SPATIOTEMPORAL GRAPH PHYSICS-INFORMED NEURAL NETWORKS FOR DATA CENTER THERMAL MODELING
DOI:
https://doi.org/10.37943/VRFC6399%20Keywords:
physics-informed machine learning, data center, thermal modeling, predicationAbstract
Accurate thermal modeling is the precondition for improving energy efficiency and operational reliability of data centers (DCs). However, airflow circulation and workloads cause the temperature field of DCs to exhibit complex spatiotemporal variations. Conventional data-driven models formulate temperature prediction as a multivariate time-series task, without explicitly accounting for spatial heat interactions. In comparison, physics-based methods require detailed geometric information and high computational resources. To address these limitations, we propose a spatiotemporal graph physics-informed neural network (SGPINN) framework for multi-node and multi-step DC temperature prediction. Inspired by resistance-capacitance thermal modeling, SGPINN represents the thermal field as a graph, where graph aggregation captures spatial thermal interactions, and a physical residual combines temperature prediction with graph-guided thermal propagation. The temporal temperature derivative is obtained through automatic differentiation, while learnable equivalent thermal capacitance and heat-flow terms are introduced to characterize heterogeneous node dynamics without requiring directly measured physical parameters. A learnable multi-task strategy is adopted to adaptively balance the data loss and physics loss. The model is evaluated using both a real-world DC dataset and a simulated DC dataset and compared with four representative baseline models. SGPINN achieves the best overall performance on both datasets, with average MAEs of 0.1097 °C and 0.3043 °C, respectively. In the real-world case, it reduces average MAE and RMSE by 3.57% and 3.58% compared with APINN, while ablation results verify the contributions of graph modeling, physical constraints, and adaptive loss weighting. These findings suggest that SGPINN improves predictive accuracy while providing a physics-guided framework for DC thermal modeling.
References
Kumar, A., & Pindoriya, N. M. (2026). Toward sustainable data center operation: A review on existing infrastructures, integrated smart energy management frameworks, and future perspectives. Renewable and Sustainable Energy Reviews, 230, 116664. https://doi.org/10.1016/j.rser.2025.116664
Bogmans, C., Ganpurev, G., Gomez-Gonzalez, P., Melina, G., Pescatori, A., & Thube, S. (2026). Power hungry: How AI will drive energy demand. Energy Economics, 158, 109278. https://doi.org/10.1016/j.eneco.2026.109278
Hao, Y., Zhou, H., Tian, T., Zhang, W., Zhou, X., Shen, Q., Wu, T., & Li, J. (2025). Data centers waste heat recovery technologies: Review and evaluation. Applied Energy, 384, 125489. https://doi.org/10.1016/j.apenergy.2025.125489
Wang, N., Guo, Y., Huang, C., Tian, B., & Shao, S. (2025). Multi-scale collaborative modeling and deep learning-based thermal prediction for air-cooled data centers: An innovative insight for thermal management. Applied Energy, 377, 124568./https://doi.org/10.1016/j.apenergy.2024.124568.
Wu, L., Mao, R., Wu, H., Zhang, J., Li, C., Yi, X., Huang, G., & Zhang, Z. (2024). Optimization of airflow organization in fan-wall data center via baffles. Applied Thermal Engineering, 244, 122745. https://doi.org/10.1016/j.applthermaleng.2024.122745.
Wang, Y., Bai, X., Fu, Y., Tang, Y., Jin, C., & Li, Z. (2023). Field experiment and numerical simulation for airflow evaluation in a data center with row-based cooling. Energy and Buildings, 294, 113231. https://doi.org/10.1016/j.enbuild.2023.113231.
He, W., Zhang, J., Li, H., Liu, S., Wang, Y., Lv, B., & Wei, J. (2022). Optimal thermal management of server cooling system based cooling tower under different ambient temperatures. Applied Thermal Engineering, 207, 118176. https://doi.org/10.1016/j.applthermaleng.2022.118176.
Tang, Y.-Q., Fang, W.-Z., Zheng, C.-Y., & Tao, W.-Q. (2025). Applications of POD-based reduced order model to the rapid prediction of velocity and temperature in data centers. Applied Thermal Engineering, 263, 125310. /https://doi.org/10.1016/j.applthermaleng.2024.125310.
Zhao, J., Chen, Z., Liu, D., Li, H., Li, Y., & Yang, Z. (2025). The short-term prediction method for the characteristic temperature of data center server rooms based on the information entropy property. Applied Thermal Engineering, 280, 128396. https://doi.org/10.1016/j.applthermaleng.2025.128396.
Fang, L., Xu, Q., Li, S., Xia, Y., & Chen, Q. (2024). Temperature prediction in data center combining with deep neural network. Applied Thermal Engineering, 244, 122571. https://doi.org/10.1016/j.applthermaleng.2024.122571.
Szeliga, W., Oleksiak, A., Roszak, R., & Górzeński, R. (2026). CFD4DC: Automated CFD framework for heat transfer analysis of data centers. Applied Thermal Engineering, 283, 128982. https://doi.org/10.1016/j.applthermaleng.2025.128982.
Meng, C., Griesemer, S., Cao, D., Seo, S., & Liu, Y. (2025). When physics meets machine learning: a survey of physics-informed machine learning. Machine Learning for Computational Science and Engineering, 1(1), 20. https://doi.org/10.1007/s44379-025-00016-0.
Chen, D., Chui, C.-K., & Lee, P. S. (2025). Adaptive physically consistent neural networks for data center thermal dynamics modeling. Applied Energy, 377, 124637. https://doi.org/10.1016/j.apenergy.2024.124637.
Le, D. Van, Liu, Y., Wang, R., Tan, R., & Ngoh, L. H. (2022). Air Free-Cooled Tropical Data Center: Design, Evaluation, and Learned Lessons. IEEE Transactions on Sustainable Computing, 7(3), 579–594. https://doi.org/10.1109/TSUSC.2021.3132927.
Mao, M., & Cui, D. (2026). Cooling load prediction method for hybrid-cooled data centers based on hybrid neural network model. Energy and Buildings, 358, 117219. https://doi.org/10.1016/j.enbuild.2026.117219.
Zhou, Y., Wei, F., Li, S., Wang, Z., Liu, J., & Yu, D. (2025). Data center load modeling through optimal energy consumption characteristics: A path to simultaneously enhance energy efficiency and demand response quality. Applied Energy, 393, 126095. https://doi.org/10.1016/j.apenergy.2025.126095.
Rodríguez Jara, E. Á., Sánchez de la Flor, F. J., Álvarez Domínguez, S., Molina Félix, J. L., & Salmerón Lissén, J. M. (2016). A new analytical approach for simplified thermal modelling of buildings: Self-Adjusting RC-network model. Energy and Buildings, 130, 85–97. https://doi.org/10.1016/j.enbuild.2016.08.039.
Wang, Z., Chen, Y., & Li, Y. (2019). Development of RC model for thermal dynamic analysis of buildings through model structure simplification. Energy and Buildings, 195, 51–67. https://doi.org/10.1016/j.enbuild.2019.04.042.
Yang, J., Wang, H., Cheng, L., Gao, Z., & Xu, F. (2024). A review of resistance–capacitance thermal network model in urban building energy simulations. Energy and Buildings, 323, 114765. https://doi.org/10.1016/j.enbuild.2024.114765.
Joe, J., Im, P., Cui, B., & Dong, J. (2023). Model-based predictive control of multi-zone commercial building with a lumped building modelling approach. Energy, 263, 125494. https://doi.org/10.1016/j.energy.2022.125494.
Wang, Y., Zhang, Y., Nörtershäuser, D., Le Masson, S., & Menaud, J.-M. (2022). Model and data driven transient thermal system modelings for contained data centers. Energy and Buildings, 258, 111790. https://doi.org/10.1016/j.enbuild.2021.111790.
Zhang, B., Liao, W., Li, X., & Li, Z. (2025). Rapid prediction of air temperature distribution in data center via feature-integrated machine learning. Energy and Buildings, 342, 115891. https://doi.org/10.1016/j.enbuild.2025.115891.
de Amorim, L. B. V, Cavalcanti, G. D. C., & Cruz, R. M. O. (2023). The choice of scaling technique matters for classification performance. Applied Soft Computing, 133, 109924. https://doi.org/10.1016/j.asoc.2022.109924.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Articles are open access under the Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Authors who publish a manuscript in this journal agree to the following terms:
- The authors reserve the right to authorship of their work and transfer to the journal the right of first publication under the terms of the Creative Commons Attribution License, which allows others to freely distribute the published work with a mandatory link to the the original work and the first publication of the work in this journal.
- Authors have the right to conclude independent additional agreements that relate to the non-exclusive distribution of the work in the form in which it was published by this journal (for example, to post the work in the electronic repository of the institution or publish as part of a monograph), providing the link to the first publication of the work in this journal.
- Other terms stated in the Copyright Agreement.