(708h) Improved Batch-Centric Formulation for Multiproduct Pipeline Scheduling

Compared to conventional transportation modes such as highway and railway, the long-distance transportation of large volumes of refined products by pipeline has lower cost and higher reliability. With the sharp increase of energy demand, the construction of multiproduct pipelines has been growing vigorously all over the world. They should be operated based on optimized schedules, generated according to detailed information of supply, demand and the characteristics of the system. For a given network, the quality of the schedule reflects the level of management and facility utilization, strongly affecting its profit as well as its safety.

Over the past two decades, a large number of mathematical models adopting different representations of time have been proposed to deal with liquid multiproduct pipeline systems. Continuous-time models [1-4] started to attract a great deal of attention due to their better computational performance. They can be divided into product-centric [4-5] and batch-centric models [6-7]. Batch-centric models have the advantage of allowing for more than one batch of a product inside a segment, which may be needed for feasibility. More importantly, multiple batches can enter or leave a segment during a slot, thereby substantially reducing the number of time slots required to represent a schedule [7]. Rigorous constraints for enforcing forbidden product sequences have also been proposed [8] with the help of Generalized Disjunctive Programming [9-10].

This work extends the models by Castro and Mostafaei [8] dealing with branched pipeline systems with a single input node, and by Mostafei and Castro [7] dealing with straight pipeline systems with multiple input and output nodes. The main novelty results from allowing multiple node injections/deliveries during a slot to reduce the required number of slots and improve computational performance. To keep the model linear and accurate, intermediate nodes with an active downstream segment can receive at most one batch during a slot [11]. Conversely, all input nodes with an active upstream segment can inject at most one batch during a slot. We also trace batch migration through line coordinates rather than segment coordinates to reduce problem size, a feature shared with other models in the literature [12-13].

Eight benchmark problems from the literature involving different pipeline configurations are solved to evaluate the performance of the proposed formulation for the objective of makespan minimization. For the five problems from branched systems, we were able to obtain new best solutions in four of them, while rigorously enforcing forbidden product sequences. For the three problems from straight systems, new best solutions are reported in two, while the computational time has been reduced by two orders of magnitude for the third problem.

For straight systems, we also perform a comprehensive comparison between the global and line batch-numbering alternatives. We show that global numbering is an overall better option but the other was able to suggest a non-trivial initial batch sequence that drove the model to further improve one makespan by 6.0%.

Acknowledgments: Financial support from the National Natural Science Foundation of Chine, Grant No. 51874325 and Fundação para a Ciência e Tecnologia (FCT) through projects IF/00781/2013 andUID/MAT/04561/2019.

References:

[1] Rejowski R, Pinto JM. A novel continuous time representation for the scheduling of pipeline systems with pumping yield rate constraints. Computers & Chemical Engineering. 2008; 32(4): 1042-1066.

[2] Cafaro VG, Cafaro DC, Méndez CA, Cerdá J. Detailed Scheduling of Operations in Single-Source Refined Products Pipelines. Industrial & Engineering Chemistry Research. 2011;50(10):6240-6259.

[3] Cafaro DC, Cerdá J. Rigorous formulation for the scheduling of reversible-flow multiproduct pipelines. Computers & Chemical Engineering. 2014; 61: 59-76.

[4] Castro PM, Mostafaei H. Product-centric continuous-time formulation for pipeline scheduling. Computers & Chemical Engineering. 2017; 104: 283-295.

[5] Herrán A, de la Cruz JM, de Andrés B. Global Search Metaheuristics for planning transportation of multiple petroleum products in a multi-pipeline system. Computers & Chemical Engineering. 2012; 37: 248-261.

[6] Cafaro DC, Cerdá J. Rigorous scheduling of mesh-structure refined petroleum pipeline networks. Computers & Chemical Engineering. 2012; 38: 185-203.

[7] Mostafaei H; Castro PM. Continuous-time scheduling formulation for straight pipelines. AIChE J. 2017; 63: 1923−1936.

[8] Castro PM, Mostafaei H. Batch-centric scheduling formulation for treelike pipeline systems with forbidden product sequences. Comput. Chem. Eng. 2018. DOI: 10.1016/j.compchemeng.2018.04.027.

[9] Balas E. Disjunctive Programming. Annals of Discrete Mathematics. 1979;5(2):3-51.

[10] Raman R, Grossmann IE. Modelling and computational techniques for logic based integer programming. Computers & Chemical Engineering. 1994; 18(7): 563-578.

[11] Liao Q, Castro PM, Liang Y, Zhang H. Computationally Efficient MILP Model for Scheduling a Branched Multiproduct Pipeline System. Ind. Eng. Chem. Res. 2019. DOI: 10.1021/acs.iecr.8b06490.

[12] MirHassani SA, Fani Jahromi H. Scheduling multi-product tree-structure pipelines. Computers & Chemical Engineering. 2011; 35(1): 165-176.

[13] Cafaro DC, Cerdá J. A Rigorous Mathematical Formulation for the Scheduling of Tree-Structure Pipeline Networks. Ind. Eng. Chem. Res. 2011: 50 (9): 5064−5085.