The article presents a mathematical model for on-line control of a heat exchanger network (HEN), aiming at maximization of heat recovery. The model assumes that HEN operates at steady state. The paper presents the method of processing measurement data in order to separate periods of time, which corresponds to the steady state of HEN. The existing mathematical models of the heat exchanger give an inaccurate results, due to large calculation errors in the estimation of the heat transfer coefficient (10-40%). Therefore, the authors introduced a heat exchanger model in which measurements were used to improve the correlation describing the heat transfer coefficient. The heat exchanger model with improved correlation was applied for estimation of the thermal fouling resistances. This approach allows calculating the exchanger with a much smaller calculation error (below 1%). The mathematical model and algorithm of HEN control were presented. The proposed on-line control model was tested on the example of HEN belonging to Crude Distillation Unit, processing 780 t/h of crude oil. In the proposed process control system, the manipulated variables are mass flows of crude oil flowing through HEN branches. The proposed method of HEN control gives opportunity to increase the heat recovery by 1.5%. (C) 2019 Elsevier Ltd. All rights reserved.