Multiphysics-embedded co-optimization of data centers and energy storage systems

Tools

Chen, Yi and Wang, Wenting and Pan, Li and Song, Jie and Yan, Yuejun and Hong, Qiteng and Booth, Campbell and Wang, Zhaoyang and Wang, Jianxiao (2026) Multiphysics-embedded co-optimization of data centers and energy storage systems. IEEE Transactions on Industry Applications. pp. 1-12. ISSN 0093-9994 (https://doi.org/10.1109/tia.2026.3676947)

[thumbnail of Chen-etal-2026-Multiphysics-embedded-co-optimization-of-data-centers-and-energy-storage-systems]
Preview
 Text. Filename: Chen-etal-2026-Multiphysics-embedded-co-optimization-of-data-centers-and-energy-storage-systems.pdf
Accepted Author Manuscript
License: Creative Commons Attribution 4.0 logo
Download (755kB)| Preview

Abstract

This paper proposes an aging-aware joint op-timization framework for data center-grid-energy storage systems that dynamically balances operational economy with battery lifetime preservation. Unlike traditional open-loop degradation models, the proposed method embeds the complex P2D battery degradation model into the optimiza-tion loop. A simplified surrogate model is established within the system-level scheduling, and the P2D model is used to collect input parameters for more accurate aging simula-tions. The updated aging parameters are then fed back into the surrogate model, enabling real-time correction and up-dating. The optimization problem incorporates thermal unit generation and ramping costs, state-of-charge and depth-of-discharge constraints for battery energy storage systems (BESS), and power flow limits, producing schedul-ing strategies that adapt to both macro-level grid conditions and micro-level degradation mechanisms. The proposed framework is validated on the IEEE 30-bus system with co-located data center nodes, and four cases are compared to examine the impacts of aging cost modeling and dynamic electrochemical feedback. Results show that enabling stor-age reduces total operating cost by up to 5.21% compared with the baseline, while incorporating aging cost suppresses deep cycling and mitigates battery wear. The dynamic feed-back strategy further reduces daily aging cost by 16.7% rel-ative to the static model and extends projected battery life-time by about 17%, demonstrating its long-term cost-effec-tiveness. Overall, the proposed study establishes a scalable multiphysics-embedded optimization framework that cou-ples system-level dispatch with real-time electrochemical degradation feedback, providing a practical pathway for lifecycle-oriented operation in future power systems with high data center penetration.

ORCID iDs

Chen, Yi, Wang, Wenting, Pan, Li, Song, Jie, Yan, Yuejun, Hong, Qiteng ORCID logoORCID: https://orcid.org/0000-0001-9122-1981, Booth, Campbell ORCID logoORCID: https://orcid.org/0000-0003-3869-4477, Wang, Zhaoyang and Wang, Jianxiao;
CORE (COnnecting REpositories)