Estimating Load Distributions for Retaining Structures Subjected to Railroad Live Loads

Abstract

Retaining structures in proximity to railroads experience a variety of lateral loading intensities depending on the live load surcharge. Current design guidelines recommend using the Boussinesq solution for computing lateral loads due to stress influence of the surcharge load. Although this method provides a conservative approach for design of retaining structures, it employs various assumptions that are no longer valid in the cases of non-uniform soil conditions and structures with flexible responses to loading. As a result, a model that captures a closer estimate of soil-structure interaction behavior is desired.

An analytical model derived from beam theory is presented in this thesis. This model implements the method of initial parameters to solve for a beam equation for a 3rd-order distributed load. A program was written in Python to solve for the coefficients in the beam equation using a least squares regression. The inputs for this program are strain measurements to be obtained from a test site, and the outputs are the regression coefficients and stresses associated with the input strains. The motivation behind this approach is to analyze future experimental data for a retaining wall constructed at a test site in proximity to a railroad. Analysis of sample strain values produced a regression curve that closely matched the expected distributions associated with strain values. It was also found that the order of the regression could be adjusted if needed to reduce error of the resulting curves. Ultimately, the model produced in this research can be used for estimating loads on a full-scale test wall.