Application of Reliability Analyses for Serviceability Design of Column-Supported Embankments: Fid-Bau Portal

Application of Reliability Analyses for Serviceability Design of Column-Supported Embankments

McGuire, Michael P. / Sloan, Joel A. / VandenBerge, Daniel R.

Proper design of column-supported embankments (CSEs) with or without a geosynthetic-reinforced load transfer platform involves careful consideration of the interaction among several natural and manmade materials. Variation in material properties, the thickness of soil layers, and the layout of columns can significantly affect service loads in the columns, tension developed in the reinforcement, and post-construction settlements. Probabilistic analyses enable the effects of uncertainty to be incorporated into design for serviceability by estimating the probability of acceptable performance. In cases where explicit functions relating parameter values to performance do not exist or are impractical to implement, Monte Carlo simulation or simplified reliability methods can be used in conjunction with a response surface to estimate the reliability of the design. A response surface is a function developed from the results of parametric analyses that defines the relationship between the design parameters and a performance parameter of interest. This paper describes the development of response surfaces for tension in the reinforcement and total settlement based on the load-displacement compatibility method for a hypothetical CSE constructed over a soil profile comprised of sand overlying soft clay. The design parameters selected for inclusion in the response surfaces frequently involve high degrees of uncertainty, especially, the thickness of the sand and clay, the stiffness of the sand and clay, and the interface friction angle between the sand and the columns. The response functions are applied using Monte Carlo simulation, the Hasofer-Lind method, and the Taylor Series method to estimate the reliability of the design with respect to the performance parameters. The analyses revealed that the thickness of a sand layer overlying the soft clay foundation and compression ratio of the clay have the greatest effect on total surface settlement and geogrid strain out of the five variables considered.