A comparison of metaheuristics for the optimal capacity planning of an isolated, battery-less, hydrogen-based micro-grid. (1st February 2020)

• Record Type: Journal Article
• Authors: Mohseni, Soheil; Brent, Alan C.; Burmester, Daniel
• Is Part Of: Applied energy. Volume 259(2020)
• Publication Date: 2020-02-01
• Subjects: Design optimization -- Micro-grids -- Smart grids -- Metaheuristics -- Hydrogen storage -- Fuel cell vehicles; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2019.114224
• Languages: English
• ISSNs: 0306-2619

Abstract

Graphical abstract: Highlights: A metaheuristic-based model is proposed to optimally size a hydrogen-based microgrid. The microgrid system is equipped with an innovative hydrogen refuelling station. The performances of eight metaheuristics are studied in terms of accuracy and speed. The moth-flame optimizer is found to outperform the popular algorithms in this area. The system's levelized costs of electricity and hydrogen are $0.09/kWh and $4.61/kg. Abstract: There has been growing interest in the development of sustainable energy systems using the potential of renewable energy sources. However, due to the intermittent nature of renewable energy sources, they must be accompanied by appropriate storage devices to be optimally integrated into so-called smart micro-grid systems. The optimal design problem of sustainable micro-grids is associated with several nonlinearities and non-convexities, and therefore is not amenable to exact methods of optimization. Accordingly, this paper proposes a metaheuristic-based approach for optimizing the size and typology of the components of an off-grid hydrogen-based micro-grid, backed up with super-capacitors and stationary fuel cell systems, in the presence of a hydrogen refuelling station. The paper also compares the performance of six recent metaheuristics. The simulations are carried out for the climatic conditions of the Feilding area, New Zealand using MATLAB. Based on the comparative results, the moth-flame optimization algorithm is found to result in a significant reduction in the total net present cost of the system in comparison with other investigated algorithms. Notably, it outperforms the genetic algorithm and the particle swarm optimization in terms of solution quality by ~2.1% and ~3.2% (equating to cost savings of $123, 910 and $188, 129 from the target system). The results also demonstrate both the technical feasibility and cost-effectiveness of the proposed stand-alone micro-grid architecture. Particularly, the levelized costs of electricity and hydrogen of the conceptualized system are found to be $0.09/kWh and $4.61/kg, respectively, which are well below the current tariffs in New Zealand.