The presented workflow focuses on the automated evaluation of robust material parameter identification tasks. A dedicated preprocessor manages all required input files and applies bootstrapping by sampling data blocks with replacement. This approach generates multiple independent input sets, each representing a statistically varied version of the original dataset.

During parameter identification, each bootstrapped dataset is assigned to an individual HTCondor worker, where an independent optimization is performed. All optimizations run concurrently, resulting in a distribution of calibrated parameter combinations rather than a single deterministic solution. This ensemble of bootstrapped parameter sets enhances robustness and provides insight into the uncertainty of the identified parameters.

Following optimization, the results are used to run finite element simulations on structural components. These simulations rely on UMAT implementations generated earlier in the workflow. A GPU-based postprocessor then evaluates the damage evolution for all simulation results in a massively parallel manner, thereby concluding the workflow.