Generation bidding game with flexible demand

Yuquan Shan, Jayaram Raghuram, George Kesidis, Christopher Griffin, Karl N. Levitt, David J. Miller, Jeff Rowe, Anna ScaglionearXiv 2014 · arXiv

Abstract

This paper investigates the dynamics of wholesale electricity markets where consumer demand is elastic and responsive to price signals. By modeling power generators as strategic players in a noncooperative game, the study reveals that even simple grid topologies can produce a variety of stable Nash equilibria and complex, oscillating market behaviors.

Introduction

Traditional Optimal Power Flow (OPF) models typically assume static demand, where consumption remains constant regardless of price. However, the integration of smart grid technologies, such as electric vehicles and responsive appliances, has introduced significant demand elasticity. This research models a "demand-response" mechanism where consumers adjust usage based on average clearing prices, creating a strategic environment for power generators.

System Model and Methodology

The study utilizes the IEEE 9-bus system as a standardized laboratory for testing market behavior. This system includes three generators and three loads, providing a realistic framework for analyzing physical constraints and transmission limits.

The IEEE 9-bus system used as the benchmark for the generation bidding game simulation.

The problem is formulated as an iterative Bertrand-type game with the following components:

  • Generators: Strategic players setting prices to maximize net utility (revenue minus quadratic costs).
  • Independent System Operator (ISO): A neutral entity solving the Economic Dispatch problem to minimize total system costs.
  • Consumers: Modeled with a linear demand-response function where total demand decreases as prices rise.

Two strategic approaches are compared: Best-response play, where generators jump to the previous round's optimal price, and Better-response play, where generators make incremental adjustments based on the profit function's gradient.

Key Insights and Market Dynamics

Unlike classic Bertrand competition where prices often spiral to zero profit, the inclusion of strictly convex (quadratic) generation costs ensures a "desirable interior Nash equilibrium" where generators remain profitable.

Plurality of Equilibria and Stability

The research highlights that the system can settle into multiple stable Nash equilibria depending on initial bidding conditions. Interestingly, 2-player duopolies exhibited greater instability and larger price oscillations compared to 3-player configurations.

Market Jitters and Limit Cycles

A significant finding is the emergence of "limit cycles" in the Best-response strategy. Instead of converging to a single point, prices may oscillate indefinitely between two or more states. This phenomenon is highly sensitive to the "step size" of price changes in the iterative model.

Visualization of price trajectories exhibiting limit cycles, where prices oscillate between specific points instead of converging.

Results

Numerical experiments on the 3-generator system showed that the market often converges to an equilibrium where generators set nearly identical prices (approx. 3.15 units). However, this stability is fragile; excessive step sizes in the Better-response model can trigger small limit cycles rather than convergence. Furthermore, physical transmission limits were found to significantly constrain strategic bidding by capping the power allocation of competitive generators, regardless of their price.

Limitations and Future Directions

The current model employs a DC power flow approximation, which simplifies voltage and reactive power complexities. Additionally, demand is treated as a passive linear response rather than active strategic bidding by consumers. Future work aims to incorporate AC power flow models and explore the impact of strategic "prosumers" on market stability.

Summary updated

Abstract

For a simple model of price-responsive demand, we consider a deregulated electricity marketplace wherein the grid (ISO, retailer-distributor) accepts bids per-unit supply from generators (simplified herein neither to consider start-up/ramp-up expenses nor day-ahead or shorter-term load following) which are then averaged (by supply allocations via an economic dispatch) to a common "clearing" price borne by customers (irrespective of variations in transmission/distribution or generation prices), i.e., the ISO does not compensate generators based on their marginal costs. Rather, the ISO provides sufficient information for generators to sensibly adjust their bids. Notwithstanding our idealizations, the dispatch dynamics are complex. For a simple benchmark power system, we find a price-symmetric Nash equilibrium through numerical experiments.
BibTeX
@misc{shan2014generation,
  author = {Yuquan Shan and Jayaram Raghuram and George Kesidis and Christopher Griffin and Karl N. Levitt and David J. Miller and Jeff Rowe and Anna Scaglione},
  title = {{Generation bidding game with flexible demand}},
  year = {2014},
  eprint = {1408.6689},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1408.6689},
}
Figure