Test SM-S v 9-C4 (Adoption of Sand Cushions)

Figure 6.13 presents the photos of the test results of the test SM-Sv9-C4. In this test, sand cushion thickness of 4 cm (replacing 36% of the marginal backfill with sand) for each reinforcement layer was applied. Tension cracks were found to start developing at 82 mins after the test started and eventually propagated to form the potential sliding surface. Afterward, no significant displacement occurred until a slight movement observed from t = 420 mins to t = 480 mins.

(a)  Start of the test (t = 0 min)

(b)  Tension cracks and potential failure surface appeared (t = 82 mins)

εxy

(c)  Water reached the bottom (t = 100 mins)

(d)  Displacement reached steady state (t = 660 mins)

(e)  End of the test (t = 1080 mins) Figure 6.13 Pictures of the model test SM-S 9-C4

0.02

Figure 6.14 presents the variation of the volumetric water content with time. VWC3, 2, and 1 showed an increase in succession as the water flowed from the top to the bottom.

Notably, VWC1, locating at the bottom of the wall, registered the highest θ; this can be attributed to the high permeability of the sand cushions. The water flowed down from the filter layer and seeped into the reinforced wall through each sand cushion layer.

Figure 6.14 VWC with time elapse

Figure 6.15 shows the wall displacement profile. Before water reached the bottom of the wall (t = 100 mins), tension cracks and potential failure surface had appeared (t = 82 mins). The maximum wall displacement at the end of the test was 8% of the wall height and it occurred at the topmost layer, similar to the test SM-Sv9 (without sand cushions).

Notably, the wall displacement had turned into a more uniform deformation along the wall face compared with the model test without adopting sand cushions. With the adoption of sand cushions, the difference between the wall displacement was 2%, whereas a difference of 8% was observed in such case without sand cushions.

Figure 6.15 Wall displacement profile

Figure 6.16 presents the maximum wall displacement and settlement versus time.

The figure suggests that there were two major displacements, at t = 60~120 mins and at t

= 540~600mins. The first corresponded to the development of the tension cracks, resulted from the loss of matric suction and soil shear strength due to rainfall infiltration. The second issued from the development of the soil shear strain. The soil eventually exceeded its failure strain; therefore, the second wall displacement commenced. Next, the wall displacement reached steady state at t = 660 mins.

Figure 6.16 Maximum wall displacement and settlement versus time

Figure 6.17 shows the mobilized reinforcement tensile strain. The maximum tensile strain was less than 3% and was located at the topmost layer. The displacement-induced maximum tensile strain in each layer decreased from the crest to the bottommost layer, which corresponded to the wall displacement from visual observation. Moreover, a potential sliding surface was shown as the red dashed line by connecting the locus of the maximum tensile strain in each layer. The failure surface by visual inspection and the maximum strain contour from the PIV results were also presented; the three were found to coincide. The maximum strain analyzed by the PIV analysis was around 26%, which was beyond the failure strain for the marginal backfill and meanwhile exceeded the peak strain of the sand.

Figure 6.17 Reinforcement tensile strain and comparison of potential sliding surface

Figure 6.18 displays the maximum reinforcement tensile force profile. At t = 1080 mins when the rainfall stopped, the mobilized reinforcement tensile strain in each layer was larger than the values suggested by the earth pressure method. Moreover, the pattern was changed from a cantilever type to a more uniform type.

Figure 6.18 Maximum reinforcement tensile force profile

In summary, the soil-geogrid interface friction was enhanced in the test SM-Sv9-C4 because of the adoption of sand cushions. The high friction between the sand cushions and the geogrid reinforcement effectively restrained the excess deformation. The failure mode, however, developed into a compound of sudden tension cracks and progressive deformation. Not only was the wall displacement reduced, but the time the displacement reached the maximum was postponed. Tension cracks were found because of the high volumetric water content confined in the marginal backfill. Moreover, the developed soil

6.2.2 Comparison

Table 6.3 summarizes the factor of safety against breakage and pullout for the tests SM-Sv9 and SM-Sv9-C4, and Figure 6.19 presents the pictures at the end of the two tests.

The thickness of the sand cushions was 4 cm, accounting for 36% of the backfill. The factor of safety against pullout was significantly larger in the test SM-Sv9-C4 because of the high interface shearing characteristic between sand-geogrid interface. Visual observation indicates that larger deformation occurred in the test SM-Sv9 while only minor displacement was observed in the test SM-Sv9-C4.

Table 6.3 Factor of safety

Model test FSbreakage FSpullout (dry) FSpullout (wet)

SM-Sv9 2.1 3.6 2.5

SM-Sv9-C4 2.1 4.4 4.3

(a) (b)

Figure 6.19 Pictures of tests: (a) SM-Sv9; (b) SM-Sv9-C4

Figure 6.20 displays the wall displacement profile with and without the adoption of sand cushions. Visual inspection indicated that the wall displacement was smaller in the case with sand cushion applied. The maximum wall displacement in the test SM-Sv9 was observed at the topmost layer, with a value of 15% wall height. After the adoption of sand cushions, the maximum wall displacement was reduced to 8% wall height. For Sv = 9 cm, the wall displacement was reduced by 40% after adopting sand cushions.

Figure 6.20 Wall displacement profile

Figure 6.21 demonstrates the maximum displacement versus time. The maximum displacement was 8 cm in the test SM-Sv9; however, after the adoption of sand cushions, the maximum displacement was reduced to 5 cm in the test SM-Sv9-C4. Aside from the reduction in the maximum wall displacement, the progression of the displacement had become gentler. The curve for the test SM-Sv9 indicated that at t = 120 mins the major wall displacement had commenced and reached a value of 5 cm. Contrarily, the curve for the test SM-Sv9-C4 only had a displacement of 2.5 cm at t = 120 mins, and it reached the maximum value at t = 660 mins.

Figure 6.21 Maximum wall displacement versus time

Figure 6.22 presents the mobilized reinforcement tensile strain in the tests SM-Sv9 and SM-Sv9-C4. The maximum tensile strain mobilized in the test without cushion was around 4%, while it was only 3% if sand cushions were adopted. The mobilized reinforcement tensile strain in the tests SM-Sv9 and SM-Sv9-C4 was significantly different at the topmost layer, as the displacement of the test SM-Sv9 was significantly larger than that of the test SM-Sv9-C4 in the topmost layer. Moreover, the major potential sliding surface is more shallow in the test SM-Sv9. Notably, a second potential sliding surface indicated by the locus of maximum tensile strain was observed in the test SM-Sv9.

Figure 6.22 Reinforcement tensile strain

Figure 6.23 presents the maximum reinforcement tensile force profile. The mobilized reinforcement tensile strain displayed different patterns before and after the adoption of sand cushions. In the tests SM-Sv9 (before the application of sand cushions), the reinforcement tensile force at the top layers was much larger than the earth pressure method. After the adoption of sand cushions, the values in each layer became uniform;

however, the values were all larger than that estimated from the earth pressure method.

Figure 6.23 Maximum reinforcement tensile force profile

Figure 6.24 demonstrates the strain contour estimated from the PIV analysis. For the model test without cushion, the maximum strain concentrated at the top layer owing to the large displacement induced by the accumulation of water infiltration at the top. With sand cushion applied, the maximum strain contour propagated through all the layers. This is similar to the strain propagation of the model test SP-Sv9, in which sandy soil was used as backfill. Furthermore, the adoption of sand cushions reduced the maximum strain by 30%, from a maximum of 36% to 26%.

(a)

(b)

Figure 6.24 Strain contour from PIV analysis: (a) SM-Sv9; (b) SM-Sv9-C4 εxy

In summary, the displacement of the wall can be significantly reduced with the application of sand cushion owing to the high interface friction between the soil-geogrid interface. Table 6.4 summarizes the performance of the test SM-Sv9 and SM-Sv9-C4. The failure modes in both tests were progressive. The main difference between the two tests is the development of the soil shear strain. For the test SM-Sv9, the rainfall infiltrated into the topmost layer, causing the loss of matric suction and soil shear strength. Moreover, the maximum displacement and the maximum soil shear strain developed at the top.

Regarding the test SM-Sv9-C4, the application of sand cushions can provide drainage and prevent water accumulation on the topmost layer. In addition, the development of the soil shear strain resembled the pattern in the test SP-Sv9, in which the potential sliding surface propagated through all the layer.

Table 6.4 Summary of test results

Test SM-Sv9 SM-Sv9-C4

Failure mode Progressive Progressive