Multi-criteria experimental evaluation of soil–reinforcement interface systems under direct shear loading

This study presents a systematic experimental and multi-criteria evaluation of soil–reinforcement interface systems under direct shear loading. Direct shear tests were conducted on two granular soils, namely a well-graded sand with silt (SW–SM) and a poorly graded sand (SP), using 100 × 100 and 300 × 300 mm shear boxes under dry and optimum moisture content (OMC) conditions. Four interface configurations were examined: unreinforced soil, an HDPE geogrid, a unidirectional carbon-fiber-reinforced polymer (CFRP) sheet, and an epoxy-impregnated CFRP sheet. The geogrid primarily enhanced friction/interlocking response through particle interlocking and dilation, whereas the CFRP-based systems provided greater post-peak retention; epoxy impregnation produced an epoxy-modified interface response consistent with an adhesion-related contribution. Representative shear stress–horizontal displacement and vertical displacement responses confirm distinct peak, softening, and contraction–dilation patterns among the interfaces. Mohr–Coulomb fits based on 50, 100, and 200 kPa normal stresses show very high coefficients of determination, while the limited number of stress levels is explicitly recognized when interpreting apparent cohesion and friction angle. A TOPSIS multi-criteria framework was applied using peak strength, large-displacement shear resistance, initial interface shear-stiffness index, strength retention, peak displacement, and an interface toughness index defined as the area under the shear stress–displacement curve from 0 to 10 mm. The baseline ranking placed CFRP+Epoxy first (Ci* = 0.680) and geogrid second (Ci* = 0.661). A 45-scenario sensitivity analysis showed that these two systems consistently formed the leading pair under the investigated criteria and weighting assumptions, with geogrid ranked first in 24 scenarios and CFRP+Epoxy in 21. Overall, the results demonstrate that reinforcement selection should distinguish between friction/interlocking-dominated peak resistance and epoxy-modified post-peak stability rather than relying on a single strength parameter.