Calibration of Soil-Steel Grid Pullout Models Using a Statistical Approach

Abstract

Steel grids in the form of bar mat and welded wire mesh are a common reinforcing material in mechanically stabilized earth (MSE) walls. This paper uses a large database of laboratory steel grid pullout tests from multiple sources to evaluate the accuracy of different ultimate pullout capacity models in a consistent statistical framework. Included in the study is a recent empirical-based pullout model for steel grid pullout in frictional soils that is now extended to include cohesive-frictional soils. This model is shown to be more accurate than previous models with default parameters based on analysis of bias statistics. The same calibrated model is shown to be accurate when compared with the results of in situ steel grid pullout tests performed in production walls constructed with cohesive-frictional soils. The new default model is shown to be more accurate than previous models, including the current Federal Highway Administration (FHWA) default model that is most common in North America for design against pullout in MSE walls constructed with frictional (granular) soils. The bias statistics reported for each pullout model in this study are also a prerequisite to compute resistance factors in load and resistance factor design (LRFD) for the pullout limit state using reliability theory–based calibration methods.