No abstract

Accelerated life testing (ALT) is an effective method of demonstrating and improving product reliability in applications where the products are expected to perform for a long period of time. ALT accelerates a given failure mode by testing at amplified stress level(s) in excess of operational limits. Statistical analysis (parameter estimation) is then performed on the data, based on an acceleration model to make life predictions at use level. The acceleration model thus forms the basis of ALT methodology. Well-established accelerated models such as the Arrhenius model and the Inverse Power Law (IPL) model exist for key stresses such as temperature and voltage, but there are other stresses, like subsea pressure, where there are no clear models of choice. This research proposes a pressure-life (acceleration) model for the first time for life prediction under subsea pressure for key mechanical/physical failure mechanisms.Three independent accelerated tests were conducted, and their results were analyzed to identify the best model for the pressure-life relationship. The testing included material tests in standard coupons to investigate the effect of subsea pressure on key physical, mechanical, and electrical properties. Tests were also conducted at the component level on critical components that function as a pressure barrier. By comparing the likelihood values of multiple reasonable candidate models for the individual tests, the exponential model was identified as a good model for the pressure-life relationship. In addition to consistently providing good fit among the three tests, the exponential model was also validated with over 10 years of field data and demonstrated several characteristics that enable robust life predictions in a variety of scenarios. In addition, the research also used the process of Bayesian analysis to incorporate prior information from field and test data to bolster the results and increase the confidence in the predictions from the proposed model.

No abstract

No abstract

No abstract