Abstract: In the air separation system, the air separation equipment may provide continuous oxygen, nitrogen and argon for steel plants through the pipe network. However, the frequent fluctuation of gas demand that often occurs in steel plants will arise the imbalance of supply and demand, resulting in the resource waste and economic loss. At the present stage, the air separation production scheduling mainly relies on the experience of operators, which cannot efficiently deal with the changes in the market and production process. Therefore, it is necessary to use mathematical models and computers to assist the production scheduling of air separation systems. In this paper, we introduce the resource task network (RTN) as a modeling method to describe the available equipment and the flow direction of material flow unit in the air separation systems. By means of historical data and prior knowledge, we establish a combined integer programming model for the production process. This model, composed of oxygen, nitrogen and argon pipe networks, contains the material conservation in the production process and the capacity constraints of equipment such as liquefier, gasifier and compressor. The goal of the model optimization is to maximize profits, which may be efficiently solved using SCIP. Finally, an example study of steel plant in China is performed to verify the feasibility of the proposed model. Under different demand scenarios, the production process is optimized to make decisions on the start-up and shut-down of equipment and the load dispatch of air separation device group. The research results are of great significance to the production scheduling of air separation systems.