Integrating Data Center Flexibility into Power System Operations: A Framework for Temporal Redispatch
Abstract
Rapid growth in data center electricity demand, driven by artificial intelligence (AI) training and inference, is expected to increase stress conditions for power grids. Operational flexibility from data centers offers a near-term lever to relieve such stresses. Existing models of data center flexibility tend to bundle economic incentives with physical constraints, obscuring the tradeoff between economic willingness to shift load and technical limitations to do so. This paper proposes a temporal-redispatch focused framework that decouples data center demand flexibility into two independently parameterized dimensions: (1) systematic flexibility, capturing the economic environment through penalties for temporal shifting of data centers' scheduled demand, and (2) operational flexibility, capturing the physical envelope through the demand shifting time window and the limit on shifted data center demand. The framework relies on a direct current optimal power flow (DC-OPF) linear program which enables annual parametric analysis at hourly resolution. The framework is studied using a synthetic grid mimicking the Texas transmission system populated with the 2026 Adjusted Forecast from ERCOT (Electric Reliability Council of Texas). Results show that operational flexibility imposes a physical lower bound on unsatisfied demand that no economic incentive can overcome; longer shifting windows and higher limits on shifted data center demand reduce operational cost with different impacts on grid utilization; and, temporal redispatch shaves system-wide net load peaks while partially compensating for localized transmission bottlenecks that emerge under modeled coal retirement scenarios.
Related
- A novel machine learning-based framework for integrating scheduling and control in continuous multiproduct chemical process operations
- Cognitive Warfare, Hybrid Pressure, and Sovereign Resilience: An Operations Research Framework Applied to the Iranian Case (2017--2026)
- An LLM-assisted real-time simulation framework for demand-responsive transport operations
- Integrating Ultraviolet and Solar Radiation Monitoring Data with an MMS-GRU Framework for Multi-Scale Dengue Forecasting
- A Framework Integrating the Dynamic Stiffness Matrix with Physics-Informed Neural Networks for Solving Eigenvalue Problems and Analysing Dynamic Response
BibTeX
@misc{li2026integrating,
title = {{Integrating Data Center Flexibility into Power System Operations: A Framework for Temporal Redispatch}},
author = {Huansong Li and Erhan Kutanoglu and Michael Baldea and Ilias Mitrai},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.02228},
eprint = {2611.02228},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.02228},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Integrating Data Center Flexibility into Power System Operations: A Framework for Temporal Redispatch
AU - Huansong Li
AU - Erhan Kutanoglu
AU - Michael Baldea
AU - Ilias Mitrai
PY - 2026
DA - 2026/11/01
PB - arXiv
DO - 10.48550/arxiv.2611.02228
UR - https://arxiv.org/abs/2611.02228
AB - Rapid growth in data center electricity demand, driven by artificial intelligence (AI) training and inference, is expected to increase stress conditions for power grids. Operational flexibility from data centers offers a near-term lever to relieve such stresses. Existing models of data center flexibility tend to bundle economic incentives with physical constraints, obscuring the tradeoff between economic willingness to shift load and technical limitations to do so. This paper proposes a temporal-redispatch focused framework that decouples data center demand flexibility into two independently parameterized dimensions: (1) systematic flexibility, capturing the economic environment through penalties for temporal shifting of data centers' scheduled demand, and (2) operational flexibility, capturing the physical envelope through the demand shifting time window and the limit on shifted data center demand. The framework relies on a direct current optimal power flow (DC-OPF) linear program which enables annual parametric analysis at hourly resolution. The framework is studied using a synthetic grid mimicking the Texas transmission system populated with the 2026 Adjusted Forecast from ERCOT (Electric Reliability Council of Texas). Results show that operational flexibility imposes a physical lower bound on unsatisfied demand that no economic incentive can overcome; longer shifting windows and higher limits on shifted data center demand reduce operational cost with different impacts on grid utilization; and, temporal redispatch shaves system-wide net load peaks while partially compensating for localized transmission bottlenecks that emerge under modeled coal retirement scenarios.
ER -