**Cement stabilization **

The main factor that causes the swelling-shrinkage potential in clay soils is the clay minerals forming them (Montmorillonite or Smectite). According to Table 2, the predominant clay mineral of Bentonite soil was Montmorillonite. So, it was the clay soil with high swelling-shrinkage potential. One of the methods to deal with the swelling-shrinkage potential problem of clay soils was to stabilize them with cement.

**Strength**** parameters**** evaluation**

Initially, the Bentonite soil was stabilized with 4, 8, and 12% cement and cured for 7 and 28 days. The UCS tests results conducted on them are shown in Fig. 3. Fig. 3(a) shows their stress-strain curves. The diagrams of changes in their amounts of qu, toughness, *ε*f, and E50 are displayed in Figs. 3(b, c, d, and e), respectively. The Bentonite clay soil sample had significant ductility due to the high water absorption and completely plastic behavior. It showed the failure strain with a value of about 5%. The applied stress after the failure point caused it to take on the shape of a barrel, and no crack was created on it. The unstabilized Bentonite sample was not evaluated as suitable due to the low strength and high strain. With the addition of 4% cement and curing for 7 days, the E50 and qu values of Bentonite soil increased by 5 and 6 times, respectively, so that the toughness value against failure tripled. After 28 days of curing, the qu, *ε*f, and E50 values compared to 7 days of curing increased by 65, 22, and 72%, respectively. As a result, the toughness value increased by 51%.

By adding 8% cement and curing for 7 days, the qu and E50 values became 3 and 2.7 times that of the Bentonite sample with 4% cement, respectively. Its stress-strain curve was on top of that of the 7-day stabilized sample with 4% cement. As a result, the amount of toughness increased by 80%. For the stabilized sample with 8% cement after 28 days of curing, the qu, E50, and toughness values increased by 48, 12, and 78% compared to the 7-day stabilized sample, respectively. Considering that the increases of the qu and stiffness values were less than 50%, this significant increase of the toughness value of the 28-day stabilized sample compared to the 7-day stabilized sample was due to the rise of the *ε*f value by 27.8%. After 7 days of curing, the qu, E50, and toughness values of the stabilized sample with 12% cement compared to the stabilized sample with 8% cement increased by 65, 21, and 35%, respectively. They increased by 32, 46, and 18% after 28 days of curing, respectively. These increases of the qu, E50, and toughness values of the sample with 12% cement compared to the sample with 8% cement against multiply increases of these parameters for the sample with 8% cement compared to the sample with 4% cement were negligible. So, the optimum cement content was 8%, which was economical. By adding more cement than 4%, the qu and also *ε*f values increased, but the increase of the *ε*f value was not significant. The stabilization of natural Bentonite with cement caused the failure strain values of both 7-day and 28-day stabilized samples to decrease by about 40% and reach less than 3%. According to their stress-strain curves in Fig .3(a), the stress dropped sharply after the failure point. Despite the ductile behavior of the natural Bentonite sample, the cement stabilization resulted in the brittle behavior.

The curing time had little effect on the strength parameters values of the sample with 8% cement. The qu and E50 values of the 28-day stabilized sample with 12% cement increased by 19 and 35% compared to the 7-day stabilized sample, which were slight amounts. After 28 days of curing, the increase of the strength parameters values for both stabilized samples with 8 and 12% cement was less than 47% compared to corresponding values of their 7-day stabilized samples. So, the pozzolanic reactions that increase the strength and stiffness values to more than 70% in the long term have not occurred or have occurred at a weak level. According to Table 2, the Bentonite soil contained approximately 45 wt% Montmorillonite mineral of the total amount. The high content of clay mineral not only prevented the complete mixing of the cement with the soil, but even it increased the cement content needed to stabilize. In general, if the range of the soil PI is more than 30%, the mixing of soil with cement is difficult, and stabilization is not well done (Little et al. 1987). After 28 days of curing, the sample with 12% cement had the minimum progress of the pozzolanic reactions. Therefore, by increasing the cement percentage, the curing time slightly affected pozzolanic reactions between the cement and the Bentonite soil containing the predominant mineral of Montmorillonite. They gained their maximum strength parameters within 7 days of curing.

**Evaluation of durability against W/D cycles**

The cement-stabilized Bentonite samples were subjected to the durability process after 28 days of curing. As soon as the stabilized samples with 4 and 8% cement were soaked in water in the first cycle, they collapsed. The surface of the stabilized sample with 12% cement was filled with cracks due to the severe shrinkage of Bentonite soil after 24 h of drying in the first cycle. These cracks were not superficial and deep. As soon as the sample was immersed in water for the second cycle, it collapsed (see Figs. 5(a and b)). In order to find the cement percentage with which Bentonite soil was stabilized to last up to six cycles, the 28-day stabilized samples with 20 and 30% cement were also tested. Still, they eventually lasted up to two cycles of W/D. The pozzolanic reactions that occur over time by exchanging the cations between the soil particles and cement lead to the cementation compounds formation. These compounds increase the strength parameters values of the 28-day stabilized samples compared to the 7-day stabilized samples by more than 70%. Also, they bind the soil particles to each other to provide the strength and rigidity of the stabilized sample to last against W/D cycles. The pozzolanic reactions in the stabilization of Bentonite soil with cement had not been performed, or they had been performed at a poor level. It was due to the slight increase of the long-term strength parameters of cement-stabilized Bentonite soil compared to their short-term strength parameters. On the other hand, the 28-day stabilized samples did not withstand six cycles of W/D even with 30% cement.

Despite the stabilization of the Bentonite soil sample with cement, its shrinkage potential was so severe that it created deep cracks on the sample surface. When this stabilized sample was placed in water, the water penetrated through these cracks, and the sample lost the rigidity and collapsed. Therefore, another adhesive additive was needed to bond the Bentonite soil particles together to overcome the shrinkage potential during drying in each cycle. In this study, epoxy resin was employed. In order to use the lowest amount of epoxy resin that improved the durability of the stabilized samples, its combination with cement was investigated. The long-term curing time on improving the strength parameters values of the stabilized Bentonite soil samples with cement was not effective. On the other hand, the clogging of epoxy resin is in the early hours. So, the stabilized samples with cement and epoxy resin were treated for 7 days. They were then subjected to the durability process.

**Stabilization with cement and epoxy resin**

In order to find the lowest amount of epoxy resin that the cement-stabilized samples did not collapse after six cycles of W/D, the addition of epoxy resin was started from the ER/W ratio equal to 0.25. The cement-stabilized samples were then tested by adding epoxy resin with the ER/W ratios of 0.5, 1, and 2. The results of the uniaxial tests performed on 7-day cement and epoxy resin-stabilized samples with the ER/W ratios of 0.25, 0.5, 1, and 2 are shown in Fig. 4. Figs. 4(c, d, and f) show the changes in the amounts of qu, toughness, *ε*f, and E50 of the stabilized samples with 4, 8, and 12% cement according to different ratios of the ER/W, respectively.

**Investigation of the strength parameters after 7 days of curing**

**Cement and epoxy resin with the ER/W ratio ****equal to**** 0.25**

According to the stress-strain curves in Fig. 4(a), unlike the 7-day cement-stabilized samples without epoxy resin, the stress of cement and epoxy resin-stabilized samples with the ER/W ratio equal to 0.25 did not experience the sudden sharp drop after the failure point. The addition of epoxy resin with the ER/W ratio equal to 0.25 prevented the sudden sharp decline of the stress after the failure point in the stress-strain curves. So, their toughness against failure increased. At the ER/W ratio of 0.25 and for different percentages of cement, the amount of failure strain varied from 3 to 5%, while for 7-day cement-stabilized samples without epoxy resin, it was less than 3%. As a result, the failure of cement and epoxy resin-stabilized Bentonite samples with the ER/W ratio of 0.25 was much ductile compared to the cement-stabilized samples without epoxy resin.

For the stabilized samples with 4, 8, and 12% cement at the ER/W ratio equal to 0.25, the increases of qu, E50, and toughness were in the amounts of 68% to 4 times, 2.3 to 4.1 times, and 32 to 74%, respectively, compared to 7-day cement-stabilized samples without epoxy resin. The presence of epoxy resin at the ER/W ratio equal to 0.25 in cement and epoxy resin-stabilized Bentonite samples increased the strength and toughness values of them to several times that of the 7-day cement-stabilized samples without epoxy resin. By adding 8% cement, the qu, toughness, and E50 values increased by 66%, 30%, and 2 times compared to the sample with 4% cement, while by adding 4% more cement (i.e., stabilization with 12% cement), these corresponding values increased by 25, 4 and 42%, respectively. Therefore, the optimum amount of cement by adding epoxy resin with the ER/W ratio equal to 0.25 was 8%.

With the addition of cement, the *ε*f values decreased so that by adding 12% cement, it decreased by 30% compared to adding 4% cement. So, at the ER/W ratio of 0.25, the failure of the stabilized samples by adding cement was brittle compared to each other. The toughness values against failure by adding 8 and 12% cement were in the range of the stabilized sample with 4% cement. It was due to the high value of the *ε*f of the sample with 4% cement, and it was not evaluated as suitable in terms of strength.

**Cement and epoxy resin with the ER/W ratio equal to 0.5**

According to the stress-strain curves shown in Fig. 4(a), the stress after the failure point for the cement-stabilized samples with the ER/W ratio of 0.5 experienced a much smaller sudden drop compared to the stabilized samples with similar cement percentages and with the ER/W ratio of 0.25. The smaller sudden drop indicated that the failure of cement-stabilized samples with the ER/W ratio of 0.5 was more ductile than the stabilized samples with the ER/W ratio of 0.25. Also, by adding epoxy resin from the ER/W ratio of 0.25 to 0.5 and for different percentages of cement, the amounts of qu, E50, and toughness increased in the range of 85% to 2 times, 36% to 3.4 times, and 80% to 4.6 times, respectively.

With the addition of cement from 4 to 8% at the ER/W ratio of 0.5, the amount of qu increased by less than 27%, and the amounts of toughness and E50 not only did not increase but even decreased by less than 20%. With the addition of 8% cement, the strength parameters values had slight changes, but with 12% cement, they increased significantly. The toughness and E50 values of the sample with 12% cement became 2.6 and 2.1 times that of the sample with 8% cement, respectively. Therefore, at the ER/W ratio equal to 0.5, the optimum content of cement was 12%. At the ER/W ratio of 0.5, unlike the ER/W ratio of 0.25, the failure strain value increased by increasing the cement percentage, and it reached a maximum value of 4%.

**Cement and epoxy resin with the ER/W ratio equal to 1**

According to the stress-strain curves shown in Fig. 4(b), the stress after the failure point did not decrease abruptly for the cement and epoxy resin-stabilized samples with the ER/W ratio equal to 1. Its reduction rate considerably decreased so that at the strain interval with the length of about 1% after the failure point, the stress drop rate was almost zero. A plastic region expanded at the strain interval with the length of about 2%, in which the difference between the sample stress and the qu value was negligible. As a result, the strength of the samples against failure was considerable, and they had failures with significant ductility. By increasing the ER/W ratio from 0.5 to 1 and for different percentages of cement, the qu, E50, and toughness values increased in the range of 70% to 2.25 times, 37% to 3.3 times, and 93% to 5 times, respectively. The changes that were more evident by adding epoxy resin from the ER/W ratio of 0.5 to 1 were the significant increase of the ductility and toughness. At the same time, their strength parameters also had high values.

According to the stress-strain curves in Fig. 4(b), at the ER/W ratio of 1, the first part of the stress-strain curve until the failure point for the sample with 12% cement was above that of the sample with 4% cement, and for the sample with 8% cement, it was on top of that of the sample with 12% cement. Therefore, the stress increase rate of the sample with 8% cement until the failure point was significantly higher than samples with 4 and 12% cement. The stress-strain curve of the sample with 8% cement from the strain of zero to about 6% was on top of that of the sample with 4% cement with a significant distance. Also, its stress-strain curve differed slightly from that of the sample with 12% cement at many strain intervals. According to the stress-strain curves, at the ER/W ratio equal to 1, it was expected that the optimum content of cement would be 8%.

Although by adding 8% cement at the ER/W ratio equal to 1, the qu, toughness, and E50 values increased by 68, 44, and 88% compared to the sample with 4% cement, with the addition of 4% more cement (i.e., stabilization with 12% cement), the qu and toughness values increased by 9 and 4.5%, respectively. Even the hardness value decreased by 12.2%. Therefore, by adding 12% cement, the rates of increase in qu, toughness, and hardness values were stopped, and the optimum amount of cement was 8%. With the addition of 8% cement at the ER/W ratio of 1, the *ε*f amount decreased by less than 8%, and by adding 12% cement, the *ε*f value change was less than 4%. Therefore, at the ER/W ratio equal to 1, adding the cement content by more than 4% had a negligible effect on the failure strain value, and so it remained almost constant. For different percentages of cement, it was almost less than 4%.

**Cement and epoxy resin with the ER/W ratio equal to 2**

By adding epoxy resin from the ER/W ratio of 1 to 2 and for different percentages of cement, the amount of qu increased in the range of 40% to 2.85 times, and the stiffness value increased in the range of 22 to 78%. The change that was more evident by increasing the ER/W ratio to the content of 2 was the increase in the failure strain value in the range of 44 to 60%. Consequently, the increase in the toughness value was in the range of 2.3 to 2.5 times for different cement percentages. As a result, the cement and epoxy resin-stabilized Bentonite samples with the ER/W ratio of 2 had high strength. At the same time, they had significant ductility and high toughness against failure.

According to the stress-strain curves of Fig. 4(b), at the ER/W ratio equal to 2, the stress-strain curve of the sample with 8% cement at the strain interval from zero to approximately 4% was on top of that of the sample with 4% cement with a considerable distance. The stress-strain curve of the sample with 12% cement almost, with a slight difference, matched the stress-strain curve of the sample with 8% cement. The stress after the failure point for the sample with 4% cement decreased sharply compared to the samples with 8 and 12% cement. However, the stress in the vicinity of the failure point for the samples with 8 and 12% cement at the strain interval with the length of approximately 3% was in the range of the qu value equal to 12 MPa. As a result, the significant broad of the plastic region was achieved at the strain interval with a length of approximately 3%. Therefore, by adding epoxy resin in the amount of the ER/W ratio equal to 2, the ductility and strength of samples with 8 and 12% cement increased significantly.

By adding 8% cement at the ER/W ratio equal to 2, the qu value decreased slightly by 8% compared to the sample with 4% cement. The toughness and hardness values increased by 30 and 28%, respectively. Although the increase in the strength parameters values of the sample with 8% cement compared to the sample with 4% cement was less than 31%, its stress at the large strain interval was approximately in the amount of qu value, and its drop rate after the failure point was low. It was indicative of the high ductility and strength of the sample with 8% cement against failure. Therefore, at the ER/W ratio equal to 2, the optimum amount of cement was 8%.

**Evaluation of durability against W/D cycles**

**Cement and epoxy resin with the ER/W ratios of 0.25 and 0.5**

Deep cracks were created on the surface of the samples stabilized with 4 and 8% cement and epoxy resin with the ER/W ratio equal to 0.25 due to the high shrinkage of Bentonite soil during drying in the first cycle. As shown in Fig. 5(c), the sample with 8% cement lost the rigidity in the wetting phase of the second cycle. When the water penetrated through deep cracks into the sample, it gradually disintegrated. After a few hours, it disintegrated completely. Despite the significant increase of the strength parameters values of the samples with 4 and 8% cement by adding epoxy resin with the ER/W ratio equal to 0.25, they did not last longer than one cycle. They collapsed in the wetting phase of the second cycle.

At the ER/W ratio equal to 0.25, the optimum amount of cement was 8%. Adding 4% more cement (i.e., stabilization with 12% cement) did not increase the after-curing strength parameters of the sample but improved the sample durability. It lasted the second cycle of W/D. The shrinkage potential of Bentonite soil was very high, and many deep cracks appeared on its surface after 24 h of drying in the second cycle (See Fig. 5(e)). As shown in Fig. 5(h), the sample collapsed as soon as it was immersed in water for the third cycle. By increasing the epoxy resin from the ER/W ratio of 0.25 to 0.5, the sample with 4% cement still did not last. As shown in Fig. 5(d), it disintegrated in the wetting phase of the second cycle. Although the after-curing strength parameters of the sample with 8% cement were slightly different from the sample with 4% cement, it lasted the second cycle of W/D. Therefore, increasing the percentage of cement at the ER/W ratio equal to 0.5 improved the durability of the samples.

The optimum amount of cement at the ER/W ratio equal to 0.5 was 12%. This sample lasted the second cycle of W/D. By comparing its image after 24 h of drying in the second cycle in Fig. 5(g) with the sample with 8% cement in Fig. 5(f), it was observed that there were fewer cracks on its surface. Still, like the sample with 8% cement, as shown in Fig.5(h), it disintegrates after a few hours in the wetting phase of the third cycle. The main point to consider in this part is that the qu value of the sample got to approximately 7 MPa. Despite the considerable qu value, it did not overcome the swelling-shrinkage potential, and neither did it retain the stiffness during the wetting phase in the third cycle and collapsed.

According to Fig. 5, the development of deep cracks on the surface of the cement and epoxy resin-stabilized samples with the ER/W ratios of 0.25 and 0.5 was due to the severe shrinkage of Bentonite soil during drying in the early cycles. As soon as the water penetrated these samples through the cracks in the wetting phase of the primary cycles, they lost their cohesiveness and collapsed. Increasing the ER/W ratio to the values of 1 and 2 for different percentages of cement was with this purpose that despite the severe shrinkage of Bentonite soil, the stabilized sample retained the rigidity and did not collapse. The cementing and bonding materials produced by the stabilization held the Bentonite soil particles and stuck them together. The better the stabilization, the more substantial cementitious and adhesive materials were made. So, fewer cracks on the sample surface were created during the drying phase of each cycle, and less water penetrated the sample during the wetting cycle. When water penetrated the stabilized sample, the moisture came in contact with the Bentonite clay, and so it became softer, resulting in high strain and low strength.

The cement and epoxy resin-stabilized samples with the ER/W ratios of 1 and 2 achieved the durability standard of this study. They lasted up to six cycles of W/D. In order to evaluate the effect of W/D cycles on the strength parameters of these stabilized samples, the uniaxial tests were performed on them after 24 h of wetting in the third and sixth cycles. The results of their uniaxial tests, including the stress-strain curves and the changes of the strength parameters values at the end of wetting in the third and sixth cycles, are shown in Fig. 6 and Fig. 7, respectively.

**Estimation of the swelling-shrinkage potential of Bentonite soil**

In order to estimate the swelling-shrinkage potential of Bentonite soil, the durability results of the stabilized Bentonite samples that lasted up to six cycles of W/D were employed. The strength parameters values required to overcome the swelling-shrinkage potential of Bentonite soil were estimated using the following equations.

$$\left(1\right){ \text{C}}_{SP}= \frac{{SP}_{after-curing} }{{SP}_{wetting}}; \left(2\right){\left(\text{S}\text{P}\right)}_{e}={\text{C}}_{SP}\times {SP}_{after-curing}$$

Where \({SP}_{after-curing}\) is the value of the strength parameter of the stabilized Bentonite sample with cement and epoxy resin after 7 days of curing, and \({SP}_{wetting}\) is its corresponding value after 24 h of wetting in the sixth cycle.\({\text{C}}_{SP}\) is the estimated coefficient of the strength parameter calculated to obtain \({\left(\text{S}\text{P}\right)}_{e}\). The strength parameters were (q*u*)*e*, (E*50*)*e*, and (Toughness)*e* estimated using Eq. (2).

After stabilization of Bentonite soil by any method such as the combined use of chemical and adhesive additives, it was predicted that the strength parameters had to attain at least the values of (q*u*)*e*, (E*50*)*e*, and (Toughness)*e* to overcome the swelling-shrinkage potential of Bentonite soil. The Overcoming of the swelling-shrinkage potential of Bentonite soil at the level was the target that no cracks were created on the surface of the Bentonite soil during drying in each cycle. The water did not penetrate the sample through these cracks during wetting, and neither did it become soft and low-strength as the wet clay.

**Cement and epoxy resin with the ER/W ratio equal to 1**

According to Fig. 6(a), the stress-strain curve of the Bentonite sample stabilized with 4% cement and epoxy resin with the ER/W ratio of 1 significantly dropped at the end of wetting in the third cycle. Also, the stress-strain curve of the sample after six cycles of W/D was at a considerable distance below that of after three cycles of W/D. According to Fig. 7, the qu and E50 values of this sample after three cycles decreased by 44.8 and 72.7%, and after six cycles, they decreased by 67.5 and 86%, respectively. Also, after the three cycles, the *ε*f value of this sample increased by 37.4%, and the toughness value decreased by 68.8%. After the six cycles, the *ε*f value increased by 62%, and the toughness value decreased by 84%.

The development of cracks on the sample surface was low and was not visible until the drying phase of the second cycle. From the third cycle of drying, the cracks became highly visible. The images of the Bentonite sample stabilized with 4% cement and the ER/W ratio of 1, from the end of drying in the third cycle to the end of the wetting in the sixth cycle, are shown in Fig. 8. Although the crack development, as shown in Fig. 8(c), appeared to divide the sample into two parts, it was on the shell of the sample surface, and it was not deep, causing the sample to collapse and lose rigidity. After 24 h of submerging the sample in the fifth cycle, more water penetrated the sample through the cracks so that the width of the cracks increased after 24 h of drying of the sample, as shown in Fig. 8(e). It was due to the severe shrinkage of the Bentonite soil and the loss of more water that penetrated the sample in the previous cycle. These cracks were on the surface shell of the sample. With the sequence of cycles, their depth on the sample surface increased but did not cause it to lose rigidity and collapse.

By applying more cycles, the strength and rigidity of the sample with 4% cement decreased due to cracks created in the drying phase of each cycle and water penetration into the sample through these cracks in the wetting phase. After five cycles of W/D, the penetration of water into the sample in the wetting phase of the sixth cycle was so much that it showed more strain than moist Bentonite and about 7%. Due to the significant reduction of the qu and E50 values after six cycles of W/D, the toughness value against failure decreased by 85%. According to the stress-strain curves in Fig. 6(a), the stress of the sample after three and six cycles faced a sharp drop after the failure point compared to the after-curing sample despite the significant increase of the failure strain value. Therefore, the after-curing sample showed significant strength against failure compared to its sample after three and six cycles of W/D. No sudden drop occurred in the stress of the after-curing sample after the failure point, so the failure was far more ductile.

According to Fig. 6(b), the stress-strain curves of the sample with 8% cement and with the ER/W ratio equal to 1 after three and six cycles of W/D dropped significantly. According to Fig. 7, the qu and E50 values of this sample decreased by 44 and 66% after three cycles. These values declined by 67 and 89% after six cycles. Also, the *ε*f value increased by 22 and 84% after three and six cycles, respectively. The toughness value decreased by 56 and 59% after three and six cycles of W/D, respectively, despite the significant increase of the *ε*f values. The considerable decrease in the toughness value was due to the significant decrease of the qu and E50 values. Also, the slight reduction in the sample’s toughness in the amount of 3% after six cycles compared to three cycles was due to the increase of the sample’s* ε*f in the amount of 52%. It did not indicate that after three additional cycles, the strength of the sample against failure did not change. So, it was not evaluated as appropriate.

The drop in the reduction rates of the qu and E50 values of the sample with 8% cement after three and six cycles of W/D had slight changes compared to the sample with 4% cement. By comparing the images of the stabilized samples with 4 and 8% cement in Figs. (8 and 9), the amount and width of cracks of the sample with 8% cement were not less than the sample with 4% cement after similar cycles. Even as shown in Fig. 9, the condition of the cracks on the surface of the sample with 8% cement was worse than the sample with 4% cement. The drop rates in the qu, E50, and toughness values of the sample with 12% cement at the ER/W ratio equal to 1 compared to the sample with 4% cement after three cycles of W/D decreased by 20, 17, and 9%, respectively. After six cycles of W/D, they declined by 12.5, 6.5, and 13%, respectively. With the addition of 8% more cement (i.e., stabilization with 12% cement), the rates of decrease in the strength parameters values after three and six cycles of W/D compared to the sample with 4% cement had slight changes. As shown in Figs. (8 and 10), there were cracks on the surface of the sample with 12% cement during W/D cycles despite adding 8% more cement. The location and type of cracks development on the surface of the sample with 12% cement during the third to the sixth cycles of W/D differed from the sample with 4% cement. Therefore, the optimum cement content at this concentration of epoxy resin was 4% in the durability process against six successive cycles of W/D.

For the initial estimation of the swelling-shrinkage potential of Bentonite soil, the durability results of the Bentonite samples stabilized with cement and epoxy resin with the ER/W ratio equal to 1 were employed in Eqs. (1) and (2). The estimated coefficients, including Cq*u*, CE50, and Ctoughnes, for the sample with 4% cement were obtained 3, 7, and 6, respectively. For the sample with 12% cement, they were obtained 2.44, 5.1, and 1.85, respectively. To overcome the swelling-shrinkage potential of Bentonite soil, the amounts of (qu)e, (E50)e, and (Toughness)e, according to the results of the sample with 4% cement, were estimated at least 15, 832, and 1.6 MPa, respectively. According to the results of the sample with 12% cement, they were estimated at least 22, 1000, and 1.35 MPa, respectively. Therefore, it was predicted that to overcome the swelling-shrinkage potential of Bentonite soil, the additives were needed as stabilizers which increase the qu value to at least approximately the ultimate compressive strength of normal concrete.

At the ER/W ratio of 1 and for different percentages of cement, the *ε*f value at the end of wetting in the sixth cycle reached approximately to the amount of 7%. The penetration of water into the sample and the wetting of the Bentonite soil caused the stabilized sample, such as the soft wet clay, to have high strain and low strength. According to the stress-strain curves of Figs. 6(b and c), the failure of the samples stabilized with 8 and 12% cement at the end of wetting in the third and sixth cycles compared to the after-curing samples was brittle despite their significant failure strain values. It was due to their stress dropping at a higher rate after the failure point compared to the after-curing sample. Also, according to the stress-strain curves of Fig. 6, at the ER/W ratio equal to 1 and for different percentages of cement, the stress after the failure point in the sixth cycle decreased at a rate approximately equal to the third cycle. However, due to the significant increase of the failure strain in the sixth cycle compared to the third cycle, the sample failure at the end of wetting in the sixth cycle was more ductile than the sample at the end of soaking in the third cycle. The creation of more cracks on the sample surface due to three additional cycles and more water penetration into the sample caused more wetting and softening of the soil.

According to Fig. 7(a), at the ER/W ratio equal to 1, the qu values of the sample with 8% cement after curing and at the end of wetting in the third and sixth cycles were approximately 70% more than the corresponding values of the sample with 4% cement. The qu values of the sample with 12% cement at the end of wetting in the third and sixth cycles were 25 and 36% more than the corresponding values of the sample with 8% cement, respectively. The addition of cement more than 4% at the ER/W ratio equal to 1 did not affect the reduction of the drop rate of the qu value after three and six cycles of W/D. Still, the optimum cement amount for the samples tested at the end of wetting in both the third and sixth cycles in terms of the qu parameter was 8%. According to Fig.7(b), the toughness values of the sample with 8% cement at the end of wetting in the third and sixth cycles were 2 and 3.65 times that of the sample with 4% cement. They were 13 and 31.6% more than the sample with 12% cement, respectively. Although the addition of cement from 8 to 12% had a negligible effect on reducing the drop rate of the toughness value after three and six cycles of W/D, for the toughness parameter as same as the qu parameter at the end of wetting in the third and sixth cycles, the optimum amount of cement was 8%.

According to Fig. 7(c), at the ER/W ratio of 1 and for different percentages of cement, the failure strain value was an ascending function of the number of W/D cycles. Due to the penetration of water into the sample caused softening of the sample, the failure strain value increased by increasing the number of W/D cycles. At the end of wetting in the third cycle, the failure strain value of the sample with 8% cement 18.1% was less than the sample with 4% cement, and for the sample with 12% cement, it was 11.3% more than the sample with 8% cement. However, at the end of wetting in the sixth cycle, the addition of cement had little effect on the failure strain values of the samples, and their values were less than 7%. According to Fig. 7(d), the hardness value of the sample with 8% cement at the end of wetting in the third cycle was 2.36 times that of the sample with 4% cement, and it was 12% less than the sample with 12% cement. At the end of wetting in the sixth cycle, the hardness value of the sample with 8% cement was 47% more than the sample with 4% cement, and it was for the sample with 12% cement was 57% more than the sample with 8% cement. The addition of cement more than 4% did not reduce the drop rate of the hardness value of their samples after three and six cycles of W/D. Still, in terms of the hardness parameter at the end of wetting in the third cycle, the optimum amount of cement was 8%. Also, at the end of soaking in the sixth cycle, the optimum amount of cement was 12%.

**Cement and epoxy resin-stabilized samples with the ER/W ratio equal to 2**

According to the stress-strain curves of Fig. 6(a), despite the slight decrease of the qu value after the three cycles of W/D, the stress behavior of the Bentonite sample stabilized with 4% cement and epoxy resin with the ER/W ratio equal to 2 was considered more suitable than the after-curing sample. It was unlike the stabilized sample with 4% cement and the ER/W ratio equal to 1. At the strain ranging from zero to about 3.7%, the stress-strain curve of the sample with 4% cement and with the ER/W ratio equal to 2 after the three cycles of W/D was above that of the after-curing sample. Therefore, the stress increase rate in the first part of the stress-strain curve of its sample at the end of wetting in the third cycle was more than the corresponding value of the after-curing sample. On the other hand, after the three cycles of W/D, in the wide range of the strain near the failure point, its stress was equal to the qu value. The plastic region expanded at the strain interval with a length of about 3%. The stress drop rate after the failure point for its sample after the three cycles of W/D was much lower than the after-curing sample. Also, its stress-strain curve dropped after six cycles of W/D and was completely below that of the after-curing sample. However, the amount of drop was much less than the sample with the ER/W ratio equal to 1.

According to Fig. 7, the qu value of the sample with 4% cement and with the ER/W ratio of 2 decreased by 25% after three cycles, and the hardness value not only did not decrease but even increased by 21%. After six cycles, the qu and E50 values decreased by 45 and 50%, respectively. Also, the *ε*f value of this sample increased in the small amount of 8% after three cycles, and the toughness value increased by 20%. After six cycles of W/D, the *ε*f value decreased by 8%, and the toughness value decreased by 53%. For the sample with 4% cement by doubling the ER/W ratio and increasing it from 1 to 2, the decrease rates of the qu and E50 values in terms of percentage after three and six cycles of W/D were almost halved.

The development of cracks on the surface of the sample with 4% cement and the ER/W ratio equal to 2, from the third cycle of drying to the sixth cycle of wetting, is shown in Fig. 11. By comparing the images of this sample in Fig. 11 with the stabilized sample with 4% cement and the ER/W ratio equal to 1 in Fig. 8, it was observed that the type of crack development on it became different and was not transverse but longitudinal. Also, the amount, depth, and width of cracks development on it were much less than the sample with the ER/W ratio equal to 1 so that after 24 h of the sample immersing and swelling in each cycle, they were not visible. In fact, by doubling the ER/W ratio due to the shallow depth of crack development, the cracks disappeared by falling the thin crust from the sample surface within 24 h of submerging. No trace of them was seen at the end of the wetting cycle. As shown in Figs. 11(b, d, and f), which are the sample images at the end of wetting in the fourth to the sixth cycles, the crack development created on the sample surface during the drying phase due to the shrinkage of the Bentonite soil was not visible after 24 h of wetting. Therefore, by doubling the ER/W ratio, the amount, width, and depth of cracks development, which were the weaknesses of Bentonite samples due to the penetration of water through them into the sample, decreased.

The failure strain value of the stabilized sample with 4% cement and the ER/W ratio equal to 1 after six cycles of W/D increased by 61%, while for the stabilized sample with the ER/W ratio equal to 2, it decreased by 8%. At the ER/W ratio of 2, due to the decrease in the amount, depth, and width of cracks created in the drying phase during the sequence of the W/D cycles, the water penetration into the sample decreased. As a result, the softening of it diminished. So, the sample with the ER/W ratio of 2 not only did not have an increase of the *ε*f value at the end of wetting in the sixth cycle, but it decreased.

According to Fig. 6(b), the stress-strain curve of the Bentonite sample stabilized with 8% cement and the ER/W ratio equal to 2, dropped after three and six cycles of W/D. After three cycles, it was entirely below the stress-strain curve of the after-curing sample, and after six cycles, it was below the stress-strain curve of the sample subjected to three cycles of W/D. The stress after the failure point for this sample after three and six cycles of W/D had a significant sudden drop compared to the after-curing sample. So, the strength against failure decreased significantly after three and six cycles of W/D. As a result, the failure of the sample with 8% cement after three and six cycles of W/D was brittle compared to the after-curing sample. The stress after the failure point of this sample after three and six cycles of W/D decreased at a rate almost equal to each other. Also, their *ε*f values were approximately equal. Therefore, the ductility of the sample stabilized with 8% cement did not change after six cycles of W/D compared to the sample subjected to three cycles of W/D. Based on Fig. 7, the qu and E50 values of the sample with 8% cement after three cycles declined by 11 and 41%, and they decreased by 32 and 53% after six cycles, respectively. The after-curing *ε*f value of this sample was approximately 6%, and its rate of change after three and six cycles of W/D was negligible. The toughness value of the sample with 8% cement after three and six cycles of W/D decreased by 52 and 61%, respectively.

At the ER/W ratio of 2, for the sample with 4% cement after three cycles of W/D, the qu value decreased in a slight amount, and the stiffness and toughness values of the sample not only did not decrease but also improved slightly. The qu value of the sample with 8% cement decreased slightly after three cycles of W/D, but the stiffness and toughness values of the sample suffered relatively significant reductions. The decrease rates of the strength parameters values of the sample with 8% cement after six successive cycles of W/D compared to the sample with 4% cement had slight changes. Therefore, adding 4% more cement (i.e., stabilization with 8% cement) had a negligible effect on reducing the drop rate of the strength parameters values of the stabilized sample with the ER/W ratio of 2 for the durability against W/D cycles. By comparing the images of the samples stabilized with 4 and 8% cement after three and six cycles of W/D in Figs. (11 and 12), despite the difference in the crack development created on these samples, it was observed that the width and depth of cracks on the surface of the sample with 8% cement were not less than the sample with 4% cement after similar cycles. Even as shown in Fig. 12(e), the crack development on the surface of the sample with 8% cement was more critical than the sample with 4% cement shown in Fig. 11(e). It should be noted that the development of cracks on the sample shown in Fig. 12(e) was superficial. The cracks were not so deep that after 24 h of the sample wetting in the sixth cycle and the falling of the thin shell of its surface, no traces of them were seen (see Fig. 12 (f)).

According to Fig. 6(c), the stress-strain curves of the Bentonite sample stabilized with 12% cement and the ER/W ratio equal to 2, dropped to some extent after three and six successive cycles of W/D. Both were entirely below the after-curing stress-strain curve. The first parts of the stress-strain curves of the samples, after three and six cycles of W/D, were at a slight distance from each other. Even at the strain ranging from zero to approximately 4.4%, the stress-strain curve of the sample after six cycles was on top of that of the sample after three cycles with a slight distance. The stress of the sample subjected to three successive cycles of W/D after the failure point encountered a significant sudden drop compared to the after-curing sample. Therefore, its failure was brittle, and it had less strength against failure than the after-curing sample. Also, the stress after the failure point for the sample after three cycles faced a sudden drop compared to the sample after six cycles of W/D. Therefore, its failure relative to the sample subjected to six cycles was also brittle. According to Fig. 7, the qu and E50 values of the sample with 12% cement after three cycles decreased by 4 and 37%. After six cycles, they declined by 10.1 and 33.2%, respectively. The changes of the *ε*f values after three and six cycles of W/D were less than 10%. The toughness value of the sample decreased by 55% after three cycles and decreased by 27.5% after six cycles. Due to the ductile failure of the sample subjected to six cycles of W/D compared to the sample subjected to the three cycles, its toughness value was far more than the sample after three cycles.

By adding 12% cement at the ER/W ratio equal to 2, the reductions in the drop rates of qu and E50 values after three cycles of W/D compared to the sample with 8% cement were in the amounts of 60 and 11.5%. After six cycles of W/D, they were in the contents of 68 and 37%, respectively. The reduction in the drop rate of the toughness value of this sample after three cycles of W/D compared to the sample with 8% cement was negligible, but after six cycles, it was in the amount of 55%. Therefore, by adding 12% cement, the drop rate in the strength parameters values of the sample decreased after three and six cycles of W/D compared to the sample with 8% cement. Adding 12% cement at this epoxy resin concentration was appropriate. The images of the Bentonite sample stabilized with 12% cement and the ER/W ratio equal to 2, from the third cycle of drying to the sixth cycle of wetting, are shown in Fig. 13. By comparing them with the images of the stabilized samples with 4 and 8% cement in Figs. (11 and 12), it was observed that the development of the cracks on the surface of the sample with 12% cement at the drying phase of the third, fourth and fifth cycles, unlike the samples with 4 and 8% cement, significantly decreased so that they were not clearly visible. Therefore, at the ER/W ratio equal to 2, the addition of 12% cement was very effective in reducing the width and depth of cracks. Still, the coherence of the sample was not enough not to create the crack development at the drying phase of each cycle.

For another estimation of the swelling-shrinkage potential of Bentonite soil, the durability results of the Bentonite samples stabilized with cement and epoxy resin at the ER/W ratio equal to 2 were employed in Eqs. (1) and (2). The estimated coefficients, including Cq*u*, CE50, and Ctoughnes, for the sample with 4% cement were obtained 1.82, 2, and 2.12, respectively, and for the sample with 8% cement, they were determined 1.47, 2.1, and 2.57, respectively. For the sample with 12% cement, they were calculated 1.11, 1.5, and 1.38, respectively. To overcome the swelling-shrinkage potential of Bentonite soil, the amounts of (qu)e, (E50)e, and (Toughness)e, according to the results of the sample with 4% cement, were estimated at least 25.4, 422, and 14 MPa respectively. According to the sample results with 8% cement, they were predicted at least 18.8, 571, and 2.2 MPa, respectively, and based on the sample results with 12% cement, were estimated at least 13.9, 427, and 1.24 MPa, respectively. The required amounts of qu, hardness, and toughness values after the stabilization using any method to overcome the swelling-shrinkage potential of Bentonite soil were estimated at least 25.4, 571, and 2.2 MPa, respectively.

Based on Fig. 7(a), at the ER/W ratio equal to 2, the qu values of the samples with 8 and 12% cement were 10 and 15% more than the sample with 4% cement at the end of wetting in the third cycle, respectively. These increases of the qu values were negligible by adding 4 and 8% more cement. The after-curing qu value of the sample with 4% cement was more than the samples with 8 and 12% cement. So, for both after curing and after the third cycle, adding cement more than 4% had little effect on the qu value of the sample. At the end of wetting in the sixth cycle, the qu values of the stabilized samples with 8 and 12% cement were 29 and 47% more than the samples with 4% cement, respectively. After three additional cycles (i.e., at the end of wetting in the sixth cycle), the addition of 12% cement had some effect on improving the qu value. Therefore, for the qu parameter of the samples subjected to six cycles of W/D, the optimum amount of cement was 12%.

According to Fig. 7(b), at the ER/W ratio equal to 2, the toughness value of the sample with 8% cement was 48% less than the sample with 4% cement, and for the sample with 12% cement, it was in the slight amount of 1% less than the sample with 8% cement at the end of wetting in the third cycle. At the end of wetting in the third cycle, the reduction in the toughness value of the sample with 8% cement was due to the decrease in the *ε*f in the amount of 15%. Therefore, it could not be inferred that increasing the cement percentage had reduced the sample strength against failure. At the end of wetting in the sixth cycle, the toughness of the sample with 8% cement was in the slight amount of 7% more than the sample with 4%. Still, for the sample with 12% cement, it was almost four times that of the sample with 4% cement. The *ε*f value of the sample in the sixth cycle had slight changes with increasing the cement percentage. After the three additional cycles (i.e., at the end of wetting in the sixth cycle), adding cement in the amount of 12% significantly affected the toughness value of the sample against failure. As a result, for the toughness parameter of the stabilized sample subjected to six cycles of W/D, the optimum cement amount was 12%.

According to Fig. 7(c), at the end of wetting in the third cycle, the failure strain value of the stabilized sample with 8% cement and with the ER/W ratio of 2 decreased in the amount of 15% compared to the sample with 4% cement. By adding 12% cement, it decreased in the amount of 3% compared to the sample with 8% cement. At the end of wetting in the sixth cycle, the addition of cement more than 4% had almost no effect on the failure strain value, and its changes were negligible. According to Fig. 7(d), at the end of wetting in the third cycle, the hardness values of the samples with 8 and 12% cement were 37 and 29% less than the sample with 4% cement, respectively. At the end of wetting in the sixth cycle, the hardness of the sample with 8% cement in the slight amount of 21% and the hardness of the sample with 12% cement in the significant amount of 78% was more than the sample with 4% cement. Finally, according to Fig. 7, at the ER/W ratio equal to 2, the optimum amount of cement was 12%, which improved the strength parameters of the Bentonite sample for durability up to six cycles of W/D.

**Evaluation of epoxy resin-stabilized Bentonite soil without cement and water**

In this part, the stabilization of the Bentonite soil sample was so that the total amount of the optimum moisture content required for the compaction was replaced with the epoxy resin additive. They were then treated for 7 days under the same conditions as the other stabilized Bentonite samples. The purpose of stabilizing Bentonite soil with only epoxy resin additive without cement and water was to increase the strength parameters values at the level of about the normal concrete. Also, if this occurred, the swelling-shrinkage potential of Bentonite soil was checked by exposing the epoxy resin-stabilized sample to six successive cycles of W/D as in the previous sections. The results of uniaxial tests performed on them after curing and at the end of wetting in the third and sixth cycles are given in Fig. 14. Their stress-strain curves are shown in Fig. 14(a). The changes in the values of their parameters containing qu, toughness, E50, and *ε*f are presented in Figs. 14 (b, c, d, and e), respectively.

**Evaluation of the strength parameters after 7 days of curing**

According to the stress-strain curve of Fig. 14(a), the stress of the stabilized Bentonite sample with only epoxy resin at the strain interval with the length of approximately 5%, which was a wide strain range, was almost in the amount of the qu. As a result, the plastic region expanded over a wide strain range with a length of approximately 5%, in which the difference between the sample stress and the qu value was negligible. Therefore, it showed a lot of ductility, toughness, and strength against failure. Its qu and the failure strain values were 28 MPa and 6.44%, respectively. After the performance, the dry unit density and strength are two important properties of lightweight structural concrete. The advantages of any materials are the high ratio of strength to dry unit density and the low cost of concrete. Usually, in the mixing plan, the 28-day compressive strength of normal concrete is in the range of 20 to 35 MPa, and the dry unit density of most lightweight structural concrete is between 1600 to 1760 kg/m3. Therefore, the qu value of Bentonite soil stabilized with only epoxy resin and no cement and water was in the range of normal concrete, while the failure strain value was approximately 32 times that of the normal concrete. Also, its dry unit density was about 70% of that of concrete. Therefore, the Bentonite sample stabilized with only epoxy resin, in addition to having high strength and ductility, was much lighter than normal concrete. It was consequently cost-effective from the strength and economic point of view.

At the ER/W ratio of 2, the optimum amount of cement was 12%. By adding epoxy resin in the amount of the optimum moisture, the qu, hardness, and toughness values became two times that of this sample. The amount of change in its failure strain compared to the sample with 12% cement was small and less than 11%. For the sample with the ER/W ratio of 2, the strain interval length in which the plastic region expanded was approximately 3%. In contrast, for the stabilized sample with only epoxy resin, it was about 5%. Therefore, the failure of both samples was ductile, but the ductility and the strength against failure of the sample stabilized with only epoxy resin was much higher.

**Evaluation of durability against W/D cycles**

According to Fig. 14(a), the stress-strain curve of the Bentonite sample stabilized with only epoxy resin at the end of the wetting in the third cycle did not decrease in the wide strain ranging from zero to approximately 3.5%. Even, it was on top of that of the after-curing sample, with a slight distance. The plastic region in which the stress was in the range of the qu value expanded at the strain interval with the length of about 3%. However, for the after-curing sample, it was approximately 5%. Therefore, after three successive cycles of W/D, the weakness that the stress-strain curve indicated that had occurred was the reduction in the toughness of the sample against failure. According to Figs. 14(b to e), the qu value of this sample decreased in the tiny amount of 5% after three cycles of W/D. The failure strain value had slight changes, but the toughness value against failure decreased by 31.5%. Its hardness value not only did not decrease but even increased by 16.3%. It could be concluded that the sample maintained the rigidity and strength after three cycles of W/D.

According to Fig. 14(a), the stress-strain curve of the Bentonite sample stabilized with only epoxy resin at the end of wetting in the sixth cycle dropped significantly over a wide strain range. Along with this, it showed very high ductility. According to the Figs. 14(b to e), at the end of wetting in the sixth cycle, the qu, hardness, and toughness values of this sample decreased by 27.6, 44, and 11%. The failure strain value increased by 55% and gained approximately 10%. The stress drop of the sample was at a low rate after the failure point. The increase of the failure strain value was significant. Although, after six cycles of W/D, the Bentonite sample stabilized with only epoxy resin showed very considerable ductility, the stabilization did not have the expected efficiency due to the significant decrease of the strength parameters values of the sample.

The images of the Bentonite sample stabilized with only epoxy resin from the third cycle of drying to the sixth cycle of wetting are shown in Fig. 14. At the end of drying in the third cycle to the fifth cycle, no crack development was observed on the sample surface. So, to overcome the shrinkage potential of Bentonite soil and prevent the crack development on the sample surface during the drying phase of each cycle, the strength parameters values of the sample should attain at least equal to that of the normal concrete.

The loss rates of qu and hardness values for Bentonite samples stabilized with only epoxy resin after six cycles of W/D were 27.6 and 44%, respectively, and did not decrease compared to the sample stabilized with 12% cement and the ER/W ratio equal to 2. Although no visible crack development was created on the surface of the stabilized Bentonite sample with only epoxy resin, it was not yet wholly impenetrable. Therefore, to make the Bentonite soil impenetrable against the swelling-shrinkage potential, additives were needed that increased the stiffness and strength to values higher than the normal concrete. The estimated coefficients, including Cq*u*, CE50, and Ctoughnes, for the stabilized Bentonite sample with only epoxy resin, were calculated 1.4, 1.8, and 1.36, respectively. In order to become Bentonite soil impenetrable against wetting in each cycle and thoroughly overcome the swelling-shrinkage potential, it was estimated that the amounts of (qu)e, (E50)e, and (Toughness)e of it after stabilization using any method should get to at least 38.6, 1074, and 2.45 MPa.