Efficient methodology for probabilistic analysis of consolidation considering spatial variability

The inherent spatial variability of natural soil deposits should be considered to obtain realistic probabilistically-based estimates of the time for soil consolidation. Although a variety of commercial software packages for solving consolidation analyses within numerical framework exist, they cannot incorporate spatial variability in a robust manner. Therefore, the consideration of the spatial variability is difficult and time consuming. This study proposes two efficient approaches for probabilistic analysis of consolidation considering the spatial variability of coefficient of consolidation (c), as well as a technique for calculating a representative c for combined vertical and radial consolidation. The proposed approaches, which include a first-order reliability (FORM) and stochastic response surface method (SRSM), were applied to examples of 1-D vertical consolidation and combined vertical and radial consolidation problems, and the applicability of the models and spatial variability are identified. As a result, the random field and numerical consolidation analysis were decoupled, and the proposed approaches were able to perform probabilistic consolidation analysis considering spatial variability using existing deterministic finite element codes without any modification. The probability of under-consolidation can be predicted by only 6 and 11 consolidation analysis using the FORM and SRSM approaches, respectively, and the accuracy of these two approaches was shown to be identical to the results of more computationally-expensive, conventional Monte Carlo simulation (MCS) results. It was found that the probabilistic consolidation analyses are sensitive to the spatial variability of c, and the variability of average degree of consolidation increases as the scale of fluctuation increases, which can be effectively identified and treated using either of the proposed approaches.