Whenever simulation assists an engineer in the course of a particular development process it offers models which serve as substitutes for the actual physical entities under consideration. The quality of these models determines the usefulness of the simulation for the entire development procedure. Usefulness in this context can have several meanings. For some TCAD applications the predictability is a major concern which means that the simulated quantities need to be predicted with a certain accuracy. Whereas it might be sufficient for other applications that the models are able to qualitatively predict the behavior of the involved physical quantities.
Anyway, there need to be mechanisms to assure the quality of the used models. Based on as much measurement data as available, the models need to be verified against those measurements. Since a lot of data will be involved, a sophisticated infrastructure is necessary to perform this task with minimal effort.