ABSTRACT
Renewable Energy Communities (RECs) have emerged as a promising vehicle to democratize the energy transition. By enabling citizens, municipalities, and small businesses to collectively produce, share, and consume renewable energy, RECs offer pathways for reducing emissions, lowering costs, and enhancing local empowerment. However, their effective deployment is constrained by challenges in forecasting demand, optimizing planning and operations, ensuring fair allocation, and aligning optimization models with broader social and environmental goals. To address these challenges, this thesis develops and integrates computational, data-driven methods tailored to REC contexts. First, a novel forecasting methodology was designed to predict household-level, day-ahead electricity demand using only smart meter and weather data. This approach incorporates behavioural clustering, machine learning (XGBoost and ANN), and robust techniques for handling missing values, thereby avoiding intrusive data collection. Second, problem-specific optimization algorithms were developed for both REC planning and operation. Genetic algorithms were adapted to participant selection and energy allocation, explicitly considering regulatory frameworks that require predefined allocation coefficients. Complementary studies examined allocation mechanisms across diverse scenarios using clustering-based simulations, while a systematic review of REC optimization research mapped methodological trends, blind spots, and the extent to which environmental and social objectives are integrated into existing models
Committee
- President: Dr Jordi Pascual Pellicer
- Secretary: Dr Álvaro Luna Alloza
- Member: Dr Jordi Mateo Fornés
PhD Advisors:
PHD CANDIDATE
Florencia Lazzari is a PhD candidate at CIMNE’s Innovation Unit in Building, Energy and Environment (BEE Group).