Hybrid modelling and parameter identification for the decentralized heat supply network of an industrial facility

Abstract

Within this work, a model of heat supply network of an industrial facility was developed, that consisted of ow rate, supply temperature and return temperature sub-models. Available for the modelling data was taken from measurements at the ve points of the network and at heat generators. The challenge for modelling was the unavailability of the information about the heat consumers and often change of ow direction at one of the points, producing di erent operational modes of the network. This work is the second attempt for the modelling of this network, and in this thesis the problem was approached by black-box and grey-box switched system modelling to approximate the real dynamics of ow rate, supply and return temperature at the ve points of the network. The chosen approach made the automatic switch between di erent operational modes possible, and the simulation results revealed the points, where the chosen approach yield good results and where more study of network behaviour and possibly di erent approach is needed in order to approximate it. Also for some of the points, the chosen complex approach did not show much increase in simulation accuracy, meaning that the less complex approach, developed in the rst attempt for modelling of this network, could be more suitable for further implementation of model predictive controller on the network.