Vibration Fatigue from Random Excitation for Battery Pack Housing in an Electric Vehicle

Any vehicle traveling on roads interacts with various profiles of surface roughness, which can be best characterized by randomness. The resulting random vibrations not only expose passengers to unpleasant physical shakes and noises, but also impart fatigue damage to nearly everything installed on the vehicle. In today’s robust design process, it is highly desirable to predict fatigue damage in the early design phase, in order to prevent any durability problems in the future, especially for electric vehicles. Historically, the conventional approach to tackling the problem of fatigue damage has involved cycle-counting stress or strain responses, obtained through step-by-step numeric solutions in the time-domain. However, the most effective method of predicting fatigue in random vibration lies in the frequency domain. Such a spectrum-based approach is greatly advantageous because it does not have to deal with expensive and tedious simulations involving millions of time instants of excitation to obtain vibration responses. In this work, the spectral method will be applied to a crucial component in any current electric vehicle—battery pack housing—to demonstrate how efficiently it handles the fatigue prediction in this context. Two algorithms based on different assumptions for the probability density functions of stress in the random response, will be compared to assess their impact on the accuracy of fatigue predictions. Additional case studies will demonstrate that modal damping ratios can dramatically affect the outcome of fatigue damage. Finally, it will also be shown how to account for residual stress due to manufacturing or assembling processes as a non-relaxed mean stress correction to cyclic randomly-varied stresses.