Abstract : In polymer processing, control of thermal regulation appears as one of the most relevant ways to increase the productivity of processes and to enhance the quality of products. In the framework of the development of numerical tools dedicated to mold cooling optimization, the purpose of this work is twofold. Firstly, in order to describe the thermal behavior of semicrystalline polymers during the cooling stage and the influence of this stage on the material‘s final microstructure and properties, a numerical model of crystallization under nonisothermal flows is developed. The 1st invariant of the extra-stress tensor, resulting from the viscoelastic behavior of the polymer, is considered as the driving force of flow induced nucleation added to the thermally induced nucleation. The growth of the nuclei is described by two sets of Schneider equations. The model is then applied to the nonisothermal crystallization of an isotactic polypropylene in a shear flow configuration (Couette flow). Thermal and rheological effects on crystallization are quantified in terms of kinetics enhancement and final morphological distribution (type, density and average size of crystallites). Secondly, the heat transfer measurement techniques used for mold cooling optimization must be accurate, reliable and compliant with the constraints of the process. A new thermal instrumentation technique is implemented and tested as a heat flux sensor integrated to a mold insert. The aim is to evaluate the local heat transfer between the polymer and the mold wall. The influence of the process parameters on the sensor response is analyzed. The “thermal footprints” of the process are used to validate a simplified heat transfer model of the cooling stage.