As the third pillar of science, computational simulation has allowed scientists to explore, observe, and test physical regimes previously thought to be unattainable. High-fidelity models are derived from physical principles and calibrated to experimental data. However, missing or unknown physics and measurement, experimental, and numerical errors give rise to uncertainties in the model form and parameter values in even the most trustworthy models. Thus, rigorous calibration and validation of a computational model is paramount to its effective use as a predictive tool. The popularity of the Bayesian paradigm stems from its natural integration of measurement and model uncertainties. An iterative, grouped sensitivity analysis is presented to identify key modelling parameters and reduce parameter space dimensionality and tractability. Parameters are grouped using subject matter expertise to determine their relative importance. Groups are redefined at each iteration until the most important parameters have been identified. The parameters shown to be the most sensitive are then optimized using Bayesian calibration and the resulting predictions are compared to high-fidelity simulation data. This systematic approach to model calibration, progressing from parameter and quantity of interest identification to sensitivity analysis and calibration is applied to a drift-diffusion simulation code called Charon. Charon allows the computational qualification of semiconductor devices subjected to displacement damage. *Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.