Technology Category
Computer Vision: 3D Computer VisionMachine Learning: Hardware-aware MLKnowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
Fast 3D Surrogate Modeling for Data Center Thermal Management
🤖 AI Summary
Real-time prediction of 3D thermal fields in data centers remains challenging due to the high computational cost and expert-dependent meshing/boundary setup of traditional CFD solvers, which preclude millisecond-scale responsiveness. Method: We propose an end-to-end vision-driven 3D surrogate modeling framework that operates directly on voxelized spatial data—eliminating manual mesh generation—and supports cross-data-center generalization. The architecture integrates 3D CNN U-Net, 3D Fourier Neural Operator, and 3D Vision Transformer modules, conditioned on server workload, fan speeds, and HVAC setpoints. Contribution/Results: Our model achieves <10 ms inference latency, <0.8°C mean temperature prediction error, and >92% hotspot detection accuracy; real-world deployment reduces PUE by 7% and corresponding carbon emissions. It accelerates thermal simulation by 20,000× over CFD while enabling the first high-fidelity, deployable, and generalizable real-time thermal-aware closed-loop control.
Technology Category
Application Category
📝 Abstract
Reducing energy consumption and carbon emissions in data centers by enabling real-time temperature prediction is critical for sustainability and operational efficiency. Achieving this requires accurate modeling of the 3D temperature field to capture airflow dynamics and thermal interactions under varying operating conditions. Traditional thermal CFD solvers, while accurate, are computationally expensive and require expert-crafted meshes and boundary conditions, making them impractical for real-time use. To address these limitations, we develop a vision-based surrogate modeling framework that operates directly on a 3D voxelized representation of the data center, incorporating server workloads, fan speeds, and HVAC temperature set points. We evaluate multiple architectures, including 3D CNN U-Net variants, a 3D Fourier Neural Operator, and 3D vision transformers, to map these thermal inputs to high-fidelity heat maps. Our results show that the surrogate models generalize across data center configurations and achieve up to 20,000x speedup (hundreds of milliseconds vs. hours). This fast and accurate estimation of hot spots and temperature distribution enables real-time cooling control and workload redistribution, leading to substantial energy savings (7%) and reduced carbon footprint.
Problem
Research questions and friction points this paper is trying to address.
Developing fast 3D thermal modeling for real-time data center temperature prediction
Replacing computationally expensive CFD solvers with vision-based surrogate models
Enabling real-time cooling control to reduce energy consumption and carbon emissions
Innovation
Methods, ideas, or system contributions that make the work stand out.
Vision-based surrogate modeling on 3D voxels
Evaluating 3D CNNs, Fourier Neural Operators, Transformers
Achieving 20,000x speedup for thermal prediction
🔎 Similar Papers
No similar papers found.
