Viabilidade fotovoltaica residencial: uma abordagem envolvendo volatilidade estocástica e opções evolucionárias

Abstract

The development of power plants based on renewable sources gave rise to new paradigms for energy generation in the world. However, despite the benefits of this type of generation, uncertainties about the future mitigate further growth in the sector, leaving governments to promote these projects. The laws of the electricity sector in Brazil underwent a reformulation in recent years and came into force in 2023, when they began to burden the returns of agents, especially residential producers, earning the nickname “sun tax”. This new regulatory framework raised the question for newcomers to residential solar generation: "Should I or should I not invest in the solar project and under what conditions?" Thus, it remains for the decision-maker to evaluate scenarios and analyze the managerial flexibilities inherent to the process, such as, for example, carrying out the investment in full or in part, postponing and/or considering an investment in a risky asset in the financial market. Therefore, this work evaluates the initiative of investing in a photovoltaic system, adopting the Theory of Real Options to more faithfully represent this decision-making context in Brazil, considering the perspective of a residential consumer who consumes an average of 360kWh/month and its flexibility managerial. Thus, a malleable and dynamic tool is developed that aims to minimize the agent’s sum by incorporating related uncertainties and that can be replicated for agents with higher energy consumption or in several other contexts. The methodological approach used in this work consists of simulating stochastic processes and optimizing the decision set by evolutionary metaheuristics. Initially, benchmark results from the literature are compared with trigger curves considering different genotypic combinations created for different optimizer algorithms. Subsequently, the uncertainties are modelled considering and comparing the solutions by applying the Heston and MGB+Jumps models to represent the price of electricity and applying the MGB model to represent the IVVB11 index. To calibrate Heston’s parameters, an approach based on DeepONet was developed. The results demonstrate an impact of around 20% to 30% in the savings obtained by the agent, depending on the stochastic approach considered in the project evaluation. Each stochastic approach generates different behaviours in the decision set, however, they agree on the delay in the moment of exercising the options. This indicates that the new Brazilian policy should slow down the growth of the residential photovoltaic sector, but it is not capable, by itself, of making such investments unfeasible.