ABSTRACT
Various numerical techniques used for analysis of flexible pavement use constitutive relationships to compute stress-strain fields in pavement under given traffic loading, material properties and environmental conditions. Even with recent improvements in computational algorithms, more time and resources are required to achieve a viable numerical solution. Under such circumstances, Response Surface Method (RSM) is a convenient method to use. Published literature in other areas of engineering has indicated that accuracy of RSM model is highly influenced by sampling method. This work presents details about various sampling strategies used while constructing response surfaces for predicting strain in flexible pavement. A three layered pavement structure was considered for this study. Numerical values of all significant input variables obtained from different sampling methods were used to predict strain at critical locations. Further, response surfaces of different orders were developed and evaluated using statistical techniques for their accuracy.
