Impact Performance Prediction and Optimization of a Circumferentially Corrugated Tube with Variable Wall Thickness Using Support Vector Machine

Abstract

Based on the hypothesis that multi-corner and multi-cell structures can effectively improve energy absorption behavior, this paper designed a multi-cell circumferentially corrugated tube (MCCT) for energy absorption. The MCCT was designed as a variable thickness form to study the influence of the materials distribution on the cross section on energy absorption. The energy absorption performance of the MCCT was investigated under impact condition with finite element simulation validated by a drop hammer test. Support vector machine, a machine learning technique, was used to predict the energy absorption performance and was further used for optimization of the MCCT. The results show that under the same mass, the MCCT with variable decreasing wall thickness (corners thicker than other regions) shows 4.81%, 30.67% and 37.70% improvement, respectively, in PCF, SEA and CFE, compared to the MCCT with variable increasing wall thickness (corners thinner than other regions). Moreover, the optimization results show that most samples in Pareto front lie in the region of tc > tm. These results all indicate that the MCCT with variable decreasing wall thickness performs better than with increasing wall thickness with regards energy absorption. In conclusion, arranging more materials in the corner element area can effectively improve the energy absorption characteristics of the thin-walled tube. This paper highlights the importance of designing thin-walled tubes as multi-corner and variable thickness configurations for energy absorption.