Integrating photovoltaic inverter reliability into energy yield estimation with Markov models

Sairaj V. Dhople, Ali Davoudi, Patrick L. Chapman, Alejandro D. Domínguez-García

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Markov reliability models to estimate Photovoltaic (PV) inverter reliability are proposed for utility-interactive systems. These are then extended to generate a unified PV energy-yield model. The integrated reliability-energy yield framework is superior to conventional methods as it accounts for inverter failures and repairs. The proposed analytical framework is utilized to compare conventional central inverter architectures to emerging architectures that employ microinverters. Case studies applied to a 9 kW residential system indicate that over a 25 year period, in typical operating conditions, microinverters provide higher energy yield as compared to a conventional system. Additionally, the analysis demonstrates that the energy yield is more sensitive to the repair time compared to the mean time to failure of the inverters.

Original languageEnglish (US)
Title of host publication2010 IEEE 12th Workshop on Control and Modeling for Power Electronics, COMPEL 2010
DOIs
StatePublished - Oct 22 2010
Event2010 IEEE 12th Workshop on Control and Modeling for Power Electronics, COMPEL 2010 - Boulder, CO, United States
Duration: Jun 28 2010Jun 30 2010

Publication series

Name2010 IEEE 12th Workshop on Control and Modeling for Power Electronics, COMPEL 2010

Other

Other2010 IEEE 12th Workshop on Control and Modeling for Power Electronics, COMPEL 2010
CountryUnited States
CityBoulder, CO
Period6/28/106/30/10

Keywords

  • Energy yield estimation
  • Markov reliability models
  • Photovoltaic energy conversion
  • Utility-interactive inverters

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'Integrating photovoltaic inverter reliability into energy yield estimation with Markov models'. Together they form a unique fingerprint.

Cite this