(385d) Scheduling of Oil-Refinery Operations with Logistic Details

The overall refinery system can be decomposed into three parts: 1) the crude-oil unloading and blending, 2) the production unit operations and 3) the product blending and lifting (Jia and Ierapetritou (2003)). The crude-oil unloading and blending includes the unloading of crude from vessels to the storage tanks, transfer from storage tanks to the charging tanks, and charging schedule to the crude-distillation units. The second part of refinery system consists of the production unit scheduling, which includes both reaction and fractionation processes. Where the reactions sections modify the molecular structure of hydrocarbons and fractionation sections separate the reactor effluent into streams of different properties and values. The third part represents short-term scheduling and blending of finish products, storage and shipping. The blending process is mixing of various intermediate products from refinery, along with some additives, such as antioxidants and corrosion inhibitors, to produce blends with certain qualities (Dewitt et al., (1989)). The main objective for gasoline blending problem is to find the best way of mixing different intermediates products in order to minimize blending costs while satisfying product quality and demand requirements.

Blending is a critical step in refinery operation as the gasoline can yield 60-70% of a refinery's profit. However, the blending and scheduling of oil-refinery operations is generally the most complex sub-problem, where its complexity arises from 1) the large number of product demands with specific quality specifications, 2) limited number of resource available for production and 3) logistic details such as multi-purpose intermediate and products tanks and parallel units. The variables associated with logistics details are the combinatorial or 0/1 variables and have exactly one-to-one correspondence with quantity sizing variables. This combinatorial characteristic of the logistic problem makes the optimization problem NP hard.

Glismann and Gruhn (2001) proposed a two-level optimization approach where a mixed-integer linear model (MILP) is used to solve the short-term blending scheduling problem and a non-linear model is utilized to solve recipe optimization problem. The result of recipe optimization problem is returned to scheduling problem with an aim of generating the optimal solution of blending problem. The logistic complexity of swing-blenders was addressed in their proposed decomposition technique. Jia and Ierapetritou (2003) developed a MILP formulation based on continuous time representation of time domain to solve gasoline blending problem which takes into account the swing-product tanks and swing-blenders. They assumed fixed product recipes for blending which means that blending decisions were not incorporated in their model. The simultaneous optimization of the of-line blending and the short-term scheduling problem was addressed by Mendez et al. (2006). They proposed a successive LP or MILP iterative procedure to deal with non-linear product properties and variable recipes. Their model can used either in discrete or continuous time domain representation.

Solving logistic, quantity and quality aspects simultaneously for large-scale problems is not possible in reasonable time with currently commercially available optimization software or techniques (Kelly and Mann, 2003a,b). Kelly and Mann (2003a,b) proposed a hierarchical decomposition of the problem into logistics and quality sub-problems. The logistic sub-problem considers only the quantity and logic variables and constraints of the problem and quality sub-problem considers product specifications and quantity constraints and bounds. The logistics sub-problem is solved first to optimality and then quality sub-problem is solved by fixing logic variables. The application of this proposed approach is presented by a small blendshop example, where logistics scheduling optimization solution without penalty was generated in 30 seconds or less (Kelly, (2006)).

A large-scale realistic refinery scheduling problem including logistics details is addressed in our work. Different decomposition methodologies are considered and compared for the solution of this case study. These include a hierarchical decomposition method based on the ideas proposed by Kelly and Mann (2003a,b); spatial decomposition exploiting the special structure of the production network (Shah et al., (submitted, 2008)); mathematical decomposition involving specialized benders decomposition (Saharidis et al., (In press )), and Largangian decomposition to simultaneously address quality, quantity and logistic optimization. The case study studied in this work has parallel units, multipurpose components and products tanks. Detailed comparison of the different decomposition methods is presented based on the quality of the solution and the computational solution time.

References

Dewitt, C. W., Lasdon, L. S., Waren, A. D., Brenner, D. A. and Melhem, S. A. (1989). "Omega: An improved gasoline blending system for Texaco." Interfaces 19(85).

Glismann, K. and Gruhn, G. (2001). "Short-term scheduling and recipe optimization of blending processes." Computers & Chemical Engineering 25(4-6): 627-634.

Jia, Z. and Ierapetritou, M. (2003). "Mixed-Integer Linear Programming Model for Gasoline Blending and Distribution Scheduling." Industrial & Engineering Chemistry Research 42(4): 825-835.

Kelly, J. D. (2006). "Logistics: the missing link in blend scheduling optimization." Hydrocarbon Processing(June): 45-51.

Kelly, J. D. and Mann, J. L. (2003a). "Crude-oil blend scheduling optimization: An application with multi-million dollar benefits: Part I." Hydrocarbon Processing(June): 47-53.

Kelly, J. D. and Mann, J. L. (2003b). "Crude-oil blend scheduling optimization: An application with multi-million dollar benefits: Part II." Hydrocarbon Processing(July): 72-79.

Mendez, C. A., Grossmann, I. E., Harjunkoski, I. and Kabore, P. (2006). A simultaneous optimization approach for off-line blending and scheduling of oil-refinery operations. 30: 614-34.

Saharidis, G., Minoux, M. and Ierapetritou, M. (In press ). "Accelerating Benders decomposition using covering cut bundle generation." International Transactions in Operational Research

Shah, N., Saharidis, G., Jia, Z. and Ierapetritou, M. G. (submitted, 2008). "Centralized - decentralized optimization for refinery scheduling." Computers & Chemical Engineering Ms. Ref. No.: 5255.