A high-accuracy online transient simulation framework of natural gas pipeline network by integrating physics-based and data-driven methods. (1st March 2023)

• Record Type: Journal Article
• Authors: Yin, Xiong; Wen, Kai; Huang, Weihe; Luo, Yinwei; Ding, Yi; Gong, Jing; Gao, Jianfeng; Hong, Bingyuan
• Is Part Of: Applied energy. Volume 333(2023)
• Publication Date: 2023-03-01
• Subjects: Natural gas pipeline -- Online transient simulation -- Integrated methods -- Parameter identification -- Model error compensation; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2022.120615
• Languages: English
• ISSNs: 0306-2619

Abstract

Highlights: An online simulation framework is proposed for natural gas pipeline network. Kriging-GA method is used to enhance the efficiency of parameter identification. NARX model is adopted for error compensation caused by physics-based model. Two cases are performed to verify the effectiveness of the framework. Abstract: The natural gas pipeline network is an important part of the integrated energy system. The optimal control of the integrated energy system requires accurate transient online simulation of the natural gas pipeline network. However, the existing relevant research does not make enough use of the measured data, resulting in low simulation accuracy. Therefore, a high-accuracy online transient simulation framework is proposed from four perspectives: transient simulation model selection, model parameters settings, measured data processing, and model error compensation. In this framework, measured data is fully used to improve simulation accuracy by integrating multiple physics-based and data-driven methods. First, the simulation model based on the finite volume method is selected as the basic physics-based model. Then measured data are processed based on the 3 times standard deviation method and moving average filtering to drive the simulation model. To set suitable model parameters, a data-driven method combining with the Kriging model and genetic algorithm is proposed to simultaneously identify the two key parameters of roughness and heat transfer coefficient. In addition, the final simulation errors of the physics-based model are compensated by combining the data-driven model in three modes, namely supplement, embedment, and integration. Two cases of an actual single pipeline and a complex pipeline network illustrate the applicability of the proposed framework and verify the advantage of the integrated data-driven and physics-based methods. Compared with measured data, the final simulation errors of mass flow and temperature are 0.3279% and 0.1036%, respectively. Under the trend of delicacy and intelligent operation of natural gas pipeline network, the proposed framework provides a theoretical basis for the online simulation application.