Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Fast 3D Surrogate Modeling for Data Center Thermal Management
Soumyendu Sarkar, Antonio Guillen-Perez, Zachariah J Carmichael, Avisek Naug, Refik Mert Cam, Vineet Gundecha, Ashwin Ramesh Babu, Sahand Ghorbanpour, Ricardo Luna Gutierrez
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| apply the same 3-channel voxel input design (power, CRAC set point, fan speed) across U-Net variants, FNO and Swin UNETR | 3D surrogate models · Data center 3D temperature test set | Mean AE (C): Mean AE 1.89 to 5.98 across architectures | Improved |
| use the ML surrogate instead of the CFD solver | data center thermal simulation · Data center test configuration | inference time: 35 minutes -> 100 ms | Improved |
| use U-Net++ instead of plain 3D U-Net | 3D surrogate · Data center 3D temperature test set | Mean AE (C): 1.89 -> 5.98 | Hurt |
| add residual blocks to 3D U-Net | 3D U-Net · Data center 3D temperature test set | Mean AE (C): 1.89 -> 2.15 | Hurt |
| add squeeze-and-excitation blocks to Residual-UNet | Residual-UNet · Data center 3D temperature test set | Top-t AE (C): 2.54 -> 2.84 | Hurt |
| use FNO instead of 3D U-Net | 3D surrogate · Data center 3D temperature test set | 3D SSIM: SSIM 0.8171 -> 0.8260; Mean AE 1.89 -> 2.15 | Mixed |
| optimize server workload placement with a genetic algorithm guided by the surrogate | data center workload distribution · Data center hotspot minimization | hotspot reduction: 7.70% reduction vs random baseline | Improved |
Closest papers by meaning
- Data Center Cooling System Optimization Using Offline Reinforcement Learning (ICLR 2025 · similarity 0.84)
- Carbon Footprint Reduction for Sustainable Data Centers in Real-Time (AAAI 2024 · similarity 0.83)
- Hierarchical Multi-Agent Framework for Carbon-Efficient Liquid-Cooled Data Center Clusters (AAAI 2025 · similarity 0.82)
- LC-Opt: Benchmarking Reinforcement Learning and Agentic AI for End-to-End Liquid Cooling Optimization in Data Centers (NeurIPS 2025 · similarity 0.82)
- DCcluster-Opt: Benchmarking Dynamic Multi-Objective Optimization for Geo-Distributed Data Center Workloads (NeurIPS 2025 · similarity 0.82)
- NeuralBES: A Differentiable, Control-Aware Emulator for Scalable Building Energy Modeling (NeurIPS 2026 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
