Verification of residual stress compliances in GFRP laminates using a thermal load

Abstract

Composite materials, particularly Fibre Reinforced Polymers (FRPs), have become widely used due to their favourable material properties such as an excellent strength-to-weight ratio. The manufacturing process for FRPs induces residual stresses into the manufactured component. These residual stresses are mainly attributed to the difference in Coefficients of Thermal Expantion (CTEs) between the fibres and the matrix, and matrix shrinkage due to cross-linking of the polymers during curing. Incremental Hole Drilling (IHD) is commonly used to measure these residual stresses due to its ease of application, accuracy, and relatively low cost. Calibration coefficients are necessary for the IHD method to relate the strain release measured on the surface of a specimen to the residual stresses present at a given depth, and are typically found using Finite Element (FE) analysis. The series expansion computational method makes use of varying temperature profiles to induce eigenstrains, and hence obtain calibration coefficients for known stress distributions. Tensile and bending loads have previously been used to verify the validity of the residual stresses found by incremental hole drilling. The use of thermal loads to validate the residual stress found through IHD has not yet been assessed, however. This approach has the advantage of requiring less cumbersome equipment than that required to apply bending or tensile loads. The primary objective of this work is to assess how well zeroth-order calibration coefficients (and hence surface strains) predicted by FE correspond to those measured experimentally in the presence of uniform thermal loading. In pursuit of this aim, a Dual-Chamber Controllable Temperature Environment (DCCTE) was built to expose predrilled Glass-fibre Reinforced Polymer (GFRP) laminates to a large change in temperature, which induces measurable surface strains at each hole. The strains are compared with those obtained by FE. The trend in the measured strains largely align with those obtained by FE, however, the magnitudes of the experimental strains at each hole are all offset from those predicted by FE by a similar magnitude. The undrilled strain gauge rosette used for reference measurements was isolated as the source of this error, leading to the finding that selecting alternative rosettes for reference measurements brings the experimentally measured strains closer to those found by FE. Further improvement is possible once individual corrections for hole diameter and offset are implemented. Outliers were identified as having irregularly shaped holes and their results were found to be partially correctable by modelling them more precisely in FE. Overall, a high level of agreement is observed between the measured strains and those found by FE. Some outliers still remain, indicating that some uncertainties may have been underestimated, and others are not included, and their effect is unknown. The level of agreement between strains obtained experimentally and by FE validates the zeroth-order calibration coefficients found using FE in the series expansion computational method.