-SGlife
Spearman's rho

Correlation Coefficient
Sig. (2-tailed)
N

**

-,187

,003
26

The total mineralogical content of the soil samples, as expressed by the VHNR parameter, is
found to result in a statistically significant, medium to good, correlation with soft ground tool
life (Figure 54, Table 17). However, the correlation shows that tool life increases with
increasing VHNR (representing the presence of a higher fraction of harder minerals). This
trend is also observed in the correlation with quartz content (Figure 53), and is clearly the
opposite of what is expected. The reason for this is most likely the lack of sufficient data sets
on which to base the correlation.

75

Discussion

Figure 54. Correlation between VHNR and soft ground tool life.
Table 16. SPSS output data on the bivariate correlation between VHNR and soft ground tool life.
VHNR-SGlife
SGlife

Pearson Correlation

.387

Sig. (2-tailed)

.041

11

4.5 Comments and analyses

The results from the current data sets described in Sections 4.2 and 4.4 demonstrate that there
is a relationship between SAT values and the C u coefficient and soft ground tool life, and
that we observe statistically insignificant correlations between the parameters overburden
thickness and TBM excavation diameter, and soft ground tool life. The SAT and Cu
parameters are found to be independent variables (Figure 55), and can thus be used as
independent predictors in a multivariate regression analysis.

76

Discussion

Figure 55. Correlation between the Geotechnical Uniformity Index (C u ) and SAT values.

Based on the linear relationships encountered between SAT, C u and soft ground excavation
tool life, and their possible application as tool life estimators, the following empirical formula
is proposed (see Table 17):
Equation 3

where SGTL is soft ground tool life, SAT is the SAT value, and C u is the Geotechnical
Uniformity Index. The formula is obtained on the basis of the value ranges 0.5 < SAT < 35
and 1.6 < C u < 12 taken from the field data.
Figure 56 shows the validity (or applicability) of Equation 3 for SAT values between 0.5
and 50, and C u values between 1 and 12. The estimation results for the 90 percentiles of the
SAT and C u values (31.8 and 5.33) indicate a negative soft ground excavation tool life. The
equation is found valid for the 85 percentiles of the SAT and C u values (27 and 4.58) ,
resulting in a soft ground excavation tool life estimate of 200. Thus, the equation is found
valid for current 85 percentile values, but not valid for values greater than the 90 percentiles
of the SAT and C u values. The reason for the invalidity of the 90 percentile values can be
explained by the low incidence of (very) well-graded soil material in the data sets. As already
mentioned in Section 1.3, the scope of this thesis is limited to relatively homogenous soils,
excluding the influence of boulders and large tunnelling obstacles. The reliability of Equation
3 is discussed in Section 5.4.

77

Discussion

Table 17. SPSS data output of the multiple regression between soft ground tool life (dependent
variable) and SAT values and C u (independent variables)
B

Sig

(Constant)

2244,544

,000

SAT

-44,724

,002

Cu

-181,382

,013

Figure 56. Validity analysis for equation 3.

The bivariate relationships shown in Figures 45, 46 and 50 are logarithmic and exponential,
whereas equation 3 exhibits a linear relationship between SAT, C u values and soft ground
excavation tool life.

78

Discussion

5 Discussion
5.1 General
The estimation of soft ground tool life is important at nearly all stages of a TBM tunnelling
project. A description or quantification of the abrasivity properties of the soil is essential
during the design and preparation of tender documents. This information is assessed by
tendering contractors and is used to plan project schedules and costs. Disputes related to
ground conditions that arise during tunnelling projects often result in a process to obtain
quantitative measurements of soil properties in order to make comparisons with data
submitted in the tender documents. Hopefully, the results obtained from this PhD study can
provide documentation and a methodology to assist the calculation of estimates of soft ground
TBM tool wear based on soft ground and soil abrasivity. These results can be evaluated and
applied in pre-investigations, disputes and risk management, as well as in further research.
Some of the soil samples used for abrasivity testing have been derived from excavated muck
containing soil conditioning additive residues such as foam and bentonite. A series of tests has
been carried out to investigate the reliability of the results. The sample preparation procedure
was as follows;
x The drying of virgin soil samples not exposed to bentonite or foam
x The addition of 15 weight per cent of water to the virgin soil samples
x The addition of FIR 50 foam and a corresponding 50 volume per cent of bentonite
suspension to the moist virgin soil sample
x The drying of conditioned soil samples
x The addition of 15 weight per cent water to soil samples prior to testing.
Figure 58 shows SGAT results demonstrating that the influence of foam and bentonite
residues is within the expected variation of SGAT apparatus results as described by Jakobsen
et al. (2013b). Similarly, SAT values for soil samples containing conditioning residues are
found to be within the expected variation of SAT test results.

79

Discussion

Figure 57. SGAT wear measurements on soil both with (foam and bentonite)
and without (water only) soil conditioning additives

The C u values were obtained by sieving and grain size distribution evaluations carried out in
Trondheim. Clients may have removed coarse particles and fragments from some of the
samples prior to shipment, which would result in lower than true C u values for the in-situ
soils.
The parameters overburden thickness and TBM diameter did not produce statistically
significant correlations with soft ground tool life. However, it is proposed here that in the
presence of constant geological conditions and geotechnical properties, an increase in
overburden thickness should in fact decrease tool life, since an increase in overburden
thickness should promote compaction of in-situ soil. As a result, overburden probably does
influence tool life in some projects, given that ground conditions along a tunnel are often
fairly constant, having a similar mineralogical content, grain shape and grain size. Thus, in
projects involving tunnels driven through a homogenous geology (with constant geotechnical
properties) it is likely that increased overburden thickness will in fact decrease tool life. This
is also a likely explanation for some results published in the literature (Khler et al. 2012).
The predictive model for tool wear developed by the Japanese Tunnelling Society provides an
interesting approach worthy of further study and development. However, due to language
difficulties, attempts to obtain empirical data from the model were unsuccessful. The main
weakness of the Japanese model is the use of the wear coefficient (K), which is not at present
a measureable entity. A measurement and evaluation of the wear coefficient, involving a
relatively simple test (SAT or LCPC) would be of great interest.
80

Discussion

In the following, inconsistencies found during the research, and the validity of the results, will
be discussed.

5.2 Inconsistencies observed during research

Some inconsistencies have been observed during both development of the SGAT apparatus
and parallel attempts to develop an estimation model for soft ground and soil abrasivity using
this device.
Figure 58 shows that there are inconsistencies between the range of measured wear values
obtained from the SAT and SGAT tests. Soil sample 3 exhibits the lowest wear value using
the SAT test, but the highest wear using the SGAT test. This can be explained by the
following;
x
x

SGAT wear incorporates the compactibility of the soil sample, which in turn also
influences wear
The mineralogy of soil sample 3 consisted of fine particles (silt) combined with
minerals exhibiting low abrasivity such as mica and calcite, and coarser particles (sand
and fine gravel) containing quartz. In the SAT procedure a fraction of the abrasive
minerals are sieved out, thus reducing the wear value obtained from this test. In the
SGAT test, the fines created a cohesive paste resulting in higher abrasivity and
consequently a higher wear value.

Figure 58. Relationships between wear values derived from SAT tests
and weight loss measured using the SGAT apparatus.

81

Discussion

As is demonstrated in Paper 8 and Figure 49 (this thesis), the SGAT apparatus has the ability
to evaluate the effects and benefits of soil conditioning. An increase in wear and required
torque is observed in the case of all dry soil samples tested containing conditioning additives
(various foams and bentonite). This is the opposite of what is expected in the presence of soil
conditioning. The likely explanation for this is that dry soil absorbs the moisture in the
conditioning medium, thus increasing the cohesion of the soil. This effect is also reported
from two EPB projects carried out in Switzerland and Germany, where the introduction of
low concentrations of soil conditioners reduced TBM tool life (Jakobsen et al. 2012).
The correlation between Vickers Hardness Number Rock (VHNR) and soft ground tool life
(Figure 54) demonstrates that tool life increases with increasing VHNR. This is contrary to
observations made of the relationship between quartz content and soft ground tool life. The
reason for this inconsistent result is explained by insufficient data.

5.3 Literature
The initial literature search revealed a lack of available material addressing estimations of
abrasive wear in connection with soft ground TBM tunnelling. During the period of this PhD
study (2010-2013) the volume of relevant literature has increased. Test apparatuses used prior
to my PhD studies were designed to determine hard rock abrasivity, and as such lacked the
ability to measure the influence of in-situ conditions such as water content,
density/compaction and undisturbed grain size distribution.
During the latter phase of the study (2012-2013), some in preparation literature concerning
dedicated test apparatuses for the determination of soil and soft ground abrasivity was
encountered. Most of the papers and other available literature address the details of laboratory
tests, wear problems related to a specific project, or relevant re-published empirical
experience and theories. There is a lack of literature dealing with comparisons of estimation
models for tool life in connection with real soft ground TBM projects. The likely reasons for
this are as follows;
x
x

This topic is a relatively new area of industrial research, resulting from the
construction of longer and more complex tunnels in soil and soft ground
It is difficult to obtain soil samples and corresponding field data, and data owners
(mainly contractors, sometimes clients) wish to keep their data confidential

5.4 Estimation of soft ground TBM tool life

The results presented in Figures 45 and 46 indicate that the relationship between the Soil
Abrasion Test (SAT) value for wear and soft ground tool life is logarithmic. Current
correlations have used a total of 49 TBM downtimes due to tool changes. These downtimes
includes several replaced tools. Since the existing correlation has not reached empirical
82

Discussion

saturation, the addition of more data is likely to influence the result. The best correlation
coefficients shown in Figures 45 and 46 lie in the range 0.4 to 0.5. This means that 40 to 50
per cent of the variation in the recorded soft ground excavation tool life data is accounted for
by the estimator derived from the SAT tests. The remaining 50 to 60 per cent can be
explained by the following;
x The SAT samples are selected as being representative for the longer section of the
tunnel in question. However, samples are in practice derived from a very small portion
of the tunnel, and as such laboratory testing cannot account for all the variation in the
soil material along a tunnel.
x The influence of TBM design is not taken into account.
x The influence of TBM operation is not taken into account.
x The influence of the types and concentrations of soil conditioning additives is not
taken into account.
x Some of the projects consist of soil material larger than 4 mm, meaning that a portion
of the sample is removed prior to SAT testing.
By combining SAT values and the geotechnical uniformity coefficient, an estimation of
tool life is derived which incorporates both the grading curve (representing compactibility),
and the abrasive properties of the soil. The tool life estimate derived from this approach is
found to be more reliable than that derived from bivariate analysis. One advantage of using
this estimate is that it only requires a small quantity of sample. This greatly speeds up sample
collection, shipping and laboratory testing. The disadvantages of this estimate are the lack of
precise soil compaction data and the fact that it fails to take the influence of soil conditioning
additives into account.
The reason for not introducing approaches such as multivariate logarithmic regression is that
the current data sets are too small to enable the interpretation of cut-off values. A cut-off
value divides a data set in two groups, and is a prerequisite for the multivariate logarithmic
regression approach. In the context of this study, such groups might be characterised as nonproblematic soft ground excavation tool life and problematic soft ground excavation tool
life. Such studies may be evaluated at a later stage as the number of data sets available
increases.
Tool life data from an ongoing slurry TBM project are used to evaluate the validity of
equation 3 and the empirical regression formulae presented in Figures 45, 46 and 50.The
project consists of a 2-tube sub-river tunnel excavated using a single 12.5 m-diameter slurry
shield TBM (Figure 59). The ground conditions consist mainly of three strata; 1) silt and sand,
2) sand, and 3) sand and fine gravel. After excavating 470 metres at drive 1, a total of 208
scraper tools were replaced due to abrasive wear and some impact wear (Figure 60). The tool
life at tunnel metre 470 is calculated as 57650/208 = 277 sm3/t.

83

Discussion

Figure 59. The slurry shield TBM being used to excavate the ongoing TBM project used for
assessment of the validity of the estimators in the PhD study. (Photo taken from
www.tunneltalk.com).

Figure 60. Worn-out scraper tools from slurry TBM used for assessment of the validity of the
estimators in the PhD study (Photo by Wojtek Smolen).

SAT values have been measured at tunnel metres 10, 25, 70, 100, 250 and 350, and exhibit
an average of 28.7 with a standard deviation of 2.5%. The low standard deviation indicates
that abrasivity encountered between tunnel metres 10 and 350 is relatively constant.
The geotechnical uniformity index (C u ) has also been measured, exhibiting an average value
of 4.08 with a standard deviation of 32.2%. This relatively high standard deviation is the
result of variation among the grading curves between tunnel metres 10 and 350.
Figure 61 shows a comparison between different soft ground tool life estimates and the
recorded actual tool life from the followed up slurry TBM project. The validities of the soft
84

Discussion

ground tool life equation (equation 3), and the C u (see Figure 46) and SAT (see Figure 45)
parameters are good, as their values depart from the real tool life data by less than 20%.

Figure 61. Various soft ground excavation tool life estimates compared with real tool life data
from the followed up onging slurry project in Europe.

The Soft Ground Abrasion Tester (SGAT) has been developed even though the various
simplified estimators exhibit good validity (Figure 61), because in the form of a single test it
provides a direct value which takes into account the influence of factors such as water
content, density/compaction and the use of soil conditioning additives on steel life. It is
expected that this test procedure will exhibit improved validity compared to SAT estimates,
since the real world contact relationship between the steel and the soft ground contact is better
reflected by this approach. At present no measurements are available to confirm or otherwise
as to whether the SGAT provides more valid results than other estimators.
The SGAT test requires approximately 8 kg of soil material, compared to the 0.5 kg
requirement of the SAT. In the case of abrasivity measurements and assessment, the SAT
test remains dominant, while the SGAT apparatus is useful for specific applications such as
the optimisation of soil conditioning additive concentrations. Test results obtained using the
SGAT apparatus indicate a reduction in wear in the presence of high water content, indicating
that water is a good conditioning additive. However, exposure of a tunnel face in soil to too
much water will significantly affect soil rheology. If the soil is too fluid, face stability may be
affected, creating difficulties for soil transportation through the screw conveyor or slurry
lines.

85

Conclusions and recommendations

6 Conclusions and recommendations

6.1 Comparison between thesis findings and the PhD plan
The initial main goals of this study on abrasive wear on TBM tools used to excavate soil and
soft ground were as follows;
1. To find a reliable and versatile methodology for determining the potential of soil and
soft ground to cause abrasive wear on TBM tools.
2. To propose an index or simplified model for estimating tool life in connection with
soil and soft ground excavation projects (Jakobsen 2010).
In order to achieve this, three underlying objectives and sub-tasks were defined and planned;
a) To obtain TBM field data and soil samples for laboratory testing in sufficient amounts
to generate a statistical model. The term sufficient at this initial stage is taken to mean
data and corresponding samples from 5-10 projects carried out in varying ground
conditions (e.g. clay, silt and sand).
b) To evaluate the results of current laboratory methods against observed abrasive wear,
based on correlations between laboratory measurements and field observations.
c) To propose a new laboratory method which enables the testing of abrasive wear
resulting from reconstructed in-situ soil and soft ground (involving the properties soil
density, pressure, large grain size distribution range, and the use of soil conditioning
additives). To evaluate if the suggested new laboratory tests provide a better estimate
than current (2010) procedures.
Main goal 1 has been achieved by means of the SAT test procedure, 313 SAT tests and
accompanying documentation. Main goal 2 has been met by means of the correlations
documented between SAT values and soft ground tool life, and between SAT values and
the geotechnical uniformity indexs relationship to soft ground tool life.
Sub-task (a) has been achieved following the acquisition of relevant field data from 16 unique
projects (Table 4). Sub-task (b) has been achieved following the results presented in Chapter 4
and summarised in Paper 6. Sub-task (c) has been partly achieved by means of the
development of the SGAT apparatus. However, no clear conclusion can be drawn at present
as to whether estimations of wear from the SGAT apparatus are better than those generated by
other, more simplified, approaches such as SAT or LCPC.

86

Conclusions and recommendations

6.2 Conclusions
The main contributions of this research fall into three groups; 1) general findings, 2)
recommendations related to the Soil Abrasion Test and simplified estimators, and 3) the
development of the Soft Ground Abrasion Test and subsequent recommendations.
1) General findings
x
x

Soil and soft ground abrasive wear results in reduced TBM performance, which often
leads to disputes between project owners.
TBM downtime resulting from tool replacement is expensive, and should be taken into
account both in tendering contractors bid documents and project owners schedules.
Pre-investigations should thus be carried out prior to the tender process to provide a
basis for assessing tendering contractors.
There are several methods which can be used to provide an estimate of soil abrasivity
in order to assess soft ground TBM tool wear. These include the NTNU Soil Abrasion
Test, the LCPC abrasivemeter, and the Penn State Soil Abrasion System.

2) Recommendations related to the Soil Abrasion Test and simplified estimators

x
x
x
x
x
x

Results from the Soil Abrasion Test (SAT) provide a good correlation with recorded
soft ground excavation tool life.
Results from the SAT can be used to schedule downtime and maintenance of TBM
excavation tools.
SAT values correlate well with grain mineralogy and grain shape.
SAT values should be assessed together with in-situ soil parameters, such as the
geotechnical uniformity index (C u ), in order to estimate soft ground tool life.
The C u value correlates well with recorded soft ground tool life.
Quartz content provides a medium to weak correlation with recorded soft ground tool
life.

3) The development of the Soft Ground Abrasion Tester and subsequent

recommendations
x
x
x
x

The development of the Soft Ground Abrasion Tester (SGAT) has been an important
contribution of the PhD study.
Steel wear measured by the SGAT is influenced by properties of the soil such as
mineralogy, grain size distribution and compaction.
The moisture content of the soil influences wear due to its influence on compactibility.
Soil conditioning additives and their applications can reduce rate of wear, and this
reduction can be measured using the SGAT apparatus.

87

Conclusions and recommendations

The SGAT apparatus provides an indication of the thrust and torque necessary to drill
a soil sample to a given penetration. There is a clear correlation between wear and the
required torque.

6.3 Further work and future perspectives

The scope of, and results included in, this study are limited to soft ground and soil fragments
less than 10 mm in diameter, and abrasive wear on soft ground TBM tools. A natural
extension of this work would be to conduct a systematic and extensive study on all the wear
processes listed in Table 2, and to include the influence of boulder and mixed face conditions
on TBM excavation tools.
Studies addressing the impact on excavation tools caused by boulders and mixed face
conditions are lacking, even though the tunnelling industry is eager to obtain knowledge on
this issue. Estimates based on pre-investigations are available in part in publications such as
(Gudbjartsson and Iversen 2003; Zhao et al. 2007; Ozdemir 2008; Jung et al. 2011; Tth et al.
2013). However, these articles direct their focus on single project issues with no attempt to
collate and normalise data from other projects for parameters such as variation in ground
conditions and TBM diameter. Such research would provide a sound basis for a doctorate
thesis provided that funding for field work is made available.
It may be possible to obtain an evaluation of the influence of mixed face at small laboratory
scale using the Soft Ground Abrasion Tester. One possibility might be to introduce a loose
material in to hardened and cemented material, and evaluate the effect by the SGAT. Another
possibility might be to use a Hoek cell containing a mixed material (e.g. a variety of gravel
clast sizes of varying hardness) and conduct Cerchar Abrasivity Index (CAI) tests according
to the procedure suggested by Alber (2008).
In terms of the influence of soil conditioning additives, the work presented in this thesis
consists only of a few laboratory trials and the consideration of a limited number of
qualitative field observations. In order to fully understand, and provide quantitative estimates
of, soil conditioning additives on TBM thrust, torque and tool life, further field and laboratory
studies will be required. Future laboratory studies on soil conditioning additives should
include testing to assess soil rheology.
Human factors linked to TBM operation and design have not been addressed in this thesis. It
would be impossible, or at best extremely difficult, to carry out large-scale trials comparing
different TBM designs and operational approaches under constant ground conditions.
However, the work conducted on EPB tunnelling and soil conditions at TU Ruhr provides
some indications about soil rheology behaviour, also following conditioning. These results
indicate that there may be opportunities to extrapolate conditions in front of the TBM cutter
head. Numerical analyses may provide another approach to the assessment of the influence of
88

Conclusions and recommendations

TBM operation and design. The Research Council of Norway, Robbins TBM, BASF,
Jernbaneverket (the Norwegian State Railways Administration), Scana Steel Stavanger and
BMS Steel have together funded the FAST-Tunn research project which includes the
development of a numerical model designed to evaluate rock breaking under disc cutters. This
model may be adapted to address excavation in soft ground using drag bit tools, and to
analyse the influence of TBM operation (Grv et al. 2013).
The relatively small amounts of tool life data available to this thesis demand further studies to
establish the ability of the SAT to estimate wear on ripper and scraper tools, and to provide
distinct tool life wear estimates for EPB and slurry shield tunnelling. Moreover, the findings
presented in Table 11 and Figure 45 indicates that more SAT values and recorded tool life
data are required to improve the current estimator. However, it is highly unlikely that there
will ever be sufficient data and samples available to provide an estimate with 100% predictive
certainty.
The recently developed Soft Ground Abrasion Tester (SGAT) has not been evaluated against
real TBM operations. Current test results (Jakobsen et al. 2013b) are promising, but do not
provide a quantitative comparative analysis with real TBM data. However, data is currently
being obtained from the previously mentioned ongoing slurry TBM project. The results of this
work, including the influence of bentonite additives on thrust, torque and excavation tool life,
will be published. There is also lack a comparative assessment of the SGAT conditioning
scheme with real TBM operations. It is hoped that this will be addressed as part of the
ongoing follow up at the slurry TBM project (which involves the collection of soil samples at
50 - 200 metre intervals) after tests on the soils mechanical behaviour, mineralogy and/or
grain size distribution variations have been carried out. Measured torque, thrust and wear
values from the SGAT will be correlated with real TBM operations data taken from the
project.
Concerning the reliability of the SGAT, this is currently under evaluation by PhD student
Javier Macias as part of the Future Advanced Steel for Tunnelling Applications (FASTTunn) project funded jointly by the industry and the Research Council of Norway. This study
will involve a systematic evaluation of test result variation, the re-use of steel tools, and the
SGATs response to crushed rock fragments.

89

References

7 References
Alber, M. (2008). "Stress dependency of the Cerchar Abrasivity Index (CAI) and its effects
on wear of selected rock cutting tools." Tunnelling and Underground Space
Technology(23): 351-359.
ASTM (2001). G75-01 Standard Test Method for Determination of slurry abrasivity (Miller
number) and slurry abrasion response of materials (SAR number).
ASTM (2011). Standard Practice for Classification of Soils for Engineering Purposes (Unified
Soil Classification System) ASTM D2487 -11.
Babendererde, T. (2010). "Difference in wear types and recognition in data
monitoring." BASF TBM Conference, London 28th Oct. 2010.
Barzegari, G., Uromeihy, A. and Zhao, J. (2013). "A newly developed soil abrasion testing
method for tunnelling using shield machines." Quarterly Journal of Engineering
Geology and Hydrogeology DOI: 10.1144/qjegh2012-039.
Becker, T. E. E. and Jakobsen, P. D. (2013). "Overview of pipe-jacking performance - review
of tunnel projects." NO-DIG Berlin
Bruland, A. (1998b). "Hard Rock Tunnel Boring - Performance and Back-mapping." NTNUAnleggsdrift: Project report 1E-98.
Bruland, A. (1998c). "Hard Rock Tunnel Boring - The Boring Process." NTNU-Anleggsdrift:
Project report 1F-98.
Bruland, A. (1998e). "Hard Rock Tunnel Boring - The boring process." NTNU-Anleggsdrift:
Project report 1E-98.
Bruland, A. (2000). "Hard Rock Tunnel Boring - Background and Discussion." Doctoral
Thesis at NTNU.
Cesano, D., Olofsson, B. and Bagtzoglow, A. C. (2000). "Parameters regulating groundwater
inflows into hard rock tunnels - a statistical study of the Bolmen tunnel in southern
Sweden." Tunnelling and Underground Space Technology 9(No. 4): 471-482.
Dahl, F., Bruland, A., Jakobsen, P. D., Nilsen, B. and Grov, E. (2012). "Classifications of
properties influencing the drillability of rocks, based on the NTNU/SINTEF test
method." Tunnelling and Underground Space Technology 28: 150-158 DOI: DOI
10.1016/j.tust.2011.10.006.
Deketh, H. J. R. (1995). Wear of rock cutting tools. Rotterdam, Balkema.
Deutsche Vereiningung fr Wasserwirtshcaft, A. u. E. e. V. (2008). Pipe Jacking and Related
Techniques. German DWA Rules and Standards DWA-A 125E. A. u. E. e. V.
Deutsche Vereiningung fr Wasserwirtshcaft. Hennef, Germany, DWA: 70.

90

References

Dowden, P. B. and Robinson, R. A. (2001). "Coping with boulders in soft ground TBM
tunneling." Proceedings of Rapid Excavation and Tunnelling Conference (RETC):
961-977.
Drucker, P. (2011). "Validity of the LCPC abrasivity coefficient through the example of a
recent Danube gravel / Aussagekraft des LCPC-Abrasivittskoeffizienten am Beispiel
eines rezenten Donauschotters." Geomechanik und Tunnelbau 4(6): 681-691 DOI:
10.1002/geot.201100051.
Emdal, A. (2002). "Introduksjon til Geoteknikk." NTNU Institutt for geoteknikk (Norwegian).
Espallargas, N., Jakobsen, P. D. and Macias, J. (under review). "Influence of corrosion on
abrasion of cutter steels used in TBM tunnelling."
Evans, I. (1962). "A theory on the basic mechanics of coal ploughing." International
Symposium on Mining Research 2: 761-798.
Evans, I. (1965). "The force required to cut coal with blunt wedges." International Journal of
Rock Mechanics and Mining Sciences 2: 1-12.
Fowler, D. W., Allen, J. J., Lange, A. and Range, P. (2006). "The Prediction of Coarse
Aggregate Performance by Micro-Deval and Other Aggregate Tests." ICAR Report
507-1F, The University of Texas at Austin.
Frenzel, C. and Babendererde, T. (2011). "Tool wear in TBM tunnelling." North American
Tunnel Journal(April/May): 36-43.
Frenzel, C., Ksling, H. and Thuro, K. (2008). "Factors Influencing Disc Cutter
Wear." Geomechanik und Tunnelbau 1(1): 55-60 DOI: 10.1002/geot.200800006.
Gennari, A. (2004). "Tesi di Laurea - The Ball Mill Abrasivity." MSc Thesis,
UniversitaDegli Studi Di Trent and NTNU.
Gharahbagh, E. A., Rostami, J. and Palomino, A. M. (2010). "Tool Wear Issue in Soft Ground
Tunneling, Developing a Reliable Soil Abrasivity Index." Presented at the North
American Tunneling Conference.
Gharahbagh, E. A., Rostami, J. and Palomino, A. M. (2011). "New Soil Abrasion Testing
Method for Soft Ground Tunneling Applications. Tunneling and Underground Space
Technology Journal." Tunnels and Underground Space Technology 26(5): 604-613.
Gharahbagh, E. A., Rostami, J., Talebi, K. and Ibarra, J. (2013). "Experimental and Practical
Study of Impact of Soil Conditioning on Soil Abrasion and Cutter Wear of EPB
TBMs." RETC Conference June 23-26 Washington D.C.
Grdal, C., Equey, S., Armada, S. and Espallargas, N. (2012). Effect of soil and rock
composition on the wear process of cutter tool steel used in tunnel boring
machines. Presented at the NordTrib conference, Trondheim.

91

References

Grv, E., Jakobsen, P. D., Kane, A. P., Hoang, H., Smading, S. and Sagen, T. B. (2013).
"TBM Cutter Steel - a challenge for Norwegian steel suppliers." Tunnelling Journal
June/July: 40-45.
Gudbjartsson, J. T. and Iversen, K. (2003). "High Quality wear ressistant paving blocks in
Iceland." International Conference in Concrete Block Paving, Sun City, Oct 2003.
Gwildis, U. G., Sass, I., Rostami, J. and Gilbert, M. B. (2010). "Soil Abrasion Effects on
TBM Tunnelling." World Tunnelling Congress, Vancouver Canada.
Helbk, M. and Westgaard, S. (2008). Statistikk - Kort og godt. Oslo, Universitetsforlaget.
Henriksen, H. (2008). "Late Quaternary regional geodynamics and hydraulic properties of the
crystalline rocks of Fennoscandia." Journal of Geodynamics 45: 49-62.
Herrenknecht (2013). Discussion about soil abrasivity BAUMA. P. D. Jakobsen and T. E. E.
Becker. Mnich.
Herrenknecht. (2013). "Selection of slurry and EPB."
from http://www.herrenknecht.com/process-technology/geotechnology.html.
Holmy, K. H. (2008). "Significance of geological parameters for predicting water leakage in
hard rock tunnels." PhD Thesis, NTNU: 187-208.
Holzhuser, J. and Nilsen, B. (2006). "Abrasivitt von Lockergestein bei Tunnelvortrieben
Aktuelle Erfahrungen aus der Baupraxis und Entwicklung eines neuen
Laborversuchs." Baugrundtagung, Bremen.
Hood, M. C. and Roxborough, F. F. (1992). Rock Breakage: Mechanical. SME Mining
Engineering Handbook. H. L. Hartmann. Littleton, Colorado USA, Society for
Mining, Metallurgy, and Exploration, Inc. 1: 680-721.
Hutchings, I. M. (1992). Tribology: friction and wear of engineering materials, Edward
Arnold.
Impregilo (2011). Feedback after presentation/dissemination of the Soil Abrasion Test to
Impregilo S.p.A. P. D. Jakobsen. Abu-Dhabi.
ISRM (1978). "Suggested methods for the quantitative description of discontinuties in rock
masses." International Journal of Rock Mechanics and Mining Sciences 15(6): 319369.
Jakobsen, P. D. (2007). "TBM Tunnelling in Soil, Abrasiveness and Performance." MSc
Thesis, NTNU.
Jakobsen, P. D. (2010). "Prediction of Abrasive Wear in Mechanised Soil Tunnelling." PhD
Project Description, NTNU.
Jakobsen, P. D. (2012). "Overview of Methods and Approaches used at NTNU/SINTEF to
Estimate Soil Abrasivity in TBM Tunnelling." NORDTRIB Proceedings, Trondheim.
92

References

Jakobsen, P. D. and Becker, T. E. E. (2012). "Tunnelling in abrasive soils - review of a tunnel

project in Germany." The 13th KTA International Symposium on Mechanized
Tunnelling Technology, Seoul.
Jakobsen, P. D., Bruland, A. and Dahl, F. (2013a). "Review and assessment of the
NTNU/SINTEF Soil Abrasion Test (SAT) for determination of abrasiveness of soil
and soft ground." Tunnelling and Underground Space Technology 37(0): 107-114
DOI: http://dx.doi.org/10.1016/j.tust.2013.04.003.
Jakobsen, P. D. and Dahl, F. (2010). "Soil Abrasion in TBM Tunnelling." Korean Tunnelling
Association International Symposium on Mechanized Tunnelling Technology 11: 8.
Jakobsen, P. D., Langmaack, L., Dahl, F. and Breivik, T. (2012). "Prediciting the abrasivity of
in-situ like soils." Tunnels & tunnelling International(June): 41 - 44.
Jakobsen, P. D., Langmaack, L., Dahl, F. and Breivik, T. (2013b). "Development of the Soft
Ground Abrasion Tester (SGAT) to predict TBM tool wear, torque and
thrust." Tunnelling and Underground Space Technology 38(0): 398-408
DOI: http://dx.doi.org/10.1016/j.tust.2013.07.021.
Jakobsen, P. D. and Lohne, J. (2013). "Challenges of Methods and Approaches for Estimating
Soil Abrasivity in Soft Ground TBM Tunnelling." WEAR In press.
Jung, H.-S., Choi, J.-M., Chun, B.-S., Park, J.-S. and Lee, Y.-J. (2011). "Causes of reduction
in shield TBM performance A case study in Seoul." Tunnelling and Underground
Space Technology 26(3): 453-461 DOI: http://dx.doi.org/10.1016/j.tust.2011.01.001.
Khaligi, B. (2011). "Effective deisgn & development of EPBM's " Technical Webinar
presented on tunnellingjournal.com. Available
at http://tunnellingjournal.com/webinar/webinar-archives/.
Kim, Y. (2009). "Tunnel Contour Quality Index in a Drill & Blast tunnel (Definition,
Analysis and Effects)." PhD Thesis, NTNU.
Klados, G. and Yeoh, H. K. (2006). "Unirqueness of SMART project in the logistic and
construction challenges encountered during TBM North and South drive." 7th
International Congress on Rock Mechanics: 209-222.
Klemetsrud, K. (2008). Soil Abrasion in Shield Tunneling. MSc Thesis, NTNU.
Ksling, H. and Thuro, K. (2010). "Determining rock abrasivity in the
laboratory." Proceedings of the European Rock Mechanics Symposium EUROCK
2010, Lausanne, Switzerland.
Khler, M., Maidl, U. and Martak, L. (2011). "Abrasiveness and tool wear in shield tunnelling
in soil." Geomechanik und Tunnelbau 4(February): 36-53 DOI: 10.1002

93

References

Khler, M., Maidl, U., Scholz, M. and Wendl, K. (2012). "Geotechnische Erkenntnisse aus
den Hydroschild -vortrieben der Baulose H 3-4 und H 8 im
Unterinntal." Geomechanik und Tunnelbau 5 DOI: 10.1002/geot.201200045.
Langmaack, L. (2002). "Soil Conditioning for TBM: Chanches & Limits." Association
Francaise des Travaux en Souterrain.
Langmaack, L. (2009). "The truth about soil conditioning: Dos and Dont'ts." World
Tunnelling Congress, Budapest, Hungary.
Langmaack, L. (2011). Discussion about soil abrasivity during the Mechanised Tunnelling
Seminar organised by the Norwegian Tunnelling Society. P. D. Jakobsen. Bergen.
Langmaack, L., Grothen, B. and Jakobsen, P. D. (2010). "Anti-wear and anti-dust soilutions
for hard rock TBMs." World Tunnelling Congress, Vancouver Canada.
LCPC (1990). "LCPC abrasivemeter standard." Normalisation Francaise P18-579.
LCPC (2006). Communication with LCPC about LCPC abrasivemeter. October.
Lislerud, A. (1997). "Principles of Mechanical Excavation." POSIVA 97-12.
Meng, H. C. and Ludema, K. C. (1995). "Wear models and predictive equations: their form
and content." Wear 181183, Part 2(0): 443-457
DOI: http://dx.doi.org/10.1016/0043-1648(95)90158-2.
Michalakopoulos, T., Anagnostou, V., Bassanou, M. and Panagiotou, G. (2006). "The
influence of steel hardness on the Cerchar Abrasivity Index Value." International
Journal of Rock Mechanics and Mining Sciences(43): 321-327.
Min, F., Zhu, W. and Han, X. (2013). "Filter cake formation for slurry shield tunneling in
highly permeable sand." Tunnelling and Underground Space Technology 38(0): 423430 DOI: http://dx.doi.org/10.1016/j.tust.2013.07.024.
Nakamura, T. (2011). Wear in soil tunnelling. P. D. Jakobsen.
Nilsen, B., Dahl, F., Holzhuser, J. and Raleigh, P. (2006a). "Abrasivity of soils in TBM
tunnelling." Tunnels & tunnelling International(March): 36-38.
Nilsen, B., Dahl, F., Holzhuser, J. and Raleigh, P. (2006b). "Abrasivity testing for rock and
soils." Tunnels & tunnelling International(April): 47-49.
Nilsen, B., Dahl, F., Holzhuser, J. and Raleigh, P. (2006c). "SAT: NTNU's new soil abrasion
test." tunnels & tunnelling International May: 43-45.
Nilsen, B., Dahl, F., Raleigh, P. and Holzhuser, J. (2007). "The New Test Methodology for
Estimating the Abrasiveness of Soils for TBM Tunnelling." Proceedings of the Rapid
Excavation and Tunneling Conference (RETC): 104 - 106.

94

References

normalization, A. f. d. (1990). Normalisation Francaise P18-579. Granulats: Essai

dabrasavite et de broyabilite.
Ojala, N., Kuokkala, V.-T., Valtonenen, K., Kivikyt-Reponen, P. and Vuorinen, P. (2012).
"High speed slurry-pot type erosion wear tester." Presented at the NordTrib
conference, Trondheim.
Olsen, V. (2009). "Rock Quarrying - Prediction Models and Blasting Safety." Doctoral Thesis
at NTNU 96(1).
Ozdemir, L. (2008). "Influence of boulders in soft ground tunnelling. Trials from Colorado
School of Mines Linear Cutter Test." Presented at BASF TBM Seminar, Seattle.
Peila, D., Picchio, A., Chieregato, A., Barbero, M., Dal Negro, E. and Boscaro, A. (2012).
"Test procedure for assessing the influence of soil conditioning for EPB tunneling on
the tool wear." World Tunnelling Congress, Bangkok Thailand.
Plinninger, R. J. (2007). "Geotechnische Einflsse auf den Werzeugverschleiss beim Bohren
in Festgestein." Berg- und Httenmannischer Tag. TU Bergakademie Freiberg.
Plinninger, R. J., Ksling, H., Thuro, K. and Spaun, G. (2003). "Testing conditions and
geomechanical properties influencing the Cerchar abrasiveness index (CAI)
vlaue." International Journal of Rock Mechanics and Mining Sciences(40): 259-263.
RETC (2007). Discussion about soil abrasivity after presentation of "New Test Methodology
for Estimating Abrasiveness of Soils for TBM-Tunnelling. P. D. Jakobsen, F. Dahl
and P. Raleigh.
Robbins (2013). Discussion about soil abrasivity BAUMA. P. D. Jakobsen. Munich.
Rostami, J., Gharahbagh, E. A., Palomino, A. M. and Mosleh, M. (2012). "Development of
soil abrasivity testing for soft ground tunneling using shield machines." Tunnelling
and Underground Space Technology 28(0): 245-256 DOI: 10.1016/j.tust.2011.11.007.
Rostami, J., Mosleh, M., Palomino, A. M. and Gharahbagh, E. A. (2012). "Study of Soil
Abrasivity and Development of a Reliable Soil Abrasivity Index." NSF Soil Abrasion
Study Report, Grant Number: #0928757.
Rostami, J., Ozdemir, L., Bruland, A. and Dahl, F. (2005). "Review of issues related to
Cherchar Abrasivity testing and their implications on geotechnical investigations and
cutter cost estimates." RETC Conference.
Samset, K. (2008). Prosjekt i tidligfasen - Valg av konsept. Trondheim, Tapir Akademisk
Forlag.
Serveal-Instruments. (2013). "Micro-Deval abrasion testing machine." Retrieved 08.15.2013,
2013.
Smading, S. (2013). Communication about soft ground tool composition. P. D. Jakobsen.
95

References

Stachowiak, G. and Batchelor, A. W. (2004). Experimental Methods in Tribology, Elsevier

Science.
Steingrimsson, J. H., Grv, E. and Nilsen, B. (2002). "The significance of mixed-face
conditions for TBM performance." World Tunnelling 9: 435-441.
Tamrock, S. (1999). Rock Excavation Handbook, Sandvik Tamrock Corp.
Tarkoy, P. (2008). "The boulder facts of life." World Tunnelling (December 2008): 25-28.
Thewes, M. and Budach, C. (2010). "Soil conditioning with foam during EPB tunnelling /
Konditionierung von Lockergesteinen bei Erddruckschilden." Geomechanik und
Tunnelbau 3(256-267) DOI: 10.1002/geot.201000023.
Thuro, K., Singer, J., Ksling, H. and Bauer, M. (2007). "Determining abrasivity with the
LCPC Test." Proceedings of the 1st Canada - U.S. Rock Mechanics Symposium.
Tth, ., Gong, Q. and Zhao, J. (2013). "Case studies of TBM tunneling performance in
rocksoil interface mixed ground." Tunnelling and Underground Space Technology
38(0): 140-150 DOI: http://dx.doi.org/10.1016/j.tust.2013.06.001.
Ugur, I., Demirdag, S. and Yavuz, H. (2010). "Effect of rock properties on the Los Angeles
abrasion and impact test characteristics of the aggregates." Materials Characterization
61(1): 90-96 DOI: http://dx.doi.org/10.1016/j.matchar.2009.10.014.
Verhoef, P., N., W., (1997). Wear of rock cutting tools - implications for the site investigation
of rock dredging projects. Rotterdam, A.A. Balkema.
Vinai, R., Oggeri, C. and Peila, D. (2008). "Soil conditioning of sand for EPB applications: A
laboratory research." Tunnelling and Underground Space Technology 23(3): 308-317
DOI: http://dx.doi.org/10.1016/j.tust.2007.04.010.
Walpole, R. E., Myers, R., H., and Myers, S. L. (1998). Probability and Statistics for
Engineers and Scientitsts. New Jersey, Prentice Hall.
West (1989). "Rock abrasiveness testing for tunnelling." International Journal of Rock
Mechanics and Mining Sciences Abstr. 26: 321-327.
Zare, S. (2007). "Prediction Model and Simulation Tool for Time and Cost of Drill and Blast
Tunnelling." Doctoral Thesis at NTNU.
Zeal-International. (2013). "Dorry Abrasion Testing Machine: ZL6013." Retrieved
15.08.2013, 2013.
Zhao, J., Gong, Q. M. and Eisensten, Z. (2007). "Tunnelling through a frequently changing
and mixed ground: A case history in Singapore." Tunnelling and Underground Space
Technology 22(4): 388-400 DOI: 10.1016/j.tust.2006.10.002.

96

8 Published papers

9

Paper 1
Classifications of properties influencing the drillability of rocks,
basedon the NTNU/SINTEF test method
Dahl, Filip
Bruland, Amund
Jakobsen, Pl Drevland
Nilsen, Bjrn
Grv, Eivind
Published in Tunnelling and Underground Space Technology 2012

Tunnelling and Underground Space Technology 28 (2012) 150158

Contents lists available at SciVerse ScienceDirect

Tunnelling and Underground Space Technology

journal homepage: www.elsevier.com/locate/tust

Classications of properties inuencing the drillability of rocks, based

on the NTNU/SINTEF test method
Filip Dahl a,, Amund Bruland b, Pl Drevland Jakobsen b, Bjrn Nilsen c, Eivind Grv a
a
b
c

SINTEF Building and Infrastructure, Rock Engineering, N-7465 Trondheim, Norway

Department of Civil and Transport Engineering, NTNU, N-7491 Trondheim, Norway
Department of Geology and Mineral Resources Engineering, NTNU, N-7491 Trondheim, Norway

a r t i c l e

i n f o

Article history:
Received 5 April 2011
Received in revised form 29 September
2011
Accepted 6 October 2011
Available online 8 November 2011
Keywords:
Underground excavation
Site investigations
Cost estimates
TBM
Drill and blast
Brittleness
Surface hardness
Abrasion

a b s t r a c t
The demand for representative rock property parameters related to planning of underground excavations
is increasing, as these parameters constitute fundamental input for obtaining the most reliable cost and
time estimates. The Brittleness Value (S20), Sievers J-Value (SJ), Abrasion Value (AV) and Abrasion Value
Cutter Steel (AVS) have been used extensively at NTNU/SINTEF since the 1960s in connection with drillability testing of rock samples. Nearly 3200 samples originating from projects in 50 countries have so far
been tested, and the method and associated prognosis model are internationally recognised for giving
reliable estimates of time and cost for tunnelling. A classication of the NTNU/SINTEF drillability indices
Drilling Rate Index (DRI), Bit Wear Index (BWI) and Cutter Life Index (CLI) has been available since
1998, but until now no ofcial classication has been available for the individual tests used to calculate
these indices. In this paper, classications of the NTNU/SINTEF drillability test methods Brittleness Value
(S20), Sievers J-Value (SJ), Abrasion Value (AV) and Abrasion Value Cutter Steel (AVS) tests will be
described in detail. The presented classications of the individual tests are based on statistical analysis
and evaluations of the existing test results recorded in the NTNU/SINTEF database.
2011 Elsevier Ltd. All rights reserved.

1. Introduction
Producing reliable and robust prognoses on cutter wear, drilling
progress and related costs is becoming an increasingly sensitive issue for machine manufacturers and contractors dealing with operation of mechanical excavation systems in mining, tunnelling, and
underground construction. Equally important for the clients and
owners; cost and time estimates must have adequate quality. Representative and trustworthy parameters describing various rock
properties, along with rock mass properties, are crucial as these
constitute the fundamental input for obtaining the most reliable
cost estimates. This is equally important when it comes to risk
assessments described by terms such as low and good in tender documents, and situations where claims are led during or following the construction period (Dahl et al., 2010).
The original NTNU/SINTEF drillability test method, formerly
known as the NTH test (Selmer-Olsen and Lien, 1960), was developed in 19581961 for evaluation of the drillability of rocks by percussive drilling. The Drilling Rate Index (DRI) (Selmer-Olsen and
Blindheim, 1970) is assessed on the basis of two laboratory tests,
the Brittleness Value (S20) test (Matern and Hjelmer, 1943) and
Corresponding author. Tel.: +47 73 59 48 35; fax: +47 73 59 71 36.
E-mail address: lip.e.dahl@sintef.no (F. Dahl).
0886-7798/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tust.2011.10.006

the Sievers J-Value (SJ) miniature drill test (Sievers, 1950). The
DRI may be described as the S20 of rocks, also dened as the ability to be crushed by repeated impacts, corrected for the surface
hardness determined by the SJ. The Bit Wear Index (BWI), which
is used to estimate the wear rate of drill bits, is assessed on the basis of the DRI and the Abrasion Value (AV) (Selmer-Olsen and
Lien, 1960). The AV is a measure of time dependent abrasion on
tungsten carbide by crushed rock powder. The development of
the Cutter Life Index (CLI) (NTH, 1983), which took place in the
years 19801983, was based on the original NTH test method.
The CLI has since the 1980s provided the possibility of estimating
cutter life in connection with rock excavation by use of TBM. The
CLI is assessed on the basis of SJ and the Abrasion Value Cutter
Steel (AVS). The AVS test uses test pieces of steel from TBM disc
cutter rings with specic properties, and it is regarded as a measure of time dependent abrasion on cutter ring steel.
Performance prediction and cost evaluation models for drilland blast tunnelling, TBM tunnelling and rock quarrying have been
developed by correlating laboratory tests and in situ geological
data with production data from tunnelling projects. The models
are continuously updated and revised as new tunnelling data become available (Dahl et al., 2010). In recent years the NTNU/SINTEF
method has been used extensively in connection with cost/time
estimates and planning of major international underground

F. Dahl et al. / Tunnelling and Underground Space Technology 28 (2012) 150158

projects, and it is gaining acceptance as a recognised and widely

used method for TBM performance prediction testing.
The NTNU/SINTEF drillability indices have recently been registered as trademarks (Dahl et al., 2010). NTNU/SINTEF is committed
to ensure that all end users have condence in these quantitative
methods for describing drillability characteristics of rock by performing consistent and repetitive testing. Quality assured and reliable drillability test results is of vital importance in order to obtain
the best possible time and cost estimates, which is the main reason
for labelling the NTNU/SINTEF indices as trademarks.
A classication of the NTNU/SINTEF drillability indices DRI,
BW I and CLI has been available since 1998 (Bruland, 1998),
but there has until now not been any ofcially published classication available for the individual test values used to calculate these
indices. The increasing use of the NTNU/SINTEF drillability test
method has generated a demand for a classication of the individual tests, in addition to the existing classication of the drillability
indices. This paper presents classications of the S20 test, SJ-Value
test, AV test and AVS test, based on the existing test results, recorded in the NTNU/SINTEF database.
It is essential for everyone involved in planning of underground
excavations to gain a proper understanding of how various individual rock properties can inuence the drillability of rock and hence
time and cost. The classications are intended to act as an important aid in that respect.

2. Principle and specications of the NTNU/SINTEF test methods

2.1. Rock brittleness determined by the Brittleness Value (S20) test
There are several different methods used for determination of
rock brittleness (Yarali and Kahraman, 2011). The brittleness test
method, utilised by NTNU/SINTEF, was originally developed in
Sweden by Matern and Hjelmer (1943). The original test was initially intended for determination of strength properties of aggregates, but several modied versions of the test have later been
developed for various purposes. The version of the S20 test developed for determination of rock drillability has been used since
the end of the 1950s (Fig. 1).

151

S20 constitutes a measure of the rock brittleness or ability to be

crushed by repeated impacts, and it is determined by use of an impact apparatus. S20 is dened as the percentage of a pre-sieved
fraction that passes through the ner sieve after 20 impacts. The
S20 test is normally performed on three extractions from one representative and homogenised sample of crushed and sieved rock
material. The reported S20 is hence the mean value of three parallel
tests. A screening of tests performed at the NTNU/SINTEF laboratory indicates that this test, correctly performed, normally will
show a standard deviation of three parallel tests on homogenised
material (homogenised through the crushing and sieving process),
less than 2 units (i.e. 4% for a mean S20 of 50). Local variations in
the lithology and texture of the sample would most likely provide
larger differences, if the sample material not was homogenised
prior to the testing.
The lowest and the highest S20 of the 3002 values recorded in
the NTNU/SINTEF database are 15.0 (amphibolite) and 95.1 (limestone). The measuring range and the distribution of the recorded
test results are shown in Fig. 2.

2.2. Rock surface hardness determined by the Sievers J-Value (SJ) test
The Sievers J-miniature drill test (Sievers, 1950) was originally
developed by Sievers (1950s). SJ constitutes a measure of the rock
surface hardness or resistance to indentation. SJ is dened as the
mean value of the measured drillhole depths in 1/10 mm, after
200 revolutions of the 8.5 mm miniature drill bit, see Fig. 3. The
standard procedure is to use a pre-cut surface of the sample which
is perpendicular to the foliation of the rock. SJ is hence measured
parallel to the foliation. The SJ test is normally performed as 48
drillings, depending on variations in the texture of the sample.
The SJ values may however in some specic cases show a variability, which necessitates more than 8 drillings in order to achieve a
representative average value.
SJ is reported as the mean value of the performed drillings. Foliated rocks like gneiss or schist can often show a texture with distinct bands of minerals with different hardness. This can result in
signicant variations in penetration depth. It should therefore always be aimed at placing the drillholes in soft and hard layers

Fig. 1. Outline of the Brittleness Value (S20) test (www.drillability.com, 2003).

152

F. Dahl et al. / Tunnelling and Underground Space Technology 28 (2012) 150158

Fig. 2. Cumulative distribution of the recorded Brittleness Values (S20).

Fig. 5. Distribution of the recorded Sievers J-Values (SJ).

The lowest and the highest SJ of the 3046 values recorded in the
NTNU/SINTEF database are 0.5 (quartzite) and 260 (alumn schist),
respectively. The measuring range and the distribution of the recorded test results are shown in Fig. 5.

2.3. Rock abrasion on tungsten carbide and cutter ring steel

determined by the Abrasion Value (AV) and Abrasion Value Cutter
Steel (AVS) tests

Fig. 3. Outline of the Sievers J-Value (SJ) miniature drill test (www.drillability.com,
2003).

The AV test developed at the Department of Geology at NTH in

the beginning of the 1960s, constitutes a measure of the rock abrasion or ability to induce wear on tungsten carbide. The development of the AVS test was based on the AV test method. The same
test equipment as for the AV measures the AVS, but the latter uses
a test piece of steel taken from a TBM cutter ring. The AVS constitutes a measure of the rock abrasion or ability to induce wear on
cutter ring steel. The abrasion powder used for both the AV and
the AVS is normally prepared by use of test material from the
extractions used to determine S20 and should hence be regarded
as representative and homogenised sample material.
An outline of the AV and the AVS tests is shown in Fig. 6.
AV is dened as the weight loss of the test piece in milligrams
after 5 min testing. AVS is dened as the weight loss of the test
piece in milligrams after 1 min of testing. The AV and AV S tests
are normally performed on 24 test pieces. The variation is found
to be very low and it should, if the testing is correctly performed,
not exceed 5 units (milligrams of weight loss). The reported AV
and AVS are the mean value of 24 parallel tests.
The lowest and the highest AV of the 2621 recorded values in
the NTNU/SINTEF database are 0.0 (limestone) and 116.0 (quartzite). The measuring range and the distribution of the recorded test
results are shown in Fig. 7.
The lowest and the highest AVS of the 2621 recorded values in
the NTNU/SINTEF database are 0.0 (limestone) and 68.5 (quartzite).
The measuring range and the distribution of the recorded test results are shown in Fig. 8.

Fig. 4. Sievers J-Value (SJ) drillings in a granite and a gneiss sample.

3. Classications of drillability parameters

according to a visual interpretation of the composition of the rock.
Drilling in the soft/hard combination should be avoided, but this
might sometimes be impossible due to e.g. thin layers of alternating mineral composition.
The very low degree of variation found in the SJ-Value drillings
for a sample of homogenous granite (red lines) and the high degree
of variation found for a sample of mica gneiss with garnet (blue
lines) are shown in Fig. 4.

The NTNU/SINTEF database does presently contain recorded

test results for nearly 3200 samples from various rock excavation
projects. Approximately 60% of the samples are originating from
projects in Norway and the remaining from projects in 49 other
countries. The increasing use of the NTNU/SINTEF drillability test
method has, as mentioned in the introduction, generated a demand
for a classication of the individual tests as supplement to the
existing classication of the drillability indices.

F. Dahl et al. / Tunnelling and Underground Space Technology 28 (2012) 150158

153

Fig. 6. Outline of the Abrasion Value (AV) and Abrasion Value Cutter Steel (AVS) test (www.drillability.com, 2003).

Fig. 7. Distribution of the recorded Abrasion Values (AV).

These properties are as previously mentioned the input for calculations of the DRI, BWI and CLI (Dahl et al., 2010). The
NTNU/SINTEF drillability indices represent combinations of various
rock properties, as opposed to e.g. UCS or CAI, which represents
one specic rock property.
The NTNU/SINTEF tests show very good reproducibility and
consistency. The variations found in the individual values from
each test are, as shown in Section 2, normally very low and the
measured individual test values will usually not deviate much
from the reported average test value.
The measuring ranges (dened as the range from the lowest to
the highest recorded average test value) of the NTNU/SINTEF tests
are very extensive. An extensive measuring range contributes substantially to a test methods ability to distinguish and classify specic properties. The NTNU/SINTEF tests methods are in that respect
well suited to classify rock properties which inuence the drillability. This is illustrated by the box whisker charts in Figs. 912,
where the recorded distribution and range of average test results
for some selected, common metamorphic, eruptive and sedimentary rock types are shown. The large variation found within each
rock type emphasises that certain rock properties cannot be generalised solely based on a determination of rock type.

Fig. 8. Distribution of the recorded Abrasion Values Cutter Steel (AVS).

The respective NTNU/SINTEF drillability tests provide consistent and reproducible measurements of rock properties:





Brittleness or ability to be crushed by repeated impacts.

Surface hardness or resistance to indentation.
Abrasivity or ability to induce wear on tungsten carbide.
Abrasivity or ability to induce wear on cutter ring steel.

Fig. 9. Box whisker chart showing the recorded range of Brittleness Value (S20) for a
selection of common rock types.

154

F. Dahl et al. / Tunnelling and Underground Space Technology 28 (2012) 150158

3.1. Statistical evaluation and classication of test results

The NTNU/SINTEF database does currently contain 3125 rock
samples. The samples have been analysed for various purposes
and based on different test series. The total number of recorded
test values from each test hence vary from 3046 (SJ) to 1590
(AVS). The available results from each test have been statistically
analysed and the distribution of the results are shown in Figs. 2,
5, 7 and 8. The classications given in Tables 14 are based on
the statistical distribution (cumulative percentage) of the available existing values from each test. It was, due to the close relationship between the drillability tests (SJ, S20, AV and AVS) and
the drillability indices (DRI, BWI and CLI), decided to use
the same method of classication of intervals and categories as
originally used for the existing classication of the indices
(Bruland, 1998).

3.1.1. Brittleness Value (S20)

The S20 is inuenced by the mineralogical composition of the
rock as well as grain size and grain binding, but also to a great extent by the degree of weathering/alteration, microfracturing and
foliation.
The classication given in Table 1 is based on the distribution of
the recorded results of the 3001 samples which have been tested
for determination of S20.
The S20 ranges for a selection of common metamorphic, igneous
and sedimentary rock types are given in Fig. 9.
The box whisker charts given in Figs. 912, illustrate the distribution and range of recorded test results for a selection of common
rock types. An explanation of how to read these charts and some
examples of the information which can be found are given in the
following.
The lowest and highest recorded value for each rock type is
shown by the start and end point of each vertical line in Fig. 9.
The lowest and highest S20 for e.g. the 106 recorded amphibolites
are respectively 15.0 and 73.0, indicating that amphibolites can
show S20 ranging from extremely low to extremely high. The marker above the lowest value indicates the upper limit of the 10th
percentile (25.8 for amphibolites). The split box in the middle starts
at the 25th percentile and ends at 75th percentile. The colour change
in the middle is the median of the recorded values (37.7 for
amphibolites). The marker below the maximum value indicates
the lower limit of the 90th percentile, which means that only 10%
of the recorded amphibolites have S20 above 56.2.

3.1.2. Sievers J-Value (SJ)

SJ is inuenced by the same factors as the S20. The mineralogical
composition has however normally the most signicant inuence
on the surface hardness and hence on the SJ.
The classication given in Table 2. is based on the distribution
of the recorded results of the 3046 samples which have been tested
for determination of SJ.

Table 1
Classication of rock brittleness, or the ability to be crushed by repeated impacts.
Category brittleness

S20-value (%)

Cumulative percentage (%)

Extremely high
Very high
High
Medium
Low
Very low
Extremely low

P66.0
60.065.9
51.059.9
41.050.9
35.040.9
29.134.9
629.0

95100
8595
6585
3565
1535
515
05

Table 2
Classication of rock surface hardness, or resistance to indentation.
Category surface hardness

SJ value (mm/10)

Cumulative percentage (%)

Extremely high
Very high
High
Medium
Low
Very low
Extremely low

62.0
2.13.9
4.06.9
7.018.9
19.055.9
56.085.9
P86.0

05
515
1535
3565
6585
8595
95100

Table 3
Classication of rock abrasion or the ability to induce wear on tungsten carbide.
Category abrasion
on tungsten carbide

AV (mg)

Cumulative percentage
(%)

Extremely high
Very high
High
Medium
Low
Very low
Extremely low

P58.0
42.057.9
28.041.9
11.027.9
4.010.9
1.13.9
61.0

95100
8595
6585
3565
1535
515
05

Table 4
Classication of rock abrasion or the ability to induce wear on cutter steel.
Category abrasion on cutter steel

AVS (mg)

Cumulative percentage (%)

Extremely high
Very high
High
Medium
Low
Very low
Extremely low

P44.0
36.044.0
26.035.9
13.025.9
4.012.9
1.13.9
61.0

95100
8595
6585
3565
1535
515
05

The SJ ranges for a selection of common metamorphic, igneous

and sedimentary rock types are given in Fig. 10.
3.1.3. Abrasion Value (AV) and Abrasion Value Cutter Steel (AVS)
The mineralogical composition of the rock is normally the factor
which has the most signicant inuence also on the AV and AVS.
The AV and AVS tests use test pieces of different material and hardness, tungsten carbide and cutter ring steel respectively, and
quartz and other hard minerals will cause different amount of
abrasion on the test pieces. The general rule is that low quartz content will give AV and AVS of the same magnitude, while high
quartz content will give an AV higher than the AVS. The explanation for this observation is most likely related to the high hardness
of quartz, which enables quartz to cause a signicantly higher degree of abrasion on tungsten carbide than other minerals are able
to do. Grain shape, size and grain binding are other factors which
are believed to contribute substantially to the abrasiveness of
rocks.
The classication given in Table 3 is based on the distribution of
the recorded results of the 2621 samples which have been tested
for determination of AV.
The classication given in Table 4 is based on the distribution of
the recorded results of the 1590 samples which have been tested
for determination of AVS.
The AV and AVS ranges for a selection of common metamorphic,
igneous and sedimentary rock types are given in Figs. 11 and
12.

F. Dahl et al. / Tunnelling and Underground Space Technology 28 (2012) 150158

Fig. 10. Box whisker chart showing the recorded range of Sievers J-Value (SJ) for a
selection of common rock types.

155

Fig. 12. Box whisker chart showing the recorded range of Abrasion Value Cutter
Steel (AVS) for a selection of common rock types.

Fig. 11. Box whisker chart showing the recorded range of Abrasion Value (AV) for a
selection of common rock types.

4. Correlations and comparisons with other test methods

There are also several other internationally recognised test
methods for determining properties inuencing the drillability of
rocks. Uniaxial Compressive Strength (UCS), Point Load Index
(Is50) and Cerchar Abrasivity Index (CAI) are probably some of the
most well known and widely used test methods in that respect.
Correlations of the recorded results for samples which have been
tested for determination of UCS, Is50 and CAI in addition to the
NTNU/SINTEF test methods are shown as charts in Section 4
(Figs. 1317). All tests are performed at the NTNU/SINTEF Engineering Geological Laboratory. UCS and Is50 are determined in
accordance with ISRM standards and CAI according to West (West,
1989). At present, no internationally recognised standard for the
CAI exists, and there are currently several different approaches to
this test (Rostami et al., 2005). It should however be noted that
an ASTM standard for CAI recently has been proposed.

4.1. Correlation of tests used to determine strength properties

S20 and UCS are two very different test methods for determining
strength properties of rock. There are currently 57 samples, representing 18 different rock types, in the NTNU/SINTEF database
which have been tested by use of both test methods. The tested
samples have S20 values in the range from 25.3 to 72.7 (extremely
low to extremely high) and UCS values in the range from 12.4 MPa
to 308.8 MPa (low to extremely high strength, according to ISRM).
The coloured boxes which are given in the chart in Fig. 13 show
the span of the two respective classications, S20 (Table 1) and
UCS (ISRM, 1978), while the red points indicate the arithmetic

Fig. 13. Correlation between UCS and S20. The coloured boxes refer to UCS
classication ranges given by ISRM and corresponding ranges in the suggested S20
classication.

midpoints of the classications. Figs. 1417 show the same presentation for the other classications of mechanical rock properties
which have been correlated. Fig. 13 shows a relatively poor relation
between S20 and UCS. This might be explained by the fact that S20 is
performed by applying repeated impacts on the sample material,
causing crushing of the sample material, while UCS is performed
by applying load on the sample, at a relatively slow constant rate,
until failure occurs.
The values presented in Table 5 outlines numbers of corresponding values in the two classications, S20 and UCS, e.g. there
are eight tested samples which have a very high UCS and a very
low S20. Tables 69 show the same presentation for the other classications of mechanical rock properties which have been
correlated.
The Point Load Strength (Is50) is another commonly used test
method for determination of rock strength properties. There are currently 23 samples, representing seven different rock types, in the
NTNU/SINTEF database which have been tested for determination
of both S20 and Is50. The samples show S20 values in the range from
28.6 to 58.4 (extremely low to high) and Is50 values in the range from
4.8 MPa to 15.7 MPa (high strength to very high strength, according to
Bieniwiaski, 1984). The two tests (Fig. 14) show no explicit relation.
Is50 is a determination of the indirect tensile strength of the rock
and it is, as the UCS, performed by applying load on the sample, at
a constant relatively slow rate, until failure occurs. The lack of correlation might be explained by the same differences as described for
S20 and UCS. It should however also be noted that the available statistical basis for this correlation analysis is limited.

156

F. Dahl et al. / Tunnelling and Underground Space Technology 28 (2012) 150158

Fig. 14. Correlation between Is50 and S20. The coloured boxes refer to Is50
classication ranges given by Bieniawski (1984) and corresponding ranges in the
suggested S20 classication. (For interpretation of the references to colour in this
gure legend, the reader is referred to the web version of this article.)

Fig. 15. Correlation between CAI and SJ. The coloured boxes refer to CAI
classication ranges given by Cerchar Institute, 1986 and the corresponding ranges
in the suggested SJ classication. (For interpretation of the references to colour in
this gure legend, the reader is referred to the web version of this article.)

Fig. 16. Correlation between CAI and AV. The coloured boxes refer to CAI
classication ranges given by Cerchar Institute, 1986 and corresponding ranges in
the suggested AV classication. (For interpretation of the references to colour in this
gure legend, the reader is referred to the web version of this article.)

4.2. Correlations of tests used to determine surface hardness and

abrasiveness
The SJ and the CAI are used to determine the surface hardness
and the abrasiveness of rock, respectively. The correlation shown
in Fig. 15 could hence be regarded as prevailing for two different
rock properties. The SJ test can however also be used to determine
the abrasiveness of rock, by calculating a Sievers J Interception

Fig. 17. Correlation between CAI and AVS. The coloured boxes refer to CAI
classication ranges given by Cerchar Institute, 1986 and corresponding ranges in
the suggested AVS classication. (For interpretation of the references to colour in
this gure legend, the reader is referred to the web version of this article.)

Point (SJIP) (Dahl et al., 2007). On the basis of this observation

these two tests can be regarded as related, and to a great extent affected by the same rock properties. This hypothesis is also supported by the fairly clear relation found between SJ and CAI.
There are currently 66 samples, representing 20 different rock
types, in the NTNU/SINTEF database which have been tested by
use of both test methods. The samples show SJ values in the range
from 1.6 to 95.5 (extremely high to extremely low surface hardness)
and CAI values in the range from 1.5 to 6.9 (medium abrasiveness
to quartzitic, according to Cerchar Institute, 1986).
The AV and the CAI are two different test methods used to
determine abrasiveness of rock. There are currently 60 samples,
representing 16 different rock types, in the NTNU/SINTEF database
which have been tested by use of both test methods. The samples
show AV values in the range from 0.5 to 72 (extremely low to extremely high) and CAI values in the range from 2.3 to 6.9 (very abrasive
to quartzitic, according to Cerchar Institute, 1986). The correlation
shown in Fig. 16 is for two test methods which to some degree
can be expected to be comparable. The relatively clear relation
found for the two tests is regarded as logical, and it emphasises
that the AV and the CAI to a great extent are affected by the same
rock properties and characteristics.
The Abrasion Value Cutter Steel (AVS) is, like the AV and the
CAI, a test method used to determine the abrasiveness of rock.
These three tests do however make use of test pieces consisting
of different materials. AV is performed by use of test pieces of tungsten carbide, AVS by use of test pieces of TBM cutter ring steel and
CAI by tempered steel pins of HRC 43 (according to West, 1989, but
commonly performed by pins of HRC 56).
There are currently 66 samples, representing 19 different rock
types, in the NTNU/SINTEF database that have been tested by use
of both test methods. The samples show AVS values in the range
from 0.5 to 49.5 (extremely low to extremely high) and CAI values
in the range from 1.5 to 6.9 (medium abrasiveness to quartzitic,
according to Cerchar Institute, 1986). The correlation between the
two tests (Fig. 17) show a similar relation as the one found for
AV and CAI. This can be regarded as logical and indicates that
AVS and CAI also are affected by the same rock properties and
characteristics.
The lowest degree of correlation between the NTNU/SINTEF
tests used to determine abrasiveness and the CAI was found for
AVS. It is difcult to nd an explicit explanation of this and it
should be noted that the correlations in general show relations
which are in the same range. It is also likely that the found variation to some extent is related to the spread and distribution of the
tested samples.

157

F. Dahl et al. / Tunnelling and Underground Space Technology 28 (2012) 150158

Table 5
Number of corresponding classications between Brittleness Value (S20) and Uniaxial Compressive Strength (UCS).
UCS

S20

Extremely low

Low strength
Medium strength
High strength
Very high strength
Extremely high strength

Very low

Low

Medium

1
2
2

8
1

High

4
3
10
1

10
1

2
3
4

Very high

Extremely high

1
1

Table 6
Number of corresponding classications of Brittleness Value (S20) and Point Load Strength (Is50).
Is50

S20

Very low strength

Low strength
Medium strength
High strength
Very high strength

Extremely low

Very low

Low

Medium

High

1
4

3
3

5
3

1
2

Very high

Extremely high

Table 7
Number of corresponding classications of Sievers J-Value (SJ) and Cerchar Abrasivity Index (CAI).
CAI

SJ

Extremely low

Not very abrasive

Slightly abrasive
Medium abrasive to abrasive
Very abrasive
Extremely abrasive
Quartzitic

2
1

Very low

Low

Medium

High

Very high

Extremely high

5
2

2
1

2
5

7
9

6
9
8

6
1

Table 8
Number of corresponding classications of Abrasion Value (AV) and Cerchar Abrasivity Index (CAI).
CAI
Not very abrasive
Slightly abrasive
Medium abrasive to abrasive
Very abrasive
Extremely abrasive
Quartzitic

AV

Extremely low

Very low

Low

3
1

7
1

6
4
1

4
14

Medium

High

Very high

6
5

2
3

Extremely high

Table 9
Number of corresponding classications of Abrasion Value Cutter Steel (AVS) and Cerchar Abrasivity Index (CAI).
CAI

AVS

Not very abrasive

Slightly abrasive
Medium abrasive to abrasive
Very abrasive
Extremely abrasive
Quartzitic

Extremely low

Very low

Low

Medium

High

Very high

Extremely high

4
1

3
2

1
6
2
1

1
10
14
1

9
3

3
2

1
2

5. Conclusions
Different samples from one single rock type can show extensive
variation when it comes to properties which are affecting the drillability, and it should be emphasised that it neither is recommended nor possible to generalise and predict the drillability of
rock samples only by determining the rock type. Extensive and
reliable laboratory testing is an essential factor when it comes to
obtaining reliable predictions.
The NTNU/SINTEF tests have been proven to have very good
reproducibility and consistency. The extensive measuring ranges
of the tests are well suited to distinguish and classify specic rock
properties which signicantly inuence the drillability. The NTNU/

SINTEF tests have therefore been used in connection with numerous underground excavation projects, and the method is gaining
increasing international acceptance and recognition as a reliable
tool for predicting the drillability of rocks.
The extensive amount of recorded data in the NTNU/SINTEF
database provides unique possibilities for analyzing correlations,
research and further developments. The classications given in this
paper are based on statistical analysis of the test values recorded in
the database so far. The reliability and consistency of the data used
for the classications are regarded as being very high, since they
are originating from one single laboratory using the original reference test apparatuses. It should however be emphasised that the
NTNU/SINTEF database still has a certain predominance of re-

158

F. Dahl et al. / Tunnelling and Underground Space Technology 28 (2012) 150158

corded samples which could be regarded as hard rock types and

that the classications presented in this paper hence are not
equally representative for soft rock types.
It is of vital importance for everyone involved in planning of
underground excavations to gain as much knowledge and understanding as possible on how specic properties will inuence the
drillability of rock. The classications are in that respect intended
to provide new valuable information and to constitute an important aid and guideline.
References
Bieniawski, Z.T., 1984. Rock Mechanics Design in Mining and Tunneling. A.A.
Balkema, Rotterdam, 272 p.
Bruland, A., 1998. Project Report 13A-98 Drillability Test Methods, p. 18.
Cerchar Institute, 1986. Cerchar Abrasivity Index Plan. Report issued by Centre d
Etudes de Charbonnages de France (CERCHAR) 1986 GAI-JTr/JS no.
86-538.
Dahl, F., Grv, E., Breivik, E., 2007. Development of a new direct test method for
estimating cutter life based on the Sievers J miniature drill test. Tunnelling and
Underground Space Technology 22 (1), 106116.
Dahl, F., Bruland, A., Grv, E., Nilsen, B., 2010. Trademarking the NTNU/SINTEF
drillability test indices. Tunnels & Tunnelling International (June), 4446.
ISRM, 1978. Suggested methods for the quantitative description of discontinuities
in rock masses. International Journal of Rock Mechanics and Mining Sciences &
Geomechanics Abstracts 15 (6), 319368.

Matern, N. von, Hjelmer, A., 1943. Frsk med pgrus (Tests with Chippings),
Medelande nr. 65, Statens vginstitut, Stockholm, 65 pp. (English summary, pp.
5660).
NTH, 1983. Hard Rock Tunnel Boring, Project Report 1-83. Norwegian Institute of
Technology, Div. of Construction Engineering, Trondheim, p. 94.
Rostami, J., Ozdemir, L., Bruland, A., Dahl, F., 2005. Review of issues related to
Cerchar abrasivity testing and their implications on geotechnical investigations
and cutter cost estimates. Proceedings of the RETC, 738751.
Selmer-Olsen, R., Lien, R., 1960. Bergartens borbarhet og sprengbarhet. Teknisk
Ukeblad nr. 34, Oslo, pp. 311
Selmer-Olsen, R., Blindheim, O.T., 1970. On the drillability of rock by percussive
drilling. Proc. 2nd Congress of ISRM, Belgrade.
Sievers, H., 1950. Die Bestimmung des Bohrwiderstandes von Gesteinen, Glckauf
86: 37/38, pp. 776784. Glckauf G.M.B.H., Essen.
West, G., 1989. Rock abrasiveness testing for tunnelling. International Journal of
Rock Mechanics and Mining Sciences & Geomechanics Abstracts 26 (2), 151
160.
Yarali, O., Kahraman, S., 2011. The drillability assessment of rocks using the
different brittleness values. Tunnelling and Underground Space Technology 26,
406414.

Web Reference
Dahl, F., 2003. The suggested
www.drillability.com>.

DRI,

BWI,

CLI

Standard.

<http://

Paper 2
Soil Abrasion in TBM Tunnelling
Jakobsen, Pl Drevland
Dahl, Filip
Presented at Korean Tunnelling Association, Mechanised
Tunnelling Symposium, 2010

Soil Abrasion in TBM Tunnelling

Pl Drevland Jakobsen 1)
Filip Dahl 2)
1)

The Norwegian University of Science and Technology (NTNU), Trondheim/Norway

2)

SINTEF Building and Infrastructure, Geology and Rock Engineering, Trondheim/Norway

ABSTRACT: Worldwide urbanization demands more infrastructures in densely populated

areas. When the surface is fully utilized, subsurface construction often proves itself to be the
most viable solution. In urban areas, excavation with tunnel boring machines (TBMs) is in
many cases preferred compared to drill and blast or cut and cover. Full face excavation in soil
now ranges from less than 2 metres (pipejacking) to over 15 metres in diameter. In the last
decades, there has been a focus on predicting abrasive wear on hard rock TBMs, and there are
currently several recognized methods to predict cutter life. In 2005, SINTEF and NTNU
together with Jacobs Associates and Babendererde Engineers, started to investigate how soil
abrasion can be predicted. Since 2005, the engineering geology laboratory at NTNU/SINTEF
has tested over 200 different soil samples originating from 20 projects in 8 different countries.
The intention of this paper is to examine various methods to predict abrasive wear, show the
consequences of abrasive wear on TBMs and how project owners can describe abrasive wear
in their tender documents and geotechnical baseline reports, as well as how contractors can
assess the available information.

1. INTRODUCTION
Worldwide urbanisation demands more infrastructure (road, railroad, light railroad, wasteand potable water transport and cable tunnels) in densely populated areas. If the surface
ground is fully utilised subsurface construction often proves itself to be the most viable
solution. Many of the worlds urban areas are situated on soil or weathered rock. The reason
for that is quite logical and originating from old settlements which were settled close to rivers,
sea and good agricultural areas. The demand for more subsurface infrastructure is confirmed
by Home (2010), where the total number of Tunnel Boring Machines (TBMs) working in soil
the last five years are approximately 350 units.
In addition to utilisation of TBMs for urban infrastructure projects there are also ongoing
research activities on tunnel boring to off-shore oil fields, in order to be able to exploit the oil
from a tunnel system instead of todays off-shore production oil platforms.

350
300
250
U
n
i
t
s

200
EPB
150

Slurry

100
50
0
1990-1995

1995-2000

2000-2005

2005-2010

Figure 1, Units of TBMs working in soil (after Home 2010).

Abrasive wear causes replacement and repair of cutter tools on TBMs. A cutter tool is a piece
of hardened steel or a steel matrix with carbide inserts used to break and excavate the soil.
Experiences from several TBM projects (Johansen 2000, Klemetsrud 2007, Log 2010) in hard
rock conditions shows that cutter tool life can vary by several hundred percent. In hard rock
tunnelling the abrasiveness of the rock mass can decide whether a contractor can achieve a
profit or loss in a tunnel project.
For TBM tunnelling in soil the actual cutter consumption (in replacement per metre tunnel or
by amount of excavated soil) is generally less than in hard rock conditions, due to the
relatively soft behaviour of a soil compared to a rock mass. However, the consequences of
worn out tools in soil excavation by TBM is more challenging, more time consuming and
requires maintenance in extreme working conditions. The reason for this is lack of stand up
time or stability, and water and ground pressure of a soil in comparison with hard rock.
Hence, most repair work and changing of cutter tools have to be done by divers in hyperbaric
conditions.

2. CONSEQUENCES OF TUNNELLING IN ABRASIVE SOIL

According to Hutchings (1992) abrasive wear is a process where material is removed or
displaced from a surface by hard particles or sometimes by hard protuberances on a counter
surface. Abrasive wear is according to tribological literature dependent on the following
particle properties (i.e. the following soil properties); hardness, shape and size.
In contradiction to hard rock excavation, which is a combination of compressive crushing and
inducing of tensile fracturing, soil excavation is a ripping of plastic or elastic material if the
soil is cohesive. Friction soils such as sand flows more or less if it is dry, single graded and
without shear strength. Thus mechanical excavation of soil requires less thrust on the cutter

tools compared to hard rock excavation. Still, the machines have powerful cutterheads with
high torque. The torque is needed as the soil together with bentonite or additives are rotated in
the TBM cutter chamber in closed face mode. The other reason is that the ripping of soil
demands higher torque as some soils are relatively ductile materials.
Excavation of soil generally causes less wear on cutter tools than excavation in hard rock, by
comparing the frequency of cutter tool changes. However, the consequences of worn out tools
in soil excavation compared with hard rock excavation are higher with more time consuming
and challenging repair and replacement work, in extreme working conditions. The reason is
the lack of stand up time of the tunnel face in soil compared to the rocks self supporting
behaviour. This limits the repair works either to be done in shafts or as hyperbaric
interventions. As an example, an intervention in 3 bars should not last more than 2.8 hours,
following 2 hours of decompression according to some given regulations in Europe and
America. One diver, working under hyperbaric condition, is able to change 1 - 2 disc cutters
per hour and roughly 6 soft ground tools (ripper and teeth) per hour. The working space is
confined and regarded as an extremely harsh environment. Own studies shown in Figure 3,
shows that the majority of replacement of soft ground cutter tools is done as hyperbaric
interventions.

Figure 2. Working environment for cutter tools changes and interventions at face. (After
Babenderede et al.2010).

100 %
90 %
80 %
70 %
60 %

Hyperbaric
interventions

50 %
40 %

Shaft access

30 %
20 %
10 %
0%
Single cutter Double cutter
discs
discs

Pics and
scrapers

Buckets

Figure 3. A graph showing distribution of worn out tools on a slurry TBM and the condition
where the replacement took place.
Another wear related problem experienced in TBM tunnelling in soil is destruction and wear
on the cutterhead itself. If the wear reaches a certain level, a complete tunnelling project can
be jeopardized. Figure 4 shows a cutterhead exposed to abrasive wear, resulting in a gap
between the cutterhead structure and the steel shield of approximately 4 cm. The
consequences of this are a smaller diameter tunnel which complicates installation of concrete
elements, wear on the shield and lower advance rates (Dahl et al.2007). The repair work can
take several months, and it will in some cases justify a construction of an extra shaft in order
to be able to conduct the repair work.

Figure 4. Wear on the cutterhead structure (Dahl et al.2007).

3. STATE OF THE ART

LABORATORY METHODS AND PREDICTION OF ABRASIVE WEAR IN SOIL
NTNU/SINTEF together with two tunnel consultants initiated some research on the subject
soil abrasion in 2005. (Nilsen et al.2006a, b, c). The research was initiated by a contractor
experiencing severe abrasive wear problems in connection with a project in the US. The
contractor wanted to convince their current Client that the soil was extraordinary abrasive.
Some minor revisions on the subject soil abrasion was introduced in 2007 (Dahl et al.2007)
by the Soil Abrasion Test (SATTM). The PhD candidate also made his Master thesis on the
soil abrasion subject (Jakobsen 2007). Another Master thesis was conducted on the same
subject, but by use of another approach using a different laboratory apparatus (Klemetsrud
2008) and (Jakobsen et al.2009). Since the start in 2005, the NTNU/SINTEF Engineering
Geological Laboratory has tested over 200 soil samples originating from 8 different countries
and 20 tunnel projects, i.e. the testing is commercially available. However, the current stage
of testing has several limitations causing a healthy scepticism from the tunnel industry.
Firstly, the current method is based on testing on disturbed soil, which means that in-situ
density, water content and in-situ water and soil pressure are not taken into account. Secondly,
there are not yet published any classifications or relations to the laboratory measured soil
abrasion and the actual observed wear on TBMs excavating soil.
The TBM manufacturer Herrenknecht has together with the Technical University of Mnich
currently a PhD student on the subject boulder abrasion. The study seems to be focusing on
abrasive wear caused by boulders interlaying soil matrix (Kppl et al.2009).
Another parallel research is conducted at the German geotechnical institute CDM. The
research focuses mostly upon actual conditions along the tunnel versus the conditions
described in tender documents and Geotechnical Baseline Reports (GBRs) (Gwildis et
al.2010). The subject does amongst other include observed variation between actual abrasive
wear and potential abrasive wear in the tender documents. The university partner to CDM on
this research is the Technical University of Darmstadt, Germany. Gwildis (2010) also
addresses the use of a test procedure called Miller Slurry Test to test the abrasion properties of
slurries. The article shows a relation between energy consumption for the excavation
correlated against tool life, i.e. if a soil requires high amount of energy to be excavated it will
cause more abrasive wear than a soil requiring a small amount of energy.
Additionally the Technical University of Mnich has performed testing on soil by their hard
rock abrasiveness equipment LCPC (Thuro et al.2007, Frentzel et al.2008). At the current
stage their laboratory methodology is limited to testing on the 4.0 6.3 mm fraction. The
NTNU/SINTEF Engineering Geology Laboratory has purchased the same laboratory
apparatus as used by TU Mnich, in order to evaluate if the test procedure can be altered
allowing a wider span of soils to be tested. The intention is also to evaluate soil abrasion by
use of various apparatuses.
The Penn State University in the US has initiated a study of soil abrasion and its impact, and
how soil abrasion can be predicted. Rostami (2010) presents an evaluation and limitations of
the two currently existing test methods, the SAT and the LCPC. The Penn State University
will proceed their study by designing a new apparatus, which intend to allow testing on
similar conditions as in-situ, and also by allowing hyperbaric test conditions up to 15 bars
earth/water pressure.

CURRENT TENDER DOCUMENT PRACTICE:

The content of the Geotechnical Baseline Report 1 (GBR) or geotechnical support documents
for the tender documents should contractually define the predicted ground conditions along a
project (Freeman 2009).
There are no standardised content of tender documents and geological baseline reports in
tunnelling. The content and layout of this valuable information depends on clients and project
owners and their consultants, hence abrasion properties of a soil can or cannot be quantified
dependent of the project owners and their consultants.
Generally, project owners quantify their soil with respect to some geotechnical values (soil
strength, water content, grain size distribution etc.) and from known problems in neighbouring
projects. Data from neighbouring projects is a valuable source of information, but it
sometimes causes the following statement in the tender documents; The soil is expected to
have a medium abrasivity, but it should be expected extreme abrasivity. This tends to overprotect the project owners and move the risk to the tendering contractors.
Several project owners in the US are now using the NTNU/SINTEF Soil Abrasion Test
(SAT) to describe the pre-investigated soil abrasion properties in their projects (Dahl et al.
2007, Gwildis et al.2010, Caufield et al.2009). However, the outcome of this test has been an
abrasion value which can range various sections along the tunnel drive with each other, but
without any specific soil classification. A suggested preliminary classification of SAT will
therefore be presented in the following chapter.
Baselines can also be conflicting with respect to abrasive wear and prediction of cutter life
(Freeman 2009). For hard rock cutter life prediction both the Cerchar Abrasivity Index (CAI)
and the NTNU/SINTEF Cutter Life Index (CLI) are widely used and acknowledged. If the
two estimates are not corresponding, the contractor is free to choose the most suitable
estimate for him (Freeman et al.2009) If several soil abrasion estimates such as SAT, LCPC
and the SAR number from the Miller Slurry Test is presented in the geotechnical report this
provides more information than using a single method, but it might also cause conflicting
abrasion properties and cutter life estimates.

4. THE NTNU/SINTEF SOIL ABRASION TEST

TEST PROCEDURE
The NTNU/SINTEF Soil Abrasion Test (SAT) is a further development of the existing
abrasion tests for rock. Compared with the AVS test only one detail has been changed: instead
of crushed rock powder <1 mm, a sieved soil sample <4 mm is used in the SATTM test. The
initial SATTM tests were performed with an upper grain size limit of 1 mm (Nilsen et al.
2006a, b, c), but was in 2007 modified to allow grains up to 4 mm. (Dahl et al.2007).

Term widely used in the US

Figure 5. The NTNU/SINTEF Abrasion Test apparatus.

Figure 6. Interaction between soil fragments and steel test piece. (After Dahl et al.2007)
To enable comparison with previous test results and to take advantage of the extensive
NTNU/SINTEF database it is considered important to follow the standardized
NTNU/SINTEF abrasion test procedures as closely as possible. The following preparation of
soil samples is therefore recommended, and has been followed for the soil testing described
here:
In order to reduce or avoid changes of the original grain shape and size, soil samples should
be dried gently in a ventilated oven at 30o C for 2 - 3 days. The following techniques should
be used after drying in order to disintegrate and separate the particles for the abrasion
powder:
1)

Disintegration by use of a soft hammer.

2)
Sieving with steel balls as gentle milling/disintegration aid.
3)
Initial disintegration in a jaw crusher if the samples contain very hard lumps of
cohesive material after drying. Crushing of intact grains should be avoided
..
SATTM testing of the sieved fraction is carried out according to the same procedures as for
AVS-testing, and the SAT-value is calculated as the mean value of the measured weight loss in
mg (to be accepted, the results of 2-4 parallel tests should not deviate by more than 5 units).
PRELIMINARY CLASSIFICATION OF TEST RESULTS
The number of samples that has been tested according to this procedure is currently 210. The
samples are originating from 20 different tunnelling projects in 8 countries. This provides a
sufficient span in sample types for suggesting a preliminary classification of SAT results.
The given classification is based only on the SAT value. Other factors that might influence
the abrasive wear on TBMs such as in-situ soil density, water content and pressure and
compaction are not taken into account.
The preliminary classification is based on a cumulative distribution of the test results recorded
in the NTNU/SINTEF soil abrasion database.
100 %
90 %
80 %

High SAT > 21,

68 observations

Cumulative percentage

70 %
60 %
50 %

Medium
6<SAT<21,
71 observations

40 %
30 %

Low < 6,
72 observations

20 %

Preliminary
classification, based
on SAT database
August 2010. Includes
all soil samples tested
for SAT, n = 211
SINTEF / NTNU

10 %
0%
0

10

20

30
SAT values

40

50

Figure 7. Cumulative distribution and classification of recorded SAT values.

60

14000
s
m
3
/
c
u
t
t
e
r
t
o
o
l

12000
10000
8000
6000
y = 10435e-0,105x
R = 0,6635

4000
2000
0
0

10

15

20

25

30

35

SAT value

Figure 8. Exponential relation between SAT values and cutter tool life based on 4 TBMs in
3 different projects.
The R2 value indicates 66 % of the total variation of cutter life (average solid cubic metres per
cutter tool) in terms of excavated cubic metres can be explained with an exponential relation
to the SAT value.

5. ONGOING DEVELOPMENT WORK

NTNU/SINTEF is working towards finding a recognized and accepted methodology to
determine a soils potential to cause abrasive wear on cutter tools. We are currently evaluating
several methods, such as the SAT, the LCPC, the Ball Mill Test and various combinations
of geotechnical parameters, in order to find the most accurate and suitable approach. Common
for all these tests are that the testing is done on disturbed soil material. I.e. water content,
compaction, in-situ density, water and ground pressures are disregarded. The disadvantages of
testing disturbed soil are of course lack of influence on in-situ factors on the cutter tool life
and the benefits from excavation additives. An advantage however is simple and affordable
testing and comparable values. Therefore, a major part of the further development work will
be to gather more samples for laboratory testing and corresponding field observations for
comparison and correlations. A more accurate cutter tool life estimate can most likely be
achieved by a combination of a laboratory measured abrasion value with in-situ soil strength.
We are in addition currently designing an apparatus which will allow testing conditions which
are more similar to conditions experienced in-situ. The scope of this apparatus is to evaluate
how and to what extent water and soil pressure, compaction and density influence abrasive
wear processes in soil tunnelling, and also to try to determine torque requirements for various
soils.

Finally, at the current stage of research we intend to try to determine whether the abrasive
process in the laboratory is similar to excavation with TBMs by SEM analyses on cutter tool
steel.

6. CONCLUSIVE REMARKS
The impact of abrasive wear on steel material in contact with soil needs to be coped with in
tunnel projects. It is of great interest for the tunnelling industry to establish a model that can
predict and estimate tool and support structure replacements and repairs.
Given that surface condition allows it, additional shafts can be designed and constructed if the
soil is predicted to be extremely abrasive. This can in some cases give less construction time
and lower project costs.
If an acknowledged prediction tool is available for abrasive wear in soil tunnelling, the project
owners should take it into account and classify the soil as extremely low abrasive to extremely
high abrasive. Tendering contractors should assess the abrasiveness properties of the soil in
relation to pricing and scheduling.
For the total tunnelling industry a reliable prediction tool for abrasive wear in soil will reduce
the risk in tunnelling project related to cost, construction time, health and safety, and also
reduce the risk of claims.

References
Babendererde, T., Elsner, F., 2009. Strait Crossings with TBMs Requirements and technical
solutions. Strait Crossing Conference, Trondheim.
Caufield, J., R., Pawlik, I., Wong, J., I., 2009. Bay Tunnel Design Challenges. Rapid
Excavation and Tunneling Conference (RETC). 60 71.
Dahl, F., Nilsen, B., Holzhauser, J., Raleigh, P., 2007. The New Test Methodology for
Estimating the Abrasiveness of Soils for TBM Tunnelling. Rapid Excavation and Tunneling
Conference (RETC). 104 - 116.
Freeman T., Klein, S., Korbin, G., Quick, W., 2009. Geotechnical Baseline Reports A
Review. Rapid Excavation and Tunneling Conference (RETC). 232 - 241.
Frentzel, C., Ksling, H., Thuro, K., 2008 Factors Influencing Disc Cutter Wear.
Geomechanic und Tunnelbau volume 1, Issue 1 2008.
Gwildis, U., Maday, L., E., Newby, J., E., 2010. Actual vs. baseline tracking during TBM
tuneling in highly variable glacial geology. World Tunneling Conference, Vancouver 250
262.
Home L., 2010. Trends in the use of TBMs worldwide. Presentation NFF TBM Seminar
Bergen.
Hutchings, I., M., 1992. Tribology Friction and Wear of Engineering Materials. ISBN 0 340
56184 X. Butterworth and Heinemann.
Jakobsen, P.D., Dahl, F., Langmaack, L., 2009. Reduzierung der Hartgesteins-Abrasion durch
Einsatz von Schaum. Tunnel 8 40 - 47 /2009 .
Jakobsen, P., D., 2007. Master Thesis TBM Tunnelling in Soil, Abrasiveness and
Performance, NTNU (unpublished).
Johansen, E., D., 2000. Dul Hasti Hydroelectric Project i Kashmir. Fjellsprengnignsdagen (In
Norwegian).
Klemetsrud, K., 2007. Project work The Manhattan Tunnel of New York City Water Tunnel
No. 3, NTNU (unpublished).
Klemetsrud, K., 2008. Master Thesis Soil Abrasion in Shield Tunneling, NTNU
(unpublished).
Kppl, F., Frentzel, C., Thuro, K., 2009. Statistische Modellierung von Gesteinparametern fr
die Leistungs- und Verschleissprognose bei TBM Vortrieben, Tagung fr Ingenieuregeologie
und Forum Junge Ingenieurgeologen Zittau.
Log, S., 2010 Master Thesis, The effects of using large cutter diameters in hard rock TBM
tunneling, NTNU (unpublished).

Nilsen, B., Dahl, F., Raleigh, P., Holzhauser, J., 2006a. Abrasivity of soils in TBM tunnelling.
Tunnels & Tunnelling International, March, 36 38.
Nilsen, B., Dahl, F., Raleigh, P., Holzhauser, J., 2006b. Abrasivity testing for rock and soils.
Tunnels & Tunnelling International, April, 47 49.
Nilsen, B., Dahl, F., Raleigh, P., Holzhauser, J., 2006c. SAT: NTNUs new soil abrasion test.
Tunnels & Tunnelling International, May, 43 45.
Rostami, J., Gharahbagh, E., A., 2010. Tackling a new soil abrasivity index. North American
Tunneling Journal September. 16 18.
Thuro, K., Singer, J., Ksling, H., Bauer, M., 2007. Determining abrasivity with the LCPC
test. 1st Canada-US Rock Mechanics Symposium, 827 -834

Paper 3
Review and assessment of the NTNU/SINTEF Soil Abrasion Test
(SAT) for determination of abrasiveness of soil and soft ground
Jakobsen, Pl Drevland
Bruland, Amund
Dahl, Filip
Published at Tunnelling and Underground Space Technology,
2013

Tunnelling and Underground Space Technology 37 (2013) 107114

Contents lists available at SciVerse ScienceDirect

Tunnelling and Underground Space Technology

journal homepage: www.elsevier.com/locate/tust

Review and assessment of the NTNU/SINTEF Soil Abrasion Test (SAT) for
determination of abrasiveness of soil and soft ground
Pl Drevland Jakobsen a,, Amund Bruland a, Filip Dahl b
a
b

Department of Civil and Transport Engineering, NTNU, N-7491 Trondheim, Norway

SINTEF Building and Infrastructure, Rock Engineering, N-7465 Trondheim, Norway

a r t i c l e

i n f o

Article history:
Received 25 June 2012
Received in revised form 27 February 2013
Accepted 1 April 2013

Keywords:
Wear
Abrasion
Abrasivity
Tribology
Cost estimates
Time estimates
Soft ground tunnelling

a b s t r a c t
Measuring soil abrasivity for excavation tool life estimation, is becoming more necessary as tunnels are
longer and with limited access to execute interventions and tool changes from shafts. The tool life is a
major contributor to the tunnel excavation costs and tunnelling progress. The aim of this paper is to
explain the consequences of tunnelling in abrasive soil and soft ground conditions, and explain the
NTNU/SINTEF Soil Abrasion Test (SAT) which is one approach to measure soil abrasivity. In this paper
a total of 254 different soil samples (clay, silt, sand and gravel are represented) originating from 8 different countries have been tested and included in the discussion of the SAT procedures applicability. Further, the paper relates the SAT test values to commonly known tribological theories regarding abrasion
as well as presenting trends and correlations between the measured SAT value and measured scraper,
ripper and disc cutter life from a total of nine completed TBM/pipe jacking projects, excavated with slurry
shield face support.
2013 Elsevier Ltd. All rights reserved.

1. Introduction
During the last decade a variety of methods for determination
and prediction of abrasivity of soil and soft rock has been published. The use of the Labroatoire Central de Ponts et Chausses
(LCPC) abrasivemeter (Thuro et al., 2007), the use of linear cutter
test for determining boulder abrasivity in soft ground matrix (Ozdemir, 2008) and the Soil Abrasion Test (SAT) (Nilsen et al.,
2007) are some of the relatively new methods. The LCPC abrasivemeter and SAT methods are testing a steel samples resistance to
abrading soil, where the soil is disturbed. Disturbed means that the
soil samples are lacking several of their in situ properties, e.g. natural water content, water pressure, original grain size distribution,
in situ density and compaction. Gharahbagh et al. (2010, 2011)
have suggested a new method for testing soil abrasivity on
in situ-like soils, including the water content, possible excavation
additives and a wider range of grain sizes and compaction of the
soil material. A similar approach is also recently suggested by Barzegari et al. (2013). There are also some purely experience-based
models (e.g. published by the Japanese Tunnelling Society) where
the soil abrasivity coefcient is chosen based on experience, and

Corresponding author. Tel.: +47 924 33833; fax: +47 73 59 70 21.

E-mail address: pal.jakobsen@ntnu.no (P.D. Jakobsen).
0886-7798/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tust.2013.04.003

without any direct measure or observation (personal communication Nakamura March 2010).
One reason for this relatively recent approach of the research
subject soil abrasion can be explained by the worldwide urbanization, which constantly demands more infrastructure (road, railway, light rail, waste and potable water transport and cable
tunnels). Where the urban surface area is fully utilized, subsurface
excavation often proves to be the most viable and less disturbing
solution. Home (2010) clearly illustrates this with an example of
the total units of earth pressure balanced (EPB) and slurry shield
TBMs operating in ve year periods. From 2005 to 2010 the operating soft ground TBMs were approximately 350 units, while from
1990 to 1995 the approximate number was 150.
In order to estimate TBM tool life for the booming market of soft
ground and soil TBM tunnelling, SINTEF and the Norwegian University of Science and Technology (NTNU) initiated a soil abrasion
test in 2005. Since the initial introduction of the SAT in 2005 a
total of 254 samples have been tested, originating from eight countries (2011).
The intention of this paper is to show the distribution of laboratory results with respect to soil lithology, in order to suggest a classication system for soil abrasivity when it comes to predicting
TBM tool life. Further, some qualitative explanations on how the
SAT values correspond with tool life in soil TBM tunnelling will
be addressed together with suggestions for further work on this
relatively new subject.

108

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 37 (2013) 107114

2. Abrasive wear on TBM soft ground excavation tools

2.1. Dening abrasion and wear
According to Hutchings (1992), abrasive wear is a process
where material is removed or displaced from a surface by hard particles or hard protuberances on a counter surface. The potential of
abrasive wear is dependent on the abrasion properties of the material such as particle hardness, shape and size, meaning that abrasivity is a material property while wear is a physical result of a
degradation process.
Nilsen et al. (2007) denes primary wear and secondary wear.
Primary wear refers to wear on excavation tools such as drag bits,
scraper tools, disc cutters and excavation buckets. These tools are
designed to be changed and replaced as they are wearing down.
Secondary wear refers to wear processes which degrade steel
structures, e.g. the cutterhead. Frentzel and Babendererde (2011)
uses the same denition of primary wear, but denes secondary
wear as a wear process due to the passive interaction of the excavation tool with the tunnel face or muck in the excavation
chamber.
TBM tools are also exposed to other degradation mechanisms
than abrasive wear, which inuence the tool life. The mechanisms
can be mechanical overload of the tools due to poor operation of
the TBM, instable face with large fragments that break the tools,
blocked disc cutters and fractured tools due to poor metal processing, and blocked disc cutters due to worn out cutter bearings or
blocking due to adhesive soil behaviour. Such breakdown mechanisms should not be neglected, but they are not a part of the scope
for this paper, which focuses on the abrasion properties of soil and
soft ground and their inuence on TBM excavation tools.
2.2. Dening soil, soft rock and hard rock
ISRM (1978) denes rock and soil according to the Uniaxial
Compressive Strength (UCS); where Soil is dened as UCS less than
0.25 MPa, soft rock is dened as UCS between 0.25 and 25 MPa and
hard rock is dened as UCS over 25 MPa.
The ISRM denitions of soil, soft rock and hard rock based on
laboratory strength measurements do not take into account how
the soil is deposited. For instance basal till, stiff or hard soils can
have considerable higher UCS than 0.25 MPa. A clear denition of
soil soft rock hard rock is therefore difcult to give. The denition of soil and soft rock applicable for the Soil Abrasion Test is
dened according to Table 1.
2.3. Tool types and wear mechanisms
While hard rock excavation is a combination of compressive
crushing and inducing tensile stresses and thus fractures in the

rock mass, excavation of soil is based on ripping plastic or elastic

material, and in some cases just control the ow rate of a single
graded material with low or no cohesion. For soil excavation only
a limited amount of intact soil grains is crushed down due to the
excavation. The tool life in soil excavation without boulders is
therefore assumed to be more inuenced by the material ow,
rather than high contact forces between the excavation tools and
the geo-material.
Tunnelling in soil and soft ground utilizes a variety of different
cutter tools. Use of disc cutters is generally necessary in rock
masses exceeding 20 MPa compressive strength (Khaligi, 2011).
Disc cutters are also used in mixed geological conditions with a
mix of rock and soil, and where boulders needs to be broken down
into smaller fractions in order to enter the pressurized muck system. Disc cutters are also sometimes tted to the TBM cutterhead
to penetrate the reception shaft, after the actual tunnel boring. In
soil excavation the discs are occasionally tted with carbide inserts, in order to achieve a higher friction between the tools and
the tunnel face, avoiding blockage due to clogging around the disc
cutter, and thus a at edged worn out cutter ring. The disc cutters
are exposed to several types of wear. The most common wear processes for the disc cutters are abrasive wear which results in an
even material loss around the disc prole, and blockage of the
bearing resulting in a at edged or single side worn ring.
Drag bits, teeth or picks (soft ground tools) are utilized in cohesive ground conditions where the majority of the material consists
of clay and silt with a plastic behaviour. Scrapers are commonly
used in sandy ground conditions and ripper tools are used in coarser ground conditions typically including gravel (Babendererde,
2010). The periphery of soil excavation cutterheads is also often tted with reamers to ensure a suitable overcut (sufcient space for
the shield and concrete lining rings). The reamers are also exposed
to wear, and due to their position on the outer edge of the cutterhead they are exposed to longer travel distances compared to tools
tted closer to the centre of the cutterhead. The most common
wear processes causing tool replacements in soft ground are worn
out steel matrix holding the carbide steel inserts, and worn out carbide steel inserts.
The wear process can accelerate on the cutterhead structure if
the tools are not changed in proper time. The repair work on the
cutterhead structure is challenging with surrounding soil and
water, often resulting in long downtimes (Fig. 1).

2.4. Indicative TBM downtime and cost of tool changes

The most obvious consequence of tunnelling in abrasive ground
conditions is the increased demand for tools, and the tool cost. The
TBM downtime, i.e. unproductive time for the TBM crew and potential need for hyperbaric interventions are also contributing to

Table 1
Simple identication of soil rock description and approximate strength (UCS). The descriptions are adapted from ISRM (1978) and the corresponding identication and
approximately UCS originates directly from ISRM (1978).
Description

Identication

Approx.
UCS (MPa)

Remarks

SAT: Very soft soil to soft soil

ISRM (1978): Very soft clay (S1) and Soft clay
(S2)
SAT: Firm stiff soil
ISRM (1978): Firm clay (S3), Stiff clay (S4) and
Very stiff clay (S5)
SAT: Extremely weak to very weak rock
ISRM (1978): Hard clay (S6), Extremely weak
rock (R0) and Weak rock (R1)
SAT: Weak rock
ISRM (1978): Weak rock (R2)

Easily penetrated by st or thumb

0.0025
0.05

Applicable for SAT testing

Penetrated by thumb with effort

0.050.5

Applicable for SAT testing

Indented by thumbnail easily peeled by pocket knife.

0.55

Applicable for SAT testing

Can be peeled by a pocket knife. Shallow indentations

made by point of geological hammer

525

Not applicable for SAT testing without

crushing of intact grains

109

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 37 (2013) 107114

Fig. 1. Damaged and worn cutterhead structure on a pipe jacking machine (outer
diameter 2 m) due to worn out disc cutters and scraper tools (Photo by Tim
Becker). The red rings marks fractures in the cutterhead structure itself. (For
interpretation of the references to colour in this gure legend, the reader is referred
to the web version of this article.)

Table 2
Inputs and assumptions for calculation of TBM depreciation cost.
Input

Quantity

Interest rate per annum

TBM purchase cost
Calculated depreciation cost of TBM per shift (8 h)

5%
7 mill USD
1973USD

Fig. 2. Distribution of percentage of cost for material, work and downtime for TBM
tool changes. Scenario 1 has a total estimated cost of 10,973 USD, and Scenario 2 has
a total estimated cost of 88,020 USD.

The indicative TBM downtime and cost estimation is done for

two scenarios. In scenario 1 it is assumed a 1000 m section with
low abrasivity (1200 s m3/c), and in scenario 2 it is assumed a
1000 m section with high abrasivity (200 s m3/c). For a hyperbaric
intervention of 8 h, it is assumed that 15 tools can be replaced For
the low abrasivity this involves one intervention to change 10 tools
and for the high abrasivity a total of six interventions is needed.
The TBM depreciation cost (TBMdc) per 8 h is calculated by the
following equation:
y

TBMdc TBM pc  1  buyback%  8 h 

increase the cost for TBM tunnelling in abrasive ground conditions
(see Table 2).
In the following, two scenarios of tool change operations will be
presented together with a cost estimates for the tool changes and
cost of TBM downtime (Table 3). The investment costs for the TBM
and the tools have been obtained from the Robbins Company (personal communication Lok Home, August 2011) and from Herrenknecht
Utility
Tunnelling
Department
(personal
communication Lutz Zur Linde, December 2011). In the calculation
it is assumed that one tool cost 540 USD.

interest
y  365  24 h

In the nancial scheme it is assumed that the TBM is nanced

with an interest rate of 5% (interest) and a 20% buy-back agreement,
and that the TBM is depreciated after 3 years operation (y) with a
purchase cost (TBMpc) of 7,000,000 USD.
The total cost for replacing 10 tools in Scenario 1 is 10,973 USD,
and for replacing 63 tools in Scenario 2 is 88,020 USD. Fig. 2 shows
Scenario 1 and 2 and h.
The indicative calculation of the tool change cost is included in
this paper to show the economic effect worn out tools may have on
tunnelling projects.
3. The NTNU/SINTEF Soil Abrasion Test (SAT)

Table 3
Estimation of tool change operations.

3.1. Background

Items
Number of TBM crews
Number of crew used for tool changes and other maintenance
during downtime
Daily wage per tunnel worker incl. insurance, taxes etc.

6
4

Cost of non-productive TBM crew (2  700 USD/shift)

1400 USD/
shift
1
1.5 bars
2h
6h

Number of divers
Assumed face pressure at intervention
Time needed for diver preparation and decompressiona
Time needed for tool changes
Assumed mobilization cost for divers
Daily wage for hyperbaric divers
Payment cost for divers
Tool cost
a

After Babendererde and Elsner (2009).

700 USD/
shift

1000 USD/
diver
1200 USD/
shift
1200 USD/
shift
540 USD/
pcs

The NTNU/SINTEF Soil Abrasion Test (SAT) is an outcome and

further development of the NTNU/SINTEF abrasion tests for rock,
referred to as the Abrasion Value (AV) and Abrasion Value Cutter
Steel (AVS). Fig. 3 shows the schematic overview of the test apparatus, which consists of a rotating steel disc, which is fed with soil
grains less than 4 mm. As the soil is transported on the rotating
disc it passes underneath a steel sample, originating from a TBM
disc cutter. This contact cause abrasive wear on the steel sample.
Compared to the AVS test, which is performed by use of a crushed
rock powder less than 1 mm, a dry and disintegrated soil sample
with grain size less than 4 mm is used in the SAT test. Fig. 4
shows the interaction between the steel and soil sample for the initial AVS and the modied SAT test piece.
Initially the SAT test was performed with an upper grain size
limit of 1 mm (Nilsen et al., 2006a,b,c). Due to the relatively limited
fraction size (01 mm), the test pieces were modied in order to
include grain sizes up to 4 mm (Nilsen et al., 2007), meaning that
the SAT is applicable for testing clay, silt and sand fractions.
The test cannot be performed on fragments larger than 4 mm. Dur-

110

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 37 (2013) 107114

Fig. 3. Schematic overview of the SAT test procedure.

Fig. 4. The initial AVS (left) and modied and current SAT (right) test pieces.

In Fig. 5, the green point is a clay sample with some coarser

grains in the matrix, the yellow points are from well graded samples where varying grain sizes most likely have varying abrasivity
properties, and the red point is a single graded sand with 100%
quartz. The amount of tests are too few in order to conclude
whether the SAT test procedure systematically differentiates
SAT1mm from SAT4mm.
3.2. Test procedure

Fig. 5. Relationship between 15 soil samples tested by SAT = 1 mm and

SAT = 4 mm.

ing the modication of test pieces some studies were conducted in

order to investigate if there was a systematical change of the measured abrasivity with the two test pieces. The deviation shown in
Fig. 5 can1 be explained by either different abrasivity properties of
the particles less than 1 mm compared to particles <4 mm (for green,
red and yellow points), or by normal variation of properties in geomaterials, which are within the test procedures variation as explained in Section 3.2.

1
For interpretation of colour in Fig. 5, the reader is referred to the web version of
this article.

To be able to make comparisons with previous test results and

the NTNU/SINTEF database for hard rock it is decided to follow the
NTNU abrasion test procedure as closely as possible. The following
preparation of soil samples is therefore suggested.
In order to reduce and avoid changes of the original soil properties, the soil samples should be dried gently in a ventilated oven at
30 C for 23 days. In order to disintegrate and separate particles
into an abrasion powder the following techniques are suggested:
(1) Disintegration by use of a soft hammer (plastic head) or
manually by hand.
(2) Sieving the sample with steel balls as a gentle disintegration
aid. The sample material is sieved on 4.0 mm and 1.0 mm
sieves. The disintegration aid is applied by use of 20 small
steel balls with a diameter of 15 mm weighing 14 grams
each.
(3) Disintegration in a jaw crusher if the sample contains hard
lumps of cohesive material after initial drying. Crushing of
intact and original grains should be avoided, in order to keep
the original grain shape and size (Nilsen et al., 2007).

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 37 (2013) 107114

111

After the drying and disintegration procedures, the material

should be sieved on 4.0 mm and 1.0 mm in order to verify the grain
size distribution after preparation. Mark that if disintegration of
the sample requires jaw crushing and the soil sample is not indentable by hand, nger or nail the abrasion result should be reported
as Abrasion Value Cutter Steel (AVS), used for measuring rock
abrasivity.
Fractions above 4 mm in soil samples are sieved out prior to
testing. In order to include the mineralogy for the grains larger
than 4 mm, crushing is an option. However, by crushing the grains
it is difcult to achieve the correctly distributed grain sizes for each
mineral, as well as the original grain shape is inuenced. If the soil
sample consists of a large portion of fragments larger than 4 mm,
the SAT test should be used with caution. Hence in this paper,
it is assumed that mineral composition of the fraction larger than
4 mm is identical to the fraction less than 4 mm.
The SAT value is calculated as the mean value of the measured
weight loss on the steel bit in milligrams after testing. To be accepted as a SAT value, the results of 24 tests should not deviate
by more than 5 mg (Nilsen et al., 2007).

The steel type used in the SAT test originates from a disc cutter. The steel type is a heat treatable, low alloy steel containing
nickel, chromium and molybdenum. The steel has a good resistance against wear and fatigue degradation and is relatively ductile, similar to steel types used for TBM excavation tools. The
steel type in the SAT test is chosen due to its similar properties
compared to steel on cutter tools (disc cutters and soft ground tool
steel matrix), and due to its properties making it possible to detect
and measure weight losses over a wide range of geological
material.
The feed of soil to the rotating disc is controlled by the operator.
The criterion for adjusting the vibrating feeder is to avoid steel
against steel contact (test piece against the rotating disc) and to
avoid a pile of grains in the front of the test piece. These two criterions make the test dependent on the operator. However, the steel
of the rotating disc is a softer steel than the SAT test piece. When
the test is performed without any soil fed to the disc (steel against
steel testing), there has not been any measurable weight loss of the
SAT test piece recorded.

3.3. Test dynamics and relation to TBM excavation

4. Assessment of the Soil Abrasion Test (SAT)

The drive length of one test is 20 m with a velocity between the

soil particles and steel bit of 0.33 m/s. The velocity is relatively low
compared to a TBM cutter tool located on the periphery on the cutterhead. However, the velocity is of the same magnitude as a TBM
tool close to the average radius of the cutterhead.
The contact force between the steel and soil particles is 100 N,
which is distributed over a line (SAT test piece). The Herzian contact equations can be used when one material indent into a second
material and causes changes in the contact area (Hutchings, 1992).
The contact area is given by the following equation (Wood, 2009):

4.1. Statistical overview of SAT values


1=2
W=l  r
after Wood 2009
pE

The maximum contact pressure is given by the following

equation:


1=2
W=l  E
Pm
after Wood 2009
pr

where a is the contact area between the spherical specimen and the
plane. W the normal load, for SAT (100 N), l the length of the
SAT test piece (30 mm), r the radius of the SAT test piece
(15 mm), and E is the elastic modulus which is dependent on the
Youngs modulus and the Poissons ratios for the SAT test specimen and the dry soil given by the following equation:


 

1
1  v 21
1  v 22

after Hutchings 1992

E
E1
E2

The NTNU/SINTEF Soil Abrasion Test database contained 254

unique abrasivity measurements on soil samples by medio 2011
(Fig. 6). The maximum recorded value is 52, which is a quartz rich
sand with sharp edged grains. The lower values (SAT < 5) are
generally representing clay samples with low content of quartz
or other hard minerals.
By dening classication limits at the 33 and 66 percentile the
following classication of SAT results is obtained.
SAT 6 7 is classied as low.
7 < SAT < 22 is classied as medium.
22 6 SAT is classied as high.
The reason for suggesting only three classication categories is
due to relatively low amount of data. For the classication of Abrasion Value Cutter Steel (AVS), the categories are seven and the
number of observations is more than 1500 (Dahl et al., 2012).
Some studies on correlation between SAT values and soil
properties like mineralogy and grain shape have also been done.
Figs. 7 and 5 indicate a relationship between SAT values and
quartz content and SAT and Vickers Hardness Number Rock
(VHNR). The VHNR is a factor taking into account the percentage

For the calculation of the contact pressure the following elastic

properties have been used:

E1 Esteel 210 GPa

t1 tsteel 0:3
0:2 < tsoil t2 < 0:5 and
0:1 GPa < Esoil E2 < 5 GPa
The outcome of the Herzian maximum pressure calculation is
then that the maximum contact pressure is ranging from 200 to
370 MPa (dependent on the elastic properties of the soil sample).
The contact pressure range is close to what a TBM disc cutter faces
in hard rock application (Gong et al., 2006).

Fig. 6. Cumulative distribution of the recorded SAT values in the NTNU/SINTEF

soil abrasion database (medio 2011).

112

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 37 (2013) 107114

Fig. 10. Box plots showing the range of SAT values for clay, silt, sand and gravel.
Fig. 7. Correlation between SAT value and content of quartz. N = 62.

Fig. 11. Overview of sub groups of clay, silt, sand and gravel recorded in the NTNU/
SINTEF Soil Abrasion Database.
Fig. 8. Correlation between SAT value and Vickers Hardness Number Rock
(VHNR). N = 30.

Fig. 9. Correlation between SAT values and grain roudness. N = 15.

of each mineral and its hardness. There is also some relationship

between SAT values and grain roundness as shown in Fig. 9.
The grain roundness, R, is measured according to Eq. (5) after
Russ (2006):

4  grain area

p  grain max diameter2

The content of quartz is measured by Differential Thermal

Analysis (DTA) and other minerals by X-ray Diffraction (XRD).

The correlation between the SAT and quartz content for 62 samples proves to be good, see Fig. 7.
The correlation between SAT values and the sample mineralogy, represented by the Vickers Hardness Number Rock (VHNR)
(Fig. 8) is found less signicant than the SAT quartz correlation.
This is unexpected, as the SAT value is expected to be highly
inuenced by the total mineralogical content of soil samples. One
likely explanation for this is the relative low amount of data points
in Fig. 9.
The grain shape (according to Eq. (5)) correlates well with the
SAT values for the relative few data points shown in Fig. 9.
The relation between the measured abrasivity with the SAT
test, mineralogy and grain shape corresponds well with general tribological laws in Hutchings (1992), where a materials abrasion
properties is dened by particle hardness, shape and size. A clear
and well dened inuence of particle size is not proven by the
SAT procedure see Fig. 10.
In Fig. 10 the term gravel could be confusing, as the SAT procedure does not allow gravel particles to be tested. The results for
gravel in Fig. 10 refer to the sieved fraction less than 4 mm from
the gravel samples, meaning that the distribution is not based on
gravel size particles. Fig. 10 also shows that the grain size of the
soils do not seem to inuence the SAT value systematically
regarding grain size, but it gives an indication that sand particles
in general are more abrasive than clay and silt particles according
to the SAT test.
Fig. 11 shows sub-groups of clay, silt, sand and gravel in the
NTNU/SINTEF Soil Abrasion Database. The total amount of samples
including grains larger than 4 mm (gravely sand, gravel, silty gravel
and sandy gravel) is 25. As mentioned in Section 3.2, results

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 37 (2013) 107114

113

obtained from samples containing fractions above 4 mm should be

used with caution.
4.2. Relation between SAT values and actual tool life
Some correlations between SAT values and recorded tool life
from TBMs have been prepared to show the possible applicability
of SAT as a direct estimation method for tool life. The current correlation is based on a total of nine TBM projects (three projects in
Northern America, ve in Germany and one in the Middle East).
The tunnel lengths are ranging from 200 m (pipe jacking) up to
5500 m with relatively small excavation diameters ranging from
2.0 m to 5.5 m. The total tunnelling length in the data set is approximately 17,500 m. All projects are based on pressurizing the tunnel
face with bentonite slurry, meaning that there are no Earth Pressure Balanced (EPB) TBMs included in the data set. The relative
short tunnel drives and small TBM diameters might also inuence
the total applicability of the charts to some extent. The data set do
not contain any projects with boulders. This is most likely contributing to the correlation coefcient R2 positively, as excavating
boulders is not only an abrasive process, but something which is
still inuencing the TBM tool life.
The tool life presented in Figs. 1216 is presented as solid cubic
metre per cutter tool (s m3/c) in order to include various TBM
diameters in the correlations. The tool life has been calculated as
instantaneous cutter consumption meaning that each tunnel section between tool changes has a calculated tool life (Bruland,
1998a).
Due to the relatively low amount of data for tool changes, the
correlations are distinguished between disc cutter life and soft
ground tool life (rippers and scrapers). The data set at present is

Fig. 14. Correlation between quartz content and recorded soft ground tool life. The
red trend line represents all available data sets. The blue dashed trend line
represents data sets without the two outliers shown in Fig. 12. (For interpretation of
the references to colour in this gure legend, the reader is referred to the web
version of this article.)

too small in order to introduce tool life for varying tool sizes,
shapes, design and metallurgy.
The two outliers (marked as triangles) in Fig. 12 are from two
projects where the SAT values were high (25.5 and 33.3) and
with high tool life. The soils at the two specic projects were single
graded sand without nes. Such soils are very easy to excavate as it
is almost owing by itself, saving the TBM tools from high contact
forces and impacts. The grain size distribution and cohesive properties of the soils are not taken into account in the SAT test,

Fig. 12. Correlation between SAT values and recorded soft ground tool life.
Fig. 15. Correlation between the Vickers Hardness Number Rock (VHNR) and the
recorded soft ground tool life.

Fig. 13. Correlation between SAT values and recorded soft ground tool life after
removing the two outliers in Fig. 12.

Fig. 16. Correlation between SAT values and recorded disc cutter life.

114

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 37 (2013) 107114

although it seems to have a high inuence and importance when

making a soft ground and soil tool life estimation.
By removing the two projects from the data set the R2 increases
and the correlation becomes better, see Fig. 13. The increase of the
R2 by removing to points also shows the sensitivity of the current
correlation when introducing new data. It should therefore be
emphasized that the current correlation is based on relatively
few data and by later adding tool life data and SAT values from
other projects the relation may be signicantly changed.
For comparison, a similar correlation has been prepared to compare quartz content with recorded soft ground tool life. The correlation between quartz content and tool life based on the collected
data is non-existing by including all available data sets, as shown
in Fig. 14.
An attempt of showing tool life correlated to the mineralogical
hardness by the VHNR is shown in Fig. 15. The correlation is found
better than the quartz-tool life correlation. The correlation coefcient is however highly inuenced by the low amount of data.
In Fig. 16, a correlation between SAT values and disc cutter
life is presented. There is no relationship with the current data
set. The work done by disc cutters in soil is limited to crushing
boulders, enter the reception shafts and tackling possible other
hard obstacles along the tunnel alignment. In the disc cutter
changes included in Fig. 16 a common reason for changing the cutters were blocked bearings, assumed to be caused by soil clogging
around the tools.
4.3. Further work and conclusive remarks
The main ndings of this paper are:
 The Soil Abrasion Test (SAT) has a good correlation to
recorded soft ground tool life, thus the SAT value can be used
to schedule and price tool cost and downtime for soft ground
tunnelling projects.
 The SAT values correlates good with grain roundness and soil
mineralogy.
However, there are some limitations basing a TBM excavation
tool life estimate purely on abrasion testing of a disturbed soil
sample. The soil samples applicable for SAT testing do not contain water and the in situ compaction/density. The test is done
without soil conditioning additives, and the applicable grain size
for testing is limited to 04 mm. The ndings presented in Figs. 12
and 13 clearly show the effect of the grading of the soil sample,
which can be related to the in situ compaction of the soil. Thus,
in order to make the estimates more reliable, the SAT value
should be adjusted with cohesion/in situ density, potential inuence of boulders and use of excavation additives.
The inuence of cohesion/in situ density can be done by introducing geotechnical parameters or measurement for the soil, for
example the grade of compaction, Cu (ratio of grain size D60 and
D10). The parameters can then be included in a multivariate regression or as input into a Soil Abrasivity Index. The boulders inuence can be introduced by a statistical factor related to the
downtime to crush boulders into smaller fractions and how much
this can affect the tool life.
Testing on in situ like soils as suggested by Gharahbagh et al.
(2010) and Barzegari et al. (2013), is also an approach which
should be followed up. By including as many as possible soil
parameters, e.g. more representative grain size distribution, water
content, compaction and density together with possible excavation
additives, the measured test results may be more understandable
and less dependent on empirical relationships. NTNU/SINTEF is
currently working on a similar approach as suggested by Gharah-

bagh et al. (2010), in order to measure the inuence of soil compaction, excavation additives and the soil connement on the tool life
(Jakobsen et al., 2012).
In an estimator for tool life in TBM tunnelling, the inuence of
the TBM operator and tool replacement scheme will also inuence
the tool life. How to include the human inuence into the estimate
will be a challenge, but might be done, as more empirical data is
available.
A continuation of collecting soil samples for testing together
with corresponding TBM production and tool data is necessary to
build a better model for predicting tool life in soil and soft ground.
Acknowledgements
The authors would like to thank the following: The Norwegian
Tunnelling Society for nancing travels to tunnelling projects to
gather tool life data; Mr. Nakamura Toshiaki of Obayashi Corporation for providing a Japanese method of predicting tool wear in soil
TBM tunnelling; Master student Helge Ivar Frostad for programming a software for quick and precise instantaneous tool life calculations; and Dipl.-Ing. Tim Becker in Becker Engineering +
Consulting for his help and effort in gathering valuable eld data.
References
Babendererde, T., 2010. Difference in wear types and recognition in data
monitoring. BASF TBM Seminar, London, October 2010.
Babendererde, T., Elsner, P., 2009. Strait Crossings with TBMs requirements and
technical solutions. Presented at Strait Crossing Conference, Trondheim, 2009.
Barzegari, G., Uromeihy, A., Zhao, J., 2013. A newly developed soil abrasion testing
method for tunnelling using shield machines. Quarterly Journal of Engineering
Geology and Hydrogeology Volume 46, 6374.
Bruland, 1998a. NTNU-Anleggsdrift: Project Report 1E-98 HARD ROCK TUNNEL
BORING Performance Data and Back-Mapping. NTNU Trondheim.
Dahl, F., Bruland, A., Jakobsen, P.D., Grv, E., Nilsen, B., 2012. Classications of
properties inuencing the drillability of rocks. Tunnelling and Underground
Space Technology 28 (March), 150158.
Frentzel, C., Babendererde, T., 2011. Tool wear in TBM tunnelling. Tunneling Journal,
3643 (April/May).
Gharahbagh, E.A., Rostami, J., Gilbert, M., 2010. Tool wear issue in soft ground
tunneling, developing a reliable soil abrasivity index. In: North American
Tunneling Conference, 1923 June, Portland, OR.
Gharahbagh, E.A., Rostami, J., Palomino, A.M., 2011. New soil abrasion testing
method for soft ground tunneling applications. Tunnelling and Underground
Space Technology 26 (5), 604613.
Gong, Q.M., Zhao, J., Hefny, A.M., 2006. Numerical simulation of rock fragmentation
processes induced by two TBM cutters and cutter spacing optimization.
Tunnelling and Underground Space Technology 21, 18.
Home, L., 2010. Trends in the use of TBMs worldwide. Norwegian Tunnelling Society
TBM Seminar, Bergen Norway, June 2010.
Hutchings, I.M., 1992. Tribology, Friction and Wear of Engineering Materials.
Edward Arnold. ISBN: 9780340561843.
ISRM, 1978. Suggested methods for the quantitative description of discontinuities
in rock masses. International Journal of Rock Mechanics and Mining Sciences &
Geomechanics Abstracts 15 (6), 319369.
Jakobsen, P.D., Langmaack, L., Dahl, F., Breivik, T., 2012. Predicting the abrasivity of
in situ like soils. Tunnels & Tunnelling International, 4144 (June).
Khaligi, B., 2011. Technical webinar, Effective Design & Development of EPBMs.
Presented on Tunnelling Journal. <http://tunnellingjournal.com/webinar/
webinar-archives/>.
Nilsen, B., Dahl, F., Raleigh, P., Holzhuser, J., 2006a. Abrasivity of soils in TBM
tunneling. Tunnels & Tunnelling International, 3638 (March).
Nilsen, B., Dahl, F., Raleigh, P., Holzhuser, J., 2006b. Abrasivity testing for rock and
soils. Tunnels & Tunneling International, 4749 (April).
Nilsen, B., Dahl, F., Raleigh, P., Holzhuser, J., 2006c. SAT: NTNUs new soil abrasion
test. Tunnels & Tunneling International, 4345 (May).
Nilsen, B., Dahl, F., Holzhuser, J., Raleigh, P., 2007. The new test methodology for
estimating the abrasiveness of soils for TBM tunnelling. In: Rapid Excavation
and Tunneling Conference (RETC), pp. 104116.
Ozdemir, L., 2008. Seminar Presentation about Boulder Abrasion, BASF TBM
Seminar, Seattle.
Russ, J.C., 2006. The Image Processing Handbook, fth ed. CRC Press.
Thuro, K., Singer, J., Ksling, H., Bauer, M., 2007. Determining abrasivity with the
LCPC test. In: 1st CanadaUS Rock Mechanics, Symposium, pp. 827834.
Wood, R.J.K., 2009 Surfaces and contact mechanics, course material from Intensive
Course on Tribology, national Centre for Advanced Tribology at Southhampton
(nCATS). Trondheim, August 2009.

Paper 4
Tunnelling in abrasive soils review of a tunnel project in
Germany
Jakobsen, Pl Drevland
Becker, Tim E.E.
Presented at Korean Tunnelling Association, Mechanised
Tunnelling Symposium 2012

Tunnelling in abrasive soils review of a tunnel project in Germany

Pl Drevland Jakobsen1) and Tim E. E. Becker 2)
1)
2)

The Norwegian University of Science and Technology (NTNU), 7491 NO Trondheim Norway
BEC | Becker Engineering + Consulting, D-13595 Berlin

Abstract
Pipe jacking is a well-recognized method for mechanical excavation of small cross-section
tunnels with diameters from 0.25 to 4 m, with limited length from 50 to 1,200 m. The
excavation method is suitable in soft soil (clay, silt and sand), soft rock (UCS < 50 MPa) and
even hard rock (50 < UCS < 200). The pipe jacking advance is based on applying thrust on
pipe elements which are installed from a start shaft, oppositely to typical segmental lining
tunneling for larger diameters. The geotechnical conditions form a major factor for the
excavation rates which may vary between 50 m per day in favorable conditions and only a
few centimeters in worst cases. In unfavorable geotechnical conditions such as varying water
pressure, occurrence of boulders or abrasive soil conditions the performance can be highly
influenced. The aims of the paper are to (1) give a brief introduction to the pipe jacking
methodology, (2) introduce how unfavorable ground conditions can be pre-investigated, and
(3) how contractors can adapt the both the tendering and decision taking at site having the
necessary information about the ground conditions. The subjects are also relevant to
segmental lining tunnelling as well as for EPB and slurry shield TBMs.

Introduction
Pipe jacking, also called Microtunnelling is a commonly used method for excavation of
tunnels with small cross-sections (< 4 m diameter). The method is based on a moving front
shield which has a rotating cutterhead with a variety of cutter tools. The selection of cutter
tools is adapted to the expected ground conditions. Ripper tools are used in cohesion soils
(clay and silt), scraper tools are commonly used in friction soils (sand and gravel) and disc
cutters are installed on the cutter head when the tunnel alignment is subjected to boulders and
hard rock conditions. The advance of the pipe jacking shield is driven by static energy applied
to the jacking-pipes by a hydraulic cylinders located at the starting shaft. The jacking pipes,
which are usually made out of concrete, are inserted piece by piece at the starting shaft and
have a length of up to 4.0 m.

Figure 1 Scheme of pipe jacking project (starting and reception shaft, main and intermediate jacking stations
(marked as yellow). (Scheme: Herrenknecht)

The soil is removed by a rotating cutterhead (Figure 2) which is capable to loose the soil at
the shield front. The soil is then mixed with the transport slurry (bentonite suspension) and
pumped to the starting shaft and the ground surface to the separation plant unit. The
separation plant cleans the excavated soil from the bentonite slurry, which is recycled.

Figure 2 Newly manufactured pipe jacking machine, AVN 1600/1800 TB by Herrenknecht. (Photo: Tim E. E.
Becker)

The operation of the pipe jacking machine, also referred to as AVN1, includes the advance
rate of the hydraulic station, and all other machine parameters. The operation and continuous
monitoring of machine parameters are monitored from a control cabin located at the starting
shaft. Since all necessary works are performed around the starting shaft, pipe jacking is
regarded as very efficient in terms of required personnel, time consumption and capacity, for
excavation of small cross-section tunnels.

Figure 3 Typical pipe jacking set up). (Scheme: Herrenknecht)

The tunnel length between shafts gives the main limitation of pipe jacking. As there is a
maximum allowed jacking force that can be applied to the pipes, intermediate jacking stations
are installed in the tunnel. These small hydraulic stations are commonly activated in order to
reduce the actual pushing length of the tunnel. Typically a 2.5 m diameter machine would
require an intermediate jacking station every 50 to 200 m, depending on the pipe material and
ground conditions. The distance between starting and reception shafts can be as high as up to
1,200 m.
In order to keep the machine steerable, the rotating cutter head produces a small overcut
(annular gap) relatively to the actual outside diameter of the jacking pipes. During the
complete jacking period, the annular gap of the tunnel is constantly lubricated by a bentonite
suspension. This is done in order to support the annular gap, and secondly, in order to
minimize settlements at the ground surface. The bentonite injection also decreases the friction
between the pipes and the surrounding soil. The automatic lubrication system takes care of
injecting the bentonite suspension. The quality of the bentonite lubrication needs to be
checked in regular intervals and adapted if necessary.

1
AVN is a German abbreviation for Automatisches Vortriebsystem mit Nassfrderung, which in English means
automatic advance system with slurry.

The presented pipe jacking project in this paper was located in Hamburg, Germany. The
project name is Sammler Ost with Hamburg Wasser as the project owner and client.
The client, Hamburg Wasser, has a lot of experience in acquiring pipe jacking projects for the
city of Hamburgs water and wastewater ways.

General Project Preparation for Pipe Jacking

During the planning process of a tunneling project, the owner needs to determine the
requirement of the tunnel. Several purposes are typically required:

Gravitation line for rain, mixed (rain & sewage) water or sewage water
Pressure pipeline
Mantle pipe for pressure pipelines (gas, oil, water, etc.)
Mantle pipe for cables
Pipe for infrastructure ways (pedestrians)

These individual project boundary conditions enable the demands of the needed pipe
diameter, material, alignment and distance between shafts.

Figure 4 Reinforced concrete jacking pipe (inner diameter (ID) 2800 mm and outer diameter (OD) 3000 mm)
(Photo:Tim E. E. Becker)

The tender/ project documents form an integral part of the project and describe the contractual
boundaries for the contract between the client and the contractor. Within German contracts,
The German DWA Rules and Standards A 125 E for Pipe Jacking and Related Techniques
provides comprehensive guidelines for planning and construction that are:

Surface and underground structures and systems (e.g. foundations, underground

railways, existing networks as channels, cables, bunkers, anchors)
Soil and groundwater conditions
Supply and disposal facilities (e.g. water, waste water, power, roads/ streets)
Jacking distance and line (e.g. alignment and elevation plans including results of soil
survey)
Dimensions and material of the jacking pipes
Third party plannings and approvals
Environmental aspects (e.g. degradable hydraulic oil for machines, noise limitations)
Requirements by authorities (e.g. traffic regulation, working hours)

The careful preparation of the above mentioned documents represent an important duty of the
client in order to implement a proper call for bids, which are highly important to achieve a
successful project. A well-experienced engineer with expertise for such projects shall be able
to perform a differentiated evaluation of the bids with a sense of proportion between the
technical capability of the bidders and economical limitations of the client as shall be shown.
A thorough description of the soil and groundwater conditions is of great importance for
tendering contractors. As the client delivers the material soil to the project, it should be one
of his initial interests to produce a geological report including of the following parameters and
information (according to DWA-A 125 E, here for soft soils):

Maximum and minimum groundwater level and hydrograph curves

Contamination level of soil, soil gas and groundwater
Disposal advice according to legislation
Concentration of abrasive minerals and quartz content to determine abrasiveness
Deformation module
Aggressiveness reaction of soil and groundwater
Swelling potential
Borehole logs
Weight per unit volume
Fault zones and cavities
Particle size distribution and particle shape
Water permeability coefficient
Compactness
Plastic limits and water content
Shear parameter, friction angle and cohesion
Earth pressure coefficient
Cobble size and cobble proportion, uniaxial compressive strength
Water content and water pressure
Organic components and lime content
Tendency towards liquefaction

The above-mentioned parameters shall be combined in a geology report, which provides

sufficient information for a careful project preparation. Nevertheless, the deviation between
the geological report and the actual conditions found on site may lead to disputes and possible
claims between clients and contractors.

Tender documents of the Sammler Ost tunnel project in Hamburg,

Germany
The project was according to the German Construction Contract Procedures, thus the project
information consisted of a technical and a legal part, which is common in Germany:

Legal project information:

o Special Conditions of Contract
o Additional conditions of Contract and
o Standard Conditions of Contract
Technical project information:
o Technical Specifications
o Additional Technical Specifications

o Standard Technical Specification

Due to the experienced client, the tender documents were thoroughly prepared. The technical
specification contained information regarding:
1.
2.
3.
4.
5.
6.
7.
8.
9.

General Information
Information on project boundaries
Available construction site areas
Special area conditions
Execution of the works
a. Time schedule
b. Order of works
Materials and Parts
Documents for execution
Technical rules, checks and permits
Attachments (e.g. drawings and maps)

Additionally, the bill of quantities described all necessary works to be executed and specified
additional requirements to each element.
The documents establishing the formal framework for the pipe jacking works consisted of:

General overview
Ground view and longitudinal section (including borehole information) for each pipe
jacking drive
Technical requirements for all materials including pipes are defined in the Additional
Technical Specification
Geology report (as part of chapter 4. Special area conditions)

The geology report represented a major part of the project information. It contained detailed
information about soil parameters. The geology report did also include a recommendation for
machine type and technology for the Sammler Ost project. As the geology report was finished
before the new DWA-A 125 was finally published, it did not provide any information
regarding the concentration of abrasive minerals and quartz content, or any statement
referring to the level of abrasivity. In our opinion, these shortcomings led to great problems in
the later project process.

Project parameters of the Sammler Ost tunnel project in Hamburg,

Germany
Client:
Contractor:

Hamburg Wasser, Germany

Hans Lemme Hoch-, Tief- und Stahlbetonbau GmbH & Co. KG, Berlin,
Germany
Construction period: January 2008 September 2010
Works:
2,400 m sewage channel ID 1,600, OD 2,240, reinforced concrete
jacking pipes with inner PEHD-liner, length 3.50 m, 3 jacking lines
(910 m, 958 m and 532 m), curved drives with R = 1,000 500 m, 3
round jacking
10 m/ reception
7 m shafts by bore piles, depth up
to 20 m

Figure 5 Project overview of the three pipe jacking drives. (Layout plan by Hamburg Wasser with additions by
Tim E. E. Becker)

Factsheet for each drive:

Drive 1
Length: 910 m from shaft 2.2 to shaft 2.1
Alignment:
radius R = 1,000 m and 3,000 m
Slope:
straight 1.8 per mill
Depth:
20 m
Geology:
95% glacial drift, partly fluvial deposited sand and silt
Groundwater:
appr. 10-12 m above pipe invert level
Drive 2
Length: 958 m from shaft 2.2 to shaft 2.3
Alignment:
straight
Slope:
straight 1.8 per mill
Depth:
20 m
Geology:
70% glacial drift, rest fluvial deposited sand, fine sand and basin silt
Groundwater:
appr. 10-12 m above pipe invert level
Drive 3
Length: 532 m from shaft 2.4 to shaft 2.3
Alignment:
R = 500 m and 600 m
Slope:
straight 1.8 per mill
Depth:
12 m
Geology:
20% fluvial deposited sand, 80% fluvial deposited fine sand and basin silt
Groundwater:
appr. 4-10 m above pipe invert level

Figure 6 The Sammler Ost Tunnel (Photo: Tim E. E. Becker)

The geology report recommended the use of a pipe jacking machine with slurry transport of
the AVN type. The contractor, Hans Lemme Hoch-, Tief- und Stahlbetonbau GmbH & Co.
KG had a newly ordered and manufactured machine AVN 1600/1800 TB by Herrenknecht
(Figure 2). The AVN machine was designed with a center door to get access to the shield
front, and to the cutter tools as well as of an integrated air pressure regulation station. These
provisions were essential for the later success of all three drives.

Figure 7 Photo of the cutter head of the Herrenknecht pipe jacking machine AVN 1600/1800.
(Photo: Tim E. E. Becker)

Execution of the Sammler Ost project

Figure 8 shows the pipe jacking performance for all three drives in the Sammler Ost project.
By comparing the performance in the three drives, one can see that the performance is varying
even though the excavation is done by the same contractor, with the same equipment, and
with experienced personnel in similar pre-investigated soil properties.
Drive 1 was completed with reasonable progress rates and without any interruptions. The
excavated glacial drift produced additional separation effort and stones and boulders were
passed successfully without damages or noticeable wear on the cutter head and its tools.

The following Drive 2 started similar as Drive 1. However, when the machine reached the
shaft wall at shaft 2.3, the machine wasnt able to pass the concrete bore piles of the shaft
wall. It was then decided to enter the cutter head chamber under air pressure (up to 1.2 bars)
in order to check the condition of the tools. The cutter head and the tools were in a disastrous
state, resulting in immediately replacement of 5 cutter discs in order to be able to enter the
reception shaft. The damage of the cutter head structure was tremendous, resulting in a full
replacement of the cutter head before Drive 3.

Figure 8 Summary of production rates for the three tunnels in the reviewed tunnel project.

Figure 9 Damaged cutter head after the second drive at the Sammler Ost project.
(Photo:Tim E. E. Becker)

Figure 10 Photo showing blocked and destroyed double disc cutter (up left), worn out ripper tools (up middle),
worn out scraper tools (up right), reduced cutter head diameter (down left) and worn out front of cone
crusher (down right). (Photos: Tim E. E. Becker)

Drive 3 was completed successfully and without any problems.

Project Aftermath and Discussion

In order to show, and try to understand the reasons for the extraordinary problems that did
occur at Sammler Osts second drive, a set of tests were done at NTNU/SINTEF. The testing
comprised measuring soil abrasivity with the Soil Abrasion Test (SAT), grain mineralogy
by x-ray diffraction (XRD) and Differential Thermal Analysis (DTA), grain shape, grain size
distribution.
Table 1 Summary of test results from the Sammler Ost project
Sample
#62
#57
#36
#90

SAT value
9
18.5
20
23.5

Quartz content
24
37
N.A.
38

Soil type
Silt
Silt
Silt
Silt

Cu
4.11
1.6
5.33

Figure 11 Photo of typical grain shape of the sand and silt sections in the analyzed project (orange grid is
1/10 mm)

The SAT values indicate that the soil samples has medium to high abrasivity according to
Jakobsen and Dahl (2010). The failure of the tools occurred after approximately 950 m of
tunnelling which is equal to approximately 3,600 sm3 (solid cubic meter) of soil. The cutter
head was fitted with a total of 10 double disc cutters, 8 scraper tools and 10 ripper tools. This
indicates an average tool life of 130 sm3/tool2. This is a lower tool life that would be expected
by studying Figure 12. However, by comparing it with Figure 13 and Figure 14 the Sammler
Ost tool life corresponds better to the trend line.

Figure 12 Exponential relation between SAT values and cutter tool life (ripper, scrapers drag-bits and disc
cutters) (Jakobsen and Dahl 2010)

In Figure 12 the tool life is presented as tool life per solid cubic meters. The points with tool
life above 1,500 sm3 are dragbits mounted on a cutter head spoke. One spoke typically has 5
20 tools dependent on the size of the machine. Therefore a tool life of 14 000 sm3/c may refer
to 14,000 sm3/c / 5 or 10. The data in Figure 12 also includes disc cutters mounted on TBMs.
Figure 13 and Figure 14 shows updated relations between the SAT value and recorded tool
life (soft ground tools, thus disc cutters are not included).
At the time the wear problems occurred at the Sammler Ost project, the limited amount of
data presented in Figure 12 was available for relating SAT values into tool life. Later, an
extensive data collection from 9 projects, Figure 13 and Figure 14 shows the current empirical
relation between SAT and tool life, based on tool life data from 9 TBM and pipe jacking
projects (TBM diameter < 5.5 m).

Tools includes disc cutters, scrapers and rippers in this calculation.

Figure 13 Correlation between SAT values and recorded soft ground tool life (to be published by (Jakobsen
et al. under review)

The two outliers (marked as triangles) in Figure 13 are from two projects where the SAT
values were high (25.5 and 33.3) and with high tool life. The soils at the two specific projects
were single graded sand without fines. Such soils are very easy to excavate as it is almost
flowing by itself, saving the TBM tools from high contact forces and impacts. The grain size
distribution and cohesive properties of the soils are not taken into account in the SAT test,
although it seems to have a high importance when making a soft ground and soil tool life
estimation.
By removing the two projects from the data set the R2 increases and the correlation becomes
better. The increase of the R2 by removing to points also shows the sensitivity of the current
correlation when introducing new data. It should therefore be emphasized that the current
correlation is based on relatively few data and by later adding tool life data and SAT values
from other project the relation may be significantly changed.

Figure 14 Correlation between SAT values and recorded soft ground tool life. (loose compacted material
removed from the dataset).

The problems, that were observed in the Sammler Ost project led to conflicts between the
involved parties regarding of who had to take care of the tremendous costs which resulted in
the abrasivity of the soil. The client persisted that all necessary information was given in the
provided geology report and that the abrasivity, that was met, did not exceed the commonly
expected value. The contractor Lemme, which performed many projects in similar geology,
had until then never experienced such wear on cutter tools and took up the position, that such
high abrasivity was not be expected by the information of the geology report. Finally, a
compromise finished the dispute. Latest tenders by Hamburg Wasser include information
according to DWA-A125:2008.

Conclusive remarks
Pipe jacking is a common method for the excavation of tunnels. A thorough project
preparation by the client is essential in order to obtain viable and comparable offers by the
contractors. Comprehensive guidelines for the preparation and execution of such projects are
given e.g. in the German DWA Rule and Standard A 125 for Pipe Jacking and Related
Techniques.
The client provides the soil as a material to the project so that it should be described as
detailed as possible in a geology report. Experiences show that the abrasivity of the soil may
result in major wear of the cutter tools and the cutter head. As presented in the Sammler Ost
project, missing parameters of the abrasivity that ought to be expected, led to significant extra
costs, and conflict between the client and the contractor.
Latest research may provide values for the abrasivity which quality is so promising that they
may give a conclusion to the expected in-situ parameters and actual wear. Contractors are
now able to make a more precise estimation of the expected cutter tool wear, and are therefore
able to define the necessary intervals of tool inspection and/ or tool changes and its costs. This
helps in obtaining a better quality of the contractors bid calculation, and avoiding arguments
about additional costs.
The use of the exponential relations between SAT values and recorded tool life for tool life
prediction should be used with caution. The SAT values from Sammler Ost correspond
somewhat to the recorded tool life for drive 2. The reason is that a simplified abrasion test
only takes into account the abrasivity on a cohesionless powder, which is not close to a soils
in-situ properties. After the completion of the Sammler Ost project the contractor moved on to
excavate other projects in Germany with the same abrasive soil (measured by the SAT test),
but did not encounter any wear problems.
On the other hand, the empirical relations are still quite promising. This means that further
research should be done to include other in-situ parameters into the tool life estimate (e.g. insitu soil density). Due to this, and the relative low amount of data in the correlation charts
tunnel excavations should not be scheduled without any interventions.
To have a safe excavation the following schedule for cutter head interventions have been
suggested (Babendererde 2010):
First directly after the shaft wall if the TBM is launched from a shaft
50 75 m later
Every 150 m (2-3 times)
Then as frequent as necessary based on the experience from the first km or after
change in geology/geotechnical properties of the soil.
This approach may be considered to be time consuming, but it would only involve some hours
downtime (for the intervention itself).
Since the initial findings presented in Figure 12, our research indicates that the influence of
soil compaction and density influences the tool life. A loose compacted soil (e.g. single
graded sand) would most likely not cause severe problems even though the SAT value is
high. Opposite, if the SAT value is in the medium range and the in-situ density of the soil is
high (above 2000 kg/m3) the wear problems may be severe. The influence of the soils in-situ

properties on the TBM tool life is the scope of current research activities at NTNU/SINTEF,
in close cooperation with the tunnelling industry.

References
Babendererde, T. (2010). "Difference in wear types and recognition in data monitoring."
BASF TBM Conference, London 28th Oct. 2010.
Jakobsen, P. D., Bruland, A. and Dahl, F. (under review). "Review and assessment of the
NTNU/SINTEF Soil Abrasion Test (SAT(TM)) for determination of abrasiveness of
soil and soft ground." Tunnels and Underground Space Technology.
Jakobsen, P. D. and Dahl, F. (2010). "Soil Abrasion in TBM Tunnelling." Korean Tunnelling
Association International Symposium on Mechanized Tunnelling Technology 11: 8.
StandardDWA-A 125 E "Pipe Jacking and Related Techniques", German Association for
Water, Wastewater and Waste, December 2008

Paper 5
Overview of pipe-jacking performance review of tunnel projects
Becker, Tim E.E.
Jakobsen, Pl Drevland
Presented at the NO-DIG convention Berlin, 2013

1st NO DIG BERLIN 2013

Conference and Exhibition
23 - 26. April 2013

Paper 2A-2

Overview of pipe-jacking performance review of tunnel projects

Tim E. E. Becker
BEC Becker Engineering + Consulting GmbH, Berlin, Germany
Pl Drevland Jakobsen
The Norwegian University of Science and Technology (NTNU), Trondheim, Norway
Pipe jacking is a well-recognized method for mechanical excavation of small crosssection tunnels with diameters from 0.25 to 4 m, with limited length from 50 to 1200
m. The excavation method is suitable in soft soil (clay, silt and sand), soft rock
(Uniaxial Compressive Strength (UCS) < 50 MPa) and even hard rock (50 < UCS <
200). The pipe jacking advance is based on applying thrust on pipe elements which
are installed from a start shaft, oppositely to typical segmental lining tunneling for
larger diameters. The geotechnical conditions form a major factor for the excavation
rates which may vary between 50 m per day in favorable conditions and only a few
centimeters in worst cases. In unfavorable geotechnical conditions such as varying
water pressure, occurrence of boulders or abrasive soil conditions the performance
can be highly influenced. This paper gives a brief introduction to the pipe jacking
methodology, general project preparation for pipe jacking, and it shows recorded
performance and tool life from 7 projects, with a total length of approximately 7,400
m. The subjects are also relevant to segmental lining tunnelling as well as for EPB
and slurry shield TBMs.

Introduction
Pipe jacking, also called Microtunnelling is a commonly used method for tunnel
excavation with small cross-sections (< 4 m diameter). The method is based on a
moving front shield with a rotating cutterhead containing a variety of cutter tools. The
selection of cutter tools is adapted to the expected ground conditions. Ripper tools
are used in cohesion soils (clay and silt), scraper tools are commonly used in friction
soils (sand and gravel) and disc cutters are installed on the cutter head when the
tunnel alignment is subjected to boulders and hard rock conditions. The advance of
the pipe jacking shield is driven by static energy applied to the jacking-pipes by a
hydraulic cylinders located at the starting shaft. The jacking pipes, which are usually
made out of concrete, are inserted piece by piece at the starting shaft and have a
length of up to 4.0 m.

-1-

Figure 1 Scheme of pipe jacking project (starting and reception shaft, main and intermediate
jacking stations (marked as yellow). (Scheme: Herrenknecht AG)

The soil is removed by a rotating cutterhead (Figure 2) which is capable to loose the
soil at the shield front. The soil is then mixed with the transport slurry (bentonite
suspension) and pumped to the starting shaft and the ground surface to the
separation plant unit. The separation plant cleans the excavated soil from the
bentonite slurry, which is reused.

Figure 2 Newly manufactured pipe jacking machine, AVN 1600/1800 TB by Herrenknecht AG.
(Photo: Tim E. E. Becker)

The operation of the pipe jacking machine, also referred to as AVN 1, includes the
advance rate of the hydraulic station, and all other machine parameters. The
operation and continuous censoring of machine parameters are monitored from a
1

AVN is a German abbreviation for Automatisches Vortriebsystem mit Nassfrderung, which in

English means automatic advance system with slurry.

-2-

control cabin located at the starting shaft. Since all necessary works are performed
around the starting shaft, pipe jacking is regarded as very efficient in terms of
required personnel, time consumption and capacity, for excavation of small crosssection tunnels.

Figure 3 Typical pipe jacking set up). (Scheme: Herrenknecht AG)

The tunnel length between shafts gives the main limitation of pipe jacking. As there is
a maximum allowed jacking force that can be applied to the pipes, intermediate
jacking stations are installed in the tunnel. These small hydraulic stations are
commonly activated in order to reduce the actual pushing length of the tunnel.
Typically a 2.5 m diameter machine would require an intermediate jacking station
every 50 to 200 m, depending on the pipe material and the ground conditions. The
distance between starting and reception shafts can be as high as up to 1200 m.
In order to keep the machine steerable, the rotating cutter head produces a small
overcut (annular gap) relatively to the actual outside diameter of the jacking pipes.
During the complete jacking period, the annular gap of the tunnel is constantly
lubricated by a bentonite suspension. This is done in order to support the annular
gap, and secondly, in order to minimize settlements at the ground surface. The
bentonite injection also decreases the friction between the pipes and the surrounding
soil. In order to achieve an optimized lubrication, use of an automatic lubrication
system is utilized for injecting the bentonite suspension. However, it is required to
check the quality of the bentonite lubrication in regular intervals, and if necessary
adapt the quantity and quality of the bentonite suspension.

-3-

General Project Preparation for Pipe Jacking

During the planning process of a tunneling project, the owner needs to determine the
requirement of the tunnel. Several purposes are typically required:
x
x
x
x
x

Gravitation line for rain, mixed (rain & sewage) water or sewage water
Pressure pipeline
Mantle pipe for pressure pipelines (gas, oil, water, etc.)
Mantle pipe for cables
Pipe for infrastructure ways (pedestrians)

These individual project boundary conditions enable the demands of the needed pipe
diameter, material, alignment and distance between shafts.

Figure 4 Reinforced concrete jacking pipe (inner diameter (ID) 2800 mm and outer diameter
(OD) 3000 mm) (Photo:Tim E. E. Becker)

The tender/ project documents form an integral part of the project and describe the
contractual boundaries for the contract between the client and the contractor. Within
German contracts, The German DWA Rules and Standards A 125 E for Pipe Jacking
and Related Techniques provides comprehensive guidelines for planning and
construction that are:
x
x
x
x
x
x
x
x

Surface and underground structures and systems (e.g. foundations,

underground railways, existing networks as channels, cables, bunkers,
anchors)
Soil and groundwater conditions
Supply and disposal facilities (e.g. water, waste water, power, roads/ streets)
Jacking distance and line (e.g. alignment and elevation plans including results
of soil survey)
Dimensions and material of the jacking pipes
Third party plannings and approvals
Environmental aspects (e.g. degradable hydraulic oil for machines, noise
limitations)
Requirements by authorities (e.g. traffic regulation, working hours)

-4-

The careful preparation of the above mentioned documents represent an important

duty of the client in order to implement a proper call for bids, which are highly
important to achieve a successful project. A well-experienced engineer with expertise
for such projects shall be able to perform a differentiated evaluation of the bids with a
sense of proportion between the technical capability of the bidders and economical
limitations of the client as shall be shown according to Deutsche Vereiningung fr
Wasserwirtshaft (2008) (DWA-A 125 E).
A thorough description of the soil and groundwater conditions is of great importance
for tendering contractors. As the client delivers the material soil to the project, it
should be one of his initial interests to produce a geological report including of the
following parameters and information (according to DWA-A 125 E, here for soft soils):
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x

Maximum and minimum groundwater level and hydrograph curves

Contamination level of soil, soil gas and groundwater
Disposal advice according to legislation
Concentration of abrasive minerals and quartz content to determine
abrasiveness
Deformation module
Aggressiveness reaction of soil and groundwater
Swelling potential
Borehole logs
Weight per unit volume
Fault zones and cavities
Particle size distribution and particle shape
Water permeability coefficient
Compactness
Plastic limits and water content
Shear parameter, friction angle and cohesion
Earth pressure coefficient
Cobble size and cobble proportion, uniaxial compressive strength
Water content and water pressure
Organic components and lime content
Tendency towards liquefaction

The above-mentioned parameters shall be combined in a geology report, which

provides sufficient information for a careful project preparation. Nevertheless, the
deviation between the geological report and the actual conditions found on site may
lead to disputes and possible claims between clients and contractors.
As shown above, the new DWA-A 125 demands the determination of parameters
apparently relevant to abrasion during pipe jacking which are amongst others:
x Concentration of abrasive minerals and quartz content to determine
abrasiveness
x Particle size distribution and particle shape

-5-

Recently, clients include such references in their geology reports but still leave the
estimation of abrasive wear up to the bidder / contractor. Mineral composition, grain
curve, grain shape, density and water content are important factors to determine the
abrasiveness of the soil. At the moment, there are no recognized prediction models
taking all this parameters into account. However, a prediction model for estimating
tool life for pipe jacking and soft ground TBMs are under development at NTNU.
The following presented results of the performed research at NTNU shall be able to
reduce the calculative risk by estimating the expected wear of cutter tools and the
necessity of tool changes during pipe jacking operation.
Estimation of abrasive wear
For all the 7 reviewed projects presented in this paper, abrasiveness properties have
been measured by the NTNU/SINTEF Soil Abrasion Test (SAT) (Nilsen et al.
2007)
and
(Holzhuser
and
Nilsen
2006).
Figure 5 shows the schematic overview of the test apparatus, which consists of a
rotating steel disc, which is fed with soil grains less than 4 mm. As the soil is
transported on the rotating disc it passes underneath a steel sample, originating from
a TBM disc cutter. This contact causes abrasive wear on the steel sample.

Figure 5 Schematic overview of the SAT test procedure, (Nilsen et al. 2007)

In addition to the SAT test, quantification of quartz content by Differental Thermal

Analyses (DTA) has been executed except for Project No. 7, see
Figure 6.

-6-

Figure 6 Comparision of SAT values and quartz content on samples from the reviewed pipe
jacking projects.

The SAT values has been correlated to the grain sample mineralogy, represented
by the Vickers Hardness Number Rock (VHNR) and the grain roundness, see
Figure
7
and
Figure 8. The VHNR is determined combining the percentage of each mineral and its
corresponding Vickers Hardness in a soil sample (Dahl et al. 2012).

Figure 7 SAT values correlated with VHNR showing a linear relation between the measured
abrasivity (SAT) and the mineralogy (VHNR)

-7-

Figure 8 SAT values correlated with grain roundness showing a linear relation between the
measured abrasivity (SAT) and the grain shape. A grain roundness of 1.0 indicates a perfect
circular grain.

The reviewed projects

A total of 7 projects is reviewed in this paper. Projects No. 6 and 7 are finished pipejacking tunnels in the Middle-East, where the Client and contractors have asked to
keep site-specific data confidential. Thus, the amount of information from Projects
No. 6 and 7 is less than for Projects No. 1 5.
Project No. 1 (Sammler Ost, 2. BA)
Client:
Contractor:

Hamburg Wasser, Germany

Hans Lemme Hoch-, Tief- und Stahlbetonbau GmbH & Co. KG,
Berlin, Germany
Construction period: January 2008 September 2010
Works:
2,400 m sewage channel ID 1,600, OD 2,240, reinforced
concrete jacking pipes with inner PEHD-liner, length 3.50 m, 3
jacking lines (910 m, 958 m and 532 m), curved drives with
R = 1,000 500 m, 3 round jacking 10 m/ reception 7 m
shafts by bore piles, depth up to 20 m
Drive 1
Length:
Alignment:
Slope:
Depth:
Geology:
Groundwater:

910 m from shaft 2.2 to shaft 2.1

radius R = 1,000 m and 3,000 m
straight 1.8 per mill
20 m
95% glacial drift, partly fluvial deposited sand and silt
appr. 10-12 m above pipe invert level

-8-

Drive 2
Length:
Alignment:
Slope:
Depth:
Geology:
Groundwater:
Drive 3
Length:
Alignment:
Slope:
Depth:
Geology:
Groundwater:

958 m from shaft 2.2 to shaft 2.3

straight
straight 1.8 per mill
20 m
70% glacial drift, rest fluvial deposited sand, fine sand and basin
silt
appr. 10-12 m above pipe invert level
532 m from shaft 2.4 to shaft 2.3
R = 500 m and 600 m
straight 1.8 per mill
12 m
20% fluvial deposited sand, 80% fluvial deposited fine sand and
basin silt
appr. 4-10 m above pipe invert level

Project No. 2 (Nebensammler Bergedorf, Abschnitt Billbrook)

Client:
Contractor:

Hamburg Wasser, Germany

Joint Venture Nebensammler Bergedorf
Hans Lemme Hoch-, Tief- und Stahlbetonbau GmbH & Co. KG,
Berlin, Germany and Grund- und Sonderbau GmbH, Berlin,
Germany
Construction period: June 2009 June 2012
Works:
2,100 m sewage channel ID 1,600, OD 2,240, reinforced
concrete jacking pipes with inner PEHD-liner, length 2,00 3.50 m, 3 jacking lines (740 m, 660 m and 700 m), curved drives
with R = 600 300 m, 1 round jacking shaft 10 m, one
reception shaft 8 m and one rectangular shaft, all by bore piles
(one additional pipe jacking with open shield under air pressure
into existing operation shaft, appr. 15 m, sewage channel
ID 1,600, OD 2,240)
Drive 1
Length:
Alignment:
Slope:
Depth:
Geology:
Groundwater:

740 m from shaft BS1 to shaft BS2

radius R = 500 m
straight 0.33 per mill
16 m
fluvial deposited middle and fine sand, partly organic silt
appr. 10-11 m above pipe invert level

-9-

Drive 2
Length:
Alignment:
Slope:
Depth:
Geology:
Groundwater:

660 m from shaft BS3 to shaft BS2

radius R = 600 m - 500 m
straight 0.33 per mill
15 m
fluvial deposited middle and fine sand, partly coarse sand and
organic silt
appr. 10-11 m above pipe invert level

Drive 3
Length:
Alignment:
Slope:
Depth:
Geology:
Groundwater:

700 m from shaft BS3 to shaft BS4

radius R = 300 m
straight 0.33 per mill
14 m
fluvial deposited middle and fine sand, partly clay
appr. 10 m above pipe invert level

Project No. 3 (Harbour Undercut Lubmin, Germany)

Client:
Contractor:

Wingas GmbH & Co. KG, Kassel

Hans Lemme Hoch-, Tief- und Stahlbetonbau GmbH & Co. KG,
Berlin, Germany
Construction period: August 2009 May 2010
Works:
2x 190 m mantle pipe ID 2,400, OD 3,000 for high pressure gas
pipeline ID 1400 (OPAL), reinforced concrete jacking pipes,
length 4,00 m, straight alignment and gradient, one round
jacking shaft 22 m, one reception shaft 14 m
Length:
Alignment:
Slope:
Depth:
Geology:
Groundwater:

2x 190 m parallel, light distance 3,00 m

straight
straight in level
22 m
glacial drift, sandy, peat clay
appr. 21 m above pipe invert level

Project No. 4 (Undercut of River Lcknitz, Kienbaum, Germany)

Client:
Contractor:

Wingas GmbH & Co. KG, Kassel

Hans Lemme Hoch-, Tief- und Stahlbetonbau GmbH & Co. KG,
Berlin, Germany
Construction period: May 2010 October 2010
Works:
365 m mantle pipe ID 2,400, OD 3,000 for high pressure gas
pipeline ID 1400 (OPAL), reinforced concrete jacking pipes,
length 4,00 m, straight alignment and curved gradient, one
rectangular jacking shaft (sheet piles), machine recovery near to
surface at reception point

- 10 -

Length:
Alignment:
Slope:
Depth:
Geology:
Groundwater:

365 m
straight
radius R = 2,100 m
up to 11 m
fluvial deposited middle and fine sand, partly organic silt
appr. 10 m above pipe invert level

Project No. 5 (Undercut of Channel Oder-Havel, Oderberg, Germany)

Client:
Contractor:

Wingas GmbH & Co. KG, Kassel

Hans Lemme Hoch-, Tief- und Stahlbetonbau GmbH & Co. KG,
Berlin, Germany
Construction period: June 2010 May 2011
Works:
190 m mantle pipe ID 2,400, OD 3,000 for high pressure gas
pipeline ID 1400 (OPAL), reinforced concrete jacking pipes,
length 4,00 m, straight alignment and curved gradient, one
rectangular jacking shaft and one rectangular reception shaft
(sheet piles)
Length:
Alignment:
Slope:
Depth:
Geology:
Groundwater:

190 m
straight
radius R = 2,100 m
up to 13 m
fluvial deposited middle and fine sand
appr. 11 m above pipe invert level

Project No. 6 (Located in the Middle-East. Ocean outfall)

Length:
Alignment:
Slope:
Depth:
Geology:
Groundwater:

1230 m
Straight
NA
up to 18 m
Fine single graded sand with high quartz content
Subsea with overburden from 18 to 0 m.

Project No. 7 (Located in the Middle-East)

Length:
Alignment:
Slope:
Depth:
Geology:
Groundwater:

710 m
NA
NA
Appr. 9 15 m
silt, sand and gravel
partly

- 11 -

Recorded performance and tool life at the reviewed projects

In the following, an overview of the recorded performance and tool life for the 7
projects will be given.
Performance
The project with the best performance (highest gross production per day) is Project 3
- Drive 2. The geology along Project 3 is a combination of marine clay with some
sand. The geology was uniform along the drive, meaning that the pipe jacking
machine could be operated at more or less constant thrust, face support and
bentonite injection. Project 3 Drive 1 had the same geological conditions, thus with
lower performance. The reason for this is assumed to be learning curve of the crew
operating the pipe jacking machine at both drives.
The project with the lowest performance is Project 1 Drive 2. The first major
downtime is due to Christmas holiday, which was scheduled. The second major
downtime was due to worn out tools, which were an unscheduled downtime.
For Project 2, 4, 5, 6 and 7 the performance has been good and in relatively stable
geological conditions. The soft ground has consisted mainly of single graded sand,
which is easy to excavate, with some intrusions of finer material (silt) and coarser
material (gravel).

Figure 9 Overview of the performance for the reviewed projects.

- 12 -

Tool life
The recorded tool life is varying from approximately 50 to 1850 sm 3/tool. The unit
sm3/tool is chosen as it makes it easier to compare tool life from varying excavation
diameters. The recorded tool life has been correlated with the SAT value and the
quartz
content,
see
Figure
10
and
Figure 11.

Figure 10 SAT correlated with excavation tool life for 7 projects.

- 13 -

Figure 11 Quartz content correlated with excavation tool life for 6 projects (Project No 7 is not
included due to missing measurements of quartz content).

The relation between the SAT value and tool life, as well as quartz content and tool
life is relative poor for the current data set. It has been shown in a recent paper, by
including the recorded tool life for other projects (slurry shield machines and EPB)
the correlation is found better (Jakobsen and Becker 2012).
In order trying to achieve a better correlation for the 7 pipe jacking projects, the
geotechnical index Cu has been used to adjust the SAT values. The Cu index
indicates the uniformity of a soil sample, and is calculated by.

Cu

d 60 d 75
|
d10 d 25

In
Figure 12, a very good correlation between the SAT value multiplied with C u and
the tool life is shown. One likely explanation for the improvement of the correlation, is
that simplified abrasion tests are done on a cohesionless material lacking a lot of
the in-situ properties of the soil. A low Cu index indicates a uniform soil, which is
relatively easy to excavate and oppositely a high Cu can indicate that higher thrust
and torque is needed to excavate a soil.

- 14 -

Figure 12 Correlation between the SAT value multiplied with Cu correlated with excavation
tool life.

Conclusive remarks
Pipe jacking is a common method for the excavation of tunnels. A thorough project
preparation by the client is essential in order to obtain viable and comparable offers
by the contractors. Comprehensive guidelines for the preparation and execution of
such projects are given e.g. in the German DWA Rule and Standard A 125 for Pipe
Jacking and Related Techniques.
The main findings of this paper are:
x Pipe jacking performance is highly variable and influenced on the ground
conditions.
x Gross production varies from 4.75 m/day to 17 m/day in the reviewed projects.
x Pipe jacking excavation tool life highly variable. For the reviewed projects, a
combination of an abrasivity measurement (in this paper by the Soil Abrasion
Test (SAT) and the geotechnical index Cu seems to provide a good
estimate. However, the current estimate is only based on few data originating
from only 7 projects in silt and sand. Thus, further studies and data collection
is needed to validate this finding, and also to include gravely soils.

- 15 -

References
Dahl, F., Bruland, A., Jakobsen, P. D., Nilsen, B. and Grov, E. (2012).
"Classifications of properties influencing the drillability of rocks, based on the
NTNU/SINTEF test method." Tunnelling and Underground Space Technology
28: 150-158.
Deutsche Vereiningung fr Wasserwirtshaft, A. u. E. e. V. (2008). Pipe Jacking and
Related Techniques. German DWA Rules and Standards DWA-A 125E. A. u.
E. e. V. Deutsche Vereiningung fr Wasserwirtshaft. Hennef, Germany, DWA:
70.
Holzhuser, J. and Nilsen, B. (2006). "Abrasivitt von Lockergestein bei
Tunnelvortrieben Aktuelle Erfahrungen aus der Baupraxis und Entwicklung
eines neuen Laborversuchs." Baugrundtagung, Bremen.
Jakobsen, P. D. and Becker, T. E. E. (2012). "Tunnelling in abrasive soils - review of
a tunnel project in Germany." The 13th KTA International Symposium on
Mechanized Tunnelling Technology, Seoul.
Nilsen, B., Dahl, F., Raleigh, P. and Holzhuser, J. (2007). "The New Test
Methodology for Estimating the Abrasiveness of Soils for TBM Tunnelling."
Proceedings of the Rapid Excavation and Tunneling Conference (RETC): 104
- 106.

- 16 -

Paper 6
Challenges of Methods and Approaches for Estimating Soil
Abrasivity in Soft Ground TBM Tunnelling
Jakobsen, Pl Drevland
Lohne, Jardar
Published in WEAR Journal 2013

Wear ()

Contents lists available at SciVerse ScienceDirect

Wear
journal homepage: www.elsevier.com/locate/wear

Challenges of methods and approaches for estimating soil abrasivity

in soft ground TBM tunnelling
P.D. Jakobsen n, J. Lohne
The Norwegian University of Science and Technology, Department of Civil and Transport Engineering, Hgskoleringen 7A, 7491 Trondheim, Norway

a r t ic l e i n f o

a b s t r a c t

Article history:
Received 14 December 2012
Received in revised form
19 June 2013
Accepted 20 June 2013

The increase of urbanization constantly demands more infrastructure, which often requires to utilize the
underground. The complex functioning of Tunnel Boring Machines (TBM) and the complexity of their
working environment make arriving at wear predictability a challenging affair. The economic consequence of tool wear is, on the other hand, signicant. There are several methods and approaches to
estimate geo-material's abrasion properties in tunnelling with TBMs. The methods are mainly based on
using empirical observations and experiences from completed tunnelling projects correlated with
simplied laboratory test results. The simplied laboratory tests means that the soil's in-situ properties
such as the cohesion, density, adhesion, water content and original grain size distribution are lacking or
is disturbed. The intention of this research paper is to give a brief overview the respective test procedures
of the different approaches and their pros and cons. In light of these, we assess a new methodological
approach and examine the extent to which it is capable of advancing the understanding of predicting
wear on TBMs with respect to general tribological experiences such as the inuence of hard minerals or
particles, grain shape and grain size. The applicability of the laboratory experiments from a practitioner's
point of view is equally briey discussed.
& 2013 Elsevier B.V. All rights reserved.

Keywords:
Wear
Abrasivity
TBM
Tunnelling

1. Introduction
Full face tunnelling with tunnel boring machines (TBMs) is a
well-known and widely utilized method to excavate tunnels. 60
80% of tunnels build in the world today are excavated by TBMs [1].
According to Home [2] the total amount of TBMs working to
excavate soil the last 5 years is approximately 350 units, with
diameters ranging from approximately 3 m to over 15 m (Fig. 1).
In order to excavate the soil and soft ground, a combination of
rotating disc cutters, scraper tools and ripper tools are installed on
the TBMs cutterhead. Generally, the disc cutters are designed to
apply a high thrust force (approximately 300 kN per disc) [3] in to
rock mass, inducing tensile failures and chipping from rock, while
scraper tools are designed to scrape coarse soil from the tunnel
face and ripper tools are ripping cohesive soil material (clay and
silt). These tools are exposed to relative high contact forces with
abrasive material causing various degradation processes such as
abrasive wear (which is the most pre-dominant), impact wear and
chipping. Experiences from tunnel projects excavated with hard
rock TBMs during the Norwegian hydropower era (1980 s), show
that the abrasion properties of the rock mass can inuence
excavation cost and time in the range of 730%.

Corresponding author. Tel.: +47 924 338 33; fax: +47 735 970 21.
E-mail addresses: pal.jakobsen@ntnu.no, paldrevl@gmail.com (P.D. Jakobsen)

The direct consequence of tunnelling in abrasive soil conditions

is an increased demand for replacement of the excavation tools
(disc cutters, scrapers and rippers). The disc cutters are equipped
on TBMs if the rock mass exceeds a compressive strength of
20 MPa [4]. However, disc cutters are quite often installed to cope
with boulders (large rock blocks) inside a soil matrix as well.
At the time being there is no recognized standard method or
approach to estimate tool life in soil and soft ground TBM
tunnelling. This is not surprising given the variety of determining
factors, such as the cost of tools varying depending on size,
manufacturercustomer relations, material quality and type. Generally speaking, the complex functioning of TBMs and the complexity of their working environment make arriving at wear
predictability a challenging affair. The economic consequence of
tool wear is, on the other hand, so signicant that the need for
such predictability is pressing. In addition to the direct cost for
tools, the down time of a TBM, unproductive time for workers and
possible delay of the tunnelling project should be included. In a
recent hard rock TBM project in Hong Kong it has been stated 1 s
downtime has the cost of USD 1, meaning that one working shift's
downtime can cost as much as USD 30,000 excluding the direct
cost for the tools [5].
Today, tunnelling works are in the most cases based on tender
documents with pre-investigations of the soil or rock along the
tunnel drive, and various contractors then have to assess the preinvestigation values in order to make a tender. However, there are

0043-1648/$ - see front matter & 2013 Elsevier B.V. All rights reserved.
10.1016/j.wear.2013.06.022

Please cite this article as: P.D. Jakobsen, J. Lohne, Challenges of methods and approaches for estimating soil abrasivity in soft ground
TBM tunnelling, Wear (2013), 10.1016/j.wear.2013.06.022i

P.D. Jakobsen, J. Lohne / Wear ()

currently numerous disputes between contractors and owners/

consultants when it comes to TBM performance and tool life. The
level of conicts and their economic consequence highlight, in
effect, the need for robust methodological approaches for predictability of TBM tool life in soil excavation.
A further complicating factor for predicting tool life is that in
soft ground tunnelling the TBMs utilizes various soil conditioners.
The objects of soil conditioners are several; to make the excavated
soil more transportable and impermeable, to reduce the torque
requirement of the TBM cutterhead, to avoid nes to clog at the
cutterhead and cutter tools and to reduce the abrasive wear of the
soil. The most common groups of soil conditioners are polymer
enriched foams or anti-clay-agents and bentonite (a natural clay
mineral). These additives are added in front of the cutterhead,
which make it an important part of the steelsoil tribological
system in tunnel excavation [6] (Fig. 2).
In soft ground tunnelling, the tools and cutterhead interventions are quite often subjected to hyperbaric working environments, meaning that the actual intervention and tool replacement
is a diver operation. Hyperbaric support is needed to stabilize and
avoid a collapse of the tunnel face. By assuming an 40 m overburden with hydrostatic water pressure, the working environment
for divers involve mechanical work in 4 bars pressure in a mix of

soil, additives and water. Such operations are risky with respect to
both health and safety for workers. Equally they are more costly
and time consuming to conduct compared with the same intervention in a self-stable tunnel without the need of hyperbaric
stabilization of the tunnel face. In hard rock TBM tunnelling there
is generally a higher consumption of tools than for soil tunnelling.
However, the consequences of worn out tools in soil tunnelling are
higher, due to the complicating hyperbaric operation that is
needed.

2. Research methods and problem formulation

The complex behaviour of soft ground and soil material (ranging from plastic clay to soft but brittle sandstones) conrms
ndings and general tribological experiences [7]. The variety of
soil material imposes a demand for numerous empirical relations
between real tribological events, and simplied laboratory measurements in order to establish parameters of predictability.
A limitation prevailing for all tests mentioned in this paper is
the relative limited amount of soil material tested. Excavating a
5000 m long tunnel involves moving of 15,000 to 850,000 m3 of
soil material depending on the tunnel diameter. In addition,
inuence of geotechnical parameters that are difcult to recreate
in a laboratory environment should be taken into account as they
inuence the soil mass behaviour and strength.
This paper results from 3 year research project on TBM wear
prediction. Extensive empirical testing is being carried out in order
to provide the data that permit predictable real life wear assessment [811,1,5,1216].
This research paper aims at examining the performance of four
of the most common methods for measuring soil abrasivity.
Further, comments on the various methods pros and cons and
applicability from a practical point of view are presented. More in
detail the paper examines rstly:

 To what extent do the existing laboratory methods and

approaches answer the challenges involved in real-life TBM
tunnelling?
In order to answer this, we assess the:
 Pros and cons of 4 state-of-the-art methods and approaches.
Methodological restraints imposed by the in-vitro testing as
opposed to the real wear phenomena on soil excavating
TBMs. The 4 methods are the LCPC Abrasivemeter, the
NTNU/SINTEF Soil Abrasion Test, mill tests and mineralogical analyses. Test procedures will be presented in
Chapter 3.

Fig. 1. Earth Pressure Balance shield (EPB) TBM with foam nozzles (Courtesy by the
Robbins TBM Company).

Fig. 2. Double disc cutter and scraper tools in contact with a mixed face consisting
of claystone and gypsum. The simplied laboratory approaches presented in this
paper cannot fully recreate the complex tribological system shown in the photo.
Such approaches enable researchers to assess different soil grains abrasivity and
impact on TBM tools separately (Photo by Pl Drevland Jakobsen).

Further, based on the results from this going-through, we wish

to assess to what extent the last years of laboratory-based
research has brought the tunnelling industry closer to a robust
wear estimate for soil tunnelling with TBMs. The question
guiding this analysis is:
To what extent can the limitations identied in existing
methods be remedied?

In order to examine this, we provide an initial presentation and

initial test results of a new method for determining soil and soft
ground abrasivity and soft ground matrix toughness against tunnel
boring in terms of torque.
In general terms, whether simplied index tests such as LCPC
abrasivemeter, SAT and mill tests can provide a good estimation
approach for TBM tool life excavating soil needs to be concluded
when more data and experiences are available.

Please cite this article as: P.D. Jakobsen, J. Lohne, Challenges of methods and approaches for estimating soil abrasivity in soft ground
TBM tunnelling, Wear (2013), 10.1016/j.wear.2013.06.022i

P.D. Jakobsen, J. Lohne / Wear ()

The choice of examined methodologies is done both in order

to answer the ever-present questions of delimitation and
representativeness.
Firstly, including all existing methodologies for in-vitro TBM
testing will surpass the limits of a research paper such as the
present one. Assessing the four methodologies in this paper,
however, serve to illustrate the general challenges to laboratory
test procedures and their relevance to the TBM industry.
Secondly, the NTNU/SINTEF tunnelling research cluster (the
GEMINI-centre)1 has been at the forefront of tunnelling and
underground research since the mid-1970 s, and the methodologies used within this centre are widely recognised in the tunnelling industry. Since the mid-1970s NTNU/SINTEF has tested over
3000 unique rock samples in connection with hard rock TBM
tunnelling performance and tool life estimates [17]. The assessment of existing methods therefore represents an up-to-date
image of present day research capabilities and restraints.

3. State of the art in soil abrasivity testing

for TBM applications
Several in-vitro approaches to the soil abrasivity testing for
TBM applications exist. In this section we present 4 of these
approaches, the LCPC abrasivemeter, the NTNU/SINTEF Soil Abrasion Test, mill tests and a general assessment of soil mineralogy
and grain properties, and discuss briey their main advantages
and disadvantages. On basis of this discussion on the state of the
art within the eld we present a novel approach. This approach
seems to represent one step forward for in-vitro assessment of soil
abrasivity testing allowing for a more precise predictability of real
life wear on TBM tools.
The test results obtained on the LCPC abrasivemeter originates
from other researchers, mainly at the Technical University of
Mnchen. The results obtained on the NTNU/SINTEF Soil Abrasion
Test, the Ball Mill Test and using mineralogical analyses originates from the ongoing research on NTNU and SINTEF.

3.1. The LCPC abrasivemeter

The LCPC abrasivemeter test procedure and apparatus is
described in the French Standard P18-570 [5]. The test is based
on exposing 500 g of 4.06.3 mm fraction of crushed rock or
natural soil for a rotating steel impeller for 5 min. The steel
impeller is 25 mm  50 mm  5 mm.
The rotation speed of the impeller is 4500 rpm and the steel
consists of a relative soft steel alloy with Rockwell B hardness of
6075. For coarse soils the 4.06.3 mm fraction can be sieved out,
while for clay, silt and sand samples the LCPC Abrasivemeter
standard test procedure is not suitable (Fig. 3).
The weight loss of the steel impeller is measured after each
test, and this is the measured abrasivity parameter. The LCPC
abrasivity coefcient (LAC) is calculated as
LAC m0 m=M
where (m0  m) is the weight loss of the steel impeller after one
test and M is the soil or rock materials weight (0.0005 t) [19]. The
soil material's brittleness properties can also be measured by the
LCPC Abrasivemeter by comparing the sieve curves of the initial
1
GEMINI-centre at NTNU/SINTEF is a formalized strategic cooperation agreement between SINTEF and NTNU departments doing parallel activities. For more
information see http://www.sintef.no/home/Building-and-Infrastructure/Infrastruc
ture/Rock-and-Soil-Mechanics/Gemini-Centre-for-Underground-Technology/ and
http://www.drillability.com

Fig. 3. The LCPC test apparatus (after [18]. (1) shows the motor, (2) the rotating
impeller, (3) the jar containing the abrasive, and (4) the funnel tube [18].

4.06.3 mm sample fraction with the particle size distribution

after the test.
The effect of different grain sizes outside the standard 4.0
6.3 mm has been studied by Thuro et al. [20]. The study indicates
higher abrasivity for larger grain sizes. Whether this is a result of
grain's abrasivity properties or caused by the inuence of heavier
grains which has the potential to cause more damage and
degradation on the test steel bit is uncertain. Misra and Finnie
[21] presents several observations on two- and three-body abrasive wear and the size effects of the abrasives. Their conclusion is
that the abrasive wear rates decreases as the size of abrasives
decreases below about 100 mm. Misra and Finnie's [21] general
results correspond with [20] ndings. However, Misra and Finnie's
results are obtained on 20250 mm abrasives sizes, while Thuro's
trials are done on more natural composed soils ranging from
60 mm to 60,000 mm. In order to validate the degradation mechanism on the LCPC test, SEM analyses of worn out LCPC test pieces
should be carried out.
Khler et al. [22] studied the relationship between the LCPC
abrasivity coefcient (LAC) and equal quartz content (EQC). The
correlation proves very poor with no clear trend in the current
data set consisting of 22 samples from the recently completed
Inntal railroad tunnel project in Austria. Further, a relation
between the Cerchar Abrasivity Index (CAI), which is a commonly
used scratch test for measuring hard rock abrasivity, and LCPC is
shown based on ndings by Thuro et al. [20]. The CAI testing
principle is based on a steel pin which is scratching a rough or cut
rock surface over a 10 mm distance under 70 N load. The Cerchar
Abrasivity Index is calculated as the measured diameter of the
resulting wear at on the pin [18], see Fig. 4.
The use of the correlation between CAI and LCPC values did not
provide a good tool life estimate (based on back calculations of the
real tool life in the Inntal project [22]). Some of the reasons found
for this are:

Please cite this article as: P.D. Jakobsen, J. Lohne, Challenges of methods and approaches for estimating soil abrasivity in soft ground
TBM tunnelling, Wear (2013), 10.1016/j.wear.2013.06.022i

P.D. Jakobsen, J. Lohne / Wear ()

 Soil abrasivity measured in lab does not always reect the soil



mass abrasivity.
Inuence of TBM operational parameters is not taken into
account.
Inuence of face support is not taken into account (Fig. 5).

The LCPC abrasivemeter can be used to nd gravel (4.06.3 mm)

soil grain's ability to cause steel wear under high rotation speed. The
applicability of the LCPC abrasivemeter is restricted to run tests
under the 4.06.3 mm grain size, meaning that clay, silt and sand
particles have to be removed from the soil sample.
3.2. The NTNU/SINTEF Soil Abrasion Test
The NTNU/SINTEF Soil Abrasion Test is a development of the
existing test procedure for hard rock Abrasion Value Cutter Steel test
(AVS). Compared with the hard rock abrasivity test AVS, some details
have been changed with respect to sample preparation and the steel
test piece [11]. The relatively small changes have been carried out in
order to achieve possibilities of comparing hard rock abrasivity
measurements (test database contains around 2000 unique test

Fig. 4. Principal set up of the Cerchar Abrasivity Index (CAI) apparatus (after West
[23]). (1) load, (2) pin guidance, (3) steel pin, (4) rock sample, (5) vice sled, and (6)
hand crank [18].

values from hard rock samples) with soil abrasivity measurements

(SAT database contains around 250 unique test values).
In order to reduce or avoid changes of the original grain shape
and size, soil samples need gently drying in an ventilated oven at
30 1C for 23 days. After the drying process, the following
techniques are used in order to disintegrate and separate the
particles for the abrasion powder:
(1) Disintegration by using a soft hammer.
(2) Sieving with steel balls.
(3) Initial disintegration in a jaw crushed if the samples contain
hard lumps of cohesive material after the initial drying.
Crushing of intact grains should be avoided.
The SAT value is calculated as the mean value of the measured
weight loss in mg. In order to be accepted as a test results, weight
losses of 24 parallel tests should not deviate by more than 5 mg
(Fig. 6).
The maximum contact pressure between soil samples and the
steel test piece ranges from 200 to 370 MPa, depending on the
elastic properties of the soil material. The steel test pieces are
machined from a TBM cutter disc ring, and it is a heat treatable,
low alloy steel containing nickel, chromium and molybdenum. The
hardness of the steel piece ranges from 5456 HRC. The length of
one SAT test is 20 m.
The relation between SAT values and the AVS classication
system for hard rock abrasivity is shown in Fig. 7. The hard rock
samples are generally found to be more abrasive than the soil
samples. There is no clear explanation of this, but one indication
can be that soil originates from degraded rock material. Softer
rock, (which generally is less abrasive) will be transformed into
soil before harder rocks. Another indication can be that in the
hard rock abrasivity database, TBM projects with abrasivity problems are overrepresented in the data set.
The SAT value has good correlation coefcient with the soil
samples' quartz content and Vickers Hardness Number Rock
(VHNR). Relations between SAT values and general tribological
parameters such as, grain mineralogy and grain size can be found
in Chapter 3.4.
In order to evaluate the SAT procedure's applicability, some
indications of SAT values relation to recorded tool life have been
prepared (Fig. 8). However, the current data is based on few cutter
changes, meaning that introduction of more data will inuence
both the strength of the correlation as well as the regression
equation. The tool life has been calculated as instantaneous tool
consumption according to [24]. Instantaneous tool consumption
along is calculated between respective tool changes along a tunnel,
causing different tool life for the different tunnel sections.
The SAT test can be used to nd clay, silt and sand abrasivity
properties. For grain sizes above 4 mm the test procedure cannot
be used without crushing the coarse grains prior to testing. The
SAT also shows a relation between the measured abrasivity value
and tool life based on empirical wear data from the eld and lab,
but the current relation does not contain any inuence of in-situ
soil density and use of soil conditioning additives.
3.3. Mill tests

Fig. 5. LAC plotted against the content of abrasive (crystalline) components and CAI
values in gravel samples [20].

There are several mill tests available for determining soil and
rocks resistance against crushing and abrasivity properties. At NTNU/
SINTEF, the Ball Mill Test has been used in order to determine the
inuence of soil conditioning additives inuence on the abrasivity
properties of crushed rock and natural soil samples. The test
procedure is easy and straight forward: A 1500 g sample consisting
of grains less than 16 mm is exposed to 20 steel bits (consisting of
ordinary construction steel) for 5400 revolutions which are equal to

Please cite this article as: P.D. Jakobsen, J. Lohne, Challenges of methods and approaches for estimating soil abrasivity in soft ground
TBM tunnelling, Wear (2013), 10.1016/j.wear.2013.06.022i

P.D. Jakobsen, J. Lohne / Wear ()

Fig. 6. Schematic overview of SAT and AVS (for measuring hard rock abrasivity)
test procedure [11]. The test apparatus consists of a rotating steel disc which is fed
by disintegrated soil powder (04 mm), which passes underneath a steel bit
originating from a TBM disc cutter.

60 min test duration. The rotation speed of the drum is 0.97 m/s, and
the steel bits are circular with 16 mm diameter and length. The
weight loss of the steel bits is measured after the testing. The
measured weight loss is the abrasivity value detected in the Ball
Mill Test (Fig. 9).
The inuence of foam enriched soil conditioner clearly reduces
the weight loss of the steel samples, shown in Fig. 11 [25]. Test
results also indicates that the abrasivity of most geo materials
(rock and soil) increases to a certain level of moisture content as
shown in Fig. 10.
For the clay sample in Fig. 10 no weight loss was recorded
under the present testing conditions. The test results in mill tests
are highly inuenced by which grain size that is included in the
test. The procedure can provide good indications of various soil
conditioning foams and moisture contents inuence on abrasive
wear for specic samples with a dened grain size distribution.
The contact forces between the steel and soil particles are also
relatively low (by gravity and tumbling of the drum). The low
contact force is not in accordance with reality where TBMs have a
relative high thrust and torque ripping and scraping the soil
during excavation.
3.4. Abrasive wear estimate based on sample mineralogy

Fig. 7. Cumulative distributions of measured soil abrasion values and AVS classication. In the current data sets, sand and silt is found to have higher abrasivity
than e.g. clay. Friction soils refers to sandy and gravely soils, while cohesion soil
refers to silty and clayey soils.

In rock engineering, the use of Vickers Hardness Number (VHN)

is commonly used. The VHN is used for measuring for the
abrasiveness of each mineral in a rock or soil. By combining the
percentage of each mineral and its hardness it is possible to
calculate a Vickers Hardness Number Rock/Soil [26].
For determining the amount of different minerals in a rock or
soil sample the most common measuring techniques are X-ray
diffraction (XRD), differential thermal analyse (DTA) and thin
section visual analyse. The XRD is a semi-quantitative approach
to nd minerals and mineral quantities in rock and soil samples.
DTA is a quantitative approach to measure the amount of quartz in
a rock or soil sample.
The correlation between SAT values and the quartz content
proves good, see Fig. 12. The correlation between SAT values and
the total content of mineral and their respective hardness, represented by the Vickers Hardness Number Rock (VHNR) proves less
good, see Fig. 13. This nding is unexpected for the authors, as the
abrasivity properties are expected to be highly inuenced by the
total mineralogical content of soil samples. The most likely explanation is that the collection of VHNR values originates from relative
few projects compared the collection of quartz content values.
Worldwide, there is several laboratories and institutions that offer
mineralogical content analyses. This is an advantage as project owners
and contractors does not need to send soil samples to laboratories
offering special tests and services. On the other hand, estimation of
mineralogical content with XRD and thin section analyses, are semiquantitative approaches, which acknowledge the subjectivity of the
person that is analysing a sample.

4. Suggestion for a new test method taking into acount

in-situ properties of soil

Fig. 8. Indications of SAT value's relation to recorded tool life (scrapers and
rippers) for TBMs o 5 m diameter and TBMs 45 m diameter. From hard rock TBM
tunnelling it is known the large diameter TBMs have a relatively higher amount of
excavation tools in the outer part of the cutterhead, would have a lower tool life
than a small diameter TBM. The reason for this is that tools in the cutterhead
periphery have a relative longer travel length due to higher speed.

The last few years more advanced test approaches has been
introduced in order to provide self-explanatory test result on soil
abrasivity. The intention of such test is to rebuild an in-situ like
soil, in order to perform a direct test on a similar material as the
TBM will encounter. The tests allows the abrasivity measurement
to be conducted on in-situ like soil meaning that water content,
soil compaction and density, use of soil conditioner and inuence
of pressure can be modied and adjusted to simulate the real life
conditions in front of a TBM cutterhead [28,29,16].

Please cite this article as: P.D. Jakobsen, J. Lohne, Challenges of methods and approaches for estimating soil abrasivity in soft ground
TBM tunnelling, Wear (2013), 10.1016/j.wear.2013.06.022i

P.D. Jakobsen, J. Lohne / Wear ()

Fig. 12. Correlation between SAT value and content of quartz. N 62. The data
shows a clear inuence on SAT values by the quartz percentage in a soil [27].

Fig. 9. Ball Mill Test apparatus can be used to assess the reduction of abrasivity by
introducing soil conditioning additives. The test apparatus consists of a rotating
drum lled with soil (016 mm) and 20 steel bits. Water and/or soil conditioning
additives can be applied.

Fig. 13. Correlation between SAT value and Vickers Hardness Number Rock
(VHNR) N 30. The data shows the inuence on SAT values by the mineralogical
content in a soil [27].

Fig. 10. Inuence of water content on steel wear in the Ball Mill Test for 4 different
soil and crushed rock samples [25]. The Figure shows a clear tendency of increased
steel wear by increasing water content up to a certain level.

At NTNU/SINTEF, in close cooperation with BASF Construction

Chemicals, there is an on-going work in developing such an
apparatus. Some initial test results are available. The results clearly
demonstrate the inuence of moisture and various types of foam
on tool life. More tests are being conducted to study the inuence
of soil compaction, soil moisture content, various foam types, as
well as different soil conditioning scenarios. The test apparatus does
also monitor the required torque for drilling with specied thrust
and rotation speed, which also will be a valuable parameter in
estimating tool life and performance. Fig. 14 (left) shows the
principal set-up of the test with the range of operational parameters. Testing of 4 different soil samples have been carried out,
and based on this the following test parameters will be used:
40 mm/min xed penetration, 50 rpm xed rotation speed, varying
torque, varying thrust. The selection of parameters is done in order
to measure torque and thrust requirements for various soils, which
can be an important soil parameter for TBM tunnelling (Figs. 1517).
In sum, based on the results obtained from the laboratory
experimentations, the suggested Soft Ground Abrasion Test
appears a major step forward for laboratory assessment of reallife tunnelling conditions.

5. Discussion
Fig. 11. Reduction of abrasivity due to introduction of soil conditioning foam [25].
The Tonstad clay sample did not show cause any steel weight loss.

The tests LCPC abrasivemeter, the Soil Abrasion Test (SAT) and
the mill tests are all based on a contact between loose soil missing

Please cite this article as: P.D. Jakobsen, J. Lohne, Challenges of methods and approaches for estimating soil abrasivity in soft ground
TBM tunnelling, Wear (2013), 10.1016/j.wear.2013.06.022i

P.D. Jakobsen, J. Lohne / Wear ()

Fig. 14. Schematic setup of the test apparatus (left) and picture of drilling tool and drill rod (right). The test chamber is lled with soil material (010 mm) which is
compacted to the desired density. The drilling tool has nozzles for using continuous foam injection during testing.

Fig. 15. SEM (Scanning Electronic Microscope) image showing a combination of abrasive wear and corrosion (left) on the suggested new tests steel tool after exposure to a
soil sample with 8 wt% water, compared to abrasive wear on the HSAT steel tool after exposure to a dry soil sample (right) (Photos Christian Kreyberg Grdal) [30].

Fig. 16. Measured relation between soil compaction grade (density), abrasivity
(weight loss) and average torque on one soil sample.

several of its in-situ conditions (e.g. water content, density/

compaction, original grain size distribution) and a steel sample.
The contact pressure between the soil and the steel sample is also
varying as different soil grains have different hardness and
mechanical properties.
The simplied tests, represented by the SAT test, indicate
promising correlations between measured abrasivity and TBM tool
life based on data collection from 9 projects. However, the use of
the exponential relations between SAT values and recorded tool
life for tool life prediction should be used with caution [14]. The
SAT values from a specic project in Germany corresponded

Fig. 17. Abrasivity (weight loss on the steel tool) for different moisture contents on
one soil sample [16]. Soil conditioning foam clearly reduces the wear and structures
the soil at higher moisture contents.

quite well with the trends that can be observed in Fig. 8, but for a
similar project (small tunnel diameter, abrasive soil conditions and
the same contractor and equipment) it did not correspond well.
The main reason for this observed inconsistency can be explained
by that the SAT value is a measurement on a cohesionless
powder. This effect is expected to be prevailing for the other tests
and approaches. The LCPC test, the Ball Mill test and the use of
VHNR does not take in to account the mechanical properties (e.g.
cohesion, strength, plasticity) of the soil.
On the other hand, the empirical relations are still quite
promising. This means that further research should be done to

Please cite this article as: P.D. Jakobsen, J. Lohne, Challenges of methods and approaches for estimating soil abrasivity in soft ground
TBM tunnelling, Wear (2013), 10.1016/j.wear.2013.06.022i

P.D. Jakobsen, J. Lohne / Wear ()

include other in-situ parameters into the tool life estimate (e.g. insitu soil density). Due to this, and the relative low amount of data
in the correlation charts tunnel excavations should not be scheduled without any interventions.
The suggested NTNU/SINTEF/BASF Soft Ground Abrasion Test,
makes it possible to evaluate the water content, density/compaction and utilization of soil conditioning additives on steel tool life.
The initial results obtained by the suggested test indicate that
denisity/compaction is highly inuencing the tool wear. It is
expected that such a test would have a better relation to reality,
but this needs to be proven as more data from both the eld and
laboratory is collected.
In order to assure that the simplied tests and the suggested
new test really represents the reality, surface analyses on the steel
samples (test pieces) should be carried out. Further, the worn out
micro structure of the test pieces should be compared with worn
out tools from real TBM tunnelling application.

6. Conclusive remarks
There are several available methods for measuring soil abrasivity. Currently, there are no model recognized by the tunnelling
industry, including contractors, project owners, consultants, TBM
machine manufacturers and academia.
Generally speaking, due to the complex conditions characterising tunnelling conditions, predicting real-life wear on TBM tools
from laboratory experiments is inherently complex. In this paper,
we have examined the degree to which four existing laboratory
approaches succeed in providing predictability.
NTNU, in close connection with SINTEF and BASF Construction
Chemicals are trying to meet this demand from the tunnelling
industry by (1) continuation of the existing abrasion testers used in
connection with soil TBM tunnelling, and (2) continue the development of the suggested Soft Ground Abrasion Tester, which is trying to
meet the demands for the tunnelling industry and criticism of the
existing simplied tests and approaches.
The initial results are promising and the suggested test obtains
abrasivity and torque measurements for a larger span of grain sizes
than the other methods. The method is also sensitive to the water
content of soft ground samples, the compaction/density of the
samples as well as the use of soil conditioning additives.

Aknowlegdements
The author would like to thank Lars Langmaack in BASF Construction Chemicals for funding and providing ideas on the new
hyperbaric soil abrasivity test apparatus, as well as the Norwegian
Tunnelling Society (NFF) for providing funds for laboratory tests and
eld trips in connection with my ongoing Ph.D. study.
References

[2] L., Home, Trends in the us of TBMs worldwide. Presented at NFF TBM seminar
Bergen, 2010.
[3] A., Bruland, Hard Rock Tunnel BoringThe Boring Process, NTNU-Anleggsdrift: Project report 1F-98, 1998.
[4] B., Khaligi, Effective deisgn & development of EPBM's Technical Webinar
presented on tunnellingjournal.com. Available from: http://tunnellingjour
nal.com/webinar/webinar-archives/, 2011.
[5] P.D. Jakobsen, A. Bruland, F. Dahl, Review and assessment of the NTNU/SINTEF
Soil Abrasion Test (SAT(TM)) for determination of abrasiveness of soil and soft
ground., Tunnels and Underground Space Technology, 2013.
[6] L. Langmaack, The truth about soil conditioning: Dos and Dont'ts, World
Tunnelling Congress, Budapest, Hungary, 2009.
[7] H.C. Meng, K.C. Ludema, Wear models and predictive equations: their form
and content, Wear 181183 (2) (1995) 443457.
[8] B. Nilsen, F. Dahl, J. Holzhuser, P. Raleigh, Abrasivity of soils in TBM
tunnelling, Tunnels & Tunnelling International (2006) 3638.
[9] B. Nilsen, F. Dahl, J. Holzhuser, P. Raleigh, Abrasivity testing for rock and soils,
Tunnels & Tunnelling International (2006) 4749.
[10] B. Nilsen, F. Dahl, J. Holzhuser, P. Raleigh, SAT: NTNU's new soil abrasion test,
Tunnels & Tunnelling International (2006) 4345.
[11] B. Nilsen, F. Dahl, P. Raleigh, J. Holzhuser, The new test methodology for
estimating the abrasiveness of soils for TBM tunnelling, in: Proceedings of the
Rapid Excavation and Tunneling Conference (RETC), 2007, pp. 104106.
[12] T.E.E. Becker, P.D. Jakobsen, Overview of Pipe-jacking PerformanceReview of
Tunnel Projects, NO-DIG, Berlin, 2013.
[13] P.D., Jakobsen Prediction of Abrasive Wear in Mechanised Soil Tunnelling,
Ph.D. Project Description, NTNU, 2010.
[14] P.D. Jakobsen, T.E.E. Becker, Tunnelling in abrasive soilsreview of a tunnel
project in Germany, in: Proceedings of the 13th KTA International Symposium
on Mechanized Tunnelling Technology, Seoul, 2012.
[15] P.D. Jakobsen, F. Dahl, L. Langmaack, Reduction of hard rock abrasiveness using
foam agents, Tunnel 8 (2009) 4047.
[16] P.D. Jakobsen, L. Langmaack, F. Dahl, T. Breivik, Prediciting the abrasivity of insitu like soils, Tunnels & Tunnelling International (2012) 4144.
[17] F. Dahl, A. Bruland, P.D. Jakobsen, B. Nilsen, E. Grov, Classications of properties inuencing the drillability of rocks, based on the NTNU/SINTEF test
method, Tunnelling and Underground Space Technology 28 (2012) 150158.
[18] H. Ksling, K. Thuro, Determining rock abrasivity in the laboratory, in:
Proceedings of the European Rock Mechanics Symposium EUROCK 2010,
Lausanne, Switzerland, 2010.
[19] A.F.D. Normalization, Normalisation Francaise P18-579, Granulats: Essai dabrasavite et de broyabilite, 1990.
[20] K. Thuro, J. Singer, H. Ksling, M. Bauer, Determining abrasivity with the LCPC
Test, in: Proceedings of the 1st CanadaU.S. Rock Mechanics Symposium, 2007.
[21] A. Misra, I. Finnie, On the size effect in abrasive and erosive wear, Wear 65 (3)
(1981) 359373.
[22] M. Khler, U. Maidl, L. Martak, Abrasiveness and tool wear in shield tunnelling
in soil, Geomechanik und Tunnelbau 4 (February) (2011) 3653.
[23] West, Rock abrasiveness testing for tunnelling, International Journal of Rock
Mechanics and Mining Sciences Abstracts 26 (1989) 321327.
[24] A. Bruland, Hard Rock Tunnel BoringPerformance and Back-mapping.
NTNU-Anleggsdrift: Project report 1E-98, 1998.
[25] K. Klemetsrud, Soil Abrasion in Shield Tunneling, in Civil and Transport
Engineering, NTNU, Trondheim 85.
[26] A. Bruland, Hard Rock Tunnel BoringDrillability Test Methods, in 13A-98,
The Norwegian University of Science and Technolgoy, Trondheim 25.
[27] P.D. Jakobsen, A. Bruland, F. Dahl, Review and assessment of the NTNU/SINTEF
Soil Abrasion Test (SAT) for determination of abrasiveness of soil and soft
ground, Tunnelling and Underground Space Technology 37 (0) (2013) 107114.
[28] E.A. Gharahbagh, J. Rostami, A.M. Palomino, Tool wear issue in soft ground
tunneling, developing a reliable soil abrasivity Index, Presented at the North
American Tunneling Conference, 2010.
[29] J. Rostami, E.A. Gharahbagh, A.M. Palomino, M. Mosleh, Development of soil
abrasivity testing for soft ground tunneling using shield machines, Tunnelling
and Underground Space Technology 28 (0) (2012) 245256.
[30] C. Grdal, S. Equey, S. Armada, N. Espallargas, Effect of soil and rock
composition on the wear process of cutter tool steel used in tunnel boring
machines, Presented at the NordTrib Conference, Trondheim, 2012.

[1] E., Grv, F. Dahl, P.D. Jakobsen, Norsk kompetanse p geologiske materialers
slitasjeegenskaper. Byggeindustrien, 2008. Available from: http://www.sintef.
no/upload/Byggforsk/Berg-%20og%20geoteknikk/Artikler/Geologiske%20mate
rialers%20slitasjeegenskaper.pdf.

Please cite this article as: P.D. Jakobsen, J. Lohne, Challenges of methods and approaches for estimating soil abrasivity in soft ground
TBM tunnelling, Wear (2013), 10.1016/j.wear.2013.06.022i

Paper 7
Overview of Methods and Approaches used at NTNU/SINTEF to
Estimate Soil Abrasivity in TBM Tunnelling
Jakobsen, Pl Drevland
Published in NORDTRIB proceedings Trondheim, 2012

Overview of Methods and Approaches used at NTNU/SINTEF to Estimate Soil

Abrasivity in TBM Tunnelling
1

P.D. Jakobsen1
PhD Candidate:The Norwegian University of Science and Technology, Department of civil and transport engineering,
Hgskoleringen 7A 7491 Trondheim, Norway.

Abstract
The increase of urbanization constantly demands more infrastructure, which often requires to utilize the underground.
There are several methods and approaches to estimate geo-materials abrasion properties in tunnelling with Tunnel
Boring Machines (TBM). The methods are mainly based on using empirical observations and experiences from
completed tunnelling projects correlated with simplified laboratory test results. The simplified laboratory tests means
that the soils in-situ properties such as the cohesion, density, adhesion, water content and original grain size
distribution are lacking or is disturbed. The intention of this research paper is to give a brief overview on the
approaches and methods from a Norwegian perspective, the test procedures and their pros and cons. The up to date
methods will be evaluated with respect to applicability from a practical view and with some general tribological
experiences such as the influence of hard minerals or particles, grain shape and grain size.
Keywords: TBM Tunnelling, abrasivity, soil, rock.
*Corresponding author: Pl Drevland Jakobsen (pal.jakobsen@ntnu.no).

clays, and to make the excavated soil more

transportable through the pressurized excavation
system [3.] The introduction of such additives is
therefore an important part of the steel soil
interaction in soil tunnelling.

1. INTRODUCTION
Full face tunnelling with tunnel boring machines
(TBMs) is a well-known and widely utilized method to
excavate tunnels. Approximately 70 % of all tunnels
excavated by TBMs are removing soil and soft rock
material [1]. According to [2] the total amount of
TBMs working to excavate soil the last 5 years is
approximately 350 units.
In order to excavate the soil and soft rock, a
combination of rotating disc cutters, scraper tools and
ripper tools are installed on the TBMs cutterhead.
Generally the disc cutters are designed to apply a high
load in to rock mass, inducing tensile failures and
chipping from rock, while scraper tools are designed to
scrape coarse soil from the tunnel face and ripper tools
are ripping cohesive soil material (clay and silt). These
tools are exposed to relative high contact forces with
abrasive material causing various degradation
processes.

Figure 1: EPB TBM with foam nozzles (Photo by Robbins

TBM Company)

In soil tunnelling the TBMs utilize various soil

conditioners. Slurry shield TBMs utilizes bentonite,
which has lubricating properties and consists of a fine
clay mineral. Utilizing bentonite is mainly done to
stabilize the tunnel face during excavation. However,
by applying the right amount of bentonite with the
right pressure it can also increase the life of tools on
the TBM.

At the time being there is not recognized standard

method or approach to estimate tool life in soil TBM
tunnelling. However, the tunnelling works are in most
cases based on tender documents with preinvestigations of the soil or rock along the tunnel
drive. Various contractors then have to assess the preinvestigation values in order to make a tender.
However, there are currently numerous disputes
between contractors and owners/consultants when it
comes to TBM performance and tool life.

Earth pressure balance shield TBMs utilizes various

soil conditioning additives to reduce the torque
demand on the TBMs. The addiives can also change
the soil matrix in order to reduce clogging effects of

environments, meaning that the actual intervention and

tool replacement is a diver operation. The reason for
the hyperbaric environment is to stabilize and avoid a
collapse of the tunnel face. By assuming an 40 meter
overburden with hydrostatic water pressure the
working environment for divers involve mechanical
work in 4 bars pressure in a mix of soil, additives and
water. Such operations are both risky with respect to
health and safety for workers, and are more costly and
time consuming to conduct compared with the same
intervention and work in a self-stable tunnel without
the need of hyperbaric stabilization of the face. In hard
rock TBM tunnelling there are generally a higher
consumption of tools than for soil tunnelling.
However, the consequences of worn out tools in soil
tunnelling are higher due to the hyperbaric operation
that is needed.

This research paper aims at presenting some of the

most common methods for measuring soil abrasivity.
Further, some comments on the various methods pros
and cons and applicability from a practical view are
presented.

2. CAUSES AND
WORN
OUT
TUNNELLING

CONSEQUENCES OF
TOOLS
IN
SOIL

The direct consequence of tunnelling in abrasive soil

conditions is an increased demand for replacement of
the excavation tools (disc cutters, scrapers and
rippers). The disc cutters are equipped on TBMs if the
rock mass exceeds a compressive strength of 20 MPa
[4]. However, disc cutters are quite often installed to
cope with boulders (large rock blocks) inside a soft
soil as well.

3. THE LCPC ABRASIVEMETER

The cost of tools are varying depending on size,

manufacturer-customer relations, material quality and
type. In addition to the direct cost for purchasing tools,
the down time of a TBM, unproductive time for
workers and possible delay of the tunnelling project
should be included. In a recent hard rock TBM project
in Hong Kong it was stated that 1 second downtime
has the cost of USD 1, meaning that one working
shifts downtime can cost as much as USD 30 000
excluding the purchasing cost for the tools (further
details on down time cost of TBMs is to be published
in 2012).

The LCPC abrasivemeter test procedure and apparatus

is described in the French Standard P18-570 [5]. The
test is based on exposing a 4.0 6.3 mm fraction of
crushed rock or natural soil for a rotating steel impeller
for 5 minutes. The rotation speed of the impeller is
4500 rpm and the steel consists of a relative soft steel
alloy with Rockwell B hardness of 60 75. For coarse
soils the 4.0 6,3 mm fraction can be sieved out, while
for clay, silt and sand samples the LCPC
Abrasivemeter standard test procedure is not suitable.

Figure 2: Double disc cutter and scraper tools in contact

with a mixed face consisting of claystone and gypsum
(Photo by Pl Drevland Jakobsen)

Figure 3:The LCPC test apparatus (Photo by Filip Dahl).

The weight loss of the steel impeller is measured after

each test, and this is the measured abrasivity
parameter. The LCPC abrasivity coefficient (LAC) is

In soil tunnelling the tools and cutterhead interventions

are quite often subjected to hyperbaric working
2

x Influence of TBM operational parameters is

not taken into account
x Influence of face support is not taken into
account

calculated as LAC = (m0 m) / M where (m0 - m) is

the weight loss of the steel impeller after a test and M
is the soil or rock materials weight (0.0005 tons) [6]
The soil materials brittleness properties can also be
measured by the LCPC Abrasivemeter by comparing
the sieve curves of the initial 4.0 6.3 mm sample
fraction with the particle size distribution after the test.
The effect of different grain sizes outside the standard
4.0 6.3 mm, has extensively been studies in [6]. The
study indicates higher abrasivity for larger grain sizes.
Whether this is a result of grains abrasivity properties
or caused by the influence of heavier grains which has
the potential to cause more damage and degradation on
the test steel bit is uncertain.
In [7] the relationship between the LCPC abrasivity
coefficient (LAC) and equal quartz content (EQC) is
studied. The correlation is very poor with no clear
trend in the current data set consisting of 22 samples
from the recently completed Inntal railroad tunnel
project in Austria. Further a relation between the
Cerchar Abrasivity Index (CAI) and LCPC is shown
based on [6] findings. The CAI value is a scratch test
used to determine disc cutter tool life on hard rock
TBM tunnelling (Fig. 4 and Fig. 5).

Figure 5: LAC plotted against the content of abrasive

(crystalline) components and CAI values in gravel samples.
[6]

4. NTNU/SINTEF SOIL ABRASION TEST

The NTNU/SINTEF Soil Abrasion Test is a
development of the existing test procedure for hard
rock (AVS (Abrasion Value cutter Steel) test).
Compared with the hard rock abrasivity test AVS some
details have been changed with respect to sample
preparation and steel test piece. [8] The relatively
small changes has been done in order to have
possibilities of comparing hard rock abrasivity
measurements (test database contains around 2000
unique test values from hard rock samples) with soil
abrasivity measurements (SAT database contains
around 250 unique test values).
In order to reduce or avoid changes of the original
grain shape and size, soil samples should be dried in an
ventilated oven at 30C for 2-3 days. After the drying
process the following techniques should be used in
order to disintegrate and separate the particles for the
abrasion powder:
1. Disintegration by using a soft hammer
2. Sieving with steel balls
3. Initial disintegration in a jaw crushed if the
samples contain hard lumps of cohesive
material after the initial drying. Crushing of
intact grains should be avoided.

Figure 4: Cerchar Abrasivity Index (CAI) apparatus (Photo

by Filip Dahl).

The use of the correlation between CAI and LCPC

values did not provide a good tool life estimate (based
on back calculations of the real tool life in the Inntal
project) [7]. Some of the reasons for this are according
to [7]:
x Soil abrasivity measured in lab does not
always reflect the soil mass abrasivity

The SAT value is calculated as the mean value of

the measured weight loss in mg. In order to be
accepted as a test results, weight losses of 2 4
parallel tests should not deviate by more than 5 units.
3

Figure 6: Schematic overview of SAT test procedure [8].

Figure 7: Cumulative distribution of measured soil

abrasion values and AVS classification.

The maximum contact pressure between soil samples

and the steel test piece is ranging from 200 to 370
MPa, dependent on the elastic properties of the soil
material. The steel test pieces are machined from a
TBM cutter disc ring, and it is a heat treatable, low
alloy steel containing nickel, chromium and
molybdenum. The hardness of the steel piece ranges
from 54 56 HRC.
The relation between SAT values and the AVS
classification system for hard rock abrasivity is shown
in (Fig. 7) [8]. The abrasivity properties of hard rock
samples are generally found to be more abrasive than
the soil samples. There is no clear explanation of this,
but some indications can be:
1. Soil originates from degraded rock material.
Softer rock, (which generally is less
abrasive) will be transformed into soil before
harder rocks.
2. In the hard rock abrasivity database TBM
projects with abrasivity problems is over
represented in the data set.

Figure 8: Indications of SAT valuesrelation to recorded

tool life (scrapers and rippers) for TBMs < 5 m diameter
and TBMs > 5 m diameter.

5. MILL TESTS
There are several mill tests available for determining
soil and rocks resistance against crushing and
abrasivity properties. At NTNU/SINTEF the Ball Mill
Test has been used in order to determine the use of
anti-abrasive soil conditioning foams influence on geo
materials abrasivity properties. The test procedure is
easy and strait forward: A 1500 g sample consisting of
grains less than 16 mm is exposed to 20 steel bits
(consisting of ordinary construction steel) for 5400
revolutions which are equal to 60 minutes test
duration. The weight loss of the steel bits is measured
after the testing. The measured weight loss is the
abrasivity value detected in the Ball Mill Test.

The SAT value has good correlation coefficient with

the soil samples quartz content and Vickers Hardness
Number Rock [9] (VHNR). Relations between SAT
values and general tribological parameters such as
grain shape, grain mineralogy and grain size is to be
published in 2012.
As more soil abrasivity data is gathered the
distributions might converge into each other.
In order to evaluate the SAT procedures
applicability to the reality some indications of SAT
values relation to recorded tool life has been prepared
(Fig. 8). However, the current data s is based on few
cutter changes, meaning that introduction of more data
will influence both the strength of the correlation as
well as the regressions equation. The tool life has
been calculated according to [10].

The influence of foam enriched soil conditioner clearly

reduces the weight loss of the steel samples [11]. The
test also indicates that the abrasivity of most geo
material increases to a certain level of moisture
content.

content, soil compaction and density, use of soil

conditioner and influence of pressure can be modified
and adjusted to simulate the real conditions in front of
a TBM cutterhead [13] and [14].

Figure 9: Ball Mill Test apparatus can be used to show the

reduction of abrasivity by introducing foams.

Figure 10: Reduction of abrasion due to introduction of soil

conditioning foam. [11] and [12].

Figure 11: Hyperbaric test chamber for abrasivity testing

on in-situ like soil(up) and drawing of drilling tool (down).

For the clay sample in Fig 10 no weight loss was

recorded under the present testing conditions. The test
results in mill tests are highly influenced by which
grain size that is included in the test. The procedure
can provide good indications of various soil
conditioning foams and moisture contents influence on
abrasive wear for specific samples with a defined grain
size distribution. The contact forces between the steel
and soil particles are also relatively low (by gravity
and tumbling of the drum). The low contact force is
not in accordance with reality where TBMs have a
relative high thrust and torque ripping and scraping the
soil during excavation.

At NTNU/SINTEF there is an ongoing work in

developing such an apparatus. Some initial test results
are available (Fig. 12) that clearly is showing the
influence of moisture and soil conditioning foam on
tool life. More tests are being conducted to study the
influence of soil compaction, soil moisture content and
other foam types. The test apparatus also monitor the
required torque for drilling with specified thrust and
rotation speed, which also can be a valuable parameter
in estimating tool life and performance.

6. TEST PROCEDURES ON IN-SITU LIKE

SOIL AND SOFT GROUND
The last few years more advanced test approaches has
been introduced in order to provide self -explanatory
test result on soil abrasivity. The intention of such test
is to rebuild an in-situ like soil, in order to do a direct
test on a similar material as the TBM has to encounter.
The tests allows the abrasivity measurement to be
conducted on in-situ like soil meaning that water

Figure 12: Initial abrasivity test result on one in-situ like

soil sample.

6. K. Thuro, J. Singer, H. Ksling, M. Bauer.

Determining abrasivity with the LCPC test. 1st
Canada-US Rock Mechanics Symposium. 2007
7. M. Khler, U. Maidl, L. Martak. Abrasiveness and
tool wear in shield tunnelling in soil.
Geomechanics and Tunnelling. Volume 4. 2011.
8. B. Nilsen, F. Dahl, J. Holzhuser, P. Raleigh. The
New Test Methodology for Estimating the
Abrasiveness of Soils for TBM Tunnelling. Rapid
Excavation and Tunneling Conference (RETC).
Toronto 2007.
9. F. Dahl, A. Bruland, P. D. Jakobsen, B. Nilsen, E.
Grv. Classifications of properties influencing the
drillability of rocks, based on the NTNU/SINTEF
test method. Tunnelling and Underground Space
Technology 28. 2012.
10. A. Bruland. Hard Rock Tunnel Boring
Performance Data and Back-mapping. NTNUAnleggsdrift Project Report 1E-98. 1998.
11. P. D. Jakobsen, F. Dahl, L. Langmaack.
Reduction of hard abrasivness using foam agents.
Tunnel 8. 2009.
12. K. Klemetsrud. Soil Abrasion in Shield
Tunneling. Masters thesis, NTNU 2008.
(unpublished)
13. E. A. Gharahbagh, J. Rostami. Tackling a new soil
abrasivity index. North American Tunneling
Conference 2010.
14. J. Rostami, E. A. Gharahbagh, A. M. Palomino,
M. Mosleh. Development of soil abrasivity testing
for soft ground tunneling using shield machines.
Tunnels and Underground Space Technology 28,
2012.
15. H.C. Meng, K. C. Ludema. Wear models and
predictive equations: their form and content. Wear
181 183, 1995.

7. CONCLUSIVE REMARKS
The complex behavior of soil material (ranging from
plastic clay to soft but brittle sandstones) confirms
findings and general tribological experiences in [15].
There is a demand for numerous empirical relations
between real tribological events, and simplified
laboratory measurements.
A limitation prevailing for all tests mentioned in this
paper is the relative limited amount of soil material
tested. Excavating a 5 000 m long tunnel involves
moving of 15 000 to 850 000 m3 of soil material
dependent of the tunnel diameter. In order to estimate
the tool life of a TBM excavating such amount of soil
demands several simplified laboratory tests, as well as
adjusting the soil abrasivity to geotechnical
parameters that may influence tool life and how the
TBM is operated with respect to soil conditioning,
thrust rpm and torque ratio, scheme for cutterhead
interventions.
Whether simplified index tests such as LCPC
abrasivemeter, SAT and mill tests can provide a
good estimation approach for TBM tool life excavating
soil needs to be concluded when more data and
experiences are available. However, the findings in
(Fig. 8) is promising, meaning that simplified index
tests should not be neglected at the moment.

8. REFERENCES
1- E. Grv, F. Dahl, P. D. Jakobsen. Norsk
kompetanse
p
geologiske
materialers
slitasjeegenskaper. Byggeindustrien 2008.
http://www.sintef.no/upload/Byggforsk/Berg
%20og%20geoteknikk/Artikler/Geologiske%
20materialers%20slitasjeegenskaper.pdf (last
cited 28.03.2012)
2. L. Home. Trends in the use of TBMs worldwiede.
Presentation NFF TBM Seminar Bergen 2010.
3. L. Langmaack. The truth about soil conditioning:
Dos and Donts. World Tunnelling Congress
Budapest, Hungary, 2009.
4. B. Khaligi. Effective Design & Development of
EPBMs. Technical webinar presented on
tunnellingjournal.com.
Available
at
http://tunnellingjournal.com/webinar/webinararchives/
5. Normalisation Francaise P18-579. Granulats: Essai
dabrasavite et de broyabilite. Assocation francaise
de normalization. 1990.

AKNOWLEGDEMENTS
The author would like to thank Lars Langmaack in
BASF construction chemicals for funding and
providing ideas on the new hyperbaric soil abrasivity
test apparatus, as well as the Norwegian Tunnelling
Society (NFF) for providing funds for laboratory tests
and field trips in connection with my ongoing PhD
study.

Paper 8
Predicting the abrasivity of in-situ like soils
Jakobsen, Pl Drevland
Langmaack, Lars
Dahl, Filip
Breivik, Torkjell
Published in Tunnels and Tunnelling International, 2012

EPB TBM ABRASION | TECHNICAL

Predicting the
abrasivity of insitu like soils
Jakobsen, Langmaack, Dahl and Breivik report on further work undertaken at the Norwegian University
of Science & Technology (NTNU Trondheim) and SINTEF on better ways to study tool-soil interaction
on abrasivity, particularly relevant to soft ground abrasion

etermining the abrasiveness of

soil and predicting TBM tool life
in shield tunnelling has become
a popular research subject in
the last few years. Nilsen et al (2006a,
2006b, 2006c and 2007) and Jakobsen and
Dahl (2010) converted NTNU/SINTEFs
existing abrasivity measurement technique
for hard rock and made it applicable for
measuring soil abrasivity. Thuro et al (2007)

converted the LCPC abrasivemeter, also

intended for hard rock abrasivity testing, to
measure soil abrasivity. Gharahbagh and
Rostami (2010) and Rostami et al (2012)
developed a new soil abrasivity measuring
technique that takes into account in situ
soil properties. Jakobsen et al (2009) and
Langmaack et al (2010) presented results
on the influence of anti-abrasion foams on
wear by the use of the Ball Mill Test and

Below, left: Figure 1, steel-soil interaction with the current Soil Abrasion Tester;
Below, right: Figure 2, drilling tool for new proposed test

www.tunnelsandtunnelling.com

NTNU/SINTEF abrasivity test. The main

conclusions from this work were that the
results showed a clear reduction of rock
and soil abrasiveness by utilising polymer
enriched foams. However, further
evaluations are needed due to difficulties in
controlling the amount of foam as well as
difficulties in comparing abrasiveness
results from samples with different grain
sizes in the Ball Mill Test. NTNU together
with SINTEF and BASF Construction
Chemicals are, as a continuation of this
initial work, developing a new measuring

JUNE 2012 Tunnels & Tunneling International

41

TECHNICAL | EPB TBM ABRASION

Table 1: Summary of soil properties of the tested soils

Soil 1
(from the Trondheim area)

Soil 2
(from a tunnelling
project in Europe)

44 per cent

34 per cent

4.66

15.5

d50

4.8 mm

4.9 mm

d25

1.5 mm

2.5 mm

26

23.5

Quartz content
measured by DTA
Cu = d60 /d10

Soil Abrasion Test

(SAT) value

technique which takes several different soil

properties in to account.
The reason for developing a new
measuring technique is to try to study tool
soil interaction in a better way, as well as
research on how various geotechnical
parameters influence the steel wear due to
soil abrasivity. The technique that is
currently used at NTNU/SINTEF is based
on testing disturbed soil material: meaning
dry powder, lack of cohesion, limited grain
size distribution (< 4 mm) and no influence
of soil compaction. Additionally, the current
technique does not allow for the testing of
the possible influence of chemical
additives, which are commonly used in
Earth Pressure Balance (EPB) machine.
The new apparatus will be able to handle
larger grain sizes (up to 12mm), different
soil moisture contents, soil compaction or
reproduced in-situ soil density and also the
use of chemical or natural additives such as
foams and bentonites.
42

Tunnels & Tunneling International JUNE 2012

The test is designed to take into account

hyperbaric conditions, as the test chamber
can be pressurised up to six bar.
Additionally the required torque for drilling
in the soil sample at variable rotation and
thrust will be monitored, thus indicating
different torque requirements in various
compacted soil-additive-water matrices.
The interaction between soil and steel
will be based on both rotation and
penetration of the drilling tool. The drilling
tool consists of four steel spokes, which
drill into the soil to a depth of 200mm in
each test. The steel used in the test piece
has a Vickers hardness of 227.
Preliminary atmospheric tests have been
performed in order to evaluate the test
procedure. The tests were conducted on
two natural soils. One (Soil 1) is from a local
source in the Trondheim area and the other
(Soil 2) from site investigation of a planned
European EPB TBM tunnelling project. The
grain size distribution of Soil 1 ranges from

Left, top: Figure 3, hyperbaric test chamber

150 x 300 mm; Above: Figure 4, boring in to
a compacted soil sample with soil
conditioning foam under atmospheric test
conditions
0 to 6.5mm and for Soil 2 from 0 to 30mm.
Particles above 12mm were removed from
the Soil 2 prior to testing.

Test procedure
The soil specimens to be tested were
assembled in four layers, each compacted
by ten strokes of a Proctor Hammer of
50mm diameter and 2.5kg load. The
rotation speed of the drilling tool was set at
90 rev/min, and a constant penetration rate
of 0.63mm/rev or 56mm/min was applied.
The test was conducted by drilling stepwise
to depths of 50, 100, 150 and 200mm. The
reason for the stepwise drilling method was
to achieve better mixing between the foam
and soil.
www.tunnelsandtunnelling.com

EPB TBM ABRASION | TECHNICAL

0.100
0.090

Without foam

0.080

With foam

Weifgt loss (g)

0.070
0.060
0.050
0.040
0.030
0.020
0.010
0.000

10
15
Moisture content (%)

20

25

0.100
0.090

Without foam

0.080

With foam

Weight loss (g)

0.070
0.060
0.050
0.040
0.030
0.020
0.010
0.000

10

15
Moisture content (%)

20

25

0.120
Without foam
0.100

With foam

0.080
Weight loss (g)

Dry soil application: In order to obtain a

dry soil, the present soil samples were
gently dried in a ventilated oven at 30C
for 48 hours.
Humid soil application: Various amounts
of water were added and mixed with the
specimens to be tested in order to
achieve homogeneous moist soil
samples.
Conditioned soil application: Foam with
an expansion ratio FER = 10 (air/liquid
ratio = 10:1) was applied on top of the
compacted and humidified soil sample
for the tests performed with additives.
The foam injection ratio FIR = 30 per

0.060

0.040

Right, top: Figure 5, soil abrasivity

development for conditioned Soil 1;
Right, centre: Figure 6, Adjusted
moisture content and soil abrasiveness
for Soil 1; Right, bottom: Figure 7:
Weight loss for varying water contents
with and without foam for Soil 2
www.tunnelsandtunnelling.com

0.020

0.000

8
10
Moisture content (%)

12

14

16

18

JUNE 2012 Tunnels & Tunneling International

43

TECHNICAL | EPB TBM ABRASION

cent (foam:soil ratio = 300 litre per cubic

metre) was used meaning that one litre
of foam was utilised for excavating
approximately 3.3 litres of soil. The foam
used was BASF Meyco SLF41, which is
a polymer enriched soil-conditioning
additive. Generally the concept of using
soil-conditioning foams is that their high
surface area will confine fines in the soil
matrix and subsequently reduce the
friction between both soil grains and the
interaction between the soil and steel
tool.
As shown by Jakobsen et al (2009), and
Rostami et al (2012), soil and rock
abrasiveness increases up to a certain level
generally expected with increasing moisture
content. Subsequent to the abrasiveness
peak level a general decrease of soil and
rock abrasiveness occurs. A similar
phenomenon is reproduced with the test
apparatus described here. For Soil 1 the
soil abrasiveness peak occurs at
approximately six per cent moisture
content and then reduces as more water is
added. The peak is assumed to be
influenced by the soils grading curve as
well as general abrasion properties such as
grain mineralogy and grain shape (see
figure 5).

The testing performed with addition of

soil conditioning foam leads to a reduction
of the soil abrasivity peak (reduction shown
in red shaded area in figure 5, page 43),
and hence provides a good indication of
the functionality of soil conditioners.
However, at low moisture contents the
introduction of soil conditioners in certain
cases may intend to increase the abrasivity,
here shown as green patterned area in
figure 5. This phenomenon has also been
observed at some EPB drives executed in
dry soil open excavation in Switzerland and
Germany where the introduction of small
amounts of soil conditioners reduced the
TBM cutter tool life.
The increase of soil abrasivity in low
moisture contents is most likely caused by
the introduction of additional moisture by
using soil conditioning agents. Taking the
total moisture into account, the soilconditioning agent reduces the wear,
except for the high moisture content area.
The increased soil cohesion probably leads
to a slower decrease of wear than just
using water. The adjusted picture is shown
in figure 6 (page 43).
Soil 1 and Soil 2 are quite similar with
respect to the measured abrasivity
properties according to the existing Soil
Abrasion Test procedure, and the initial

References
Gharahbagh, EA, Rostami, J & Gilbert, M (2010). Tool wear in soft ground tunnelling - Developing a
reliable soil abrasivity index, Proc North American Tunneling conference 2010, SME UCA, Colorado,
USA
Gharahbagh, EA & Rostami J (2010). Tackling a new soil abrasivity index, North American
Tunneling Journal, Sep 2010, pp1618
Jakobsen PD, Dahl F, & Langmaack L (2009). Reduzierung der Hartgesteins-Abrasion durch Einsatz
von Schaum, Tunnel 8, 2009, pp4047, Bauverlag, Gutersloh, Germany
Jakobsen PD & Dahl F (2010). Soil Abrasion in TBM Tunnelling, Korean Tunneling Association on
Mechanized Tunneling.
Langmaack L, Grothen B & Jakobsen PD (2010). Anti-wear and anti-dust solutions for hard rock
TBMs Proc World Tunnelling Cong, Vancouver, 2010, ITA-AITES & Tunneling Assoc of Canada
Nilsen B, Dahl F, Raleigh P & Holzhauser J (2006a). Abrasivity of soils in TBM tunnelling, Tunnels &
Tunnelling International, Mar 2006, pp3638, Progressive Media Group, London.
Nilsen B, Dahl F, Raleigh P & Holzhauser J (2006b). Abrasivity testing for rock and soils, Tunnels &
Tunnelling International, Apr, 2006, pp4749, Progressive Media Group, London
Nilsen B, Dahl F, Raleigh P & Holzhauser J (2006c). SAT: NTNUs soil abrasion test, Tunnels &
Tunnelling International, May, 2006, pp4345, Progressive Media Group, London.
Nilsen, B, Dahl F, Holzhauser J & Raleigh P (2007). The new test methodology for estimating the
abrasiveness of soils for TBM tunnelling, Proc Rapid Excavation and Tunneling Conference (RETC),
2007, pp104116, SME.
Rostami J, Gharahbagh, EA, Palomino, AM & Mosleh M (2012). Development of soil abrasivity
testing for soft ground tunneling using shield machines, Tunnelling and Underground Space
Technology 28, 2012, pp245256, Elsevier.
Thuro K, Singer J, Kasling H & Bauer M (2007). Determining abrasivity with the LCPC test, Proc 1st
Canada-US Rock Mechanics Symposium, 27, pp827-834.

44

Tunnels & Tunneling International JUNE 2012

The authors
Pal Drevland Jakobsen, the lead author, is
a PhD candidate at the Dept of Civil &
Transport Engineering, Norwegian
University of Science & Technology
(NTNU Trondheim);
Lars Langmaack is technical manager for
TBMs with BASF Construction
Chemicals;
Filip Dahl is laboratory manager at
SINTEF Building & Infrastructure
SINTEF is the largest independent
research organisation in Scandinavia;
Torkjell Breivik is senior engineer at the
Dept of Geology & Mineral Resources
Engineering, NTNU

measurements by the new test. The quartz

content is also in the same range. However
Soil 2 has a more uniformly graded sieve
curve that might cause a higher resistance,
and thus higher steel wear on the
excavating tools. In general a reduction of
abrasive wear in the range of 25-45 per
cent was observed by applying adequate
soil conditioners to the soil samples.

Conclusions
The few tests conducted clearly indicate
that moisture content influences the
abrasive wear in a steel soil interaction,
similar to the Rostami et al (2012) findings.
The introduction of soil conditioners
reduces the potential abrasivity peak of the
tested soil samples. However the current
tests are so far only based on sandy soils,
meaning that further testing on other
materials is needed.
The introduction of an additional soil
abrasion test is, at present, not intended as
a new index test. The idea of introducing
this test procedure is to provide selfexplanatory test results taking into account
several in-situ soil properties.
Based on the promising results
NTNU/SINTEF, together with BASF, intends
to perform extensive testing in order to
show the influence of various soil
conditioners on abrasivity and TBM torque,
as well as the effects of confinement
pressures and soil compaction.

Acknowledgements
The authors would like to thank Ivar Sletta for
conducting the tests, and Reidar Frog at the
Faculty of Natural Sciences and Technologys
workshop, as well as the Norwegian
Tunnelling Society that is funding Pl
Drevland Jakobsens PhD study on
prediction of soil abrasion in TBM tunnelling.

www.tunnelsandtunnelling.com

Paper 9
Development of the Soft Ground Abrasion Tester (SGAT) to
predict TBM tool wear, torque and thrust
Jakobsen, Pl Drevland
Langmaack, Lars
Dahl, Filip
Breivik, Torkjell
Published in Tunnelling and Underground Space Technology,
2013

Tunnelling and Underground Space Technology 38 (2013) 398408

Contents lists available at ScienceDirect

Tunnelling and Underground Space Technology

journal homepage: www.elsevier.com/locate/tust

Development of the Soft Ground Abrasion Tester (SGAT) to predict TBM

tool wear, torque and thrust
Pl Drevland Jakobsen a,, Lars Langmaack b, Filip Dahl c, Torkjell Breivik d
a

Department of Civil and Transport Engineering, NTNU, N-7491 Trondheim, Norway

BASF Construction Chemicals Europe Ltd., Zuerich, Switzerland
SINTEF Building and Infrastructure, Rock Engineering, N-7465 Trondheim, Norway
d
Department of Geology and Mineral Resources Engineering, NTNU, N-7491 Trondheim, Norway
b
c

a r t i c l e

i n f o

Article history:
Received 26 February 2013
Received in revised form 12 July 2013
Accepted 17 July 2013

Keywords:
Soft ground tunneling
Soil conditioning
Wear
Abrasion
EPB

a b s t r a c t
The Norwegian University of Science and Technology (NTNU), SINTEF Rock Engineering and BASF Construction Chemicals have jointly developed a new test device called the Soft Ground Abrasion Tester
(SGAT). The ambition and purpose of the design of the test and the applied test procedure is to replicate
an in situ soil TBM excavation tool contact, in a small and simplied scale. The current development is
attempting to bridge a gap when it comes to estimating soft ground and soil abrasivity, as earlier research
on e.g. the NTNU/SINTEF Soil Abrasion Test (SAT) shows that it does not catch up all driving factors for
soft ground and soil abrasivity directly. The paper summarizes the development of the SGAT apparatus,
and shows its capabilities to evaluate, quantify and compare how the soil mineralogy, water content,
pressure, compaction, and the use of soil conditioning additives inuences the wear rate on the SGAT
excavation tool. During testing the required torque and thrust are monitored and logged, making it possible to measure various soilsoil conditioning matrixes requirement for operational parameters.
2013 Elsevier Ltd. All rights reserved.

1. Introduction

1.2. State of the art on soil abrasion prediction based on hard rock test
methods

1.1. General
Predicting soft ground TBM tool life is a complex matter. In order to study and quantify in situ soft ground abrasivity, The Norwegian University of Science and Technology (NTNU), SINTEF
Rock Engineering and BASF Construction Chemicals have developed a test device called the Soft Ground Abrasion Tester (SGAT).
The intention for developing the apparatus is to provide a reliable
test method for determination of in situ like soils abrasivity, as
well as various soils and soil conditioners torque requirement
for soft ground TBM applications. The apparatus has the capability
of evaluating how soil abrasivity is inuenced by water content,
air-pressure, compaction or soil density as well as introduction of
soil conditioning additives. The developing consortium has been
successful and worked in the following manner: NTNU has managed the development based on a BASF design concept. The development has been quality assured by SINTEF. Generally, the SGAT is
an open source development and other suppliers, contractors, clients and TBM manufacturers are invited to run tests on the
apparatus.

Corresponding author. Tel.: +47 92 43 38 33; fax: +47 73 59 70 21.

E-mail address: pal.jakobsen@ntnu.no (P.D. Jakobsen).
0886-7798/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tust.2013.07.021

So far, the research on soil abrasivity and TBM tool life on soft
ground tools at NTNU/SINTEF has been limited to the Soil Abrasion
Test (SAT) (Nilsen et al., 2006c,2007; Jakobsen and Becker, 2012),
and the Ball Mill Test for determining the inuence of soil conditioning additives and presence of water on hard rock and soil abrasivity (Jakobsen et al., 2009; Jakobsen and Lohne, in press). The
initial development of the SAT test procedure results from a request from a contractor, which would like to evidence that a specic soil condition was highly abrasive. All these test procedures
and approaches originate from NTNU/SINTEFs research on hard
rock TBM tunneling performance and tool life estimates, which
have been an ongoing research activity for several decades. In
2011, there has also been initiated research on the effect of tribo-corrosiveness of rock and soil in interaction with steel (Grdal
et al., 2012). The intention of this present work is to achieve a further understanding of the mechanisms which are degenerating
TBM excavation tools.
Similar to the development of the NTNU/SINTEF Soil Abrasion
Test (SAT), the Technical University in Munich introduced the
LCPC abrasivemeter (LCPC, 1990) for determining soil abrasivity
(Thuro et al., 2007). The LCPC approach has some similarities to
the SAT procedure available at NTNU/SINTEF, as both test methods use dried soil samples in limited fractions (LCPC 4.06.3 mm/

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 38 (2013) 398408

SAT < 4.0 mm). The sample used for the LCPC test is however exposed to a steel impeller rotating at 4500 rpm. The high speed rotation of the steel impeller is causing crushing of soil or hard rock
fragments, and this interaction causes wear on the steel.
Gwildis et al. (2010) present tool wear data from the Brightwater conveyance tunnel project, which indicate that cutterhead
energy consumption together with abrasivity descriptors (e.g.
SAT, quartz content or Miller Slurry Test) are the driving factors
for tool wear. The simplied test approaches such as the SAT test
and the LCPC abrasivemeter do not have the ability to directly include the soil materials need for cutterhead energy, as the methods are based on testing the interaction between steel and loose
soil particles.
Khler et al. (2011) present experiences from the tunneling project Lower Inn valley in Austria, and conclude that there are no recognized prediction models for estimating tool wear in shield
tunneling in soil. They also consider the possibility to establish correlations between small-scale laboratory index values and real-life
TBM wear rates to be unlikely, if not impossible.
1.3. New developed soft ground abrasion test methods
The rst approach of developing an apparatus purely intended
for soil and soft ground abrasive wear prediction was performed
and published by Gharahbagh et al. (2010, 2011, 2013) and
Rostami et al. (2012a,b). The Penn state soil abrasion testing system consists of a rotating blade at a xed position which is in contact with a soil sample. The apparatus has the possibilities of
evaluating the inuence of various water contents, rotation speeds,
higher ambient pressures and various excavation tool hardness.
However, the soil sample is not consolidated prior to testing
according to the test suggested by Gharahbagh et al. (2010). The
soil sample density/consolidation is therefore not a controllable
variable. Furthermore, the rotating tool is in a xed position during
testing (not penetrating into fresh soil sample material) and soil
conditioners can only be used as an already preconditioned soil
sample.
A more recent approach is suggested by Barzegari et al. (2013).
The test device consists of rotating steel plates in contact with soil
samples or crushed rock. The soil sample can be tested under pressure, and the test device allows utilization of additives.
Due to the assessment of simplied abrasion measurements
presented by Khler et al. (2011), Gwildis et al. (2010) and Jakobsen and Becker (2012), as well as the lacking possibility to run tests
on a consolidated sample in the Penn State system, a development
of a more advanced prediction method is needed. The development
of the new SGAT is an attempt to develop a laboratory approach
that after further assessment and work, may work as a pre-investigation tool on tool life for soft ground and soil TBM tunneling.
1.4. Research questions
Jakobsen and Becker (2012) and Jakobsen et al. (2013) evaluated the SAT values against observed tool life for some recently
completed tunneling projects with bentonite slurry face support.
In this evaluation, one of the reasons for empirical outliers were
identied as the inuence of the soil grading. Single graded soils
with high SAT values did not cause any reduction in excavation
tool life. This effect is, as stated by Gwildis et al. (2010), explained
by the relative low amount of energy the TBM needs to apply in order to excavate such soils, and thus relatively low contact pressures between the soft ground tunnel face and the TBM
excavation tools.
These previously missing effects of soil and soft ground compaction, together with inuence of soil conditioning additives are
the main reason for developing the apparatus. If the development

399

proves to provide valid and reliable predictions of tool life, a secondary effect of the apparatus can be to obtain laboratory data
about how soil conditioning additives, compaction, water inuence
isolated inuences tool life, and use these experiences on SAT
values. The research questions we intend to answer in this paper
are:
 To what extent does the soft ground and soil compaction inuence the soft ground TBM excavation tool life?
 Is the excavation tool life inuenced by the amount of energy
the TBM utilizes in order to excavate the soil and soft ground?
 To what extent does the water content inuence the soft ground
TBM excavation tool life?
 To what extent can the use of soil conditioning additives
increase the soft ground TBM excavation tool life and inuences
other TBM parameters like torque and thrust?
2. The New Soft Ground Abrasion Tester (SGAT)
The SGAT apparatus consists in the actual status of a drive unit
(rotation and vertical movement), a shaft attached to an exchangeable cutterhead-like tool consisting of two steel bars of Vickers
Hardness 227 equal to 20 HRC, a testing chamber for the soil sample with a lid which is airtight up to 6 bars pressure, and a foam
pump, see Fig. 1. During testing, water, bentonite or soil conditioning additives can be added continuously and directly at the cutterhead-like tool, replicating the real TBM operation. The current steel
type, which the results in this paper comprise, is a carbon steel
with the chemical composition presented in Table 1.
The drilling tool consists of two steel bars attached to a holder.
The tool is designed in order to achieve mixing between the soil
sample and the possible used soil conditioning additives, and to
achieve relatively high contact forces between the lower steel
bar (Fig. 2) and the compacted soil sample during the test. The
use of two separate steel bars to form the drilling tool does also
provide a possibility to distinguish between primary wear, wear
on the lower steel bar, and secondary wear recorded on the upper
steel bar. The length of the steel bars is 13 cm, which allows large
grains (620 mm) to pass between the drilling tool and the periphery of the testing chamber. The inside periphery of the test vessel
consist of steel. For verication issues, some tests have been run
without the lid in order to see whether the soil sample rotates
along with the tool, which has not been the case.
The rotation speed is variable between 0 and 100 rpm. The xed
maximum speed of 100 rpm is chosen in order to avoid erosive
wear, and to reduce the possibility of high impacts between the
steel and soft ground and soil fragments. Running tests on
100 rpm results in a travel speed of approximately 0.7 m/s, which
is in the range of a TBM excavation tool, which typically ranges between 0.1 and 1.5 m/s dependent on the tool position.
Several techniques have been tried in order to apply soil conditioning additives, during the development of the SGAT apparatus.
Fig. 4 shows the three main approaches, (a) applying the soil conditioning additive on top of the compacted soil sample prior to
testing, (b) injecting foam continuously during testing and (c)
pre-mix the soil and soil conditioning additive prior to testing.
The by far closest to reality technique for applying soil conditioning additives is by injecting through the points shown in Fig. 2,
equal to the method shown in Fig. 3b.
2.1. Preliminary test procedure
Generally, all soil samples have been dried for 48 h in a ventilated oven at 30 C prior to testing. After the drying, grains above
10 mm are removed from the sample. The next step is to add water
and properly mix water and soil. Similarly to Rostami et al.

400

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 38 (2013) 398408

Fig. 1. Outline of the new Soft Ground Abrasion Tester (SGAT) (left) and photo of the test rig (right). The height of the test rig is 210 cm and the width is 75 cm, and the test
chamber is 30 cm high and with inner diameter of 15 cm (photo by Simon Alexander Hagen).

Table 1
Chemical composition of the steel type used for the SGAT tool in the initial testing.
C

Si

Mn

Ni

Cr

Mo

0.43
0.45

Max
0.4

0.5
0.8

Max
0.4

Max
0.045

Max
0.045

Max
0.4

Max
0.1

Fig. 2. The SGAT drilling tool. The test pieces have 1  1 cm cross-section, and the
holes on the lower steel bar is the nozzles for soil conditioning additives.

(2012a,b), the mixing of water and soil were done carefully in order to ensure an uniform distribution of water. In order to avoid
crushing of soil grains, thereby introducing more nes in-to the
sample, the mixing were done carefully by hand1. For soil samples
with the desired water content, testing have been conducted on the
original soil sample without drying.

1
For the soil in sample 3, lumps of sedimented clay and silt were mechanically
crushed from gravel and stone size to soil <10 mm.

After nishing the sample preparation, the soil samples were

assembled in four layers with different grade of compaction as
shown in Table 3. Fixed volumes of soil samples have been applied
to the test bucket, causing the sample weight to vary (between
6500 g and 8000 g dependent on the grain density and level of
compaction). An evaluation of various compaction levels along
the sample has not yet executed. The authors expect an increase
of compaction towards the bottom which is backed up by increasing torque and thrust data.
The SGAT test can be run under different operational schemes.
In this paper, the rotation speed and vertical penetration has been
xed, causing the torque and thrust to vary. Oppositely, it would be
possible to run tests under a xed torque with varying vertical
penetration or varying rpm. The tool penetration is about 15 cm
for the results presented in this paper, with a penetration rate of
40 mm/min. The apparatus has the possibility to reduce the penetration rate or even run tests without any penetration, see Table 4.
The edges on the steel bars on the drilling tool are sharp edged,
prior to use. In order to avoid replacement of the steel bars after
one single use, the tools need to be runned-in for 2 h in an abrasive
soil sample prior to the rst test.
As a standard test procedure the penetration speed and rotation speed were xed, while the thrust force and torque varied
dependent on the soil properties and possible use of soil conditioning additives. This approach is carried out in order to compare different soil samples torque requirement, which is
thought to be a good indicator of how easy or hard a soil is to
excavate mechanically, as well as indicating the inuence on
the steel wear rate.

2.2. Data collection and software

The rotation and penetration are driven by two separate servo
motors with a gear ratio. The control of the motors use standard
analog IO (010 V) for position, rotation, penetration speed, thrust
and torque. These data are together with a dedicated signal for
measuring the air pressure inside the SGAT test chamber continuously logged and presented in the control software (Fig. 4). The
software is written in LabVIEW 2012 and utilizes NI CompactDAQ
as interface for the control of the motors.

401

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 38 (2013) 398408

Fig. 3. Overview of possibilities to add soil conditioning additives in the SGAT apparatus. (a) Shows addition of foam on top of the soil sample, (b) shows a continuous
addition of foam through nozzles, and (c) shows a premix of foam and the soil sample).

Fig. 4. Screen view of the SGAT operational and data collection software.

Table 2
Mineralogy of the soil sample obtained by X-ray diffraction (XRD), and measured
abrasivity with the Soil Abrasion Test.

Quartz (%)
Mica (%)
Plagioclase (%)
Chlorite (%)
Kali-feldspar (%)
Amphibolite (%)
Calcite (%)
Albite
SAT value

Soil sample 1

Soil sample 2

Soil sample 3

44
18
15
12
5
3
3
NA
26 (high)

42
<1
36
6
15
NA
NA
NA
23.5 (high)

76
16
NA
NA
NA
NA
7
<1
6.5 (low)

Table 3
Example of inuence of soil compaction and density on wear and torque for Soil
sample 1.
Density (kg/m3)

Compaction proc.

Wear (mg)

Avg. torque (Nm)

1544
1886
1958
2058
2109
2228

No compaction
5 Blows/4 layers
10 Blows/4 layers
15 Blows/4 layers
20 Blows/4 layers
30 Blows/4 layers

52
82
75
92
92
115

8.7
10.7
11.2
13.2
11.4
17.0

402

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 38 (2013) 398408

Table 4
Comparison of the new SGAT test procedure and the Penn state soil abrasion testing system.

Tool design
Tool steel
Rpm
Length of penetration
through the soil sample
Penetration rate
Thrust force
Torque variation
Ambient pressure
Maximum grain size
Soil compaction
Addition of soil conditioners

Soft Ground Abrasion Tester (SGAT)

Penn state soil abrasion (SAI) testing system (personal

communication Jamal Rostami September 2012)

4 Steel spokes
Standard construction steel. Vickers hardness 227  HRC 23 has
been used so far to limit the testing time
1100
Up to 200 mm

Propeller blade with var. pitch angle. Standard pitch angle is 10

17, 31, 43, 51 and 60 HRC

0200 mm/min
03000 N
032 Nm
Atm 6 bars (4 bars with cont. foam injection)
10 mm (for consistent and comparable results)

Not applicable
Not applicable
Not known, but torque is measured
Atm 10 bars
Published results include D50 ranging from 0.5 to 7 mm (Rostami
et al., 2012a,b)
Not applicable (compaction under the propeller blade during the
test)
Premix and continuous addition through pre-installed ports

Manually by proctor hammer prior to testing. Compaction as

desired.
Continuous addition through the drilling tool

60180 (tested so far)

Fixed position, with 150 mm soil above and below the propellers

3. Initial test results

Fig. 5. Example of relation between thrust force and required torque for achieving a
xed penetration of 40 mm/min for one soil sample.

The continuous data collection enables the analyst to nd how

varying operation parameters inuence each other. Fig. 5 shows an
example of thrust force and torque correlated in the SGAT software.

The initial results are obtained by testing a soil sample with

grain size between 0 and 6.4 mm (Fig. 7), a soil from an ongoing
European soft ground TBM project and a soil originating from an
upcoming soft ground project in Asia. As reference, Soil Abrasion
Test, mineralogy by XRD and grain size distribution analyses
have been performed in addition to the SGAT values, see Table 2
for sample mineralogy and SAT values.
In the initial testing scheme, some SEM images have been taken,
in order to show the degradation mechanisms on the steels microstructure. Fig. 8 shows abrasive wear, and Fig. 9 shows tribo-corrosive wear, which is a synergy of abrasive wear and corrosive wear.
There has been observed degradation in the micro-structure due to
corrosion in short tests (Grdal et al., 2012). The SEM photos
showed in Figs. 7 and 8 originates from SGAT tests with a 40 min
duration. The corrosive effect has not been possible to detect quantitatively by weight loss, meaning that the SEM photos are the only
evidence to show the effect (Grdal et al., 2012).
Observations and explanations on how the soil compaction/
density, pressure, and introduction of soil conditioning additives
inuences the abrasivity and torque measurements are presented
in the following.

Fig. 6. The grain size distribution of Soil samples 1, 2 and 3.

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 38 (2013) 398408

403

3.2. Inuence of rpm

The inuence of rpm on the recorded wear by SGAT has been
evaluated on Soil sample 3, see Fig. 9. The testing conditions are
xed at 40 mm/min penetration and 9% water content. The gure
also illustrates the variation of test results at identical soil conditions, which seems to increase with increasing rpm.
3.3. Inuence of earth pressure mode

Fig. 7. SEM image showing abrasive wear on the SGAT tool steel after exposure to a
dry soil sample. The photo is of the lower steel bar on the SGAT tool (photo
Christian Kreyberg Grdal).

Fig. 8. SEM image showing a combination of abrasive wear and corrosion on the
SGAT steel tool after exposure to a soil sample with 8 weight% water. The photo is of
the lower steel bar on the SGAT tool (photo Christian Kreyberg Grdal).

In order to simulate the face pressure system at shielded TBMs,

the test procedure allows running tests in pressurized mode. The
pressurized mode utilizing air which is applied through a valve
in the lid of the test chamber. A series of tests have been conducted
by adding an over-pressure from atmospheric conditions to 5 bars.
The rotation speed of the drilling tool was 50 rpm, with a varying
torque, 400 mm total vertical travel length (200 mm downwards
and 200 mm upwards) of the drilling tool, and 5-weight% water
content in the soil sample. The up and down movement was performed in order to evaluate the pressures possible inuence on
direction inside the test chamber.
The weight loss (wear) at different pressures is presented for
the lower steel bar (test piece A in Fig. 10) and the upper steel
bar (test piece B), as well as the total weight loss in Fig. 10.
The ndings in Fig. 10 corresponds well with the Rostami
et al. (2012a,b) ndings, which conclude that the amount of
additional steel wear due to increased ambient pressure is not
signicant. However, the ndings presented in this paper only
take into account a few tests on one single, relatively uncompactable soil sample. It is therefore necessary to conduct further
testing to conclude, if hyperbaric pressure inuences the wear
rate.
The example of the face pressures inuence on the tool wear is
based on the relatively single-graded soil lacking nes (sample 1 in
Fig. 6). The possible effect of the support pressures inuence on the
tool wear will be evaluated for a more compactable soil with water
content close to the saturation point, at a later stage.
3.4. Inuence of moisture and soil conditioning additives

Fig. 9. Recorded inuence of rpm on the SGAT weight loss for sample 3.

3.1. Inuence of different soil compaction

The inuence of compaction on tool life and torque has been
measured on Soil sample 1. A proctor hammer has been used to
compact the tested soil samples. The varying grade of compaction
has been achieved by varying the number of applied blows.
The results summarized in Table 3 are based on a rotation speed
of 50 rpm, a varying torque, 200 mm vertical travel length and
40 mm/min penetration rate of the drilling tool and 5 weight%
water content in the soil sample.

The development of soil abrasivity for various moisture contents, when exposing the SGAT drilling tool to 1000 mm drilling,
50 rpm and 100 mm/min penetration speed, with stepwise drilling
(50, 100, 150 and 200 mm) has been evaluated by Jakobsen et al.
(2012). The stepwise drilling involves 50 mm drilling down,
50 mm retraction, 100 mm drilling down, 100 mm retraction,
etc., until the 200 mm depth is reached and retracted. The stepwise
drilling was used in 2012, in order to mix the soil conditioning
additives with the soil, prior to the development of continuous
conditioning (Jakobsen et al., 2012). For comparison the same
development has also been evaluated for 400 mm drilling length,
50 rpm and 40 mm/min penetration speed. Fig. 11 shows the inuence and importance of moisture content on the measured weight
loss and torque on the SGAT.
The development of abrasiveness for varying water contents
(Figs. 11 and 12) corresponds well with Rostami et al. (2012a,b)
tests on the Penn state abrasion testing system. The increase of
water content has previously showed a general increase of wear
by using the Ball Mill Test. (Jakobsen et al., 2009) and (Klemetsrud,
2008). However, the reduction of wear after reaching a specic
water content has not been observed previously with the Ball Mill
test.
The Norwegian Geotechnical Institute (NGI) performed a study
on soil compactibility, dependent on different moisture contents in
the early 1980s in order to evaluate the tightness of rock ll dams
(Damgruppen, 1983). The main conclusions of this study were that

404

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 38 (2013) 398408

Fig. 10. Example of relation between weight loss (abrasion) and face support pressure (bars) for Soil sample 1.

Fig. 11. Example of soil abrasivity development for various moisture contents with
the same compaction procedure (5 blows with the proctor hammer in 4 layers). The
graph also presents the development of torque for Soil sample 3 with different
moisture contents.
Fig. 13. Inuence of the compression work on soil density. Relatively low water
content gives a higher density. For higher water contents pores will be easier to
close with less compression work (Damgruppen, 1983).

Fig. 12. Compaction curves from a natural moraine (Damgruppen, 1983).

the dry density obtained by compaction is highly inuenced on the

water content in the soil sample. Single grains have a high strength
if the soil is relatively dry. This makes it impossible to ll voids be-

tween the hard lumps. Thus, the dry density is relatively low. If the
water content is increased, the soil gets more plastic and during
the compaction the voids will be closed, resulting in a higher density (Damgruppen, 1983). This nding can explain the inuence of
water content on soil density, and thus the soils potential to cause
abrasive wear on an excavation tool. See Fig. 12 for density development related to water content and saturation, and Fig. 13 for the
inuence of compaction work on soil density.
An evaluation of the possible benets by adding soil conditioning additives was carried out. The additives were added as (a)
foam on top at the soil sample or (b) as a continuous foam injection through small foam injection nozzles 2 cm behind the drilling tool. Initially a pre-mix of soil conditioning additives and
soil had been tried, but the results from this approach were discarded as the sample rheology deviated from the reality in front
of a TBM.

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 38 (2013) 398408

405

Fig. 14. Soil abrasivity development for conditioned soils. Left gure shows a soil from a natural deposit close to Trondhein, and Right shows results on a soil sample
originating from a tunneling project in Europe (Jakobsen et al., 2012).

3.4.1. Tests with foam addition on the top of the sample (testing
condition a)
The results presented in Fig. 14 shows that adding foam on the
top of the soil sample reduces the weight loss of the drilling tool, as
well as the torque. It was however discovered that the foam did not
properly mix with the bottom part of the soil sample. This might
indicate a too high Foam Injection Rate (FIR) in the upper part of
the soil sample and a non-existing conditioning of the lower part
of the soil sample. The densities of the two soils presented in
Fig. 13 are approximately 2100 and 1900 kg/m3, and the results
do not have torque measurements. The tests were done in atmospheric pressure conditions with foam expansion ratio (FER) of
10 and foam injection ratio (FIR) of 30%. The results shown in
Fig. 14 are obtained on moisted soil sample prior to testing.
The results obtained by foam addition on the top of the sample
seem to indicate benets of soil condition, in terms of reduced
wear and torque. However, the upper part of the soil sample is
likely to be over-conditioned, while the lower part remains under-conditioned to unconditioned. Quantication of varying grades
of soil conditioning, subsequently the test is not done. However, in
all the conducted tests, it appears that the top of the soil sample
(10 cm) is over-conditioned and the lower part of the soil sample
is gradually exposed to less soil conditioning additives. This effect
will again not correctly indicate the effects and benets of soil conditioning agents.
3.4.2. Continuous foam injection (test condition b)
In order to achieve a proper continuous foam injection and
hence an evenly conditioned soil sample, a total of three different

Fig. 15. Example of soil abrasivity development for Soil sample 3 for different
moisture contents, and the inuence of the foam injection ratio (FIR).

tool designs have been evaluated. So far, the most successful tool
design is shown in Fig. 2. Prior to the design showed in Fig. 2, ejection of soil conditioning additives was tried from pipes 1 and 4 cm
behind the lower steel bar. The results achieved by running tests
with continuous soil conditioning are shown in Figs. 15 and 16.
Fig. 15 repeats the indication of the strong relation between
moisture content and wear, and shows the high inuence of soil
conditioning additives injection rate. In the example showed in
Fig. 16, the wear is reduced down to less than 20% of the initial value dependent on the moisture content. The torque and thrust
were reduced by more than 30% for the xed penetration rate of
40 mm/min, on Soil sample 3 with 15% water.
Several techniques have been evaluated for the continuous
addition of soil conditioning additives into the soil. The rst attempt was based on foam injection through two holes at the SGAT
drill-shaft about 5 cm above the drilling tool. The second version
used foam injection through the upper drilling tool (part B),
whereas the nal and currently used version (Fig. 4b) uses a foam
nozzle at the level of the lower drilling tool (part A) which is in
contact with the compacted soil directly corresponding with
the foam injection at the TBM cutterhead. Only this modication,
by being able to apply the additives exactly at the contact zone between the drilling tool and the compacted (virgin) soil, allows
SGAT test results to be directly translated to effects in EPB TBM
tunneling.
In Fig. 15 the strong inuence on wear by the soils water content can be observed together with the inuence of continuous
foam injection. Further testing in this manner needs to be carried

Fig. 16. Example of the inuence of proper soil conditioning for Soil sample 3. The
wear is reduced to approximately 1/5, and the torque and thrust to approximately
2/3 of the untreated soil.

406

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 38 (2013) 398408

out, in order to nd good and bad soil conditioning for various

soft ground samples.
3.4.3. Premix of soil and soil conditioning additive (test condition c)
In order to evaluate the possibility of premixing soil samples
with soil conditioning additives some trial tests have been conducted with FER 10 and FIR of 50%. The tests were conducted with
rotation speed of 50 rpm and with a total of 400 mm vertical
movement of the drilling tool (200 mm downwards and 200 mm
retraction upwards). A small concrete mixer was used to make
the premix of soil and soil conditioning additives.
The results obtained by testing a premix of soil and soil conditioning additives do not show the reality between soft ground
excavation tools and the conditioned soil, as the recorded weight
loss and torque are very low, and in the same range as what is expected for dry testing. For Soil sample 1 the weight loss ranged
from 5 to 7 mg and the torque from 6 to 7 Nm in premixed soil
and soil conditioning additives. Therefore, the reduced torque
and wear is not relevant for TBM tool life research or other phenomena taking place at the cutterhead level. However, these results can only be relevant for estimating the conditions in the
EPB TBM working chamber behind the TBM cutterhead.
In order to evaluate the inuence of soil conditioning additives
correctly, their introduction technique is of high importance. The
only way to obtain comparable results to the TBM cutterhead situation is using a continuous injection of soil conditioners at the cutterhead tool which is penetrating into a consolidated ground.
4. Discussion
4.1. General
The ambition and purpose of the design of the test and the applied test procedure is to replicate an in situ soil TBM tool contact, in a small and simplied scale. The drilling tool was
designed in a way which is causing a relative small area of initial
contact between the tool and the soil.
The main differences between the SGAT and the existing Penn
state soil abrasion testing system are the design of the drilling tool,
the rotation speed and penetration of the tool and the possibilities
to introduce soil conditioning additives (e.g. foam or bentonite)
during the test. The new Soft Ground Abrasion Tester (SGAT) does
in addition allow testing of soil samples with a dened compaction. Table 3 shows the similarities and differences between the
SGAT and Penn state soil abrasion testing system.
The SGAT apparatus has the possibility of drilling through soil
and soft ground samples, hence close to real TBM conditions in soft
ground.

The limitation of the presented SGAT test procedure as compared to the real life TBM boring process consists mainly in the
limit of the soils grain size. The current tool allows 10 mm large
grains to be included in the soil sample. Thus, grains above
10 mm need to be removed from the soil sample material prior
to testing with the current drilling tool design. The limitation will
not be substantial, as the test is designed to test soft ground conditions. However, in soils containing large amount of gravels and
stones, the current test may not be equally suitable.
Our analysis nds the test to be torque sensitive of the position
of the drilling tool, indicating an increasing soil compaction towards the bottom. This effect is most likely induced by the layered
Proctor hammer compaction procedure.
Equally, we nd a clear relation between the measured tool
wear and the required torque, as well as increasing tool wear by
increasing rpm. As the torque increases, the contact forces between
the steel tool and soil increases, causing a higher potential for degradation of the steel. The torque has also proven to show grain size
variations in the soil sample quite well. A limitation in the torque
measurement is an uneven compaction through the soil sample.
The lowest part of the sample probably has a higher compaction
than the upper part, due to the layered compaction procedure.

4.2. Test relevance and repeatability

The SGAT apparatus still lacks of a detailed test procedure intended for commercial use. The test procedure presented in this
paper is preliminary, and might be changed as more data are measured and compared to real torque, thrust and tool life data from
TBMs. The test apparatus is designed to evaluate several variables
inuence on abrasive wear and torque. Thus, the test procedure
should be decided prior to testing of a new batch of sample material, based on what the test results should show or not.
In order to quantify the reliability a total of 10 tests on Soil sample 1 with 10% water content, 50 rpm and 40 mm/min penetration
were carried out. The standard deviation of these tests were 6.3,
which is acceptable taking into account the sources of errors present in testing of geo material with possible varying distribution of
water content and compaction.
For assessing the validity of the automatic torque and thrust
recording, manually measurements with a scale and torque
wrench have been carried out. The ndings in Fig. 17 show that
there are not any inconsistency between the collected thrust and
torque values in the SGAT apparatus.
A few relations between SAT values and SGAT values are
shown in Fig. 18. The measured Soil Abrasion Test (SAT) values
do not range in accordance with measured wear on the Soft
Ground Abrasion Tester (SGAT), Soil sample 3 has the lowest SAT

Fig. 17. Relation between data log values and measured values for torque and thrust on the SGAT apparatus.

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 38 (2013) 398408

407

pressure, compaction and soil density. In addition, the important

inuence of different types of soil conditioning additives can be
evaluated.
The initial results presented and discussed in this paper are very
promising for evaluating various geotechnical parameters inuence on soft ground abrasivity. The TBM operations inuence on
tool wear can also be evaluated by adjusting the apparatus rpm,
penetration rate, thrust and soil conditioning parameters (Foam
Expansion Rate (FER) and Foam Injection Rate (FIR)).
5.1. Main ndings

Fig. 18. Relations between SAT values and weight loss measured on the SGAT
apparatus.

value and the highest measured SGAT wear. The main reasons for
this are most likely related to the inuence of compaction and soil
humidity, which are not taken into account by the SAT, as well as
mineralogy of Soil sample 3, consisting of nes consisting of lowabrasive minerals mica and calcite, with coarser particles consisting of quartz. It is believed that the nes creates a cohesive paste
holding the coarser abrasives, causing the high weight loss. The
SGAT ndings has also shown that a calculation of a more reliable
wear index could be achieved by combining the measured SAT
values with factors for in situ soft ground properties, like humidity
and compaction.
4.3. Suggestions for further work
The initial testing of the SGAT apparatus and method comprises
only three soil samples. In order to gain more experience and
knowledge on how various soil types (clay, silt, sand and gravel)
behave when they are exposed to various compaction grades, use
of soil conditioning additives and pressure, there is a need for further testing.
For pressurized testing conditions, the tests presented in this
paper are conducted on 5% water content. This is relatively far from
reality as most pressurized TBM performances are below the
ground water table. Further testing on more soil types and with
water content close to the saturation point is therefore needed.
The apparatus enables a unique testing procedure being very
close to the excavation conditions at a real TBM. The test results
obtained with the SGAT apparatus is, however, so far not correlated
or validated against any real TBM excavation. This needs to be done
in order to evaluate the scaling effect between the SGAT apparatus
and a real EPB TBM. Such a study will also comprise an evaluation
of the necessity and relevance of distinguishing between primary
and secondary wear on SGAT test pieces.
In order to evaluate the relation between SAT and SGAT values, more testing is needed. The authors are currently working
on a SAT based estimate on tool life, where the SAT values
are adjusted with other relevant geotechnical values.
In this current paper, the soil rheology is missing. Generally it
should be evaluated in connection with pre-investigation and evaluations of soil conditioning additives. For the further research on
the SGAT apparatus, we will therefore initiate to run ow table
mortar testing according to EN 413-2 and EN 459-2 in order to
check the conditioned soils rheology for EPB TBM applicability.
5. Conclusions
The set-up and design of the apparatus has the capability to
evaluate how soft ground abrasivity is inuenced by water content,

 Wear on steel excavating soft ground in the new SGAT apparatus is clearly inuenced by
The nature of the soil (e.g. mineralogy, quartz content, abrasiveness, grain size distribution, compaction).
The moisture of the soil inuences the wear (weight loss) as
high as 500%.
Type and method of soil conditioning (soil conditioning type,
FER, FIR) can reduce the wear rate down to 20% of the unconditioned sample.
 The pressure added to the test chamber did not show any significant inuence on the measured soft ground abrasivity for the
soil material with 5% moisture content used in this initial
research.
 There is a clear correlation between the measured wear and the
recorded torque, as well as rpm by the SGAT apparatus.
 The correct use of soil conditioning additives, apart from the
above mentioned wear reduction, has clear effects on
Reduction of torque by approximately 40% in some cases.
Reduction of necessary SGAT penetration thrust by approximately 40% in some cases.
 Furthermore, the differences between good or bad soil conditioning can now be quantied, and results from the SGAT apparatus can be used to evaluate and to improve the effect of soil
conditioning additives.

Acknowledgements
The authors would like to thank the Norwegian Tunnelling Society for partly nancing the development of the Soft Ground Abrasion Tester, Dr. Art Jardar Lohne for proof reading the paper, as well
as the students Ivar Sletta, Andreas Hauso, Ole Fredrik Brattberg
and Leon Eide for their assistance in running tests and developing
the test apparatus.
References
Barzegari, G., Uromeihy, A., Zhao, J., 2013. A newly developed soil abrasion testing
method for tunnelling using shield machines. Quarterly Journal of Engineering
Geology and Hydrogeology.
Damgruppen, N., 1983. Fyllingsdammer. Norwegian Geotechnical Institute, Oslo
Norway.
Gharahbagh, E.A., Rostami, J., Palomino, A.M. 2010. Tool wear issue in soft ground
tunneling, developing a reliable soil abrasivity index. In: Presented at the North
American Tunneling Conference.
Gharahbagh, E.A., Rostami, J., Palomino, A.M., 2011. New soil abrasion testing
method for soft ground tunneling applications. Tunneling and underground
space technology journal. Tunnels and Underground Space Technology 26 (5),
604613.
Gharahbagh, E.A., Rostami, J., Talebi, K., Ibarra, J. 2013. Experimental and practical
study of impact of soil conditioning on soil abrasion and cutter wear of EPB
TBMs. In: RETC Conference June 2326, Washington, DC.
Grdal, C., Equey, S., Armada, S., Espallargas, N. 2012. Effect of soil and rock
composition on the wear process of cutter tool steel used in tunnel boring
machines. In: Presented at the NordTrib Conference, Trondheim.
Gwildis, U.G., Sass, I., Rostami, J., Gilbert, M.B. 2010. Soil abrasion effects on TBM
tunnelling. World Tunnelling Congress, Vancouver, Canada.

408

P.D. Jakobsen et al. / Tunnelling and Underground Space Technology 38 (2013) 398408

Jakobsen, P.D., Becker, T.E.E. 2012. Tunnelling in abrasive soils review of a tunnel
project in Germany. In: The 13th KTA International Symposium on Mechanized
Tunnelling Technology, Seoul.
Jakobsen, P.D., Bruland, A., Dahl, F., 2013. Review and assessment of the NTNU/
SINTEF Soil Abrasion Test (SAT) for determination of abrasiveness of soil and
soft ground. Tunnelling and Underground Space Technology 37, 107114.
Jakobsen, P.D., Dahl, F., Langmaack, L., 2009. Reduction of hard rock abrasiveness
using foam agents. Tunnel 8, 4047.
Jakobsen, P.D., Langmaack, L., Dahl, F., Breivik, T., 2012. Prediciting the abrasivity of
in-situ like soils. Tunnels and Tunnelling International (June), 4144.
Jakobsen, P.D., Lohne, J. 2013. Challenges of methods and approaches for estimating
soil abrasivity in soft ground TBM tunnelling. WEAR, in press.
Klemetsrud, K. 2008. Soil Abrasion in Shield Tunneling. M.Sc. Thesis, NTNU.
Khler, M., Maidl, U., Martak, L., 2011. Abrasiveness and tool wear in shield
tunnelling in soil. Geomechanik und Tunnelbau 4 (February), 3653.

LCPC, 1990. LCPC Abrasivemeter Standard. Normalisation Francaise P18-579.

Nilsen, B., Dahl, F., Holzhuser, J., Raleigh, P., 2006c. SAT: NTNUs new soil abrasion
test. Tunnels and Tunnelling International (May), 4345.
Nilsen, B., Dahl, F., Raleigh, P., Holzhuser, J. 2007. The new test methodology for
estimating the abrasiveness of soils for tbm tunnelling. In: Proceedings of the
Rapid Excavation and Tunneling Conference (RETC), pp. 104106.
Rostami, J., Gharahbagh, E.A., Palomino, A.M., Mosleh, M., 2012a. Development of
soil abrasivity testing for soft ground tunneling using shield machines.
Tunnelling and Underground Space Technology 28, 245256.
Rostami, J., Mosleh, M., Palomino, A.M., Gharahbagh, E.A. 2012b. Study of soil
abrasivity and development of a reliable soil abrasivity index. NSF Soil Abrasion
Study Report, Grant Number: #0928757.
Thuro, K., Singer, J., Ksling, H., Bauer, M. 2007. Determining abrasivity with the
LCPC test. In :Proceedings of the 1st Canada U.S. Rock Mechanics Symposium.

Paper 10
Anti-wear and anti-dust solutions for hard rock TBMs
Langmaack, Lars
Grothen, Brad
Jakobsen, Pl Drevland
World Tunnelling Congress, Vancouver 2010

Anti-wear and anti-dust solutions for hard rock TBMs

L. Langmaack1, B. Grothen2, P. D. Jakobsen3
BASF Construction Chemicals, Zuerich, Switzerland; 2The Robbins Company, Solon, USA;
3
The Norwegian University of Science and Technology (NTNU),Trondheim, Norway

1. Introduction
Abrasive wear on cutter tools and high levels of dust are common processes causing problems in
hard rock TBM tunneling. Wear and damage on cutter tools are mainly caused by crushed rock
powder in the cutter grooves and from falling rock from the tunnel face. Wear rates on cutter
tools are also observed to increase when boring without water, most likely because of high
temperatures and inability to remove fines in the cutter grooves [1]. To cope with abrasive wear
from fines, dust suspension and high temperatures, one approach can be utilization of chemicals.
The use of water, water based agents or foam on hard rock tunnel boring machines is not new
technology, and dates back to 1973 and earlier. They were or are still used, more or less
successfully, in order to reduce the amount of dust and to reduce the dust-related problems.
2. State of the Art Methods
2.1. Wear Reduction
Improved performance of cutters is an area of continued research with substantial improvements
made over the past fifty years. Recent research has addressed multiple facets of cutter design
with cutter life being an important design objective. Some examples of areas that are evaluated
with respect to cutter life are the cutterhead profile, cutter tip profile and cutter metallurgy. The
number of cutters changed versus their positions is tracked, with a higher rate of wear being seen
at the positions where face cutters transition to gage cutters. This reduction in cutter life in the
transition region is due to an increase in both the rate of cutter ring wear and the number of
blocked cutters that occur in the region. The area has been shown through the use of strain
gauges to have higher loading than face cutters with a comparable spacing. Cutterhead design
has evolved over time using tests and experience to refine position and mounting to extend cutter
life.
A second area that has been developed to deliver performance while providing a good service life
is the cutter profile. The tip width is chosen to give good penetration into the rock while still
providing sufficient strength to maintain the integrity of the rings and minimize edge chipping,
which is seen on the high cutter loading of modern machines. Cutter metallurgy is also used to
combat this issue, with many different steels being used over time. Disc rings were historically
made from bearing quality steels. Robbins cutters are made of steel with a proprietary chemistry
and hardening process that has been progressively refined to provide higher hardness without
the loss of fracture toughness, thereby minimizing chipping.
2.2 Dust Reduction
Dust is an inherent part of tunneling in rock. New regulations in many countries highlight the
need to control airborne dust, with specific focus on quartz-containing dust. Dust is currently
controlled in two different ways on a TBM. First, it is captured at point of origin to limit excessive
airborne particles at the face of the tunnel. Second, water is brought to the face through a rotary
union, where it is distributed to equally-spaced spray nozzles on the face of the cutterhead. As

the rock is cut the water spray captures the dust and removes it through the mucking system.
Water spray bars may also be installed at dust producing areas, such as conveyor transition
points on the backup, and in the muck car loading area. Once the dust becomes airborne it is
removed though the use of a dust scrubber. Again this is done primarily in the area of the
cutterhead. Large fans pull the dust-laden air through ducting to a dust scrubber located on the
back up. Fresh air is brought forward to help replace the air that is removed. The scrubber can
be either a wet scrubber, which again uses a water spray to capture and remove the airborne
dust, or a dry scrubber that forces the air though a series of filters. In both cases the captured
dust is added back to the muck stream and removed from of the tunnel.
The following methods are of special interest [2]:
- Water sprays: wetting and airborne capture
Of the two, wetting of the broken material is far more effective. Adequate wetting is extremely
important for dust control. The vast majority of dust particles created during breakage are not
released into the air, but stay attached to the surface of the broken material [3]. Wetting this
broken material ensures that the dust particles stay attached. As a result, adding more water can
usually (but not always) be counted on to reduce dust [4,5,6].
- Water additives: foam and wetting agents
For dust control, foam works better than water. It provides dust reductions of 20% to 60%
compared to water. Foam also can produce similar results at lower water use. The amount of
water needed to make the foam is less than the equivalent water spray. High-expansion foam,
when compared to water sprays at a belt transfer point, averaged an additional 30% dust
reduction [7]. Foam released from a longwall shearer drum cut the dust an additional 50%
compared to conventional water sprays on the drum [8]. Also, the system used one-half the water
of the conventional sprays. The drawback of the foam was high cost. Like water, foam works best
when it is mechanically mixed with the broken material. A comprehensive review of foam for dust
control in mining and minerals processing has been given [9].
Wetting agents receive a disproportionate amount of attention, perhaps because they seem to
offer an easy fix to dust problems. Most of the interest has been in coal mining because of the
hydrophobic nature of coal. The effectiveness of wetting agents has been the subject of
considerable research over the years, without much of a definitive answer on how well they work.
3. The use of Foams and Polymers
BASF has taken a fresh view on the use of foam on hard rock TBMs, believing that its use can be
much more effective than the common water sprays and also more (cost) effective than described
above. The effectiveness of foam strongly depends on the way the foam is generated and on how
it is used improvements can be made here. Furthermore, a possible incorporation of anti-wearadditives into the foams or the development of foamable polymers represents an interesting dual
role for the new additives. The increase of dust catching effectiveness together with their new
anti-wear-capacities will reduce the above mentioned high cost draw-back of the use of these
additives.
3.1. Laboratory Results and Interpretation on Construction Time and Wear
Imitations of the process of hard rock drilling with TBMs have been tried by several universities
and researchers. Two of the most common methodologies for imitation of hard rock TBM
tunneling with respect to advance rate, cutter consumption and cost estimates are the NTNU
method consisting of the Drilling Rate Index (DRI) and Cutter Life Index (CLI), and the
Colorado School of Mines method based on individual cutter forces to determine drilling
advance[10]. In addition the Cerchar Abrasivity Index (CAI) is recognized as a quick measure on
a rocks abrasivity and expected cutter consumption on TBMs.
The NTNU model is based on empirical relations between rock parameters obtained from
laboratory and field, such as DRI, CLI, porosity, fracture class, quartz content and TBM
parameters [11]. The empirical relations are established with a basis of 30 different tunnels (250
km) with respective TBM production data and wear records on cutter tools. It has been tried to
use the NTNU model to check theoretically how the use of anti abrasive agents influence
advance rate, cutter consumption and relative tunneling cost.

The DRI is based on the rocks brittleness and surface hardness, and the CLI is based on the
surface hardness and abrasion properties of the rock type. The brittleness value is evaluated in
an apparatus which is based on 20 stroking impacts on a known fraction of a rock sample. After
the impacts, the percentage of the crushed down rock represents the measured brittleness value.
The surface hardness measure is obtained by the Sievers J apparatus, which drills miniature
holes in a rock sample. The depth of these holes is the surface hardness measure.
The final rock property needed to be established before calculation of DRI and CLI is the
abrasion. The NTNU model uses an in-house built apparatus consisting of a rotating steel disc
applied with crushed rock powder or soil. The soil or rock powder has to pass a cutter ring piece,
which causes a measureable weight loss - the result of which is our abrasion value. The
abrasion apparatus was used to measure reduction of abrasion by introducing the MEYCO ABR5
anti-abrasion agent on one rock sample from Ltschberg in Switzerland and one rock sample
from the AMRII project in India. Results and classification of results are showed in Table 1. The
use of additives influences the abrasion test and subsequently the CLI, whilst brittleness values
and surface hardness properties are the same with and without additives.

Figure 1. Schematic introduction of brittleness testing (left), Sievers J (top right) and abrasion
testing (bottom right).
For further and detailed description of laboratory test procedures and pictures needed for the
NTNU model please refer to http://www.drillability.com.
Table 1. Drillabillity indicies and classification for evaluated rock samples.

The NTNU advance rate, cost and cutter consumption model have been used in order to go one
further level in the evaluation of ABR5. The evaluation has been done with a software called
fullprof which is provide quick estimates of the NTNU model [12]. The summary of the estimation

is showed in Figure 2, and the estimation indicates an increase of weekly advance rates and
increased cutter life.
The input parameters are drillability indices as presented in Table 1, with rock mass classification
I, which is equal to average spacing of 80 cm between fissures and joint system in the rock mass.
To excavate the Ltschberg rock it is assumed a hard rock gripper TBM with 51 cutters of 19
inches and average cutter thrust of 260 kN per cutter. For the excavation of India rock it is
assumed a hard rock gripper TBM with 67 cutters of 19 inches and 312 kN per cutter in average
thrust.
140%
130%

Relativepercentage

120%

Ltschberg
Ltschbergfoam
India

110%

Indiafoam

100%
90%
80%
70%
60%
50%
Netadvancerate

Weeklyadvancerate

Cost

Averagecutterringlife[h]

Figure 2. Relative comparison of estimated advance rate, cost and cutter ring life.
3.2.

Site Example No1: Guadarrama High Speed Rail Tunnel, Spain

The Guadarrama tunnels belong to the new high speed railway link between Madrid and Oviedo.
For this project the Guadarrama mountains between Madrid and Valladolid had to be crossed
with a 28.400m twin tube hard rock tunnel. Totally 4 TBMs 2 Herrenknecht and 2 Wirth were
used and finished boring in June 2005.
characteristics Wirth TBM:
diameter: 9,46m; total installed power: 5.600 kW; cutterhead torque: 27.000 kNm
characteristics Herrenknecht TBM:
diameter: 9,51m; total installed power: 5.500 kW; cutterhead torque: 20.000 kNm
3.2.1.

Project description

The geology along the alignment showed 85% metamorphic and igneous rock, 10% weathered
rock conditions as well as various fault zones. The 620m long Umbria fault with nearly loose
ground conditions represents the longest of its kind, several others last for 10-20m only leading
to a double shield TBM concept.
Nevertheless, the intact granite sections showed UCS values of up to 280 MPa, in addition
containing quartz contents of up to 80%. This resulted in measured Cerchar Abrasivity Index
values between 5 and 6, classified as extremely abrasive to quartzitic.
In light of these predictions, the JV decided to consider the possibility of using anti-wear-additives
and to evaluate their benefits on-site. In order to use these anti-wear-additives efficiently, the
TBM needs to be adapted to their use.
3.2.2.

Necessary adoption of the TBM

The MEYCO ABR 5 anti-wear-additive must be supplied to the cutter head as foam. In
consequence, it requires some modifications of the TBM and additional installation:
- Foam System
A foam system is necessary in order to foam up the anti-wear-agent. Similar to the foam systems
used on Earth Pressure Balance (EPB) TBMs, the following components are mandatory to be

installed: water (supplied by water booster pump), dosing pump (for the correct dilution of the
anti-abrasion-agent into the water), compressed air, foam guns (which create the foam out of
compressed air and the foaming solution) and a regulation system for each foam gun.
Unlike the fully computerised versions on the EPB TBMs, the system installed on the hard rock
TBM was manually operated due to lower investment costs and quite steady output values.
- Foam nozzles
The foam has to be injected through special designed foam nozzles on the TBM cutterhead. The
existing water sprinkler nozzles as standard equipment on the cutterhead will destroy the foam
and must be replaced. In the case of the 9,5m diameter Guadarrama TBM, the above indicated 5
injection points on the cutterhead have been chosen to be changed into foam injection points in
order to ensure a homogeneous and even foam distribution on the cutterhead.
- Rotary coupling (rotary swivel)
The normally installed water splitter box cannot be used for the anti-wear-additives. Only the
installation of a rotary coupling ensures specific outputs per foam injection point representing a
key factor for the successful use of the anti-wear-additives.
Generally, existing hard rock machines can be upgraded to the use of modern anti-dust and antiwear-additives. Nevertheless it is strongly recommended to study especially the installation of the
rotary coupling during the TBM design stage, reducing considerably the later upgrading costs
without increasing the total TBM costs significantly.
3.2.3.

Site test results

Altogether, some 600 tons of MEYCO ABR 5 anti-wear and anti-dust-agent were used on the
Guadarrama High Speed Railway Project in Spain.
The following benefits associated with their use were reported:
- Cutter wear reduction
A wear reduction of > 15 % was achieved, resulting in 25 30 hours per month less down time
due to less cutter changes. This downtime was then used for additional excavation.
Disappearance of blocked cutters using MEYCO ABR 5.
- Clean cutter tools
The rock dust with is created during the boring process, can agglomerate on the disks as shown
above when using water. This implicates a time consuming cleaning process before these disks
can be changed, there may be the risk of clogging the disk window and last but not least the
grinding paste formed by the stone dust & water increases wear. At Guadarrama the use of
MEYCO ABR 5 prevented the stone dust from creating this paste and the tools remained clean
(see figure 3).

Figure 3. dust agglomeration difference: left side water use, right side MEYCO ABR 5.
- Temperature reduction
The use of MEYCO ABR 5 resulted in a significant temperature decrease from 90 150C to
around 70C, resulting in shorter down time due to less cooling and waiting time.

- Improved muck transfer and a dust free working environment (see Figure 4).

Figure 4. Dust & transport differences: left side water use, right side MEYCO ABR 5.
- Reduction of water usage and less water reclamation
When using MEYCO ABR 5, the amount of injected water was reduced from originally 310
litres/m3 excavated rock down to 50-100 l/m3.
Key assumptions made for the following roughly-estimated benefit calculation:
- TBM speed of around 50mm/minute
- monthly advance rate of 500m
- fixed TBM costs around 2000 /h
With 70.000 MEYCO ABR 5 monthly product costs it was possible to reduce the wear &
maintenance in this case of more than 15%, which can be back-calculated to a reduction of
maintenance and material costs higher than 50.000.
Anti-abrasion-agents are still useful because the reduction of maintenance has not only a direct
cost influence but realises also considerable time savings. The 15% of reduced downtime can
directly be translated into 80-90m of extra excavation per month turning the above calculation
with an initial loss of 10-20.000 into final savings of some 40.000 per month.
In addition, the above quick benefit calculation does not even take into account important benefits
such as a nearly dust free environment, more convenient and quicker changing of discs (due to
lower temperature and clean discs), drastic reduction of sprinkling water and reduced energy
consumption due to less exhausting.
Knowing that these effects do have a significant cost influence in many projects, BASF and
Robbins decided to have a deeper and broader look into these parameters by launching a
copious on-site test program at the Indian AMR II project.
3.3 Site Example No. 2: AMR II Water Transfer Tunnels, India
3.3.1 Project Description
The Alimineti Madhava Reddy (AMR) Project is a water project located near the city of
Hyderabad in Andhra Pradesh, India and is part of a much larger water transportation scheme.
The region is one of the most arid in India with only 925 mm of annual rainfall. Local water
supplies to 500 area villages are contaminated with fluoride levels that far exceed guidelines.
This is being addressed by a system of canals and tunnels, which contains over 100 km of canals
and one of the longest TBM driven tunnels ever constructed in India.
There are two main projects emanating from a common reservoir to supply water to four districts
in Andhra Pradesh, one of which is the AMR project. The main tunnel will be constructed using
two Robbins 10 meter diameter double shielded machines boring from opposite ends. The main
tunnel is 43.5 km long and will connect the Srisailam Reservoir to a balancing reservoir on the
Dindi River for transfer during the monsoon months. A second 7.3 km long tunnel will then
distribute the water to a network of canals to the plains of the Nalgonda District, where it will be
used to irrigate farmland and provide potable water to 516 villages.
The geology is generally very stable, as this section of the country is part of the South Indian
Peninsular Shield made up of two primary rock types: quartzites and granite. The machines will

excavate in both rock types with the quartzite zones having compressive strength up to 450 MPa,
and layered with shale for about 60% of the tunnel length. The granite is expected to have a
range of 160 to 190 MPa (23-28 ksi) and makes up the remainder of the tunnel. The quartzite
sections are of particular concern with respect to cutter cost due to significant abrasiveness and
high strength. The quartzite section can also be very blocky in nature, which can increase cutter
wear due to the damage caused by impact loading. The Robbins DS325-317 machine will start
at the outlet of the tunnel and bore up to meet the DS325-318 machine, which will start at the
reservoir at the opposite end. Both machines were delivered in 2007 and the -317 began boring
in November of 2007. While the second TBM parts had been delivered, problems encountered in
obtaining access to land necessary for commencing excavation of the inlet portal delayed the
start of assembly of the -318 machine until June of 2009.
In early October, the assembled machine, located in the assembly portal, was flooded and
covered by 10 m of water. Currently efforts are underway to repair the damage so that boring
can commence (see Figure 5).

^
Figure 5. AMR II Inlet portal before flood (left) and after flood (right).
3.3.2 Necessary Mechanical Changes
The delay in the ability to gain site access to the portal provided the opportunity for required
modifications to be made prior to machine assembly, and new systems to be added during the
assembly process. The most notable change to the machine was the addition of foam nozzles to
the cutter head. Water spray nozzles cannot be used with the foam, as they damage the foam
and their locations cannot be modified for use with foam, as the water alone is used as a bench
mark. Engineers decided that four new locations would be provided on the cutterhead. The
modification required 200 mm holes through 100 mm of structural plate and hard plating.
Mounting plates were then welded into place for the nozzle assemblies. Further modifications to
the structure were required to allow for the routing of additional plumbing and the addition of a
manifold. The new design provides dedicated passages for each foam location and the existing
water spray system through a new five passable rotary union.
3.3.3 Necessary Foam Installation
Similar to the Guadarrama foam installation, also here a manually controlled foam system will be
used. Nevertheless, a couple of new features are installed to increase the effectiveness and user
friendliness:
- data measurement: the foam system is equipped with magnetic flow meters for water, anti-wear
agent and compressed air. This ensures correct data logging without any need of calibration.
- Data display: in order to ensure an easy survey, the flow of water, anti-wear-agent and
compressed air is displayed at the dosing unit. The specific flow values for each foam gun are
displayed at the generator itself.
- Remote control: the foam system can be switched ON and OFF via remote control from the
drivers cabin. If switched on again, the foam system climbs automatically back to the latest
installed output quantities. Furthermore, the remote control indicates the function (or nonfunction) of all main dosing components in order to detect defects as early as possible.

- Foam guns: in order to increase the foam quality of the anti-wear agents, a special foam core
design has been developed. The foam quality has got a high influence on the efficiency of the
anti-wear agents and their lifetime. This is especially difficult for low expansion ratios.
3.3.4 Additional Data Recording
The -317 machine was provided with a data collection system that captures machine data. This
includes date and time of day, cutterhead rotation speed, cutterhead power, start / stop time (i.e.
propel pressure greater than X), boring stroke position, penetration rate, thrust pressure and
gripper pressure, most of which is applicable in evaluating cutter performance. In addition to
existing monitored parameters, additional sensors were added for the testing. To evaluate
changes in water use, analog flow meters were provided with the foam generation unit, as well as
flow meters to monitor compressed air usage. Additional flow meters were added to the TBM
water system to monitor water flow to the cutterhead spray system. The flow meters used in
combination with the added dust monitor, are then used to evaluate the effectiveness of the foam
product in reducing airbone dust and possibly reducing water usage at the face. The final set of
sensors added to the machine is the Robbins Cutter Instrumentation System, which supplies real
time vibration data, cutter rpm, and cutter temperature. From this it is possible to infer the rock
face condition and how it is affecting cutter operation, as well as the state of cutter wear, without
entering the cutterhead to inspect the cutters. The cutters will also be closely measured manually
during excavation.
4.

Summary and Outlook

Laboratory research as well as site data illustrate the possibility to reduce the three main
problems on hard rock TBMs: abrasion, temperature and dust.
This can be realized already today by traditional measurements like water sprinkling, metallurgic
improvements and exhausting but there is a chance of significant improvements by using
modern foams and polymers.
In order to prove the promising laboratory data, further on-site evaluations are necessary and will
be given in the near future by detailed monitoring of the above described AMR II project in India.
5.

References

[1] Bruland, A. [1998], Hard Rock Tunnel Boring, The boring process, Trondheim
[2] Fred N. Kissell [2003], Handbook for Dust Control in Mining IC9465, U.S. Department of
Health and Human Services, www.cdc.gov/niosh
[3] Cheng L, Zukovich PP [1973]. Respirable dust adhering to run-of-face bituminous coals.
Pittsburgh: U.S. Department of the Interior, Bureau of Mines, RI 7765. NTIS No. PB 221883.
[4] Jankowski RA, Organiscak JA [1983]. Dust sources and controls on the six U.S. longwall
faces having the most difficulty complying with dust standards. Pittsburgh, PA: U.S.
Department of the Interior, IC 8957. NTIS No. PB84-142058
[5] Zimmer RA, Lueck SR, Page SJ [1987]. Optimization of overburden drill dust control systems
on surface coal mines. Int J Surf Min 1:155-157.
[6] Ruggieri SK, Jankowski RA [1983]. Fundamental approaches to longwall dust control. In:
Proceedings of the Symposium on Control of Respirable Dust (Beckley, WV, Oct. 4-8, 1983).
[7] Seibel RJ [1976]. Dust control at a transfer point using foam and water sprays. Pittsburgh, PA:
U.S. Department of the Interior, Bureau of Mines, TPR 97. NTIS No. PB-255-440
[8] Mu Mukherjee SK, Singh MM [1984]. New techniques for spraying dust. Coal Age June:54-56.
[9] Page SJ, Volkwein JC [1986]. Foams for dust control. Eng Min J 187(10):50-52, 54.
[10] Rostami, J., Ozdemir, L. & Nilsen, B. [1999], Comparision between CSM and NTH Hard
Rock TBM Performance Prediction Models.
[11] Bruland, A. [1998], Hard Rock Tunnel Boring, Advance Rate and Wear, Trondheim
[12] Fullprof software [2009]. Trial and full version available at http://www.anleggsdata.no

3DSHU

,QIOXHQFHRIFRUURVLRQRQDEUDVLRQRIFXWWHUVWHHOVXVHGLQ7%0
WXQQHOOLQJ

(VSDOODUJDV1XULD
-DNREVHQ3nO'UHYODQG
0DFLDV)UDQFLVFR-DYLHU
/DUV/DQJPDDFN


Under Review LQ
5RFN0HFKDQLFVDQG5RFN(QJLQHHULQJ

5RFN0HFKDQLFVDQG5RFN(QJLQHHULQJ


,QIOXHQFHRIFRUURVLRQRQDEUDVLRQRIFXWWHUVWHHOVXVHGLQ7%0WXQQHOOLQJ
0DQXVFULSW'UDIW


0DQXVFULSW1XPEHU
)XOO7LWOH

,QIOXHQFHRIFRUURVLRQRQDEUDVLRQRIFXWWHUVWHHOVXVHGLQ7%0WXQQHOOLQJ

$UWLFOH7\SH

2ULJLQDO3DSHU

.H\ZRUGV

7%0WXQQHOERULQJDEUDVLRQFXWWHUVWHHOFRUURVLRQWULERFRUURVLRQ

&RUUHVSRQGLQJ$XWKRU

1XULD(VSDOODUJDV
7URQGKHLP125:$<

&RUUHVSRQGLQJ$XWKRU6HFRQGDU\
,QIRUPDWLRQ
&RUUHVSRQGLQJ$XWKRU
V,QVWLWXWLRQ
&RUUHVSRQGLQJ$XWKRU
V6HFRQGDU\
,QVWLWXWLRQ
)LUVW$XWKRU

1XULD(VSDOODUJDV

)LUVW$XWKRU6HFRQGDU\,QIRUPDWLRQ
2UGHURI$XWKRUV

1XULD(VSDOODUJDV
3O'UHYODQG-DNREVHQ
/DUV/DQJPDDFN
)UDQFLVFR-DYLHU0DFLDV

2UGHURI$XWKRUV6HFRQGDU\,QIRUPDWLRQ
$EVWUDFW

$EUDVLRQRQ7%0FXWWHUVFDQEHFULWLFDOLQWHUPVRIWLPHDQGFRVW6HYHUDOUHVHDUFKHUV
DUHVWXG\LQJGHJUDGDWLRQRIFXWWHUWRROVIRU7XQQHO%RULQJ0DFKLQHV7%0H[FDYDWLQJ
KDUGURFNVRIWJURXQGDQGORRVHVRLO6RIDUWKHUHVHDUFKFRPSULVHVPDLQO\
GHJUDGDWLRQGXHWRDEUDVLYHZHDU$EUDVLYHZHDULVDKLJKO\SUHYDLOLQJSURFHVVLQ7%0
H[FDYDWLRQEXWNHHSLQJLQPLQGWKDWWKHHQYLURQPHQWWKHH[FDYDWLQJWRROVDUHZRUNLQJ
LQFRUURVLRQPLJKWKDYHDQLPSDFW7KURXJKWKLVSDSHULWLVSUHVHQWHGVRPHPHWKRGV
WKDWFDQEHXVHGWRHYDOXDWHWKHLQIOXHQFHRIFRUURVLRQRQDEUDVLRQRQ7%0H[FDYDWLRQ
WRROV7KHSDSHUDOVRSUHVHQWVWKHLQIOXHQFHRIFRUURVLRQRQDEUDVLYHZHDUIRUVRPH
LQLWLDOWHVWVZKHUHWKHVWHHODQGJHRPDWHULDOKDYHEHHQNHSWFRQVWDQWDQGZLWKYDU\LQJ
SURSHUWLHVRIWKHH[FDYDWLRQIOXLGVVRLODQWLDEUDVLRQDGGLWLYHVDQGZDWHU7KHUHVXOWV
VKRZWKDWWKHFRQWHQWRIFKORULGHLQWKHZDWHUJUHDWO\LQIOXHQFHVWKHDPRXQWRIZHDU
SURYLQJWKHLQIOXHQFHRIFRUURVLRQLQWKHDEUDVLRQRIWKHFXWWLQJWRROV7KHDGGLWLRQRI
FRQGLWLRQLQJDGGLWLYHVWDLORUPDGHIRUURFNRUVRLOFRQGLWLRQVUHGXFHWKHDPRXQWRIZHDU
KRZHYHUZKHQFKORULGHLVSUHVHQWLQWKHZDWHUWKHDGGLWLYHVPLQLPL]HWKHZHDUUDWHV
EXWWKH\GRQRWVXSUHVVWKHFRUURVLRQSURFHVVLQWKHFXWWLQJWRROV

Powered by Editorial Manager and ProduXion Manager from Aries Systems Corporation

 0DQXVFULSW
&OLFNKHUHWRGRZQORDG0DQXVFULSWWULERFRUURVLRQSDSHU7%0BILQDOGRF[

Influence of corrosion on abrasion of cutter steels used in TBM tunnelling

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

  

1

NTNU, Faculty of Engineering Science and Technology, Dept. Engineering Design and Materials. Tribology Lab, N7491 Trondheim, Norway.
2 NTNU, Faculty of Engineering Science and Technology, Dept. of Civil Engineering and Transport, N-7491
Trondheim, Norway.
3 BASF Construction Chemicals Europe Ltd., Zurich, Switzerland.


*


Abstract

               


              
             

             

           
              


  


Keywords:

1. Introduction
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

            
        
             
 
             

              
                 
       
      
               
             
 

            
            
              



1.1. Concept definitions for tunnel boring: tribology and tribocorrosion

            
           

             

 


 
    
        Figure 1   abrasive wear,
            


     

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

 
         
 


           


             
          
             
              
                
tribocorrosion.
 
         

           

              


1.2. Why tribocorrosion in TBM applications
             
            


             Figure 1   
            
          


          


            
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65






            

 



               
               

 





            
             
            
          
              









Figure 1.



2. Experimental set-up and materials

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

                   






2.1. Tribocorrosion tests applicable for geological materials
            
           
    
        Figure 2     




Metal sample

Normal load

Rubber
wheel

Reciproca ng
movement

Weight

Chemical
environmentt

Steel ball

Chemical environment
with soil

Rock




Figure 2.


2.1.1. Reciprocating ball-on-plate
                  

Figure
2a 
                 


1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

    
               

 


2.1.2. Rubber Wheel

           Figure 2b   
               
       
               





2.2. Tests with field materials in field chemical environment
 

        
     

               
             
Table 1


            
                
            
              
               


             
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

            

            
  

                 


                


Table 1.

Material

Steel

Soil

Rock

Water 1

Water 2

Nomenclature













Conditioning
additives

 







































Abrasiveness
(AVS/SAT)
Hardness
Composition
pH
Tribocorrosio
n (sliding)
Abrasioncorrosion
(rubber wheel)


2.3. Tests in laboratory controlled conditions
            
 
             


          
 
              
                


1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65



 

          


Table 2


Table 2.

Material

Nomenclature
Abrasiveness
(AVS)
Hardness
Composition
before testing
Composition
after testing
pH
Viscosity
Tribocorrosion
(sliding)

Steel

Rock

Distilled
water

3.4 wt.% Conditioning Conditioning

NaCl
additives
additives +
3.4% NaCl







 

 



















































2.4. Materials characterization tests and chemical analysis
           
             


2.4.1. Steel
             
         
 
            


        
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

             


2.4.2. Soil/rock
    
             
            
            


2.4.3. Liquid media
              
             
         
 




    



          
 
              


3. Test results and discussion
3.1. Steel and geological samples characterization
                
 
 


Figure 3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

            
           




 



                  
          
              


              
     



Figure 3.

Table 3
              
              
 


Table 3. 


Plagioclase

K-feldspar

Clino-pyroxene

Magnetite

Calcite

Actinolite

Muscovite

Albite

Magnesia

Microline

Mica

VHNR

Feldspar

Abrasive
reference
soil

Scandinavian
jobsite

Iddefjord
granite

Soil/rock
name

Quartz

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

3.2. Chemical analysis and pH of the liquid media


                 

                
              
               

 
 




    

                 




1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65







       Figure 4       
Figure 4a

               

 Figure 4b
             

 





1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65



Figure 4.







 
            
               

              
 





 




1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

               
     
   
            


3.3. Viscosity of the conditioning additives and their mixtures
 
             
              

Table 4, 
             

                


Table 4.               

ABR5
Dynamic
viscosity
Pa s [10-3]

SLF41

3 vol.% ABR5
in distilled
water

3 vol.% ABR5
in 3.4 wt.%
NaCl

3 vol.% SLF41
in distilled
water

3 vol.% SLF41
in 3.4 wt.%
NaCl


3.4. Abrasive wear of the steel in dry and wet conditions
3.4.1. Abrasion ranking in dry conditions (AVS and SAT tests)
 
 
               
              
Figure 5Figure 5 
           



45
40
35
Abrasion value [mg]

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

30
25
20
15
10
5
0
Scandinavian Jobsite

Iddefjord granite

Abrasive reference soil

Figure 5.

Table 3

                 





3.4.2. Abrasion in wet conditions: Rubber Wheel
             

                 

   
Figure 6 


Figure 6a

Figure 6b
6c




1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

                










1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65



Figure 6.
                




Figure 7              
  

Figure 7b
Figure 6

            
  
             
               
      Table 4       
     Figures 7c  7f       

  Figures 7e 7f        
              
 


 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65



















Figure 7.              
 
 




3.4.3. Sliding tribocorrosion: Reciprocating ball-on-plate

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

Figure 8             
               
              
  

Figure 9a, 9b, 9d 9e


Figure 9f

             
Figure 8a 8c

                 
               

           
Table
4

 
 

              
 
Figure 9


1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65










Figure 8.              





1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65



























Figure 9.




 
 





3.1.1. Tribocorrosion of cutter disc steel used in TBM

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65


                
              





              
               
        Table 5    
            
Figure 7

               
              



Table 5.

Reference soil
SC water
Reference soil
ME water
SC water & SLF41
SC water & ABR5
ME water & SLF41
ME water & ABR5


Fe [P
Pg/mL]

Cr [P
Pg/mL]

Ni [P
Pg/mL]

Cu [P
Pg/mL]






























Table 6.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

100% SC water
3% ABR5
SC water
100% distilled
water
3.4 wt.% NaCl
3% ABR5
distilled water
3% SLF41
distilled water
3% ABR5
3.4 wt.% NaCl
3% SLF41
3.4 wt.% NaCl


Fe [P
Pg/mL]


Cr [P
Pg/mL]


Ni [P
Pg/mL]


Cu [P
Pg/mL]


4. Conclusive remarks
 
          
 

x

              



            



            


 
          


5. Acknowledgements
 
          
  
 


      
      
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65



6. References




  



 
         


          
 
   
        





  


            




         



           
  

           
         

   

 


   
        

 

 









 



 






 

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

 










             
 
         
   
 


Paper 12
TBM Cutter Steel a challenge for Norwegian steel suppliers
Grv, Eivind
Jakobsen, Pl Drevland
Kane, Alexandre-Pascal
Hoang, Hiue
Smading, Steve
Sagen, Trine Bye
Published in the Tunnelling Journal 2013