Correlation between the viscoelastic properties and cracking potential of engineered cementitious composites

Construction and Building Materials 71 (2014) 375–383

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Correlation between the viscoelastic properties and cracking potential of engineered cementitious composites Suleyman Bahadir Keskin a, Mustafa Sahmaran b,⇑, Ismail Ozgur Yaman c, Mohamed Lachemi d a

Department of Civil Engineering, Mugla Sitki Kocman University, Mugla, Turkey Department of Civil Engineering, Gazi University, Ankara, Turkey c Department of Civil Engineering, Middle East Technical University, Ankara, Turkey d Department of Civil Engineering, Ryerson University, Toronto, Canada b

h i g h l i g h t s  Both microcracking and viscous shear theories are related to the deformability of ECC.  Dimensional compatibility depends on tensile creep, elastic and shrinkage properties.  Cracking potentials calculated by tensile creep and shrinkage tests reflected the real behavior.  The type and amount of mineral admixture have significant effect on dimensional and mechanical properties.

a r t i c l e

i n f o

Article history: Received 20 June 2014 Received in revised form 22 August 2014 Accepted 27 August 2014

Keywords: Engineered Cementitious Composites (ECC) Dimensional stability Cracking potential

a b s t r a c t Although Engineered Cementitious Composites (ECC) offer a number of advantages over ordinary and fiber reinforced concrete in many respects, it is not cost-effective to build a whole structure with ECC, currently. Thus, ECC can potentially be used in repair systems or in bi-material systems which require it to be used together with a dimensionally stable material. High shrinkage, together with the restraining effect brings about cracking a critical phenomenon for ECC. In this study, along with the mechanical properties of ECC, viscoelastic properties like autogenous shrinkage, drying shrinkage and tensile creep which were used to calculate ECC’s cracking potential were studied. At the same time, the tendency of ECC mixtures to crack under restrained shrinkage conditions was also investigated using restrained shrinkage rings. It was concluded that creep, elastic properties, and shrinkage data should be together used to evaluate the dimensional compatibility. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction As new civil engineering materials, cementitious materials of different types being developed and brought onto the scene with the goal of constructing safer and more durable structures. Engineered Cementitious Composites (ECC), which possesses high ductility and durability, may be the most striking of those materials. Although ECC is similar to conventional fiber reinforced concrete in terms of ingredients, the characteristics of the materials such as aggregate size, fiber type etc. used in its production lead to superior tensile ductility properties. To provide those properties, ECC is designed using a micromechanical design method that elaborates the properties of each single ingredient to achieve composite properties. In addition, the compressive strength of ECC is high enough ⇑ Corresponding author. Tel.: +90 312 582 3249; fax: +90 312 231 9223. E-mail address: [email protected] (M. Sahmaran). http://dx.doi.org/10.1016/j.conbuildmat.2014.08.089 0950-0618/Ó 2014 Elsevier Ltd. All rights reserved.

to be used in all engineering structures. Its main advantage over ordinary and fiber reinforced concrete is tensile strain capacity, which is 200–500 times greater than that of conventional concrete accompanied with tensile strain hardening property; hence ECC can be considered as a ductile material [1–4]. Another important property of ECC is that self-controlled crack widths remain under 100 lm under tensile stress regardless of ultimate tensile strain, which is the key property behind the material’s enhanced durability [5]. Despite the superior properties that make ECC a preferable material in most respects, special constituents make it too costly to use instead of concrete in whole structures. For this reason, ECC is generally used in critical parts of structures or for repair work. Potential uses of ECC have been demonstrated in multiple studies, and include bridge deck, coupling beams, bridge deck link slabs, patch repairs, retrofits, layered repair systems or overlays, stud connections, shotcrete repair systems, high early strength

376

S.B. Keskin et al. / Construction and Building Materials 71 (2014) 375–383

repairs and ductile strips [6–13]. These applications all require ECC to be used adjacent to another material like concrete or steel, which brings about early restrained cracking problems. Durability of a repair material or a material used in bi-material systems is dependent on its dimensional compatibility with the substrate, repaired material or the materials which are in contact with it. When a material is placed on or adjacent to a dimensionally stable material – which has lower shrinkage due to long-term exposure to environmental conditions and advanced hydration – the repair material should have appropriate shrinkage, tensile creep and elastic properties. However, due to the high amount of binder, low water to binder ratio and lack of coarse aggregate, ECC may potentially suffer from dimensional instability problems. Several researchers have investigated the required properties for cementitious materials to be used for repair purposes, as well as factors affecting their performance. In general, compatibility of the repair material to the substrate – especially dimensional compatibility – was emphasized. Shrinkage, creep and instantaneous elastic properties [14,15] and strength [16–18] were suggested to be effective on compatibility. Also criteria for repair materials in terms of strength, shrinkage, creep, and elasticity properties were suggested by researchers [19,20]. Dimensional compatibility may be the most important property of a material used for repair purposes or in bi-material systems. If there is a mismatch between adjacent materials, any variations in the conditions such as temperature, hydration, environmental, loading and restraining effect yield internal stresses resulting in the formation of tension cracks. Li [21] and Li and Stang [22] defined cracking potential as superposition of time-dependent strains and formulated the cracking potential of a strain-hardening material under restrained shrinkage conditions as:

(GGBFS) were also used as binders. Physical properties and chemical compositions of the PC, FA and GGBFS can be found in Table 1. Aggregate used in the production of ECC was silica sand with a SiO2 content of 99.8%. Specific gravity and absorption capacity of the silica sand were 2.60 and 0.3%, respectively. Although silica sand with maximum and average aggregate sizes of 250 lm and 110 lm are used in the production of standard ECC mixtures [30], maximum and average aggregate sizes of the locally available silica sand used in this study were about 400 lm and 200 lm. Polyvinyl alcohol (PVA) fibers 8 mm in length and 39 lm in diameter, which are commonly used in ECC production, were also used. Nominal tensile strength and specific gravity of the fibers were 1610 MPa and 1.3, respectively, as reported by the manufacturer. PVA fibers were provided coated with a 1.2% (by weight) hydrophobic oiling agent. In order to obtain a workable fresh mixture with adequate plasticity for uniform fiber distribution without bundling of fibers, polycarboxylate ether type high range water reducing admixture (HRWR) with a solid content of 40% and specific gravity of 1.1 was used, as the water to binder ratio of the ECC mixtures was quite low. In order to investigate the effect of mixture proportions on time-dependent and mechanical properties, four different ECC mixtures were prepared. For an ECC specimen to attain strain hardening behavior with multiple fine microcracks, mixture proportions are determined by a micromechanical design principle which utilizes strength and energy criteria [3,31,32]. In the literature the ECC mixture named ‘‘M45’’ satisfying both criteria is comprehensively studied and is frequently referred. For this reason ‘‘M45’’, exhibiting an average crack width of about 60 lm and tensile strain capacity of about 5% under direct tensile stress, was selected as one of the mixtures in this study [5]. In addition to the ‘‘M45’’ (F 1.2 in this study), having a fly ash to cement ratio of 1.2 where about 55% of the total cementitious materials is fly ash, the amount of FA was increased to about 70% of the cementitious materials in F 2.2, to observe the effect of mineral admixture amount while the proportions of other ingredients (except HRWR) were kept constant. The remaining two mixtures replicated the first two mixtures, except the mineral admixture used was GGBFS instead of FA. All four mixtures possessed a water to binder ratio of 0.27, and the amount of silica sand was kept constant, for a sand to binder ratio of 0.36. An adequate amount of the HRWR, varied for all mixtures, was used to reach the required consistency. All mixtures contained 2% of PVA fibers by volume. Proportions and designations of the ECC mixtures are provided in Table 2. 2.2. Test specimen preparation and testing

p ¼ esh  ðee þ ei þ ecp Þ where esh is the shrinkage strain, ee is the elastic strain capacity, ei is the inelastic tensile strain capacity and ecp is the tensile creep strain. According to the given formulation, shrinkage (autogenous, plastic, drying and carbonation, thermal) is the main driver for cracking potential, while the resistors are elastic and inelastic strain capacity and material creep. Cracking due to lack of dimensional compatibility may be reduced or even prevented by altering the material parameters that affect cracking potential [17,21–23]. Although the mechanical properties of ECC have been studied intensively in the literature, there are limited studies concerning its viscoelastic properties. Despite the fact that autogenous shrinkage [24,25] and tensile creep [26–29] of ECC have attracted the attention of researchers, the results of these studies were not correlated to cracking potential. This paper presents the results of the experimental study that concentrated on the cracking potential of ECC, together with the viscoelastic properties that affect the cracking potential. Viscoelastic properties including drying and autogenous shrinkage and tensile creep were studied along with mechanical properties. Cracking potential values of the tested ECC specimens were calculated using experimentally determined shrinkage and creep values in accordance with the formulation given in Li [21] and Li and Stang [22]. All specimens were subjected to restrained shrinkage, and the tendency of the specimens to cracking was determined by this way. Calculated values of cracking potential (p) were then compared with the results of the restrained shrinkage test. 2. Experimental program 2.1. Materials and mixture proportions The Portland cement (PC) used in ECC production was European type CEM I 42.5R, which is similar to ASTM Type I. Fly ash (FA) corresponding to Class-F (according to ASTM C 618 standard) and ground granulated blast furnace slag

All ECC mixtures were prepared in a planetary type mixer with a 25-liter capacity. All solid materials were introduced and blended in the mixer, and water was added afterwards. Following the preparation of the stiff mortar, HRWR was added until the desired consistency was achieved. Finally, PVA fibers were added and the mixing operation was continued until reaching a fiber distribution without bundling of fibers. The compressive and flexural strengths of the specimens were determined at 7, 14, 28 and 90 days. Cubic specimens with 50 mm side lengths were prepared for compressive strength tests. The flexural strengths of the mixtures were determined on 360  75  50 mm beam specimens under four point bending. Specimens were demolded 24 h after casting. For the first seven days, they were kept in 95 ± 5% relative humidity (RH) at 23 ± 2 °C in plastic bags to avoid moisture transfer, and stored at 50 ± 5% RH at 23 ± 2 °C until the age of testing. During the four-point bending tests, beams were placed on two supports 300 mm away from each other, and load was applied symmetrically from two 100 mm spaced points with a loading rate of 0.005 mm/s on a deformation controlled closed-loop testing machine. A linear variable differential transformer (LVDT) was attached to each flexural strength specimen and mid-span deflections of the beams were recorded simultaneously during loading. The viscoelastic and dimensional properties of the ECC mixtures determined in this study include autogenous and drying shrinkage and tensile creep. Additionally, restrained shrinkage tests were performed to experimentally evaluate the cracking potential of the produced ECCs. Free autogenous shrinkages of the specimens were

Table 1 Chemical composition and physical properties of the cementitious materials. Chemical composition

PC

FA

GGBFS

CaO (%) SiO2 (%) Al2O3 (%) Fe2O3 (%) MgO (%) SO3 (%) K2O (%) Na2O (%) Loss on ignition (%) Specific gravity Blaine fineness (m2/kg)

61.43 20.77 5.55 3.35 2.49 2.49 0.77 0.19 2.2 3.06 325

1.64 56.22 25.34 7.65 1.8 0.32 1.88 1.13 2.1 2.31 290

34.48 38.4 10.96 0.81 7.14 1.48 0.86 0.18 3.0 2.79 425

377

S.B. Keskin et al. / Construction and Building Materials 71 (2014) 375–383 Table 2 ECC mixture proportions by weight. Mix. Id.

MA type

PC

W/CM

MA/PC

S/CM

PVA fibers (%) (by volume)

HRWR (kg/m3)

F 1.2 F 2.2

FA

1 1

0.27 0.27

1.2 2.2

0.36 0.36

2 2

5.1 3.0

S 1.2 S 2.2

GGBFS

1 1

0.27 0.27

1.2 2.2

0.36 0.36

2 2

5.8 4.7

MA: Mineral Admixture; PC: Portland Cement; W: Water, CM: Cementitious Material; S: Sand; PVA: Polyvinyl alcohol; HRWR: High Range Water Reducing Admixture.

measured with linear prismatic molds called shrinkage drains; similar setups have been used for early age shrinkage [33] and autogenous shrinkage measurements [34] in the literature. Details of the shrinkage drains used in this study were previously described in another study by Keskin et al. [25]. Shrinkage drains consist of two coaxial prismatic molds, with one mold placed inside another. The inner mold has a length of 1000 mm, depth of 60 mm and width of 90 mm. The outer mold serves as a way of locating the LVDT to measure the deformation at the open end of the inner mold. It also provides a space for the thermocouples placed under the inner mold to measure temperature variations while the fresh mixture is cast inside the inner mold. All the interior surfaces of the inner mold were covered with Teflon sheets to avoid restraining effect due to friction. One end of the drain was left open, and an LVDT was attached for continuous acquisition of length data. Details of shrinkage drains are depicted in Fig. 1. Before placing the fresh ECC mixture, aerosol Teflon was sprayed over the Teflon sheet to further reduce friction, and interior surfaces were covered with a thin plastic sheet to encase the fresh mixture and prevent moisture transfer between specimen and environment. Finally, a piece of styrofoam was placed in the open end to prevent loss of fresh material and bleed water. A small metal plate was embedded inside the fresh mixture and brought in contact with a rod surrounded by a spring to provide a link between the specimen and the LVDT. After casting the fresh ECC mixture, the top surface of the drain was immediately covered with another Teflon sheet. The styrofoam used to hold the fresh mixture was removed after the specimens reached their final setting time, and autogenous shrinkage measurements began immediately after that. The LVDT reading corresponding to the maximum temperature recorded was taken as the zero reading for the length change measurements, and autogenous shrinkage tests were continued until the end of 28 days. LVDT readings were taken at 10-min intervals. Autogenous shrinkage test setups were kept in an isolated room at 23 ± 2 °C, and autogenous shrinkages were determined on two replicated specimens for each of the ECC mixtures tested. A tensile creep test frame (Fig. 2) in form of a lever arm was designed and manufactured with steel I-beam sections. Each frame was capable of simultaneously loading three specimens with a maximum load of 2000 kgf. 5 and 10 kg steel plates were used to apply the load with a sensitivity of about 35 kgf. To provide loading without any eccentricity, hinge connections were used in the joints between the frame and the specimens. Tensile creep frames were calibrated with a load cell before use, and calibration charts were prepared to ensure accurate loading. Ishaped prism specimens were prepared by using a special mold, and steel plates with bolts at one face were mounted to the top and bottom surfaces of each specimen with quick-set epoxy resin. In order to decrease the amount of tensile stress acting on the steel-ECC connection, the I-shaped prisms were designed so the cross

sectional areas at the top and bottom were 75  75 mm and the prism measured 55  55 mm in the middle. Two steel discs were located 250 mm apart on all four faces where the cross sectional area was smaller. Length change measurements were taken over these steel discs using a mechanical strain gage with the capability of measuring 250 mm and with a sensitivity of 0.001 mm. Detailed information on tensile creep tests can be found in Keskin [35]. Two sets of tensile creep tests, one for early age and the other for late age, were conducted to observe the behavior of ECC mixtures under sustained direct tensile stresses. In the first set of experiments, which took place 24 h after casting, specimens were demolded and cured under lime saturated water for the two days. After being removed from the water, the specimens were left to dry for one day under laboratory conditions to ensure that the epoxy resin had the dry surfaces it requires to work properly. A steel plate was mounted to one end of each specimen with epoxy resin and left for one day to allow the epoxy resin to set. Another steel plate was mounted to the opposite end in the same manner, and specimens were left for epoxy resin to set and gain strength until the age of seven days. At seven days, specimens were placed into the tensile creep loading frame and loaded to 30% of their ultimate direct tensile strength, as determined on briquet specimens. Tensile strengths were also determined at 14 and 28 days in the same manner, and tensile creep loads were modified to sustain a tensile load of 30% of the ultimate tensile strength until the end of 90 days. Tensile creep tests were conducted in an isolated room where the temperature and relative humidity were fixed at 23 ± 2 °C and 50 ± 4%, respectively. In the second set of tensile creep experiments, specimens were demolded 24 h after casting and cured under lime saturated water for 21 days. After 21 days, specimens were prepared for tensile creep test in the following seven days, as described in the first set. At the age of 28 days, specimens were placed in the loading frame and loaded with a sustained tensile stress of 1.5 MPa for 90 days. For the drying shrinkage test, two additional specimens with the same dimensions of the tensile creep specimens were prepared. Drying shrinkage specimens were stored under the same conditions as the tensile creep specimens without loading until the end of the tensile creep tests. Measured drying shrinkage strains were used to calculate tensile creep since the specimens were inevitably subjected to drying shrinkage during tensile creep loading. Restrained shrinkage tests were conducted to determine the tendency of ECC mixtures to crack. For this purpose, restrained shrinkage rings similar to those described in AASHTO PP-34-99.66 [36] were used. A restrained shrinkage ring consists of two concentric steel rings with a height of 140 mm, located on a non-absorbent steel base. The inner ring has a thickness of 12.5 mm and an outer diameter of 305 mm. The outer ring has an inner diameter of 355 mm that allows the casting of

Rod and Spring Mold

Fig. 1. Details of the autogenous shrinkage drains.

LVDT

378

S.B. Keskin et al. / Construction and Building Materials 71 (2014) 375–383

Fig. 2. Tensile creep test specimen (a) and test frame (b) (all units in mm).

fresh materials, with a thickness of 25 mm between the two rings. Immediately after casting the fresh mixture between the rings, the top surface of the test material was covered with wet burlap. Twenty-four hours after casting, the outer mold was removed, the top surface of the test material was sealed with a silicone-based sealant, and the specimen was subjected to drying from the outer surface. Ambient temperature and relative humidity were kept at 23 ± 2 °C and 50 ± 4% during the test. In the restrained shrinkage test, short curing time and early age testing reflected real life in situ conditions. Crack onset time, width and number were recorded for 28 days. Crack width measurements were taken daily from three different points on each crack using a hand held microscope with an accuracy of 5 lm. Restrained shrinkage tests were performed on two replicates of each ECC mixture.

3. Results and discussion 3.1. Mechanical performance Results of the compressive and flexural strength tests conducted on ECC mixtures at 7, 14, 28 and 90 days are presented in Table 3. Each result is the average of at least five specimens. It

can clearly be seen from the table that mixtures containing GGBFS have higher compressive strengths compared to the FA mixtures, especially at early ages. The observed high early strength of GGBFS mixtures may be attributed to the higher reactivity of GGBFS at early ages due to its high specific surface area. Also, considering the oxide composition of GGBFS, since the amount of SiO2 is higher and CaO is lower than that of Portland cement, hydration of GGBFS Portland cement mixtures yield higher amounts of C–S–H gel and lower amounts of Ca(OH)2 compared to Portland cement hydration alone; resulting in a denser structure [37]. Additionally, compared to FA, the self-cementing property of GGBFS may also be a reason for higher strength. Furthermore, alkali hydroxides of the Portland cement may react with GGBFS before the production of Ca(OH)2 [38]. On the other hand, rate of compressive strength development between 7 and 90 days are higher for mixtures containing FA. Mixtures containing higher amounts of cement possess higher compressive strength as expected, since the compressive strength of

Table 3 Mechanical properties of the ECC mixtures. Basic mechanical properties

Compressive strength (MPa)

Tensile strength (MPa)

Flexural strength (MPa)

Mid-span deflection (mm)

7 days 14 days 28 days 90 days

40.1 51.3 69.3 75.8

4.26 4.54 4.59 4.62

8.5 10.4 12.7 12.8

5.68 5.14 4.83 4.49

7 days 14 days 28 days 90 days

30.2 35.9 46.1 54.7

3.13 3.49 3.49 3.51

7.4 9.1 10.2 11.6

7.02 6.13 5.89 5.29

7 days 14 days 28 days 90 days

59.9 72.4 88.5 91.2

4.57 4.91 5.91 5.94

9.9 11.3 12.6 12.6

3.73 3.51 3.10 3.04

7 days 14 days 28 days 90 days

51.8 64.2 76.3 80.3

3.84 4.20 4.50 4.59

8.4 10.2 11.5 11.8

4.31 4.14 3.98 3.44

F 1.2

F 2.2

S 1.2

S 2.2

S.B. Keskin et al. / Construction and Building Materials 71 (2014) 375–383

379

the mixtures with mineral admixture to cement ratio 1.2 are higher than that of 2.2 at all testing ages. Qian and Li [39] showed that deflection under bending can be correlated to the tensile strain capacity of ECC; therefore each ECC mixture tested in this study may be considered as strain hardening. It is obvious that mixtures containing GGBFS possess slightly higher flexural strength at early ages than mixtures containing similar amounts of FA. Contrary to the compressive strength test results, the difference in flexural strength disappears at later ages. This may be attributed to the fact that, especially in the case of strain-hardening cementitious materials like ECC, flexural strength is dependent on more complex material properties such as tensile first cracking strength, ultimate tensile strength and strain capacity [40]. 3.2. Autogenous shrinkage Autogenous shrinkage developments of the mixtures for a duration of 28 days are depicted in Fig. 3. There are only few studies in the literature on the autogenous shrinkage of ECC mixtures, one of which was conducted by Sahmaran et al. [24] using sealed ECC bars. Another study was conducted by Keskin et al. [25], using the same test set-up described in this study. Both FA- and GGBFS-containing ECC mixtures exhibit high autogenous shrinkages even at early ages; autogenous shrinkage must not be underrated as the specimens have lower tensile strength and strain capacity due to low matrix-fiber bond, which increases their tendency to crack. Autogenous shrinkage of specimens containing GGBFS was found to be higher than those containing FA, which may be an outcome of the high reactivity and early hydration of GGBFS compared with FA. Wei et al. [34] showed that GGBFS tends to increase the long term autogenous shrinkage as a result of the reduction in pore humidity due to the reactivity of GGBFS. This argument may also be valid for this research, since the GGBFS used in this study also exhibited high reactivity, which consumes water in the system and drops relative humidity. Similar results were found in the study conducted by Li et al. [41], where it was concluded that autogenous shrinkage is mainly dependent on the volumetric percentage of the pores between 5 nm and 50 nm. The pores in this range were concluded to be effective on the capillary stresses. To investigate the effect of pore structure on ECCs’ dimensional stability properties, the porosity and pore size distribution study was also carried out with mercury intrusion porosimetry (MIP) testing at the age of 28 days. Fig. 4 shows the normalized volumes of mercury intrusion in a specified range of pores (greater than 50 nm and 5–50 nm) for ECC mixtures. As seen from the figure, pore sizes of the specimens are well correlated with the autogenous shrinkage test results.

Fig. 4. Pore size distributions of ECC mixtures.

3.3. Drying shrinkage Drying shrinkage tests were conducted as a part of tensile creep tests and the same I-shaped prismatic specimens used in the tensile creep tests were also prepared for drying shrinkage determination to avoid experimental variation due to size. Two different test procedures were conducted. In the first procedure, specimens were demolded 24 h after casting, cured under water for two days and kept under the same laboratory conditions as the tensile creep specimens. Initial readings were taken at the age of seven days. In the second procedure, specimens were cured 21 days under water and kept in under same laboratory conditions until the age of 28 days, when initial readings were taken. Each drying shrinkage test continued for 90 days after initial reading. Total drying shrinkage strains recorded at the end of each test are presented in Table 4. Testing revealed that the use of additional mineral admixture resulted in a reduction in drying shrinkage. This may be attributed to lower autogenous shrinkage of the ECC specimens produced with a higher amount of mineral admixture. Another possible reason for this behavior may be the high amount of unhydrated particles acting as fine aggregates that provide additional restraints for drying shrinkage [42–44]. Considering the effect of mineral admixture type on drying shrinkage, ECC mixtures with GGBFS tended to have higher drying shrinkage compared to similar mixtures with FA. This may be attributed to the improved pore size refinement mechanism of GGBFS. GGBFS decreases pore sizes (Fig. 4) and lowers the total porosity, which increases drying shrinkage proportionally [34–41]. In addition, as discussed earlier, autogenous shrinkage of GGBFS-containing mixtures is higher and a significant amount of the measured drying shrinkage (around 600 le) is actually due to autogenous shrinkage. 3.4. Restrained shrinkage

Fig. 3. Autogenous shrinkage development of ECC mixtures.

Restrained shrinkage tests are indicators of the cracking potential of a material, since they represent the ability of the material to compensate for its own shrinkage deformations with its deformability, under restrained conditions. According to the ASTM C1581 [45] standard method, restrained shrinkage testing is terminated when the first crack initiates, as any other crack formation is not possible for brittle cementitious composites. On the other hand, ECC exhibits strain hardening with multiple cracking under increasing stresses, and therefore restrained shrinkage tests were not terminated until the end of the designated 28 days. In this study, all ECC mixtures experienced multiple cracking to some extent. Generally speaking for the ECC mixtures in this study, when

380

S.B. Keskin et al. / Construction and Building Materials 71 (2014) 375–383

Table 4 Elastic modulus (at the age of testing), tensile creep (after 90 days sustained loading) and free drying shrinkage of ECC mixtures. Mix ID

Elastic modulus (GPa)

Tensile creep coefficient

Specific tensile creep (le/MPa)

Drying shrinkage (le)

7 days F 1.2 F 2.2 S 1.2 S 2.2

18.9 17.5 23.0 21.0

6.09 9.03 3.87 5.90

322 516 168 280

1372 1185 1560 1436

28 days F 1.2 F 2.2 S 1.2 S 2.2

19.7 18.1 24.9 22.8

4.12 5.94 3.31 3.57

209 328 133 157

1188 998 1461 1324

Fig. 5. Average crack width development in restrained shrinkage specimens.

10 mm

Fig. 7. Microcrack observed on a tensile creep specimen after 90 days of loading.

Strain

Total tensile strain (B-C)

Adjusted tensile creep (B-A-C) Elasc Strain (A)

Time

Drying tensile creep strain + elasc strain (B) Fig. 8. Correlation between deformability and tensile creep.

Drying shrinkage (C)

Drying tensile creep (B-A)

Fig. 6. Schematic representation of elastic and time-dependent strains of a tensile creep specimen.

mixtures containing FA. It is important to note that GGBFS had an adverse effect on cracking potential, which may be attributed to the development of high tensile stresses due to the high drying and autogenous shrinkage of GGBFS mixtures. 3.5. Tensile creep

a crack opened up, the width increased quickly to a value which eventually stabilized. Fig. 5 shows that the onset time of the first crack was earlier for the GGBFS-containing mixtures. It also clearly shows that FA usage shifted the onset time of first crack, and that the average width of the cracks was smaller in the case of ECC

As mentioned earlier, two sets of experiments were conducted for tensile creep testing. In the first set, specimens were loaded at the age of seven days, with a tensile stress equal to 30% of their tensile strength, and the stress level was gradually adjusted to a

381

S.B. Keskin et al. / Construction and Building Materials 71 (2014) 375–383

(a) FA 1.2

(b) FA 2.2

(c) S 1.2

(d) S 2.2 Fig. 9. Specific tensile creep of ECC mixtures with time.

level equal to 30% of the tensile strength at 14 and 28 days. In the second setup, specimens were loaded with a 1.5 MPa tensile stress at the age of 28 days. Tensile loads were sustained for 90 days after the initiation of the tests. Strains were calculated by using the averages of length changes measured on the total of twelve surfaces of three specimens. Strains calculated for each surface were sufficiently close to each other, indicating that the specimens were loaded without eccentricity. All specimens were contracted during the test, since simultaneously measured drying shrinkages were higher than tensile creep. Superposition of drying shrinkage and drying tensile creep, referred to as ‘‘adjusted tensile creep,’’ was considered to evaluate the tensile creep behavior of the specimens. Fig. 6 illustrates the time-dependent and elastic strains during the tensile creep tests. Creep data can be denoted by specific creep, which is defined as creep strain per unit stress and also by creep coefficient, which is defined as creep strain per elastic strain [46]. After calculating the adjusted tensile creep strains, development of specific creep of the ECC mixtures with time was also determined in order to assess the first and second tensile creep test sets together. The elastic portion of the total strain was separated by the use of the elastic modulus determined by using cylindrical specimens with a diameter of 10 cm and height of 20 cm. Specimens used for determining the elastic moduli were kept under the same conditions as the tensile creep specimens and subjected to loading at the same age, which corresponds to seven and 28 days after casting. The mechanism of tensile creep has been studied by several researchers. Ward and Cook [47] based the mechanism of tensile creep on microcracking. Bissonnette et al. [48] also claimed that

the microcracking theory might be valid for unsealed tensile creep specimens, as they determined that secant modulus of elasticity of tensile creep specimens were slightly lower than that of the companion specimens, which may be a result of microcracking. Boshoff and Van Zijl [26], in their study on tensile creep of ECC, showed that time-dependent crack formations, which may be a source of creep deformation, may occur in ECC specimens under sustained tensile stresses. Although no cracks could be observed under ocular inspection in this study, several microcracks were detected under heavy light source, as seen in Fig. 7. Additionally, there may be more cracks that were not visible, even under a heavy light source, as the stress rate was limited to 30% of the tensile strength or 1.5 MPa. Bissonnette et al. [48] found that viscous shear theory, which relies on the sliding of gel particles against one another due to the disruption of adsorbed water layers by the applied load, was successful in describing tensile creep. Both microcracking and viscous shear theories are related to the deformability of ECC, as the high tensile strain capacity of ECC is due to multiple microcracks. As seen from Fig. 8, there is a very close relationship between the deformability and tensile creep of ECC. As seen from Fig. 9, specimens loaded at seven days had higher specific creep values compared to specimens loaded at 28 days. This may be explained by the deformability of the ECC mixtures again, since ECC mixtures are more deformable at early ages as a result of their low matrix fracture toughness. This phenomenon may also be used to explain the higher tensile creep values of mixtures containing FA compared to those containing GGBFS, as well as the higher tensile creep values of mixtures containing higher amounts of mineral admixture.

382

S.B. Keskin et al. / Construction and Building Materials 71 (2014) 375–383

3.6. Cracking potential Results of the restrained shrinkage tests provide an indicator of a material’s cracking potential. Creep coefficient, defined as the ability of a material to relax stresses [49], may also give information about cracking behavior. It can therefore be concluded that the ECC mixtures with high creep coefficients (Table 4) are the same mixtures that perform better under restrained shrinkage conditions, showing that the induced tensile stresses due to drying relaxed well in those mixtures. According to the restrained shrinkage test results, specimens with a lower tendency for cracking also had low autogenous and drying shrinkage and high total tensile creep strains, which is consistent with the commonly accepted facts on cracking potential. However, restrained shrinkage tests were conducted within the first 28 days after casting, while tensile creep and drying shrinkage tests were started at seven and 28 days. The first set of tensile creep and drying shrinkage tests is more convenient since it reflects the early age tensile creep and drying shrinkage behavior, including the ages after seven days. Moreover, in order to be able to calculate the cracking potential according to the formulation given by Li [21] and Li and Stang [22], data from different tests should belong to same duration. However, it was not possible to start tensile creep tests before seven days, as sufficient time was needed to prepare the creep specimens. Even so, the first 21 days of the drying shrinkage and tensile creep tests overlapped with the restrained shrinkage test, which provided the opportunity to

a. Specimens loaded at 7 days

b. Specimens loaded at 28 days Fig. 10. Cracking potential development of ECC mixtures.

calculate cracking potential and compare it with the results of the restrained shrinkage test. In order to calculate cracking potential (p), resistors of cracking potential (total tensile strains including both instantaneous and creep strains) were subtracted from the drivers of the cracking potential (drying shrinkage). Absolute values of the strains were used in the calculations. The development of the cracking potentials of the specimens for 90 days from the beginning of the test is presented in Fig. 10. As seen from the figure, the cracking potentials of the specimens agree well with the results of the restrained shrinkage tests. Although it was concluded that the cracking potential of ECC should be very low, even showing negative values [21,22], it should be noted that cracking potential values were calculated considering tensile strains (elastic and creep) corresponding to a stress level of 30% of the ultimate tensile strength of the material or 1.5 MPa. The inelastic strain capacity of ECC, which is up to 5%, was not included in the calculations, which further decreased the cracking potential to negative values. 4. Conclusions The following conclusions were drawn from the results of the experimental study:  The compressive strength of the mixtures containing GGBFS was higher than that of the mixtures containing FA due to the earlier reaction, self-cementing property and high specific surface area of the GGBFS compared to FA. As mineral admixture to cement ratio was increased, Portland cement was diluted and compressive strength decreased. The type and amount of mineral admixture significantly affected the flexural performance of ECC mixtures. GGBFS mixtures had higher flexural strengths compared to FA mixtures, while in the case of midspan beam deflection, the situation was opposite.  When the amount of the mineral admixture was increased, the drying shrinkage of the mixtures decreased significantly. This may be attributed to unreacted mineral admixture particles, which restrained shrinkage strains and lowered autogenous shrinkage. GGBFS mixtures had higher drying shrinkage than FA mixtures due to enhanced pore size refinement, which resulted in higher capillary tension.  Autogenous shrinkage of mixtures containing GGBFS was higher than that of FA mixtures. Autogenous and drying shrinkages were affected by the mix proportions of ECC in the same manner, since the mechanisms of drying and autogenous shrinkage are similar. However, the development of autogenous shrinkage strains differed for FA and GGBFS mixtures due to their rates of reactivity.  During the restrained shrinkage test, all specimens exhibited multiple cracking. The onset time of first crack was earlier for GGBFS-containing mixtures. As the amount of mineral admixture increased, onset time of first crack was delayed and average crack width decreased. Low compatibility of GGBFS is attributed to lower ductility, lower tensile creep and higher shrinkage, which in turn resulted in higher induced tensile stresses.  Mixtures containing FA had higher tensile creep compared to the mixtures with GGBFS. As the amount of mineral admixture increased, tensile creep strains increased as well.  Specific creep values of the specimens loaded at seven days were higher than those of the specimens loaded at 28 days. Both the effect of mineral admixture amount and age of the first tensile creep loading are attributed to the deformability of the specimens and explained by microcracking and viscous shear theories.

S.B. Keskin et al. / Construction and Building Materials 71 (2014) 375–383

 Viscoelastic properties of the mixtures were found to be related to the cracking potential of the ECC mixtures. Cracking potentials calculated by tensile creep and shrinkage tests reflected the real cracking behavior of ECC specimens under restrained shrinkage conditions. Consequently, ECC mixtures may potentially suffer from early age cracking due to shrinkage under restrained conditions as a result of low water to cementitious material ratio. However, besides shrinkage, deformability is also important for the performance of a material under restrained shrinkage conditions as cracking potentials calculated by using the data obtained by the tests on viscoelastic properties can be correlated to the experimentally determined cracking behavior. Thus, shrinkage data alone may not be sufficient to interpret the compatibility of a material to be used adjacent to another construction material. Although restrained shrinkage test is still practical for assessment of cracking potential, tensile creep and instantaneous deformation, together with shrinkage data, can also be used to evaluate cracking potential besides restrained shrinkage test. ECC mixtures containing FA, especially at high volumes, were found to be more appropriate for repair and strengthening work owing to their low autogenous and drying shrinkages, high tensile creep and better performance under restrained shrinkage conditions. For mixtures with high potential of cracking, shrinkage reducing admixtures may also be used as a remedy to lower cracking potential which is recommended for further studies. Acknowledgments The authors gratefully acknowledge the financial assistance of the Scientific and Technical Research Council (TUBITAK) of Turkey provided under Project: MAG-112M035, Turkish Academy of Sciences, Young Scientist Award program and Feyzi AKKAYA Scientific Activates Supporting Fund (FABED) Young Investigator Research Award. References [1] Li VC, Kanda T. Engineered cementitious composites for structural applications. J Mater Civil Eng 1998;10(2):66–9. [2] Li VC, Mishra DK, Naaman AE, Wight JK, LaFave JM, Wu HC, et al. On the shear behavior of engineered cementitious composites. Adv Cem Based Mater 1994;1(3):142–9. [3] Li VC. ECC—tailored composites through micromechanical modeling. In: Banthia N, editor. Proceeding fiber reinforced concrete: present and the future conference. Montreal: CSCE Press; 1998. p. 64–97. [4] Li VC. Advances in ECC research. ACI special publication on concrete: material science to applications 2002;206–23:373–400. [5] Sahmaran M, Li VC. Durability properties of micro-cracked ECC containing high volumes fly ash. Cem Concr Res 2009;39(11):1033–43. [6] Lepech MD, Li VC. Application of ECC for bridge deck link slabs. Mater Struct 2009;42(9):1185–95. [7] Kim YY, Fischer G, Li VC. Performance of bridge deck link slabs designed with ductile engineered cementitious composites. ACI Struct J 2004;101(6):792–801. [8] Kim YY, Fischer G, Lim YM, Li VC. Mechanical performance of sprayed engineered cementitious composite using wet-mix shotcreting process for repair. ACI Mater J 2004;101(1):42–9. [9] Qian S, Li VC. Influence of concrete material ductility on shear response of stud connections. ACI Mater J 2006;103(1):60–6. [10] Lim YM, Li VC. Durable repair of aged infrastructures using trapping mechanism of engineered cementitious composites. Cem Concr Compos 1997;19(4):373–85. [11] Li M, Li VC. Behavior of ECC/concrete layered repair system under drying shrinkage conditions. J Restor Build Monum 2006;12(2):143–60. [12] Li M, Li VC. High-early-strength ECC for rapid durable repair – material properties. ACI Mater J 2011;108(1):3–12. [13] Zhang J, Li VC, Nowak A, Wang S. Introducing ductile strip for durability enhancement of concrete slabs. J Mater Civ Eng 2002;14(3):253–61. [14] Rizzo EM, Sobelman MB. Selection criteria for concrete repair materials. Concr Int 1989;11(9):46–9. [15] Emmons PH, Vaysburd AM, McDonald JE. A rational approach to durable concrete repairs. Concr Int 1993;15(9):40–5.

383

[16] Mangat PS, Limbachiya MK. Repair material properties which influence longterm performance of concrete structures. Constr Build Mater 1995;9(2):81–90. [17] Morgan DR. Compatibility of concrete repair materials systems. Constr Build Mater 1996;10(1):51–61. [18] Decter MH, Keeley C. Durable concrete repair – importance of compatibility and low shrinkage. Constr Build Mater 1997;11(5–6):267–73. [19] Emmons PH, Vaysburd AM, McDonald JE, Poston RW, Kesner KE. Selecting durable repair materials: performance criteria. Concr Int 2000;22(3):38–45. [20] Poston RW, Kesner K, McDonald JE, Vaysburd AM, Emmons PH. Concrete repair material performance—laboratory study. ACI Mater J 2001;98(2):137–47. [21] Li VC. High performance fiber reinforced cementitious composites as durable material for concrete structure repair. Int J Restor Build Monum 2004;10(2):163–80. [22] Li VC, Stang H. Elevating FRC material ductility to infrastructure durability. In: Proceedings of 6th international RILEM symposium on fibre-reinforced concretes (BEFIB’2004). Varenna, Lake Como, Italy; 2004. [23] Li M, Li VC. Influence of material ductility on the performance of concrete repair. ACI Mater J 2009;106(5):419–28. [24] Sahmaran M, Lachemi M, Hossain KMA, Li VC. Internal curing of engineered cementitious composites for prevention of early age autogenous shrinkage cracking. Cem Concr Res 2009;39(10):893–901. _ Effect of pre-soaked expanded [25] Keskin SB, Sulaiman K, Sßahmaran M, Yaman IÖ. perlite aggregate on the dimensional stability and mechanical properties of ECC. J Mater Civ Eng 2013;25(6):763–71. [26] Boshoff WP, van Zijl GPAG. Time-dependent response of ECC: characterisation of creep and rate dependence. Cem Concr Res 2007;37(5):725–34. [27] Boshoff WP, Mechtcherine V, van Zijl GPAG. Characterising the timedependant behaviour on the single fibre level of SHCC: Part 1: Mechanism of fibre pull-out creep. Cem Concr Res 2009;39(9):779–86. [28] Boshoff WP, Mechtcherine V, van Zijl GPAG. Characterising the timedependant behaviour on the single fibre level of SHCC: Part 2: The rate effects on fibre pull-out tests. Cem Concr Res 2009;39(9):787–97. [29] Boshoff WP, Adendorff JC. Effect of sustained tensile loading on SHCC crack widths. Cem Concr Res 2013;37:119–25. [30] Li VC. Integrated structures and materials design. Mater Struct 2006;40(4):387–96. [31] Lin Z, Kanda T, Li VC. On interface property characterization and performance of fiber reinforced cementitious composites. Concr Sci Eng 1999;1(3):173–84. [32] Yang EH, Li VC. Numerical study on steady- state cracking of composites. Compos Sci Technol 2007;67(2):151–6. [33] Wongtanakitcharoen T, Naaman AE. Unrestrained early age shrinkage of concrete with polyproplene, PVA, and carbon fibers. Mater Struct 2007;40(3):289–300. [34] Wei Y, Hansen W, Biernacki JJ, Schlangen E. Unified shrinkage model for concrete from autogenous shrinkage test on paste with and without groundgranulated blast-furnace slag. ACI Mater J 2011;108(1):13–20. [35] Keskin SB. Dimensional stability of engineered cementitios composites. Ph D Thesis. Ankara, Middle East Technical University, 2012. [36] AASHTO PP 34–99. Standard practice for estimating the cracking tendency of concrete. Washington, DC: American Association of State Highway and Transportation Officials, 1999. [37] Neville AM. Properties of concrete. Essex: Pearson Educational Limited; 2003. [38] Roy DM, Idorn GM. Hydration, structure, and properties of blast furnace slag cements, mortars, and concrete. ACI J 1983;79(6):445–57. [39] Qian S, Li VC. Simplified inverse method for determining the tensile strain capacity of strain hardening cementitious composites. J Adv Concr Technol 2007;5(2):235–46. [40] Qian S, Zhou J, Rooij MR, Schlangen E, Ye G, van Breugel K. Self-healing behavior of strain hardening cementitious composites incorporating local waste materials. Cem Concr Compos 2009;31(9):613–21. [41] Li Y, Bao J, Guo Y. The relationship between autogenous shrinkage and pore structure of cement paste with mineral admixtures. Constr Build Mater 2010;24(10):1855–60. _ Tokyay M. Development of high volume low-lime and [42] S ß ahmaran M, Yaman IÖ, high-lime fly-ash-incorporated self consolidating concrete. Mag Concr Res 2007;59(2):97–106. [43] Zhang MN. Microstructure, crack propagation, and mechanical properties of cement pastes containing high volumes of fly ashes. Cem Concr Res 1995;25(6):1165–78. [44] Bisaillon A, Rivest M, Malhotra VM. Performance of high-volume fly ash concrete in large experimental monoliths. ACI Mater J 1994;91(2):178–87. [45] ASTM C1581. Standard test method for determining age at cracking and induced tensile stress characteristics of mortar and concrete under restrained shrinkage. West Conshohocken (PA): American Society for Testing and Materials; 2009. [46] Mehta PK, Monteiro PJM. Concrete; microstructure, properties, and materials. New York: McGraw-Hill; 2006. [47] Ward MA, Cook DJ. The mechanism of tensile creep in concrete. Mag Concr Res 1969;21(68):151–8. [48] Bissonnette B, Pigeon M, Vaysburd AM. Tensile creep of concrete: study of its sensitivity to basic parameters. ACI Mater J 2007;104(4):360–8. [49] D’Ambrosia MD, Lange DA, Grasley ZC. Measurement and modeling of concrete tensile creep and shrinkage at early age. ACI Spec Publ 2004;2004(220):99–112.