A Capacity-Price Game for Uncertain Renewables Resources

Pan Li, Shreyas Sekar, Baosen Zhang

TL;DR

In decentralized energy markets, selfish behavior by small producers usually leads to high prices and inefficient infrastructure. This paper proves that the inherent randomness of renewable energy (like clouds passing over solar panels) actually forces producers into a "conservative" Nash equilibrium that is socially optimal and highly efficient as the number of participants grows.

What Problem Does This Paper Address?

As we move toward a grid dominated by distributed energy resources (DERs) like rooftop solar, we face a coordination challenge. Traditionally, a "Social Planner" (a utility or central operator) decides how much capacity to build to minimize total costs. However, in the real world, thousands of individual owners make these decisions based on their own profit motives.

Previous game-theoretic research on deterministic systems (like communication networks) suggested that decentralized competition is "bad"—it leads to inefficient over-investment or price-gouging. The authors of this paper ask: Does the unpredictable nature of wind and sun make this problem worse, or could it actually help?

Methods

The authors model the market as a two-level capacity-price game:

  1. Upper Level (Capacity Game): Producers decide how much solar capacity ($C_i$) to install, balancing the cost of investment against expected future revenue.
  2. Lower Level (Price Sub-game): Once capacity is built, producers compete to sell their energy using a Bertrand competition model (competing on price). Because renewable marginal costs are near zero, this is modeled as a mixed-strategy Nash equilibrium where producers bid randomly within a range.

The researchers use mathematical proofs (including the Berry-Esseen Theorem) to analyze the efficiency of this game compared to a centralized "Social Planner" benchmark. They evaluate the system using two metrics: Social Cost (investment + grid imbalance costs) and Market Efficiency (the ratio of investment costs to consumer payments).

Datasets

The paper validates its theoretical findings using two types of data:

  • Synthetic Data: Uniformly distributed random variables for generation and demand to test the basic mechanics of the 2-player and N-player games.
  • Real-World PV Data: 5-minute resolution solar generation profiles from distributed PV sites in California, provided by the National Renewable Energy Laboratory (NREL). This data was used to account for the high spatial correlation typical of weather patterns.

Key Insights and Ideas

  • Randomness as a Regulator: In a deterministic world, producers can exploit "corner cases" to maximize profit at the consumer's expense. In a renewable world, the uncertainty of their own production (will the sun shine?) induces a natural conservatism. This prevents them from exercising excessive market power.
  • The Efficiency of Large Numbers: As the number of producers ($N$) increases, the decentralized market's total capacity and pricing converge to the Social Planner’s optimal solution. The "Price of Anarchy" effectively disappears.
  • Correlation Doesn't Break the Model: Even when solar production is highly correlated (e.g., a cloud covers an entire neighborhood), the market remains efficient. The individual-level randomness is enough to maintain healthy competition.
  • The "Long-Run" Benefit: In deterministic models, as technology gets cheaper (lower investment cost $\gamma$), market efficiency actually gets worse because producers keep prices high while costs drop. In this stochastic model, efficiency remains high regardless of falling tech costs.

Key Results

  • Existence of Optimal Equilibrium: The authors prove that a Nash equilibrium always exists that exactly matches the Social Planner’s minimum social cost.
  • Asymptotic Efficiency: They demonstrate that as $N \to \infty$, the total payment made by consumers approaches the actual investment costs of the producers ($\xi \to 1$).
  • Simulation Performance: Using real California solar data, the market efficiency ($\xi$) was calculated at 0.83 for 5 producers and rose to 0.98 for 120 producers.
  • Capacity Buffer: Unlike deterministic markets where capacity equals demand, the "uncertainty game" results in a total capacity that is ~30% higher than mean demand, providing a necessary buffer for reliability.

Limitations and Future Work

  • Auction Format: The study focuses specifically on "pay-as-bid" auctions. The authors note that other formats, like Cournot (quantity-based) or clock auctions, might behave differently under uncertainty.
  • Equilibrium Multiplicity: While they prove an efficient equilibrium exists, the game can technically have multiple equilibria. The paper "side-steps" this by focusing on the efficiency of the symmetric case, but future work could explore which equilibrium is most likely to emerge in practice.
  • Storage: The current model does not explicitly account for the strategic impact of battery storage, which could allow producers to "time-shift" their randomness, potentially changing the competitive dynamics.

Summary updated

Abstract

Renewable resources are starting to constitute a growing portion of the total generation mix of the power system. A key difference between renewables and traditional generators is that many renewable resources are managed by individuals, especially in the distribution system. In this paper, we study the capacity investment and pricing problem, where multiple renewable producers compete in a decentralized market. It is known that most deterministic capacity games tend to result in very inefficient equilibria, even when there are a large number of similar players. In contrast, we show that due to the inherent randomness of renewable resources, the equilibria in our capacity game becomes efficient as the number of players grows and coincides with the centralized decision from the social planner's problem. This result provides a new perspective on how to look at the positive influence of randomness in a game framework as well as its contribution to resource planning, scheduling, and bidding. We validate our results by simulation studies using real world data.
BibTeX
@article{li2022capacity,
  author = {Pan Li and Shreyas Sekar and Baosen Zhang},
  title = {{A Capacity-Price Game for Uncertain Renewables Resources}},
  journal = {IEEE Trans. Sustain. Comput.},
  year = {2022},
  url = {https://openreview.net/forum?id=OxQmClmFue},
}

Loading related content...

Figure