Aplicação de redes neurais na otimização de dutos de sucção de turbinas Francis

Abstract

This study proposes the application of artificial neural networks to optimize the performance of Francis turbines, with particular emphasis on draft tube efficiency. The research is motivated by the need to enhance the energy efficiency of these systems, supporting sustainable power generation in alignment with the sustainable development goals. The methodology integrates computational fluid dynamics (CFD) simulations with machine learning techniques, utilizing operational data from a Francis turbine provided by HACKER Industrial, with specific evaluation focused on the NW parameter. The results demonstrate neural networks' capability to optimize critical performance parameters of the draft tube, presenting an innovative approach for hydraulic system optimization. The study highlights the promising potential of combining computational modeling with artificial intelligence as an advanced tool for the renewable energy sector.