Authors

Max Fu, Akanksha Patil

Research Question

How can Rome’s power grid flexibility be optimized at the territorial cell level to support rising demand driven by electric vehicles, heat pumps, photovoltaic systems, and battery energy storage systems while minimizing inefficiencies and ensuring reliable operation?

Background

Growing demand for renewable energy and electricity necessitates a shift to decentralized smart grids, where customers actively participate alongside multiple local energy sources. Prosumers must collaborate with grid operators to optimize energy flows and maintain stability.

Methodology

Following data preprocessing, descriptive analysis was used to explore the geographic distribution and temporal progression of key variables—customers, agreement power, and produced power—to analyze trends and model customer consumption evolution. This analysis is further categorized by customer features: intended use, charging stations, and supply type. The study analyzes geographical distribution across urban areas and annual growth trends and rates. Variables include customers and available power categorized by intended use (domestic or other), charging stations (public or private), and supply type (consumer or producer). Predictive analysis is used for demand forecasting, and feature engineering incorporates time-based features, lag features, and external factors to enhance forecasting accuracy. For model selection, ARIMA is used for short-term, stationary data, and SARIMA accounts for seasonal adjustments. Hyperparameter tuning is conducted using a grid search to optimize AIC for ARIMA and SARIMA to identify the best parameters. Model evaluation metrics include RMSE, MAE, and R² score to assess accuracy.

Deliverables

• Data Dashboard
• Technical Report