Modelagem tridimensional com o algoritmo structure from motion (SfM) para análise do potencial de geração de energia fotovoltaica em telhados: estudo de caso na vila caiçara de Picinguaba, Ubatuba-SP.

Abstract

The conversion of the world electrical matrix to renewable sources has stood out in recent decades as a safe way for nations – especially those that seek to comply with the seventh Sustainable Development Goal of the United Nations' 2030 Agenda – to become less dependent of polluting and exhaustible resources. In this context, photovoltaic generation stands out, which from sunlight, a free and constant resource, is able to supply energy at low cost due to the technological advances obtained in the last decades. Although Brazil has 62% of its matrix in the hydroelectric modal, classified as renewable, the recent variations in the water balance have put in check the capacity of the concessionaires to distribute electricity to the user, which has resulted in the increase of consumption tariffs. The continental dimensions associated with high rates of solar irradiation throughout the year make a conversion to this method favorable in Brazil. In order to encourage the use of photovoltaic energy in the country, this study seeks to provide a set of tools based on low-cost mapping technologies and three-dimensional modeling in a computational environment to detect the individual potential of roofs in buildings able to produce their own energy. The focus of the work is to model the terrain surface of Picinguaba village, in Ubatuba-SP, using the Structure from Motion (SfM) computer vision algorithm with aerial images captured by drones, and to submit them to processes of solar radiation contained in the ArcGIS software. The positional quality of the three-dimensional model was checked using topography techniques obtained by high-precision GNSS and total stations. The results found ensure that the method is feasible to be applied and replaces more expensive mappings, such as the one subsidized by laser embedded in piloted aircraft (LIDAR). The practical implications show that the photovoltaic study must be developed in more detailed scales of mapping, such as the one addressed in the proposal, mainly due to the natural and physical obstacles that sunlight encounters to the roof surfaces. Some free platforms on the web already provide geolocalized data on solar irradiation, but because they involve small mapping scales, they do not consider these elements in the calculations and, with this, they can lead to serious sizing errors of the microgenerator plants, causing losses to the consumer. The research stands out as a suggestion for a new mapping method for solar potential, able to be replicated in any urban area, which was once developed only in restricted urbanized spaces, as it required a high investment in the collection of this data. As a dissemination mechanism for the local society, the study also includes the publication of all of the analyzed content in an interactive map hosted on a freely accessible web platform, where the user will be able to consult the potential of annual solar radiation of any property in the village.