Energy-based heuristic system optimizer for off-grid photovoltaic system design: HINTEDIS

HINTEDIS is a decision support tool that aims to provide a methodology to optimize photovoltaic (PV) systems at the project design phase to establish the most suitable solution for a high energy performance toward Zero Energy Building (ZEB). In addition, via a user-friendly web interface, this tool assists architects and engineers with a fast, easy and precise way to gain insight into the photovoltaic potential of their projects. It is presented via a user-friendly web interface. The web stack is written with popular open-source frameworks, React and Django, for the front-end and back-end respectively. In this study, the authors implemented a deterministic, numerical, energy-based economic system optimizer that minimized the cost of off-grid PV systems and guaranteed an arbitrarily small Loss of Load Probability (LLP). This optimizer extended typical numerical approaches by adding a set of heuristics that considerably reduced the search space and, in return, improved its computational efficiency. HINTEDIS also implements NREL’s EnergyPlus as a heat-transfer simulation engine for load estimation. Provided with enough information about a project, the numerical energy model estimates power consumption using historical local data. By estimating the power generation of each PV module with weather data and an improved solar radiation estimation model for Panama, the program could assess whether the system was supplying the load, charging the batteries, dropping the excess energy, or losing the load. This was simulated hourly for a period of a year and the data gathered was used to calculate the LLP, which accepted or rejected candidates. The list of viable candidates was then evaluated by the economic model. Each candidate was ranked by the total estimated capital cost of the system and the first one was selected and presented to the user. The economic model referenced a database of typical costs for modules, inverters, batteries, installation, and other expenses. Given a user-selected configuration, the model scaled each cost by the system size, added all expenses, and returned the expected cost for the candidate system.