This project focuses on the design of empirical (fundamental) models of industrial manufacturing equipment for optimal control, based on kinetic equations of process dynamics, combined with data-driven modelling based on measured data. The modelled plant is a polypropylene polymerisation reactor for the production of polypropylene granulate. The control variable is the temperature of the reaction mixture in the plant, the operating conditions of which are satisfied by manipulating the coolant flow in the heat exchanger to remove the heat generated by the reaction. The dynamics of the exothermic polymerization reactor is fast and the process is characterized by unstable dynamics, which makes it impossible to identify a model of the process out of the control loop. These limitations lead us to design a model of the polypropylene reactor based only on available industrial data and empirical models based on the mechanical-structural properties of the process. Combining them will result in a multi-model that will deterministically describe the dynamics of the controlled process and adaptively predict the evolution of the response with respect to the measured data when applied in optimal control.