Abstract: |
Subgrade reaction methods, as implemented through the p-y curve analysis, remain the most globally utilized analytical tool to characterize the lateral response of deep foundations. As recognized by many researchers, traditional procedures are known pose challenges in obtaining suitable data fits, or mathematical difficulties associated with data differentiation. The results obtained from a model-scale lateral loaded test on a concrete pile in sand are presented and analyzed through an optimization technique. A genetic algorithm framework is developed to facilitate data interpretation in presence of disturbed data readings and pile nonlinearity. This approach overcomes existing challenges affecting the experimental derivation of p-y curves (e.g. choice of fitting technique and input parameters) by using an ensemble of statistical methods and minimizing an objective fitness function. Experimentally derived p-y curves are then compared with traditional analytical models found in literature. © 2019 Associazione Geotecnica Italiana, Rome, Italy. |