Linear Programming for Real-Time Renewable Energy Operation in Low Energy Buildings

Abstract

This study presents a model for day-ahead load and renewable energy scheduling, utilizing Support Vector Regression to enhance 24-hour electricity prediction. The optimization of scheduling incorporates linear programming for real-time renewable energy operations, exploring the scenarios of photovoltaic without storage, photovoltaic with storage and no-feed-in-tariff, and photovoltaic with storage and feed-in-tariff in low energy buildings. The envisioned real-time efficient renewable energy operation involves the adoption of microgrids for electricity export and energy management. Through experimental validation on a residential building in Ifrane, Morocco, the forecasting model demonstrates an average error rate of ±8.71%. Notably, with a 4 kWp rated photovoltaic power system, the building operates autonomously during the day, acting as a positive energy structure by exporting surplus energy to the grid. In the photovoltaic with a battery and no-feed-in-tariff scenario, the building aims for energy autonomy, pursuing a zero-energy building status while exporting excess energy. In the feed-in-tariff scenario, the algorithm manages electricity storage and export, but post-midnight, grid electricity remains necessary, categorizing the residential building as a low-energy structure. Finally, while the model has broad applicability, the findings of the case study are limited.