ABSTRACT
Past studies showed high variability of calculated settlements estimated by different settlement formulas. Such variability, mostly attributed to model errors, has been studied comparing predictions with actual settlements from field tests by means of the ratio R, defined as:
where Sc represents the measured settlement and Sm the predicted one corresponding to the same load level. In this paper, starting from previous works on such topic (Cherubini & Greco 1998, Sivakugan & Johnson 2004, Cherubini & Vessia 2005) statistical tools have been used in order to reduce the scatter into R datasets relating to different formulas and to estimate the most efficient settlement formulation. Finally, on the basis of this latter some charts of acceptable settlements have been suggested for design purpose.
