Modelo matemático para o scheduling integrado de parque de tancagem e polidutos

Abstract

This work presents a Mixed-Integer Linear Programming (MILP) model with continuous time representation to address the tank farm scheduling of finished products in refineries. The model is part of a larger solution process that involves other elements from oil industry supply chain (SC), where strategic, tactical, and operational decisions must be taken. In this context, the standard refinery system can be divided into four subsystems involving the flow of products among refinery internal (tank farm, production units) and external (vessels, pipelines, final costumers, other refineries) entities. The tank farm scheduling is part of one of these subsystems and its solution contributes to the operational activities of the refinery tank farm, enabling the integration that occurs between domestic production, domestic demand, and product transport by pipelines. In this work the scheduling of loading and unloading operations in the tank farm of finished products at each network node (refinery) is determined. Real scenarios are considered, which were obtained from the planning of refineries and external pipeline network scheduling, proposed by Boschetto (2011). The scenarios present volumes and values of stored product inventories, maximum capacity tanks, and start and end times to product movements at the refinery interfaces (production, demand, and pipelines). Thus, the pipeline scheduling determined by Boschetto (2011) is an input parameter to the considered approach. The proposed MILP model searches a scheduling that minimizes the movements within the refinery tank farm in order to respect the imposed operational and structural constraints. Further, for making feasible the scheduling in a smaller computational time, an iterative algorithm is developed and cutting restrictions are inserted in a new model approach, named MILP-IA. The results allow us to analyze the model computational time, the temporal and structural violations, and the number of product movements for each scenario. For the studied cases, we can also check for attending to time and monthly volume constraints to each interface. Finally, the results also indicate that the proposed MILP-IA approach finds solutions in computational times in the order of minutes. The obtained solutions contribute to improve the transfer and storage activities (TS) on two main points: (i) they minimize the number of movements, facilitating the plant operational tasks (searching for routes); and, (ii) they provide feedback to the pipeline scheduling.