A methodological framework for the validation of predictive simulations

This paper presents a methodological framework for the validation of predictive simulations that synthesizes concepts from the fields of Conventional Model Verification and Validation, Testing of Scientific Theories, Design of Experiments, Error/Sensitivity, and Statistical Exploratory Data Analysis. This framework presents validation as a five-phase process that starts with the often overlooked phase ‘Is the simulation subject to validation?’. The remaining phases are: a priori tests, design and execution of experiment, comparison with predictions, and lastly simulation improvement.