The presented research results are related to the modelling and simulation of technological processes using recurrent neural networks (RNN). This paper presents the application of such modelling technique for a flow-through heater. The structure of the RNN is presented and the correctness of the obtained modelling results is verified. The application of such a model for the classical PI algorithm controller tuning is presented. The results were compared with the popular PI algorithm tuning rule (Chien-Hrones-Reswick). The advantage of the proposed approach to tuning controllers over traditional methods was pointed out.