Probabilistic evaluation of distribution power transformers reliability indices considering load transfers and mobile unit substations

• Published in: Electric power systems research 2020-10, Vol.187, p.106501, Article 106501
• Main Authors: Costa, João Guilherme de C., Leite da Silva, Armando M., Hamoud, Gomaa A., Pureza, Izabela M., Silva Neto, Noé
• Format: Article
• Language: English
• Subjects: Algorithms; Distribution substations; Electric power distribution; Electricity distribution; Feeders; Mobile unit substations; Monte Carlo simulation; Probabilistic methods; Reliability analysis; Reliability assessment; Spare transformer requirements; Statistical analysis; Stock sizing; Substations; Transformer failures; Transformers
• ISSN: 0378-7796; 1873-2046
• DOI: 10.1016/j.epsr.2020.106501

•More realistic system representation when compared to previous assessment methods for distribution transformer stock sizing.•The effects of using load transfers and mobile unit substations on the reliability indices are demonstrated.•Any probability distribution can be used to represent operating and waiting times.•Exponential and non-exponential times to failure are tested and the results discussed.•Aging exerts a significant influence on reliability indices and operating costs.•The proposed method is tested in a real system through several examples to demonstrate its flexibility and practicality. This paper proposes a new and practical probabilistic method for calculating reliability indices and costs associated with distribution substations that share spare transformers of a same group. Based on Monte Carlo simulation, the proposed algorithm considers the chronological representation of failure occurrences, and enhances previous methods by modeling emergency actions, such as transferring load to neighboring feeders and connecting a mobile unit substation to ensure electricity supply to customers during the replacement of failed transformers. Thus, spare transformer requirements with new acquisitions to replenish the stock or even to expand the system are probabilistically assessed. All waiting times associated to these actions can be duly represented by any probability distribution, including the usual exponential model. A more accurate calculation of reliability indices and costs associated with the interruption of consumers is achieved. The corresponding computational algorithm is used in several numerical applications, considering a class of transformers of a Brazilian electricity utility. The results of these applications are deeply discussed to demonstrate the practical relevance of the proposed method.