A model predictive control framework with customer-priority tiers for virtual power plant resilience during extreme weather: A UK heatwave case study

Edward Moroshko, Weizhe Qin, Desen Kirli, Mohammed Qais, Sotirios A. Tsaftaris, Aristides E. KiprakisarXiv 2025 · DOI · arXiv

Introduction

As climate change intensifies, electrical grids are increasingly subjected to extreme weather events like heatwaves and severe storms. These events create a "double whammy" for grid stability: electricity demand spikes (e.g., for air conditioning) while supply becomes precarious due to infrastructure strain or thermal limits on generation and storage.

Virtual Power Plants (VPPs)—which aggregate distributed energy resources like solar PV, battery energy storage systems (BESS), and flexible loads—have emerged as a promising solution for grid support. However, most existing VPP control strategies are optimized for "blue sky" economic profit. During "dark sky" emergencies, these traditional controllers often fail because they rely on static plans and perfect forecasts. This paper proposes a Model Predictive Control (MPC) framework specifically designed to prioritize grid resilience and essential services during extreme weather.

The Proposed MPC Framework

The authors introduce a framework formulated as a Multi-Objective Mixed-Integer Quadratically Constrained Program (MIQCP). Unlike static optimization, this approach uses a receding horizon, re-optimizing the system every hour based on real-time measurements and updated forecasts. This allows the system to "course-correct" when reality deviates from predicted solar generation or load levels.

System architecture of the proposed VPP resilience framework, illustrating the interaction between weather events, the error simulation model, and the MPC control system.

Key Technical Innovations

  • Customer-Priority Tiers: The objective function explicitly weights loads into three categories: Critical (e.g., hospitals), High, and Low. This enables "surgical" load shedding, ensuring essential services remain powered even when total demand exceeds supply.
  • Enhanced Delta LinDistFlow: To maintain computational efficiency without sacrificing accuracy, the researchers developed a "delta" version of linearized distribution flow. This method optimizes changes around the most recent operating point rather than a fixed nominal point, significantly reducing linearization errors in the power flow model.
  • Thermal-Aware Battery Modeling: Recognizing that high temperatures degrade battery performance, the framework incorporates a time-varying efficiency model. This is crucial during heatwaves, where standard constant-efficiency assumptions lead to significant energy shortfalls.

Case Study: The 2022 UK Heatwave

The framework was validated using a simulation of the IEEE 33-bus distribution network integrated with real-world data from the record-breaking July 2022 UK heatwave. Temperature records from Coningsby, UK (reaching 40.3°C), were used alongside hourly electricity prices from the Elexon BMRS Market Index and realistic UK-based demand and solar generation patterns.

The simulation identified a critical "Extreme Weather Alert" window. Five hours before an expected outage, the MPC automatically shifts from "economic mode" (minimizing costs) to "resilience mode" (aggressive battery pre-charging). This proactive stance ensures maximum energy availability for the duration of the weather event, even if it requires importing power at high market prices.

Key Results and Insights

  • Resilience Boost: Under realistic conditions involving forecasting errors and modeling uncertainties, the MPC framework improved the grid's resilience metric by 11–20% compared to traditional full-horizon optimization methods.
  • Robustness to Forecast Bias: Even when forecasts consistently underestimated demand (negative bias), the MPC maintained a high resilience score (R ≈ 0.90), whereas static approaches saw performance degrade significantly (R ≈ 0.75).
  • Addressing the Energy Shortfall: The study demonstrated that ignoring temperature effects on batteries can lead to an energy shortfall of up to 150 kWh. Because the MPC monitors the actual state of charge every hour, it successfully compensates for these losses by adjusting import schedules.
  • Voltage Stability: The "delta" optimization approach ensured that voltages remained closer to nominal values (0.9607 p.u. average) compared to baseline methods, maintaining power quality under extreme stress.

Conclusion and Future Work

The research highlights the "illusion of optimality" in static grid planning. While full-horizon plans may look perfect on paper, they often become physically impossible to execute during real-world emergencies. The proposed MPC framework offers a robust alternative that prioritizes human safety and critical infrastructure.

Future research directions include moving toward Stochastic MPC to better handle the probabilistic nature of outage timings and integrating a wider variety of distributed resources, such as Electric Vehicles (EVs) and wind generation, into the VPP coordination framework.

Summary updated

BibTeX
@misc{moroshko2025model,
  author = {Edward Moroshko and Weizhe Qin and Desen Kirli and Mohammed Qais and Sotirios A. Tsaftaris and Aristides E. Kiprakis},
  title = {{A model predictive control framework with customer-priority tiers for virtual power plant resilience during extreme weather: A UK heatwave case study}},
  year = {2025},
  eprint = {2511.22528},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2511.22528},
}
Figure