Addressing Product Demand Uncertainty: An Insight into an Ethylene Plant

Resumen

The present work addresses the simultaneous planning of production and scheduling of the furnaces’ maintenance shutdowns of an ethane-fed ethylene plant under demand uncertainty. The main processing units (eight parallel cracking furnaces) have decaying performance over time due to coke deposition inside the coils. The present Mixed Integer Nonlinear Programming (MINLP) model determines the furnaces' run length between decoking procedures, as well as production and storage decisions. The objective function is to maximize the profit, which includes sales, production costs, inventory costs, and penalties for not meeting product demands.

The higher the demand uncertainty, the more likely it is to unmeet the demand. Similarly, the risk of over-production rises, resulting in increased inventory requirements and costs. A two-stage stochastic programming strategy is employed with the probabilistic distribution of demands represented by a series of scenarios that are solved simultaneously.