Eventos Anais de eventos
ENCIT 2020
18th Brazilian Congress of Thermal Sciences and Engineering
Direct Normal Irradiance forecasting using Numerical Weather Prediction and separation models
Submission Author:
João Humberto Serafim Martins , SC
Co-Authors:
João Humberto Serafim Martins, Roberto Velasquez, Fabricio Polifke da Silva, Júlio Passos
Presenter: João Humberto Serafim Martins
doi://10.26678/ABCM.ENCIT2020.CIT20-0266
Abstract
The solar irradiation assessment, especially the Direct Normal Irradiance (DNI) component, represents a major step in the conception of Concentrated Solar Power (CSP) facilities, since they only take advantage of the DNI for power conversion. This process is vital to the accurate plant sizing and the power output estimate through its life cycle. Also, the DNI forecasting can reduce the power output uncertainty caused by solar irradiance intermittency, which leads to profitability gains and assists the facility operation and maintenance planning. In such a context, this study applies the Weather Research and Forecasting (WRF) model to assess the Global Horizontal Irradiance (GHI) within a 48-hour horizon. The site examined is Roque Gonzales, in southern Brazil, where a 250 kW-electrical rated power CSP plant is under development. GHI from the WRF model is dissociated in its beam and diffuse components with the application of the Boland-Ridley-Lauret (BRL) model, previously tuned to best fit the Brazilian weather. The forecasts performed better for clear skies, being the lowest overall errors obtained with the coarser spatial grid resolution. For clear skies, the Normalized Root Mean Square Error (nRMSE) ranged from 17% to 41% for GHI and 25% to 48% for DNI, while for cloudy conditions it ranged from 203% to 209% for GHI and from 218% to 237% for DNI, corroborating the challenges of cloudy weather forecasting.
Keywords
Direct Normal Irradiance, Numerical Weather Prediction, Weather Research and Forecasting, BRL model, concentrated solar power
DOWNLOAD PDF VIEW PRESENTATION
