The study presents a successful approach used in development of new design of thermal product. The high uncertainty of dynamic thermofluid process analysis and the need of fast exploration of large dataset of design parameters require elaboration of this new approach. It combines detailed thermoCFD simulation models that are validated with physical experiments, with simplified thermostructural simulation models. The need for such an approach is predetermined by the request for fast processing of big data with certain level of accuracy and detail. The approach is demonstrated by a case study of resistance temperature detector design development.