Uncertainty and Autarky: Cooperative Game Theory for Stable Local Energy Market Partitioning

Saurabh Vaishampayan, Maryam Kamgarpour

TL;DR

This paper explores how to best divide a power distribution grid into Local Energy Markets (LEMs) when energy production and consumption are uncertain. Using cooperative game theory, the authors demonstrate that while a single large market is most efficient under perfect conditions, uncertainty and grid constraints often make smaller, "autarkic" (self-sufficient) coalitions more stable and cost-effective for both the grid operator and prosumers.

What Problem Does This Paper Address?

As more households become "prosumers" (producing and consuming energy via solar panels, batteries, etc.), regulations are shifting to allow them to trade energy locally. However, a major question remains: What is the right scale for these markets? Should an entire city be one market, or should it be broken down by neighborhood or even individual street feeders?

The challenge is a tug-of-war between two parties:

  1. The Distribution System Operator (DSO): They want to minimize total grid costs, including the risk of line overloading and voltage violations caused by unpredictable solar generation or demand spikes.
  2. Prosumers: They want to minimize their individual energy bills. If a prosumer feels they are paying too much to support a large market, they might "break away" to form a smaller, more beneficial group.

Current research often ignores the physical constraints of the wires (the grid) or the "externalities"—how one neighborhood's trading affects the voltage levels of another. This paper provides a framework to find a "stable" partition where no group of prosumers has an incentive to leave.

Methods

The authors combine Cooperative Game Theory with Stochastic Optimization through a two-stage model:

  • Two-Stage Cost Model:
    • Ex-ante stage: Markets are planned based on energy forecasts to minimize expected costs.
  • Ex-post stage: Real-world uncertainty hits. If the actual sun doesn't shine as predicted, the grid incurs "penalty costs" for energy imbalances and physical constraint violations (like overvoltage).
  • Stability Analysis (The Core): They use the game-theoretic concept of the "Core." A partition is stable only if the costs are allocated such that no subset of prosumers can achieve lower costs by forming their own independent market.
  • Linear DistFlow: To keep the math manageable while respecting physics, they use a linearized version of AC power flow equations to model how energy moves through the radial lines of a distribution grid.
  • Algorithm 1: Since finding the optimal stable partition is NP-hard (the number of ways to group prosumers grows exponentially), they propose an iterative algorithm to evaluate partitions and check for core stability under uncertainty.

Datasets

The researchers evaluated their framework on two main setups:

  1. Modified IEEE 33-bus system: A standard benchmark for power systems research, modified to include three distinct branches (Nightclub/Wind, Residential/PV, and Commercial/Biowaste).
  2. Lausanne Digital Twin: A real-world dataset from a 400V low-voltage network in Lausanne, Switzerland. It includes 43 prosumers with residential loads, heat pumps, and solar PV, using anonymized smart meter data.

Key Insights and Ideas

  • Uncertainty as a Centrifugal Force: Under perfect information, the "Grand Coalition" (one giant market) is always best because it maximizes trading opportunities. However, as forecast errors (noise) increase, the risk of expensive grid violations grows. This pushes the optimal setup toward smaller, more localized markets that can manage their own imbalances locally—a state called autarky.
  • Externalities Matter: In power grids, your neighbor’s trading isn't just their business; it changes the voltage at your plug. The paper shows that these physical "externalities" make market stability much harder to achieve because a group’s costs depend on what everyone else in the grid is doing.
  • The "Sweet Spot" of Localization: The results suggest that neither extreme—one giant market nor total individual isolation—is usually optimal. Instead, moderate localization (grouping prosumers on the same feeder) often provides the best balance of trading gains and risk mitigation.

Key Results

  • Theorem of the Grand Coalition: The authors mathematically prove that if forecasts are perfect and there are no grid externalities, the largest possible market is always the most stable and efficient.
  • Impact of Noise: In the IEEE 33-bus simulation, as forecast noise increased from 0% to 20%, the optimal stable partition shifted from one large market to three separate, branch-specific markets.
  • Real-World Validation: In the Lausanne case study, a partition involving a small neighborhood coalition of 5 prosumers reduced costs for that neighborhood by 39% compared to a single grid-wide market, while also reducing total grid-wide costs.

Limitations and Future Work

  • Computational Complexity: The current algorithm uses an iterative search, which is "computationally prohibitive" for very large grids. The authors note that developing more efficient, scalable algorithms is a primary direction for future work.
  • Simplified Physics: While Linear DistFlow is standard, it ignores power losses and the non-linearities of full AC power flow, which could be significant in highly stressed grids.
  • Investment Decisions: The paper focuses on short-term operational costs. It does not yet account for long-term decisions, such as where the DSO should invest in new transformers or where prosumers should install larger batteries.

Summary updated

Abstract

Local energy markets empower prosumers to form coalitions for energy trading. However, the optimal partitioning of the distribution grid into such coalitions remains unclear, especially in constrained grids with stochastic production and consumption. This analysis must take into account the interests of both the grid operator and the constituent prosumers. In this work, we present a cooperative game theoretic framework to study distribution grid partitioning into local energy market coalitions under uncertain prosumption and grid constraints. We formulate the optimal stable partitioning problem to balance the interests of the grid operator with that of prosumers. Under deterministic load and generation, we show that the largest market coalition is the optimal stable partition. For the case of stochastic loads and generation, we provide an algorithm to evaluate the optimal stable partition. Numerical experiments are performed on benchmark and real world distribution grids. Our results help in understanding how uncertainty affects local energy market partitioning decisions in constrained distribution grids.
BibTeX
@misc{vaishampayan2026uncertainty,
  author = {Saurabh Vaishampayan and Maryam Kamgarpour},
  title = {{Uncertainty and Autarky: Cooperative Game Theory for Stable Local Energy Market Partitioning}},
  year = {2026},
  eprint = {2603.05169},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2603.05169},
}
Figure