Metamodels are replacing costly validation simulations and experiments in clinch joint design. If materials or conditions change, existing metamodels may no longer be reliable. This paper presents an approach that uses model uncertainty, the Coefficient of Prognosis and the R2 score to decide if a model should be reused or recalibrated, or if fine-tuning is needed. Two case studies show that the framework can provide sufficient recommendations and reused, recalibrated and fine-tuned models can match new models while reducing simulation and training effort.
