Sunteți pe pagina 1din 10

Materials and Design 56 (2014) 10391048

Contents lists available at ScienceDirect

Materials and Design


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

Inuence of mineral additions and different compositional parameters


on the shrinkage of structural expanded clay lightweight concrete
J. Alexandre Bogas , Rita Nogueira, Nuno G. Almeida
DECivil/ICIST, Instituto Superior Tcnico, 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 2013
Accepted 5 December 2013
Available online 13 December 2013
Keywords:
Lightweight aggregate concrete
Shrinkage
Fly ash
Silica fume
Nanosilica

a b s t r a c t
A comprehensive experimental study was carried out on the shrinkage behaviour of structural expanded
clay lightweight aggregate concrete (LWC), taking into account different compositions, types and initial
wetting conditions of lightweight aggregates (LWA). The inuence of different compositional parameters
on shrinkage, such as the type and volume of aggregate, the w/c ratio, the binder content and the partial
replacement of normal aggregates by LWA was analysed. The shrinkage of LWC depends on how the volume of LWA varies. The inuence of different pozzolanic materials was also studied, namely, silica fume,
nanosilica and y ash. Depending on the type, content and reactivity of the pozzolanic additions, the
shrinkage was higher or lower than that of LWC without admixtures. The initial wetting condition of
LWA had little inuence on the long-term shrinkage. The LWC with the most porous aggregates is more
affected 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 coefcients of about 1.3 for the most common structural LWA and about 1.6 for more porous LWA are suggested, 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 essentially
governed by the paste, which is the source of contraction of the
system, and by the aggregates that oppose this contraction. Since
structural lightweight aggregate concrete (LWC) is modied at
these two factors, it is expected that its behaviour differs from that
observed in normal density concrete (NWC).
On the one hand, the use of less rigid porous aggregates decreases the restriction effect on paste deformation [13]. On the
other hand, either for strength purposes or for reasons of workability and stability of the mixes, LWC is usually characterized by larger volumes of better quality paste and lower volumes of coarse
aggregates than NWC [4,5]. Therefore, the long-term shrinkage of
LWC should be higher than that of NWC of the same strength.
However, the paste quality is higher in LWC, which increases the
matrix stiffness. Moreover, the water absorbed by lightweight
aggregates (LWA) is later released into the paste by internal curing,
which compensates for the initial water lost by drying and selfdesiccation.
According to Kayali et al. [2], the improved aggregate-paste
interface in LWC helps to enhance the restriction effect of the
LWA. Moreover, because of the internal curing and the prolonged
Corresponding author. Tel.: +351 218418226; fax: +351 218418380.
E-mail address: abogas@civil.ist.utl.pt (J.A. Bogas).
0261-3069/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.matdes.2013.12.013

hydration of the paste the deformation resistance of the matrix is


higher and less water is available for evaporation [6]. The combination of all these factors, together with the high variability of the
different types of LWA, explains why some authors report higher
shrinkage in LWC (e.g. [1,79]), while others report shrinkage in
LWC that is lower than or similar to shrinkage in NWC (e.g.
[1,1014]).
Most studies have looked at the global shrinkage of LWC of a given composition and type of aggregate. However, thus far, only a
few studies have been published on the detailed analysis of the
inuence of different parameters on LWC shrinkage, such as the
volume and type of LWA, w/c ratio and the amount and type of binder, especially the use of different admixtures.
Hoff [15] and Holm [8] reported higher shrinkage in watercured LWC exposed to 50% RH, produced with y ash, than in concrete without this admixture. However, Malhotra [16] has found
lower one year shrinkage in high-strength LWC with y ash.
According to Zhang et al. [1] the addition of 5% of silica fume can
signicantly reduce concrete shrinkage, an effect that is more relevant in LWC than in NWC. Lower shrinkage in LWC with silica
fume is also reported by Hoff [15]. However, according to Zhutovsky et al. [17] the addition of silica fume can lead to a signicant
reduction of internal curing in LWC produced with pumice LWA.
In the main standard documents, such as ACI 209R [18], EN
1992 [19] and MC 2010 [20], the shrinkage of LWC is usually
roughly estimated from the expressions dened for normal density

1040

J.A. Bogas et al. / Materials and Design 56 (2014) 10391048

concrete (Table 1). Regarding EN 1992-1 [19], the nal drying


shrinkage of LWC is multiplied by an empirical factor of 1.2. This
factor is also suggested by MC 2010 [20], but with respect to the
total shrinkage. Standard EN 1992 (2010) assumes that the autogenous shrinkage of LWC can be much smaller than that of NWC if
the LWA is pre-soaked, but no suggestion is given for its estimation. The method suggested in ACI 209R [18] was calibrated from
experimental results obtained in NWC and LWC of low to moderate
strength, under controlled exposure conditions. The expressions
are empirically based and do not model shrinkage phenomena.
This document does not provide any coefcient to penalize the
long-term shrinkage of LWC.
The main purpose of our work was to study the exact inuence
on LWC shrinkage of different compositional parameters such as
the w/c ratio, type and volume of aggregate, type and content of
binder, use of additives and the partial replacement of normal
aggregates by LWA. This is analysed for a wide range of common
structural lightweight concretes with about 3070 MPa and density classes from D1.6 to D2.0. The accuracy of the main standard
expressions 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 total
porosity, PT, particle density, qp, bulk density, qb, and 24 h water
absorption, wabs,24 h, are indicated in Table 2. A more detailed
microstructural characterisation of these aggregates is presented
elsewhere [4,21]. In terms of their specic properties, the selected
LWA 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 ne aggregates (NA) were also used. For
the reference normal density concrete, two crushed limestone
aggregates of different sizes were combined so as to have the same
grading curve as Leca (20% ne and 80% coarse gravel). Fine aggregates consisted of 2/3 coarse and 1/3 ne sand. Their main

properties are listed in Table 2. The two fractions of type C LWA


were also combined to have the same grading curve as type B
LWA (35% 24 and 65% 38F, Table 2). Fly ash (FA) from the Pego
thermoelectric 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 materials are listed in Table 3. A water dispersed RHEOMAC VMA 350
nanosilica (NS) was also used. Basically, this admixture is a dispersion of non-agglomerated spherical nanoparticles of pure silica
fume with an average density of 1.1 and about 16.1% solids
content.
2.2. Concrete mixing and mixture proportions
The concretes were produced in a vertical shaft mixer with bottom discharge. Except for initially dry or pre-wetted aggregates,
the LWA was pre-soaked for 24 h to better control the workability
and effective water content of the concrete. The aggregates were
then surface dried with absorbent towels and placed in the mixer
with sand and 50% of the total water. After 2 min of mixing, the
cementitious materials and 40% of the water were added. The Sp
was added slowly with 10% of water. The total mixing time was
7 min. When used, NS was added with about 20% of water after
6 min of mixing and mixing then continued for a further 4 min.
Thirty-two different compositions were designed according to
Bogas and Gomes [22,23], in order to take into account: different
binder volumes and w/c ratios; volume, initial water content and
type of aggregate; addition of silica fume, nanosilica or y ash;
use of Sp; partial replacement of coarse or ne aggregates by
LWA. All concrete compositions are listed in Table 4. The water/
binder ratio (w/b) relates to the effective water available for cement hydration. The Sp/c is the percentage of superplasticizer by
cement weight. The denominations NWC, A, B and C represent
the mixes with NA and type A, type B and type C LWA. The prex
V refers to different volumes of aggregates for constant amounts
of water and cement (variable coarse aggregate/sand), and the prex Vm to different volumes of aggregate for constant composition
of the mortar (same cement/aggregate ratio). Except for mixture
BS450, natural sand was used in combination with coarse LWA.

Table 1
Shrinkage estimation main standard documents.
Standard

Total shrinkage estimation

Observations

ACI 209R [18]


(NWC/
LWC)

esh,t = esh,u(t  ts)/(f + t  ts)


f = 35 for concrete water-cured during 7 days
esh,u = 780csh  106
csh = ctccRHcvscscwccca
Sugestion: f = 26exp[1.42  102(v/s)] to take into account
the size and geometry on the drying of concrete elements

csh Represents the product of several factors that take into account the curing time

EN1992 [19]
(NWC)

ecs (t, ts) = ecd + eca


ecd (t) = Kh.ecd,0.bds(t,ts)
ecd,0 = 0.85bRH[(220 + 110ads1)exp(ads2fcm/10)]106

Kh, bRH, ads1 e ads2 Are coefcients depending on the notional size (h0), the relative
humidity (RH) and the type of cement (ads1;ads2)

(tc), the relative humidity (RH), the cross-section geometry (vs), the slump (s), the
percentage of ne aggregate (w) and the air content (a)

3=2

(LWC)
MC 2010 [20]
(NWC)

bds t  t s t  t s =t  ts 0:04  h0 
eca(t) = [1exp(0.2t0.5)]eca0
eca0 = 2.5(fck10)106
ecs (t, ts) = g3ecd + eca
ecs (t, ts) = ecd + eca
ecd (t) = ecd,0bRHbds (t, ts)
ecd,0 = [(220 + 110ads1)exp(ads2fcm/10)]106

g3 = 1.2 for fck > LC20/22


bRH, aas, ads1 e ads2 Are coefcients depending on the relative humidity (RH) and the
type of cement (aas; ads1; ads2)

2 0:5

(LWC)

bds t  t s t  ts =t  t s 0:035  h0 
eca(t) = [1exp(0.2t0.5)]eca0
eca0 = aas[(fcm/10)/(6+fcm/10)]2.5  106
ecs (t, ts)LWC = gecs (t, ts)NWC

g = 1.2 for fck > LC20/22

esh,t shrinkage strain after a period of time t; esh,u ultimate shrinkage strain; (v/s) volume/area ratio of the concrete element.
ecs (t, ts) shrinkage strain between the age t and the begining of drying ts; ecs0 ultimate shrinkage strain.
ecs total shrinkage strain; ecd drying shrinkage strain; eca autogenous shrinkage strain.
fcm or fck average or characteristic compressive strength at 28 days.

1041

J.A. Bogas et al. / Materials and Design 56 (2014) 10391048


Table 2
Aggregate properties.
Property

Normal weight aggregates


Fine sand

Particle dry density, qp (kg/m3)


Loose bulk density, qb (kg/m3)
24 h Water absorption, wabs,24 h (%)
Total porosity, PT (%)
Granulometric fraction (di/Di)
Los Angeles coefcient (%)

2620
1416
0.2

0/2

Coarse sand

2610
1530
0.5

0/4

Lightweight aggregates
Fine gravel

Coarse gravel

2631
1343
1.4

4/6.3
33.3

2612
1377
1.1

6.3/12.5
30.5

Type A

Type B

AF7

03

412

24

Type C
38F

1290
738
12.1
52
3/10

1060
562

59
0.5/3

1068
613
12.3
60
4/11.2

865
423
22.9
67
4/8

705
397
23.3
73
6.3/12.5

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

Standard

Fly ash

Silica fume

Cement I 52.5 R

Residue on the 45 lm sieve, (%)


Blaine specic surface, (cm2/g)
Compressive strength of reference mortar, (MPa)

EN 451-2
EN 196-6
EN 196-1

10.2

83.7b
103.1b
0.5a
6.5
83.0
3.38
0.36
2.33

92.0a

106.7c

3.7
94.0
0.83
not detected
2.25

1.1
5102
40.4
62.7

0.5
1.64
29.1
61.6
1.45
3.11

2 days
28 days

Activity index at 28 days, (%)


Activity index at 90 days, (%)
Expansion, (mm)
Loss on ignition (LOI), (%)
SiO2 + Al2O3 + Fe2O3, (%)
CaO, (%)
Free CaO, (%)
Density, (g/cm3)
a
b
c

EN
EN
EN
EN
EN

EN
EN

196-1
196-1
196-3
196-7
196-2
451-1
196-6

Residue on the 90 lm sieve.


Mortar with CEM I42,5R + 25% Fly ash.
Mortar with CEM I42,5R + 10% silica fume.

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

a
b

ecs (106 m/m)


Age (days)b

Mixes

Coarse
aggreg.
(m3/m3)

Cement
(kg/m3)

Addition
(kg/m3)

wag
(%)a

Sp/c
(%)

Effective
w/b (L/m3)

Oven dry
density, qd
(kg/m3)

fcm,28d
(MPa)

30

90

A350
A450
A450PM
A450PD
A525
VA425
ANS
AFA22
A35 (35%LWA)
A65 (65%LWA)

0.35
0.35
0.35
0.35
0.35
0.425
0.35
0.35
0.35
0.35

350
450
450
450
525
450
414
350
450
450

5.4
100

13.4
13.1
0
0
13.4
13.4
14.0
14.0
16.0
16.0

0.7
0.7
0.7
0.7
0.8
0.7
1.2
0.6
0.7
0.7

0.45
0.35
0.30
0.30
0.30
0.35
0.35
0.35
0.35
0.35

1789
1840
1839
1846
1872
1764
1838
1798
2150
2008

57.6
65.8
63.5
65.1
68.5
57.9
65.5
60.0
72.3
66.5

41
14
34
36
14
31
28
9
36
7

73
36
111
79
10
1
77
50
97
63

231
130
350
278
51
63
176
238
299
245

498
292
502
451
180
258
326
424
389
410

563
490
603
530
360
408
403
510
454
490

640
508
589
509
458
476
408
534
466
502

B350
B450_0.55
B450_0.45
B450
B450PM
B450PD
B525
B450_Wsp0.35
B450_Wsp0.5
VB400
VmB250
VmB425
BSF
BNS
BFA22
BFA40
BS450
B35 (35%LWA)
B65 (65%LWA)

0.35
0.35
0.35
0.35
0.35
0.35
0.35
0.35
0.35
0.4
0.25
0.425
0.35
0.35
0.35
0.35
0.35
0.35
0.35

350
450
450
450
450
450
525
450
450
450
522
414
414
350
270
450
450
450

36
5.4
100
180

23.4
19.5
19.5
24.5
0
0
24.0
20.0
20.0
24.0
23.2
23.9
23.4
24.0
24.0
24.0
24.0
22.2
22.2

0.6
0.0
0.2
0.7
0.7
0.7
0.8
1.0
0.0
0.6
0.6
0.7
0.9
1.0
0.7
0.7
0.6
0.6
0.6

0.45
0.55
0.45
0.35
0.30
0.30
0.30
0.30
0.50
0.30
0.35
0.35
0.35
0.35
0.35
0.35
0.35
0.35
0.35

1712
1540
1676
1740
1744
1753
1770

1678
1876
1675
1741
1754
1703
1668
1435
2112
1950

43.1
36.1
41.9
48.6
46.5
46.5
50.0
43.3
39.0
45.7
43.4
38.6
47.6
46.7
42.4
37.1
37.5
59.8
53.3

15
49
51
18
48
43
38
12
58
18
39
5
8
8
27
24
61
17
59

62
150
87
35
146
136
38
49
138
52
93
41
38
22
66
77
113
67
104

271
583
243
133
369
322
26
128
462
114
322
149
132
86
330
300
374
266
284

531
919
601
333
472
429
49
322
791
303
488
292
381
381
503
421
605
359
441

624
1065
759
493
586
496
184
563
928
511
558
415
537
448
594
468
733
422
529

658
1148
721
591
565
555
369
561
977
596
613
442
587
562
631
462
825
446
569

C450
NWC350
NWC450

0.35
0.35
0.35

450
350
450

19.4
0.3
0.4

0.7
0.8
0.7

0.35
0.45
0.35

1610
2264
2299

31.2
65.8
76.2

37
44
51

91
112
123

190
260
238

470
306
327

616
348
376

638
410
385

wag Intial aggregate water content.


The negative sign refers to contraction.

180

365

1042

J.A. Bogas et al. / Materials and Design 56 (2014) 10391048

For BS450, coarse sand was replaced by the lightweight sand (LWS)
indicated in Table 2 (type B 0-3). The maximum concrete aggregate
size was 12.5 mm.
LWC with initially dry LWA (PD) or pre-wetted LWA (PW) were
also produced to study their inuence on shrinkage. The PD aggregate was added during mixing after being dried at 200 C and the
PW aggregate was rst wetted for 3 min with 50% of the total
water, before mixing. Based on the method suggested by Bogas
et al. [24], the absorption of LWA in the mix was estimated beforehand to take into account the correction of the total mix water.
Modied normal density concretes (MND) were produced with
partial 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 nanosilica
and 22% or 40% of y ash by weight of cement are used. The NS
content was dened based on previous experimental studies with
NWC carried out by the supplier. According to these studies,
about 0.15 kg of NS had the same effect on the compressive
strength 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 were
produced to measure the total shrinkage according to E398 [25].
To analyse the inuence of the specimen size, two additional specimens of 150  150  600 mm were produced for the reference
mixes with Type B and NA aggregate (B/NWC450_PR).
After demolding at 24 h, the specimens were placed in a controlled chamber with a temperature of 22 2 C and a relative
humidity of 50 5%, according to E398 [25]. In addition, two specimens of each of the B/NWC 450 reference mixes were previously
cured for 7 days in water before they were placed in the chamber.
The total axial shrinkage was monitored by a demountable

mechanical strain gauge (DEMEC) with a precision of 1 lm and a


gauge length of 5 mm. The DEMEC was placed over two steel pins,
200 mm apart, which had been glued onto one of the concretes
moulded surfaces (Fig. 1). Compressive tests were also carried
out on 150 mm water-cured cube specimens at 28 days. The dry
density, qd, the compressive strength, fcm, and the total shrinkage,
ecs, at different ages are listed in Table 4.

3. Results and discussion


3.1. Inuence of the aggregate type
As expected, the shrinkage of LWC is lower than that of NWC at
earlier ages and higher at later ages. The lower initial shrinkage
rate of LWC is offset in the long-term (Fig. 2). The shrinkage of
LWC 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 relationships
were 1.32 (type A), 1.54 (type B) and 1.66 (type C). This behaviour
has also been documented by other authors (e.g. [1,8,13]); it is
essentially related to the water supplied by LWA during the initial
ages and the lower restriction effect imposed by the less rigid LWA
at later ages. The additional partial replacement of natural sand by
lightweight sand led to increased shrinkage at all ages, with values
about twice as high as for NWC, after one year.
The shrinkage of LWC is lower than that of NWC up to about
40 days for LWC with type C LWA and up to about 3 months for
LWC with the most common types A and B LWA. Taking into account the initial water percentage of LWA (Table 4) and their particle density (Table 2), the type B aggregate has the highest initial
water content (262 L/m3) followed by types A and C, with about
169 L/m3 and 147 L/m3, respectively. The highest water content
of the LWC with type B aggregate is conrmed by the weight loss
(DM) curves shown in Fig. 2.

100
0
-100
-200
-300
-400
-500
-600
-700
-800
-900

M (%)

cs (x10-6)

Fig. 1. Specimen geometry: prisms of 80  80  330 mm (left); shrinkage monitoring (right).

30

60

90

180

365

days (log scale)


B450
A450
BS450
C450
NWC450
NWC350

10
8
6
4
2
0
3

30

60

90

180

365

days (log scale)


Fig. 2. Inuence of the type of aggregate on the total shrinkage and weight loss up to one year old.

1043

J.A. Bogas et al. / Materials and Design 56 (2014) 10391048

w/c
0.25
0

cs (x10-6)

When the internal curing becomes less relevant (after about the
rst 730 days) there is a sudden increase in the shrinkage rate of
LWC compared to NWC. After this transition period, the shrinkage
of LWC is proportional to the water lost by evaporation (curve
inection in Fig. 3) and the less rigid type C LWA has the lowest
restriction effect on the paste deformation, which leads to higher
shrinkage rates (more stepped curve).
In the initial ages, the weight loss, DM (water lost by evaporation) in the LWC is not followed by a proportional shrinkage increase, conrming the effectiveness of the internal curing. But
this proportionality is found in the NWC from the rst days of drying. Because of the severe drying and the small equivalent thickness of the specimens, the shrinkage tends to stabilize after
about 912 months. This stabilization is slower in the LWC due
to 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 volume
of water has a small effect on shrinkage when the w/c ratio ranges
between 0.35 and 0.45 [26]. In this case, the reduction of the volume of paste is offset by the increment of the w/c ratio.

0.3

0.35

0.4

0.45

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

-200
-400
-600

R = 0.81

-800
Fig. 5. One year shrinkage versus w/c ratio.

ratio the longer the drying delay. Moreover, there is no relevant


shrinkage or even a slight expansion for concrete mixes with a
w/c ratio of 0.3. In these concrete mixes, the drying is less effective
and the water movement to the exterior is temporarily halted.
As expected, more weight is lost in concrete mixes with a higher
w/c ratio and higher initial water content (Fig. 4). In fact, for a higher w/c ratio but the same volume of water, the concrete is more
permeable and consumes less hydration water, which leads to
greater losses of evaporable water.

3.2. Inuence of the cement content


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

For the same cement content, the shrinkage increases with


higher w/c ratio (Figs. 6 and 7). In fact, there is an increment of
the volume of paste and a corresponding reduction of the aggregate content. Moreover, for the same cement content, the higher
the w/c ratio the lower the mortar stiffness and the higher the volume of evaporable water. A small reduction of the w/c ratio also
means a signicant delay in drying shrinkage. In this case, varying

cst (x10-6)

B450
A450
BS450
C450
NWC450
NWC350

M (%)

-100
-200
A450
ANS
ACZ22
B450
BSF
BNS
BFA22
BFA40

-300
-400
-500
-600
-700

cs (x10-6)

Fig. 3. Total shrinkage versus weight loss, DM.

100
0
-100
-200
-300
-400
-500
-600
-700

30

60

90

180

365

days (log scale)


A350
A450
A525
B350
B450
B525

M (%)

6
4
2
0
3

30

60

90

180

365

days (log scale)


Fig. 4. Inuence of the cement content on the total shrinkage and weight loss up to one year old.

1044

J.A. Bogas et al. / Materials and Design 56 (2014) 10391048


3

cs (x10-6)

0
-100
-200
-300
-400
-500
-600
-700
-800
-900
-1000
-1100
-1200

30

60

90

180

365

days (log scale)

B450
B450_0.45
B450_0.55
B450_Wsp0.35
B450_Wsp0.5

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

w/c

cs (x10-6)

0.3

0.35

0
-200
-400
-600
-800
-1000
-1200

0.4

0.45

0.5

0.55

Type B (1 year)

R = 0.92

B450Wsp0.5

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

the w/c ratio from 0.55 to 0.35 led to a shrinkage reduction of


77.3% at 30 days and 48% at one year (Fig. 7).
The magnitude and progression of the shrinkage were quite
similar in the dry mixes without Sp (B450_Wsp0.35) and in the
uid mixes of the same composition with Sp (B450), Fig. 6.
The main effect of superplasticizers should be related to how
they can change the water content and the w/c ratio of concrete
[27,28]. To better study this effect, a concrete without Sp but with
higher water content (B450_Wsp0.5) was produced to have the
same slump as the reference mix B450. In this case, the introduction of Sp leads to a reduction of the shrinkage by about 39.5% after
one year (Table 4). Fig. 7 shows that B450_Wsp0.5 follows the
same trend as the other concrete mixes with different w/c ratios,
regardless the use of Sp.

higher the volume of LWA the greater the efciency of the internal
curing.
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 inuence on the shortterm and long-term shrinkage.
Therefore, there is a greater inuence of the volume of aggregate on the shrinkage of LWC with type A aggregate than with type
B LWA. On the one hand, since the stiffness of the type A aggregate
is higher, its replacement by natural sand has less inuence on the
restriction effect on the free deformation of the paste. On the other
hand, the greater amount of particles and the smaller maximum
aggregate 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 mortar
characteristics implies a proportional increase of the volume of
sand and paste, with a corresponding increase of the shrinkage
(Fig. 9). In fact, the reduction of the LWA volume leads to a simultaneous increase of the paste content and a reduction of the volume of water available for internal curing. The inclusion of

Aggregate volume (L/m3)


200
0

cs (x10-6)

3.4. Inuence of the volume of aggregate

0
-100
-200
-300
-400
-500
-600
-700

M (%)

cs (x10-6)

Considering concrete mixes with different volumes of aggregates but with the same w/c ratio and cement content. On one
hand, since the replacement of sand by lightweight coarse aggregate implies a reduction of the mortar stiffness, increased
shrinkage would be natural. But on the other hand, the

-200

250

300

350

Type A (1 year)

-400
R = 0.78

-600

VB400

-800
Fig. 9. One year shrinkage versus the volume of aggregate.

30

60

90

180

365

days (log scale)

6
4
2
0

450

Type B (1 year)

A450
VA425
VB400
B450
VmB425
VmB250

400

30

60

90

180

365

days (log scale)


Fig. 8. Inuence of the volume of aggregate on the total shrinkage and weight loss up to one year old.

1045

J.A. Bogas et al. / Materials and Design 56 (2014) 10391048

M (%)

cs (x10-6)

3
0
-100
-200
-300
-400
-500
-600
-700

30

60

90

180

365

days (log scale)


A450
ANS
AFA22
B450
BSF
BNS
BFA22
BFA40

6
4
2
0
3

30

60

90

180

365

days (log scale)


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

concrete mix VB400 in Fig. 9 shows that the shrinkage of LWC


depends on how the volume of LWA varies.
3.5. Concrete with admixtures
The replacement of 8% of cement by silica fume had little
inuence in terms of both shrinkage and weight loss (Fig. 10).
Similar ndings were reported by Hooton [29] for normal concrete. Note that compressive strength was also little affected
(Table 4). The absence of initial water curing may have inhibited
the regular development of the pozzolanic reactions. It is also
likely that there was no effective dispersion of silica fume in
concrete.
There is a long-term reduction of the shrinkage in LWC with
nanosilica, especially in LWC with type A aggregate. After one
year, the reduction was 19.7% in LWC with type A LWA and 5%
in LWC with type B LWA (Fig. 10). The addition of nanosilica
led to a slight reduction of the volume of paste. Moreover, there
may have been a renement of the matrix porosity with a consequent increase of the mortar stiffness. In addition to the ller
effect, the high reactivity of nanosilica means it reacts at early
ages when there is more water available within the concrete. In
this case, the migration of water from the LWA to the paste
may be relevant. Because of the nano-size of the NS, part of this
admixture may be absorbed by the more porous LWA (type B),
thus reducing its effectiveness.

%FA
0

22

40

cs (x10-6)

0
Type B (1 year)
Type A (1 year)

-200
-400
-600
-800

Shrinkage increases at all ages in the LWC with 22% replacement of cement by y ash, regardless the type of LWA. However,
the difference is more relevant in the LWC with more porous
LWA (type B). The shrinkage increase in LWC with y ash is also
reported by Holm [8] and Haque [30], although the opposite was
reported by Malhotra [16]. Since the hydration reactions are slower
in y ash mixtures, their porous structure tends to be coarser, at
least in the early ages. Therefore, the drying rate increases and
the restriction on the paste deformation decreases (Fig. 10). On
the other hand, since the hydration reactions occur later, there is
less compensation for the water stored in the LWA and meanwhile
eliminated by evaporation. The absence of initial water curing and
the increasing volume of paste also affect the results.
Interestingly, although the shrinkage is lower in the LWC with
type A LWA, the water lost is slightly lower in the LWC with type
B LWA. Moreover, the long-term shrinkage of the LWC with 40% of
y 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 y ash,
especially in the rst 6 months. This behaviour can be related to
the differences in the level of the paste hydration.
Since the mixes with a higher volume of y ash have a lower degree of hydration, there are more non-hydrated particles that work
as non-shrinkage microaggregates. Therefore, the volume of particles that act as a source of contraction is lower. Termkhajornkit
et al. [31] have conrmed that the shrinkage in NWC is affected
by the extent to which y ash hydrates, which is lower for higher
amounts. In part, this phenomenon may explain the difference observed in how shrinkage evolves in BFA22 and BFA40, after 30 days
(Fig. 10). In addition, the lower hydration level of the paste justies
the higher drying rate (Fig. 10) and the lower effect of the water
lost in the shrinkage of BFA40 (Fig. 3). The behaviour of the LWC
with type A LWA and 22% of y ash is between that found for
BFA22 and BFA40. This is probably because there is less water
available in type A LWA than in type B LWA, which means lower
levels of hydration in the LWC with type A LWA.

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

cs (x10-6)

3
0
-100
-200
-300
-400
-500
-600

30

60

90

180

365

days (log scale)


A450
A35%
A65%
B450
B35%
B65%
NWC450

Fig. 12. Inuence of the total or partial replacement of normal weight coarse aggregate by LWA on the total shrinkage up to one year old.

1046

J.A. Bogas et al. / Materials and Design 56 (2014) 10391048

3.8. Inuence of the curing conditions and specimen size

VLWA (L/m3)
0

35

65

100

cs (x10-6)

0
Type B (1 year)
Type A (1 year)

-200
-400

R = 0.73

-600
Fig. 13. One year shrinkage versus the percentage replacement of LWA.

3.6. Total or partial replacement of NA by LWA


Figs. 12 and 13 show that the higher the percentage replacement of NA by LWA the higher the long-term shrinkage. But the
higher the percentage replacement the lower the initial shrinkage.
On the one hand, the greater the percentage replacement of NA by
LWA the higher the water available to the paste at early ages. On
the other hand, the replacement of NA by the less rigid LWA reduces the restriction effect on the paste deformation. From
Fig. 12, it can be concluded that 35% replacement of type B LWA
and 65% replacement of type A LWA are enough for the water lost
by evaporation to be practically offset in the rst 7 days.
3.7. Inuence of the initial water content of LWA
As expected, there was a greater delay of the drying shrinkage
and a reduction of the initial shrinkage in the LWC with pre-saturated LWA (about 53.5% lower in the LWC with type A LWA and
58.8% in the LWC with type B LWA, Fig. 14). However, the pre-saturation of the LWA had little inuence on the long term-shrinkage.
After one year, the shrinkage of the LWC with pre-dried aggregates
is 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 lower
shrinkage of LWC with pre-dried aggregates can be related to the
improvement of the aggregate-paste transition zone that promotes
a more effective restriction effect by the aggregates. By comparing
the LWC compressive strength for different initial wetting conditions of LWA, it is conrmed that the mortar characteristics should
be quite similar (Table 4).

cs (x10-6)

3
0
-100
-200
-300
-400
-500
-600
-700

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 was
previously cured for 7 days in water (B/NWC450_7D, Fig. 15). For
these curing conditions, the period in which the shrinkage of
LWC is lower than that of NWC was extended from about 3 to
5 months.
As shown in Fig. 15, the shrinkage evolution is signicantly affected by the specimens geometry. The higher the specimen thickness the lower the drying shrinkage, especially in the centre of the
concrete, which behaves as if it was isolated from the exterior.
However, the difference in the shrinkage evolution is more relevant 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 drying
shrinkage is only effective after about 3 months (Fig. 15).
In this study, the one year shrinkage of LWC was 34% lower or
54% higher than that of NWC, depending on the size of the specimen. These differences may explain the apparently contradictory
results reported in the literature. Moreover, the results show that
under real conditions LWC can be quite slow to dry and only the
long-term shrinkage should be more meaningful.
4. Normative analysis
The shrinkage curves obtained for LWC with type A and type B
LWA are presented in Figs. 16 and 17. These graphs cover w/c ratios
between 0.3 and 0.55 (type B LWA) and between 0.3 and 0.45 (type
A LWA). Figs. 16 and 17 also show the estimated shrinkage curves
according to ACI 209R [18], EN1992-1 [19] and MC 2010 [20].
To build the normative curves shown in Figs. 16 and 17, the following 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) and
h0 = 2Ac/u = 40 mm, with Ac = 80  80 mm2 (cross-sectional
area) and u = 80  4 mm2 (cross-sectional perimeter);

30

60

90

180

365

days (log scale)


A450
A450PM
A450PD
B450
B450PM
B450PD

Fig. 14. Inuence of the initial water content of LWA on the total shrinkage up to one year old.

100

30

60

90

180

365

cs (x10-6)

days (log scale)

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

B450
B450_7D
B450_PR
NWC450
NWC450_7D
NWC450_PR
Fig. 15. Inuence of the specimen size on the total shrinkage up to one year old.

1047

cs (x10-6)

J.A. Bogas et al. / Materials and Design 56 (2014) 10391048

100
0
-100
-200
-300
-400
-500
-600
-700
-800
-900
-1000
-1100
-1200

30

60

90

180

365

days (log scale)


MaxTypeB (w/c=0.45)
MaxTypeB
MinTypeB
ACI 209R
EN1992 (min)
EN1992 (max)
MC2010 (min)
MC2010 (max)

cs (x10-6)

Fig. 16. Shrinkage curves and normative estimations for LWC with type B aggregate (w/c of 0.30.55).

100
0
-100
-200
-300
-400
-500
-600
-700
-800
-900

30

60

90

180

365

days (log scale)


MaxTypeA
MinTypeA
ACI 209R
EN1992 (min)
EN1992 (max)
MC2010 (min)
MC2010 (max)

Fig. 17. Shrinkage curves and normative estimations for LWC with type A aggregate (w/c of 0.30.45).

In general, the expressions in the standards tend to overestimate the shrinkage of LWC with a w/c ratio up to 0.45, especially
at early ages. These expressions are not suitable for lightweight
concrete, since they do not take into account the shrinkage delay
caused by the internal curing.
Of the methods analysed, the one suggested by ACI 209R [18]
led to the best tted shrinkage curves at early ages. However, the
shrinkage estimation at later ages may be too conservative, particularly for LWC with a low w/c ratio or with LWA of high initial
water content. It is noteworthy that this report does not, for practical purposes, take into account the concrete composition and the
water content of the aggregates. The method is empirically based
and less reasonable for moderate to high-strength LWC.
The long-term shrinkage estimates from the expressions suggested by EN1992-1 [19] and MC 2010 [20] are within the experimental results obtained for low w/c lightweight concrete,
especially after a long drying period. However, the expressions
were totally inadequate for early ages. Moreover, the shrinkage
of common LWC with higher w/c ratio can be greatly underestimated in these recommendations. In fact, the long-term shrinkage
of LWC with w/c ratio above 0.45 is underestimated by more than
50% (Fig. 16). However, since multiplier factors for LWC are provided in EN1992-1 [19] and MC 2010 [20] these differences can
be smaller. Comparing just the one year shrinkage of LWC with
that of NWC with the same composition (Table 4), multiplier

f cm (MPa)
30 35 40 45 50 55 60 65 70 75 80

cs,1year (x10-6 m/m)

 f = 35 was assumed for the expression of ACI209R (1992);


 percentage of ne 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

0
-200
-400
-600
-800
-1000
-1200

VmB425
BFA40
Type A
Type B
MND (Type A)
MND (Type B)

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

coefcients of about 1.3 for higher density LWA (type A) and about
1.6 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 the
compressive strength. However, shrinkage does not depend on the
compressive strength but on the parameters related to the microstructure 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 compressive
strength, as shown in Fig. 18.

5. Conclusions
The inuence of different compositional parameters on the
shrinkage of lightweight concrete was analysed by means of an
extensive experimental programme. The internal curing and the
lower stiffness of LWA are responsible for the lower initial shrinkage and the higher long term shrinkage of the LWC. Depending on
the type of LWA, the one year shrinkage of LWC was about 3065%
higher than that of NWC and about twice as high as that of NWC
when additionally there was a replacement of natural sand by
lightweight ne aggregates. The partial replacement of coarse NA
by LWA led to intermediate shrinkage curves relative to those obtained for concrete produced with only NA or LWA.

1048

J.A. Bogas et al. / Materials and Design 56 (2014) 10391048

Keeping the effective volume of water, the shrinkage increase of


LWC with the w/c ratio is more effective the higher the initial
water content of LWA. Concrete shrinkage is affected by the introduction of superplasticizer, but only when this implies a reduction
in the volume of mixing water.
The shrinkage of LWC depends on how the volume of LWA varies. Keeping the same volume of paste and increasing the volume
of aggregate, the shrinkage of LWC can either decrease or be about
the same, depending on the type of LWA and its water content. For
the same mortar characteristics there is a proportional increase of
shrinkage the LWA volume is reduced.
The addition of 8% of silica fume had little inuence on shrinkage at all ages. However, the long-term shrinkage of LWC with the
more reactive nanosilica tends to be lower. The shrinkage of LWC
with y ash strongly depends on the percentage of replacement
and type of LWA. The shrinkage increases at all ages for 22% of
y ash replacement, and is more relevant in LWC with more porous
LWA. However, for 40% of y ash replacement the shrinkage is lower than that of concrete without y ash.
The pre-saturation of LWA reduces the early shrinkage and delays the drying shrinkage. However, the initial wetting condition of
LWA had little inuence on the long-term shrinkage. It was also
shown that the shrinkage can be strongly affected by the geometry
of the specimens, a phenomenon that is more relevant in LWC than
in NWC. This leads to a greater susceptibility of LWC to differential
shrinkage.
The current methods suggested in the main standards and
codes cannot properly predict the shrinkage behaviour of LWC. In
particular, the suggested methods do not take into account the typical delayed shrinkage of LWC. It has also been shown that there is
a weak correlation between the shrinkage and the compressive
strength in LWC, contrary to the philosophy adopted in EN19921 [19] and MC 2010 [20]. It is necessary to nd new solutions that
can better take the inuence of the w/c ratio on shrinkage into
account.
A rough but simple procedure to take into account the shrinkage delay in LWC can be to consider a ctitious higher curing period in the codes recommendations. Based on the experimental
results, multiplier coefcients of about 1.3 for the higher density
LWA and about 1.6 for the more porous LWA are suggested as a
way of taking into account the higher long-term shrinkage of lightweight concrete.
Acknowledgements
The authors wish to thank ICIST-IST for funding the research
and the companies Argex, Saint-Gobain Weber Portugal, Soarvamil,
BASF and SECIL for supplying the materials used in the experiments. The research work presented herein was supported by the
Portuguese Foundation for Science and Technology (FCT), under
grant PTDC/ECM-COM1734/2012.
References
[1] Zhang M-H, Li L, Paramasivam P. Shrinkage of high-strength lightweight
aggregate concrete exposed to dry environment. ACI Mater J
2005;102(10):8692.
[2] Kayali O, Haque MN, Zhu B. Drying shrinkage of bre-reinforced lightweight
aggregate concrete containing y ash. Cem Concr Res 1999;29(11):183540.
[3] Virlogeux M. Gnralits sur les caractres des btons lgers. In: Granulats et
betons legers-Bilan de dix ans de recherches, Arnould et Virlogeux, Presses de
lcole nationale des ponts et chausses; 1986, p. 111246.

[4] Bogas JA. Characterization of structural lightweight expanded clay aggregate


concrete. Lisbon: PhD thesis in civil engineering. Technical University of
Lisbon, Instituto Superior Tcnico; 2011 [in Portuguese].
[5] Holm, TA, Bremner TW. State-of-the-art report on high-strength, highdurability structural low-density concrete for applications in severe marine
environments. Us Army corps of engineers. Structural Laboratory, ERDC/SL TR00-3; 2000. 104 p.
[6] Selih J, Bremner TW. Drying of saturated lightweight concrete: an
experimental investigation. Mater Struct 1996;29(7):4015.
[7] Coquillat G. Inuence des caractristiques physiques et mcaniques des
granulats lgers sur les proprits des btons legers de structure. In:
Granulats et betons legers-Bilan de dix ans de recherches, Arnould et
Virlogeux, Presses de lcole nationale des ponts et chausses; 1986, p. 25598.
[8] Holm TA. Performance of structural lightweight concrete in a marine
environment. In: International symposium ACI SP-65. St. Andrews By-TheSea, Canada; 1980. 12p.
[9] Hossain KMA, Lachemi M. Mixture design, strength, durability, and re
resistance of lightweight pumice concrete. ACI Mater J 2007;104(5):44957.
[10] Nobuta Y, Satoh K, Hara M, Sogoh S, Takimoto K. Applicability of newly
developed high-strength lightweight concrete for civil structures. In: S.
Helland et al., editors, Second international symposium on structural
lightweight aggregate concrete, Kristiansand, Norway, 1822 June; 2000. p.
396405.
[11] Wegen GJL, Bijen JMJM. Properties of concrete made with three types of
articial PFA coarse aggregates. Int J Cem Compos Lightweight Concr
1985;7(3):15967.
[12] Nilsen AU, Aitcin PC. Properties of high-strenght concrete containing light,
normal, and heavy weight aggregate. Cem Concr Aggre 1992;14(1):812.
[13] FIP. FIP manual of Lightweight aggregate concrete. Fdration internationale
de la prcontrainte (FIP). 2nd ed., Surrey University Press; 1983.
[14] EuroLightCon R31. Long-term effects in LWAC: Strength under sustained
loading shrinkage of high strength LWAC. Eur. Uni. Brite EuRam III, BE963942/R31; 2000.
[15] Hoff GC. High strength lightweight aggregate concrete for artic applications
Part1,2,3. In: Holm, Vaysburd, editors, Structural lightweight aggregate
concrete performance, ACI SP-136; 1992, p. 1245.
[16] Malhotra VM. Properties of high-strength lightweight concrete incorporating
y ash and silica fume. In: Hester, editors, High-strength concrete, second
international symposium, SP-121. Michigan, American Concrete Institute;
1990, p. 64566.
[17] Zhutovsky S, Kovler K, Bentur A. Inuence of cement paste matrix properties
on the autogenous curing of high-performance concrete. Cem Concr Compos
2004;26(5):499507.
[18] ACI 209.2R-08. Guide for modeling and calculating shrinkage and creep in
hardened concrete. American Concrete Institute; 2008.
[19] EN 1992-1-1. Eurocode 2: Design of concrete structures Part 11: General
rules and rules for buildings. European Committee for standardization CEN;
2010.
[20] MC 2010. Model Code (MC) prepared by b special Activity Group 5 Final
draft; 2011.
[21] Bogas JA, Mauricio A, Pereira MFC. Microstructural analysis of Iberian
expanded clay aggregates. Microsc Microanal 2012;18:1190208.
[22] Bogas JA, Gomes A. A simple mix design method for structural lightweight
aggregate. Mater Struct 2013;46(11):191932.
[23] Bogas JA, Gomes A. Compressive behavior and failure modes of structural
lightweight aggregate concrete characterization and strength prediction.
Mater Des 2013;46:83241.
[24] Bogas JA, Gomes A, Gloria MG. Estimation of water absorbed by expanding clay
aggregates during structural lightweight concrete production. Mater Struct
2012;45:156576.
[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;36:43376.
[27] Collins TM. Proportion high-strength concrete to control creep and shrinkage.
ACI Mater J 1989;86(55):57680.
[28] Rixon R, Mailvaganam N. Chemical admixtures for concrete. 3rd ed. E&FN
SPON; 1999.
[29] Hooton D. Inuence of silica fume replacement of cement on physical
properties and resistance to sulphate attack, freezing and thawing, and
alkali-silica reactivity. ACI Mater J 1993;90(15):14351.
[30] Haque MN. Strength development and drying shrinkage of high-strength
concretes. Cem Concr Compos 1996;18(5):33342.
[31] Termkhajornkit P, Nawa T, Nakai M, Saito T. Effect of y ash on autogenous
shrinkage. Cem Concr Res 2005;35(3):47382.

S-ar putea să vă placă și