PUMICE AGGREGATES FOR STRUCTURAL LIGHTWEIGHT AND INTERNALLY

CURED CONCRETES

Samuel Green1, Nicholas Brooke2 and Len McSaveney3

ABSTRACT

This paper presents research on the utilisation of New Zealand’s abundant resources of pumice aggregates
for producing structural lightweight and internally cured concretes. Mixture designs were developed for
lightweight concrete containing both partially saturated and fully saturated pumice aggregates. A vacuum
saturation system was developed to completely saturate the normally partially saturated aggregates and
consequently avoid a loss of workability attributed to the high absorption capacity of pumice. Shear strength
and bond strength of pumice concrete was investigated, as well as the application of saturated fine pumice
aggregates as an internal curing media for both high performance concrete and ready-mixed full scale thin
concrete slabs with a low water-cementitious materials ratio. The research presented demonstrates the
potential to commercially produce structural lightweight concrete containing naturally occurring pumice
aggregates, while meeting the requirements of New Zealand Concrete Design Standards.

_______________________
1
S.G. Structural Engineer, Arup Pty Ltd, Sydney, Australia. samuel.green@arup.com.au
2
N.B. Former Lecturer, Department of Civil and Environmental Engineering, University of
Auckland, New Zealand.
3
L.M. Market Development Manager, Golden Bay Cement, Auckland, New Zealand.
 INTRODUCTION PUMICE AGGREGATES
New Zealand currently has abundant resources of The name “Pumice” is a generic term for a range
pumice aggregates that are underutilised in the of porous vesicular materials produced during
construction industry. Products containing pumice explosive volcanic eruptions. Pumice is essentially
aggregates do exist, such as lightweight masonry composed of solidified frothy lava which is
blocks and non-structural precast elements, but generally rhyolitic in composition, but can also be
there has been no recent research in New Zealand produced in a less acidic form. Aggregate strength
which focuses on the production of structural grade ranges from very weak and porous, to stronger
(25+ MPa) lightweight pumice aggregate concrete and less porous. Absorption is generally high, with
(LPAC). the specific characteristics being largely dependent
on the porosity and size of the aggregates. The
In the early 1970’s, much effort went into principal requirements for pumice to be considered
researching LPAC in New Zealand due to the large a desirable aggregate for use in structural
quantity of pumice resources spread throughout lightweight concrete are a low density and
the central North Island, and a lack of alternative relatively high strength. The density is determined
artificially produced lightweight aggregates [1]. A by the voids content, which is a function of the type
paper published in New Zealand Concrete of deposit from where the pumice is found [4, 5].
Construction [2] highlighted issues of volume
stability in pumice concrete products. Shrinkage Figure 1 shows a cross section obtained using X-
cracking of wall panels was a common problem, ray CT scanning of a typical fine pumice
which subsequently led to a decline in consumer aggregate. The porosity of the pumice is clearly
confidence. The research appeared to cease and illustrated by the relatively large internal and
ever since pumice aggregates have primarily only external voids.
been utilised for non-structural purposes and for
other non-concrete related applications [1].

The first use of structural lightweight concrete in

New Zealand was in the construction of the
Wellington “Westpac Stadium”. Built in 2000, the
Stadium was constructed out of concrete
containing artificially expanded shale aggregate
imported from California. The project clearly
demonstrated the commercial feasibility and
benefits of utilising lightweight concrete in
New Zealand construction [3]. The need for
increased research and development into Figure 1. Cross-section of pumice aggregate
lightweight concrete, in particular the utilisation of (Courtesy of Dr. N. Kikkawa)
New Zealand’s abundant resources of pumice
aggregates, has been highlighted by past
researchers [1]. The opportunity to produce an PUMICE AGGREGATES UTILISED IN THIS
economic, structural lightweight concrete for the STUDY
New Zealand market which contains naturally Pumice aggregates were obtained from the
occurring pumice aggregates undoubtedly exists. Winstone Aggregates Te Puke Quarry, otherwise
known as “Cameron’s Quarry”. Located in the
Following the purchase of a large pumice resource Central North Island, Cameron’s Quarry is the
by Winstone Aggregates, a partnership was largest licensed pumice resource in New Zealand,
formed between Winstone Aggregates, Golden with estimated reserves in excess of 150 million
Bay Cement, Humes Pipeline Systems and The tonnes [6]. Typical properties of the various pumice
University of Auckland to undertake an aggregates and normal weight Helensville beach
investigation into the production of structural grade sand used throughout this study are given in
LPAC. This paper presents the results of this Table 1. These properties were either provided by
investigation, including the development of LPAC Winstone Aggregates or found experimentally
mixture designs, the development of a vacuum according to NZS 3111 Methods of Test for Water
saturation system required to fully saturate pumice and Aggregate for Concrete [7]. Table 2 presents
aggregates to a wet moisture condition, and the the results of chemical analysis using X-ray
results of both bond and shear strength testing. fluorescence (XRF) on a range of pumice
Additionally, this paper discusses a study into the aggregates obtained from Cameron’s Quarry.
utilisation of saturated pumice as an internal curing
media for high performance concretes and thin
concrete sections, which are both typically prone to
significant shrinkage problems.
 Table 1. Pumice aggregate properties Standard Model Code for Concrete Construction
[12] and Eurocode ENV 1992-1-4 [13] impose
SSD OD 3
Aggregate Size Absorption upper limits of 2000 and 2050 kg/m respectively.
Density Density
Range 3 3 (%)
(kg/m ) (kg/m ) Mixture designs were investigated utilising pumice
Helensville
2660 2640 0.58 aggregates in the air-dry moisture condition (as
Sand
0 - 4.75 mm provided by Winstone Aggregates), with maximum
1550 1290 16.6 aggregate sizes of 16 mm and 9.5 mm adopted for
Pumice
4.75 – 9.5 mm the OS-LPAC and HP-LPAC mixtures respectively.
1380 630 54.3 Normal weight Helensville beach sand was used
Pumice
9.5 – 16 mm
1290 560 56.3
as fines for all the OS-LPAC mixtures, while sand
Pumice in combination with a small fraction of 0 – 4.75 mm
pumice was used for the HP-LPAC mixtures.
Table 2. Chemical analysis of pumice aggregates
All cement utilised in this study was Golden Bay
0 - 4.75 mm 4.75 - 9.5 mm 9.5 - 16 mm Cement Ordinary Type GP Portland cement. The
Element
Pumice Pumice Pumice HP-LPAC mixtures contained “Microsilica 600”
SiO2 65.69 67.2 66.63 (MS 600), a fine naturally occurring amorphous
Al2O3 15 14.7 14.94
Fe2O3 3.55 3.45 3.37
silica and a highly reactive pozzolan, and Grace
CaO 4.1 4.14 4.38 ADVA 142, a standard high range water-reducing
MgO 0.74 0.74 0.71 superplasticising admixture.
SO3 0.06 0.06 0.06
Na2O 3.29 3.54 3.57 The preliminary LPAC mixture designs
K2O 1.87 1.99 1.96 investigated in this research are shown in Table 3
Mn2O3 0.12 0.12 0.13 and Table 4 for the OS-LPAC and HP-LPAC
Loss on Ignition 5.59 4.06 4.26 mixtures respectively. Table 3 and Table 4 also
include the water-cementitious materials (w/cm)
DEVELOPMENT OF PRELIMINARY PUMICE ratios and the resulting 28-day f’c and unit weight
CONCRETE MIXTURE DESIGNS results. All testing was carried out according to the
A preliminary investigation based on previously relevant sections of NZS 3112 Methods of Test for
published international research [8, 9] was Concrete [14]. Unit weight is presented as the
conducted to develop an understanding of the equilibrium unit weight, which is defined by
behaviour of LPAC using New Zealand materials. ASTM C567 as the oven-dry unit weight plus
Results from this investigation provided a 50 kg/m3. Equilibrium unit weight is assumed to be
benchmark for further refinement of the mixtures obtained after ninety days of air-drying [10].
discussed later in this paper.
Table 3. OS-LPAC mixtures
Two classes of LPAC were defined for the purpose
of this research as follows: Mix Reference OS-M1 OS-M2 OS-M3
3
Ordinary Strength Lightweight Pumice Cement (kg/m ) 544 635 469
3
Aggregate Concrete (OS-LPAC) - having a unit Water (kg/m ) 201 97 211
weight not exceeding 2000 kg/m3 and 28-day Sand (kg/m 3) 575 651 695
3
compressive strength (f’c) of not less than Fine Pumice (kg/m ) - - -
3
17.5 MPa, nor greater than 35 MPa. Coarse Pumice (kg/m ) 496 558 455
w/cm 0.37 0.15 0.45
High Performance Lightweight Pumice
28-day f’c (MPa) 24.5 30.5 21.0
Aggregate Concrete (HP-LPAC) - having a unit Equilibrium unit weight
3
weight not exceeding 2000 kg/m and 28-day f’c 3 1610 1690 1610
(kg/m )
greater than 35 MPa.
Table 4. HP-LPAC mixtures
The lower limit of 35 MPa for HP-LPAC was
chosen for this research following the ASTM Mix Reference HP-M1 HP-M2 HP-M3
International Standard ASTM C567 [10] definition Cement (kg/m3) 539 573 576
for high-strength lightweight concrete, as a Water (kg/m3) 155 140 152
concrete with a 28-day f’c greater than 34.5 MPa. Sand (kg/m 3) 934 899 855
3
Although the upper limit on unit weight of Fine Pumice (kg/m ) 42 45 46
3
2000 kg/m3 defined above for both OS-LPAC and Coarse Pumice (kg/m ) 259 275 280
3
HP-LPAC is inconsistent with the New Zealand MS 600 (kg/m ) 82 87 87
3
ADVA 142 (kg/m ) 12.6 13.4 13.4
Concrete Structures Standard NZS 3101 [11]
3 w/cm 0.25 0.21 0.23
upper limit of 1850 kg/m for structural lightweight 28-day f’c (MPa) 37.5 42.5 41.0
concrete, this value has not been chosen Equilibrium unit weight
arbitrarily. Other international design standards 3 1910 1960 1890
(kg/m )
impose similar limits; for example the International
Following recommendations given by NZS 3152 stiffening is the aforementioned ability of pumice to
Specification for the Manufacture and Use of absorb relatively large quantities of water.
Structural and Insulating Lightweight Concrete
[15], and a mixing procedure used in research VACUUM SATURATION OF PUMICE
conducted by Hossain and Lachemi [8], the mixing AGGREGATES
sequence consisted of 3 minutes of mixing both Following the preliminary mixture design
the fine and coarse aggregates with 50 % of the development, the need for an effective method of
mix water, followed by an additional 3 minutes of saturating pumice aggregates to the SSD or wet
mixing with the cement, admixtures (where used) moisture state was identified as a solution to the
and remaining water until a visually workable loss of workability that occurred, and to aid in the
mixture was achieved. The initial mixing of the accurate determination of the actual free moisture
water with the aggregates was carried out in an content. Neville [16] and the American Concrete
attempt to saturate any partially filled external Institute Standard ACI 213R-03 Guide for
voids of the air-dry pumice aggregates, and to Structural Lightweight Aggregate Concrete [18]
reduce a potential loss of cement into the outer both discuss the technique of utilising vacuum
vesicles on the surface of the pumice aggregates. saturation to efficiently saturate lightweight
Total water added to the mixtures was closely aggregates and overcome the unfavourable ability
monitored to enable the calculation of water of lightweight aggregates to absorb mix water
absorbed by the pumice aggregates, and thus the during batching. Lura et al. [19] discusses the use
remaining free mix water. of vacuum saturation in the context of pumice
aggregates as they noted the inefficiency of
The results presented in Table 3 for the OS-LPAC saturating pumice by the conventional practice of
trials demonstrate that mixtures containing pumice submersion in water.
as coarse aggregates with a maximum size of
16 mm in combination with normal weight sand as A vacuum saturation procedure was subsequently
fines can produce concrete with compressive developed following a study by Williamson [20].
strength in excess of 30 MPa. Equilibrium unit The procedure adopted can be broken down into
weight was than the NZS 3101 specified upper the following five phases:
3
limit of 1850 kg/m .
Phase 1: Place aggregates in a sealed
The results presented in Table 4 for the HP-LPAC vacuum chamber
trials demonstrate that the addition of microsilica Phase 2: Apply a vacuum for a sufficient
and ADVA 142 into the mixtures developed, as period of time to evacuate all air
well as an increase in cement content and from within the chamber and
reduction in water and total pumice content results external pores of the aggregates
in compressive strengths in excess of 40 MPa. Phase 3: Seal chamber under vacuum
Equilibrium unit weight values were within the Phase 4: Release water into chamber
maximum limit of 2000 kg/m3 defined for HP- Phase 5: Immerse aggregates in water for
LPAC. 24 ± 1 hours.

Due to the aggregates being in an air-dry moisture Phase 2 (vacuum) and Phase 4 (water saturation)
state during batching, and an inherent difficulty in are illustrated graphically in Figure 2 below.
obtaining pumice aggregates in the saturated
surface-dry (SSD) or wet moisture states, Vacuum air Saturate with
calculating precise mixture proportions by out water
considering the free available moisture and
absorption of the aggregates proved to be
complicated. Free moisture content was
particularly difficult to determine as the pumice Aggregates Aggregates
appeared to be absorbing mix water during
batching. Neville [16] and Mindess et al. [17] both
discuss the great difficulty in determining the free
mix water quantities for most lightweight aggregate
concretes, with only approximate w/cm ratios
Figure 2. Phase 2 and Phase 4 of vacuum
typically given.
saturation procedure
Although not directly measured, a loss of By evacuating the air from the external voids of the
workability, which resulted in stiffening of the pumice aggregates, the water released into the
mixtures, was observed to occur within chamber should theoretically fill these empty pores
approximately 20 minutes of batching for all the and fully saturate the aggregates to a wet moisture
mixtures investigated. The most likely cause of the state [19].
A vacuum saturation system was subsequently the effectiveness of these aggregates for
developed to trial the concept on pumice producing LPAC and internally cured concretes.
aggregates, and then utilise these aggregates for As an aside, further investigations into designing a
producing LPAC and internally cured concretes vacuum saturation system capable of producing
discussed later in this paper. The system saturated pumice aggregates on a commercial
3
developed was capable of processing 0.16 m of scale are currently underway. Additionally, the
vacuum saturated pumice aggregates in potential to economically stockpile the aggregates
approximately 10 minutes. while avoiding significant surface moisture loss is
also under consideration.
The effectiveness of the vacuum saturation system
was determined by comparing the moisture VACUUM SATURATED PUMICE CONCRETE
content of the pumice aggregates, obtained in MIXTURES
accordance with Section 7 of NZS 3111 Methods Following the investigation into the preliminary
of Test for Water and Aggregate for Concrete [7], pumice concrete mixture designs and the
pre and post vacuum saturation. Aggregate size development of the vacuum saturation system, a
ranges representing the full scope of available further investigation was conducted utilising the
aggregates applicable to this research were vacuum saturated pumice aggregates to produce
utilised for verification of the system. These being: HP-LPAC. Updated properties for the vacuum
0 – 2.36 mm, 0 – 4.75 mm, 4.75 – 9.5 mm and saturated pumice aggregate are given in Table 5.
9.5 – 16 mm. The aggregates were vacuum
saturated from an initial air-dry moisture condition Table 5. Vacuum saturated pumice aggregate
as supplied by Winstone Aggregates. properties

Figure 3 presents the results of the pre and post SSD OD

Aggregate Size Absorption
vacuum saturation moisture contents. Aggregate Density Density
Range 3 3 (%)
moisture content is significantly increased in all (kg/m ) (kg/m )
instances through the use of vacuum saturation. 0 - 4.75 mm 1610 1090 32.4
Pumice
4.75 – 9.5 mm 1361 611 55.1
160 Pumice
Aggregate Moisture Content

140 Pre-vacuum saturation

Post-vacuum saturation Table 6 shows the mixture proportions, including
120
(% dry mass)

100
the w/cm ratio, for the vacuum saturated HP-LPAC
mixtures investigated. 28-day f’c and unit weight
80 results are also given.
60
40 Table 6. Vacuum saturated high performance
lightweight pumice aggregate concrete mixtures
20
0 Mix Reference HP-M1-VS HP-M2-VS
0 - 2.36 0 - 4.75 4.75 - 9.5 9.5 - 16.0 Cement (kg/m3) 518 599
Water (kg/m3) 65 71
3
Aggregate Size Range (mm) Sand (kg/m ) 958 1042
Fine Pumice (kg/m3) 70 55
Figure 3. Increase in moisture content of pumice Coarse Pumice (kg/m3) 370 276
3
aggregates through vacuum saturation MS 600 (kg/m ) 79 92
3
ADVA 142 (kg/m ) 8.7 10
The larger aggregates appeared to be absorbing a w/cm 0.11 0.10
relatively greater quantity of water when compared 28-day f’c (MPa) 39.5 46.0
to the smaller aggregates, which is most likely Equilibrium unit weight
3 1840 1860
(kg/m )
attributed to an increase in the size of the external
pores of the larger aggregates and therefore
The above results demonstrate that the application
absorption capacity [19].
of vacuum saturated pumice aggregates can
The results indicate the system developed is produce LPAC achieving a compressive strength
in excess of 40 MPa while maintaining a unit
effective in vacuum saturating pumice aggregates
weight within 1 % of the NZS 3101 specified upper
to a moisture condition representative of the wet 3
moisture state for practical purposes. The capacity limit of 1850 kg/m for lightweight structural
of the system, coupled with the reasonably short concrete. Furthermore, loss of workability and
complications with calculating mix water quantities
period of time required to process the aggregates
were not observed to occur for the vacuum
enabled the processing of sufficient quantities of
saturated pumice mixtures due to the aggregates
vacuum saturated pumice aggregates necessary
being in a wet moisture state during mixing.
to undertake further laboratory investigations into
 STRUCTURAL TESTING OF PUMICE Table 7 also compares the test results with the
CONCRETE bond requirements of NZS 3101 using values of
An experimental study was conducted to assess 1.0 for αa (less than 300 mm fresh concrete
the bond and shear performance of HP-LPAC by beneath the bars), αb (zero flexural reinforcement)
comparing the results of testing with the and αd (zero transverse reinforcement), and 1.5 for
requirements of NZS 3101. NZS 3101 contains αc (cover greater than 1.5db), for all test
various provisions for detailing and assessing the specimens.
shear strength and bond strength of structural
lightweight concrete elements. To accurately gauge the consistency of the results
across the range of different bar sizes and
Bond Strength of LPAC embedment lengths, the variation in compressive
strength f’c must be taken into consideration. As
Bond strength was assessed by performing tensile the average required bond stress found using
pullout tests on reinforcing bars embedded at Equation 2 is linearly proportional to the square
variable lengths into HP-LPAC, and calculating the root of the compressive strength, the average bond
resulting average bond stress developed along the stress can be normalised by dividing by √f’c. The
embedded length of the reinforcing bar. The normalised bond stress (ub/√f’c) is also presented
procedure adopted was similar to that utilised in in Table 7.
research conducted by Hossain [21], which also
investigated the bond strength of pumice concrete. The results from bond pullout testing indicate the
Deformed bars with nominal diameters of either bond strength of HP-LPAC is greater than the
12 mm or 16 mm were used, with both Grade 300 nominal required bond strength of lightweight
and Grade 500E bars tested. Further details on the concrete found using the provisions of NZS 3101
bond test specimen geometry and testing by an average factor of 2.37. The results were
procedure can be found in Green [22]. reasonably well spread, having a standard
deviation of 0.19 when the effects of differing
Table 7 contains data on the reinforcing bar compressive strengths were taken into
diameter (db), embedment length (le), specified consideration.
yield strength (fy) and 28-day compressive strength
(f'c) of the bond test specimens investigated. The Shear Strength of LPAC
mean average bond stress (ub) results as well as
details on the failure mode for the various Shear strength of LPAC was assessed by
configurations tested are also given. The average performing shear tests on beams constructed out
bond stress (ub) was calculated using Equation 1. of HP-LPAC. Equivalent strength normal weight
The steel stress (fs) was calculated as the concrete beams were also tested to provide control
maximum tensile force applied to the bar divided results. The shear beam test specimen details
by the nominal cross sectional area of the bar. chosen for this study were an adaptation of the
specimens used in research conducted by Li et al.
𝑑 [23]. The shear beam geometry and loading
𝑢𝑏 = 4𝑙𝑏 𝑓𝑠 (1)
𝑒 configurations are shown in Figure 4 and Figure 5
respectively.
The average bond stress relied upon by NZS 3101
for lightweight concrete (ubr) is given by Longitudinal D16 Grade 300 deformed bars were
Equation 2. included at the top and bottom of the beams to
prevent the possibility of flexural failure and to
𝛼𝑐 𝛼𝑑 𝑓𝑐′ ensure a shear mode of failure occurred. No shear
𝑢𝑏𝑟 = (2) reinforcement was provided.
𝛼 𝑎 𝛼 𝑏 2.66

Table 7. Bond strength test details and results

db le fy f'c fs ub ubr
Specimen ID ub/√f’c ub/ubr Failure Mode
(mm) (mm) (MPa) (MPa) (MPa) (MPa) (MPa)
B125-D12 12 125 300 35.5 332.8 7.99 3.36 1.34 2.38 Yield
B125-D16 16 125 300 35.5 243.7 7.80 3.36 1.31 2.32 Slip/Split
B75-D12H 12 75 500 37 264.7 10.59 3.43 1.74 3.09 Slip/Split
B75-D16H 16 75 500 37 159.6 8.51 3.43 1.40 2.48 Slip/Split
B175-D12H 12 175 500 36.5 466.1 7.99 3.41 1.32 2.35 Slip/Split
B175-D16H 16 175 500 36.5 310.3 7.09 3.41 1.17 2.08 Slip/Split
Mean 1.34 2.37
Standard Deviation 0.19 0.34
 D16 Flexural
Reinforcement
D16 Flexural
Reinforcement
(fct) and flexural tensile (Tf) strengths, as well as
38
unit weight values for the three mixtures are
summarised in Table 8.

Table 8. Shear test specimen material properties

210

134
38

Mix Mix Mix

50 100 50
SP1 SP2 SNC1
185 170 185
28-day f’c (MPa) 36.5 37.0 40.5
ELEVATION RH ELEVATION 28-day fct (MPa) 3.8 4.0 4.4
28-day Tf (MPa) 5.8 4.8 6.0
28-day SSD unit weight
2090 2050 2360
50
3
(kg/m )
3
OD unit weight (kg/m ) 1890 1840 2190
200

100

Equilibrium unit weight

1940 1890 2240
(kg/m3)
50

540

PLAN Average first crack (vcrack) and ultimate shear

strength (vultimate) results from shear testing are
given in Table 9, with results compared with the
Figure 4. Shear beam geometry (dimensions in NZS 3101 estimation of nominal shear strength
mm) calculated using both approaches (a) and (b) of
Displacement Clause 9.3.9.3.5. The average shear stress
Transducer
resisted by the beams was estimated as the shear
force at the beam centreline divided by the cross-
160 330 50
sectional area resisting the shear. When
calculating the NZS 3101 nominal shear strength,
values of 0.85 for ka (maximum aggregate size of
10 mm or less) and 1.0 for kd (effective depth less
than 400 mm) were adopted following the
provisions of NZS 3101 Clause 9.3.9.3.4. For
approach (b) of Clause 9.3.9.3.5, a reduction factor
of 0.85 was applied as the HP-LPAC mixtures
investigated are classed as "sand-lightweight"
concrete. Figure 6 illustrates the average shear
stress and crack width vs. time for a typical test
50 330 160
specimen, and is given to demonstrate the first
crack stress and ultimate failure stress.
Figure 5. Load configuration for shear tests
Table 9. Average shear strength test results
Tests were performed by applying gradually
increasing equal uniformly distributed loads onto NZS 3101
Specimen f'c vcrack vultimate
(MPa)
roller supports located on the top surface of the ID (MPa) (MPa) (MPa)
(a) (b)
shear beam specimens at the load positions SB-SP1 36.5 2.48 4.02 0.94 0.80
(arrows) shown in Figure 5. Roller supports were SB-SP2 37 2.72 4.57 0.95 0.81
also provided on the bottom surface of the SB-SNC1 40.5 2.76 3.89 0.99 0.99
specimens. A load cell was used to measure the
magnitude of forces applied during testing. Portal
displacement transducers were attached
perpendicular to the assumed crack path on each
side of the thinner central shear panel to measure
the propagation of shear cracks, and to indicate
the first crack load and ultimate load. Loading was
increased until a dramatic loss of strength occurred
in the beams. Further details on the testing
procedure can be found in Green [22].

Three different concrete mixtures were

investigated; SP1 (non-vacuum saturated HP-
LPAC), SP2 (vacuum saturated HP-LPAC) and
SNC1 (normal weight concrete control mixture).
Average 28-day compressive (f’c), splitting tensile
 5.0 5.0 media dispersed throughout the cement paste
Ultimate shear matrix. Under conventional curing conditions, the
failure
4.0 4.0 generally low permeability of HPC impedes the
Shear Stress (MPa)

Crack Width (mm)

ability of water to migrate through the concrete,
3.0 3.0 limiting the water available for cement hydration.
The addition of internal curing water therefore
First shear
greatly aids in the efficient hydration of cement
2.0 crack 2.0
[24].
1.0 1.0 Research by Bentz et al. [25] recommends using
Equation 3 for determining the mass of saturated
0.0 0.0 fine lightweight aggregates required to prevent
0 20 40 60 80 100 120 self desiccation in HPC. Further details can be
found in the literature.
Time (s)
𝐶𝑓 .𝐶𝑆.𝛼 𝑚𝑎𝑥
Shear Stress Average crack width 𝑀𝐿𝑊𝐴 = (3)
𝑆.∅𝐿𝑊𝐴

Figure 6. Comparison of average shear stress and where:

crack width MLWA = Mass of fine lightweight
aggregates required per unit
The results presented above indicate that the volume of concrete (kg/m )
3

shear strength of HP-LPAC is greater than the Cf = Cement content for concrete
nominal shear strength found using the provisions mixture (kg/m3)
of NZS 3101 for structural lightweight concrete. CS = Chemical shrinkage of cement
However, the shear test procedure adopted (g of water / g of cement)
αmax = Maximum expected degree of
appeared to over predict the shear strength of the
cement hydration (%)
concrete and therefore further research is S = Degree of saturation of
recommended to validate these findings. aggregates (0 = 0 % saturation,
1 = 100 % saturation)
APPLICATION OF PUMICE AGGREGATES FOR ϕLWA = Absorption capacity of lightweight
INTERNAL CURING aggregates (%)
Two further investigations were conducted to study
the effectiveness of 0 – 2.36 mm vacuum Based on the research by Bentz et al. [25], an
saturated pumice aggregates as an internal curing experimental investigation was carried out to
media in high performance concrete (HPC) and determine the effects of utilising saturated fine
thin concrete sections with a relatively low w/cm pumice aggregates as an internal curing media for
ratio and high surface area to volume ratio. HPC. A HPC mixture with a w/cm ratio of 0.29 was
Compressive strength, unit weight and shrinkage chosen, and the pumice aggregate requirements
properties of the internally cured concretes were were calculated using Equation 3. Properties of the
investigated. saturated fine pumice aggregates used in this
study are given in Table 10.
Internal Curing of High Performance Concrete
Table 10. Saturated fine pumice aggregate
The utilisation of HPC with a low w/cm ratio is properties
becoming ever more prevalent as advances in
concrete technology allow practical mixes to be SSD
Aggregate OD Density Absorption
developed. The combination of a low w/cm ratio Density
Size Range 3 (kg/m3) (%)
(kg/m )
coupled with the inclusion of silica fume and other
0 – 2.36 mm
admixtures alters both the internal pore structure of Pumice
1780 1330 25.0
the concrete and the chemistry of the hydration
reactions. A major issue influencing the The original, base HPC mixture contained normal
performance of HPC is early age cracking caused weight aggregates (both fine and coarse), ordinary
by autogenous shrinkage. Autogenous shrinkage Type GP Portland cement, fly ash, microsilica
is accompanied by self-desiccation during (MS 600) and a high-range water reducing
hardening of the concrete, which is characterised admixture. Three mixtures were studied: C1
by internal drying. An effective technique (control - 0 % calculated required pumice), P1
developed to help mitigate and overcome the (100 % calculated required pumice), and P2 (50 %
issues of autogenous shrinkage and self- calculated required pumice). The pumice was
desiccation is internal curing. Internal curing added to the mixtures by replacing an equal
utilises small internal reservoirs of water volume of normal weight fine aggregates,
introduced into the concrete by water bearing effectively maintaining a constant yield and
consistent mixture proportions. The resulting 600
effects on the concrete properties were assessed
by gathering data on shrinkage, compressive 500

Shrinkage (Microstrain)
strength and unit weight.
400
All mixture production was completed at the
Golden Bay Cement Portland Manufacturing Plant 300
Cement Laboratory (GBC lab) in Whangarei.
Compressive strength and unit weight 200
measurements were also completed at the GBC C1: 0 kg/m3 Pumice (Control)
Lab according to the relevant sections of 100 P1: 143 kg/m3 Pumice
NZS 3112 [14]. Early age shrinkage data was P2: 72 kg/m3 Pumice
obtained by sending 280 L x 75 W x 75 H (mm) 0
shrinkage beams to the Wellington BRANZ 0 2 4 6 8 10 12 14 16 18 20 22
laboratory for shrinkage testing. Drying Time (days)

Average 7 and 28-day compressive strength Figure 7. Internally cured HPC shrinkage
results, and 7-day unit weight results are given measurements
below in Table 11. These indicate that the
inclusion of saturated 0 – 2.36 mm pumice The effects of microsilica and fly ash on the values
aggregates within the mixtures trialled have a of shrinkage were not taken into consideration
detrimental effect on the compressive strength, when calculating the required saturated pumice
reducing the 7-day and 28-day strengths by up to contents. Previous research by Bentz et al. [25]
18.6 % and 17.3 % respectively, for the mixtures and Zhutovsky et al. [26] highlighted the significant
containing pumice, when compared to the control influence that silica fume (similar to microsilica)
mixture C1. has on the effectiveness of internal curing due to
dense packing of the silica fume particles around
Table 11. Internally cured HPC average the aggregates inhibiting moisture movement. It is
compressive strength and unit weight results therefore recommended that the influence of
microsilica on the effectiveness of pumice
C1 P1 P2 aggregates for internal curing of HPC be further
7-day f’c (MPa) 56.5 46.0 48.0 investigated.
28-day f’c (MPa) 81.0 67.0 67.0
3
7-day SSD unit weight (kg/m ) 2340 2270 2300 Internal Curing of Ready-Mixed Full Scale Thin
3
7-day OD unit weight (kg/m ) 2260 2200 2170 Concrete Slabs
7-day equilibrium unit weight
2310 2250 2220
(kg/m3)
Previous research and field experience has
demonstrated that elements having high surface
This unexpected reduction in compressive strength
area to volume ratios and relatively low w/cm ratios
could be due to the pumice aggregates being in a
can be particularly vulnerable to stresses
wet moisture state during mixing, as Zhutovsky et
developed as a result of changes in volume at
al. [26] suggested the SSD moisture state is more
early ages [27]. Based on this premise, three thin
preferable than the wet moisture state for batching
concrete slabs were constructed at the Humes
internally cured mixtures. Additionally, the specific
Pipelines Papakura Precast Manufacturing Plant to
aggregate size chosen and distribution of the
study the effects of saturated fine pumice
aggregates within the cement paste matrix effects
aggregates as an internal curing media on the
the resulting compressive strength of the mixtures.
compressive strength (both in-situ core tests and
Shrinkage results, as shown in Figure 7, indicate
laboratory based cylinder tests) and the early age
the inclusion of saturated 0 – 2.36 mm pumice
and long-term in-situ shrinkage of the concrete.
aggregates does not appear to significantly
influence the shrinkage of the concrete at early
The base mixture chosen for this study had a w/cm
ages (up to 7-days). An increase in shrinkage for
ratio of 0.395, and was originally designed and
the internally cured mixtures was observed for
utilised for pavement construction at the Auckland
drying times up to 21 days.
International Airport. An alternative approach to the
one discussed in the previous section was used for
calculating the required pumice content, and was
based on research by Jensen and Hansen [28].
Jensen and Hansen [28] developed a set of
theoretical equations for calculating the required
amount of water entrained in an internal curing
media to enable full hydration of cement and to
mitigate self-desiccation of concrete. These are
presented here as Equation 4 and Equation 5. Table 12. Average cylinder and in-situ core sample
Equation 4 is applicable for a w/cm ≤ 0.36 and compressive strength results for internally cured
Equation 5 is applicable for 0.36 ≤ w/cm ≤ 0.42. slabs

(w/cm)e = 0.18(w/cm) (4) Slab A Slab B Slab C

(w/cm)e = 0.42 - (w/cm) (5) Lab Core Lab Core Lab Core
7-day f’c (MPa) 41.5 37.5 40.0 31.0 42.5 37.0
where:
(w/cm)e = Additional water entrained in Preliminary shrinkage data (strain) is given below
an internal curing media to
in Figure 8. These limited data indicate that the
theoretically obtain complete
hydration of cement addition of saturated pumice into the mixtures
slightly increases the early age shrinkage
Three slabs were investigated (Slab A, B and C) encountered in the slabs.
having pumice contents as follows: Slab A 0.250
contained 100 % of the calculated required
pumice, Slab B contained 50 % of the calculated 0.200
required pumice, and Slab C (control mixture)
contained 0 % of the calculated required pumice.

Strain (%)
0.150
To maintain a constant yield and consistent
mixture proportions, the pumice was added to the
mixtures by replacing an equal volume of normal 0.100
weight fine aggregates. Table 10 shows the
properties of the saturated fine pumice aggregates 0.050 Slab A: 42.4 kg/m3 Pumice
Slab B: 21.2 kg/m3 Pumice
used in this investigation.
Slab C: 0 kg/m3 Pumice (Control)
0.000
All batching was completed at the Firth East 0 2 4 6 8 10 12 14 16
Tamaki ready-mix plant in Auckland. The batching Drying Time (days)
sequence consisted of addition of all aggregates to
the concrete truck mixers, followed by the cement Figure 8. Internally cured concrete slab shrinkage
and additives, and finally water. The vacuum
saturated pumice aggregates were added to the The specific site chosen for the investigation into
mixture at the same time as the sand, and were in internally cured concrete slabs may have had an
the wet moisture state during batching. effect on the resulting in-situ strength and
shrinkage of the concrete. As the site chosen had
Studs with a central sharp v-groove (similar to many variables not encountered in a laboratory
DEMEC points) were cast flush into the slabs in a environment, such as sun exposure, uneven
grid arrangement for gathering shrinkage data. A shading and precipitation, their influence on the
large vernier calliper (600 mm length) with pointed effectiveness of internal curing using saturated
jaws was used for taking measurements of the pumice aggregates needs to be studied in greater
distance between the studs. These distances were detail. At the time of writing this paper, only limited
used to calculate the average concrete strain for results were available for publication. These
each of the three slabs. Further details of the slab results are part of an on-going investigation, with
specifications and testing regime can be found in data planned to be acquired over a 1 year period.
Green [22]. This will allow more conclusive long-term data to
be obtained.
Preliminary results from both in-situ and laboratory
based compressive strength tests of the three CONCLUSIONS
mixtures used in the slabs are given in Table 12. This paper has presented research conducted on
These indicate that the inclusion of pumice the utilisation of New Zealand’s naturally occurring
aggregates into the mixtures appears to reduce the lightweight pumice aggregates for producing
in-situ compressive strength of the concrete. The structural grade (25+ MPa) lightweight concrete,
laboratory based cylinder compressive strength and as an internal curing media in both HPC and
results however were reasonably consistent across thin concrete sections with a relatively low w/cm
the three mixtures investigated. ratio and high surface area to volume ratio.
Additionally, the development of a vacuum
saturation procedure for efficiently saturating
pumice aggregates to a wet moisture state has
been discussed. The research presented
demonstrates the potential to commercially
produce high performance lightweight concrete
containing naturally occurring pumice aggregates
reaching compressive strengths greater than Congress of the Precast Concrete
40 MPa, while meeting the requirements of Industry: Istanbul, Turkey.
New Zealand Concrete Design Standards. Results 10. ASTM (2005) ASTM Standard C 567 - 05,
from both the internal curing investigations indicate in Standard Test Method for Determining
that the inclusion of saturated fine pumice appears Density of Structural Lightweight Concrete.
to cause a slight, unexpected, increase in ASTM International, West Conshohocken,
shrinkage and a small reduction in compressive PA, www.astm.org.
strength of the concrete at an early age. More 11. SNZ (2006) NZS 3101:2006 Concrete
detailed research is recommended to determine Structures Standard, in Part 1 - The
the effects of silica fume and site specific Design of Concrete Structures. Standards
conditions for the HPC and thin slab investigations New Zealand.
respectively. 12. CEB (1977) Euro-International Committee
for Concrete, in Draft International
The effective utilisation of New Zealand’s abundant Standard Model Code for Concrete
resources of naturally occurring lightweight pumice Construction. CEB: Paris.
aggregates will provide an environmentally 13. BSC (1996) ENV 1992-1-4: 1996
friendlier and more cost effective solution to the Eurocode 2: Design of Concrete
commercial production of structural lightweight Structures. General Rules., in Lightweight
concrete for the New Zealand concrete industry. Aggregate Concrete with Closed Structure.
British Standards Institution, UK
ACKOWLEDGEMENTS 14. SNZ (1986) NZS 3112.2:1986 Methods of
The authors gratefully acknowledge the financial test for concrete, in Part 2: Tests relating
support provided by Winstone Aggregates, Golden to the determination of strength of
Bay Cement, Humes Pipeline Systems and the concrete. Standards New Zealand.
New Zealand Tertiary Education Commission. 15. SNZ (1974) NZS 3152:1974 Specification
for the Manufacture and Use of Structural
REFERENCES and Insulating Lightweight Concrete.
1. CCANZ (1980). Lightweight Concrete. Standards New Zealand.
New Zealand Concrete Construction, 24 16. Neville, A.M. (1995). Properties of
(February): p. 5-12. Concrete. 4th ed., Essex: Longman Group
2. New Zealand. Concrete Construction Limited. 844.
(1970). Pumice Concrete Research. New 17. Mindess, S., J.F. Young, and D. Darwin
Zealand Concrete Construction (2002). Concrete. 2nd ed., Upper Saddle
(November): p. 191. River: Prentice Hall. 644.
3. McSaveney, L.G. (2000) The Wellington 18. ACI (2003) ACI 213R-03, Guide for
stadium: New Zealand's first use of high structural lightweight aggregate concrete.,
strength lightweight precast concrete, in in ACI Manual of Concrete Practice, Part
Second international symposium on 1: Materials and General Properties of
structural lightweight aggregate concrete. Concrete. American Concrete Institute:
Helli Grafisk, Oslo, NORVEGE: Farmington Hills, Michigan. p. 38.
Kristiansand. 19. Lura, P., et al. (2004). Pumice aggregates
4. Carryer & Associates LTD (1995) Pumice for internal water curing. in International
Resources of New Zealand. RILEM Symposium on Concrete Science
5. Moss, P.J. (1970). Lightweight Aggregate and Engineering: A Tribute to Arnon
Concrete and their use in New Zealand. Bentur. Evanston, IL.
New Zealand Concrete Construction, 20. Williamson, G.S., et al. (2007). Vacuum
February: p. 25-29. Saturated Absorption As Aggregate
6. PPS (2008) Pacific Pumice Supplies. Date Durability Indicator. ACI Materials Journal,
[cited 2008 15 March]; Available from: 104 (3).
www.pumice.co.nz. 21. Hossain, K.M.A. (2008). Bond
7. SNZ (1986) NZS 3111:1986 Methods of characteristics of plain and deformed bars
Test for Water and Aggregate for in lightweight pumice concrete.
Concrete. Standards New Zealand. Construction and Building Materials, 22
8. Hossain, K.M.A. and M. Lachemi (2007). (7): p. 1491-1499.
Mixture design, strength, durability, and 22. Green, S.M.F. (2009) High Performance
fire resistance of lightweight pumice Lightweight Pumice Aggregate Concrete,
concrete. ACI Materials Journal, 104 (5): p. in Department of Civil and Environmental
449-457. Engineering. The University of Auckland:
9. Yeginobali, A. (2002) Structural lightweight Auckland. p. 183.
concretes produce with natural lightweight 23. Li, V.C., et al. (1994). On the shear
aggregates in Turkey, in 17th International behavior of engineered cementitious
 composites. Advanced Cement Based
Materials, 1 (3): p. 142-149.
24. Kovler, K. and O.M. Jensen (2005). Novel
Techniques for Concrete Curing. Concrete
International, 27 (9): p. 39-42.
25. Bentz, D.P., L. Pietro, and J.W. Roberts
(2005). Mixture proportioning for internal
curing. Concrete International, 27 (2): p.
35-40.
26. Zhutovsky, S., K. Kovler, and A. Bentur
(2004). Influence of cement paste matrix
properties on the autogenous curing of
high-performance concrete. Cement and
Concrete Composites, 26 (5): p. 499-507.
27. Pyc, W.A., et al. (2007) Internal Curing
Study with Intermediate Lightweight
Aggregate, in ACI Special Publication SP-
256 Internal Curing of High Performance
Concrete: Lab and Field Experiences.
28. Jensen, O.M. and P.F. Hansen (2001).
Water-entrained cement-based materials:
I. Principles and theoretical background.
Cement and Concrete Research, 31 (4): p.
647-654.