Citation

Milho JF, Ambrosio JAC, Pereira MFOS. Int. J. Crashworthiness 2004; 9(5): 483-493.

Affiliation

Instituto de Engenharia Mecanica, Instituto Superior Tecnico; Universidade Lusofona de Humanidades e Tecnologias

Copyright

(Copyright © 2004, Informa - Taylor and Francis Group)

DOI

10.1533/ijcr.2004.0304

PMID

unavailable

Abstract

Advanced train crashworthiness design requires not only numerical simulation tools capable of describing the dynamic response of train sets during general crash scenarios, but also, optimization procedures that can be used efficiently in the earlier design stages. A multibody dynamics based methodology that combines optimization with efficient analysis techniques is proposed in this work, for the design of train crashworthy components. In this methodology, the components of the trains are described as rigid bodies that have their relative motion constrained by kinematic joints and among which there are nonlinear spring-damper type elements that represent the structures of the trains that deform under normal operating conditions or during the train crash. Interaction between the colliding trains components are described by contact detection and contact force models. A planar dynamics formulation is used to access out-of-direction dynamics of the train cars. Through the use of an optimization algorithm, a general design framework is developed for single objective optimization problems, applied to the design of train crashworthy components. The selection of any optimization function is allowed, particularly, the ones related with train crashworthiness such as train car accelerations, deformations of train car structures or energy absorbed during train impact. Design variables related to the characteristics of the train car structures or components are used, such as train car mass or material behavior of train car structures defined by force-displacement curves. This methodology is applied to optimize the characteristics of complete train sets to design full-scale experimental crash tests. The results are compared with those obtained in simplified unidimensional multibody train models, using optimization algorithms that do not use analytical sensitivity information.