Ethylene plant scheduling system based on a MINLP formulation

Abstract

This paper addresses a multi-period production scheduling problem of an ethylene plant by considering feedstock supplies, plant operating conditions, product inventories and demands simultaneously. The main complexity in solving this problem comes from a mixed integer nonlinear programming (MINLP) formulation of a naphtha feedstock management problem that includes scheduling vessel arrivals, storage tank assignments and blending. In addition, we also consider 4 other feedstocks, nonlinear yield models, separation constraints, product inventories and demands in the model. The number of binary variables in the model for a typical scheduling periods of one month can easily exceed a thousand. It may well be one of the largest and most rigorous model for ethylene plant production scheduling ever reported to date. Here we will discuss issues related to a problem formulation, solution strategy and industrial implementation of the ethylene plant production scheduling system.