Influence of water saturation on the strength characteristics and deformation behavior of hardened cement paste backfill


 In this study, a uniaxial compression experimental was conducted to examine the mechanical properties of hardened cemented paste backfill (CPB) with different water saturations (0.18%, 4.98%, 9.30%, 21.6%, 32.8%, and 100%). The experimental results demonstrated that water saturation loosened the overall structure of the CPB, which led to the deterioration of its mechanical properties. As the water saturation increased, the uniaxial compressive strength (UCS), residual strength, strength difference, deformation modulus, secant modulus, E50 (the secant modulus at 50% of the UCS), peak strain, and elastic strain decreased, while the plastic strain ratio increased. The UCS, E50, and peak strain demonstrated exponential function relationships with the water saturation. After the peak point, when the water saturation was less than 20%, the strength of the CPB decreased rapidly, and when the water saturation was greater than 30%, the strength decreased slowly. Lastly, the plastic strain, the strain at 50% of the UCS, and the strain at the maximum secant modulus conformed to the normal distribution, and the water saturation had a minimal impact on these three strains. The fractal dimension, D, of the cracks in the CPB increased exponentially with increasing water saturation and demonstrated a negative linear correlation with the UCS.


29
The stope-and-fill method of mining underground metal ore is commonly used worldwide, 30 especially in China, Canada, and Australia, because of its significant advantages in controlling 31 surface subsidence, reducing the discharge of waste rock and tailings, and making efficient use 32 of mineral resources [1][2][3][4][5]. Filling the mined-out area with tailings can greatly reduce the 33 exposed area of the roof, increase mining safety, improve resource recovery, and increase the 34 mine's profitability [6-10]. 35 It is important to study the mechanical properties of CPB, which not only affect its 36 stability but also establish the stability of the surrounding rock and roof, thus determining the 37 safety of the mining operation [11][12][13][14]. Numerous studies, in laboratory and field-scale tests, 38 have illustrated that the mechanical strength and stability performance of CPB is greatly 39 affected by both internal and external parameters, especially in the presence of water [15][16][17][18][19][20]. 40 Cao et al. [21][22] analyzed the influence of structural factors (number of structural planes, 41 angle of structural planes, and filling interval time) on the mechanical characteristics of CPB. 42 Full et al. [23] quantitatively analyzed the influence of sulfate on the strength development of 43 CPB and predicted the CPB's strength evolution. 44 Water is one of the most basic components of the CPB, and thus, plays a vital role [26-45 28]. First, it affects the efficiency of transporting the backfill slurry in the pipeline, and, second, 46 the backfill hydration reaction requires a large amount of water [29][30]. Shortage of water can 47 cause the cement hydration to end prematurely. However, excessive water in the CPB 48 negatively influences its strength and durability [31 -33]. As the mining continues to extend 49 horizontally and vertically, the stope will inevitably expand to the bottom of rivers, lakes, and 50 even the ocean [24][25]. When this occurs, excessive water penetrates the CPB along the cracks 51 in the rock mass. When the CPB contacts excessive water, its stability changes, which can 52 adversely impact safety. Because accurately controlling the water saturation of backfill is 53 difficult, the existing relevant research has primarily focused on dry and fully saturated backfill 54 [34][35]. Liu  with the help of damage mechanics theory, performed an in-depth study of the evolution of the 59 backfill damage for both conditions. 60 Most backfill materials are exposed to humidity, and the seepage water in the cracks of 61 the rock mass and the moisture in the air inevitably affect the mechanical properties of the 62 backfill material. At this time, the CPB in different stopes or in different positions of the same 63 stope is affected by different water saturation (0 to 100%). However, only studying the 64 influence of complete drying and saturation conditions on the mechanical properties of CPB 65 will inevitably have limitations. At present, there are relatively few studies on the effect of 66 water saturation on the mechanical properties of backfill. The environmental conditions of the 67 CPB vary with mines and stopes of the same mine. If we study the mechanical behavior of 68 CPB under completely dry (water saturation of 0%) or wet (water saturation of 100%) 69 conditions, the conclusions obtained will not be representative of most mine conditions, and 70 this creates difficulty with providing guidance to the mines. Therefore, it is essential to study 71 the mechanical behavior of CPB and analyze the mechanism of its strength deterioration under 72 different water saturation conditions to provide a reference for mines to control their CPB 73 stability. 74 From the above analysis, it is of great theoretical and practical significance to study the 75 influence of water saturation on the strength characteristics and deformation behavior of CPB. 76 The specific objectives of this research are: firstly, to qualitatively analyze the influence of 77 water saturation on stress-strain curve, strength characteristics and deformation behavior of 78 hardened CPB; secondly, to quantitatively analyze the influence of water saturation and peak 79 strength, peak strain, elastic strain and secant modulus of hardened CPB; and thirdly, to discuss 80 the strength degradation mechanism and internal crack distribution of water-saturated CPB 81 specimens. 82

83
The materials and methods section should contain sufficient detail so that all procedures 84 can be repeated. It may be divided into headed subsections if several methods are described. 85

Material characteristics 86
The tailings used in this experiment are from Shandong Province, China. The particle size 87 distribution of tailings is measured by SA-CP3 particle size analyzer (Figure1). First, the 88 tailings particles were dispersed into the coal liquid. Due to gravity, the tailings particles 89 settled, and the settling speeds of the different particle sizes were different. The differences in 90 the tailings particle settlement was obtained using the optical system. The optical signal was 91 then transformed into an electrical signal. This electrical signal was converted into data after 92 being amplified by the amplifier, and finally, the test result was output. The median particle 93 size of the tailings was 107.33 μm, and the average particle size was 144.26 μm. The weighing 94 method determined that the density of the tailings was 2.53 kg/m 3 . An x-ray diffractometer was 95 used to analyze the chemical composition of the tailings, and the results in Table 1. The binder  96 was ordinary Portland cement, type 42.5R. The test molds were transparent acrylic tubes with 97 an inner diameter of 50 mm and a height of 100 mm. The sidewall of the molds was marked 98 with a scale, and several filter holes were distributed across the bottom.  The "dry first, then wet mix" method was adopted for the CPB preparation process, to 105 make the slurry mixing more uniform. The tailings and Portland cement, type 42.5R were first 106 dry-mixed for 3 min; the proper amount of water was added, and then the mixture was stirred 107 for 3 min. The cement-to-tailings ratios of all specimens were 1:4, and the slurry concentrations 108 were 75%. The uniformly mixed slurry was poured into the molds, which were then placed in 109 a curing box having a constant temperature of 20 ± 1 ℃ and humidity of 95% ± 5 %. After 7 110 d of curing, the CPB specimens were separated from the external molds and placed in the 111 curing box again. After a total of 60 d of curing, the 24 prepared specimens were removed from 112 the curing box to undergo further treatment. 113

Specimen saturation 114
CPB specimens with different amounts of water saturation were prepared as follows and 115 as shown in Figure 2: 116 (1) All the prepared specimens were placed in the constant temperature and humidity 117 curing box for natural drying for at least 2 weeks. 118 (2) The specimens were then dried for more than 6 d in a 105 °C oven, and their masses 119 were measured every 1−2 day. When the mass of a specimen no longer changed, it was 120 removed from the oven and its mass recorded as md. 121 (3) All specimens were placed in the laboratory and dried naturally for 1 day. 122 (4) All specimens were water-saturated under a vacuum condition, and their masses were 123 measured every 1−2 day. When the mass of a specimen no longer changed, it was removed, 124 and its mass was recorded as mw. Before UCS testing, the mass of each specimen was 125 measured and recorded as mi. The water saturation of the specimen is calculated as follows: 126 where ws is the water saturation of the specimen, mi is the mass of the specimen before 128 compression, mw is the mass of the specimen at full saturation, and md is the mass of the 129 completely dry specimen. 130 (5) Five water-saturated specimens were selected for UCS testing in water (Figure3), the 131 results of which were recorded as the group WS. 132 (6) Three water-saturated specimens were placed in the laboratory for 1 day, after which 133 the UCS testing was conducted, and the results were recorded as the group AD1. 134 (7) Five water-saturated specimens were placed in the laboratory for 2 weeks. The UCS 135 tests were then performed, and the results were recorded as the group AD. 136 (8) Three water-saturated specimens were placed in the laboratory for 2 weeks. The 137 specimens were placed in water for 30 seconds, the UCS testing was conducted, and the results 138 were recorded as the group IW. 139 (9) Three water-saturated specimens were dried in a 105 ℃ oven for 6 days, and the UCS 140 testing was performed, the results of which were recorded as the group OD. 141 (10) The last five water-saturated test pieces were placed in a vacuum tube for 1 day and 142 then underwent UCS testing. The results were recorded as the group VD. 143

147
After the water saturation treatment, all the specimens were packaged in plastic wrap to 148 prevent environmental influence on their water saturation. Testing was conducted on the VD 149 group to verify the plastic wrap's effectiveness in isolating the specimens. After a specimen 150 was enclosed in plastic wrap and placed in the laboratory for 2 hours, a quality control test 151 showed that the mass of the specimen increased negligibly, by 0.006 g. Therefore, the plastic 152 wrap provided an effective isolation environment. 153

Unconfined compressive strength tests 154
The UCS test was performed on the prepared CPB specimens, which had an outer 155 diameter of 50 mm and a height of 100 mm. The equipment used was a GAW-2000 electro-156 hydraulic servo testing machine ( Figure 4). The loading rate of uniaxial compression test is 157 0.5mm/min, and the stress and strain data of CPB can be exported by Excel format. 158

161
The primary test results of this study are shown in σres is the residual strength; deformation modulus is the ratio of the peak stress to the strain at 164 the peak stress. 165

Influence of water saturation on the stress-strain curve 167
The stress-strain curves for all specimens are displayed in Figure 5, and they show that, 168 with increasing water saturation, the strength of the cemented tailings backfill gradually 169 decreased a total of 75.2%, from the peak value of 5.65 MPa (average value) to 1.40 MPa 170 (average value). With increasing water saturation, the slope of the curve before the peak point 171 decreased, that is, the modulus decreased. Additionally, with increasing water saturation, the 172 curve after the peak point changed notably. When the water saturation was less than 20%, the 173 strength decreased rapidly after the peak point. When the water saturation was greater than 174 30%, the strength decreased slowly after the peak point. The residual strength showed a pattern 175 similar to that of the peak strength; with increasing water saturation, the residual strength 176 decreased gradually. Further, the stress-strain curve of the low-saturation backfill surrounds 177 the stress-strain curve of the high-saturation backfill.

Influence of water saturation on strength 190
Referring to the related research results in the field of rock mechanics [38-39], this study 191 used exponential, logarithmic, and power functions to fit and analyze the relationship between 192 the CPB strength and water saturation. The results are shown in Figure 6. 193 Figure 6 shows that, with increasing water saturation, the UCS of the CPB decreased 194 rapidly. The UCS was particularly sensitive to water saturation in the lower saturation range, 195 with a slight change in water saturation causing a rapid decrease in the strength. As the water 196 saturation increased, the sensitivity of the UCS to the water saturation decreased significantly. 197 The residual strength demonstrated similar properties. When the water saturation increased 198 from 0.11% to 32.8%, the uniaxial compressive strength and residual strength decreased by 199 63.4% and 72.5%, respectively; when the water saturation increased from 32.8% to 100%, the 200 strength decreased by 32.4% and 47.9%, respectively. 201 The parameters and R2 values of the fitting functions are also provided in Figure 6, and 202 they show that the degree of fit of the exponential function was higher than for the logarithmic 203 and power functions. The exponential function well represents the observed characteristic of a 204 slower strength change in the later stages, and the water saturation can be taken as zero, while 205 the water saturation of the logarithmic and power functions cannot. 206 207 Figure   where εf is the strain at the peak point, εe is the elastic strain, and εp is the plastic strain. 215 The relationship between these three strain values and water saturation, and the 216 relationship between the plastic strain ratio and water saturation is shown in Figure 7. The 217 plastic strain ratio is defined as follows: Three fitting function types-exponential, logarithmic, and power functions-were used to 220 fit the relationship between the strain and water saturation; the results shown in Figure 7a-c. It 221 can be seen from Figure 7a-b that with increasing water saturation, the peak strain and elastic 222 strain demonstrate patterns of change similar to that of the strength. The greater the water 223 saturation was, the lower the peak strain and elastic strain were, and, as the water saturation 224 increased, the sensitivity of the peak strain and elastic strain to the water saturation decreased. 225 It can be seen from Figure 7a-b that there is no clear relationship between the plastic strain and 226 the water saturation, which is a discrete distribution, distributed between 0.00115 and 0.0019, 227 among which the points between 0.0014 and 0.0016 are the greatest. Figure 7d shows that 228 although the plastic strain distribution was relatively discrete, the plastic strain ratio gradually 229 increased as the water saturation increased. In the lower water saturation interval, the plastic 230 strain ratio increased rapidly, and when the water saturation exceeded 30%, the plastic strain 231 ratio decreased rapidly. 232 As can also be seen from Figure 7, the exponential function had the best degree of fit for 233 the relationships between the water saturation and the peak point and elastic strains. However, 234 the power function had the greatest degree of fit to the relationship between the water saturation 235 and the plastic strain ratio. Based on the conclusion in Section 3.2, the strength decreased with 236 increasing water saturation, and it can be inferred that the strength also decreased with 237 increasing plastic strain ratio.

247
It can be seen from Figure 8a  The two columnar zones in Figure 8a are the strain ranges of E50 (the secant modulus at 259 50% of the UCS) and the maximum secant modulus of each specimen. It can be seen from 260 Figure 8a that the strain distribution was between 0.0025 and 0.0035 when the specimen 261 reached E50 and between 0.004 and 0.005 when it reached the maximum secant modulus. 262 The E50 of each specimen was calculated as provided in Figure 8b, which shows that, with 263 increasing water saturation, the E50 demonstrated a similar relationship with the strength and 264 strain, that is, when the water saturation was low, the E50 was particularly sensitive to the water 265 saturation, and the rate of change decreased as the water saturation increased. When the water 266 saturation increased from 0.11% to 32.8%, the E50 decreased by 53.7%; when water saturation 267 increased from 32.8% to 100%, the E50 decreased by 41.1%. 268 Exponential, logarithmic, and power functions were also used to fit the relationship 269 between the E50 and water saturation. The fitting parameters and R 2 are shown in Figure 8b. 270 The exponential function had the highest degree of fit and characterizes the rule of change of 271 the E50 with water saturation better than that by logarithmic or power functions. 272 In conclusion, the water saturation had a significant effect on the mechanical properties  where σf (ws) is the UCS of the specimen having a water saturation of ws, and a, b, and c

311
Eq. 14 is used for fitting and precision analysis. Figure 9 shows the fitting results between 312 the strength difference and water saturation, and the upper right corner of Figure 9 shows the 313 residual value after fitting. It can be observed that R 2 is 0.984 after fitting with Eq. 14, and the 314 residual distribution is (−0.4, +0.4). Only one abnormal point was noted, when the water 315 saturation is 4.853%, illustrating the extremely high degree of fit of Eq. 14. Therefore, in 316 mining engineering, Eq. 14 can be used to predict the strength of CPBs having different water 317 saturations. 318

Interdependency of strain and water saturation 319
It can be seen from the results in Sections 3.3 and 3.4 that, with increasing water 320 saturation, the plastic strain, the strain corresponding to the E50, and the maximum secant 321 modulus are discrete distributions. The laws of distribution of these three variables were 322 studied as follows. 323 First, the degrees of dispersion and laws of distribution of the three strains were analyzed 324 by using a box plot [46][47]. A typical box plot is shown in Figure 10. The distribution range 325 of a group of data can be intuitively determined using the box plot, and the form of distribution 326 of the data set can also be preliminarily determined. Data points that exceed the upper and 327 lower boundaries of the box plot are considered abnormal values. 328 329 Figure 10: Box plot.

330
The box plots of the plastic strain, the strain at the E50, and the strain at the maximum 331 secant modulus are shown in Figure 11. Figure 11a shows the box plot of the plastic strain, 332 which shows that most of the plastic strains of the specimens with different water saturations 333 were distributed within the upper and lower edges. There were a total of 16 data points, of 334 which 11 were distributed within the range of the box body from 0.00155 to 0.00145, so it can 335 be preliminarily determined that the plastic strain followed a normal distribution. Figure 11b  336 is the box plot of the strain corresponding to 50% of the UCS and the strain corresponding to 337 the maximum secant modulus. It can be seen from Figure 10b that the data points for both 338 strains were within the upper and lower edges, there were no abnormal points, and the average 339 values and median lines of both strains were almost equal, indicating that the two groups of 340 data have better concentrations. The strain at 50% of the UCS was mostly distributed in the 341 range of 0.00315-0.00348. There were 13 points in total, with five data points between the 342 upper quartile and the upper edge and five between the lower quartile and the lower edge. The 343 strain at the maximum secant modulus was primarily distributed in the range of 0.0045-0.005. 344 There were 14 data points in total, with four data points between the upper quartile and the 345 upper edge and five between the lower quartile and the lower edge. Therefore, it can be 346 preliminarily determined that the two strains conformed to the normal distribution. 347 348 Figure 11: Box plots of (a) plastic strain and (b) strain at E50 and maximum secant modulus.

349
To further determine the distribution of the three strains, a distribution histogram and a 350 Q-Q plot of each strain were developed, as shown in Figure 12-14. Figure 12 is the distribution 351 histogram of the plastic strain, which shows that most of the plastic strains were concentrated 352 between 0.00140 and 0.00155, accounting for 60.9% of the total. There were few other interval 353 distributions, but, due to its non-symmetric distribution, the plastic strain was closer to a 354 lognormal distribution. It can also be seen from Figure 12b that most of the data points were 355 distributed along the reference line. Using the K-S test, the P-value was 0.384, which is greater 356 than the 0.05 reference value, indicating that it conformed to the typical lognormal distribution 357 characteristics. Similarly, the strain at 50% of the UCS and the strain at the maximum secant 358 modulus underwent distribution fitting and the K-S test, which indicated that both conformed 359 to the normal distribution. The strain at 50% of the UCS was mostly concentrated in the range 360 of 0.0032 to 0.0034 and 0.0034 to 0.0036; the total of the two accounted for 60.8%; the P-value 361 was 0.107. The strain at the maximum secant modulus was mostly distributed in the range of 362 0.0046 to 0.0048, and the P-value was 0.469. The P-values of the two types of strains were 363 greater than 0.005, indicating that they conformed to the typical normal distribution 364 characteristics. 365 In summary, the three types of strain data all conformed to the normal distribution, 366 indicating that the relationships between the three types of strain and the water saturation are 367 independent of each other, i.e., the water saturation will not be affected by these three types of 368 strain. 369  There are two principals' reasons for the strength change of the CPB with varying water 382 saturation: the chemical corrosive action of water and the mechanical action of water. Due to 383 the long curing time (60d), the internal hydration process was completed, and the chemical 384 reaction had little effect on the strength. Thus, the strength difference was chiefly caused by 385 the physical effects of the water. During the CPB drying process, the water in the inner 386 capillaries gradually evaporated, and the capillaries shrank, resulting in an increase in capillary 387 suction in the CPB. When the CPB was treated with water saturation, capillary expansion was 388 caused by the continuous penetration of the water into the internal capillary, and the capillary 389 suction in the CPB was gradually reduced, resulting in an overall strength reduction. 390 Additionally, the water permeation led to the loosening of the overall structure of the CPB and 391 the deterioration of the overall strength of the CPB. 392 Figure 15 shows the surface crack distribution of the CPB specimens with different water 394

Distribution characteristics of cracks 393
saturations. It can be seen from Figure 15 that increased water saturation led to an increased 395 number and density of cracks. It can also be observed that most of the cracks were distributed 396 around the CPB specimens' peripheries, and, with increasing water saturation, the cracks 397 extended deeper into the center of the CPB specimens.

401
With the help of fractal theory [51][52], the fractal dimensions of the crack outlines in the 402 CPB specimens with different water saturations were determined by the box-counting method, 403 and the results are shown in Figure 16. It can be seen that the cracks in the CPB specimens 404 with different water saturations demonstrated fractal characteristics. The fractal dimension, D, 405 increased with increasing water saturation, and the change in the fractal dimension, D, is 406 inverse to the change in strength. Using exponential, logarithmic, and power functions to fit 407 the relationship between the fractal dimension and water saturation, it was found that the degree 408 of fit of the exponential function was the highest. This revealed that the relationship between 409 the fractal dimension of the cracks and the water saturation is an exponential function, 410 expressed as: 411 0.042 0.23 1.31 where D is the fractal dimension of the cracks in the CPB. 413 Further, the relationship between the UCS of the CPB and its fractal dimension was 414 obtained, as shown in Figure 16.

420
Water saturation has a great influence on the mechanical properties of CPB. In this paper, 421 uniaxial compression tests were carried out on five kinds of CPB specimens with different 422 water saturation. The influence of water saturation on CPB strength, deformation and apparent 423 crack was discussed. Based on the research of this paper, the main conclusions are as follows: 424 (1) The deformation modulus of the CPB decreased with increasing water saturation. 425 After the peak point, the strength of the CPB decreased rapidly when the water saturation was 426 less than 20%, and the strength decreased slowly when the water saturation was greater than 427 30%. 428 (2) The relationship between the water saturation and the UCS, residual strength, peak 429 strain, elastic strain, and strain ratio had an exponential function: residual strength, peak strain, and elastic strain of the CPB decreased and the strain ratio of the 431 CPB increased with increasing water saturation. There was no apparent relationship between 432 the plastic strain and water saturation. 433 (3) The secant modulus of the CPB decreased with increasing the water saturation, and it 434 first increased and then decreased (the critical point was 0.5%) with increasing strain. The E50 435 decreased with increasing water saturation as an exponential function: 3 33 cx y a b e  . 436 (4) The strength difference of the CPB decreased with increasing water saturation, and 437 there was a strong exponential function relationship between them. The water saturation had 438 little effect on the plastic strain, the E50 strain, or the maximum secant modulus strain. 439 (5) Most of the cracks in the CPB specimens with different water saturations were 440 distributed around the periphery of the specimens. The greater the water saturation was, the 441 more cracks that were present in the CPB. The fractal dimension, D, of the cracks in the CPB 442 increased exponentially with increasing water saturation. The fractal dimension, D, has a 443 negative linear correlation with the UCS. 444 This study considered the effects of water saturation on the mechanical properties of CPB 445 and demonstrated that water saturation has apparent degradative effects on the mechanical 446 properties of CPB. However, the influence of water saturation on the mechanical properties of 447 CPB is closely related to the cement-to-tailings ratio, slurry concentration, and curing age, 448 among other factors. This study can provide a scientific reference for the strength design of 449 CPB when the underground is rich in water. 450

463
The authors declare that there is no conflict of interest regarding the publication of this 464 paper. 465