Probabilistic modelling for timing belt fatigue life predictions using accelerated testing

Abstract

To design low-cost and more robust products, the application of probabilistic methods should be employed. This has been successfully carried out on a timing belt, a key component of the synchronous belt system for driving camshafts, water pumps, fuel injection pumps and balance shafts. The model development process for determining the durability and reliability of a timing belt will be discussed. The timing belt model, based on Miner's rule and the Arrhenius Law, will also be described for the primary failure mode of tooth shear due to cumulative fatigue. This is a closed-form engineering parametric model that considers the belt loads, geometry, mechanical properties, environmental effects and customer usage. The model calibration and correlation against accelerated testing and fleet data, respectively, will be shown. Using an algorithm in BASIC listed in the appendix based on the Monte Carlo simulation method, the predicted timing belt durability and reliability results will be given on an existing vehicle application and an optimised design for the same vehicle, where the belt construction and geometry have been changed. The benefits of using probabilistic methods in the design process are effectively demonstrated and depend on the use of accurate reliability prediction models.