(372g) Learning Based Real-Time Batch Tracking of Multi-Product Pipeline Systems

Multi-product pipelines play a crucial role in transporting refined oil from refineries to distribution depots, which use a batch transportation process, allowing the transportation of various petroleum products in sequential batches and performing delivery operations along the pipeline [1-2]. Accurately tracking the location of each batch is essential for effective monitoring and control, ensuring safe mixed oil cutting operations, and managing the pipeline effectively [3]. Traditionally, the batch interface location is determined using a volume calculation model, which relies on upstream volume injection and downstream volume outflow. However, due to factors such as temperature variations, elevation changes along the pipeline, and constantly shifting operating conditions, the volume of oil products can vary, leading to changes in the flow dynamics within the pipeline [4]. As a result, there exist deviations between the calculated location and the true location of the batch interface.

Motivated by these considerations, this work develops a more adaptable and accurate approach for tracking the real-time batch interface in multi-product pipelines. To achieve this, an infinite-dimensional transient hydraulic model, derived from the conservation laws of mass and momentum, is utilized to capture the intricate, spatially and temporally varying flow dynamics influenced by changes in elevation, varying friction, time-dependent product viscosity, and other flow characteristics within a multi-product environment [5-6]. To enhance the estimation accuracy of batch tracking, a data-driven model is developed and implemented to compensate for the tracking errors between the predicted and the actual positions of the batch interfaces. The accuracy and efficiency of the proposed approach will ultimately be validated through a series of case studies and comparisons with industrial case available data.

References:

[1] Rejowski Jr, R. and Pinto, J.M., 2003. Scheduling of a multiproduct pipeline system. Computers & Chemical Engineering, 27(8-9), pp.1229-1246.

[2] Cafaro, D.C. and Cerdá, J., 2010. Operational scheduling of refined products pipeline networks with simultaneous batch injections. Computers & Chemical Engineering, 34(10), pp.1687-1704.

[3] Cafaro, V.G., Cafaro, D.C., Méndez, C.A. and Cerdá, J., 2015. Optimization model for the detailed scheduling of multi-source pipelines. Computers & Industrial Engineering, 88, pp.395-409.

[4] Zheng, J., Du, J., Liang, Y., Wang, B., Li, M., Liao, Q. and Xu, N., 2023. Deeppipe: A hybrid intelligent framework for real-time batch tracking of multi-product pipelines. Chemical Engineering Research and Design, 191, pp.236-248.

[5] Sundar, K. and Zlotnik, A., 2018. State and parameter estimation for natural gas pipeline networks using transient state data. IEEE Transactions on Control Systems Technology, 27(5), pp.2110-2124.

[6] Zhang, L., Xie, J. and Dubljevic, S., 2023. Tracking model predictive control and moving horizon estimation design of distributed parameter pipeline systems. Computers & Chemical Engineering, 178, p.108381.