Optimal Scheduling Of Multiple Air Separation Units Under Frequently Changing Demands

Cryogenic air separation is an important process to provide purified oxygen, nitrogen and agron for the production of iron and steel. As the demand of gaseous products is periodical and intermittent in an iron and steel plant, inappropriate operation may lead to the frequent occurrence of undersupply or release, resulting in significant energy wastage. To deal with the frequently changing demand, automatic load change (ALC) technique is developed, thereby offering more freedom but also challenging scheduling activities. Manual scheduling, which is based on the personal experience of the system operators, presents strong subjectivity and poor reliability. In this project, the optimal scheduling for multiple sets of air separation units is studied. Two types of units equipped with and without the ALC technique are included. Both of the deterministic and uncertain demand senarios are studied. The main research contents of this thesis are as follows:1. A combined ALC and scheduling strategy is proposed based on the characteristics of the ASUs to optimize the total profit margin within a certain time horizon. The production of units is represented with a set of operating modes, and each mode is described with a convex hull according to history data. Transition behaviors, especially during the load-change, are modeled. Optimal scheduling is modeled as a mixed integer linear programming (MILP).2. The scheduling model is applied to a real-world task, during which scheduling of different units is studied and scheduling rules of ALC units are acquired. Comparison with manual scheduling demonstrates that economic performance can be greatly improved by applying the optimal model. Impact of different modeling strategies is also analyzed, which shows that the proposed model can ensure the steady production of ASUs with load-change capability.3. A two-stage stochastic model is built to deal with the uncertain demand of gaseous products. Modes of operation and production levels of different units, sales and inventory levels are determined in the fist stage. Additional supply and release are determined in the second stage. Case studies show that the degree of uncertainty has great effect on the scheduling results. The solutions to the deterministic and the stochastic senarios are also compared and discussed.