The comprehensive and detailed study of the tank car DTA leads to the following conclusions.

(1) A global finite element (FE) model for a general-purpose tank car has been developed that contains all the major structural details including the top and bottom openings. Contact elements are used between the major structural pads (front sill and body bolster pads) and the tank shell to realistically represent the load transfer between pad and the tank shell. Super-elements representing the entire stub sill also have been developed and used to improve computational efficiency.

(2) Using the global FE model, major fatigue-inducing loads (VCF and LCF) have been analyzed. Specifically, eight load cases were considered consisting of four loading modes (upward VCF, downward VCF, buff and draft) and two lading conditions (loaded and empty tank). Results from the global model analysis were used to identify potential fatigue critical locations (PFCLs) in the tank and to provide cut-boundary conditions for the submodel, where the submodel joins with the global model.

(3) Based on the PFCLs identified from the global model analysis, a submodel has been developed that includes all the PFCLs in the tank section from the stub sill extension to the head revealed by the global model analysis. In the submodel, the weld filler metal along the fillet welds has also been simulated and significant mesh refinement compared to the global model has been included.

(4) All the load cases considered in the global model analysis have been further analyzed using the submodel to obtain reasonably accurate stress values and distributions for use in fatigue crack growth (FCG) analysis. A total of seventeen PFCLs have been identified through the FEA and summarized in both tabular and graphical forms. In addition to being considered in the fatigue analysis, these PFCLs also serve to identify target areas for tank car inspections.

(5) The original VCF load spectrum has been split into upward and downward spectra to account for the apparent difference in stress magnitude and distributions caused by the upward and downward VCF in the tank shell. To investigate load ordering effects under the condition that the load interaction effects (retardation and acceleration) are ignored, a reordering of the tank car load spectra has been performed. The reordered spectra represent the most severe condition when the load interaction effects are neglected.

(6) Examination of the 15,000 miles OTR report indicates that a certain degree of support has been shown to exist for the concept of displacement-controlled VCF loading, though further study and confirmation are needed.

(7) Using the neutron diffraction technique, welding residual strains were measured for an unconstrained section of tank cylinder girth welds. Both conditions with and without stress relief (post-weld heat treatment) were considered. Using the measured residual strains, a three-dimensional residual stress field was determined along a transverse cross-section of the weld. The results provided input for FCG analysis.

(8) To represent more closely the R-ratio that occurs at a specific PFCL, the available fatigue property data expressed as the Paris equation parameters was further categorized into three R-ratio ranges. A “primary R-ratio” concept was proposed based on relative fatigue damage contribution to implement the categorization of R-ratio ranges. A simplified method was proposed to consider the weld toe stress concentration effect for cracks larger than the weld toe radius.

(9) Fatigue analysis has been performed for the PFCLs identified. The results show that the most critical location is N1, caused by the downward VCF load at the toe of the corner weld between the front sill pad and tank head. The fatigue critical location (FCL) N1 has been analyzed for two materials: A516-70 and A515. Two conditions were considered for each material: with and without the welding residual stresses. Under the assumed conditions, the tensile welding residual stress (no post-weld heat treatment) increases the FCG rate by 46%. A range of initial flaw size and shape was analyzed that should cover a reasonable range of actual situations.

(10) For the A516-70 steel, the FCG analysis results showed that without welding residual stresses, failure occurred only for a 3 inch long, shallow initial crack. This resulted in an inspection interval of 10.7 years. With welding residual stresses, failures occurred for (a) a 3 inch long shallow initial crack, which required an inspection interval of 7.5 years, and (b) 1.5 inch long initial crack with an aspect ratio of a/c=0.5, which required an inspection interval of 16.4 years.

(11) For the A515 material without welding residual stresses, inspection intervals vary from 0.7 to 15.7 years, depending upon the initial flaw size and shape. With welding residual stresses, the inspection intervals were reduced and varied from 0.4 to 10 years, depending upon the initial flaw size and shape.

Full Paper