A Bayesian Approach to the Reliability Analysis of Renewables-Dominated Islanded DC Microgrids

• Published in: IEEE transactions on power systems Vol. 36; no. 5; pp. 4296 - 4309
• Main Authors: Eajal, Abdelsalam A., El-Awady, Ahmed, El-Saadany, Ehab F., Ponnambalam, Kumaraswamy, Al-Durra, Ahmed, Al-Sumaiti, Ameena S., Salama, Magdy M. A.
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.09.2021; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adequacy; Algorithms; Alternative energy sources; Bayes methods; Bayesian analysis; Bayesian network; DC microgrid; Distributed generation; distributed generator; droop control; Electric potential; Electric power supplies; Electric power systems; Energy sources; Islanding; Microgrids; Monte Carlo simulation; Network reliability; Penetration; Power flow; Power system reliability; Probabilistic logic; Reliability; Reliability analysis; renewable power; Statistical analysis; Voltage; Voltage control
• ISSN: 0885-8950; 1558-0679
• DOI: 10.1109/TPWRS.2021.3056314

The DC microgrid (DC MG) concept enables the hosting of DC-type renewable energy resources. However, their intermittent nature means that a high penetration of renewables can jeopardize supply adequacy and voltage provision during islanding. The work presented in this paper was therefore directed at developing a probabilistic graphical approach based on Bayesian networks (BNs) for the reliability analysis of renewables-dominated DC MGs. The proposed BN model incorporates a family of novel reliability indices for quantifying the impact of a high penetration of renewables on MG reliability, including loss of renewable power supply, rise in voltage, and reversal of power flow. The model is supported by a newly formulated fast and accurate linearized power flow algorithm for probability calculations. The accuracy of the BN model has been verified against a Monte-Carlo simulation (MCS). The effective application of the new BN model for reasoning and impact assessment reveals that a high penetration of renewables affects reliability indices differently. Case study results suggest that the proposed BN model shows promise as a valuable tool for the reliability analysis of renewables-dominated MGs that feature islanding capability.