A new MILP formulation for crude oil scheduling optimization:: A case study in a Turkish refinery

Abstract

Refineries are intricate industrial facilities that undergo the complex process of converting crude oil into fuels and various chemicals. The operations of refineries present unique challenges due to the utilization of advanced technologies and intricate processes. The complexity of refineries stems from the extensive array of crude oil varieties they handle, as well as the production of different products that necessitate specific quality standards. Planning activities are carried out at different levels to coordinate numerous operations that require a significant amount of energy and resources to work efficiently, effectively, and safely in refineries. These activities involve the development of plans that will improve production, distribution, and all other operations in alignment with the set targets. Subsequently, detailed schedules are developed based on these plans to determine the timing, sequence, and allocation of necessary resources for operations. Effective scheduling plays a crucial role in refinery operations as it impacts a multitude of decisions, particularly in the scheduling of production activities. The scheduling process entails determining optimal production conditions and quantities for each product, taking into consideration factors such as the availability of raw materials, capacity of relevant refinery units, and customer demands. Refinery scheduling can be classified into three main categories: crude oil scheduling, production unit scheduling, and final product scheduling. Crude oil operations, being the initial phase of the refinery process, hold significant influence over all subsequent operations and greatly impact the refinery's continuity and profitability. Therefore, it assumes a critical position among refinery operations, underscoring the importance of scheduling specifically tailored for crude oil processes within the scope of refinery planning. This study addresses a real-life problem related to the planning process of crude oil operations based on real refinery conditions. The problem is large-scale and comprehensive as it deals with a real refinery case. The crude oil operation processes in question are managed through three operations: unloading, transfer, and charging. Since it is a coastal refinery, crude oil reaches the refinery via ships, and the transfer from ships to refinery tanks is provided. This process is called crude oil unloading. Sending the crude oil mixture in the tanks to the relevant charging unit for processing is referred to as crude oil charging. Transfer, on the other hand, represents the transfer of crude oil between tanks. When the planner creates a crude oil operation plan, they consider the entire process from the unloading of crude oil from vessels to its charging into units. To handle all operations in an integrated manner, multiple electronic spreadsheets and system data are utilized. Based on this data, the planner identifies suitable tanks for unloading considering factors such as unloading limits and crude oil mixture characteristics. Additionally, to charge the appropriate mixtures in tanks to charging units, the planner determines the required crude oil compositions and the suitable tanks for charging accordingly. Due to the complexity of crude oil operations and the abundance of constraints, generating and evaluating all possible solutions within a reasonable time frame is highly challenging. Furthermore, situations arise during the implementation of a plan that necessitate the updating of the contingency plan based on the evolving circumstances. All these factors have led to the complexity of the planning process, requiring additional resources. The purpose of this study is to develop a decision support system that accurately reflects the entire structure, limits, and operations of the refinery and optimizes the scheduling process for crude oil. For this purpose, a new mixed integer linear programming model (MILP) is proposed, which schedules the crude oil operations by addressing the unloading of crude oil, transfers between tanks, and charging into units. The model is formulated with a discrete time representation, considering all the limits specific to the mentioned refinery problem, and the planning period is approached in daily increments. Unlike refinery structures where storage and charging of crude oil are conducted in separate tanks, as emphasized in most studies in the literature, in this case, the tanks are used for both storage and charging.