Planejamento da Produção de uma Refinaria Integrado com a Programação da Distribuição de Derivados

Abstract

Oil industry transforms petroleum into different types of marketable products, such as gasoline, diesel, and jet fuel. The process is composed by several processing units, each of them performing separation, conversion, treatment or blending operations, in order to obtain high quality end products. A large number of streams comprised of different properties arrive and leave the units. These streams should be blended to achieve the end products specifications. Afterwards, the end products are pumped to distribution centers, to supply local markets. Accordingly, this work develops a formulation to carry out the planning and scheduling of refinery operation. Initially, an MILNP problem is considered for refinery planning and blending operations. The decisions include crude oil selection, detailed processing units load, unload, and operational conditions and planning of blending operations. Intermediate streams qualities are tracked and so that end product qualities can be accurately calculated when mixture takes place at the blending tanks. A set of five end products are to be produced by the refinery and transferred to the distribution center, namely, gasoline, diesel with three different grades and liquefied petroleum gas (LPG). A second MILP formulation is proposed to distribution scheduling, from the refinery to distribution center, through multiproduct unidirectional pipeline. The problem considers detailed tanks management at the refinery and distribution center, operations time synchronization, product settling period, operating rules, and forbidden sequence. In this approach is also proposed a procedure to reduce the index domain in order to reduce the model size. Later, the independent hierarchical models are solved in sequence to integrate the oil industry supply chain. The objective is to maximize refinery profit, respecting final products specifications and minimize the distribution costs, scheduling the pipeline operation to meet demand values. All models were implemented in the GAMS system and tested against a real-world case study.