When Intel was looking for a thermal partner to help them develop the next generation cooling solution for their high performance computing products, they knew both the company and the technology they needed to work with. Advanced Cooling Technologies has been working on pumped two phase cooling for more than a decade and is a recognized technology leader. Pumped two phase cooling essentially uses the vapor produced from the working fluid’s phase transition to effectively and efficiently remove waste heat. To accommodate higher power density processors, Intel recognized the need for a higher performance and robust cooling solution.
The major challenges ACT had to overcome in developing a solution were the increased heat dissipation requirements within the very tight packaging footprint of Intel’s Coolant Distribution Unit (CDU). The CDU was designed for a total heat load of 4kW and for over 50,000 individual nodes in a 1U blade configuration, with a maximum weight of 12 lb. ACT utilized its wide array of thermal simulation tools to design and then prototype a pumped two phase cooling solution to meet these requirements. The resulting prototype, including the cold plate pictured above, used the available air cooling in the existing system, while increasing heat dissipation capability with an additional liquid cooling system. ACT’s solution provided even more space savings than originally anticipated, even with the addition of liquid cooling after the heat exchanger.
Dr. Devdatta Kulkarni, ACT’s main collaborator at Intel, authored a white paper on the project, “Experimental Study of Two-Phase Cooling to Enable Large-Scale System Computing Performance,” presented at ITHERM 2018. The paper includes compelling test result data, and an acknowledgement of the team at ACT for their contribution.
Advanced Cooling Technologies is a recognized leader in thermal management solutions, including pumped two phase cooling. When faced with challenging thermal requirements, ACT is the one you want to turn to. Contact the thermal experts today to discuss your application.
Single-phase cold plates are commonly used to dissipate waste heat from electronics components. To accommodate the varying footprint and output power of electronics components available, cold plates are being developed with custom internal structures to maximize the efficiency of heat rejection from the electronics components. Cold plates with these custom internal structures require thermal and hydraulic performance evaluations to ensure they meet target operating conditions before being deployed in the end application. Advanced Cooling Technologies, Inc. developed the custom test apparatus, shown in Figure 1, for a customer that enables the thermal and hydraulic performance of multiple cold plate designs to be evaluated and compared under equivalent operating conditions in order to identify the optimum cold plate design.
This test apparatus was designed with an integrated data acquisition (DAQ) panel to read and record instrument responses from thermocouples and pressure transducers, easy connection to an external chiller, and a custom manifold that enabled cold plates of various internal geometries to be connected and removed from the test apparatus. The test apparatus was outfitted with resistance temperature detectors (RTDs) at the inlet and outlet ports and differential pressure transducers such that the temperature rise and pressure drop of the cold plate can be measured.
Cold plates to be tested are prepared with heat sources and outfitted with surface mount thermocouples and incorporated into the test apparatus. The heat sources are energized with an external power supply and once steady-state is achieved, the thermal resistance of the cold plate is calculated from Equation 1 where Rth is the thermal resistance of the cold plate, Ts is the surface temperature of the heat source, Tout is the outlet temperature of the cold plate, Tin is the inlet temperature of the cold plate, is the mass flow rate of coolant, and Cp is the specific heat capacity of the coolant.
For any given cold plate configuration, there is a pumping power required to remove a certain amount of heat. The coefficient of performance (COP) of a cold plate is a dimensionless number that provides a measure of the amount of pumping power required to remove a given amount of heat and provides an equivalent comparison for different cold plate configurations. Thus, in addition to the thermal resistance of the cold plate, the COP of the cold plates can be evaluated from Equation 2, where ΔP is the pressure drop of the cold plate and is the volumetric flow rate of coolant through the cold plate.
Thus, this single test apparatus, designed, developed, and fabricated by ACT, can provide a full thermal and hydraulic performance evaluation to enable the down selection of the optimal cold plate for a given set of operating conditions.
One of ACT’s customers was interested in using spray and jet impingement cooling to remove high heat fluxes (1000W/cm2) from the surface of a CPU. A test section was designed capable of taking single jet impingement heat transfer data using Intel’s supplied 34980A thermal test platform. Figure 5 shows an overall view of the test section with the Sil-clamp bolted to the base plate. The Thermal Test Vehicle is also shown in place. The 8 bolts hold the Sil-clamp in a precise location above the base plate and also provide even pressure around the O-rings to the acrylic cylinder. Figure 17 shows the flow exiting the nozzles at the projected angle as well as the projected impact of the spray with the die. In the design process the 3-D CAD software was used to verify no flow interactions. The test apparatus was designed to test at various inlet temperatures from -10°C to 125°C and work with three different fluids (HFE-7000, Methanol, and Water). The apparatus successfully measured heat transfer coefficients at various locations along the die, providing the customer with the precise measurements required for their application; see Figure 5.
Read the full technical write-up on this project
With support from the National Science Foundation, ACT has developed a pumped two-phase cooling system for high heat flux electronic components and laser diodes and is now working on packing the system in a compact, user friendly, stand-alone platform. The two-phase system efficiently handles fluxes on ~ 300-500W/cm2, requires little pumping power, maintains device temperatures below operating limits and provides for a high degree of isothermality over the heated surface. Meeting these requirements is important in many applications including lasers whose emission wavelengths are temperature-dependent.
Unlike single-phase cooling, two-phase systems take advantage of the phase change (latent heat) of the coolant which enables it to handle higher heat fluxes for a given heat load. Two-phase cooling systems can however be more complex and prone to flow and thermal instabilities. As such, techniques to effectively manage instabilities have been developed and characterized. These include the application of engineered microporous coatings on the heated surface(s), which enhance boiling performance by increasing the number of nucleation sites together with a capillary-driven resupply of the coolant to the heated surface, which prevents or postpones dry-out (extends CHF).
A schematic of a pumped two-phase cooling system is shown in Figure 1. Key components include a pump, preheater, surge tank, evaporator (the heat sink), condenser and accumulator. The surge tank and the preheater differentiate this system from a traditional liquid cooling loop. The surge tank consists of vapor and liquid at saturation; by controlling the pressure in the tank, the saturation (boiling) temperature of the working fluid can be controlled. The preheater heats the subcooled liquid exiting the condenser to a temperature close to the saturation temperature before it enters the evaporator. This is important as boiling heat transfer is most efficient at saturation (minimal subcooling).
Figure 1: Schematic of a pumped two-phase cooling system
Also shown in Figure 2 is a representative copper minichannel heat sink coated with a microporous coating.
Figure 2: Minichannel heat sink with porous sintered power coating
The pumped two-phase cooling system shown in Figure1 was fabricated and evaluated. The heat transfer coefficient (HTC) [W/m2K] and the Incipient Wall Superheat [K] were determined as a function of the input heat flux and coolant mass flux [kg/m2s] using refrigerant R134a and others. Representative results for the HTC for coated and uncoated minichannel heat sinks are shown in Figure 3. Clearly, the HTC is higher for the coated heat sinks and the CHF is extended.
Figure 3: Heat Transfer Coefficient (HTC) associated with Minichannel Heat Sinks noting the increase in CHF for coated heat sinks.
The incipient wall superheat is also shown in Figure 4 as a function of input heat flux on coated and uncoated heat sinks. Again, the microporous coating enhances the thermal performance of the heat sinks as evident by lower values of incipient wall superheat; in other words, a heat-generating device mounted on a two phase heat sink with microporous coating will be maintained at a lower temperature compared to one that is mounted on an identical uncoated heat sink.
Figure 4: Incipient wall superheat decreases with the application of microporous coatings
In addition to improving the heat transfer, the thin copper coatings help suppress two-phase flow instabilities. Figure 5 shows flow instabilities with a bare copper cold plate, where the vapor flow occasionally reverses and enters the input plenum. In contrast, instabilities are suppressed in the cold plate with a thin porous coating, see Figure 6.
Figure 5. High heat fluxes with uncoated mini- and micro-channel systems can have unstable flow. Note the intermittent vapor flow back into the inlet plenum.
Figure 6. The maximum heat flux in two-phase systems can be increased by adding a thin porous layer. An additional benefit is that the flow becomes more stable.
In short, pumped two phase cooling systems have been developed. Flow and thermal stability issues were well managed with the use of porous coatings, which increase the heat transfer coefficient and extend the CHF. Pumping power required is minimal and the application of the coating does not increase the pressure drop in a measurable way as the coating thickness is very small compared to the channel dimensions. Additional work on flow boiling heat transfer in minichannel heat sinks is ongoing with a focus on maximizing performance and understanding the parameters (i.e., the coating properties – thickness for a given particle size, etc.) that affect performance for specific applications.
For more information on Pumped Two Phase Cooling, contact The Thermal Experts at ACT.