The need for high-precision injection molded parts is constantly increasing. According to the pvT-behavior of the polymer, local variations of temperatures and pressures lead to inhomogeneous shrinkage and thus result in warpage. Local manipulation of the specific volume is aimed by controlling the local mold temperature to homogenize shrinkage and reduce warpage. Therefore, a mold with 18 individually controllable tempering zones was developed for dynamic heating and cooling. Due to the large mass of the mold, thermal processes underlie significant dead-times. A model predictive control approach, based on a finite-difference discretization of a one-dimensional heat equation, was developed to predict temperatures within mold and part for accurate control of the tempering zones. Heat transfer coefficients between mold and part remain unknown, which is why the temperature prediction model needed to be validated. Since the temperature distribution over the part thickness is not directly measurable, injection molding simulations (FEM) were used for the validation and calibration of the prediction model. The FEM simulations were calibrated regarding their local heating power through experimental trials using thermographic recordings of both mold sides during heating. A modification of material properties and heat transfer coefficients improved the accuracy of the prediction model.