(141d) Robust Optimization for Design and Operation of Chilling Train System in an Olefin Plant

Over the past decades, olefin plants are confronted with volatile feedstock and product markets, which pose a great challenge to their stable manufacturing. Apparently, a fixed design based on economic situations during the design time period is not sufficient to cope with the ever-changing disturbances from feedstock and product markets. Furthermore, a deterministic design problem usually obtains an optimum operating point through minimizing or maximizing an economic objective. Designing control strategies after process design may be difficult because such a sequential approach excludes the simultaneous consideration of operational flexibility and controllability at the initial design stage. Thus, the problem of integrating inherent flexibility and controllability into the plant design stage is increasingly important to ensure its feasible or even optimal operations over a wide range of disturbances.

In this work, robust optimization has been applied to the efficient design and operation of a typical chilling train system, which is the biggest consumer of refrigerating capacity in an olefin plant. At first, an optimization model is built according to fundamentals of mass and energy balances coupled with rigorous simulation models. Binary variables are also employed to determine whether or not a candidate design will be selected. Next, the optimization problem will be solved under different scenarios based on probability distributions for different disturbance variables such as feed composition and operating conditions. Then, the mathematical model is solved to obtain the optimum expectation of the objective function as well as feasible operating windows for key process variables. The proposed methodology is demonstrated by a case study of chilling train system in face of feed stock disturbances, in order to achieve a balance among energy conservation, material saving and operational flexibilities.