A Bayesian accelerated degradation studies on nitrile rubber O-ring

To research nitrile rubber O-rings´ lifetime with less time and resources, and to reach a more convincing and statistic result, an accelerated degradation testing (ADT) is designed, which consists of a 90C pre-test and a step-stress test with 70 °C, 80 °C, 90 °C and 100 °C. Compression sets obtained from ADT are analyzed as degradation data, and degradation models were fitted to data using Bayesian method which easily accommodated uncertainties in the test. Then a relationship model between degradation model and temperature was built with support vector machines (SVM), so the step-stress data can be converted to data at 25 °C and the O-rings´ 25 °C degradation models were obtained. Finally, with these models, nitrile rubber O-rings´ lifetimes were studied by Bayesian method, and the lifetime´s distribution was obtained, which predicted that the nitrile rubber O-ring will reach 60% compression set by 6.2 years at 25°C based on 97.5% confidence bands.