FEDECOM - D4.2 System modelling library and model predictive controllers

This deliverable describes the work performed in T4.2 “Energy asset and system-level modelling and simulation” and T4.3 “MPC algorithms for optimized control of local energy systems” within the FEDECOM project. It focuses on the development of energy system simulation models and field-level Model Predictive Control (MPC) services for the three distinct pilot sites, each representing different types of federated energy communities.

Pilot 1 (Virtual Green H2 Federation) includes specific models for condensing boilers with efficiency depending on return temperature and partial load, heat storage relating temperature to the filling volume, conventional boilers based on efficiency values at different return temperatures, biomass boilers that account for the inertia of the asset, and electrolyzers for green hydrogen production facilities. In addition, models developed for the optimization tool, used in Puertollano and TMB plant models are presented, enabling more efficient hydrogen production scheduling.

Pilot 2 (Residential Hydropower Federation) includes models for building simulations, which consider the thermal envelope, internal heat gains, and heat losses. These models are built using a grey-box model approach guaranteeing a balance between detail and computational cost, facilitating their application in optimization. The building models in Lugaggia are enriched by the implementation of a Moving Horizon Estimator (MHE), used to integrate self-learning capabilities. This technique is used to recalibrate the model parameters before each execution of the MPC. Other modelled assets include: 1) Storage aging models, especially for electrical storage, which consider various parameters influencing aging to determine degradation with high accuracy; 2) Heat pumps created using performance curves and calibrated with measured data and calibrated with standard least squares linear regression technique and 3) Water tanks, modelled through a first order model.

Pilot 3 (Cross-country e-Mobility Federation) focuses on the electrical domain with models for electrical vehicles (EVs) and battery systems.. The EVs are created using hybrid models and include V2G scenarios. The battery models are simplified for use in an MPC framework to balance accuracy and computational efficiency. These models utilize time series data to describe operational states and historical performance.

The modelling library is completed by presenting the results and discussion of simulation and calibration efforts, with models validated using real-world data and real-time measurements for adaptive control.

The report also highlights the development and implementation of Model Predictive Control (MPC) algorithms for optimized control of local energy systems. These MPC algorithms enable the execution of optimal control of assets and systems, minimizing the tracking error with respect to optimal energy demand profiles while achieving optimization objectives such as cost and energy savings. The MPCs leverage the First Principle Reduced Order (FPRO) models from the system modeling library. The report details MPCs for the three pilot sites. The MPC also incorporates a mechanism to integrate recommendations from a cross-vector optimiser, acting as an Energy Hub. Additionally, Grid Singularity integrated and tested a Tekniker-developed grey-box heat pump coefficient of performance (COP) model as an additional option for configuring heat pump digital twins in the Grid Singularity Exchange.