Engineering practice has usually dealt with the treated soil bodies using simplistic constitutive models (e.g. elastic perfectly-plastic Mohr–Coulomb). In this paper, a more refined bonded elasto-plastic model is here applied, with emphasis on the ease of calibration. Empirical studies have identified the ratio of cement content to the cured mixture void ratio as a controlling variable for mechanical response. This observation is elaborated upon to show that measuring porosity and unconfined compressive strength is enough to initialize the state variables of a bonded elasto-plastic model. Data from cement-improved Bangkok clay is employed to illustrate and validate the calibration procedure proposed. The structure-scale consequences of the constitutive model choice for the soil–cement are explored through the parametric analysis of an idealized excavation problem. A treated soil–cement slab is characterized by increasing cement contents in the clay–cement mixture. Two sets of parametric analysis are run characterizing the clay–cement either with a linear elastic-perfectly plastic model or with the bonded elasto-plastic model. The same values of unconfined compressions strength (UCS) are specified for the two models to make comparisons meaningful. Results from both series of analysis are compared highlighting the differences in predicted behaviour of the retaining wall and the excavation stability.

Peer Reviewed