A novel physical and data-driven optimization methodology for designing a renewable energy, power to gas and solid oxide fuel cell system based on ensemble learning algorithm

Xiaoyi Ding, Yifan Wang, Pengcheng Guo, Wei Sun, Gareth P. Harrison, Xiaojing Lv and Yiwu Weng

Energy, 2024, vol. 313, issue C

Abstract: Power-to-gas (P2G) is a fast-developing energy-storage technology that could increase flexibility of renewable sources, creating green-powered bond between wind & PV sources and gas-based power system. The combination of P2G and solid oxide fuel cell/gas turbine (SOFC/GT) in a multi-energy system (MES) provides a promising solution to reduce renewable curtailment with high efficiency and low emission. However, due to the complicated topology and thermal-coupled safety risks of SOFC/GT, detailed thermodynamic modeling is commonly required to evaluate its coordination with intermittent renewable sources, yet significantly increasing computational complexity and time cost for engineering application. To overcome the challenge, this paper presents a novel physical and data-driven approach to effectively designing a MES system, with explicit considerations of safety risk and fluctuations of renewable source & local load. This framework specifically targets the computational burden associated with repeated simulations and sample evaluations, which are integral to high-fidelity models—particularly those involving detailed thermodynamic models in MES. Variations of SOFC/GT mass & power flow boundaries are extracted based on multiple safety criteria, which are used as database for training of ensemble learning (EL) algorithm. Meanwhile, under the proposed MES power management strategy, the generated data-driven model is then coupled with wind power, H2 storage and P2G module, where a multi-objective optimization is carried out to achieve low wind curtailment and life-cycle cost. Finally, the proposed method is compared against traditional physics-based optimization technique using standard genetic algorithm (GA). Results showed that with the physical and data-driven methodology, the overall computing efficiency of optimization is improved significantly by 37 times compared with traditional method. Wind curtailment rate lower than 0.5 % and life-cycle cost below 2.6∗106￡of MES could be achieved from results in Pareto front. Among the evaluated design parameters, SOFC cell number has a dominating and nonlinear effect on SOFC/GT flexibility. The proposed framework allows for quick and accurate evaluations of MES, making it beneficial for industrial applications when dealing with large data sets, thermally-coupled processes and complex system topologies.

Keywords: Machine learning; Solid oxide fuel cell; Power-to-gas; Renewable energy; Multi-energy system; Data-driven (search for similar items in EconPapers)
Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0360544224037800
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:energy:v:313:y:2024:i:c:s0360544224037800

DOI: 10.1016/j.energy.2024.134002

Access Statistics for this article

Energy is currently edited by Henrik Lund and Mark J. Kaiser

More articles in Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().