Otimização das operações de transferência e estocagem em rede de dutos

Abstract

This work addresses the problem of developing optimization structurres to aid the operational decision-making of planning and scheduling activities in real-world pipelines. Two problems are studed: (i) The first one contemplates a pipeline connecting a refinery and a depot. The pipeline transports various products. An alternative solution approach is proposed to obtain the pipeline scheduling, using only MILP (Mixed Integer Linear Programming) models. Since the resolution of the entire problem in a unique optimization model became too hard due to the high computational complexity, the problem decomposition is developed. The first proposed model provides the punping product-sequence and its volume, respectively. Fixing these variables, the MILP model is able to obtain the final scheduling of the pipeline in few seconds. The quality of the solution obtained by the MILP sequencing model is better than a previous heuristic sequencing published. (ii) The second system is composed by a network with 30 pipelins that connect different areas including 4 refineries, 2 harbors, 2 final clients, and 6 distribution terminals. They receive or send many oil derivatives. The studied pipeline network has particular features to be processed as, for instance, local constraints, surge tank operation, flow reverse operational, seasonal costs, and work shifts. Also, the pipelines are shared by different products that use different routes. In this way, pumping stoppages should be evaluated. Considering that the pipeline network complexity is higher than the unique pipeline scheduling procedure, the problem decomposition becomes essential. Scheduling details must be given, including pumping sequence in each node, volume of batches, tankage constraints and timing issues, while always respecting a series of operational constraints. Moreover, the computacional burden to determine a a short-term scheduling for the considered scenario is a difficult task. Three MILP models are proposed to address this problem: planning model, allocation and sequencing model and,finally, the timing model. The output of a model is used as input to other structures and/or MILP models. These models are managed in a hierarchical manner to obtain the desired solution. Many insights have been derived from the obtained solutions, which are given in a reduced computational time for oil industrial-size scenarios.