Analysis of stress rupture data on fiber composites: Part 1- A unified maximum likelihood method

Abstract

Stress rupture is a failure mode of increasing concern in composite overwrapped pressure vessels (COPVs) in particular and continuous unidirectional fiber composites in general. Stress rupture is catastrophic and occurs randomly with little or no advance warning. Prediction of a composite structure's resistance to stress rupture is typically based on extensive test data from samples, measuring both strength and lifetime at multiple load levels. Lifetime tests are often censored due to practical limitations on available test time, thus complicating the statistical prediction process. Furthermore extensive extrapolation within the framework of a robust statistical model is required as the desired lifetimes are typically much longer than available time for testing, the service loads are much lower and the desired reliabilities are much higher than the data alone can support. This paper presents a maximum likelihood based prediction method for stress rupture using the classic power law model in a Weibull framework. Details of the method are fully illustrated using data generated on model carbon/epoxy COPVs tested at the NASA White Sands Test Facility. This method treats the strength and lifetime data all at once in one unified procedure, as compared to current sequential industry standard approaches that typically treat strength data first and then lifetime data separately using strength estimates. Compared to such approaches when applied to typical real world data sets, this unified procedure provides much tighter estimates of failure probability at a given desired lifetime and load level, and with lower amounts of bias.