Real world pipeline scheduling with inventory management: a combined MILP and sequencing heuristic approach
Systematic methods and tools for managing the complexity
Supply Chain Management & Business Decision Support (T4-3)
Keywords: multiproduct pipeline, tank farm, MILP, heuristic, continuous time and volume
Real world crude oil supply chain is a complex network system, spread over the world and including several nodes that exchange information and goods. The effort to develop computational tools that can be used as decision support systems to manage such structures has become an important field of study. One of the main objectives is to provide the decision makers with tools that help strategic, tactical and operational decision levels. However the complexity of such systems leads to the study of simplified structures and, thus, attention has been focused in smaller problems.
This work explores the transportation and distribution of refined products, which is a challenging problem. However, studies in this field are scarce. Examples are the works by Rejowski and Pinto [Comp. Chem. Eng. 28 (2004) 1511] and Magatão, Arruda and Neves [Comp. Chem. Eng. 28 (2004) 171] as discrete approaches, and Cafaro and Cerdá [Comp. Chem. Eng. 28 (2004) 2053] as a continuous approach. The emphasis of these works is focused on the pipeline operation and few works have studied the interactions between the pipeline and the inbound and outbound of such systems.
In this paper the pipeline scheduling is optimized accounting for the outbound tank farm where capacity and operational restrictions (e.g. settling periods for quality guarantee) are present. The methodology developed comprises a MILP model formulation that is based on a continuous time and volume representations. Some model features include daily clients’ representation, variable settling period by product, pipeline stoppages and variable flowrate. This approach is an improved version of the work by Relvas et al. [Ind. Eng. Chem. Res., 45 (2006) Issue 23, 7841]. A heuristic for the definition of the products sequences is developed that uses the initial inventories and clients’ demands. This heuristics is coupled with the MILP model so as to improve its performance.
The real world scenario of CLC – Companhia Logística de Combustíveis, a Portuguese distribution company is studied where the model developed is applied. Six refined products are scheduled having in mind clients’ demands, raw-materials availability as well as operating restrictions presented both within the pipeline as well at the tank farm.
A study over six months of planning within the real system operation is performed. This is then compared with the results of the real plan data. The benefits of the developed tools are identified.
Presented Thursday 20, 11:20 to 11:40, in session Supply Chain Management & Business Decision Support (T4-3).
