Lightwight Concrete

Materials and Design 56 (2014) 1039–1048

Contents lists available at ScienceDirect

Materials and Design

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

Influence of mineral additions and different compositional parameterson the shrinkage of structural expanded clay lightweight concrete

0261-3069/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.matdes.2013.12.013

⇑ Corresponding author. Tel.: +351 218418226; fax: +351 218418380.E-mail address: [email protected] (J.A. Bogas).

J. Alexandre Bogas ⇑, Rita Nogueira, Nuno G. AlmeidaDECivil/ICIST, Instituto Superior Técnico, Technical University of Lisbon, Av. Rovisco Pais, 1049-001 Lisbon, Portugal

a r t i c l e i n f o

Article history:Received 17 July 2013Accepted 5 December 2013Available online 13 December 2013

Keywords:Lightweight aggregate concreteShrinkageFly ashSilica fumeNanosilica

a b s t r a c t

A comprehensive experimental study was carried out on the shrinkage behaviour of structural expandedclay lightweight aggregate concrete (LWC), taking into account different compositions, types and initialwetting conditions of lightweight aggregates (LWA). The influence of different compositional parameterson shrinkage, such as the type and volume of aggregate, the w/c ratio, the binder content and the partialreplacement of normal aggregates by LWA was analysed. The shrinkage of LWC depends on how the vol-ume of LWA varies. The influence of different pozzolanic materials was also studied, namely, silica fume,nanosilica and fly ash. Depending on the type, content and reactivity of the pozzolanic additions, theshrinkage was higher or lower than that of LWC without admixtures. The initial wetting condition ofLWA had little influence on the long-term shrinkage. The LWC with the most porous aggregates is moreaffected by the cross-section geometry of concrete in that it is more susceptible to differential shrinkage.Current standard expressions did not properly predict the shrinkage behaviour of LWC. Multiplier coef-ficients of about 1.3 for the most common structural LWA and about 1.6 for more porous LWA are sug-gested, to take into account the higher long-term shrinkage of LWC.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

It is well known that the shrinkage of concrete is essentiallygoverned by the paste, which is the source of contraction of thesystem, and by the aggregates that oppose this contraction. Sincestructural lightweight aggregate concrete (LWC) is modified atthese two factors, it is expected that its behaviour differs from thatobserved in normal density concrete (NWC).

On the one hand, the use of less rigid porous aggregates de-creases the restriction effect on paste deformation [1–3]. On theother hand, either for strength purposes or for reasons of workabil-ity and stability of the mixes, LWC is usually characterized by lar-ger volumes of better quality paste and lower volumes of coarseaggregates than NWC [4,5]. Therefore, the long-term shrinkage ofLWC should be higher than that of NWC of the same strength.However, the paste quality is higher in LWC, which increases thematrix stiffness. Moreover, the water absorbed by lightweightaggregates (LWA) is later released into the paste by internal curing,which compensates for the initial water lost by drying and self-desiccation.

According to Kayali et al. [2], the improved aggregate-pasteinterface in LWC helps to enhance the restriction effect of theLWA. Moreover, because of the internal curing and the prolonged

hydration of the paste the deformation resistance of the matrix ishigher and less water is available for evaporation [6]. The combina-tion of all these factors, together with the high variability of thedifferent types of LWA, explains why some authors report highershrinkage in LWC (e.g. [1,7–9]), while others report shrinkage inLWC that is lower than or similar to shrinkage in NWC (e.g.[1,10–14]).

Most studies have looked at the global shrinkage of LWC of a gi-ven composition and type of aggregate. However, thus far, only afew studies have been published on the detailed analysis of theinfluence of different parameters on LWC shrinkage, such as thevolume and type of LWA, w/c ratio and the amount and type of bin-der, especially the use of different admixtures.

Hoff [15] and Holm [8] reported higher shrinkage in water-cured LWC exposed to 50% RH, produced with fly ash, than in con-crete without this admixture. However, Malhotra [16] has foundlower one year shrinkage in high-strength LWC with fly ash.According to Zhang et al. [1] the addition of 5% of silica fume cansignificantly reduce concrete shrinkage, an effect that is more rel-evant in LWC than in NWC. Lower shrinkage in LWC with silicafume is also reported by Hoff [15]. However, according to Zhutov-sky et al. [17] the addition of silica fume can lead to a significantreduction of internal curing in LWC produced with pumice LWA.

In the main standard documents, such as ACI 209R [18], EN1992 [19] and MC 2010 [20], the shrinkage of LWC is usuallyroughly estimated from the expressions defined for normal density

1040 J.A. Bogas et al. / Materials and Design 56 (2014) 1039–1048

concrete (Table 1). Regarding EN 1992-1 [19], the final dryingshrinkage of LWC is multiplied by an empirical factor of 1.2. Thisfactor is also suggested by MC 2010 [20], but with respect to thetotal shrinkage. Standard EN 1992 (2010) assumes that the autog-enous shrinkage of LWC can be much smaller than that of NWC ifthe LWA is pre-soaked, but no suggestion is given for its estima-tion. The method suggested in ACI 209R [18] was calibrated fromexperimental results obtained in NWC and LWC of low to moderatestrength, under controlled exposure conditions. The expressionsare empirically based and do not model shrinkage phenomena.This document does not provide any coefficient to penalize thelong-term shrinkage of LWC.

The main purpose of our work was to study the exact influenceon LWC shrinkage of different compositional parameters such asthe w/c ratio, type and volume of aggregate, type and content ofbinder, use of additives and the partial replacement of normalaggregates by LWA. This is analysed for a wide range of commonstructural lightweight concretes with about 30–70 MPa and den-sity classes from D1.6 to D2.0. The accuracy of the main standardexpressions is also assessed.

2. Experimental programme

2.1. Materials

Three Iberian expanded clay lightweight aggregates were used(Leca and Argex from Portugal and Arlita from Spain). Their totalporosity, PT, particle density, qp, bulk density, qb, and 24 h waterabsorption, wabs,24 h, are indicated in Table 2. A more detailedmicrostructural characterisation of these aggregates is presentedelsewhere [4,21]. In terms of their specific properties, the selectedLWA are categorized as type A (Arlita), B (Leca) and C (Argex),which represent LWA of low, moderate and high porosity (Table 2).Normal density coarse and fine aggregates (NA) were also used. Forthe reference normal density concrete, two crushed limestoneaggregates of different sizes were combined so as to have the samegrading curve as Leca (20% fine and 80% coarse gravel). Fine aggre-gates consisted of 2/3 coarse and 1/3 fine sand. Their main

Table 1Shrinkage estimation – main standard documents.

Standard Total shrinkage estimation Obse

ACI 209R [18](NWC/LWC)

esh,t = esh,u�(t � ts)/(f + t � ts) csh R(tc),perce

f = 35 for concrete water-cured during 7 daysesh,u = 780�csh � 10�6

csh = ctc�cRH�cvs�cs�cw�cc�ca

Sugestion: f = 26�exp[1.42 � 10�2(v/s)] – to take into accountthe size and geometry on the drying of concrete elements

EN1992 [19](NWC)

ecs (t, ts) = ecd + eca Kh, bhumecd (t) = Kh.ecd,0.bds(t,ts)

ecd,0 = 0.85�bRH�[(220 + 110�ads1)�exp(�ads2�fcm/10)]�10�6

bdsðt � tsÞ ¼ ðt � tsÞ=½ðt � tsÞ þ 0:04 � h3=20 �

eca(t) = [1�exp(�0.2t0.5)]�eca0

eca0 = 2.5�(fck�10)�10�6

(LWC) ecs (t, ts) = g3�ecd + eca g3 =

MC 2010 [20](NWC)

ecs (t, ts) = ecd + eca bRH,typeecd (t) = ecd,0�bRH�bds (t, ts)

ecd,0 = [(220 + 110�ads1)�exp(�ads2�fcm/10)]�10�6

bdsðt � tsÞ ¼ ½ðt � tsÞ=½ðt � tsÞ þ 0:035 � h20 ��

0:5

eca(t) = [1�exp(�0.2t0.5)]�eca0

eca0 = �aas�[(fcm/10)/(6+fcm/10)]2.5 � 10�6

(LWC) ecs (t, ts)LWC = g�ecs (t, ts)NWC g = 1

esh,t – shrinkage strain after a period of time t; esh,u – ultimate shrinkage strain; (v/s) – vecs (t, ts) – shrinkage strain between the age t and the begining of drying ts; ecs0 – ultimaecs – total shrinkage strain; ecd – drying shrinkage strain; eca – autogenous shrinkage strfcm or fck – average or characteristic compressive strength at 28 days.

properties are listed in Table 2. The two fractions of type C LWAwere also combined to have the same grading curve as type BLWA (35% 2–4 and 65% 3–8F, Table 2). Fly ash (FA) from the Pegothermoelectric power plant in Portugal, silica fume (SF) from Spain,type I 52.5 R cement and a polycarboxylate based superplasticizer(Sp) were used. The main properties of these cementitious materi-als are listed in Table 3. A water dispersed RHEOMAC VMA 350nanosilica (NS) was also used. Basically, this admixture is a disper-sion of non-agglomerated spherical nanoparticles of pure silicafume with an average density of 1.1 and about 16.1% solidscontent.

2.2. Concrete mixing and mixture proportions

The concretes were produced in a vertical shaft mixer with bot-tom discharge. Except for initially dry or pre-wetted aggregates,the LWA was pre-soaked for 24 h to better control the workabilityand effective water content of the concrete. The aggregates werethen surface dried with absorbent towels and placed in the mixerwith sand and 50% of the total water. After 2 min of mixing, thecementitious materials and 40% of the water were added. The Spwas added slowly with 10% of water. The total mixing time was7 min. When used, NS was added with about 20% of water after6 min of mixing and mixing then continued for a further 4 min.

Thirty-two different compositions were designed according toBogas and Gomes [22,23], in order to take into account: differentbinder volumes and w/c ratios; volume, initial water content andtype of aggregate; addition of silica fume, nanosilica or fly ash;use of Sp; partial replacement of coarse or fine aggregates byLWA. All concrete compositions are listed in Table 4. The water/binder ratio (w/b) relates to the effective water available for ce-ment hydration. The Sp/c is the percentage of superplasticizer bycement weight. The denominations ‘NWC’, ‘A’, ‘B’ and ‘C’ representthe mixes with NA and type A, type B and type C LWA. The prefix‘V’ refers to different volumes of aggregates for constant amountsof water and cement (variable coarse aggregate/sand), and the pre-fix ‘Vm’ to different volumes of aggregate for constant compositionof the mortar (same cement/aggregate ratio). Except for mixtureBS450, natural sand was used in combination with coarse LWA.

rvations

epresents the product of several factors that take into account the curing timethe relative humidity (RH), the cross-section geometry (vs), the slump (s), thentage of fine aggregate (w) and the air content (a)

RH, ads1 e ads2 Are coefficients depending on the notional size (h0), the relativeidity (RH) and the type of cement (ads1;ads2)

1.2 for fck > LC20/22

aas, ads1 e ads2 Are coefficients depending on the relative humidity (RH) and theof cement (aas;ads1;ads2)

.2 for fck > LC20/22

olume/area ratio of the concrete element.te shrinkage strain.ain.

Table 2Aggregate properties.

Property Normal weight aggregates Lightweight aggregates

Fine sand Coarse sand Fine gravel Coarse gravel Type A Type B Type C

AF7 0–3 4–12 2–4 3–8F

Particle dry density, qp (kg/m3) 2620 2610 2631 2612 1290 1060 1068 865 705Loose bulk density, qb (kg/m3) 1416 1530 1343 1377 738 562 613 423 39724 h Water absorption, wabs,24 h (%) 0.2 0.5 1.4 1.1 12.1 – 12.3 22.9 23.3Total porosity, PT (%) – – – – 52 59 60 67 73Granulometric fraction (di/Di) 0/2 0/4 4/6.3 6.3/12.5 3/10 0.5/3 4/11.2 4/8 6.3/12.5Los Angeles coefficient (%) – – 33.3 30.5 – – – – –

Table 3Main characteristics of cement, fly ash and silica fume.

Parameter Standard Fly ash Silica fume Cement I 52.5 R

Residue on the 45 lm sieve, (%) EN 451-2 10.2 92.0a 1.1Blaine specific surface, (cm2/g) EN 196-6 – – 5102Compressive strength of reference mortar, (MPa) 2 days EN 196-1 – – 40.4

28 days – – 62.7Activity index at 28 days, (%) EN 196-1 83.7b 106.7c –Activity index at 90 days, (%) EN 196-1 103.1b – –Expansion, (mm) EN 196-3 0.5a – 0.5Loss on ignition (LOI), (%) EN 196-7 6.5 3.7 1.64SiO2 + Al2O3 + Fe2O3, (%) EN 196-2 83.0 94.0 29.1CaO, (%) – 3.38 0.83 61.6Free CaO, (%) EN 451-1 0.36 not detected 1.45Density, (g/cm3) EN 196-6 2.33 2.25 3.11

a Residue on the 90 lm sieve.b Mortar with CEM I42,5R + 25% Fly ash.c Mortar with CEM I42,5R + 10% silica fume.

Table 4Concrete mix proportions, dry density, compressive strength and total shrinkage.

Mixes Coarseaggreg.(m3/m3)

Cement(kg/m3)

Addition(kg/m3)

wag

(%)aSp/c(%)

Effectivew/b (L/m3)

Oven drydensity, qd

(kg/m3)

fcm,28d

(MPa)ecs (�10�6 m/m)Age (days)b

3 7 30 90 180 365

A350 0.35 350 – 13.4 0.7 0.45 1789 57.6 �41 �73 �231 �498 �563 �640A450 0.35 450 – 13.1 0.7 0.35 1840 65.8 �14 �36 �130 �292 �490 �508A450PM 0.35 450 – 0 0.7 0.30 1839 63.5 �34 �111 �350 �502 �603 �589A450PD 0.35 450 – 0 0.7 0.30 1846 65.1 �36 �79 �278 �451 �530 �509A525 0.35 525 – 13.4 0.8 0.30 1872 68.5 �14 �10 �51 �180 �360 �458VA425 0.425 450 – 13.4 0.7 0.35 1764 57.9 �31 �1 �63 �258 �408 �476ANS 0.35 414 5.4 14.0 1.2 0.35 1838 65.5 �28 �77 �176 �326 �403 �408AFA22 0.35 350 100 14.0 0.6 0.35 1798 60.0 �9 �50 �238 �424 �510 �534A35 (35%LWA) 0.35 450 – 16.0 0.7 0.35 2150 72.3 �36 �97 �299 �389 �454 �466A65 (65%LWA) 0.35 450 – 16.0 0.7 0.35 2008 66.5 �7 �63 �245 �410 �490 �502

B350 0.35 350 – 23.4 0.6 0.45 1712 43.1 �15 �62 �271 �531 �624 �658B450_0.55 0.35 450 – 19.5 0.0 0.55 1540 36.1 �49 �150 �583 �919 �1065 �1148B450_0.45 0.35 450 – 19.5 0.2 0.45 1676 41.9 �51 �87 �243 �601 �759 �721B450 0.35 450 – 24.5 0.7 0.35 1740 48.6 �18 �35 �133 �333 �493 �591B450PM 0.35 450 – 0 0.7 0.30 1744 46.5 �48 �146 �369 �472 �586 �565B450PD 0.35 450 – 0 0.7 0.30 1753 46.5 �43 �136 �322 �429 �496 �555B525 0.35 525 – 24.0 0.8 0.30 1770 50.0 38 38 26 �49 �184 �369B450_Wsp0.35 0.35 450 – 20.0 1.0 0.30 – 43.3 �12 �49 �128 �322 �563 �561B450_Wsp0.5 0.35 450 – 20.0 0.0 0.50 – 39.0 �58 �138 �462 �791 �928 �977VB400 0.4 450 – 24.0 0.6 0.30 1678 45.7 �18 �52 �114 �303 �511 �596VmB250 0.25 522 – 23.2 0.6 0.35 1876 43.4 �39 �93 �322 �488 �558 �613VmB425 0.425 – 23.9 0.7 0.35 1675 38.6 �5 �41 �149 �292 �415 �442BSF 0.35 414 36 23.4 0.9 0.35 1741 47.6 �8 �38 �132 �381 �537 �587BNS 0.35 414 5.4 24.0 1.0 0.35 1754 46.7 8 �22 �86 �381 �448 �562BFA22 0.35 350 100 24.0 0.7 0.35 1703 42.4 �27 �66 �330 �503 �594 �631BFA40 0.35 270 180 24.0 0.7 0.35 1668 37.1 �24 �77 �300 �421 �468 �462BS450 0.35 450 – 24.0 0.6 0.35 1435 37.5 �61 �113 �374 �605 �733 �825B35 (35%LWA) 0.35 450 – 22.2 0.6 0.35 2112 59.8 �17 �67 �266 �359 �422 �446B65 (65%LWA) 0.35 450 – 22.2 0.6 0.35 1950 53.3 �59 �104 �284 �441 �529 �569

C450 0.35 450 – 19.4 0.7 0.35 1610 31.2 �37 �91 �190 �470 �616 �638NWC350 0.35 350 – 0.3 0.8 0.45 2264 65.8 �44 �112 �260 �306 �348 �410NWC450 0.35 450 – 0.4 0.7 0.35 2299 76.2 �51 �123 �238 �327 �376 �385

a wag – Intial aggregate water content.b The negative sign refers to contraction.

J.A. Bogas et al. / Materials and Design 56 (2014) 1039–1048 1041

1042 J.A. Bogas et al. / Materials and Design 56 (2014) 1039–1048

For BS450, coarse sand was replaced by the lightweight sand (LWS)indicated in Table 2 (type B 0-3). The maximum concrete aggregatesize was 12.5 mm.

LWC with initially dry LWA (PD) or pre-wetted LWA (PW) werealso produced to study their influence on shrinkage. The PD aggre-gate was added during mixing after being dried at 200 �C and thePW aggregate was first wetted for 3 min with 50% of the totalwater, before mixing. Based on the method suggested by Bogaset al. [24], the absorption of LWA in the mix was estimated before-hand to take into account the correction of the total mix water.Modified normal density concretes (MND) were produced withpartial replacement of NA by 35 and 65 percent of type A LWA(A35, A65) or type B LWA (B35, B65). The designations ‘SF’, ‘NS’,‘FA22’ and ‘FA40’ appear when 8% of silica fume, 1.3% of nanosilicaand 22% or 40% of fly ash by weight of cement are used. The NScontent was defined based on previous experimental studies withNWC carried out by the supplier. According to these studies,about 0.15 kg of NS had the same effect on the compressivestrength of normal weight concrete as about 1 kg of silica fume.‘Wsp’ refers to the mixtures without Sp.

2.3. Experimental procedure

For each composition, two prisms of 80 � 80 � 330 mm wereproduced to measure the total shrinkage according to E398 [25].To analyse the influence of the specimen size, two additional spec-imens of 150 � 150 � 600 mm were produced for the referencemixes with Type B and NA aggregate (B/NWC450_PR).

After demolding at 24 h, the specimens were placed in a con-trolled chamber with a temperature of 22 ± 2 �C and a relativehumidity of 50 ± 5%, according to E398 [25]. In addition, two spec-imens of each of the B/NWC 450 reference mixes were previouslycured for 7 days in water before they were placed in the chamber.The total axial shrinkage was monitored by a demountable

Fig. 1. Specimen geometry: prisms of 80 � 80 �

3 7

-900-800-700-600-500-400-300-200-100

0100

ε cs(x

10-6

)

B450A450BS450C450NWC450NWC350

3 702468

10

ΔM

(%

)

Fig. 2. Influence of the type of aggregate on the tota

mechanical strain gauge (DEMEC) with a precision of 1 lm and agauge length of 5 mm. The DEMEC was placed over two steel pins,200 mm apart, which had been glued onto one of the concrete’smoulded surfaces (Fig. 1). Compressive tests were also carriedout on 150 mm water-cured cube specimens at 28 days. The drydensity, qd, the compressive strength, fcm, and the total shrinkage,ecs, at different ages are listed in Table 4.

3. Results and discussion

3.1. Influence of the aggregate type

As expected, the shrinkage of LWC is lower than that of NWC atearlier ages and higher at later ages. The lower initial shrinkagerate of LWC is offset in the long-term (Fig. 2). The shrinkage ofLWC at 30 days was 0.55 (type A), 0.56 (type B) and 0.8 (type C)of that obtained in NWC. But after one year these relationshipswere 1.32 (type A), 1.54 (type B) and 1.66 (type C). This behaviourhas also been documented by other authors (e.g. [1,8,13]); it isessentially related to the water supplied by LWA during the initialages and the lower restriction effect imposed by the less rigid LWAat later ages. The additional partial replacement of natural sand bylightweight sand led to increased shrinkage at all ages, with valuesabout twice as high as for NWC, after one year.

The shrinkage of LWC is lower than that of NWC up to about40 days for LWC with type C LWA and up to about 3 months forLWC with the most common types A and B LWA. Taking into ac-count the initial water percentage of LWA (Table 4) and their par-ticle density (Table 2), the type B aggregate has the highest initialwater content (262 L/m3) followed by types A and C, with about169 L/m3 and 147 L/m3, respectively. The highest water contentof the LWC with type B aggregate is confirmed by the weight loss(DM) curves shown in Fig. 2.

330 mm (left); shrinkage monitoring (right).

days (log scale)

30 60 90 180 365

30 60 90 180 365

days (log scale)

l shrinkage and weight loss up to one year old.

R² = 0.81-800

-600

-400

-200

00.25 0.3 0.35 0.4 0.45

ε cs(x

10-6

)

w/c

Type B (1 year)

Type A (1 year)

Fig. 5. One year shrinkage versus w/c ratio.

J.A. Bogas et al. / Materials and Design 56 (2014) 1039–1048 1043

When the internal curing becomes less relevant (after about thefirst 7–30 days) there is a sudden increase in the shrinkage rate ofLWC compared to NWC. After this transition period, the shrinkageof LWC is proportional to the water lost by evaporation (curveinflection in Fig. 3) and the less rigid type C LWA has the lowestrestriction effect on the paste deformation, which leads to highershrinkage rates (more stepped curve).

In the initial ages, the weight loss, DM (water lost by evapora-tion) in the LWC is not followed by a proportional shrinkage in-crease, confirming the effectiveness of the internal curing. Butthis proportionality is found in the NWC from the first days of dry-ing. Because of the severe drying and the small equivalent thick-ness of the specimens, the shrinkage tends to stabilize afterabout 9–12 months. This stabilization is slower in the LWC dueto the drying delay in early ages.

The difference between LWC and NWC stays almost the same,taking into account the equal strength concrete NWC 350(Fig. 2). In fact, varying the cement content for the same volumeof water has a small effect on shrinkage when the w/c ratio rangesbetween 0.35 and 0.45 [26]. In this case, the reduction of the vol-ume of paste is offset by the increment of the w/c ratio.

3.2. Influence of the cement content

As shown in Fig. 4, if the w/c ratio is increased by reducing theamount of cement and keeping the volume of water, the shrinkageof LWC increases at all ages. This is valid regardless the type ofLWA (Fig. 5), although the LWC with the higher water contentaggregate (type B) is more sensitive to the w/c variation. Afterone year, the increment of the cement content from 350 kg/m3 to525 kg/m3 led to a shrinkage reduction of 44% (LWC with type BLWA) or 28.5% (LWC with type A LWA), Fig. 5. The lower the w/c

ε

Δ

B450A450BS450C450NWC450NWC350

Fig. 3. Total shrinkage ver

3 7

-700-600-500-400-300-200-100

0100

ε cs(x

10-6

)

A350A450A525B350B450B525

3 70

2

4

6

ΔM

(%

)

Fig. 4. Influence of the cement content on the total

ratio the longer the drying delay. Moreover, there is no relevantshrinkage or even a slight expansion for concrete mixes with aw/c ratio of 0.3. In these concrete mixes, the drying is less effectiveand the water movement to the exterior is temporarily halted.

As expected, more weight is lost in concrete mixes with a higherw/c ratio and higher initial water content (Fig. 4). In fact, for a high-er w/c ratio but the same volume of water, the concrete is morepermeable and consumes less hydration water, which leads togreater losses of evaporable water.

3.3. Influence of w/c ratio and superplasticizer

For the same cement content, the shrinkage increases withhigher w/c ratio (Figs. 6 and 7). In fact, there is an increment ofthe volume of paste and a corresponding reduction of the aggre-gate content. Moreover, for the same cement content, the higherthe w/c ratio the lower the mortar stiffness and the higher the vol-ume of evaporable water. A small reduction of the w/c ratio alsomeans a significant delay in drying shrinkage. In this case, varying

1 42 3 5 6

-700

-600

-500

-400

-300

-200

-100

0

ε cst

(x10

-6)

ΔM (%)

A450ANSACZ22B450BSFBNSBFA22BFA40

sus weight loss, DM.

days (log scale)

30 60 90 180 365

30 60 90 180 365

days (log scale)

shrinkage and weight loss up to one year old.

3 7 30 60 90 180 365

-1200-1100-1000-900-800-700-600-500-400-300-200-100

0

ε cs(x

10-6

)

days (log scale)

B450

B450_0.45

B450_0.55

B450_Wsp0.35

B450_Wsp0.5

Fig. 6. Influence of w/c ratio and superplasticizer on the total shrinkage and weight loss up to one year old.

B450Wsp0.5R² = 0.92

-1200-1000-800-600-400-200

00.3 0.35 0.4 0.45 0.5 0.55

ε cs(x

10-6

)

w/c

Type B (1 year)

Fig. 7. One year shrinkage versus w/c ratio, for the same cement content (type BLWA).

VB400

R² = 0.78

-800

-600

-400

-200

0200 250 300 350 400 450

ε cs(x

10-6

)

Aggregate volume (L/m3)

Type B (1 year)

Type A (1 year)

Fig. 9. One year shrinkage versus the volume of aggregate.

1044 J.A. Bogas et al. / Materials and Design 56 (2014) 1039–1048

the w/c ratio from 0.55 to 0.35 led to a shrinkage reduction of77.3% at 30 days and 48% at one year (Fig. 7).

The magnitude and progression of the shrinkage were quitesimilar in the dry mixes without Sp (B450_Wsp0.35) and in thefluid mixes of the same composition with Sp (B450), Fig. 6.

The main effect of superplasticizers should be related to howthey can change the water content and the w/c ratio of concrete[27,28]. To better study this effect, a concrete without Sp but withhigher water content (B450_Wsp0.5) was produced to have thesame slump as the reference mix B450. In this case, the introduc-tion of Sp leads to a reduction of the shrinkage by about 39.5% afterone year (Table 4). Fig. 7 shows that B450_Wsp0.5 follows thesame trend as the other concrete mixes with different w/c ratios,regardless the use of Sp.

3.4. Influence of the volume of aggregate

Considering concrete mixes with different volumes of aggre-gates but with the same w/c ratio and cement content. On onehand, since the replacement of sand by lightweight coarse aggre-gate implies a reduction of the mortar stiffness, increasedshrinkage would be natural. But on the other hand, the

3 7

-700

-600

-500

-400

-300

-200

-100

0

ε cs(x

10-6

) A450

VA425

VB400

B450

VmB425

VmB250

3 70

2

4

6

ΔM

(%

)

Fig. 8. Influence of the volume of aggregate on the to

higher the volume of LWA the greater the efficiency of the internalcuring.

The increment of the volume of type A aggregate from 350 L/m3

(A450) to 425 L/m3 (VA425) led to a shrinkage reduction at all ages,although the long-term shrinkage was about the same (Fig. 8).However, varying the volume of type B aggregate from 350 L/m3

(B450) to 400 L/m3 (VB400) had a small influence on the short-term and long-term shrinkage.

Therefore, there is a greater influence of the volume of aggre-gate on the shrinkage of LWC with type A aggregate than with typeB LWA. On the one hand, since the stiffness of the type A aggregateis higher, its replacement by natural sand has less influence on therestriction effect on the free deformation of the paste. On the otherhand, the greater amount of particles and the smaller maximumaggregate size of type A LWA when compared to type B LWA(Table 2), may have induced more effective internal curing.

Reducing the volume of aggregate without changing the mortarcharacteristics implies a proportional increase of the volume ofsand and paste, with a corresponding increase of the shrinkage(Fig. 9). In fact, the reduction of the LWA volume leads to a simul-taneous increase of the paste content and a reduction of the vol-ume of water available for internal curing. The inclusion of

days (log scale)

30 60 90 180 365

30 60 90 180 365

days (log scale)

tal shrinkage and weight loss up to one year old.

3 7

-700-600-500-400-300-200-100

0

ε cs(x

10-6

)

days (log scale)A450ANSAFA22B450BSFBNSBFA22BFA40

3 7

30 60 90 180 365

30 60 90 180 3650

2

4

6Δ

M (

%)

days (log scale)

Fig. 10. Influence of mineral admixtures on the total shrinkage and weight loss up to one year old.

J.A. Bogas et al. / Materials and Design 56 (2014) 1039–1048 1045

concrete mix VB400 in Fig. 9 shows that the shrinkage of LWCdepends on how the volume of LWA varies.

3.5. Concrete with admixtures

The replacement of 8% of cement by silica fume had littleinfluence in terms of both shrinkage and weight loss (Fig. 10).Similar findings were reported by Hooton [29] for normal con-crete. Note that compressive strength was also little affected(Table 4). The absence of initial water curing may have inhibitedthe regular development of the pozzolanic reactions. It is alsolikely that there was no effective dispersion of silica fume inconcrete.

There is a long-term reduction of the shrinkage in LWC withnanosilica, especially in LWC with type A aggregate. After oneyear, the reduction was 19.7% in LWC with type A LWA and 5%in LWC with type B LWA (Fig. 10). The addition of nanosilicaled to a slight reduction of the volume of paste. Moreover, theremay have been a refinement of the matrix porosity with a conse-quent increase of the mortar stiffness. In addition to the fillereffect, the high reactivity of nanosilica means it reacts at earlyages when there is more water available within the concrete. Inthis case, the migration of water from the LWA to the pastemay be relevant. Because of the nano-size of the NS, part of thisadmixture may be absorbed by the more porous LWA (type B),thus reducing its effectiveness.

0 22 40

-800

-600

-400

-200

0

ε cs(x

10-6

)

%FA

Type B (1 year)Type A (1 year)

Fig. 11. One year shrinkage versus the percentage replacement of fly ash.

3 7

-600

-500

-400

-300

-200

-100

0

ε cs(x

10-6

) A450A35%A65%B450B35%B65%NWC450

Fig. 12. Influence of the total or partial replacement of normal weight c

Shrinkage increases at all ages in the LWC with 22% replace-ment of cement by fly ash, regardless the type of LWA. However,the difference is more relevant in the LWC with more porousLWA (type B). The shrinkage increase in LWC with fly ash is alsoreported by Holm [8] and Haque [30], although the opposite wasreported by Malhotra [16]. Since the hydration reactions are slowerin fly ash mixtures, their porous structure tends to be coarser, atleast in the early ages. Therefore, the drying rate increases andthe restriction on the paste deformation decreases (Fig. 10). Onthe other hand, since the hydration reactions occur later, there isless compensation for the water stored in the LWA and meanwhileeliminated by evaporation. The absence of initial water curing andthe increasing volume of paste also affect the results.

Interestingly, although the shrinkage is lower in the LWC withtype A LWA, the water lost is slightly lower in the LWC with typeB LWA. Moreover, the long-term shrinkage of the LWC with 40% offly ash is lower than that of the LWC produced without it (Fig. 11).However, the water loss is higher in the LWC with 40% of fly ash,especially in the first 6 months. This behaviour can be related tothe differences in the level of the paste hydration.

Since the mixes with a higher volume of fly ash have a lower de-gree of hydration, there are more non-hydrated particles that workas non-shrinkage ‘microaggregates’. Therefore, the volume of par-ticles that act as a source of contraction is lower. Termkhajornkitet al. [31] have confirmed that the shrinkage in NWC is affectedby the extent to which fly ash hydrates, which is lower for higheramounts. In part, this phenomenon may explain the difference ob-served in how shrinkage evolves in BFA22 and BFA40, after 30 days(Fig. 10). In addition, the lower hydration level of the paste justifiesthe higher drying rate (Fig. 10) and the lower effect of the waterlost in the shrinkage of BFA40 (Fig. 3). The behaviour of the LWCwith type A LWA and 22% of fly ash is between that found forBFA22 and BFA40. This is probably because there is less wateravailable in type A LWA than in type B LWA, which means lowerlevels of hydration in the LWC with type A LWA.

30 60 90 180 365

days (log scale)

oarse aggregate by LWA on the total shrinkage up to one year old.

0 35 65 100

R² = 0.73

-600

-400

-200

0ε c

s(x

10-6

)

VLWA (L/m3)

Type B (1 year)

Type A (1 year)

Fig. 13. One year shrinkage versus the percentage replacement of LWA.

1046 J.A. Bogas et al. / Materials and Design 56 (2014) 1039–1048

3.6. Total or partial replacement of NA by LWA

Figs. 12 and 13 show that the higher the percentage replace-ment of NA by LWA the higher the long-term shrinkage. But thehigher the percentage replacement the lower the initial shrinkage.On the one hand, the greater the percentage replacement of NA byLWA the higher the water available to the paste at early ages. Onthe other hand, the replacement of NA by the less rigid LWA re-duces the restriction effect on the paste deformation. FromFig. 12, it can be concluded that 35% replacement of type B LWAand 65% replacement of type A LWA are enough for the water lostby evaporation to be practically offset in the first 7 days.

3.7. Influence of the initial water content of LWA

As expected, there was a greater delay of the drying shrinkageand a reduction of the initial shrinkage in the LWC with pre-satu-rated LWA (about 53.5% lower in the LWC with type A LWA and58.8% in the LWC with type B LWA, Fig. 14). However, the pre-sat-uration of the LWA had little influence on the long term-shrinkage.After one year, the shrinkage of the LWC with pre-dried aggregatesis similar to that of LWC with pre-saturated type A LWA and 6.6%lower than that of pre-saturated type B LWA. The slightly lowershrinkage of LWC with pre-dried aggregates can be related to theimprovement of the aggregate-paste transition zone that promotesa more effective restriction effect by the aggregates. By comparingthe LWC compressive strength for different initial wetting condi-tions of LWA, it is confirmed that the mortar characteristics shouldbe quite similar (Table 4).

3 7

-700-600-500-400-300-200-100

0

ε cs(x

10-6

)

A450A450PMA450PDB450B450PMB450PD

Fig. 14. Influence of the initial water content of LW

3 7

-600

-500

-400

-300

-200

-100

0

100

ε cs(x

10-6

)

B450

B450_7D

B450_PR

NWC450

NWC450_7D

NWC450_PR

Fig. 15. Influence of the specimen size on

3.8. Influence of the curing conditions and specimen size

There was a delay and a reduction of the long-term shrinkage,both in LWC (16.5%) and in NWC (12.1%) when the concrete waspreviously cured for 7 days in water (B/NWC450_7D, Fig. 15). Forthese curing conditions, the period in which the shrinkage ofLWC is lower than that of NWC was extended from about 3 to5 months.

As shown in Fig. 15, the shrinkage evolution is significantly af-fected by the specimen’s geometry. The higher the specimen thick-ness the lower the drying shrinkage, especially in the centre of theconcrete, which behaves as if it was isolated from the exterior.However, the difference in the shrinkage evolution is more rele-vant in LWC than in NWC. Contrary to what happens in NWC,the self-desiccation in LWC can be offset by the internal curing.Therefore, the LWC suffers a slight expansion and the dryingshrinkage is only effective after about 3 months (Fig. 15).

In this study, the one year shrinkage of LWC was 34% lower or54% higher than that of NWC, depending on the size of the speci-men. These differences may explain the apparently contradictoryresults reported in the literature. Moreover, the results show thatunder real conditions LWC can be quite slow to dry and only thelong-term shrinkage should be more meaningful.

4. Normative analysis

The shrinkage curves obtained for LWC with type A and type BLWA are presented in Figs. 16 and 17. These graphs cover w/c ratiosbetween 0.3 and 0.55 (type B LWA) and between 0.3 and 0.45 (typeA LWA). Figs. 16 and 17 also show the estimated shrinkage curvesaccording to ACI 209R [18], EN1992-1 [19] and MC 2010 [20].

To build the normative curves shown in Figs. 16 and 17, the fol-lowing were assumed (Table 1):

� relative humidity of 50% (cHR = 0.89 [18], bRH = 1.36[19,20]);

� one curing day without moisture movement to the exterior(ctc = 1.2 [18]);

� specimen section of 80 � 80 � 330 mm (cvs = 1.1 [18]; Kh = 1[19]), where v/s = 18 mm (equivalent thickness) andh0 = 2Ac/u = 40 mm, with Ac = 80 � 80 mm2 (cross-sectionalarea) and u = 80 � 4 mm2 (cross-sectional perimeter);

30 60 90 180 365

days (log scale)

A on the total shrinkage up to one year old.

30 60 90 180 365

days (log scale)

the total shrinkage up to one year old.

3 7 30 60 90 180 365

-1200-1100-1000-900-800-700-600-500-400-300-200-100

0100

ε cs(x

10-6

)

days (log scale)

MaxTypeB (w/c=0.45)MaxTypeBMinTypeBACI 209REN1992 (min)EN1992 (max)MC2010 (min)MC2010 (max)

Fig. 16. Shrinkage curves and normative estimations for LWC with type B aggregate (w/c of 0.3–0.55).

3 7 30 60 90 180 365

-900-800-700-600-500-400-300-200-100

0100

ε cs(x

10-6

)

days (log scale)MaxTypeAMinTypeAACI 209REN1992 (min)EN1992 (max)MC2010 (min)MC2010 (max)

Fig. 17. Shrinkage curves and normative estimations for LWC with type A aggregate (w/c of 0.3–0.45).

VmB425

BFA40

-1200

-1000

-800

-600

-400

-200

030 35 40 45 50 55 60 65 70 75 80

ε cs,

1yea

r(x

10-6

m/m

)fcm (MPa)

Type AType BMND (Type A)MND (Type B)

Fig. 18. One year shrinkage versus the compressive strength.

J.A. Bogas et al. / Materials and Design 56 (2014) 1039–1048 1047

� f = 35 was assumed for the expression of ACI209R (1992);� percentage of fine aggregate (cw = 1.0 (ACI209R 1992) – 30

to 35% of sand);� rapidly hardening high early strength cement (ads1 = 6;

ads2 = 0.11 [19] or 0.12 [20]);� lightweight concrete (a multiplier factor g = g3 = 1.0, was

always considered, Table 1);� fcm varies between 36 and 50 MPa for LWC with type B LWA

and w/c ratio between 0.3 and 0.55;� fcm varies between 58 and 69 MPa for LWC with type A LWA

and w/c ratio between 0.3 and 0.45;� for EN1992-1 [19] and MC 2010 [20] it is assumed that the

LWC autogenous shrinkage is fully offset

In general, the expressions in the standards tend to overesti-mate the shrinkage of LWC with a w/c ratio up to 0.45, especiallyat early ages. These expressions are not suitable for lightweightconcrete, since they do not take into account the shrinkage delaycaused by the internal curing.

Of the methods analysed, the one suggested by ACI 209R [18]led to the best fitted shrinkage curves at early ages. However, theshrinkage estimation at later ages may be too conservative, partic-ularly for LWC with a low w/c ratio or with LWA of high initialwater content. It is noteworthy that this report does not, for prac-tical purposes, take into account the concrete composition and thewater content of the aggregates. The method is empirically basedand less reasonable for moderate to high-strength LWC.

The long-term shrinkage estimates from the expressions sug-gested by EN1992-1 [19] and MC 2010 [20] are within the experi-mental results obtained for low w/c lightweight concrete,especially after a long drying period. However, the expressionswere totally inadequate for early ages. Moreover, the shrinkageof common LWC with higher w/c ratio can be greatly underesti-mated in these recommendations. In fact, the long-term shrinkageof LWC with w/c ratio above 0.45 is underestimated by more than50% (Fig. 16). However, since multiplier factors for LWC are pro-vided in EN1992-1 [19] and MC 2010 [20] these differences canbe smaller. Comparing just the one year shrinkage of LWC withthat of NWC with the same composition (Table 4), multiplier

coefficients of about 1.3 for higher density LWA (type A) and about1.6 for the more porous LWA (type C) are suggested.

The expressions suggested in EN1992-1 [19] and MC 2010 [20]take the concrete composition into account indirectly through thecompressive strength. However, shrinkage does not depend on thecompressive strength but on the parameters related to the micro-structure and composition of concrete, such as the volume of paste,w/c ratio, degree of hydration and aggregate properties. Therefore,there is a weak correlation between shrinkage and compressivestrength, as shown in Fig. 18.

5. Conclusions

The influence of different compositional parameters on theshrinkage of lightweight concrete was analysed by means of anextensive experimental programme. The internal curing and thelower stiffness of LWA are responsible for the lower initial shrink-age and the higher long term shrinkage of the LWC. Depending onthe type of LWA, the one year shrinkage of LWC was about 30–65%higher than that of NWC and about twice as high as that of NWCwhen additionally there was a replacement of natural sand bylightweight fine aggregates. The partial replacement of coarse NAby LWA led to intermediate shrinkage curves relative to those ob-tained for concrete produced with only NA or LWA.

1048 J.A. Bogas et al. / Materials and Design 56 (2014) 1039–1048

Keeping the effective volume of water, the shrinkage increase ofLWC with the w/c ratio is more effective the higher the initialwater content of LWA. Concrete shrinkage is affected by the intro-duction of superplasticizer, but only when this implies a reductionin the volume of mixing water.

The shrinkage of LWC depends on how the volume of LWA var-ies. Keeping the same volume of paste and increasing the volumeof aggregate, the shrinkage of LWC can either decrease or be aboutthe same, depending on the type of LWA and its water content. Forthe same mortar characteristics there is a proportional increase ofshrinkage the LWA volume is reduced.

The addition of 8% of silica fume had little influence on shrink-age at all ages. However, the long-term shrinkage of LWC with themore reactive nanosilica tends to be lower. The shrinkage of LWCwith fly ash strongly depends on the percentage of replacementand type of LWA. The shrinkage increases at all ages for 22% offly ash replacement, and is more relevant in LWC with more porousLWA. However, for 40% of fly ash replacement the shrinkage is low-er than that of concrete without fly ash.

The pre-saturation of LWA reduces the early shrinkage and de-lays the drying shrinkage. However, the initial wetting condition ofLWA had little influence on the long-term shrinkage. It was alsoshown that the shrinkage can be strongly affected by the geometryof the specimens, a phenomenon that is more relevant in LWC thanin NWC. This leads to a greater susceptibility of LWC to differentialshrinkage.

The current methods suggested in the main standards andcodes cannot properly predict the shrinkage behaviour of LWC. Inparticular, the suggested methods do not take into account the typ-ical delayed shrinkage of LWC. It has also been shown that there isa weak correlation between the shrinkage and the compressivestrength in LWC, contrary to the philosophy adopted in EN1992-1 [19] and MC 2010 [20]. It is necessary to find new solutions thatcan better take the influence of the w/c ratio on shrinkage intoaccount.

A rough but simple procedure to take into account the shrink-age delay in LWC can be to consider a fictitious higher curing per-iod in the codes’ recommendations. Based on the experimentalresults, multiplier coefficients of about 1.3 for the higher densityLWA and about 1.6 for the more porous LWA are suggested as away of taking into account the higher long-term shrinkage of light-weight concrete.

Acknowledgements

The authors wish to thank ICIST-IST for funding the researchand the companies Argex, Saint-Gobain Weber Portugal, Soarvamil,BASF and SECIL for supplying the materials used in the experi-ments. The research work presented herein was supported by thePortuguese Foundation for Science and Technology (FCT), undergrant PTDC/ECM-COM1734/2012.

References

[1] Zhang M-H, Li L, Paramasivam P. Shrinkage of high-strength lightweightaggregate concrete exposed to dry environment. ACI Mater J2005;102(10):86–92.

[2] Kayali O, Haque MN, Zhu B. Drying shrinkage of fibre-reinforced lightweightaggregate concrete containing fly ash. Cem Concr Res 1999;29(11):1835–40.

[3] Virlogeux M. Généralités sur les caractères des bétons légers. In: Granulats etbetons legers-Bilan de dix ans de recherches, Arnould et Virlogeux, Presses del’école nationale des ponts et chausses; 1986, p. 111–246.

[4] Bogas JA. Characterization of structural lightweight expanded clay aggregateconcrete. Lisbon: PhD thesis in civil engineering. Technical University ofLisbon, Instituto Superior Técnico; 2011 [in Portuguese].

[5] Holm, TA, Bremner TW. State-of-the-art report on high-strength, high-durability structural low-density concrete for applications in severe marineenvironments. Us Army corps of engineers. Structural Laboratory, ERDC/SL TR-00-3; 2000. 104 p.

[6] Selih J, Bremner TW. Drying of saturated lightweight concrete: anexperimental investigation. Mater Struct 1996;29(7):401–5.

[7] Coquillat G. Influence des caractéristiques physiques et mécaniques desgranulats légers sur les propriétés des bétons legers de structure. In:Granulats et betons legers-Bilan de dix ans de recherches, Arnould etVirlogeux, Presses de l’école nationale des ponts et chausses; 1986, p. 255–98.

[8] Holm TA. Performance of structural lightweight concrete in a marineenvironment. In: International symposium ACI SP-65. St. Andrews By-The-Sea, Canada; 1980. 12p.

[9] Hossain KMA, Lachemi M. Mixture design, strength, durability, and fireresistance of lightweight pumice concrete. ACI Mater J 2007;104(5):449–57.

[10] Nobuta Y, Satoh K, Hara M, Sogoh S, Takimoto K. Applicability of newlydeveloped high-strength lightweight concrete for civil structures. In: S.Helland et al., editors, Second international symposium on structurallightweight aggregate concrete, Kristiansand, Norway, 18–22 June; 2000. p.396–405.

[11] Wegen GJL, Bijen JMJM. Properties of concrete made with three types ofartificial PFA coarse aggregates. Int J Cem Compos Lightweight Concr1985;7(3):159–67.

[12] Nilsen AU, Aitcin PC. Properties of high-strenght concrete containing light,normal, and heavy weight aggregate. Cem Concr Aggre 1992;14(1):8–12.

[13] FIP. FIP manual of Lightweight aggregate concrete. Fédération internationalede la précontrainte (FIP). 2nd ed., Surrey University Press; 1983.

[14] EuroLightCon R31. Long-term effects in LWAC: Strength under sustainedloading shrinkage of high strength LWAC. Eur. Uni. Brite EuRam III, BE96-3942/R31; 2000.

[15] Hoff GC. High strength lightweight aggregate concrete for artic applications –Part1,2,3. In: Holm, Vaysburd, editors, Structural lightweight aggregateconcrete performance, ACI SP-136; 1992, p. 1–245.

[16] Malhotra VM. Properties of high-strength lightweight concrete incorporatingfly ash and silica fume. In: Hester, editors, High-strength concrete, secondinternational symposium, SP-121. Michigan, American Concrete Institute;1990, p. 645–66.

[17] Zhutovsky S, Kovler K, Bentur A. Influence of cement paste matrix propertieson the autogenous curing of high-performance concrete. Cem Concr Compos2004;26(5):499–507.

[18] ACI 209.2R-08. Guide for modeling and calculating shrinkage and creep inhardened concrete. American Concrete Institute; 2008.

[19] EN 1992-1-1. Eurocode 2: Design of concrete structures – Part 1–1: Generalrules and rules for buildings. European Committee for standardization CEN;2010.

[20] MC 2010. Model Code (MC) prepared by fib special Activity Group 5 – Finaldraft; 2011.

[21] Bogas JA, Mauricio A, Pereira MFC. Microstructural analysis of Iberianexpanded clay aggregates. Microsc Microanal 2012;18:1190–208.

[22] Bogas JA, Gomes A. A simple mix design method for structural lightweightaggregate. Mater Struct 2013;46(11):1919–32.

[23] Bogas JA, Gomes A. Compressive behavior and failure modes of structurallightweight aggregate concrete – characterization and strength prediction.Mater Des 2013;46:832–41.

[24] Bogas JA, Gomes A, Gloria MG. Estimation of water absorbed by expanding clayaggregates during structural lightweight concrete production. Mater Struct2012;45:1565–76.

[25] E398. Concrete – Determination of shrinkage and expansion. LNEC, Lisboa;1993 [in Portuguese].

[26] Blanks RF, Vidal EN, Price WH, Russell FM. The properties of concrete mixtures.ACI J 1940;36:433–76.

[27] Collins TM. Proportion high-strength concrete to control creep and shrinkage.ACI Mater J 1989;86(55):576–80.

[28] Rixon R, Mailvaganam N. Chemical admixtures for concrete. 3rd ed. E&FNSPON; 1999.

[29] Hooton D. Influence of silica fume replacement of cement on physicalproperties and resistance to sulphate attack, freezing and thawing, andalkali-silica reactivity. ACI Mater J 1993;90(15):143–51.

[30] Haque MN. Strength development and drying shrinkage of high-strengthconcretes. Cem Concr Compos 1996;18(5):333–42.

[31] Termkhajornkit P, Nawa T, Nakai M, Saito T. Effect of fly ash on autogenousshrinkage. Cem Concr Res 2005;35(3):473–82.