An Application of Extended Kalman Filtering to Heat Exchanger Models

Abstract

This paper shows how an extended Kalman filter can be applied to the parameter estimation in continuous time heat exchanger models. The model is based on lumping of the heat exchanger. It is on state space form where the temperature in each section is a model state. By letting the model parameters be functions of the massflows and the temperatures one obtains a model that is capable of accurately describing the dynamics of the heat exchanger for all relevant working conditions. Since the parameters are functions of temperature, the model is nonlinear in the states and an extended Kalman filter is applied to the state estimation. Empirical relations of the heat transfer coefficients are incorporated in the model parameters in order to cope with the massflow and temperature dependence. Some of the parameters in the empirical relations are also estimated, thereby adjusting the formulas to the specific heat exchanger.