Abstract
The U.S. Department of Transportation’s (USDOT’s) Federal Railroad Administration (FRA) has sponsored a series of four full-scale side impact tests on specification DOT-113 railroad tank cars. A DOT-113 is a specially designed tank car intended to transport cryogenic liquid commodities. For each side impact test, researchers at the USDOT’s Volpe National Transportation Systems Center (Volpe Center) created a pre-test finite element (FE) model to estimate the overall force-time response of the impactor, puncture/non-puncture outcomes of the impacted tank car, global motions of the tank car, internal pressures within the tank car, and the energy absorbed by the tank car during the impact. While researchers have previously compared FE model results to test measurements for tank car side impact tests, there are currently no formal guidelines on what measurable level of agreement is an acceptable demonstration of FE model validation. This paper presents FE model validation of DOT-113 and DOT-113 surrogate side impact tests using a publicly available software named Correlation and Analysis Plus (CORA) [1] which was originally developed for automotive crashworthiness using models of anthropomorphic test devices, i.e., crash test dummies.
The authors have previously presented FE model validation frameworks for impact simulations [2] and demonstrated FE model validation for non-cryogenic tank car side impacts [3] using CORA and another software called Roadside Verification and Validation Program (RSVVP) [4]. The authors have decided to use CORA in this paper because its validation metrics and rating procedures are included in an ISO technical specification for road vehicles (ISO/TS 18571:2014) [5]. Conversely, RSVVP is not incorporated in a US or international specification.
The results indicate that CORA can be directly applied to tank car side impact model results using the procedures in ISO/TS 18571:2014 when the model does not self-terminate due to puncture. The FE models achieved excellent and good CORA scores for cases without puncture of the tank car. However, early termination of the FE model due to puncture disrupted the automated post-processing of the model results for the two tests that produced a puncture outcome. Further consideration is necessary to develop guidelines that can produce useful validation scores for FE models that include a puncture outcome.
