Please use this identifier to cite or link to this item: http://repositorio.unicamp.br/jspui/handle/REPOSIP/335282
Type: DISSERTAÇÃO DIGITAL
Degree Level: Mestrado
Title: Estudo de modelos matemáticos para análise da radiação solar e desenvolvimento de ferramenta para modelagem e simulação de sistemas fotovoltaicos
Title Alternative: Study on mathematical models for the analysis of solar radiation and development of a modeling and simulation tool for photovoltaic systems
Author: Silva, Michelle Kitayama da, 1990-
Advisor: Villalva, Marcelo Gradella, 1978-
Abstract: Resumo: The energy generated by a photovoltaic system varies according to climatic changes such as irradiance and temperature. Thus, obtaining reliable meteorological and irradiance data is one of the most important steps in the performance analysis of photovoltaic systems. Non-accurate irradiance information leads to a significant increase in uncertainties in the final energy output. This occurs not only due to imprecisions of the input data but also due to the progressive increase of uncertainties in the conversion of the irradiance from the horizontal plane to the inclined surface of the photovoltaic modules. Thus, the objective of this work is to evaluate the accuracy of several solar irradiance modeling methods, such as the decomposition models of Erbs, DISC and DIRINT, Liu and Jordan, Klucher, Hay and Davies, Reindl and Perez transposition models, and a simplified analysis of the hourly synthesis model of Aguiar and Collares-Pereira. The models were implemented in Python and are part of the design and simulation tool of Photovoltaic Systems Connected to the Electrical GRID (SFCRE) that is being developed. The results of the hourly synthesis, decomposition and transposition models were compared with irradiance data measured in Golden, Colorado, by NREL, and the results of the energy estimation were compared with energy data collected in Araçariguama - SP. The results showed that different databases present irradiance variations for the same locality. The combination of the DIRINT decomposition model and Perez transposition model showed the best results for the Golden locality, with a mean error of -1.61% and mean square error of 3.18% per year. In general, the developed tool presented satisfactory results for the simulation of photovoltaic systems and the estimation of energy generated, compared to measured energy data and simulation results using PVSYST®. The mean square error of the simulation results compared to the PVSYST ® using the Perez transposition model was 2.52%, and 6.94% compared to measured data

Abstract: The energy generated by a photovoltaic system varies according to climatic changes such as irradiance and temperature. Thus, obtaining reliable meteorological and irradiance data is one of the most important steps in the performance analysis of photovoltaic systems. Non-accurate irradiance information leads to a significant increase in uncertainties in the final energy output. This occurs not only due to imprecisions of the input data but also due to the progressive increase of uncertainties in the conversion of the irradiance from the horizontal plane to the inclined surface of the photovoltaic modules. Thus, the objective of this work is to evaluate the accuracy of several solar irradiance modeling methods, such as the decomposition models of Erbs, DISC and DIRINT, Liu and Jordan, Klucher, Hay and Davies, Reindl and Perez transposition models, and a simplified analysis of the hourly synthesis model of Aguiar and Collares-Pereira. The models were implemented in Python and are part of the design and simulation tool of Photovoltaic Systems Connected to the Electrical GRID (SFCRE) that is being developed. The results of the hourly synthesis, decomposition and transposition models were compared with irradiance data measured in Golden, Colorado, by NREL, and the results of the energy estimation were compared with energy data collected in Araçariguama - SP. The results showed that different databases present irradiance variations for the same locality. The combination of the DIRINT decomposition model and Perez transposition model showed the best results for the Golden locality, with a mean error of -1.61% and mean square error of 3.18% per year. In general, the developed tool presented satisfactory results for the simulation of photovoltaic systems and the estimation of energy generated, compared to measured energy data and simulation results using PVSYST®. The mean square error of the simulation results compared to the PVSYST ® using the Perez transposition model was 2.52%, and 6.94% compared to measured data
Subject: Radiação solar
Efeito fotovoltaico
Sistemas de energia fotovoltaica
Energia renovável
Language: Português
Editor: [s.n.]
Citation: SILVA, Michelle Kitayama da. Estudo de modelos matemáticos para análise da radiação solar e desenvolvimento de ferramenta para modelagem e simulação de sistemas fotovoltaicos. 2019. 1 recurso online (117 p.). Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação, Campinas, SP.
Date Issue: 2019
Appears in Collections:FEEC - Tese e Dissertação

Files in This Item:
File SizeFormat 
Silva_MichelleKitayamaDa_M.pdf3.73 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.