Accounting for techno-economic parameters uncertainties for robust design of remote microgrid

Abstract

This paper presents two different approaches to deal with uncertainties in the design optimization of renewable hybrid power systems in order to enhance the decision-making. The first one, Sensitivity Analysis Approach (denoted as SAA), takes the uncertainties into account after the optimization of the system. It permits first to evaluate the sensitivity on the static performances of the optimized system through uncertainty propagation. Secondly, it permits to identify the most influential uncertain parameters through Global Sensitivity Analysis (denoted as GSA). The second approach, called Robust Optimization (RO), integrates a Monte Carlo (MC) simulation into the process of optimization conducted with a Genetic Algorithm (GA). The two approaches have been set up and applied to a remote power system or microgrid, under uncertainties on techno-economic parameters. This illustrative case study is the electrical supply of a stand-alone application located in Nigeria, using photovoltaic production associated to a hybrid energy storage with a bank of batteries and a complete hydrogen chain (with an electrolyzer, a gas tank and a Fuel Cell (FC). Classically, the main source of uncertainties of such a system is associated to the temporal variation of renewable energy sources and load demands. Instead, this paper focuses on uncertainties of techno-economic parameters to improve the reliability of the optimization process for a stand-alone power system. Moreover, from a precise analysis of the state-of-the-art of such uncertainties, the authors propose to investigate complementarities of Sensitivity Analysis Approach and Robust Optimization. This study also aims to propose a methodological framework for any designer projecting to take into account uncertainties on techno-economic parameters. The results show the high interest to take into account such uncertainties for the decision-making and the ability of RO to limit their impact on system performance indicators.