Construction and Building Materials 74 (2015) 132–139

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier .com/locate /conbui ldmat

The influence of aggregate type on the strength and elastic modulusof high strength concrete

http://dx.doi.org/10.1016/j.conbuildmat.2014.08.0550950-0618/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +27 (0) 21 650 5181.

Hans Beushausen ⇑, Thomas DittmerConcrete Materials and Structural Integrity Research Unit, University of Cape Town, Department of Civil Engineering, South Africa

h i g h l i g h t s

� The coarse aggregate type has an apparently contradicting influence on strength and elastic modulus of HSC.� Aggregate with higher strength and stiffness results in lower strength but higher elastic modulus of HSC.� In contrast to compressive strength, tensile strength of concrete is not significantly affected by coarse aggregate type.

a r t i c l e i n f o

Article history:Received 14 March 2014Received in revised form 6 August 2014Accepted 23 August 2014Available online 6 November 2014

Keywords:ConcreteStrengthElastic modulusAggregateHigh strength concrete

a b s t r a c t

This paper examines the influence of 2 common South African aggregate types on the compressivestrength, split and flexural tensile strength and elastic modulus properties of high strength concrete.Two different aggregate types (Andesite and Granite) were used to produce concrete with targetstrengths ranging from 30 MPa to 120 MPa. Granite concrete was found to have a higher compressivestrength, while the stiffer Andesite aggregate produced concrete with a significantly higher elastic mod-ulus. An attempt is made to explain this phenomenon with fracture mechanic theories. No trend wasidentified for the influence of aggregate type on splitting and flexural tensile strength. The effectivenessof the SANS and EN elastic modulus prediction models was analysed against the test results and it wasfound that both prediction models accurately predicted elastic modulus values for the Andesite concrete,but produced far less accurate predictions for the Granite concrete.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

High strength concrete (HSC) has been used extensively aroundthe world, but has only recently become popular in South Africa.This form of concrete allows for much smaller concrete cross-sections in members, resulting in lower volumes of concrete beingrequired. There are obvious benefits associated with this pertain-ing to reductions in formwork, transportation, reinforcing andon-site handling costs as well as the further benefits of structuralmembers with significantly reduced self-weights that occupymuch less space. The high strength required for HSC relies heavilyon achieving a very low water: binder ratio of below 0.4 [20]. Thisallows high strength to be achieved by reducing porosity, inhomo-geneity and micro cracks in the cement paste [15]. This in turn willhave a direct effect on other intrinsic properties such as, tensilesplit and flexural strength, elastic modulus, creep deformationand shrinkage. Parameters such as, cement type, supplementarycementitious materials, chemical admixtures and curing

techniques have been identified as having an influence on thesekey properties of HSC. However, the influence of aggregate andin particular, coarse aggregate type must also be considered.

Longstanding studies by Aitcin and Mehta [2], Zhou et al. [21]and de Larrard and Belloc [9] agree on the influence of differentaggregate types on concrete compressive strength, with strongeraggregate types increasing the overall strength of the concrete.This is further supported by studies done by Ezeldin and Aitcin[10], Özturan and Çeçen [16] and Beshr et al. [6] who not onlyhighlight the direct relationship between aggregate strength andcompressive strength, but also show that there is a similar rela-tionship with respect to concrete tensile and flexural strength.

A more recent study by Kılıç et al. [12] and reviewed by Kovlerand Roussel [13] examined the influence of aggregate type on thestrength and abrasion characteristics of high strength silica fumeconcrete. Five different aggregate types (gabbro, basalt, quartzite,limestone and sandstone) were tested with the same mortar mix.It was found that the stronger gabbro aggregate concrete had thehighest compressive and flexural tensile strength while the weak-est sandstone aggregate concrete had the lowest compressive and

H. Beushausen, T. Dittmer / Construction and Building Materials 74 (2015) 132–139 133

flexural strength. This confirmed the direct relationship betweenaggregate compressive strength and the resulting concrete com-pressive and flexural strength. This is supported by findings byAhmad and Alghamdi [1], who conducted an investigation to studythe effect of two different types of coarse aggregates on the perfor-mance of concrete in terms of compressive strength, modulus ofelasticity, and steel-corrosion penetration rate. It was found thatthe strength qualities of the aggregate control the strength proper-ties of the concrete.

Uysal [18] investigated the influence of coarse aggregate typeon mechanical properties of self-compacting concrete (SCC) andalso concluded that greater concrete strength was achieved withthe use of higher strength aggregates.

Another key finding from Kılıç et al. [12] was that concretesmade with aggregates with a high compressive strength (gabbroand quartzite) had a lower compressive strength than the actualcompressive strength of the aggregate itself, while the lowerstrength aggregate concrete (basalt, limestone and sandstone)had a very similar compressive strength to the compressivestrength of the respective aggregate itself. These findings wereattributed to the idea that the strength of concretes made withhigher strength aggregates is limited by the strength of the mortarpaste and not the strength of the aggregate itself.

The elastic modulus of concrete is directly proportional to thestiffness of the individual phases that form its composition and theirinterfacial characteristics [5]. Aggregate type therefore has a directeffect on the elastic properties of HSC and Ahmad and Alghamdi[1] have shown that the effect of aggregate type on elastic modulusof concrete is very significant. This is supported by Uysal [18], whoshowed that an increase in elastic modulus of the concrete wasfound when a higher stiffness aggregate was used.

Rashid et al. [17] compared 644 elastic modulus results for con-cretes made with various aggregate types from available literature.These results were plotted against compressive strength and it wasconcluded that the impact of aggregate type on the elastic modulusof concrete at various strengths was very pronounced, with lowerstiffness aggregates such as sandstone producing concrete with asignificantly lower elastic modulus for all concrete strengths.

It is therefore necessary to consider the effect of the presence ofcoarse aggregate inclusions on the ‘localised’ mechanisms involvedwith concrete failure. Chiaia et al. [8] explains how, in heteroge-neous materials such as concrete, failure will generally occur alongthe weakest link. This weak link is commonly represented by theinterfacial transition zone (ITZ) between two dissimilar materials.In the case of concrete, the ITZ between the (mortar or cement)matrix and the aggregate particle is considered to be the weakestelement in the material. The material is most likely to fail at thelocation with the highest stress relative to the interface strength.These higher stresses are located where two different materialswith different elastic properties meet, at the interface betweenthe matrix and the aggregate, resulting in stress concentrations.These stress concentrations play a key role in concrete fractureand failure.

Giaccio and Zerbino [11] describe how coarse aggregate parti-cles arrest failure crack growth, producing meandering andbranching cracks. This influence of coarse aggregate on crack prop-agation is dependent on aggregate characteristics such as surfacetexture, shape and stiffness. As previously highlighted, the differ-ence in elastic properties of the matrix and the aggregate will havea profound influence on concrete failure. Giaccio and Zerbino [11]suggest that aggregate surface texture is one of the most importantfactors that affect matrix–aggregate bond strength, with rougheraggregates having superior bonds.

The concept of strain softening prior to failure of concrete mustalso be considered and involves the formation and nucleation ofmicrocracks at these stress concentration zones and pre-existing

microdefects which become ‘‘attractors’’ for the subsequent mac-rocracking development [8,19]. The process of macrocracking fail-ure is characterised by the growth and expansion of interfacialcracking through the matrix. During the strain softening stage, thismacrocracking will remain discontinuous until complete crackingfailure is observed when macrocracking becomes continuousthrough the matrix. Strain softening can therefore improve the per-formance of concrete and directly influence the fracture energyrequired for failure. Fracture energy is described by Wittmann[19] as the specific energy required for cracking to occur.

However, studies by Wittmann [19] and Kovler and Bentur [14]found that the energy consuming cracking failure (strain softening)in normal strength concrete (NC) was not observed in the crackingof high strength concrete (HSC). Due to the increased strength ofthe matrix associated with HSC, crack formations were observedto run through the inclusions (aggregate) and form a plane similarto that observed for the failure of fine mortars. It was deduced thatmechanisms of mechanical interaction (strain softening) betweenthe inclusions and the HSC matrix were minimal, resulting in themore brittle failure of the HSC around the stress concentrations.

Another reason for the more brittle fracture process of HSC,compared to NC may be the consumption of Portlandite in the puz-zolanic reaction in concretes using silica fume, which results in abetter bond between aggregate and cement paste. As a conse-quence, the interface transition zone, often considered being theweak link in the concrete matrix and significantly contributing tothe strain softening phenomenon, is strengthened.

This paper represents the results of a laboratory investigationthat aimed to study the effect of two different aggregate types(Andesite and Granite) on the compressive strength, splitting ten-sile strength and elastic modulus of high strength concrete. Resultsshow that aggregate type has a significant influence on concreteproperties which is not always considered in conventional viewsand prediction models.

2. Experimental details

2.1. Mix design and curing

Four mixes of varying target strengths were developed thatwould be tested using Andesite and Granite fine and coarse aggre-gate (Table 1). The 90 and 120 MPa mix designs were based on mixdesigns from North America, developed by Burg and Ost [7], andperfected using a number of trial mixes. These would be comparedto 60 and 30 MPa mixes. This would allow for comparisons to bemade between the HSC mixes and normal concrete (NC) mixes.

The coarse aggregate content was kept the same to ensure aneffective comparison. Silica fume was added to all of the mixdesigns, replacing roughly 10% of the cement content. Sika Visco-Crete – 10 superplasticiser was used for the 90 and 120 MPa mixesand Sikament – NN liquid superplasticiser was used for the 60 MPamixes.

Two types of coarse and fine aggregates were used for theinvestigation mix designs from sources in the Gauteng Provinceand the Western Cape Province of South Africa. The coarse aggre-gate type from the Gauteng source is 19 mm Andesite. The aggre-gate is very angular in shape. The fine aggregate is from the samequarry and is crusher sand made from the same Andesite deposit.Alexander et al. [4] describes how Andesite develops an excellentbond with Portland Cement (PC) and enhances the fracture tough-ness of the paste-rock composite. Andesite has a relative densityRD of 2.91, a stiffness factor (Ko) of ±26 GPa and an elastic modulusof ±81 GPa [3].

The coarse aggregate type from the Western Cape source is a19 mm Granite and is also very angular in shape. Again the fine

Table 1Final mix designs used in the investigation.

Mix: 120 MPa mix 90 MPa mix 60 MPa mix 30 MPa mix

Target compressive strength (MPa) 120 90 60 30Component materials in 1 m3 batch (kg/m3)Cementitious materials

Cement 610 500 286 197Silica fume 50 50 32 21Total 660 550 318 218

Water 145 170.5 175 180Fine aggregate 631 656 843 916Coarse aggregate 1070 1070 1070 1070Superplasticiser L/m3 7.75 3.25 – –w/b ratio by wt. 0.22 0.31 0.55 0.83Superplasticiser (% Cement wt.) 1.35 0.69 – –

Table 2Testing program.

Parameter Specimen type Specimen dimensions (mm) Ages when tested (days) No. of specimens tested (at each age)

Compressive strength Cube 100 � 100 � 100 1, 3, 7 and 28 3Splitting strength Cube 100 � 100 � 100 1, 3, 7 and 28 3Flexural strength Beam 100 � 100 � 500 1, 3 and 28 2Elastic modulus Cylinder 80 � 150 14 and 28 3

134 H. Beushausen, T. Dittmer / Construction and Building Materials 74 (2015) 132–139

aggregate is from the same quarry and is crusher sand made fromthe same deposit. Alexander et al. [4] explains how Granite devel-ops a slightly weaker bond with PC than Andesite, but the resultingpaste-rock composite is similar to Andesite. Granite has a relativedensity RD of 2.63, a stiffness factor (Ko) of ±21 GPa and an elasticmodulus of ±58 GPa [3].

Specimens were covered with wet hessian and plastic for 24 hafter initial casting. They were then demoulded and wet cured ina curing tank at 23 �C until they were tested.

Table 2 presents an overview on the type and number of testspecimens.

Fig. 2. Load positioning dimensions for flexural strength test.

2.2. Testing procedures

2.2.1. Compressive strength testCompressive strength tests were conducted at 1, 3, 7 and

28 days after initial casting. The tests were conducted inaccordance with the SABS Method 861-2:1994. Three

Fig. 1. Tensile splitting test using the AMSLER. Fig. 3. Target layout on cylinder specimens.

H. Beushausen, T. Dittmer / Construction and Building Materials 74 (2015) 132–139 135

100 � 100 � 100 mm cubes were crushed for each test using anAMSLER test machine. Loading was applied at a constant loadingrate of 15 MPa per minute until failure.

2.2.2. Splitting tensile strength testSplitting tensile strength tests were conducted at 1, 3, 7 and

28 days after initial casting. The tests were conducted in accor-dance with SABS Method 1253:1994. The 100 � 100 � 100 mmcubes were positioned, with a semicircular bar above andbelow the specimen, between the parallel platens of the AMS-LER (Fig. 1). A compressive load was then applied to the barswhich result in two compressive forces being applied alongtwo diametrically opposite lines. The failure load was thenrecorded. Three cubes were tested to failure for each test andthe splitting tensile strength could then be calculated accordingto elastic theory using the relevant equation specified in thecode.

Table 3Detailed test results.

Test Test age (days) 30 MPa mix Stdev. 60

AndesiteCompressive strength (MPa) 1 3.97 0.12 11

3 11.07 0.31 307 15.53 0.42 42

28 24.83 0.76 54

Splitting strength (MPa) 1 0.17 0.04 03 1.27 0.06 17 1.83 0.13 3

28 2.50 0.19 4

Flexural strength (MPa) 1 0.16 0.03 03 1.06 0.22 1

28 2.07 0.06 2

Elastic modulus (GPa) 14 20.77 1.38 3228 31.77 4.24 44

GraniteCompressive strength (MPa) 1 6.97 0.45 13

3 14.87 0.31 377 20.27 0.46 48

28 31.83 0.76 63

Splitting strength (MPa) 1 0.51 0.00 03 1.44 0.04 27 1.78 0.17 2

28 2.82 0.07 3

Flexural strength (MPa) 1 0.50 0.06 13 1.15 0.16 1

28 2.14 0.16 2

Elastic modulus (GPa) 14 20.24 7.76 2528 23.81 5.38 27

0

1

2

3

4

5

6

7

8

0 5 10 1

Split

ting

tens

ile st

reng

th (M

Pa)

D

30 MPa (Andesite) 60 MPa (Andesite)

30 MPa (Granite) 60 MPa (Granite)

Fig. 4. Splitting tensile strength development for mi

2.2.3. Flexural strength testFlexural strength tests were conducted at 1, 3 and 28 days after

initial casting. The tests were conducted in accordance with SABSMethod 864:1994. 100 � 100 � 500 mm beam specimens weretested using a Denison testing machine which applies a four pointload to the specimen and records the failure load (Fig. 2). Two spec-imens were tested to failure for each test and the flexural strengthwas calculated using the relevant equation from the code.

2.2.4. Elastic modulus testElastic modulus tests were conducted at 14 and 28 days after

initial casting. The tests were conducted using a method thatwas developed and adapted to make use of a Z100 Zwick/Roel Test-ing Machine. Four pairs of brass targets were attached to the80 � 150 mm cylindrical specimens using epoxy so that strainreadings could be recorded using the Demec strain gauge (Fig. 3).Initial ‘zero’ strain readings were taken and then the specimen

MPa mix Stdev. 90 MPa mix Stdev. 120 MPa mix Stdev.

.33 0.06 46.33 0.58 60.20 2.96

.47 1.40 64.67 3.21 83.77 2.63

.10 2.25 79.13 2.87 99.67 2.14

.33 0.58 92.00 1.00 114.73 0.64

.82 0.04 2.50 0.27 3.04 0.45

.99 0.39 4.27 0.17 4.80 0.15

.20 0.15 5.03 0.29 6.05 0.76

.35 0.13 5.35 0.34 7.00 0.32

.79 0.10 1.91 0.03 2.79 0.13

.64 0.10 2.86 0.16 3.87 0.06

.66 0.06 3.53 0.10 4.61 0.16

.98 8.87 44.80 2.97 54.55 10.59

.81 5.29 54.50 15.68 58.77 12.21

.80 1.40 34.73 0.23 63.27 1.53

.60 1.59 69.93 2.69 89.13 2.73

.07 0.83 84.60 2.51 100.27 6.02

.67 0.58 99.67 3.21 117.33 9.29

.87 0.07 2.48 0.23 3.91 0.33

.48 0.17 4.33 0.33 4.86 0.46

.87 0.13 5.22 0.38 5.48 0.40

.74 0.3 5.98 0.34 6.83 0.16

.24 0.16 1.73 0.03 2.84 0.06

.94 0.13 2.63 0.22 3.98 0.03

.93 0.32 3.71 0.03 4.30 0.03

.35 4.83 31.97 2.78 42.58 0.66

.03 0.83 35.87 0.58 46.78 7.09

5 20 25 30ays

90 MPa (Andesite) 120 MPa (Andesite)

90 MPa (Granite) 120 MPa (Granite)

xes made using Andesite and Granite aggregate.

0 0.5

1 1.5

2 2.5

3 3.5

4 4.5

5

0 5 10 15 20 25 30

Flex

ural

stre

ngth

(MPa

)

Days

30 MPa (Andesite) 60 MPa (Andesite) 90 MPa (Andesite) 120 MPa (Andesite)30 MPa (Granite) 60 MPa (Granite) 90 MPa (Granite) 120 MPa (Granite)

Fig. 5. Flexural strength development for mixes made using Andesite and Granite aggregate.

136 H. Beushausen, T. Dittmer / Construction and Building Materials 74 (2015) 132–139

was loaded to 7.5% ultimate compressive strength and the changein length between the targets was recorded. Finally the specimenwas loaded to 15% ultimate compressive strength and the finalchange in length between the targets was recorded. This informa-tion was used to calculate strain values for the different compres-sive loads which could then be used to plot a stress/strain curve forthe specimen and calculate the elastic modulus by determining thegradient of the graph. This process was repeated for three speci-mens for each test and the mean value used in the analysis.

This test method was based on the method outlined in the BS1881-121:1983 and the ASTM C 469-02 codes. The secant moduluswas determined.

3. Test results

Detailed test results are presented in Table 3

3.1. Tensile splitting and flexural strengths

The splitting tensile strength results calculated for all mixes areshown in Fig. 4 and the respective calculated flexural strengthresults are shown in Fig. 5. From the results it can be noted thatthere is no clear trend with respect to the influence of aggregatetype on tensile strength. The influence of aggregate type on con-crete tensile strength is expected to be limited due to the followingfactors:

� Other than in compression, in tensile split or flexural testing,stress concentrations resulting from different aggregate stiff-ness are not expected.

0

20

40

60

80

100

120

0 5 10

Com

pres

sive

stre

ngth

(MPa

)

D30 MPa (Andesite) 60 MPa (Andesite)

30 MPa (Granite) 60 MPa (Granite)

Fig. 6. Compressive strength development for mixe

� In tensile testing, a larger spread of results can generally beexpected when compared to compression testing, and hencethe scatter may also be linked to the variability that is inherentin tensile testing.

However, Giaccio and Zerbino [11] have shown that coarseaggregate factors such as particle shape, texture and stiffness willinfluence the degree of microcracking (strain softening) which willdirectly influence concrete tensile strength. Based on both thesplitting and flexural strength results, it can therefore be deducedthat the degree of strain softening associated with the Granite andAndesite concretes is very similar in magnitude. This thereforesuggests that the ITZ strength, which is particularly influenced bythe paste properties (which are the same in both mixes), is similarfor both aggregate concretes.

3.2. Compressive strength

The compression test results are shown in Fig. 6 and show thatfor all mixes, the Granite concrete had a higher compressivestrength than the Andesite concrete. This can be explained by thehigher elastic modulus of Andesite, which probably results in moresignificant stress concentrations at the interface between the stoneand paste and hence results in the earlier failure.

It was observed that at higher strength (120 MPa), the effect ofaggregate type on compressive strength was less significant. Thiscan be expected because, for higher strength concrete, the strengthof the stone becomes more important. The stress-concentrationeffect is thus partly offset by the benefits gained from the higherstrength aggregate. Furthermore, as brittle failure was observed

15 20 25 30

ays90 MPa (Andesite) 120 MPa (Andesite)

90 MPa (Granite) 120 MPa (Granite)

s made using Andesite and Granite aggregate.

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

30MPa 60MPa 90 MPa 120 MPa

Ela

stic

mod

ulus

(GPa

) Andesite

Granite

Fig. 7. 28 day elastic modulus for mixes made using Andesite and Graniteaggregate.

0

10

20

30

40

50

60

30MPa 60MPa 90 MPa 120 MPa

Ela

stic

mod

ulus

(GPa

)

SANS Equation

EN Equation*

Measured 28 Day Elastic Modulus

*20% increase applied because Andesite is a Basalt aggregate type

Fig. 9. Predicted and measured 28 day elastic modulus for mixes made usingAndesite aggregate.

H. Beushausen, T. Dittmer / Construction and Building Materials 74 (2015) 132–139 137

for the high strength concretes (120 MPa), with failure planes pass-ing directly through the coarse aggregate inclusions, it can beassumed that the influence of strain softening in the ITZ was neg-ligible for both aggregate types, thus possibly accounting for thesimilarity observed in the compressive strength results for thehigher strength concretes. This is supported by work done by Witt-mann [19] and Kovler and Bentur [14].

0

10

20

30

40

50

30MPa 60MPa 90 MPa 120 MPa

Ela

stic

mod

ulus

(GPa

)

SANS Equation

EN Equation*Measured 28 Day Elastic Modulus

*30 % reduction applied because the Granite is assumed to be slightly aggregate typeweatherd and similar to a Sanstone

Fig. 10. Predicted and measured 28 day elastic modulus for mixes made usingGranite aggregate.

3.3. Elastic modulus

The Influence of aggregate stiffness is highly evident in the28 day elastic modulus results shown in Fig. 7, confirming thataggregate type has a significant effect on the elastic properties ofconcrete.

In principle, the trend that is evident in Fig. 7 could be explainedby two factors, namely, an improved ITZ (interfacial transitionzone) stiffness or a greater aggregate stiffness. The first explanationcan be discounted as no noticeable trend was identified in the ten-sile testing so it can be deduced that the ITZ of the Andesite andGranite is similar, and will therefore be expected to have a similarstiffness for both aggregate types. As the cement paste was keptconstant for both aggregate types it is proposed that the stone stiff-ness of the Andesite and the Granite has the largest influence onthe difference in elastic qualities of the different aggregate con-cretes. This is supported by Alexander and Mindess [5] who sug-gest that the elastic modulus of concrete is indeed directlyproportional to the stiffness of the concrete paste, the aggregateand the ITZ, and as both the stiffness of the paste and ITZ are sim-ilar in this case, it is the inherent aggregate stiffness that is influ-encing the results.

The rate of development of elastic modulus for the highstrength mixes is shown in Fig. 8, and indicates that this develop-ment is similar in nature to the development of compressive andtensile strength. This supports findings by Ahmad and Alghamdi[1] and Uysal [18]. Initial rates of elastic development at youngerages (first 14 days) are much greater and tend to level off after

0

10

20

30

40

50

60

0 5 10

Ela

stic

Mod

ulus

(GPa

)

D90 MPa (Andesite) 120 MPa (Andesite)

Fig. 8. Elastic modulus development for high strength m

roughly 30 days. The noticeable influence of aggregate type onthe rate of development is observed during this first 14 days ofdevelopment where the Andesite mixes are shown to have a morerapid rate of stiffness development than the Granite mixes. Afterthe initial 14 days, the influence of aggregate type on stiffnessdevelopment is observed to be minimal as all mixes are shownto develop at a similar rate.

Figs. 9 and 10 show the comparisons of the measured 28 dayelastic modulus values compared with the respective predictedvalues that are calculated using the prediction models stipulated

15 20 25 30

ays90 MPa (Granite) 120 MPa (Granite)

ixes made using Andesite and Granite aggregate.

0

5

10

15

20

25

30

30MPa 60MPa 90 MPa 120 MPa

Com

pres

sive

stre

ngth

/tens

ile

flexu

ral s

tren

gth

Andesite

Granite

Fig. 13. Ratio between 28 day compressive strength and tensile flexural strength.

138 H. Beushausen, T. Dittmer / Construction and Building Materials 74 (2015) 132–139

in the SANS10100-1:2000 and the BS EN 1992:2004 codes. TheSouth African expression (SANS10100-1:2000) is based on theBritish BS 8110 method which specifies an equation that is a func-tion of the characteristic concrete strength. Aggregate type isaccounted for quite extensively through the inclusion of an aggre-gate stiffness constant (Ko) which is based on the elastic modulusof the aggregate itself. An aggregate specific coefficient (a) is alsoincluded.

This prediction model was shown to successfully predict elasticmodulus values for the Andesite mixes (Fig. 9), however, predic-tions for the Granite mixes (Fig. 10) were substantially over esti-mated when compared to the measured values.

The Eurocode expression (BS EN 1992:2004) is also a function ofthe characteristic strength. However, the expression is based onthe CEB-FIP MC 90 method which makes no allowances for theeffect of aggregate. The EN version does make some broad allow-ances for aggregate type with suggested percentage reductions oradditions for different aggregate groups. These are however, verybroad and unspecific. From the results it can be noted that theEN model was also far more accurate for the Andesite mixes(Fig. 9), but tended to slightly underestimate elastic modulus val-ues as the strength of the Andesite mixes was increased. In contrastto the SANS model, the EN model gave accurate predicted valuesfor the lower strength Granite mixes (30 and 60 MPa) shown inFig. 10, but it was observed that the predictions significantlyunderestimated the measured value for the higher strength mixes(90 and 120 MPa). A 30% reduction was applied to the EN predic-tion as it was assumed that the Granite aggregate was slightlyweathered and therefore similar to a sandstone type aggregate.

4. Discussion

The most notable results of this research are the different influ-ence of aggregate type on the strength and elastic properties ofconcrete.

0

0.5

1

1.5

2

2.5

3

30MPa 60MPa 90 MPa 120 MPa

Com

pres

sive

stre

ngth

/ela

stic

m

odul

us (M

Pa/G

Pa)

Andesite

Granite

Fig. 11. Ratio between 28 day compressive strength and elastic modulus.

0 2 4 6 8

101214161820

30MPa 60MPa 90 MPa 120 MPa

Com

pres

sive

stre

ngth

/tens

ile

split

ting

stre

ngth

Andesite

Granite

Fig. 12. Ratio between 28 day compressive strength and tensile splitting.

The effect of aggregate type on the elastic properties of theconcrete can be confirmed by extensive work presented in theliterature by Rashid et al. [17], Alexander and Mindess [5], Ahmadand Alghamdi [1] and Uysal [18], which confirm that aggregatestiffness has a direct impact on the elastic modulus of the concrete.This is supported by findings in this investigation which show theimpact of the stiffer Andesite aggregate which results in signifi-cantly higher elastic modulus values for all Andesite concretemix strengths. Furthermore, the impact of aggregate type on thedevelopment of elastic properties was shown to be significant atonly younger ages (first 14 days).

Findings with respect to compressive strength do, at firstglance, seem to contradict general views that stronger aggregatetypes will result in higher compressive concrete strength. Explana-tions for the findings in this study detail how the stiffer Andesiteaggregate results in higher stress concentration at the interfacebetween the stone and the paste, resulting in earlier failure.However, this does not in fact contradict the literature findings,as almost all of the previous studies propose a direct relationshipbetween specific aggregate compressive strength and the resultingconcrete compressive strength, and do not discuss in detail theeffect of aggregate stiffness on compressive strength. It is thereforepossible that the low elastic properties of the Granite aggregate aresuch that stress concentrations in the ITZ will be minimised to sucha degree that a stronger concrete compressive strength is achieved,despite the fact that the Andesite is a stronger aggregate type. Thisrelationship is illustrated in Fig. 11 which shows the ratio betweencompressive strength and elastic modulus. It can be observed thatthis ratio is significantly higher for all Granite mixes, indicating ahigh compressive strength – low elastic modulus relationshipwhich tends to increase in magnitude as the target Granite mixstrength increases.

As previously mentioned, no clear trend was identified in theresults with respect to the influence of aggregate type on concretetensile strength. This contradicts work done by Ezeldin and Aitcin[10], Özturan and Çeçen [16], Beshr et al. [6] and Kılıç et al. [12]who present findings that show that the effect of aggregate typeon tensile strength is similar to that of the effect on compressivestrength. It can be seen in trends illustrated in Figs. 12 and 13,which show the ratio of compressive strength to tensile splittingand tensile flexural strength, that the ratios of compressive to ten-sile strength are similar for both aggregate types. It is noted thatthe Granite mixes have a slightly higher ratio for all mix strengthswhich can be attributed to the higher compressive strengthsassociated with this aggregate type.

The accuracy of the SANS and EN prediction models wereshown to effectively predict the elastic properties of the Andesiteconcrete mixes. However, both prediction models were shown tobe relatively inaccurate at predicting the elastic properties of theGranite concrete. When comparing the outputs of elastic modulusprediction models with specific test results, Rashid et al. [17]

H. Beushausen, T. Dittmer / Construction and Building Materials 74 (2015) 132–139 139

warns that even with specific aggregate types, elastic modulus val-ues for concrete may vary within quite a large bandwidth.

5. Conclusions

The conclusions that can be drawn from this investigation are asfollows:

� Although Andesite aggregate has a higher aggregate elasticmodulus when compared to Granite aggregate, this stiffnessresults in stress concentrations at the interface between thestone and the paste resulting in earlier failure and a lower com-pressive strength for Andesite concrete compared to Graniteconcrete.� The effect of aggregate stiffness on compressive strength is less

apparent at higher strengths because at higher strengths, thestrength of the aggregate becomes more important, and there-fore the stress concentration effect of the stiffer aggregate ispartly offset by the increase in compressive strength due tothe higher aggregate strength.� The tensile strength of concrete made with Andesite and

Granite aggregate was similar, probably because the degree ofstrain softening and ITZ strength was similar for both aggregatetypes.� The influence of aggregate type, and specifically aggregate stiff-

ness was shown to have a profound effect on concrete elasticproperties. As the paste and ITZ stiffness was constant for allmixes it can be concluded that aggregate stiffness dictated theelastic modulus of the concrete, with the stiffer Andesite aggre-gate resulting in higher concrete elastic modulus values for thevarious strength Andesite mixes.� Aggregate type was shown to only affect the rate of develop-

ment of elastic properties of the concrete at younger ages (first14 days).� Both the SANS and EN elastic modulus prediction models were

shown to accurately predict and account for the effect of aggre-gate type for the different strength Andesite mixes.� The SANS elastic modulus prediction model significantly over-

estimated the predicted elastic modulus values for all of theGranite mixes. The EN elastic modulus prediction modelshowed some level of accuracy for the predictions for lowerstrength (30 and 60 MPa) Granite mixes, but also tended to sig-nificantly overestimate predictions for the higher strength (90and 120 MPa) Granite mixes.� Although the novelty of the research is not significant, the

above findings can be effectively applied to the applicationand use of HSC, in particular when the design requires the con-sideration of various concrete properties.

Acknowledgements

The authors gratefully acknowledge the financial support of theConcrete Materials and Structural Integrity Research Unit at TheUniversity of Cape Town (CoMSIRU), and the industrial sponsorsof CoMSIRU: PPC Cement (Pty) Ltd., the Concrete Institute, andSIKA South Africa.

References

[1] Ahmad S, Alghamdi SA. A study on effect of coarse aggregate type on concrete.Perform Arab J Sci Eng 2012;37:1777–86.

[2] Aitcin PC, Mehta PK. Effect of coarse aggregate characteristics on mechanicalproperties of high strength concrete. ACI Mater J 1990;87(2):103–7.

[3] Alexander MG, Davis DE. Properties of aggregates in concrete (Part 1). HippoQuarries Technical Publication. South Africa: University of the Witwatersrand;1990.

[4] Alexander MG, Mindess S, Diamond L. Properties of paste-rock interfaces andtheir influences on composite behaviour. Canada: University of BritishColumbia; 1995.

[5] Alexander MG, Mindess S. Aggregates in concrete. Abingdon: Taylor & Frances,Oxon; 2005.

[6] Beshr H, Almusallam AA, Maslehuddin M. Effect of coarse aggregate quality onthe mechanical properties of high strength concrete. Constr Build Mater2003;17:97–103.

[7] Burg RG, Ost BW. Engineering properties of commercially available high-strength concretes. PCA Research & Development Bull. RD104T, PortlandCement Association, Skokie, Ill; 1992.

[8] Chiaia B, van Mier JGM, Vervuurt. Crack growth mechanisms in four differentconcretes: microscopic observations and fractal analysis. Cem Concr Res1998;28(1):103–14.

[9] de Larrard F, Belloc A. Influence of aggregate on the compressive strength ofnormal and high-strength concrete. ACI Mater J 1997;94(5):417–26.

[10] Ezeldin AS, Aitcin PC. Effect of coarse aggregate on the behaviour of normaland high strength concretes. Cem Concr Aggr 1991;13(2):121–4.

[11] Giaccio G, Zerbino R. Failure mechanism of concrete: combined effects ofcoarse aggregates and strength level. Adv Cem Based Mater 1998;7:41–8.

[12] Kılıç A, Atis� CD, Teymen A, Karahan O, Özcan F, Bilim C, et al. The influence ofaggregate type on the strength and abrasion resistance of high strengthconcrete. Cem Concr Compos 2008;30(4):290–6.

[13] Kovler K, Roussel N. Properties of fresh and hardened concrete. Cem Concr Res2011;41:775–92.

[14] Kovler K, Bentur A. Cracking sensitivity of normal- and high-strengthconcretes. ACI Mater J 2009;106(6):537–42.

[15] Mehta PK, Monteiro PJM. Concrete: microstructure, properties, and materials.3rd ed. USA: McGraw-Hill; 2006.

[16] Özturan T, Çeçen C. Effect of coarse aggregate type on mechanical properties ofconcretes with different strengths. Cem Concr Res 1997;27:165–70.

[17] Rashid MA, Mansur MA, ASCE M, Paramasivam P. Correlations betweenmechanical properties of high-strength concrete. J Mater Civ Eng 2002;14:230–8.

[18] Uysal M. The influence of coarse aggregate type on mechanical properties of flyash additive self-compacting concrete. Constr Build Mater 2012;37:533–40.

[19] Wittmann FH. Crack formation and fracture energy of normal and highstrength concrete. Sadhana 2002;27(4):413–23.

[20] Wu KR, Chen B, Yao W, Zhang D. Effect of coarse aggregate type on mechanicalproperties of high-performance concrete. Cem Concr Res 2001;31:1421–5.

[21] Zhou FP, Lydon FD, Barr BIG. Effect of coarse aggregate on elastic modulus andcompressive strength of high performance concrete. Cem Concr Res1995;25(1):177–86.