Gas cylinders made of composite materials receive growing popularity in light-weight applications. Current standards are mostly based on safety determination relying on minimum amounts of endured load cycles and a minimum burst pressure of a small number of specimens. This paper investigates the possibilities of a probabilistic strength assessment for safety improvements as well as cost and weight savings. The probabilistic assessment is based on destructive testing of small sized samples. The influence of sample size on uncertainty of the assessment is analysed. Furthermore, methods for the assessment of in-service ageing (degradation) are discussed and displayed in performance charts.
Composite materials show a degradation of properties depending on service life. This creates the necessity to find tailored methods to determine strength and residual strength of composite cylinders. The determination can be done e.g. by load cycles tests. The result needs a statistical assessment for the precise description of strength. Especially the statistical assessment of load cycle strength properties has a high uncertainty. It is unclear if a Log-Normal distribution, a WEIBULL distribution or others, describe the scatter behaviour of residual strength properties correctly.
Distribution functions aim at approximating the frequency of occurrence of residual load cycle strength for high survival rates. An assumption has to be found and confirmed to prevent over-estimation of reliability.