Probabilistic Fatigue and Reliability Simulation

Abstract

This paper demonstrates the advantages of using ‘probabilistic fatigue simulation’ and ‘stochastic’ design over the traditional ‘deterministic’ design approach. It demonstrates how ‘Monte Carlo’ simulation with ‘Latin hypercube sampling’ are effective for obtaining simulated reliability tests based on standard FEA models. Particular attention is paid to recommendations for addressing the effects of fatigue endurance and static failure in the reliability analysis. It also describes how design ‘robustness’ is ensured by using a ‘factorial sampling’ technique in conjunction with a ‘response surface’ model. The requirements for a ‘Reduced Order Model’, obtainable from FEA, are discussed. Recommendations are made to address the effects of fatigue endurance, static failure and non-linearities in the fatigue life curves during Weibull analysis. A case study demonstrates how the methods are applied to an air-cooled intercooler. A comparison is made between measured reliability test results and simulations performed using both deterministic and stochastic approaches. The simulation offers excellent correlation with the experimental measurements. No further improvements in the model were deemed necessary and the model is considered a suitable platform for performing additional design simulations and for extrapolation of the reliability statistics.