Systematic solving study for the optimization of the multiperiod blending problem : a multiple mathematical approach solution guide

In this paper, six different approaches for the multiperiod blending problem are tested in terms of global optimality and computational time using a new set of problem instances. The solution methods discussed are the standard MINLP formulation, the relaxation created using McCormick envelopes, a Radix-Based Discretization, a generalized disjunctive programming (GDP) formulation, a Redundant Constraint GDP formulation and a Two- Stage MILP-MINLP Decomposition (still ongoing). The addressed problem is a non-convex MINLP which has been solved for instances with a limited number of variables; hence, determining the best approach and the best solution algorithm is desirable. Results obtained show the best method is the standard MINLP, followed by the Redundant Constraint GDP and the best solution algorithms are the MIQCP algorithms provided by Gurobi. Still, results from the Two-Stage MILP-MINLP Decomposition are still ongoing and have shown promising results so far.

• Autores: Ovalle Varela, Daniel
• Tipo de recurso: Trabajo de grado de pregrado
• Fecha de publicación: 2021
• Institución: Universidad de los Andes
• Repositorio: Séneca: repositorio Uniandes
• Idioma: eng
• OAI Identifier: oai:repositorio.uniandes.edu.co:1992/51673
• Acceso en línea: http://hdl.handle.net/1992/51673
• Palabra clave: Mezcla (Ingeniería química)-Metodología-Investigaciones; Petroquímicos-Investigaciones; Líquidos-Investigaciones; Ingeniería
• Rights: openAccess
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/