Model Calibration and Validation
A model is not useful until we understand just how well it matches reality. Model calibration and validation are two vital steps in developing a useful model.
The first step is calibration – this is simply about getting the model to reproduce the behaviour of the real system for a selection of test cases. The match between model and reality will never be perfect, and it does not need to be for the model to be useful – the allowable discrepancy depends on your objectives.
The second and more complex step is validation. It is possible to have a model that calibrates well for a particular test case, and then produces unphysical results for other cases. The causes for this can be errors in the founding equations, errors in implementation, or omission of important physics. Validation is about confirming the ability of the model to give you a reliable picture of the behaviour of the process or system under scenarios for which you have no test data, and usually involves rigorous testing of the model in parts and as a whole.
Model calibration and validation requires knowledge of the physics involved and a clear understanding of your objectives. If you are having difficulty in believing your model we can assist you with the process of model calibration and validation.