The key issue of power systems is to match power demand and supply with the minimum gap and delay. In the steelmaking industry that is well-known for intensive power demand, multiple types of power supply sources are thus prepared. One is to use gas turbine generators consuming byproduct gases, another is to purchase power from external electricity companies, and yet another is to use in-house self-generators to prevent any interruptions. Because we have to prepare the power supply before demands are actually realized, the fuels for generators should be purchased based upon a relatively long-term plan. and redundant power should be consumed by being resold or in other ways. It is economically important to predict power load accurately for the profitability of the steelworks. A load prediction model is therefore mathematically formulated as a linear programming (LP) problem with a view to minimizing Me overall power cost. A case of an actual steelmaking company in Korea is addressed to illustrate the applicability of the proposed model with some remarks.