Statistical assessment of load model accuracy for steel grid-reinforced soil walls

Abstract

The focus of this paper is on quantitative evaluation of four different methods that use closed-form equations to calculate the nominal load in steel grid-reinforced soil walls under operational (end of construction) conditions. The four methods are the Coherent Gravity Method used in the UK, the AASHTO Simplified Method (USA), the PWRC Method used in Japan and the Simplified Stiffness Method. The accuracy of the methods is quantified based on analysis of bias statistics where bias is the ratio of measured to predicted (nominal) load. A large database of 113 measured reinforcement loads collected from 11 instrumented field walls is used in the study. For walls constructed with frictional soils, the Coherent Gravity Method and PWRC Method were the least accurate. The AASHTO Simplified Method demonstrated better accuracy and the Simplified Stiffness Method was the most accurate of all methods examined. The Coherent Gravity Method and the updated Simplified Stiffness Method for steel grid walls in the current study have the advantage that they can be used with soils that have a dependable soil cohesive strength component. However, the accuracy of the Simplified Stiffness Method was much better for all soil types based on bias analyses.