Construction and Building Materials 61 (2014) 41–49

Contents lists available at ScienceDirect

Construction and Building Materials

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

Effect of internal curing on internal relative humidity and shrinkageof high strength concrete slabs

http://dx.doi.org/10.1016/j.conbuildmat.2014.02.0600950-0618/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +86 10 62797422.E-mail address: junz@tsinghua.edu.cn (J. Zhang).

Yudong Han a, Jun Zhang a,⇑, Yiming Luosun a, Tingyu Hao b

a Key Laboratory of Structural Safety and Durability of China Education Ministry, Department of Civil Engineering, Tsinghua University, Beijing 100084, Chinab Central Research Institute of Building and Construction, MCC Group Co. Ltd., Beijing 100088, China

h i g h l i g h t s

� Internal relative humidity in high strength concrete slabs with and without internal curing is experimentally measured.� Efficiency of internal curing on shrinkage reduction in high strength concrete slab is evaluated.� Internal curing can greatly improve moisture status and shrinkage development in high strength concrete slab.

a r t i c l e i n f o

Article history:Received 25 December 2013Received in revised form 12 February 2014Accepted 17 February 2014

Keywords:High strength concreteInternal curingRelative humidityShrinkage

a b s t r a c t

In the paper, the moisture content, in terms of relative humidity, inside of high strength concrete slab,with and without addition of pre-soaked lightweight aggregate (PSLWA), exposed to normal indoorenvironment is investigated by continually measuring the interior humidity of concrete immediatelyafter slab casting until 28 days. The effects of internal curing on the developments of internal relativehumidity and free shrinkage in high strength concrete slab are analyzed. The experimental results showthat the internal relative humidity of high strength concrete decreases with concrete age since casting.The developing tendency of the relative humidity inside of concrete with age follows a vapor-saturatedstage with 100% relative humidity (stage I) and a stage that relative humidity gradually decreases (stageII). A humidity gradient along the thickness of slab exists in early age high strength concrete. As PSLWA isadded, the reduction rate of interior humidity in stage II is obviously reduced and the reduction trendwith age changes from non-linear pattern to almost linear pattern. The duration of the internal humid-ity-saturated stage I is noticeably prolonged by addition of PSLWA. The humidity gradient together withthe corresponding shrinkage gradient along the thickness of the slab is reduced also. The highestreduction on internal humidity at 28 days since casting in high strength concrete slab is changed from46.5% to 26.2% with a medium PSLWA addition ratio, and finally to 7.9% with a high PSLWA addition level.Within present addition range, the more the PSLWA added, the stronger the above positive internalcuring effects.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

The moisture content in the concrete pores is a critical param-eter for most of the degradation processes suffered by concrete,for example concrete shrinkage and shrinkage related cracks.Concrete shrinks as moisture is lost to the environment or byself-desiccation. As concrete shrinks, tensile stresses will developin the structure due to restraints from adjacent materials, con-nected members or shrinkage gradient. The stresses may overcomethe tensile strength and lead the concrete to crack. The magnitude

of shrinkage strain is normally proportional to the amount of mois-ture lost [1–4]. The evaluation of the shrinkage induced stresses inthe structure requires the knowledge of the distribution ofshrinkage strain, which, in turn needs the information of moisturedistribution first. The local shrinkage can directly relate to the porehumidity [5–7].

Normally, the autogenous shrinkage of concrete is inverselyproportional to water to cement ratio. The higher the water tocement ratio, the lower the autogenous shrinkage. High strengthconcrete generally has a low water to cement ratio, thereforemarked autogenous shrinkage may occur in high strength concretestructures [6,7]. In addition, a similar problem may emerge in highperformance concrete due to its high content of cementitious

(a)

0 1 2 3 4 5 6 7 8 9 10Diameter (mm)

0

10

20

30

40

50

60

70

80

90

100Pe

rcen

t pas

sing

(%)

(b)

Fig. 1. Lightweight aggregate used in experiments, (a) photographer and (b) sievingcurve.

42 Y. Han et al. / Construction and Building Materials 61 (2014) 41–49

materials normally. In order to avoid the shrinkage induced crack-ing in high strength concrete or high performance concrete struc-tures, it is necessary to compensate for the moisture lossresulting from cement hydration. Use of pre-soaked lightweightaggregate (PSLWA) as an internal reservoir to provide water asthe concrete dries is an effectual method to reduce autogenousshrinkage of high strength concrete [8,9]. Furthermore, moisturecontent in pores directly affects strength, thermal properties andthe rate of cement hydration. It plays a role in the durability prob-lems and fire resistance [10]. In addition, frequent cracking of con-crete structures at early ages has indicated that the early-age is oneof the most critical periods of the life time of the cementitiousmaterials [11]. Therefore, the moisture content and its distributioninside of concrete, with and without taking the action of internalcuring with pre-soaked lightweight aggregate, are critically neededin order to evaluate the effectiveness of internal curing on the mit-igation of shrinkage induced cracking and on the improvements ofdurability and service-life of concrete structures.

The purpose of this paper is to investigate the effect of internalcuring using PSLWA on moisture content and its distribution ofhigh strength concrete and to contribute to the understandingand designing of concrete mixtures including the considerationof internal curing. In the paper, the moisture content, in terms ofrelative humidity (RH), inside of high strength concrete slab, withand without the addition of PSLWA, exposed to normal indoorenvironment was investigated by continually measuring theinterior humidity and temperature of concrete immediately afterslab casting until 28 days. The effect of internal curing on thedevelopments of internal relative humidity and free shrinkage aswell in high strength concrete slab is analyzed.

2. Experimental program

2.1. Details of materials and specimen

Basic concrete mixture with water to cement ratio of 0.33 was used in theexperiments. Based on the above basic mixture, two mixtures with different con-tents of PSLWA were utilized. Mixtures of concrete containing lightweight aggre-gate were designed according to the required amounts of internal curing waterbased on the above basic mixture proportion. Slightly different water to cement ra-tios compared to the reference was used for the mixtures with PSLWA in order tomaintain a comparable compressive strength among the three mixtures. Fly ashbased lightweight aggregate with the particle size of 2.5–6.5 mm, porosity of0.37, water absorption in weight of 21% after 7 days’ soaking, dry density of1375 kg/m3 was used as the carrier of internal curing water. Fig. 1 is the photographand sieving curve of the lightweight aggregate used in experiments. All mixtureswere made with the same normal Portland cement. Natural sand and crushed lime-stone with a maximum particle size of 5 mm and 20 mm, respectively, were used asnormal fine and coarse aggregates. The concrete mixture proportions used in thepresent work are listed in Table 1, where WIC/C is the mass ratio between the intro-duced water in PSLWA and the cement. A polycarboxylate superplasticizer with 30%solid content was used in all three mixtures to guarantee the fresh concrete havinga comparable slump of 90–120 mm. In the slab tests, one dimensional heat andmoisture transportation in concrete were created. Waterproof plywood mold withnet inner dimensions of 200 � 200 � 800 mm was used. To ensure the heat andmoisture to transfer only along the specimen thickness direction, the bottom andthe inner transverse and longitudinal sides of the mold were covered with fivepolystyrene boards to prevent heat loss and further with plastic sheets to preventmoisture loss. The casting face was immediately suffered to dry after finishingthe surface.

2.2. Devices for measuring the relative humidity

In this study, a digital resistance based sensor that can measure humidity andtemperature at the same time was used for relative humidity and temperature mea-surements. The measuring accuracies of relative humidity and temperature are 3%and 0.5 �C respectively. The signals from the sensor can automatically be recordedby a computer. The humidity and temperature were recorded every 10 min. In orderto keep the sensor to be at the appointed location in concrete, a PVC tube with aninner diameter of 15 mm was used to hold the sensor. One end of the PVC tube wascovered with a plastic sheet glued to the end. To keep the moisture exchanging withsurrounding concrete, three rectangular holes were made on the surface of the PVCtube. In order to prevent the fresh cement paste flowing into the PVC tube from

these oblong holes, a steel bar with a diameter a little smaller than the plastic tubewas filled into the tube first during concrete casting. After a few minutes of concretecasting, the steel bar was removed from the tube and the sensor was put in. To en-sure the measured humidity reflecting the real values inside concrete, two O-ringswith a 2 mm thickness were used to isolate the unoccupied space between the PVCtube and the sensor bar. The O-ring was slightly above the sensory section of thesensor (about 1 cm in length). In the mean time, at the top of the tube, the gapbetween the PVC tube and the sensor bar was sealed by an industry sealant todouble isolate moisture loss through the gap. Thus, the sensory section is onlyconnected to the concrete where we wish to measure the humidity. The detailedpositions of the sensors and the PVC tubes in the specimen are shown in Fig. 2.

2.3. Concrete mixing and specimen casting procedures

The LWA was soaked in water for 7 days. Before concrete mixing, the pre-soaked LWA was surface dried with a wet towel in the laboratory and kept in asealed container for use. The concrete mixing procedure can be described as fol-lows. First, the fine and coarse aggregates were mixed together. Next, the cementwas added followed by the required water with the supperplasticizer mixed inand the mixing was continued for 3 min. Then prepared LWA were added and con-tinued for another 2 min of mixing. Afterwards the fresh concrete was cast into themold in two layers and consolidated by a needle vibrator. After compacting, the PVCtubes with steel bar filled in were put into concrete at appointed depths from thecasting surface. While inserting the tubes, the vibration hammer was started againto ensure that the PVC tube surfaces were well contacted with the surroundingconcrete. After about 30 min of concrete casting, the steel bars were removed from

Table 1Mix proportions of concrete.

No. Cement(kg/m3)

Water(kg/m3)

Sand(kg/m3)

Stone(kg/m3)

Silica Fume(kg/m3)

LWA(kg/m3)

Water providedby LWA (kg/m3)

Super-Plasticizer(kg/m3)

W/C WIC/C CompressiveStrength(MPa)

C-0 450.0 150.0 580.0 1140.0 50.0 0 0 7.47 0.33 0.00 93.1C-1 468.8 143.0 408.6 965.3 52.1 177.9 37.5 8.33 0.31 0.08 97.2C-2 494.4 133.5 308.9 863.7 54.9 281.3 59.3 8.85 0.27 0.12 93.5

Fig. 3. Slabs with installed humidity and temperature sensors.

50 60 70 80 90 100Theoretical value (%)

50

60

70

80

90

100

Mea

sure

d va

lue

(%)

Sensor 1Sensor 2Sensor 3Sensor 4

Fig. 4. Calibrating curves of RH sensors.

Y. Han et al. / Construction and Building Materials 61 (2014) 41–49 43

the tubes and the sensors were put into the tubes afterwards. Five sensors wereused to measure the humidity and temperature at five different depths inside ofthe slab and one was used to measure the humidity and temperature of the envi-ronment. After finishing the surface, the casting face was kept exposed to the air.A typical room without temperature and humidity control was used to store thetest specimens and equipments. The humidity and temperature in concrete wereautomatically recorded every 10 min by a computer since concrete casting. Aphotograph showing a test specimen with installed humidity sensors is displayedin Fig. 3.

After the tests, the humidity sensors were calibrated utilizing saturated saltsolutions under constant temperature with 22 �C. The saturated salt solutions wereprepared with distilled water and analytically pure salts. The sensory section of thesensor and saturated salt solution was placed in a small sealed container made ofthe corrosion resistant and non-hydrophilic material. For each calibration process,the final displayed relative humidity that does not change with time was used asthe measured value of the sensor under that precise moisture environment. Eachsensor was calibrated at six different humidity levels, and then it’s calibrating curvewas obtained. Based on these calibrating results, the measured humidity data wasrevised. Fig. 4 shows some calibrating curves of RH sensors.

3. Results and discussion

3.1. The development of internal humidity with time

The compressive strength of the two concretes at 28 days isgiven in Table 1. Apparently, the compressive strength of the threeconcretes is higher than 90 MPa and they can be classified as highstrength concrete. The experimental results of the development ofinternal relative humidity with time since casting at four or fivetarget depths from the casting surface are shown in Fig. 5a–c forconcretes with PSLWA addition levels of WIC/C = 0, 0.08 and 0.12respectively, labeled as C-0, C-1 and C-2 correspondingly.

From the results presented in the figures, first we can observethat for high strength concrete without PSLWA addition, theinterior humidity inside of concrete decreases with time fast andheavily. The development of the internal relative humidity of con-crete can be described as a water–vapor saturated stage with 100%relative humidity (stage I) followed by a stage that relative humid-ity gradually reduces (stage II). The length of the stage I (calledcritical time) is significantly influenced by the depth from

Fig. 2. Detailed set-up of humidity measurement.

0 7 14 21 28Age (days)

0

10

20

30

40

50

60

70

80

90

100

Rel

ativ

e hu

mid

ity (%

)

C-02.5cm5cm10cm18cmAir

(a) C-0

0 7 14 21 28

Age (days)

0

10

20

30

40

50

60

70

80

90

100

Rel

ativ

e hu

mid

ity (%

) C-1 1.5cm2.5cm5cm10cm18cmAir

(b) C-1

0 7 14 21 28

Age (days)

0

10

20

30

40

50

60

70

80

90

100

Rel

ativ

e hu

mid

ity (%

)

C-21.5cm2.5cm5cm10cm18cmAir

(c) C-2

Fig. 5. Variations of internal relative humidity at different locations in concreteslab, (a) C-0, (b) C-1 and (c) C-2.

44 Y. Han et al. / Construction and Building Materials 61 (2014) 41–49

specimen surface. In the present tests, the lengths of stage I atdepths of 2.5 cm, 5 cm, 10 cm and 18 cm away from the specimencasting surface for concrete C-0 are 168 h, 144 h, 98 h and 46 hrespectively. Within the range of 2.5–18 cm from the drying sur-face, the closer to the top of the slab, the longer the length of stageI. The highest reduction on relative humidity in the slab at 28 daysafter concrete casting is about 46.5%.

For concretes with PSLWA addition, that is C-1 and C-2, thereductions on internal humidity are considerably decreased anddelayed, as displayed in Fig. 5b and c. The highest humidity reduc-tion at 28 days in the C-1 concrete slab with internal curing level atWIC/C = 0.08 occurs at 1.5 cm place from the slab top and is onlyabout 26.2%. As the internal curing level increases to WIC/C = 0.12, the highest humidity reduction at 28 days in the C-2 con-crete slab occurs at 18 cm place from the slab top and is just about7.9%. Thus, the internal curing level and depth from the drying sur-face are the two factors influencing the overall developments anddistributions of interior humidity in the slab. On the other hand,the critical time at depths of 1.5 cm, 2.5 cm, 5 cm, 10 cm and18 cm becomes 360 h, 420 h, 636 h, 672 h and 84 h for C-1 con-crete slab and 476 h, 490 h, 672 h, 672 h and 211 h for C-2 concreteslab respectively. For comparison, the critical time for differentlocations in slabs with and without PSLWA is displayed in Fig. 6.Apparently, the effect of PSLWA addition on the progress of inter-nal humidity of high strength concrete is very significant, espe-cially on the length of moisture saturated stage. But it should benoted that the surface drying can still influence the decrease ofinternal humidity in the area that is close to the exposed top faceeven in concrete with PSLWA addition, because the internalhumidity at 1.5 cm depth from the slab top started to decrease inC-1and C-2 slabs within 28 days.

Second, for a given time, obvious difference in relative humiditybetween separate locations can be observed. Fig. 7a–c display theexperimental results in term of the relationship between relativehumidity and location in concrete slab with (C-0) and without(C-1 and C-2) PSLWA addition respectively at some typical timeafter casting. From the figures, we can observe that the distributionof water (represented by relative humidity) along the specimendepth decrease markedly with time in high strength concrete with-out PSLWA addition. This water distribution and its variations withage are the results of cement hydration and water diffusion andevaporation to environment. By contrast, as the PSLWA was addedin the concrete, the reduction on internal humidity becomes muchsmaller compared with that without PSLWA and the nonlinearityof the distribution of relative humidity occurs only around the

0 0.05 0.1 0.15 0.2Location (m)

0

200

400

600

800

1000

Crit

ical

tim

e(h

ours

)

C-0 TestC-1 TestC-2 TestFit

Fig. 6. The critical time versus position in concrete.

0 5 10 15 20

Position (cm)

40

60

80

100

Re

lativ

e h

um

idity

(%

)

C-0672, 576, 480, 384, 288, 168, 72h

(a) C-0

0 5 10 15 20

Position (cm)

40

60

80

100

Re

lativ

e h

um

idity

(%

)

C-1

672, 576, 480, 384, 288, 168,72h

(b) C-1

0 5 10 15 20

Position (cm)

40

60

80

100

Re

lativ

e h

um

idity

(%

)

C-2

672, 576, 480, 384, 288, 168,72h

(c) C-2

Fig. 7. Humidity profiles of concrete slab at different ages, (a) C-0, (b) C-2 and (c) C-2.

Fig. 8. Schematic diagram of concrete hum

Y. Han et al. / Construction and Building Materials 61 (2014) 41–49 45

places close to the drying face. The moisture environment insideconcrete is considerably improved by the addition of PSLWA. Thelarger the PSLWA addition, the higher the internal curing level,the greater the mitigation of self-desiccation and surface dryingeffect, see Fig. 7b and c. The water releasing from PSLWA mayeffectively compensate the water reduction resulting from cementhydration and/or diffusion or evaporation.

3.2. The development of shrinkage strain with time

It is understood that the shrinkage strain of concrete is inver-sely proportional to the pore humidity. The lower the pore humid-ity, the higher the shrinkage. The relationship between shrinkagestrain and internal humidity of above three concretes had beeninvestigated experimentally by Zhang and Han et al. [12]. In thetests, the mold used to cast the specimens was made of plexiglasswith inner dimensions of 60 � 100 � 400 mm. The four inner sidesof the mold were covered with four pieces of removable plasticsheets with 2 mm thickness and the bottom of the mold was cov-ered with a thin vinyl sheet with 1 mm thickness to reduce the fric-tional resistance between the mold and the concrete. After theinitial set of the concrete, the inner four removable plastic sheetswere lifted to create the ‘‘free restraint’’ conditions of the shrinkagetest. In this test, the humidity and temperature at the center of thetest specimen were measured. The method used for relativehumidity measurement is described in the last section. The defor-mation was measured by two linear variable differential transduc-ers (LVDTs) mounted on the two long ends of specimen. Themeasuring range of the LVDT is 2 mm and the measuring accuracyis 1 lm. To ensure the LVDTs were in good contact with the con-crete, two small cylinder bars were pre-cast into the concrete atthe centers of the two long ends of the specimen. The sensorybar of the LVDT was directly contact with the small bar during test-ing. The schematic diagram of concrete humidity and deformationmeasurement used in [12] is shown in Fig. 8.

The test results of the development of shrinkage strain andinternal relative humidity at the center of the test specimen is pre-sented in Fig. 9a–c for concretes C-0, C-1 and C-2 respectively, andcomplete plastic sheet sealing and five faces drying curing condi-tions were used in these tests. Apparently, the addition of PSLWAcan efficiently decrease the reduction of internal humidity of theconcrete and accordingly reduce the shrinkage strain as well. Usingthe experimental data shown in Fig. 9, a simple relationship be-tween shrinkage strain (e) and internal relative humidity (RH) ofthe concretes can be expressed as:

e ¼ e0 þ k½lnðRHÞ � lnð100Þ� ð1Þ

where e0 is the shrinkage generated during the humidity developingstage I (RH = 100%), which should principally be governed by waterto cement ratio. k is a constant. In the present study, e0 = 355.5 lm/m,k = �756.0 lm/m, e0 = 251.2 lm/m, k = �869.5 lm/m and e0 =200.0 lm/m, k = �350.3 lm/m for the concretes C-0, C-1 and C-2

idity and shrinkage measurement [12].

500

600

700

800

900

e (µ

m/m

)

C-0C-1C-2Model

46 Y. Han et al. / Construction and Building Materials 61 (2014) 41–49

respectively. The comparison between model and test results ispresented in Fig. 10. Thus, if we know the internal humiditydistribution in concrete, we may use Eq. (1) to calculate the freeshrinkage field and further to calculate the shrinkage induced stres-ses in structures as well.

Specially, from the shrinkage and internal humidity-age dia-grams, obvious two-stage pattern of the development of shrinkage

0

Age (days)

0

100

200

300

400

500

600

Shrin

kage

(µm

/m)

50

60

70

80

90

100

Relative hum

idity (%)

C-0Shrinkage-sealShrinkage-dryRH-sealRH-dry

(a) C-0

0

Age (days)

0

100

200

300

400

500

600

Shrin

kage

(µm

/m)

50

60

70

80

90

100

Relative hum

idity (%)

C-1Shrinkage-sealShrinkage-dryRH-sealRH-dry

(b) C-1

0

7 14 21 28 35

7 14 21 28 35

7 14 21 28 35

Age (days)

0

100

200

300

400

500

600

Shrin

kage

(µm

/m)

50

60

70

80

90

100

Relative hum

idity (%)

C-2Shrinkage-sealShrinkage-dryRH-sealRH-dry

(c) C-2

Fig. 9. Shrinkage and interior humidity versus time diagrams of concrete, (a) C-0,(b) C-1 and (c) C-2.

4.2 4.3 4.4 4.5 4.6 4.7Ln (RH)

0

100

200

300

400

Shrin

kag

Fig. 10. Simple relationship between free shrinkage and internal humidity.

strain versus age starting from concrete setting can be observed,The first stage shrinkage strain, e0 is normally developed withinthe humidity saturated stage (RH stage I) [12]. Using Eq. (1), theevolution of shrinkage strain during the humidity reduction stageII can be obtained. The predicted shrinkage strains versus timediagrams in the three concrete slabs are shown in Fig. 11a–crespectively, where the developments of relative humidity areshown together for comparison. Apparently, the shrinkage strainis greatly reduced due to the addition of PSLWA. Under the samecuring condition, the maximum shrinkage strain at 28 days fallsfrom 820 lm/m (without PSLWA for C-0 concrete slab) to500 lm/m (with WIC/C = 0.08 for C-1 concrete slab) and furtherto 230 lm/m (with WIC/C = 0.12 for C-2 concrete slab). In orderto see the effect of PSLWA on shrinkage gradient in the slab,Fig. 12a–c display the relationship between shrinkage strain andlocation in concrete without and with PSLWA at some typical ages.Clearly, distribution of shrinkage in early age high strength con-crete is varied not only with age, but also with position. Significantshrinkage gradients exist in early age high strength concrete struc-tures, which must be taken into consideration in shrinkage stresscalculations. Generally speaking, addition of PSLWA in highstrength concrete can reduce the shrinkage and make the distribu-tion of shrinkage strain more even relatively, especially in C-2concrete slab with a higher PSLWA addition. But even within28 days in concrete slab with PSLWA addition, a shrinkage gradientstill starts to emerge around the place close to the drying faceespecially in C-1 slab. This indicates that the internal curing cannevertheless not eliminate the surface cracking in the long rununder the surface drying status, although it can effectively reducethe internal cracking induced by autogenous shrinkage. In addi-tion, it should be pointed out that the addition of PSLWA maymerely reduce shrinkage generated in the humidity developingstage II effectively. The shrinkage strain occurred in the stage I,e0 in Eq. (1), is still noticeable after utilizing PSLWA in concrete,see Fig. 10. We are required to explore the techniques in the futurethat may effectively decrease the shrinkage strain of e0.

4. Mechanism analyses

Concrete is a multiphase system where the voids of the skeletonformed by aggregates and cement powders are filled with liquidwater and partly with a gas mixture composed of air and water va-por. The relative humidity inside of concrete is the relative humid-ity of the pore where the gaseous phase in equilibrium with theinterstitial liquid phase. Therefore, the initial amount of water inthe pores after concrete casting and the rate of the pore water lossshould be the two factors governing the development of relative

Y. Han et al. / Construction and Building Materials 61 (2014) 41–49 47

humidity in concrete. In general, the size and amount of the voidsor pores are controlled by the water to cement ratio, which in turncontrols the concrete strength. High water/cement ratio results in alarger size of pores and leads a low concrete strength. In addition,the initial water content in the pores may vary also with depth inconcrete due to water bleeding during construction. Thus, for thesame concrete mixture, the initial water content should be variedalong the specimen depth. The closer to the casting face, the morethe water content. This phenomenon was experimentally observedrecently by measuring the water content in fresh concrete at differ-ent locations [13]. Therefore, for given environmental condition,the length of stage I of humidity development should be a functionof both water/cement ratio and the location from casting surface,

0 7 14 21 28

Age (days)

0

20

40

60

80

100

Rel

ativ

e hu

mid

ity (%

)

20030040050060070080090010001100120013001400

Shrinkage (µm/m

)

C-02.5cm5cm10cm18cm

RH

Shrinkage

(a) C-0

0 7 14 21 28

Age (days)

0

20

40

60

80

100

Rel

ativ

e hu

mid

ity (%

)

20030040050060070080090010001100120013001400

Shrinkage (µm/m

)

C-11.5cm2.5cm5cm10cm18cm

RH

Shrinkage

(b) C-1

0 7 14 21 28Age (days)

0

20

40

60

80

100

Rel

ativ

e hu

mid

ity (%

)

20030040050060070080090010001100120013001400

Shrinkage ( µm/m

)

C-21.5cm2.5cm5cm10cm18cm

RH

Shrinkage

(c) C-2

Fig. 11. Calculated free shrinkage strain with time curves of concrete slab atdifferent depths from the exposed face, (a) C-0, (b) C-1 and (c) C-2.

as well as internal curing level, as showed in Fig. 6. Certainly, ifthe environmental conditions are changed, such as environmentaltemperature and humidity, the length of stage I should be changedalso.

Regarding the water loss in concrete, especially in early ageconcrete, normally caused by cement hydration and water diffu-sion and evaporation to the environment. At the initial period afterconcrete casting, most of the pores in concrete are filled with liquidwater. Thus, the relative humidity in concrete should be equal orclose to 100% due to the high amount of liquid water in the pores,which makes the water–vapor pressure in the pore is equal or veryclose to the vapor pressure when the water vapor is saturated. Dueto the process of water consuming is so slow that the period with

0 5 10 15 20Position (cm)

200

300

400

500

600

700

800

900

1000

Shrin

kage

(µm

/m)

C-0672,576,480,384,288,168,72h

(a) C-0

0 5 10 15 20Position (cm)

200

300

400

500

600

700

800

900

1000

Shrin

kage

(µm

/m)

C-1

672,576,480,384,288,168,72h

(b) C-1

0 5 10 15 20Position (cm)

200

300

400

500

600

700

800

900

1000

Shrin

kage

(µm

/m)

C-2

672,576,480,384,288,168,72h

(c) C-2

Fig. 12. Free shrinkage profiles of concrete slab, (a) C-0, (b) C-1 and (c) C-2.

1 10 100 1,000 10,000 100,000 1,000,000Diameter (nm)

0

0.04

0.08

0.12

0.16

0.2

0.24

0.28

0.32

V p(m

l/g)

LWAMatrix-1dayMatrix-28day

Fig. 13. Pore size distribution of cement matrix and lightweight aggregate.

48 Y. Han et al. / Construction and Building Materials 61 (2014) 41–49

100% humidity, which is defined as stage I in the present paper, canlast quite a long time. When the water content in concrete poresachieves a critical value, the vapor pressure becomes lower thanthe saturated value and the relative humidity starts to decrease.Beginning at this point, the progress of internal humidity goes intostage II, in which the humidity gradually decreases with time. Herewe may define the above water content as the critical watercontent, and corresponding time is defined as the critical time, thatis the length of stage I.

As PSLWA was utilized, the water released from PSLWA willcomplement the water consumption due to cement hydrationand /or diffusion to environment, and thereby prolong the lengthof stage I, finally decreasing the shrinkage of concrete. The successof using lightweight aggregates as internal curing agent ordinarilydepends on the water needed to refill the empty pores caused byself-desiccation and evaporation to environment, and the abilityof water to leave the lightweight aggregate [9]. For cement hydra-tion resulted self-desiccation, Bentz et al. [14] proposed a model tocalculate the amount of lightweight aggregate required for releas-ing water to refill the pores existed in the surrounding matrix as:

M ¼Wcecsamax

S � / ð2Þ

where M is the mass of dry lightweight aggregate required to be sat-urated to provide water to fill the voids in cement paste created bychemical shrinkage, in kg/m3. Wc is the cement content of the mix-ture, in kg/m3. ecs is chemical shrinkage of the cement, in l/kg. amax

is the expected maximum hydration degree of cement and isapproximately equal to 0.6261 for the mixtures used in present pa-per [12]. / is the porosity of the lightweight aggregate, and is equalto 37% for the aggregate used in the present work. S is the water sat-uration of lightweight aggregate, and is equal to 0.77 for 7-day’ssoaking in present work. In this study, contents of cement and silicafume in mixtures with PSLWA are 468.8 kg/m3 and 52.1 kg/m3 forconcrete C-1 and 494.4 kg/m3 and 54.9 kg/m3 for concrete C-2respectively. According to Powers et al. (1948) and Jensen et al.(2001), the ultimate chemical shrinkage of Portland cement and sil-ica fume may achieve 0.06 l/kg and 0.20 l/kg respectively. Hence forconcretes C-1 and C-2 the water needed for internal curing esti-mated from (2) is 24.13 kg/m3 and 25.45 kg/m3 respectively; onthe other hand, water provided by the PSLWA in the two mixturesis 37.5 kg/m3 and 59.3 kg/m3. Clearly, it is sufficient enough to fillthe voids under sealed condition. However, as specimen suffers dry-ing (see Figs. 11 and 12), the capillary pores cannot continuingly befilled by the released water and the internal relative humidity ofconcrete should finally start to reduce, and accordingly dryingshrinkage strain must occur as well. Certainly, the released waterdoes efficiently prolong the period of moisture saturated stage withRH = 100%. This is why internal curing cannot completely eliminatethe surface drying induced cracking, but it does effectively delay thecracking time. More techniques are definitely needed regarding theprevention of surface drying cracking.

From the theoretical point of view, as long as the internalrelative humidity at the places neighboring the LWA is lower thanthe humidity inside of LWA, the water contained in LWA shouldstart to flow out to refill the empty pores in the concrete and allowthe internal humidity to rise. The ability of water to leave the light-weight aggregate may be analyzed from the comparison of porestructures of cement matrix and lightweight aggregate. Fig. 13 pre-sents the pore size distribution of the cement matrix used in thepresent work at 1 and 28 days respectively and the lightweightaggregate using mercury intrusion porosimetry. Apparently, mostof pores are below 100 nm for cement matrix and are in 1000–100,000 nm in LWA by contrast. As the internal relative humidityin the cement matrix begins to drop from 100%, the water in thecoarser pores in the LWA should be dried first and it must flow into

surrounding matrix. This distinguished difference in pore sizedistribution between the cement matrix and the LWA effectivelyguarantees the transportation of internal curing water from LWAto cement paste, resulting in extending the length of stage I andtherefore reducing shrinkage of concrete. In addition, the mainpore sizes of the LWA are lower than 0.1 mm and the particle sizeof LWA is 2.5–6.5 mm, this property should prevent the water inPSLWA from releasing too early during concrete mixing and withinthe initial period of the humidity saturated stage I, which is impor-tant for the success of internal curing method.

As stated previously, the initial size and amount of pores inconcrete are a function of water/cement ratio and location in con-crete [13]. The initial water content in the pores should obey thesame law as the pore size, for instance low water/cement ratioresults in small pores and therefore lesser quantity of water in thepores is expected. For the same water-loss rate, a longer time willbe used to reach the critical pore water content in concretes withPSLWA. Therefore, for a given water loss and/or consumption rate,the length of stage I should be substantially increased for concreteswith PSLWA. Further, within present addition range the morePSLWA added, the longer the length of stage I, see Fig. 8. On the otherhand, for certain initial pore water content, the reduction of porewater with time will be controlled by water-loss speed, which nor-mally is influenced by environmental conditions, diffusion proper-ties of concrete, concrete surface status, cement hydration speed,as well as the ability of water supplement in the case of internal cur-ing with PSLWA. The experimental findings on the progress of inter-nal humidity in high strength concretes with and without PSLWA atearly-ages shown in the present paper are the results reflecting theinfluences of all above internal and external factors, including con-crete mix proportion, cement hydration, water diffusion and evap-oration properties, environmental and internal curing conditions.

5. Conclusions

In the paper, the moisture content, in terms of relative humid-ity, inside of high strength concrete slabs, with and without addi-tion of PSLWA, exposed to normal indoor environment isinvestigated by continuously measuring the internal humidity ofconcrete immediately after slab casting until 28 days. The internalrelative humidity of high strength concrete decreases with curingage since casting. The decreasing in relative humidity inside ofconcrete is an integrated result of cement hydration and waterdiffusion to the environment. The variation of relative humidityinside of concrete with time follows a vapor-saturated stage with100% relative humidity (stage I) and a stage that relative humiditygradually decreases (stage II). A humidity gradient along the thick-ness of slab exists in early age high strength concrete. As PSLWA is

Y. Han et al. / Construction and Building Materials 61 (2014) 41–49 49

added, the length of internal humidity saturated stage I is signifi-cantly prolonged and the decreasing rate of internal humidity instage II noticeably decreased; within present addition range themore the PSLWA added, the stronger the internal curing effect.The humidity gradient along the thickness of the slab is reducedsimilarly. The highest reduction on internal humidity at 28 dayssince casting in high strength concrete slab falls from 46.5% to26.2% with a medium PSLWA addition ratio, and finally to 7.9%with a high PSLWA addition level.

Acknowledgements

This work has been supported by a Specialized Research Fundfor the Doctoral Program of Higher Education (20130002110034)and Grants from the National Science Foundation of China (No.51178248), a Grant from National Basic Research Program of China(No. 2009CB623200) to Tsinghua University.

References

[1] Ayano T, Wittmann FH. Drying, moisture distribution, and shrinkage ofcement-based materials. Mater Struct 2002;35(247):134–40.

[2] Baroghel-Bouny V, Mainguy M, Lassabatere T, Coussy O. Characterization andidentification of equilibrium and transfer moisture properties for ordinary andhigh-performance cementitious materials. Cem Concr Res 1999;29(8):1225–38.

[3] Bissonnette B, Pierre P, Pigeon M. Influence of key parameters on dryingshrinkage of cementitious materials. Cem Concr Res 1999;29(10):1655–62.

[4] Zhang ZB, Zhang J. Experimental study on the relationship between shrinkagestrain and environmental humidity of concrete. J Build Mater2006;6(6):720–3. In Chinese.

[5] Wittmann FH. The fundamentals of a model for the description of concretecharacteristics. Schriftenreihe Deutscher Ausschuss fur Stahlbeton, Heft 290,Berlin; 1977. p. 43–101. [In German].

[6] Zhang J, Hou D, Chen H. Experimental and theoretical studies on shrinkage ofconcrete at early-ages. ASCE J Mater Civ Eng 2011;23(3):312–20.

[7] Zhang J, Hou D, Han Y. Micromechanical modeling on autogenous and dryingshrinkages of concrete. Constr Build Mater 2012;29(3):230–40.

[8] Bentur A, Igarashi S, Kovler K. Prevention of autogenous shrinkage in high-strength concrete by internal curing using wet lightweight aggregates. CemConcr Res 2001;31(11):1587–91.

[9] Henkensiefken R, Bentz D, Nantung T, Weiss J. Volume change and cracking ininternally cured mixtures made with saturated lightweight aggregate undersealed and unsealed conditions. Cem Concr Compos 2009;31(7):427–37.

[10] Mehta PK. Concrete structure, properties, and materials. Prentice-Hall, NewJersey: Englewood Cliffs; 1986.

[11] Voigt T, Ye G, Sun Z, Shah SP, Klaas VB. Early age microstructure of Portlandcement mortar investigated by ultrasonic shear waves and numericalsimulation. Cem Concr Res 2005;35(5):858–66.

[12] Zhang J, Han Y, Gao Y, Luosun Y. Integrative study on the effect of internalcuring on autogenous and drying shrinkage of high strength concrete. DryTechnol 2013;31(5):565–75.

[13] Zhang J, Huang Y, Qi K, Gao Y. The interior relative humidity of normaland high strength concrete at early age. ASCE J Mater Civ Eng2012;24(6):615–22.

[14] Bentz D, Snyder KA. Protected paste volume in concrete: Extension to internalcuring using saturated lightweight fine aggregate. Cem Concr Res1999;29(11):1863–7.