The present study introduces an automated multidisciplinary optimization (MDO) workflow that, for the first time, couples an explicit dynamic bird strike analysis with a post-impact static stress check. This joint problem is solved during preliminary wing sizing by integrating batch Bayesian optimization on Kriging surrogates with a variance-based variable screening procedure. The optimization problem comprises 19 thickness design variables and two highly non-linear constraints, imposing a maximum leading edge penetration and a maximum post-impact front spar stress while minimizing wing mass. The workflow is demonstrated on a five-bay metallic wing segment, yielding a 43% weight saving over the best-performing design during initial data generation while respecting CS 25.631 crashworthiness limits. Results demonstrated substantial computational savings by variable screening and highlighted the necessity of the stress constraint, as designs satisfying only the penetration depth requirement could still experience critical post-impact stress levels.