Exploring the Potential of Single-Phase Immersion Cooling Through Quantitative Thermal Performance

Author: Presenter (unnamed on stage); Matt Donahue (co-author) Collection: 2024 OCP Global Summit - Cooling Environments & Immersion

Summary

This presentation delivers the most quantitatively dense fluid benchmarking dataset presented at OCP 2024, covering 8 commercial immersion fluids tested across a comprehensive parametric matrix on a custom-built 2U test stand. The central question is whether immersion fluids are interchangeable drop-in replacements — the answer is definitively no: fluid selection in natural convection setups can account for up to 25% differentiation in case thermal resistance, with viscosity as the primary driver. The dataset covers natural convection, forced convection, immersion heat sinks, air heat sinks, flow rates from 1–20 LPM (liters per minute), bulk fluid temperatures from 5–50°C, and TTV (Thermal Test Vehicle) powers from 200W to 1.44 kW.

Key performance ceilings established: natural convection with immersion heat sink can exceed 700W TDP (thermal design power) without forced flow; forced convection reduces thermal resistance by up to 27% relative to natural convection; and switching from an air heat sink to an immersion heat sink improves thermal resistance by up to 25%. The 8 fluids tested span viscosities from 1.6 to 5.9 cSt (centistokes) at KV100 and flash points from 141°C to 235°C (ASTM D93), reflecting the regulatory tradeoff between flash point requirements in different geographic regions and the resulting viscosity constraints that impact thermal performance.

The work introduces a non-dimensional analysis direction using Reynolds numbers to characterize forced convection behavior, with the goal of validating or developing new heat transfer correlations (Nusselt, Reynolds, Prandtl number triplet) specific to immersion environments. This is positioned as foundational science that will enable the emerging Advanced Cooling Methods workstream within OCP.