Uncertainty in Fibre Strength Characterisation Due to Uncertainty in Measurement and Sampling Randomness

Carbon fibres have exceptional mechanical properties and are used for critical structural applications such as composite pressure vessels and aerospace components. For such high performance applications, reliability-based designs and lifetime assessments require very accurate strength models. Accuracy of the predictions made by composite strength models depend on realistic material properties of constituents, which are used as input. In practice however, fibre strength properties reported by different sources show significant variations. The work described here aims at understanding the influence of measurement uncertainty and sampling randomness on the uncertainty in calculated tensile strength distribution parameters. Tensile strength data for T700 carbon fibres obtained from single fibre testing process has been analysed for uncertainties. A parametric bootstrap method has been used for the evaluation. It has been shown that although both the causes studied of uncertainty are critical, the sampling randomness has a larger influence on the uncertainty of fibre strength, as compared to the uncertainty due to measurement. Choosing an insufficient sample size for analysis can thus result in uncertain or even inaccurate fibre strength properties, which would limit the reliability of composite strength models. The knowledge of the causes and effects of these uncertainties can help in taking appropriate measures for improving the accuracy of results. This would thereby enhance the capability of composite strength models to estimate the behaviour of different composite structures more accurately.