TL;DR
This paper investigates whether renewable energy suppliers should invest in storage when competing in a local market. It reveals the counter-intuitive finding that the inherent uncertainty of renewables can actually benefit suppliers by softening market competition, and that being a "first-mover" in storage investment can sometimes be a disadvantage.
What Problem Does This Paper Address?
Renewable energy sources like wind and solar are inherently intermittent. In modern "two-settlement" energy markets, suppliers commit to a certain amount of power in a day-ahead market but are penalized in real-time if their actual generation falls short. While energy storage (like large-scale batteries) can stabilize output and avoid these penalties, it comes with a hefty price tag.
The authors tackle a strategic dilemma: In a local market where multiple suppliers compete for the same customers, should they invest in storage? Most existing research looks at storage for a single supplier in isolation. This paper fills the gap by modeling how competition changes the "to store or not to store" decision, considering both market dynamics and investment costs.
Methods
The researchers use a three-stage game-theoretic model solved via backward induction:
- Stage I (Investment): Suppliers decide whether to invest in storage at the start of a long-term horizon.
- Stage II (Bidding): On a daily basis, suppliers set their bidding prices and quantities in a day-ahead market.
- Stage III (Purchasing): Consumers choose which supplier to buy from to maximize their cost savings.
To model the price-quantity competition in Stage II, they employ the Bertrand-Edgeworth model, which handles price competition where firms have capacity constraints. A key technical contribution is solving this game for both pure and mixed strategy equilibria under the complex conditions of random generation and deviation penalties. They also introduce a probability-based method to determine the minimum storage capacity needed to meet a specific reliability threshold ($\alpha$).
Datasets
The study uses historical solar energy generation data from Hong Kong spanning 20 years (1993 to 2012). This data is used to create empirical cumulative distribution functions (CDFs) to model the "types" of renewable generation across different months and hours, ensuring the simulations reflect real-world weather patterns.
Key Insights and Ideas
- The "Blessing" of Uncertainty: Prevailing wisdom suggests storage is always better because it stabilizes output. However, the authors find that uncertainty can be a "buffer." Because suppliers without storage are afraid of penalties, they bid more conservatively. This reduces fierce price wars, allowing both suppliers to maintain higher profit margins.
- The Free-Rider Problem: If one supplier invests in storage, they stabilize the market price. A competitor who doesn't invest can "free-ride" on this stabilized price without bearing any of the massive storage investment costs, potentially ending up with higher net profits.
- Counter-Intuitive Market Drivers: Surprisingly, a higher penalty for supply shortfalls or a higher storage cost can sometimes lead to higher profits for suppliers. This happens because these factors force everyone to be more conservative, which paradoxically eases market competition.
Key Results
- Equilibrium Characterization: The paper successfully characterizes the conditions for pure and mixed strategy equilibria for all competitive scenarios (both have storage, one has storage, or neither has storage).
- Revenue Comparison: In the "mixed" case (one with storage, one without), the supplier with storage generally earns higher revenue, but not necessarily higher profit once investment costs are subtracted.
- Positive Profits Guaranteed: The authors prove that in a competitive market with random generation, suppliers are guaranteed strictly positive profits at equilibrium. Conversely, if both had perfectly stable outputs (via storage), their profits could drop to zero due to intense competition at low demand levels.
- Oligopoly Extension: The core insights—particularly the potential benefits of uncertainty—were shown to hold even when the model is expanded from two suppliers (duopoly) to many (oligopoly).
Limitations and Future Work
- Binary Investment Choice: The current model treats storage investment as a "yes/no" decision. Future work could allow suppliers to choose from a continuous range of storage capacities.
- Complete Smoothing Assumption: The authors assume that if a supplier has storage, they use it to completely smooth their output to the mean. In reality, a supplier might strategically choose to only partially smooth their output.
- Demand Response: The model assumes consumer demand is "perfectly inelastic" (it doesn't change based on price). Incorporating demand-side flexibility would add another layer of realism to the local energy market dynamics.
Summary updated