Characteristics of locational uncertainty marginal price for correlated uncertainties of variable renewable generation and demands

Xin Fang, Hantao Cui, Ershun Du, Fangxing Li, Chongqing KangApplied Energy 2020 · DOI

TL;DR

This paper introduces a novel "Locational Uncertainty Marginal Price" (LUMP) model to better price the inherent uncertainty of renewable energy sources (RES) and demand in electricity markets. By using a scenario-free approach and explicitly modeling correlations, it ensures fairer cost recovery for generators and better market equilibrium, crucial for integrating more renewables.

What Problem Does This Paper Address?

The increasing integration of variable renewable energy sources (RES) like wind and solar into electricity grids presents significant challenges for market operations and grid stability. These sources are inherently unpredictable, leading to uncertainty in their power output. Current market designs often struggle to adequately price this uncertainty, leading to several issues:

  • Difficulty in obtaining real-time probability distributions: Accurately forecasting the probability distributions of RES output and demand in real-time is extremely difficult and prone to errors.
  • Market clearing inefficiencies: Models relying on numerous scenarios to capture uncertainty can become computationally burdensome, making timely market clearing challenging.
  • Uncertainty in cost recovery: For some scenarios, conventional generators might not recover their costs, creating financial risks and disincentivizing their participation.
  • Ignoring correlations: The correlation between RES output and demand uncertainty is often overlooked, which can lead to inaccurate pricing and market outcomes.

This paper aims to address these challenges by proposing a new pricing mechanism that directly accounts for uncertainty and its correlations in a more robust and efficient manner.

Methods

The core of this paper is the development of a scenario-free stochastic market-clearing model based on chance-constrained optimal power flow (CC-OPF). Instead of relying on predefined probabilistic scenarios, the model uses distributionally-robust chance constraints to represent uncertainty. This approach allows for the modeling of ambiguity sets that encompass all possible probability distributions matching certain statistical moments of the forecasting uncertainty.

Key methodological contributions include:

  • Locational Uncertainty Marginal Price (LUMP): A new pricing mechanism is derived from the Lagrangian of the CC-OPF model. This LUMP explicitly prices the uncertainty of demand and RES power, differentiating it from traditional Locational Marginal Prices (LMP) which primarily price energy.
  • Endogenous Correlation Modeling: The model inherently captures the correlation between demand and RES uncertainties within both the market-clearing and the pricing formulations.
  • Scenario-Free Approach: By avoiding the need for numerous scenarios, the model is more computationally efficient and less vulnerable to inaccuracies in scenario generation.
  • Adaptive Reserve Compensation: The LUMP model facilitates the compensation of adaptive generation reserves, which are crucial for balancing the system in real-time due to RES variability.

Datasets

The paper evaluates its proposed model using two test systems:

  • PJM 5-bus system: A modified version of the original PJM 5-bus system, used for conceptual illustration.
  • IEEE 118-bus system: A larger, more complex system used to demonstrate the model's scalability and application in a realistic scenario.

These systems are standard benchmarks in power systems research, allowing for comparison and validation of the proposed method.

Key Insights and Ideas

  • Uncertainty has a location and a price: The LUMP explicitly shows that the price of uncertainty varies by location and by the type of uncertainty (demand vs. RES), reflecting the grid's structure and the specific characteristics of the uncertainty.
  • Correlations matter for pricing: Ignoring correlations between demand and RES uncertainty can lead to inaccurate pricing. The model's ability to capture these correlations provides a more nuanced and fairer price signal.
  • Adaptive reserves are key: The model demonstrates how to price and compensate for adaptive generation reserves, which are essential for managing the real-time imbalances caused by RES. This compensation is directly linked to the uncertainty price.
  • Guaranteed cost recovery for generators: A significant insight is that the proposed LUMP model ensures that conventional generators can recover their costs, even if they don't provide flexibility services, thereby improving market stability and participation.
  • RES should manage their uncertainty: The results show that RES profitability is directly impacted by their uncertainty levels. This provides a strong incentive for RES to strategically manage their output volatility, perhaps through storage or smarter forecasting.

Key Results

  • Demonstrated Feasibility: The model was successfully applied to both a small and a large test system, showing its practical applicability.
  • Revenue Adequacy: The model ensures revenue adequacy for various market settlements, including energy, reserve services, and congestion revenue, preventing underfunding issues.
  • Market Equilibrium: The paper proves that the proposed model maintains partial market equilibrium, meaning generators have no incentive to deviate from the dispatch determined by the market.
  • Cost Recovery: Conventional generators are guaranteed to recover their costs, regardless of whether they provide adaptive reserves or not.
  • Computational Efficiency: The case studies were solved within seconds, indicating the model's computational efficiency for practical applications.
  • Sensitivity Analysis: The study showed that system generation costs increase with higher demand and RES uncertainty levels, and that LUMPs also rise accordingly.

Limitations and Future Work

  • Gaussian Distribution Assumption: While the paper discusses distributionally robust chance constraints for unknown distributions, the specific calculation of $K_\epsilon$ often relies on the Gaussian distribution assumption. Future work could explore more general distribution assumptions.
  • Real-time Market Pricing: The paper focuses on the day-ahead market clearing and pricing. The real-time market clearing for power deviations is outlined in Appendix B, but a more detailed analysis of real-time pricing mechanisms could be beneficial.
  • Model Complexity: While computationally efficient, the CC-OPF formulation can still be complex. Further research into simplifying or approximating these formulations for even larger-scale systems might be valuable.
  • Strategic Behavior: The paper assumes rational market participants. Future work could explore how strategic bidding behavior might interact with the proposed pricing mechanism.

Summary updated

BibTeX
@inproceedings{fang2020characteristics,
  author = {Xin Fang and Hantao Cui and Ershun Du and Fangxing Li and Chongqing Kang},
  title = {{Characteristics of locational uncertainty marginal price for correlated uncertainties of variable renewable generation and demands}},
  booktitle = {Applied Energy},
  year = {2020},
  doi = {10.1016/j.apenergy.2020.116064},
}
Figure