Session: Session 02-05: Two Phase Cooling - II

Paper Number: 170206

170206 - Effect of Inundations on Enhanced Condenser Performance for Two-Phase Immersion Cooling Applications; A System-Based Modeling Approach 

The rapid growth of cloud computing, artificial intelligence, and big data analytics has significantly increased the energy consumption of data centres, necessitating advanced thermal management solutions. Among the emerging technologies, two-phase immersion cooling has gained significant attention due to its ability to efficiently manage the high heat flux densities generated by modern electronic components. Central to this system is the condenser, which plays a critical role in converting vaporized dielectric coolant back into liquid, ensuring efficient heat removal. This study presents a comprehensive system model in MATLAB for the design and performance prediction of enhanced-tube condensers in two-phase immersion cooling systems, focusing on the effect of key parameters such as flow rate, coolant inlet temperature and tube bundle arrangement on condenser efficiency.

The proposed system model can be used to evaluate the thermal-fluidic performance of the condenser, with a particular emphasis on the effect of condensate inundation - a phenomenon where condensate from upper tubes flows onto lower tubes, increasing the liquid layer thickness and reducing heat transfer efficiency. The MATLAB program simulates the local temperature distribution and heat transfer characteristics of a multi-pass condenser system consisting of a bundle of tubes between two headers. Water flows through these tubes, arranged in rows and columns, in a three-pass configuration, where the flow direction reverses after each pass after mixing at the headers. The inlet water temperature and flow rate are specified, and at each tube location, the local heat transfer coefficient for condensation is calculated using the Nusselt condensation model. The local heat transfer coefficient for condensation for lower rows in each pass is adjusted based on the previous row’s local heat transfer coefficient for condensation, following a scaling relation to accommodate the inundation effects. Thermal development effects impact the internal heat transfer coefficient, especially in the entry region and in the present model, Gnielinski correlation accounts for entrance effects. The energy balance equations determine the local wall temperature and local bulk water temperature at discrete segments along the tube length. The code also accounts for the mixing of water at the headers after each pass, ensuring an accurate representation of the thermal mixing process. The final outputs include the local and overall heat transfer rates, temperature distributions, and thermal performance of the condenser.

Further investigation using this system model reveals that closely packed tube bundles exacerbate inundation effects, leading to a decrease in the condensation heat transfer coefficient. To minimize these effects, we reduced the number of vertically aligned tubes and optimized tube spacing to prevent hydrodynamic and thermal boundary layer interactions.

These design improvements enhance condenser performance in two-phase immersion cooling systems. Additionally, our optimization approach refines tube arrangement, circuit configuration, and coolant flow rate, enabling the development of efficient, compact cooling solutions for high-performance data centres.

Presenting Author: Jimil M. Shah MARA Holdings, Inc.

Presenting Author Biography: Dr. Jimil Shah is a Senior Director of Immersion Cooling at MARA Holdings, Inc.