Optimization of Cryptocurrency Mining Demand for Ancillary Services in Electricity Markets

Ali Menati, Yuting Cai, Rayan El Helou, Chao Tian, Le XieProceedings of the ... Annual Hawaii International Conference on System Sciences/Proceedings of the Annual Hawaii International Conference on System Sciences 2024 · DOI

Abstract

We model the operation of a cryptocurrency mining facility with heterogeneous mining devices participating in ancillary services as an optimization problem. We propose a general formulation for the cryptominers to maximize their profit by strategically participating in ancillary services and controlling the loss of mining revenue, which requires taking into account the disparity in the efficiency of the mining machines. The optimization formulation is considered for both offline and online scenarios, and optimal algorithms are proposed to solve these problems. As a special case of our problem, we investigate cryptominers' participation in frequency regulation, where the miners benefit from their fast-responding devices and contribute to grid stability. Simulation results based on real-world Electric Reliability Council of Texas (ERCOT) traces show more than 20% gain in profit, highlighting the advantage of our proposed algorithms.

Introduction

Cryptocurrency mining, particularly Proof-of-Work (PoW), is often criticized for its massive energy footprint, consuming between 0.9% and 1.7% of total U.S. electricity. However, unlike traditional data centers or industrial plants, crypto mining is uniquely "interruptible." Miners can ramp their power consumption up or down almost instantly without losing data or damaging equipment. This flexibility presents a significant opportunity for grid stabilization through "ancillary services"—grid-balancing programs that help maintain electricity stability.

The challenge lies in strategic decision-making. A mining facility often owns a mix of hardware with varying efficiencies. When the grid operator requests a load reduction, the miner must decide which machines to shut down to minimize lost revenue and how much capacity to commit to different grid programs to maximize total profit. This paper provides a mathematical framework to bridge crypto-economics with electricity market requirements.

Methodology

The authors model the operation as a two-level optimization problem. First, they establish an optimal deployment strategy, proving that to minimize the loss of mining revenue, a facility should always shut down its least efficient machines first. This is formalized into a "critical type" ($k_c$) threshold.

Second, for ancillary service profile selection, the authors propose two main approaches:

  1. Stochastic Subgradient Descent: Used when the statistical distribution of grid signals and prices is known (the offline scenario).
  2. Online Gradient Descent (OGD): A "no-regret" learning algorithm for real-world scenarios where prices and grid needs are unpredictable.
Stochastic Gradient Descent path towards the optimal portfolio of ancillary service programs.

The framework specifically addresses Frequency Regulation, deriving closed-form solutions for "reg-up" (reducing load) and "reg-down" (increasing load) scenarios.

Key Insights: Correlation and Heterogeneity

A critical finding is the significant negative correlation (-0.757) between crypto mining load and system-wide net load during summer peaks in Texas. As shown in the figure below, when the grid is most stressed, miners naturally scale back, making them ideal partners for grid stability.

System-wide demand vs. total mining demand in ERCOT, showing the negative correlation during peak periods.

Furthermore, the study emphasizes that hardware heterogeneity matters. The disparity in machine efficiency creates a "buffer" where older machines serve as the primary tools for grid response, preserving the uptime of the most profitable, efficient hardware. The authors demonstrate that the expected cost function for these services is piecewise linear and convex, guaranteeing that standard optimization algorithms will converge to the true global optimum.

Results and Discussion

The study validates its models using high-fidelity, real-world data from ERCOT and Bitcoin market history. Key results include:

  • 20% Profit Boost: In simulations, the proposed optimization algorithms outperformed a standard "50-50" split strategy by more than 20%.
  • Real-Time Learning: The online optimization (OGD) showed "sublinear regret," meaning the algorithm quickly learned the optimal bidding strategy within approximately 70 days without prior market knowledge.
  • Grid Support: Miners effectively participated in both "reg-up" and "reg-down" programs, facilitating the integration of volatile renewable energy sources like wind and solar.

Conclusion

This research demonstrates that cryptocurrency mining can be a valuable asset for power grid stability. By leveraging the interruptible nature of mining and the heterogeneity of hardware, miners can significantly increase their profitability while providing essential flexibility to the grid. Future work may address the "price-maker" potential of large-scale facilities and the long-term physical impact of frequent power cycling on mining hardware.

Summary updated

Abstract

We model the operation of a cryptocurrency mining facility with heterogeneous mining devices participating in ancillary services as an optimization problem.We propose a general formulation for the cryptominers to maximize their profit by strategically participating in ancillary services and controlling the loss of mining revenue, which requires taking into account the disparity in the efficiency of the mining machines.The optimization formulation is considered for both offline and online scenarios, and optimal algorithms are proposed to solve these problems.As a special case of our problem, we investigate cryptominers' participation in frequency regulation, where the miners benefit from their fast-responding devices and contribute to grid stability.Simulation results based on real-world Electric Reliability Council of Texas (ERCOT) traces show more than 20% gain in profit, highlighting the advantage of our proposed algorithms.
BibTeX
@inproceedings{menati2024optimization,
  author = {Ali Menati and Yuting Cai and Rayan El Helou and Chao Tian and Le Xie},
  title = {{Optimization of Cryptocurrency Mining Demand for Ancillary Services in Electricity Markets}},
  booktitle = {Proceedings of the ... Annual Hawaii International Conference on System Sciences/Proceedings of the Annual Hawaii International Conference on System Sciences},
  year = {2024},
  doi = {10.24251/hicss.2024.369},
}
Figure