Settlement prediction in a soft clay deposit using back-analysis

Field-observed settlement behaviors are frequently different from the behaviors predicted in the design stage due to several uncertainties such as difficulty of acquiring soil properties, inaccuracy of numerical modeling, and error in the measuring procedures. To increase the reliability in predicting the soil properties, which significantly dictate the soil behavior, a technique, called a back-analysis method, has been used and several methods are discussed in this paper. A genetic algorithm (GA) back-analysis method is proposed to overcome the local convergence problem of the existing back-analysis methods. The applicability of the proposed method is analyzed in comparison with other back-analysis methods. The results indicate that the consolidation properties in a multi-layer soft clay ground can be properly predicted using the genetic algorithm (GA) back-analysis at a relatively early stage of surcharge process.