An MILP Optimization Model for a Petrochemical Complex

E.P. SCHULZ; M.S. DIAZ; J. A. BANDONI

MECANICA COMPUTACIONAL

Asociación Argentina de Mecánica Computacional

Lugar: Santa Fé; Año: 2003 vol. 22 p. 1841 - 1851

This paper develops a multiperiod mixed integer nonlinear programming (MINLP) model for planning the production and operation of a real world petrochemical complex. Solving the MINLP model directly results in inconsistency in solution quality and time. Therefore, a reformulation and linearization technique is first applied to the bilinear terms of the model to obtain an mixed integer linear programming (MILP) formulation which is the initial point to solve the NLP obtained by the reformulation of the MINLP. The model is solved with GAMS 2.25 using OSL and CONOPT2. Two case scenarios are shown to illustrate the scope of this model.