Uncertainty quantification of residual strength post lightning strike: A coupled stochastic thermal–electrical–mechanical simulation framework for composite laminates

The strength of composite laminates can be significantly impacted by the damage caused due to lightning strikes. Quantifying such impact of lightning strikes, taking the inevitable compound influence of material and lightning current uncertainty into consideration, is of utmost importance to ensure the operational safety and serviceability in critical composite structural applications such as aircraft and wind turbines. We introduce a machine learning-enabled stochastic framework of hybrid thermal–electrical–mechanical simulations for the uncertainty quantification of residual strength post lightning strike in composite laminates. A comprehensive probabilistic analysis is presented for accurately assessing the uncertainty associated with the residual tensile strength of carbon/epoxy laminates considering stochastic temperature-dependent material properties and lightning current waveform. The results reveal that source uncertainty of the unprotected laminates significantly influences the structural strength with considerable stochastic variability. The machine learning models are exploited further for conducting global sensitivity analysis to examine the relative impact of the influencing parameters on the residual strength after lightning strikes. Seamless coupling of the Gaussian process-driven machine learning model in the finite element based multi-physical lightning strike analysis, integrating multi-stage computationally intensive simulations, leads to an efficient quantification of uncertainty for complete probabilistic characterization of the residual strength and subsequent serviceability analysis.