Models

HAMLET’s simulation framework is built upon detailed mathematical models that define the behavior of key energy system components and decision-making processes. This section provides an in-depth look into the various models used, their assumptions, and mathematical formulations.

HAMLET’s models are categorized into two main groups:

  1. Component Models: These describe the physical behavior of energy system elements, such as energy consumption, generation, and storage.

  2. Home Energy Management System (HEMS): This includes forecasting, control strategies, and trading mechanisms that enable decision-making.

Component Models

HAMLET models various energy system components to simulate decentralized energy systems accurately. The key component models include:

  • Electricity Load: Models demand patterns based on occupancy, appliance usage, and stochastic variations.

  • PV Generation: Uses weather data and panel specifications to compute solar power generation.

  • Wind Power: Simulates wind-based electricity generation based on wind speed distributions and turbine efficiency.

  • Battery Storage: Captures state-of-charge dynamics, charge/discharge efficiency, and degradation effects.

  • Heat Demand: Computes thermal energy needs based on building properties, external temperatures, and user preferences.

  • Electric Vehicles: Models driving patterns, charging behavior, and vehicle-to-grid interactions.

  • Grid Constraints: Includes voltage and line capacity constraints affecting power flows.

Home Energy Management System (HEMS)

The HEMS models determine how agents interact with markets, grids, and their own energy assets. It is structured into:

  • Forecasting Methods: Predict future energy demand, generation, and market prices using: - Naïve methods (e.g., persistence models) - Statistical models (e.g., SARMA, ARIMA) - Machine learning techniques (e.g., Random Forest, Neural Networks)

  • Control Strategies: Optimize energy usage based on: - Rule-Based Control (predefined heuristics) - Model Predictive Control (MPC) - Reinforcement Learning (self-learning agent behavior)

  • Trading Strategies: Determine how agents buy and sell energy in markets, including: - Zero Intelligence (randomized bids) - Linear Bidding (progressively adjusted offers) - Retailer-based pricing (buying/selling at fixed prices)