Hierarchical Bayesian Approach to Estimating Variability of Liquefaction Resistance of Sandy Soils Considering Individual Differences in Laboratory Tests

Cyclic undrained triaxial tests are commonly used in research and practical design to evaluate the liquefaction resistance of sandy soils. This paper aims propose a methodology by considering variability or uncertainty associated with experimentation, using Bayesian statistics Markov chain Monte Carlo technique. In addition conventional nonhierarchical modeling, hierarchical modeling is adopted properly incorporate factor caused individual differences (e.g., difference between experimenters) into evaluation. Findings show that regression curves cyclic ratio estimated model for all experimenters’ results cooperative test program too generic poorly applicable. contrast, each experimenter’s sometimes deviate from overall trend, dragged characteristics. The demonstrates both trend individuality can be rationally considered posterior distributions input parameters) referring other results, even though number cases limited focal experimenter. Another advantage that, when different experimenter newly performs similar laboratory tests, distribution based on existing dataset as prior estimate parameters specific proposed may also difficult quantify, e.g., testing apparatus specimen size.