A comparison of theory and experiment for the stress-based Grant–Sanders method

In general the predictions given in the part A paper [Compos. Sci. Technol. (1998) 1033] agree well with test results, which were supplied by the organisers of the world-wide failure exercise, where the laminate contained fibres in the directions of the applied loadings. The agreement with experiment is more variable for situations where the matrix-dominated properties assume more significance. This is most pronounced in the comparison with experimental stress/strain curves for ±45° and ±55° glass/epoxy laminates, where the test-theory discrepancies were often large, although where these specimens achieved higher stress and sometimes appreciably higher strain levels than those at which final failure might have been expected, this behaviour was predicted with some success. Closer examination of these discrepancies has revealed that they mainly occurred where the ply transverse and shear strains exceeded the values for which stress v. strain data had been provided. For strength prediction the main shortcoming in the model used appears to be in the application of the shear-compression interaction in glass/epoxy laminates, where agreement with experimental results was improved by dropping this criterion, although results indicate a weaker interaction, rather than none at all. Agreement was also improved by dropping a delamination criterion which led to some under-predictions in the two cases involving (90±30)° glass-epoxy laminates.