This paper introduces optimal control into the process of oil heating for natural gas fractionation. There are two sources of thermal energy for heating the oil, one of which is a free resource because it uses waste heat from the cogeneration process. With anticipation of the thermal loads related to gas fractionation one day in advance, it is possible to employ supervisory predictive control for minimisation of operative costs. The underlying optimization problem is solved using linear programming through which optimal references for the low-level circuits are obtained. The performed simulation tests show that the desired control effect is achieved, that is, a constant hot oil supply with the required amount and temperature while adhering to all technical limitations and with minimal operating costs.