Construction and Building Materials 61 (2014) 260–269

Contents lists available at ScienceDirect

Construction and Building Materials

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

Autogenous shrinkage of high performance concrete containing mineraladmixtures under different curing temperatures

http://dx.doi.org/10.1016/j.conbuildmat.2014.03.0230950-0618/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel./fax: +86 571 8832 0124.E-mail address: yangyang@zjut.edu.cn (Y. Yang).

Chenhui Jiang a,b, Yang Yang c,⇑, Yong Wang c, Yuenian Zhou d, Chengchang Ma c

a College of Chemical Engineering and Materials Science, Zhejiang University of Technology, Hangzhou 310014, PR Chinab Department of Construction Engineering, Zhejiang College of Construction, Hangzhou 311231, PR Chinac College of Architecture and Construction Engineering, Zhejiang University of Technology, Hangzhou 310014, PR Chinad Zhoushan Supervision Station of Construction Engineering Quality, Zhoushan 316000, PR China

h i g h l i g h t s

�We provided a database of autogenous shrinkage of HPC under different temperatures.� We updated the measurement method of autogenous shrinkage.� FA and BS will decrease and increase autogenous shrinkage respectively.� Both the rate and the magnitude of autogenous shrinkage vary with temperature.� We proposed an equation of estimating autogenous shrinkage at different temperatures.

a r t i c l e i n f o

Article history:Received 26 November 2013Received in revised form 17 March 2014Accepted 17 March 2014Available online 5 April 2014

Keywords:Autogenous shrinkageHigh performance concrete (HPC)Fly ash (FA)Blast-furnace slag (BS)Curing temperatureWater–binder ratio (w/b)Estimation

a b s t r a c t

The present study investigated experimentally autogenous shrinkage behaviors of high performanceconcrete (HPC) containing fly ash (FA) and blast-furnace slag (BS) exposed to different isothermaltemperatures. The deformation of concrete specimen after initial setting was determined using a modi-fied method which is based on non-contact measurement technique. The results indicated that themethod can precisely monitor non-load induced deformations of HPC mixtures. The inclusions of BSand FA resulted in significant increase and decrease of autogenous shrinkage of HPC, respectively. Whileboth the rate and the magnitude of autogenous shrinkage for almost all mixtures were increased with riseof curing temperature, extents of the influence were varied with water–binder ratio, composition ofcementitious materials and age. It is noted that although the equivalent age equation was widely appliedto evaluate temperature dependence of mechanical properties of cement-based materials, its applicabil-ity on autogenous shrinkage of HPC was questionable. In addition, on a trial and error basis, a modifiedautogenous shrinkage equation was performed in terms of numerical fitting of the measured data.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction macro-properties [1,2]. As a predominant aspect, considerable

In recent years, as a typical structural material, high perfor-mance concrete (HPC) is widely used in civil engineering becauseof its excellent performance, namely high workability, highstrength, high durability and long-term performance [1–3]. Forthe purpose of improving performance of concrete, lower andlower w/b are executed [2,3], and a series of mineral and chemicaladmixtures (such as fly ash (FA), ground granulated blast furnaceslag (BS) and superplasticizer) are introduced [1–3]. Accordingly,many aspects of micro-structures of HPC are distinguished fromnormal concrete, which further results in differences of

early-age volume changes of HPC usually compromise its ‘‘highperformance’’. Due to various types of internally and/or externallyrestraints in concrete structures, these volume changes cannotrelease freely, and tensile stress often arises. As tensile stressincreases and exceeds tensile strength of HPC at a specific age,early-age cracking happens. Once cracking, the strength, long-termperformance and global stability of HPC structures will be deterio-rated seriously. In general, early-age volume changes of HPC aremainly composed of autogenous shrinkage, thermal deformationand drying shrinkage [1], which induced by self-desiccation ofcapillary porosity, by temperature and moisture gradients betweenconcrete and exterior surroundings, respectively [1–3].

It is observed that compared to thermal deformation and dryingshrinkage, autogenous shrinkage accounts for the foremost

Table 1Testing parameters and their levels.

Testing parameters Corresponding levels

Water binder ratio (w/b) 0.20; 0.30; 0.40Composition of cementitious materials

(by mass)OPa: 100%OP, no inclusions of FAand BSFAb: 65%OP + 35%FABSc: 50%OP + 50%BS

Curing temperature (�C) 10; 20; 30Age of strength test (d) 1; 3; 7; 14; 28

a OP stands for ordinary Portland cement.b FA stands for fly ash.c BS stands for blast furnace slag.

Table 2Oxide and potential mineral compositions (by mass) and physical properties ofcementitious materials.

Items Unit OP FA BS

C. Jiang et al. / Construction and Building Materials 61 (2014) 260–269 261

significance in volume change components of HPC at early ages[3,4]. Researchers have issued a great quantity of scientific litera-ture and technical reports on terminology, experimental methods,mechanism, and prediction models on autogenous shrinkage andinduced restrained stress [1,3–6] in last two decades. As a typicalaging material, concrete is subjected to vital temperaturevariations induced by heat release (temperature rise) associatedwith hydration of cementitious materials, by subsequent cooling(temperature drop) due to heat transfer [3,6,7], and by changesin ambient temperature when concreting is performed under rela-tively hot and cold climatic conditions. Autogenous shrinkagemainly results from self-desiccation in porosity, which is inten-sively related to cement hydration. Whereas the rate and degreeof hydration closely depend on the exposed temperature historyof concrete, it can be inferred that a link is existed between the ex-posed temperature history and autogenous shrinkage.

However, limited research work on this topic is available, andmost of the work focuses on cement pastes. The influences of cur-ing temperature on autogenous shrinkage have been investigatedfirstly by Tazawa and Miyazawa in Japan [8], who confirmed tem-perature effects of autogenous shrinkage can be estimated by anequivalent age equation. Jensen and Hansen demonstrated thatthe traditional maturity concept generally is not applicable toautogenous deformation and autogenous RH change of hardeningcement paste [10]. Weiss, Lura and Sant examined the influencesof curing temperature on autogenous deformation in cement pastecontaining shrinkage reducing admixtures [9,27]. Based on exper-imental investigation, Chu et al. [7] concluded that high curingtemperature at early ages results in lower autogenous shrinkageat later ages when compared to the cases subjected to relativelylow curing temperature at early ages. They suggested that theequivalent age function [10,11] is not applicable to evaluatetemperature dependence of autogenous shrinkage at early ages[9]. From the point of view of development rate, Maruyama andTeramoto [12] divided autogenous shrinkage of ultra high strengthconcrete into two stages, i.e., the earlier age followed by the laterage stage. Lower temperature increased autogenous shrinkage atearlier-age stage; higher temperature produced larger autogenousshrinkage at later-age stage. Therefore, in these literatures, no con-sensus is reached on autogenous shrinkage behavior exposed todifferent temperature conditions. Meanwhile, little work is focusedon effects of FA and BS that is commonly used as supplementarycementing materials [24,28–31] on autogenous shrinkage.

In order to explore the influence of temperature on autogenousshrinkage of HPC in depth, an experimental investigation isperformed on a series of typical HPC containing FA and BS exposedto 3 different isothermal temperatures. For the sake of monitoringvolume change strain of HPC exactly, especially at early ages, atesting method based on a non-contact measurement techniqueis used. Also, a modification based on Tazawa and Miyazawa’smodel of autogenous shrinkage [8] is performed.

CaO % 62.64 3.60 31.33SiO2 % 22.24 59.31 33.05Al2O3 % 5.34 22.10 15.47Fe2O3 % 3.20 8.26 0.27MgO % 0.63 1.82 16.11Na2O % 0.29 0.33 0.72SO3 % 2.64 0.28 0.37LOIa % 1.67 – –

Potential mineral compoundsC3S % 45.5 – –C2S % 29.5 – –C3A % 8.7 – –C4AF % 9.7 – –Specific gravity g/cm3 3.12 2.25 2.90Finenessb m2/kg 376 343 436

a LOI stands for loss on ignition of cement.b Test by the Blaine air permeability method.

2. Experimental program

2.1. Experimental parameters

As tabulated in Table 1, experimental parameters mainly including w/b, compo-sition of cementitous material and curing temperature. Corresponding levels ofeach parameter are set based on characteristic of HPC mixtures and typical expo-sure conditions, respectively. Also, substitution ratios of FA and BS are in the rangeof typical applications.

2.2. Materials and mixture proportions

2.2.1. MaterialsOrdinary Portland cement (OP) in accordance with China National Standard GB

175-2009, fly ash (FA) and ground granulated blast furnace slag (BS) are used ascementitious materials (binders). Their oxide compositions based on X-ray

fluorescence analysis and potential mineral compounds of cement are tabulatedin Table 2. Crushed stone with a maximum nominal size of 20 mm, and river sandare used as coarse and fine aggregates, respectively. Also, a liquid polycarboxylate-based superplasticizer (SP) is used to adjust workability of different w/b HPCmixtures to the same level.

2.2.2. Concrete mixture proportionsAs shown in Table 3, the mixtures without mineral admixture are viewed as

control mixtures. As far as the mixtures with the same w/b, total mass contentsof binders are constant, i.e., FA and BS replace cement by equivalent mass, respec-tively. For the purpose of gaining similar slump, slump flow and air content, thedosages of superplasticizer are properly adjusted. One more thing, in order to keepthe same level of restraint effects of aggregate to volume change, the same amountof fine and coarse aggregate are adopted in all mixtures excluding ‘‘VF-30’’.

2.3. Autogenous shrinkage

2.3.1. Testing methodsMeasurements of autogenous shrinkage have been carried out in two funda-

mentally different ways: measurement of volumetric strain and of linear strain(one-dimensional strain). Compared to measuring volumetric strain, linear strainprovides more explicit engineering definition, and its measurement is much easierto handle and control. One major disadvantage of the previous linear method [3,13–15] is the tremendous obstacle of starting the measurements immediately aftercasting, because measurement apparatus (e.g., linear variable differentialtransducer) is difficult to install when concrete is at a plastic state before setting.

Recently, early-age linear strain of concrete is generally measured with non-contact sensors [13–15,27], such as laser-based sensor and capacitive sensors. Inthis work, they are replaced by an eddy-current displacement sensor (ECDS) whichoperate with electromagnetic induction effect. The details of the testing apparatususing the ECDS are shown in Fig. 1. There has certain advantages of ECDS used overlaser-based sensors, such as the requirements on testing environment of the formerare lower than those of the latter, which cannot be applied in dirty and dusty sur-rounding, and the prices of the ECDS are generally superior to the laser-based ones[15]. To improve the reusing rate of the ECDS, up to the age of 3 days, electric

Table 3Mixture proportions and basic properties of the HPC.

Mix. ID w/b Unit contents of raw materials (kg/m3) Basic properties

OP FA BS W SPa (%) S G Slump (mm) Slump flow (mm) Air content (%) 28d fcb (MPa)

OP-20 0.20 700 – – 140 2.00 600 900 265 600 � 605 4.2 115.3OP-30 0.30 600 – – 180 1.00 600 900 270 610 � 620 3.9 85.7OP-40 0.40 500 – – 200 0.80 600 900 275 620 � 630 3.4 71.3FA-20 0.20 455 245 – 140 1.70 600 900 275 650 � 655 3.8 94.3FA-30 0.30 390 210 – 180 0.90 600 900 265 655 � 660 3.6 74.5FA-40 0.40 325 175 – 200 0.70 600 900 270 660 � 675 4.2 47.9BS-20 0.20 350 – 350 140 1.50 600 900 280 625 � 635 3.2 108.3BS-30 0.30 300 – 300 180 0.75 600 900 275 660 � 670 3.6 75.1BS-40 0.40 250 – 250 200 0.50 600 900 280 665 � 670 3.9 59.8

VF-30c 0.30 533 – – 160 1.90 656 944 – 560 � 560 – 63.4

a Dosages of SP are based on mass of total cementitious materials.b 28d fc stands for compressive strength of concrete at the age of 28 days.c A mixture is prepared for verifying the autogenous shrinkage equation.

262 C. Jiang et al. / Construction and Building Materials 61 (2014) 260–269

resistance strain gauges (ERSG) stuck on surface of the specimen is used to monitorshrinkage strain subsequently. It is proved in latter section that this method hasgreat practical value, enough precision and good reproducibility.

2.3.2. Testing detailsAs shown in Fig. 1(a) and (b), two identical sensor supports are installed and

fastened on the top of steel prism mould with interior size of100 � 100 � 400 mm. The ECDS are fixed on the sensor supports, and theirpositions could be regulated previously. Two U-shaped invar steel target seats arepositioned properly on the bottom surface of the steel mould (see also Fig. 1(c)).After casting the HPC mixtures, target seats will be embedded in concrete specimen,and will deform simultaneously with concrete. Standard targets are magneticallyattached on target seats (see also Fig. 1(f)), respectively.

The standard targets and the corresponding ECDS constitute fundamentalelements for stirring up eddy-current, and any tiny deformation of concretespecimen causing by self-desiccation, temperature and moisture gradients, etc.,can be monitored by signal conversion device, which transforms electric signalsinto displacements. Intervals of measurement data acquisition can be set freely.

Fig. 1. Autgenous shrinkage test method from initial set to 3-day age: (a) top view and (mixture casted into steel mould; (e) installation of ECDS on surface of the mould and (f

To minimize friction between inner surfaces of steel mould and concretespecimen, Teflon sheets are placed on the inner surfaces of steel mould, as shownin Fig. 1(c). Also, two buckles are used to fix the positions of target seats beforemonitoring of concrete deformation, and removed at the moment of measure-ment. Top surface of the concrete specimen is sealed by two layers of polyethyl-ene films (see Fig. 1(d)) so that the measured linear strain not contains thecomponent due to exterior drying (defined as total deformation). Correspondingto Fig. 1(a), (b) and (e) shows the state of specimen and testing apparatusimmediately before deformation monitoring. There are also close-up shots ofthe U-shaped target seat and the installed ECDS, as shown in Fig. 1(f) and (g),respectively. Round holes in the target seat make sure its anchorage effect in con-crete and uniformity of specimen. It is worth mentioning that the usage of gas-kets attached to target seat is minimizing friction between the target seat andbottom surface of the mould.

In order to separate autogenous shrinkage from the measured total deforma-tion, thermal deformation must be determined and offset. Therefore, temperaturehistory at the center of concrete specimen was concurrently monitored with athermocouple.

b) front view; (c) steel mould with pre-attached U-shaped target seats; (d) concrete) standard target attached upon target seat; (g) an amplification of ECDS.

C. Jiang et al. / Construction and Building Materials 61 (2014) 260–269 263

Concrete specimen is demolded and wrapped up using polyethylene film andaluminum foil sequentially at the age of 3 days. Two pieces of 3-wire ERSG (withgauge length of 90 mm and electric resistance of 120 X) are stuck on two lateralsurfaces of specimen carefully, and then sealed the ERSG completely. The measure-ment period lasts up to the age of 60 days in the study.

In addition, curing and measurements of linear strain of volume change are con-ducted at walk-in isothermal chambers with temperatures of (10 ± 2), (20 ± 2) and(30 ± 2) �C, and relative humidity (RH) of (60 ± 5)%, respectively.

2.4. Setting time

A technical committee on autogenous shrinkage of the Japan Concrete Institute(JCI) defined autogenous shrinkage as the macroscopic volume reduction of cemen-titious materials when cement hydrates after initial setting [3,13]. This definitionimplies that the initial setting time is the start point of autogenous shrinkage.Although there are a series of testing methods for setting time of concrete [16,17],penetration resistance method as prescribed in ASTM C403 which is simple and easyto handle is conducted in this study. To reduce moisture evaporation, a polyethylenefilm and a damp cloth are covered sequentially on the surface of the sample. Thesamples are cured in the same condition for autogenous shrinkage specimens.

2.5. Compressive strength

Compressive strength of the HPC is measured using 100 mm cubes specimens atthe ages of 1, 3, 7, and 28 days, respectively. The cubes in moulds are firstly storedin the above-mentioned chambers for 24 h and then stripped, and placed in wateruntil the age of testing.

3. Results and discussion

3.1. Validation of shrinkage measurement method

To verify the effectiveness and reproducibility of the modifiedmethod for measuring autogenous shrinkage, a trial and errorwas performed before put into practice, and results are shown inFig. 2. It is find that desirable reproducibility and stability can beachieved. In Fig. 2(a), two curves of linear strain obtained fromtwo replicate specimens based on the ECDS coincide with each

Fig. 2. Reproducibility and stability and of the shrinkage test method used: (a) shrinkageand (b) shrinkage strain on 3 different lateral sides of the same specimen, obtained from

Table 4Initial and final setting time (h) of the HPC mixtures used.

Mix. ID OP-20 OP-30 OP-40 FA-20

Curing temperature of 10 �CInitial setting 4.8 10.3 11.1 5.2Final setting 11.1 14.6 16.1 11.2

Curing temperature of 20 �CInitial setting 4.0 7.4 8.2 4.9Final setting 7.4 9.8 11.7 10.2

Curing temperature of 30 �CInitial setting 3.6 4.2 5.8 3.6Final setting 5.8 6.2 8.7 6.2

other well. After that, there is about 3.5% of maximum deviationbetween the two replicate specimens named as ‘‘specimen_1’’and ‘‘specimen_2’’. As far as ERSG measurements are concerned,3 curves plotted from 3 groups of strain data on the same specimenshow similar trends and good uniformity (see Fig. 2(b)). Although‘‘ERSG-3’’ slightly diverges from ‘‘ERSG-1’’ and ‘‘ERSG-2’’, if system-atic error is considered, their deviations are acceptable.

3.2. Basic properties of HPC mixtures

3.2.1. Setting timeInitial and final setting times of the HPC mixtures used are tab-

ulated in Table 4. These experimental data show that the higherthe curing temperature and the lower the w/b, the earlier the sethappens. Compared to the control mixtures, the mixtures contain-ing FA and BS display evident setting retardation, which resultsfrom minor pozzolanic activity of FA and BS at very early ages. Itseems as if influences of curing temperature on setting time ofthe higher w/b mixtures is greater than that of the lower w/b ones.Also, the effects of curing temperature on final setting time aremore significant than those on initial setting time.

3.2.2. Temperature characteristicAlthough constant curing temperatures are imposed on the HPC

in this study, the temperature in specimens are not necessarily con-stant. Since heat of hydration releases, temperature in specimenrise and then drop with heat exchange between specimen and sur-roundings. Peak values and rates of temperature rise are dependenton a series of physical parameters such as the phase compositionand fineness of the cement, mineral admixtures employed, heatcapacity, thermal conductivity of concrete mixtures [6].

As shown in Fig. 3, the mixtures with lower w/b generallycorrespond to higher peak values and rates of temperature rise.

strain on two identical specimens, obtained from non-contact shrinkage apparatus3 pieces of ERSGs, respectively.

FA-30 FA-40 BS-20 BS-30 BS-40

10.0 14.0 10.3 12.0 14.014.8 18.6 15.8 17.6 18.9

8.0 8.9 5.9 7.3 9.111.0 12.8 11.0 11.0 14.2

4.5 6.5 3.7 4.9 6.96.8 8.8 6.9 7.3 9.2

264 C. Jiang et al. / Construction and Building Materials 61 (2014) 260–269

Shoulders appeared in curves of temperature history demonstratethat the addition of FA and BS alleviates temperature rise signifi-cantly. The curves have obvious distinction under different curingtemperatures. Because hydration rate is much slower under lowercuring temperature, temperature rise value (i.e., the difference be-tween peak and initial values) and temperature rise rate are lower.

On the one hand, temperature variation results in thermal defor-mation, which fully coupled with autogenous shrinkage. On theother, temperature affects the hydration and self-desiccation ofcementitious materials, which further affects autogenous shrinkage.

Furthermore, at early-age stage, non-stable thermal expansioncoefficients of concrete [12,18,19] make things much more

Fig. 3. Evolution of temperature at the center of concrete specimen: (a) curing temperatusubsequent figures are interpreted in form of schematic.

Table 5Compressive strength (MPa) of HPC mixtures used.

Mix. ID OP-20 OP-30 OP-40 FA-20

Curing temperature of 10 �C1d 40.4 23.8 7.4 16.83d 92.4 55.3 33.1 55.37d 94.8 64.7 46.6 71.028d 99.2 74.6 62.6 89.9

Curing temperature of 20 �C1d 65.5 42.2 22.6 37.93d 93.6 62.3 41.7 65.97d 98.1 74.5 52.5 78.128d 115.3 85.7 71.3 94.3

Curing temperature of 30 �C1d 84.7 55.3 35.1 59.23d 95.4 64.3 47.7 78.77d 97.4 70.3 54.0 85.328d 103.9 80.6 74.7 115.2

complicated. For simplicity, the measurement of autogenous strainwas corrected for thermal strain by assuming that the thermalexpansion coefficient of concrete was 10 � 10�6/�C. The correctionis calculated assuming that no temperature gradient existedbetween the surface and the core of the specimen because of itssmaller cross section.

3.2.3. Compressive strengthCompressive strength of all 10 mixtures at specific ages under

different curing temperatures is tabulated in Table 5. The followingconclusions can be drawn from these testing data: (1) lower w/bmixtures obtained higher strength; (2) the development rate of

re of 10 �C, (b) 20 �C and (c) 30 �C for the 9 HPC mixtures. Legends in the graphs and

FA-30 FA-40 BS-20 BS-30 BS-40

13.6 5.2 20.7 8.1 2.736.0 18.4 54.6 28.6 13.648.6 28.5 72.4 48.1 24.972.1 46.2 87.1 66.2 42.7

34.8 7.2 13.5 11.0 5.143.9 21.4 56.7 33.9 21.257.7 33.2 95.5 53.9 38.574.5 47.9 108.3 75.1 59.8

34.8 18.8 45.0 27.1 14.256.1 33.3 88.3 60.3 35.361.0 44.5 104.3 76.9 44.591.3 67.5 116.1 91.0 70.7

C. Jiang et al. / Construction and Building Materials 61 (2014) 260–269 265

lower w/b mixtures is faster than that of higher w/b mixtures; (3)the effects of curing temperature on strength of the mixturescontaining FA/BS are greater than these of the control mixturesand (4) higher curing temperature is beneficial to strengthdevelopment of the mixtures containing FA/BS.

3.3. Autogenous shrinkage

3.3.1. Effects of w/b on autogenous shrinkageIt is supposed that the deformation before initial setting has

negligible influence on stress development in a restrained concrete

Fig. 4. Effects of w/b on early-age autogenous shrinkage: (a) age-dependent curves otemperature of 30 �C and (b) autogenous shrinkage ratio (i.e., the ratio of autogenous shrcuring temperature of 30 �C.

Fig. 5. Effects of mineral admixtures on early-age autogenous shrinkage: (a) mixtures witmixtures with w/b = 0.30 under 30 �C curing and (d) mixtures with w/b = 0.40 under 30

member, so the autogenous strain is zeroed at the initial set for allmixtures under different curing temperatures. Test results for con-trol mixtures in Fig. 4 show that lower w/b HPC mixtures exhibitedlarger autogenous shrinkage than higher w/b mixtures, and thatthe increasing rate of the former is superior to the latter. This isin agreement with those reported by Tazawa and Miyazawa [8],Lura [4] and Lim and Wee [17]. For all the concretes studied, 60%or more of the autogenous shrinkage strain up to 60 days occurredduring the first 2 weeks after casting. Thus, the early-age autoge-nous shrinkage should not be overlooked and underestimated, asthe concrete is most susceptible to cracking at early ages due to

f autogenous shrinkage strain for 3 control (pure cement) mixtures under curinginkage value at a specific age to the 60d value) vs. age for 3 control mixtures under

h w/b = 0.20 under 20 �C curing; (b) mixtures with w/b = 0.20 under 30oC curing; (c)�C curing.

266 C. Jiang et al. / Construction and Building Materials 61 (2014) 260–269

its low tensile strength and strain capacity. For the control mix-tures, although the same reduction of w/b happens, when itdecreases from 0.40 to 0.30 and from 0.30 to 0.20, resultedincreases of autogenous shrinkage are unnecessarily same. In otherwords, the increase of autogenous shrinkage is not proportional tothe decrease of w/b. In addition, these explanations and conclu-sions are also suitable to other mixtures under different curingtemperatures which not plotted in Fig. 4.

Occasionally, swelling can be observed in the FA and BS mix-tures, a contribution to this swelling could be re-absorption ofbleeding water. It has been observed that removing the bleedingwater reduces the swelling [3,4], but does not eliminate it totally.The residual swelling could be due to internal bleeding in the mix-ture. In the present research, a little external bleeding has beenobserved in BS and FA mixtures. Other possible explanations of thisphenomenon could be found at the scale of the hydration productsand early-age ettringite formation [4,9,10,20]. In a word, the early-age expansion is far from fully understood and it complicates theinterpretation of autogenous shrinkage measurements, leading toconfusion in the shrinkage behavior.

3.3.2. Effects of mineral admixtures on autogenous shrinkageAs partially shown in Fig. 5, all BS mixtures have higher autog-

enous shrinkage than the OP ones; all FA mixtures have lowerautogenous shrinkage than the OP ones with the same w/b andexposed to the same temperature. The higher autogenous shrink-age of concrete containing BS may be due to the greater chemicalshrinkage than that of the concrete with pure Portland cement.Thus, the greater chemical shrinkage led to faster and greaterself-desiccation, and results in larger autogenous shrinkage. More-over, the use of BS makes cement paste have a finer pore structure,

Fig. 6. Effects of curing temperature on autogenous shrinkage for: (a) all OP mixtures; (bmixtures at specific ages (1, 3, 28, 60 days) vs. curing temperature.

as confirmed by its lower permeability [1,2,17,30]. Finer pores con-tribute to a lower internal relative humidity, which increases thedegree of self-desiccation.

The behavior of autogenous shrinkage in FA mixtures seems tobe correlated with hydration characteristic of the pozzolanic mate-rial. With the substitution of cement by FA, internal relativehumidity of concrete decreases relatively slowly, self-desiccationmay not practically occurs, and consequently reduces autogenousshrinkage. Another reason for decrease of autogenous shrinkageis delayed hydration of FA (also be called as secondary hydrationor pozzolanic reaction), which must be premised on portlanditeproduced by cement hydration. The delayed hydration producesa negligible contribution to autogenous shrinkage because of im-proved stiffness of concrete. As illustrated in Tables 4 and 5,although FA often reduces autogenous shrinkage, it also bringsset retardation, lower early-age strength and accelerated carbon-ation [2,6,8].

As far as the shrinkage rate is concerned, it can be seen fromFig. 5, OP mixtures kept the maximum rate before the age of3 days, followed by BS mixtures, and FA mixtures show the slowestshrinkage rate. After that, the situation is changed, the autogenousshrinkage of the BS mixtures continued increase rapidly and whosedevelopment rate exceeded those of OP and FA ones. With lowerw/b and higher curing temperature, the characteristic is muchmore remarkable.

3.3.3. Autogenous shrinkage under different curing temperaturesFig. 6 shows the age-dependent curves of measured autogenous

shrinkage cured at 3 different temperatures. Both the rate and themagnitude of autogenous shrinkage are significantly varied withthe curing temperature. In general, higher curing temperature

) all FA mixtures; (c) all BS mixtures; (d) autogenous shrinkage strain for w/b = 0.20

C. Jiang et al. / Construction and Building Materials 61 (2014) 260–269 267

results in greater autogenous shrinkage and its increasing rate. Theinfluences of curing temperature are much more significant atearlier ages than at later ages. Also the influences of elevatedtemperature on autogenous shrinkage are more obvious in themixtures with lower w/b than the ones with higher w/b. This isdue to much higher binder contents in lower w/b mixtures. In low-er w/b mixtures, considerable self-desiccation during first severaldays after casting will inevitably generate greater autogenousshrinkage. Thereafter, the rate of development of autogenousshrinkage becomes almost same regardless of curing temperatures.In terms of early-age cracking risk, the rate of the autogenousshrinkage is even more important than its magnitude, so it couldbe speculated that higher curing temperatures increase the crack-ing risk, since the shrinkage develop more quickly. Unfortunately,cross-over effect stated by Chu et al. [7,21,26] cannot be observedin the relatively short period in this study.

It seems as if the effects of elevated curing temperature onautogenous shrinkage of concrete mixtures with different cemen-titious systems are incompatible. As shown in Fig. 6(d), the incre-ment of autogenous shrinkage of the OP-20 mixtures due to curingtemperature elevated from 20 �C to 30 �C is greater than elevatedfrom 10 �C to 20 �C, regardless of specific ages. However, the ten-dency is reversed for the BS-20 mixture. When it comes to theFA-20 mixtures, no consistent tendency is existed. Relativelyspeaking, autogenous shrinkage of FA mixtures is the least influ-enced by variation of curing temperature in 3 series of mixtures.Also, it is interesting that the change of autogenous shrinkage isnegligible when curing temperature elevated from 10 �C to 20 �Cfor mixtures with higher w/b and mineral admixtures inclusion.These phenomena can be explained from mechanism of hydrationand potential reactivity of different cementitious materials andtheir temperature dependence.

Fig. 7. Autogenous shrinkage as a function of equivalent age at different temperatures f

Based on Arrhenius’s principle [10,11] and the equivalent ageequation [7,10] proposed by Hansen and Jensen, autogenousshrinkage of a series of typical HPC mixtures other than cementpaste or mortar under different curing temperatures are evaluated,as partially shown in Fig. 7. Except the OP-30 mixture, autogenousstrain curves of the same mixture respectively exposed to 10, 20and 30 �C do not coincide with each other. It should be noted thatmajor deviations existed among three autogenous shrinkage evo-lution curves for mixtures with lower w/b and inclusion of mineraladmixtures. As a consequence, it is apparent that the temperaturedependence of autogenous shrinkage cannot be simply describedby the time–temperature transformations, as is also presented inprevious work [1,2,4,10]. As Hansen, Sant and Weiss et al.explained, although the maturity concept can reasonably appliedin interpretation of cement hydration and related properties underdifferent temperatures, with respect to temperature dependence ofautogenous shrinkage, it seems difficult to apply [4,15,27]. Whenthe maturity concept is applied to cementitious systems, it is oftenincorrectly assumed that the overall response depends exclusivelyon the extent of reaction and their microstructure formationprocesses are not influenced by temperature. Obviously, it isunreasonable as shown in many related study [4,9,15,27], self-desiccation and autogenous shrinkage of HPC mixtures arestrongly related to its microstructure evolution that is strongly af-fected by thermodynamics. In addition, it is doubtful that constantapparent activation energy is employed in the equivalent ageequation. Some authors think that apparent activation energy ofhydration is not independent on temperature, age and bindercomposition [7,13,22,23]. Therefore, in order to correctly evaluateautogenous shrinkage under different curing temperatures andtemperature history, new more reasonable theories and modelsare needed.

or (a) OP-20 mixture; (b) OP-30 mixture; (c) FA-20 mixture and (d) BS-30 mixture.

268 C. Jiang et al. / Construction and Building Materials 61 (2014) 260–269

3.3.4. Estimation of autogenous shrinkage considering curingtemperature and mineral admixtures

Many authors have tried to predict and estimate autogenousshrinkage in terms of some basic parameters (e.g., w/b, compres-sive strength and porosity) and obtained experimental data[3,13,19,23,25]. Due to differences in concrete making materials,testing methods and conditions, each prediction model is inevita-bly applicable to specific conditions, a unified equation is impossi-bly constructed. To explore estimation of autogenous shrinkageconsidering the effects of curing temperature and mineral admix-tures, and to verify the obtained data, a trial and error based onthe following prediction model proposed by Tazawa [3,17,24] isperformed.

eASðt;20Þ ¼ c � eAS;28 � bðtÞ ð1Þ

eAS;28 ¼ �3070 � exp½�7:3 � ðw=bÞ� ð2Þ

bðtÞ ¼ exp a � 1� 28� tis

t � tis

� �� �b( )

ð3Þ

where eAS (t,20) (10�6) is autogenous strain responding to curingtemperature of 20 �C at the age of t (days), c is a coefficient todescribe the effect of mineral admixtures, eAS,28 (10�6) is the 28-day autogenous strain of pure OPC concrete (i.e., OP mixtures in thispaper) cured at 20 �C, b(t) is an age-dependent function of autoge-nous shrinkage rate at 20 �C, w/b is water–binder ratio, tis (days)is initial setting time; and a and b are constants in terms of numer-ical fitting of experimental data.

As shown in Eqs. (1)–(3), the Tazawa’s prediction model hasfailed to account for the temperature dependence of autogenousshrinkage. In order to include temperature-dependence, a modifi-

Fig. 8. Estimation of autogenous shrinkage by a modification on a previous model: (a) Opresent model. It is noteworthy to mention that fitting values for the constants (c, a, b,

cation is made by comparing the differences of shrinkage strain be-tween cured at 20 �C and other temperatures. The modified modelis expressed in the following equation (Eq. (4)), i.e., a power itemincluding the temperature effects is added to the original model.

eASðt; TÞ ¼ eASðt;20Þ � 1þ c

ðt � tisÞðT�20Þ=d

" #ð4Þ

where T (�C) is actual curing temperature of concrete, and c and dare constants in terms of numerical fitting of experimental data.When T equals to 20 �C, it is assumed that the constant c equalsto zero. It is supposed that ultimate degree of hydration roughlythe same regardless of curing temperature and the same degreeof hydration responds to the same autogenous shrinkage, it is rea-sonable that the same limit-value of autogenous shrinkage strainunder various temperatures can be achieved in the model.

All constants in Eqs. (1)–(4) are obtained by means of numericalregression on the measured data. On account of a lack of theoreti-cal evidence, maybe the modification is somewhat arbitrary. How-ever, to a certain extent, the mathematical trends demonstrated bythe test data can be expressed directly. As an example, Fig. 8 showsthe comparison between estimated values and the experimentalresults, the both in agreement with each other approximately.Additionally, the mixture marked with ‘‘VF-30’’ in Table 3 is usedfor the verification of the modified equation. The comparisons be-tween evaluated values and test results are presented in Fig. 8(d).

4. Summary and conclusions

As a concluding remark, the following points from this study arehighlighted:

P-20, (b) FA-40 and (c) BS-30 mixture; (d) the VF-30 mixture is used to verify thec and d) in Eqs. (1)–(4) are also attached.

C. Jiang et al. / Construction and Building Materials 61 (2014) 260–269 269

(1) For the purpose of accurately measuring linear autogenousstrain initiated from setting, the previous method ismodified with a high effective eddy-current displacementsensor.

(2) The mixtures containing FA or BS display evident settingretardation. Higher peak values and rates of temperaturerise correspond to the mixtures with lower w/b. The addi-tion of FA and BS significantly decreases temperature rise.The effects of curing temperature on strength of the mix-tures containing FA/BS are greater than those of the controlones.

(3) Lower w/b mixtures exhibit larger autogenous shrinkageand faster development. The inclusions of BS and FA resultin increase and decrease of autogenous shrinkage, respec-tively. While both the rate and the magnitude of autogenousshrinkage for all mixtures are increase with rise of curingtemperature, their extents of influence are vary with w/b,and composition of cementitious materials. The temperaturedependence of autogenous shrinkage cannot be simplydescribed by the equivalent age function.

(4) A modified model for estimating autogenous shrinkage ofHPC is constructed, but its generalizability is a challengedue to limited data and undefined physical implications.

Acknowledgements

The authors acknowledge the National Natural Science Founda-tion of PR China (Project No. 50979098) and Department of Educa-tion of Zhejiang Province (Project No. Y201330182) for financiallysupporting this work, respectively. We are most grateful to anony-mous reviewers for useful comments that helped to improve thisarticle.

References

[1] Aïtcin P-C. High-performance concrete. E & FN SPON; 1998.[2] Mehta PK, Monteiro JM. Concrete: structure, properties, and materials. 2nd

ed. New Jersey: Prentice Hall; 1993.[3] Institute Japan Concrete. Technical committee report on autogenous

shrinkage. In: Tazawa, editor. Proc. of int. workshop. London: E&FN Spon;1998. p. 1–63.

[4] Lura P. Autogenous deformation and internal curing of concrete. PhD Thesis,Delft University of Technology, Delft; 2003.

[5] Holt EE, Leivo MT. Cracking risk associated with early age shrinkage. CemConcr Compos 2004;26(5):521–30.

[6] Bentz DP. A review of early-age properties of cement-based materials. CemConcr Res 2008;38(7):196–204.

[7] Chu I, Kwon SH, Amin MN, Kim JK. Estimation of temperature effects onautogenous shrinkage of concrete by a new prediction model. Constr BuildMater 2012;35(10):171–82.

[8] Tazawa E, Miyazawa S. Influence of cement and admixture on autogenousshrinkage of cement paste. Cem Concr Res 1995;25(10):281–7.

[9] Sant G. The influence of temperature on autogenous volume changes incementitious materials containing shrinkage reducing admixtures. Cem ConcCompos 2012;34(7):855–65.

[10] Jensen OM, Hansen PF. Influence of temperature on autogenous deformationand relative humidity change in hardening cement paste. Cem Concr Res1999;29(4):567–75.

[11] Arrhenius S. On the reaction velocity of the inversion of cane sugar by acids. ZPhys Chem 1889;11(4):226.

[12] Maruyama I, Teramoto A. Temperature dependence of autogenous shrinkageof silica fume cement pastes with a very low water-binder ratio. Cem ConcrRes 2013;50(3):41–50.

[13] Institute Japan Concrete. Technical committee report on autogenous shrinkageof concrete. In: Tazawa E, editor. Proc. int. workshop. Hiroshima (Japan): E &FN Spon; 1998. p. 1–62.

[14] Sant G, Lura P, Weiss J. Measurement of volume change in cementitiousmaterials at early ages: review of testing protocols and interpretation ofresults. Transport Res Board 2005:183–6.

[15] Eddy-current-sensors. <http://www.lionprecision.com/inductive-sensors/eddy-current-sensors.html>; 2011 [Last verified 22.06.11].

[16] Lee KM, Lee HK, Lee SH, Kim GY. Autogenous shrinkage of concrete containinggranulated blast-furnace slag. Cem Concr Res 2006;36(7):1279–85.

[17] Lim SN, Wee TH. Autogenous shrinkage of ground granulated blast furnaceslag concrete. ACI Mater J 2000;97(6):587–93.

[18] Jung HY, Seong C, Moon CW. In situ measurement of coefficient of thermalexpansion in hardening concrete and its effect on thermal stress development.Constr Build Mater 2013;38(7):306–15.

[19] Jiang CH, Yang Y, Li P. Time dependence of thermal expansion behavior ofcement mortar at early ages. J Chin Ceram Soc 2013;41(5):33–9.

[20] Gao PW, Xu SY, Chen X. Research on autogenous volume deformation ofconcrete with MgO. Consr Build Mater 2013;40(3):998–1001.

[21] Viviani M, Glisic B, Smith IFC. Separation of thermal and autogenousdeformation at varying temperatures using optical fiber sensors. Cem ConcrCompos 2007;29(6):435–47.

[22] Jiang CH, Yang Y, Wang XD. A modified equivalent age equation for estimatingtemperature dependence on properties of young HPC. Concrete2013;48(11):25–30 [in Chinese].

[23] Yoo SW, Kwon SJ, Jung SH. Analysis technique for autogenous shrinkage inhigh performance concrete with mineral and chemical admixtures. ConsrBuild Mater 2012;34(9):1–10.

[24] Lee HK, Lee KM, Kim BK. Autogenous shrinkage of high performance concretecontaining fly ash. Mag Concr Res 2003;55(10):507–15.

[25] Zhang MH, Tam CT, Leow MP. Effect of water-to-cementitious materials ratioand silica fume on the autogenous shrinkage of concrete. Cem Concr Res2003;33(4):1687–94.

[26] Lura P, Breugela K, Maruyama I. Effect of curing temperature and type ofcement on early-age shrinkage of high-performance concrete. Cem Concr Res2001;31(6):1867–72.

[27] Lura P, Sant G, Weiss WJ. The influence of temperature on volume changes andresidual stress development in cementitious materials. In: Proceedings of the2nd international RILEM conference on concrete durability and service lifeplanning, Haifa, Israel; 2009.

[28] Mehta PK. High-performance, high-volume fly ash concrete for Sustainabledevelopment. In: Proceedings of international workshop on sustainabledevelopment and concrete technology, Beijing, China; May 20–21, 2004.

[29] Yusuf MO, Johari MAM, et al. Strength and microstructure of alkali-activatedbinary blended binder containing palm oil fuel ash and ground blast-furnaceslag. Constr Build Mater 2014;52(2):504–10.

[30] Li Q, Li Z, Yuan G. Effects of elevated temperatures on properties of concretecontaining ground granulated blast furnace slag as cementitious material.Constr Build Mater 2012;35(10):687–92.

[31] Philip VH, Belie ND. A service life based global warming potential for high-volume fly ash concrete exposed to carbonation. Constr Build Mater2012;55(3):183–93.