Probabilistic strength prediction for double lap joints composed of pultruded GFRP profiles – Part II: Strength prediction

Abstract

A probabilistic method for the strength prediction of adhesively bonded double lap joints composed of pultruded GFRP adherends subjected to quasi-static axial tensile loading is presented. The prediction method is based on the experimentally determined material strength of joint materials and its statistical distribution. It takes into consideration the scale sensitivity of the strength, which is modelled using Weibull statistics, and considers not only the magnitude of the stress fields, but also the volume over which stress peaks act. The method therefore presents a mechanical explanation for the increased resistance of local material zones against high strain or stress peaks, which would normally cause failure of bulk material. The predicted joint strengths are slightly underestimated compared to the experimental results due to inaccurate upper tail modeling of the material strength by the Weibull statistical distribution. A comparison of strength results with corresponding results from a maximum stress based failure criterion confirms a statistical size effect with respect to the strength in the order of 10–25%. The probabilistic method provides reasonable results for brittle joint failure, but it cannot be used to predict joint strength of quasi-brittle or pseudo-ductile joint failures, which showed minor statistical scale sensitivity.