Despite recent advances in the ability of nonlinear finite element analysis methods to model soil-structure interaction phenomena associated with buried flexible culverts, there are currently few published full-scale field data available for evaluation of the accuracy and reliability with which nonlinear finite element methods model the influence of live vehicle loads on culvert stability. Presented in this paper are the results of finite element analyses of three full-scale field cases involving culvert failure under exceptional live (vehicle) loads. Field evidence in all three cases suggests that the live loadings applied only barely exceeded the structural capacities of the culvert systems so that these three case studies provide a good basis for evaluation of the accuracy and reliability of the finite element analysis methods employed in these studies. A new empirical procedure is employed in these studies to develop equivalent line loads used to provide a representative plane strain modeling of actual discrete vehicle wheel loads. The results of these case studies provide good support for the accuracy and reliability of the analytical procedures used in these studies.

The past 15 years have seen steady improvement in the ability of nonlinear finite element analysis methods (FEM) to model and analyze soil-structure interaction phenomena associated with buried flexible culvert structures. Full-scale case studies and observed long-term successful performance of culvert designs based on such analysis methods suggest that incremental FEM are now able to model backfill loading and compaction-induced stresses with good accuracy for many types of culvert and backfill configurations and conditions. Unfortunately, however, there are few published data currently available on the ability of nonlinear FEM to correctly model the influence of surficial live (vehicle) loads on culvert stability. This stems, in part, from the large expense associated with performance of meaningful full-scale live load tests to failure or near-failure conditions, as well as from the understandable reluctance of engineers and culvert manufacturers to disseminate information on field cases involving live-loadinduced culvert distress or failure, or both. The study of such field cases involving distress or failure would represent an excellent basis for evaluation of the ability of FEM to accurately model live load effects on culvert stability. Presented in this paper are the results or finite element analyses of three field cases involving culvert distress or failure resulting from inadvertent application of exceptional live loads, where exceptional live loads are vehicle loads considerably in excess of allowable design loading. Two of the cases consid-