Costa Júnior, João Carlos ArantesPraxedes, Kelvin da Cruz2017-12-052017-12-052017-08-25PRAXEDES, Kelvin da Cruz. Estudo de modelos de banda larga para estimativa da irradiação direta normal. 2017. 116f. Dissertação (Mestrado em Engenharia Mecânica) - Centro de Tecnologia, Universidade Federal do Rio Grande do Norte, Natal, 2017.https://repositorio.ufrn.br/jspui/handle/123456789/24419The present study has as main objective to study twelve broadband models for clean sky to estimate Direct Normal Irradiation (IDN), and to choose the most suitable for the climatic conditions of the city of Natal-RN. To do this, a revision was necessary about the main methods of estimation and measurement of IDN to determine the best among them, and used it in the research. Then the value of the Angstrom turbidity coefficient (β) was estimated using a mathematical model of parameterization proposed by Louche et al. (1987). For the estimation of β and the calculation of the IDN values of the models used in the study, data from the CTGÁS-ER meteorological station, located in Natal-RN (altitude 84 m, latitude 5.82 ° S, longitude 35.23 ° W). The solar station also provided IDN data measured from a pyreliometer, where it was possible to compare the results obtained from the models with those measured, and verify the viability of these models for the climatic conditions of the studied region. The value of β is presented in this work in monthly averages. The obtained result follows in parts the expected pattern, which is of lower values of β in the winter months, and higher values of β in the summer months. The months of December and January fall short of this pattern, mainly due to the amount of rainfall in these months being more than 40% higher in January and 200% in December compared to the average of the last ten years. The minimum and maximum values presented for β are 0.094 in July and 0.128 in March. This proximity of values is due to the fact that there is not a considerable variation in the value of water precipitation thickness throughout the year, mainly because the average temperature in Natal be practically the same throughout the year. Regarding broadband models, these were divided into two groups for analysis, considering the number of input parameters. Those that need four or fewer parameters are considered simple models, and those that need more than four parameter complex models. In order to validate the results, we used six statistical methods, the Mean Bias Error (MBE), the Mean Error Root Mean (RMSE), the t-statistic (TT), the Expanded Uncertainty (U95), the Coefficient of Pearson correlation (r) and the Determination Coefficient (R2). The models were classified and ranked according to the performance of each of the statistical methods, calculated through the Global Performance Indicator (GPI). The results show that for the city of Natal, the number of parameters does not determine the performance of the models, since the simple models obtained one of the three best results, and also the two worst ones.Acesso AbertoIrradiação direta normalModelos de céu-limpoModelos de banda largamasterThesisCNPQ::ENGENHARIAS::ENGENHARIA MECANICA