Determining representative elementary volume size of in-situ expansive soils subjected to drying-wetting cycles through field test

Cracks resulting from cyclic wetting and drying of expansive soils create discontinuities and anisotropy in the soil. The representative elementary volume (REV) defined by the continuous-media theory cannot be applied to cracked expansive soils that are considered discontinuous media. In this study, direct shear tests of three different scales (30 cm2, 900 cm2, 1963 cm2) and crack image analysis were carried out on undisturbed soil samples subjected to drying-wetting cycles in-situ. The REV size of expansive soil was investigated using the crack intensity factor (CIF) and soil cohesion. The results show that soil cohesion decreased with increasing sample area, and the development of secondary cracks further exacerbated the size effect of sample on cohesion of the soil. As shrinkage cracks developed, the REV size of the soil gradually increased and plateaued after 3–5 cycles. Under the same drying-wetting cycle conditions, the REV size determined using soil cohesion (REV-C) is 1.75 to 2.97 times the REV size determined using CIF (REV-CIF). Under the influence of shrinkage cracks, the average CIF is positively correlated with the REV size determined using different maximum permissible errors, with the coefficient of correlation greater than 0.9. A method for determining the REV-C based on crack image analysis is proposed, and the REV-C of expansive soil in the study area under different exposure times is given.