Effect of concrete workability on bond properties of steel rebar in 1 pre-cracked concrete 2

9 Although previous research has shown a considerable influence of the pre-cracking phenomenon 10 on steel-congested concrete members, only normal concrete (NC) has been considered in the 11 literature. Hence, this paper intends to study the effect of the pre-cracking phenomenon on the 12 bond response of pre-cracked NC with different slump flow values and self-consolidating concrete 13 (SCC). Initial crack widths ranging from 0.0 to 0.5 mm are studied. Results show that initial crack 14 widths larger than 0.10 mm have a significant influence on bond properties so that higher than 15 30% and 50% reduction factors are obtained for the maximum bond strength of concrete specimens 16 exposed to the initial crack widths of 0.2 mm and 0.4 mm respectively. Results show that concrete 17 mixtures with higher workability are less sensitive to the pre-cracking phenomenon as compared 18 to NC mixtures. The average bond stress of steel rebar in the pre-cracked SCC is found to be 19 similar to that of the NC with a slump flow of 200 mm, which is considerably better than for NC


27
Recently, there has been a growing tendency of studying bond characteristics of steel rebar 28 embedded in cracked concrete, denoted as "pre-cracking phenomena", in which induced initial 29 cracks generate and propagate through a plane containing a reinforcing bar (rebar) axis (Mousavi 30 et al., 2019, Brantschen et al., 2016, Lindorf, 2011, Matsumoto et al., 2016, Mousavi et al., 2020b. 31 Plastic shrinkage cracking in the steel-congested concrete region (Hadidi and Saadeghvaziri, 2005, 32 Saadeghvaziri and Hadidi, 2005) and accidental internal damages due to the previous overloading 33 (such as earthquakes and/or overstress situations) (Matsumoto et al., 2016) can cause the pre-34 cracking phenomenon, where corrosion plays no direct role in generating the internal crack. Until 35 now, different expressions have been used in the literature for describing the pre-cracking 36 phenomenon including mechanical pre-loading (Brantschen et al., 2016, Brantschen, 2016, 37 biaxial load transfer (Lindorf et al., 2009, Saadeghvaziri andHadidi, 2005), multiaxial stress states 38 (Purainer, 2005), and transverse tension (Lindorf, 2011). These situations cause the crack 39 propagation parallel to the rebar direction, resulting in internal damages at the rebar-concrete 40 interface. 41 bond properties in the pre-cracked concrete. They recommended the use of rebar rib geometry (rib 48 orientation, height, and spacing) for future studies related to the pre-cracking phenomenon 49 (Brantschen et al., 2016). In this field, Mousavi et al. (2019)  slip response in pre-cracked NC. They used rebar deformations (rib height and spacing) and crack 54 width to introduce an analytical model for predicting the maximum bond strength of rebar 55 embedded in the pre-cracked concrete. They showed that the pre-cracking phenomenon reduces 56 the lateral concrete confinement surrounding the rebar. 57 As reported in previous research, steel-congested concrete members have been mostly affected by 58 the pre-cracking phenomenon (Matsumoto et al., 2016) including the typical surface crack pattern 59 of a slab specimen reinforced with transverse elements (cracks in punching area around column) 60 (Brantschen et al., 2016), flexural reinforcement in slabs (Dawood and Marzouk, 2012), and 61 reinforced concrete (RC) beam-column joints (Joergensen and Hoang, 2015). However, only NC 62 has been considered in previous studies, while relatively new concrete generations have been 63 introduced for using in the steel-congested regions to maintain desired structural behavior, such as 64

Specimens and test set-ups
For each mixture, 3 cylindrical specimens with a diameter of 100 mm and a height of 200 mm 110 were prepared to measure the concrete compressive strength. A total number of 26 cylindrical 111 specimens with a diameter of 150 mm and a height of 113 mm were also considered for pull-out 112 tests under monotonic loading including 9 uncracked and 17 pre-cracked specimens ( Fig. 2(a)). 113 All specimens were cured for 28 days in a moisture room at 97.3% relative humidity (RH) and 23 114 ºC temperature. To simulate the pre-cracking phenomenon, an indirect tensile test (Brazilian 115 splitting test) was considered for generating cracks perpendicular to the direction of the rebar 116 placed at the center of cylindrical specimens ( Fig. 2(b)). A displacement-controlled loading with 117 a rate of 0.15 mm/min was applied to prevent unexpected splitting failure during the pre-cracking 118 loading. To measure the initial crack width, two crack gauges were installed at both sides of 119 concrete cylinders. As crack width changes with unloading, the ultimate initial crack width was 120 directly measured after stopping the pre-cracking procedure (Fig. 2(c)). Direct pull-out tests were 121 carried out by applying tensile force directly to the rebar. An MTS testing machine with a load 122 capacity of 250 kN was used to apply the tensile load through a displacement-controlled protocol 123 at the rate of 0.5 mm/min. The unloaded end slip was measured with a linear variable differential 124 transformer (LVDT). To provide a relatively uniform distribution of bond stresses, the embedment 125 length of the rebar was 50 mm (five times the nominal diameter of rebar) in all specimens. This 126 short-embedded length provides a better measurement of the local bond stress. Plastic sleeves were 127 used for covering the unbonded length ( Fig. 2(a)). An automatic data acquisition system was used 128 to record the data. Crack opening during the pull-out test was also measured by crack gauges for 129 obtaining bond stress-crack opening curves. This curve is used to obtain the bond fracture energy.

7
It was tried to impede the splitting bond failure in uncracked specimens by providing enough 131 concrete cover around the rebar ( = 7.5 ⁄ ). In all cases, the rebar was positioned at the center 132 of cylinders. The used steel rebar has a nominal diameter of 10 mm with a specified yield strength 133 and ultimate tensile strength of 432 MPa and 620 MPa respectively. The surface characteristics 134 and rib pattern of the steel rebar used are shown in Fig. 2(d). The average value of rib-face angle 135 for rebar used in the experimental program was about 55 degrees. The rib spacing-to-rib height 136 ratio ( ℎ ⁄ ) was about 7.0. 137

138
As mentioned widely in the literature (Guizani et al., 2017, Wu andZhao, 2012), for the short bond 139 region (generally five times the rebar diameter), bond stress is close to uniform distribution and 140 can be averaged along the anchorage length to get a close estimate of the local bond stress. Hence, 141 the local bond stress, (N/mm 2 ), can be calculated by Eq. (1): 142

=
(1) where is the tensile load, is the rebar diameter, and is the embedded length. As 143 recommended by RILEM (TC, 1994), arithmetic mean of bond stresses (denoted as "average bond 144 stress") is calculated by Eq. (2), in which variables 0.01 , 0.10 , and 1.00 corresponding to bond 145 stresses at slips of 0.01 mm, 0.10 mm, 1.0 mm respectively. 146 Overall, results of uncracked and pre-cracked concrete, as well as the failure modes, are 152 summarized in Table 2. Initial crack widths ranging from 0.0 (uncracked) to 0.5 mm were obtained 153 by the pre-cracking tests. Specimens are designated by the type of concrete (NC1, NC2, and SCC) 154 followed by the letter "C" and initial crack width in mm for pre-cracked concrete, or only "U" for 155 uncracked concrete. Although three specimens were considered for every initial crack width, only 156 two repetitions were obtained in some cases due to the brittle nature of the pre-cracking test and 157 to the difficulty of controlling the target initial crack width. The pull-out failure mode was observed 158 for uncracked specimens ( Fig. 3(a)). Similarly, pre-cracked specimens with initial crack widths 159 smaller than 0.15 mm ( < 0.15 mm) failed by pulling out the rebar without splitting the 160 surrounding concrete. Similar results for NC mixtures were reported by Mousavi et al. (2019), in 161 which small initial crack width has no impact on bond failure mechanism. However, large initial 162 crack widths significantly affect the maximum bond strength along with the failure mode so that 163 the splitting failure mode was obtained for pre-cracked specimens with ≥ 0.15 ( Fig. 3(b)). 164

165
Bond-slip curves of uncracked specimens for different concrete mixtures along with the 166 normalized bond properties are illustrated in Fig. 4(a). Although comparable results are obtained 167 for the normalized average bond stress, general results indicate that NC2 mixture has the highest 168 interfacial maximum bond strength and NC1 mixture has the lowest values among the other 9 mixtures ( Fig. 4(b)). Comparing mixtures with the same average bond stress ( ), as 170 recommended by RILEM, is appropriate in the section of pre-cracked concrete as this parameter 171 is meaningful notably about on both bond strength and the initial stiffness. Although SCC and 172 NC2 mixtures have an approximately similar compressive strengths of 38.80 MPa and 40.34 MPa 173 respectively (Table 1), SCC mixture has lower bond properties as compared to NC2 mixture. 174 However, the maximum and residual bond strengths of SCC mixture are higher than NC1 mixture 175 with the same water-to-powder ratio of 0.41. Many studies have reported the higher maximum 176 bond strength of steel rebar in SCC than NC (Mousavi et al., 2017, de Almeida Filho et al., 2008, 177 Valcuende and Parra, 2009, Sabău et al., 2016, Zhu et al., 2004, Desnerck et al., 2010, while some 178 studies have shown comparable or slightly lower results (Esfahani et al., 2008, Castel et al., 2006, 179 Pandurangan et al., 2010, König et al., 2001, Schiessl and Zilch, 2001, Gibbs and Zhu, 1999, 180 Lorrain and Daoud, 2002. These observed inconsistency of the test results may attributed to the 181 concrete compositions and the experimental conditions considered in the literature (Sfikas and  182 Trezos, 2013). 183

184
Bond-slip curves of pre-cracked specimens for the studied concrete mixtures are shown in Fig. 5. 185 General results show that the pre-cracking phenomenon has a significant impact on the bond-slip 186 curve so that a high reduction is observed for the bond energy in pre-cracked concrete. In the case 187 of uncracked concrete, there is a plateau after the maximum bond strength, while a sudden drop is 188 observed for pre-cracked concrete in all mixtures. As the initial crack width increases, the slope of 189 the descending branch of the bond-slip curve increases, which shows a more rapid drop in bond 190 stress with increasing slip (Fig. 5). Although specimens with small initial crack widths ( < 0.15 mm) have similar plateau at the peak, the pre-cracking phenomenon affects the post-peak bond 192 response so that the slope of the descending branch of the bond-slip curve is steeper as compared 193 to uncracked concrete ( Fig. 5(a, c)). This leads to about 39.0% and 42.0% reductions in the area 194 under the bond-slip curves (characteristic bond energy, ) ( Table 2) To determine the bond strength reduction due to the pre-cracking phenomenon, a reduction factor 198 is defined as below: 199 where and − are bond strengths of uncracked and pre-cracked concrete 200 respectively, which are listed in Table 2. Reduction factors corresponding to the maximum bond 201 strength of mixtures are illustrated in Fig. 6(a). Results show that SCC mixture has the lowest 202 reduction factor among the other mixtures which indicate that, in terms of the maximum bond 203 strength, SCC mixture is less sensitive to the pre-cracking phenomenon than NC mixture. This can 204 be attributed to the high paste content in SCC mixture as compared to NC mixture. Regarding NC 205 mixtures, as slump flow value increases, the sensitivity of concrete to the pre-cracking 206 phenomenon decreases, so that NC2 has a lower reduction factor as compared to NC1 for the 207 maximum bond strength. Regarding uncracked concrete, Fu and Chung (1998) and Pop et al. 208 (2015) reported that with an increase in the slump (fluidity) of concrete mixture, the interfacial 209 microvoids around the rebar decreases causing a higher bond strength. Uniform distribution of 210 matrix and aggregate surrounding the rebar, due to the higher slump of concrete, may increase the 211 probability of the aggregate interlocking phenomenon across the initial cracks. This phenomenon 11 causes an increase in the maximum bond strength and average bond stress of high slump flow 213 mixtures. As shown in Fig. 6(a), the reduction factor of the bond strength has a good empirical 214 correlation with the initial crack width ( ) for all mixtures as follows: 215 Similarly, in the case of the average bond stress, NC1 has the highest reduction factor ( Fig. 6(b)), 216 while SCC and NC2 mixtures have approximately similar reduction factor. Although there is no 217 precise trend for the reduction factor of the residual bond strength (Fig. 6(c)), the general tendency 218 indicates that SCC mixture has the lowest reduction factor among the other mixtures. 219 Regarding small initial crack widths ( ≤ 0.10 mm), as mentioned in Table 2, the pull-out failure 220 mode was observed for SCC specimen, while the failure was sudden for NC2 specimen causing 221 concrete splitting along with pulling-out the rebar. As shown in Fig. 7(a), initial crack widths cause 222 16.92% and 5.73% maximum bond strength reductions in NC2 and SCC mixtures respectively. 223 Similar observations are obtained for the average bond stress ( ) and the residual bond strength 224 ( ). However, comparable bond energy ( ) is obtained for both mixtures. Regarding the average 225 bond stress, even improvement in bond strength is observed for pre-cracked SCC specimens as 226 compared to the uncracked ones ( Fig. 7(a)), which may be due to the aggregate interlock at the 227 crack surfaces. Crack opening of specimens was also measured for NC2 and SCC mixtures. Two 228 crack gauges were installed at both sides of rebar during the pull-out tests. As illustrated in Fig.  229 7(b), for pre-cracked SCC specimen, similar crack width opening is observed during the tests for two gauges, while different values of crack width opening are noted for NC2 mixture. Moreover, 231 results show that an approximate linear ascending branch is observed for both mixtures prior to a 232 plateau, before reaching the maximum bond strength. The stiffness of this initial branch of the 233 curve is steeper for NC2 mixture specimens as compared to SCC mixture. The length of this 234 plateau is longer for SCC mixture specimens as compared to NC2 mixture, which means that 235 higher sustained bond strength is obtained. The area under the bond-crack opening curve 236 demonstrates the fracture energy. Similar to the bond energy shown in Fig. 7(a), for a small initial 237 crack width, pre-cracked SCC specimen shows slightly higher/comparable fracture energy for the 238 ascending branch of the bond stress-crack opening curve ending at the maximum bond strength 239 ( Fig. 7(b)), as compared to NC2 specimen. 240 Bond stress-crack opening curves of the pre-cracked specimens with large initial crack widths are 241 illustrated in Fig. 8. Results show that an unsymmetrical crack opening phenomenon happened 242 during pull-out tests (Fig. 3(c)). Results presented in Fig. 8 show that this phenomenon is more 243 crucial for NC2 specimens as compared to SCC ones, causing different bond stress-crack opening 244 curves for two crack gauges installed at both sides of the steel rebar. The results shown in Figs. 7 245 and 8 indicate that the crack opening of large initial cracks is different from the small ones. Crack 246 opening more than 6.0 mm is observed for large crack widths (Fig. 8). Crack opening is very small 247 until the peak (maximum bond strength, ), while considerable crack opening values and also 248 sudden drops in curves are observed after the peak. Additionally, crack gauges installed at both 249 sides of rebar show different behaviors before and after the maximum bond strength. This indicates 250 the existence of non-uniform (or unsymmetric) crack opening surrounding the rebar after the pre-251 cracking phenomenon that arises mostly from the brittle nature of the cracking in concrete (Fig.  252 8). The area under the bond stress-crack opening curve can be used as the "fracture energy, 13 " of the bond mechanism in the pre-cracked specimen. The maximum, minimum, and 254 average values of the fracture energy of NC2 and SCC mixtures are shown in Fig. 9, concerning 255 the initial crack widths. Results show that as the initial crack width increases, the average fracture 256 energy decreases so that 40.7%, 13.9%, 7.9%, and 2.8% values are obtained for fracture energies 257 of specimens exposed to 0.15, 0.2, 0.3, 0.4, and 0.5 mm initial crack widths. Similar results are 258 observed for the minimum and maximum values of the fracture energy. Finally, Eq. (7) is achieved 259 for predicting the fracture bond energy of pre-cracked specimens as a function of the secondary 260 crack opening (́) with a good correlation of R 2 =0.99, as follows: 261 = 42.45́− 1.62 (7)

262
As reported by Mousavi et al. (2019), rebar diameter has a considerable impact on the influence 263 of the pre-cracking phenomenon, so that normalized bond stress is related to the initial crack width-264 to-rebar diameter ratio ( ⁄ ). A similar trend is illustrated in Fig. 10 value of the initial crack width corresponding to SCC mixture shows a considerable resistance to 274 the pre-cracking phenomenon. General observations show that the initial crack width significantly 275 changes the interfacial bond behavior so that bond failure mode is changed for initial crack widths 276 larger than 0.15 mm. The results summarized in Table 2 where is a reduction ratio of the bond strength that considers the opening of cracks due to the 296 pre-cracking phenomenon. M1 and M2 are empirical coefficients for different concrete 297 compositions, which are summarized in Table 3. As shown in Fig. 2(d), value of 1.89 mm was 298 measured for the rib height of the steel rebar. 299 The performance of the proposed model is shown in Fig. 11. The dependence of the reduction ratio 300 ( ) on the ratio of the initial crack width-to-rib height (Eq. (12)) can be attributed to the fact that 301 the loss of bond strength is related to a reduction in the interfacial contact between the steel ribs 302 and the concrete. For a well-confined situation (uncracked concrete and ≤ 0.10), the interface 303 separation is very small; consequently, the reduction factor will be equal or very close to one ( ≈ 304 1.0). Different separation mechanisms can be occurred for a pre-cracked specimen, including one-305 side and both-side separations. The value of 0.50 is considered in this section, which corresponds 306 to the initial crack width of et al. (2016) as an idealized pre-cracking situation. Based on the experimental results (Table 3), 308 the domain of 2 ⁄ ≤ 0.026ℎ is obtained (for all mixtures) in the Eq. (12) for the reduction factor 309 equal to 1.0. Based on the model presented by Murcia-Delso and Benson Shing (2014), if the 310 interface separation is larger than the rib height, the contact between ribs and the concrete is lost, 311 and the bond stress disappears ( = 0). However, the values of 0.14ℎ , 0.16ℎ , and 0.19ℎ are 312 obtained for NC1, NC2, and SCC mixtures respectively, as the critical separation values for a 313 complete loss of the bond strength. These values correspond to the initial crack widths of ≈ 0.61, 314 0.52, and 0.70 mm for NC1, NC2, and SCC mixtures respectively (Figs. 6 and 10). This may be 315 attributed to the significant reduction in the concrete compressive strength due to the pre-cracking 316 phenomenon, which has been reported by several works (Vecchio and Collins, 1993, Kollegger 317 and Mehlhorn, 1990, Shirai and Noguchi, 1989, Mikame et al., 1991, Belarbi and Hsu, 1991. 318 Considerable effect of compressive strength on the rebar-concrete interfacial strength has been 319 frequently confirmed (Guizani et al., 2017, Mousavi et al., 2017. 320 It is worth mentioning that the proposed equations are validated for the experimental results of the 321 present study. More experimental investigations are necessary to determine the efficiency of the 322 proposed equations and to generalize them for different concrete mixtures. 323

324
An experimental program was conducted in the present study to determine the effect of concrete 325 workability (or flowability) on the pre-cracking phenomenon. Three concrete mixtures with 326 normal, moderate, and high flowability were considered. Specimens with different initial crack 327 widths were obtained through the Brazilian test (splitting). Pull-out results of the pre-cracked 328 specimens were compared with the uncracked ones. According to the experimental results, the 329 following conclusions can be drawn: 330 -In the case of uncracked specimens, general results show that SCC mixture has 13.2% 331 higher normalized maximum bond strength as compared to NC1 mixture (normal concrete 332 with a high slump flow of 97 mm) with the same water-to-powder ratio of 0.41. However, 333 SCC mixture has 20.4% lower normalized maximum bond strength than NC2 mixture 334

Conflict of interests
The authors declare that there are no competing interests regarding the publication of this paper. length of the top horizontal and diagonal section of ribs rebar diameter Absorbed energy by the bond mechanism tensile force of pull-out test fracture bond energy ′ concrete compressive strength ℎ rib height of rebar embedded length (5 ) M1, M2 empirical coefficients for Eq. (11) bond properties reduction factor due to the pre-cracking phenomenon 19 spacing of periodical ribs 0 effective length of crushed concrete between two adjacent ribs initial crack width ́ crack opening during the pull-out test of pre-cracked specimens w/c water-to-cement ratio w/p water-to-powder ratio bond stress average bond stress as defined by RILEM