Advanced Two-Phase Cooling System for Modular Power Electronics

Kuan-Lin Lee[1], Sai Kiran Hota[2], Andrew Lutz[3] and Srujan Rokkam[4]

Advanced Cooling Technologies, Inc, Lancaster, PA, 17601

Developing standardized electronics cooling system components is important to reduce system capital costs and reduce the operations and maintenance complexities. As such, it is important to develop components that can handle high heat loads with less temperature drop between the electronics and the heat sink. Two-phase thermal heat spreaders provide an excellent alternative to conventional conduction-based heat spreaders. Under a Small Business Innovation Research program, Advanced Cooling Technologies, Inc. (ACT) developed a pulsating heat pipe (PHP)-based heat spreader for advanced modular power electronics and compared it with ACT’s high conductivity (HiK^™) plate. The features of the test components are in accordance with modular electronics standards for space power systems (MESSPS) and applicable across various space missions. Test results showed, that both heat spreaders (HiK™ plate and PHP) can reduce the thermal resistance by half compared to an aluminum plate. The unique feature of the PHP is that the capillary channels extend onto the sharp stepped interface, which is not feasible with HiK™ plate. This improves the thermal conductance across the stepped plane between the heat spreader and the card retainer. ACT’s Isothermal Card Edge (ICE-Lok^®) was used as the card retainer. Both PHP and Hi-K plates are able to operate against gravity orientation. Also, PHP was found to be up to 10.4% lighter than the conventional card by 16 grams.

Nomenclature

Abbreviations

ACT                =    Advanced Cooling Technologies, Inc.

AMPS             =    Advanced Modular Power systems

HiK™ plate =    High conductivity card

ICE-Lok^®    =    Isothermal Card Edge

PHP                 =    Pulsating Heat Pipe

SBIR               =    Small Business Innovation Program

MESSPS         =    Modular Electronics Standard for Space Power Systems

I.     Introduction

Standardizing electronics cooling systems offers many advantages such as reducing the costs of component development and unique spare parts, easily repairing and replacing parts etc. The Advanced Exploration System Modular Power System (AMPS) project aims to construct future power control architecture of spacecraft with standardized, modularized, and interchangeable components, including electronics and auxiliary thermal management components. As power electronics evolve, the standardized cooling system must handle high heat flux. Conventional cooling system with conduction only aluminum plate heat spreader is limited by the material thermal conductivity. A recent investigation by Colozza and Gardner showed an electronics module with the conventional heat spreader can only deliver up to 16 W per card at a maximum allowable operating temperature of 44 °C [i]. The thermal resistance was found to be around 1.5 °C/W, where the heat spreader thermal resistance contributed to about half the total thermal resistance. Replacing the conventional heat spreader with advanced two-phase heat spreaders can significantly improve the heat carrying capability by reducing the thermal resistance of the system.

Under SBIR Phase I Program for NASA, two different two-phase heat spreaders: ACT’s HiK™ plate and pulsating heat pipe (PHP) were developed and tested as superior alternatives to a conduction-only aluminum plate for electronics (3U card) cooling. ACT’s HiK™ plate is essentially copper-water heat pipes embedded into an aluminum plate [ii]. HiK™ plate can have an effective thermal conductivity of 600 W/m-K to 1200 W/m-K, which is 3 to 6 times more than an aluminum plate. HiK™ plate can also handle up to a heat flux of 60-70 W/cm^{2 [iii]}. Pulsating heat pipe (PHP) on the other hand, uses oscillation/ pulsation of the saturated working fluid within a serpentine capillary to transfer heat^[iv]. The heat transport capability is limited by the working fluid and geometry but could be well in excess of 10 W/cm² up to 50 W/cm² [v]^,[vi]. In addition to the heat spreader, it was identified that Wedge-Lok is the standard solution to mechanically interface the electronic card enclosure with the chassis of the module. ACT’s ICE-Lok^® can replace COTS Wedge-Lok as it provides an additional heat transfer path which in turn reduces the thermal resistance across the interface joint by more than 30% [vii].

This manuscript describes the development, testing, and performance comparison of two-phase heat spreader components and conduction-only aluminum plate heat spreader (baseline) for 3U card electronics cooling system. During the testing, ACT’s ICE-Lok^® was used as the card retainer instead of COTS Wedge-Lok. The geometrical features of the fabricated components are in accordance with modular electronics standards for space power systems (MESSPS). The developed system components can directly benefit various missions such as Gateway-Artemis, Lunar and Mars Habitats, Solar Electric Propulsions, Electrified Aircraft Propulsions, and future ISS missions like Orbiting Habitats, etc.

[1] Lead Engineer, Advanced Cooling Technologies, Inc., 1046 New Holland Ave., Lancaster, PA 17601

[2] R&D Engineer II, Advanced Cooling Technologies, Inc., 1046 New Holland Ave., Lancaster, PA 17601

[3] R&D Engineer III, Advanced Cooling Technologies, Inc., 1046 New Holland Ave., Lancaster, PA 17601

[4] R&D Manager Advanced Cooling Technologies, Inc., 1046 New Holland Ave., Lancaster, PA 17601

Figure 1. Advanced cooling system components for modular power electronics

Under an SBIR program, ACT developed an advanced cooling system for electronics card (3U size). Two-phase-based heat spreaders based on two different heat pipe technologies, ACT’s HiK™ plate and PHP were tested as part of the proposed cooling system for the modular power electronics. ACT’s ICE-Lok^® was used as an enhanced isothermal card edge retainer as it provides an additional heat transfer path over COTS Wedge-Lok. The schematic of the system is shown in Figure 1. The heat transfer path from the electronics card source to the heat sink (coolant) is represented by red arrows.

A mock cooling system was constructed to test the proposed cooling system according to modular electronics standard for space power systems (MESSPS)[i]. The geometric details of the electronics cooling system are shown in Figure 2. The card is enclosed by the heat spreader, which provides primary heat transfer path, and with a backplane on the backside of the card.  The backplane was a standard 1/8” (3.2 mm) thick solid aluminum plate. The heat spreader and the backplane interface with the chassis rails (or teeth) by means of an ICE-Lok^®. A cold plate (aluminum plate with embedded copper tubes) was attached on the top and the bottom side of the chassis to provide a uniform heat sink temperature. Water at standard 22 °C was used as the heat transfer fluid. A 2-inch x 2-inch polyimide film heater was used to simulate heat generation in the electronics cards.

I.     Design and Development of HiK™ plate and PHP

Based on the chosen film heater configuration, a HiK™ plate with heat pipes was designed and fabricated according to ACT’s standard fabrication procedure. There were six copper water heat pipes symmetrically embedded onto the aluminum base plate.

Figure 2. Geometric details of the modular power electronics cooling system (dimensions in inches)

The schematic of the HiK™ plate heat spreader is shown in Figure 3 (a). The heat pipe condensers only extend until the edge of the heat source plane as it is difficult to incorporate sharp turns onto the stepped plane. The PHP design on the other hand, allowed for incorporating working fluid channels onto the stepped plane. The schematic of the PHP is shown in Figure 3 (b). A total of 24 PHP channels of size 1/16” (1.6 mm) were symmetrically spread out along the heat spreader. A common fluid charging port was provided to charge the PHP working fluid. ethanol. Propylene was selected as the working fluid based on the merit number calculation [i], size applicability based on Bond number and ease of handling.

Figure 3. Schematic of (a) HiK™ plate and (b) PHP for electronics cooling

A.   Heat transfer operating limit of the proposed heat spreaders

Before fabricating the two-phase heat spreaders, heat transfer operating limits were calculated for both HiK™ plate (from in-house codes) and Pulsating Heat Pipe (PHP) [ii] to assess the feasibility of using the proposed concept for modular electronics cooling.

The operating limits for one-half of one heat pipe were determined from established in-house codes, and the results are shown in Figure 4 (a). The maximum HiK™ plate heat transfer capability is limited by the capillary limit, which has an increasing trend with operating temperature. At 40 °C operating temperature, the heat transfer capability of the overall HiK™ plate is 180 W. On the other hand, the swept length limit imposes a maximum heat transfer limit on the operation of the PHP. The swept length limit decreases with increasing operating temperature. Propylene was chosen as the working fluid, as it has a Bond number limit at over 54 °C, while, the same for R134a, is only 36 °C. At 40 °C operating temperature, Pulsating Heat Pipe (PHP) heat transport limit is 80 W.

Figure 4. Heat transfer operating limits of (a) HiK™ plate and (b) PHP

A.   Fabrication of the heat spreaders

Along with the proposed heat spreader solutions, a standard aluminum base plate of thickness 0.1” was fabricated by conventional methods. The solid aluminum plate weighed 153 grams. The HiK™ plate was fabricated as per ACT’s standardized fabrication procedure. The heat pipe and the base plate production can start simultaneously at tandem. The heat pipes are soldered onto the base plate slots and standard surface treatment was applied. The prototype HiK™ plate for the testing in the SBIR program is shown in Figure 5. In the post-processing step, slots for securing the backplane and the ICE-Lok^® were provided on the stepped interfaces. The tentative heater location for simulating heat generation on the HiK™ plate is represented by the red square region at the center of the heat spreader. The HiK™ plate is now ready to be tested as the heat spreader. The total mass of the HiK™ plate was found to be 173 grams, which was 13% higher than a solid aluminum plate. Note that the mass of HiK™ can be further optimized by removing excessive base material. The optimized mass should be comparable to an aluminum plate.

Figure 5. HiK™ plate for electronics card cooling

Figure 6. 3D printed PHP for electronics card cooling

A PHP, on the other hand, was 3D printed due to the complexity of incorporating working fluid channels. The base material was standard aluminum alloy for additive manufacturing (AlSi₁₀Mg). Figure 6 shows the fabricated PHP. Post fabrication, a number of post-production steps were carried to charge and test the PHP as a heat spreader including metal power removal, x-ray inspection, de-powdering port closing, fill tube welding.  A helium leak check was performed to ensure that the PHP was leak tight. The PHP was then charged with working fluid up to 50% of the available channel volume. The PHP is now ready to be tested as the heat spreader. It was observed that the PHP weighed 137 grams, and was about 10.4% lighter than the aluminum standard base plate.

I.     IR Characterization of a Pulsating Heat Pipe (PHP) Prototype

Figure 7. Pulsating Heat Pipe (PHP) prototype for IR characterization testing

Before thermal performance testing, another prototype Pulsating Heat Pipe (PHP) with separate channels was first charged with acetone and placed under an IR camera to characterize and visualize PHP’s operating mechanism. A PHP prototype containing two separate channels as shown in Figure 7. was constructed for IR characterization. A 2” by 2” film heater was attached to the center of the test piece and the top surface was cooled by fans. For this characterization testing, only the left half PHP channel was filled with acetone up to 60% of volume. The right half was kept empty and used as an aluminum conduction-only test piece. This allows for a direct comparison of heat transfer performance between the PHP and conduction plate within the same test piece.

Figure 8. Left – Test piece temperature before Pulsating Heat Pipe (PHP) start-up; Right – Test piece temperatures when Pulsating Heat Pipe (PHP) is operating

The IR camera test results are shown in Figure 8. The figure of the left shows the increasing test piece temperature before the PHP start-up. At this time, the temperature is uniformly distributed throughout the entire test piece. Once, the PHP starts pulsating, the center temperature drops immediately from above 76 °C to below 70 °C. The temperature measurements at different locations of the test piece is shown above. Performance calculation showed that the conductance of the PHP was ~3.7 times higher than the conduction-only plate. The video file for the test characterization can be accessed at: https://www.youtube.com/watch?v=08AnO60Cw7k&feature=youtu.be

I.     Testing of heat spreaders for electronics cooling

Figure 9. Experimental test setup for testing heat spreaders

The three heat spreaders: the aluminum plate (baseline), HiK™ plate, and the Pulsating Heat Pipe (PHP) were tested with the electronics chassis cooling setup shown in Figure 9. The setup consisted of two cold plate channels at the top and the bottom of the electronics system chassis for uniform heat distribution. The 2-inch x 2-inch film heater was attached to the underside of the heat spreader. The coolant (water) temperature was set at 22 °C with a total flow rate of 0.9 gallons per minute. The heat transfer from the heat spreader to the sink was through the chassis by means of an ICE-Lok^®. Steady-state temperature measurements of the components were recorded at varying heater levels until the maximum temperature of 44 °C according to the MESSPS standards. In the case of the Pulsating Heat Pipe (PHP), propylene at 50% charge ratio (volume basis) was charged into the channels. During the testing, the system was insulated to mitigate heat losses to the environment.

Figure 10. Cooling system component temperatures with aluminum plate heat spreader

The experiments were first performed with aluminum plate as the baseline. The temperature measurements of the different components are shown in Figure 10. The dotted red line is the maximum allowable temperature of 44 °C. The steady temperature difference between the center to the edge of the heat spreader was proportional to the heat applied. It was observed that the aluminum plate can transport up to 21.2 W heat (flux: 0.8 W/cm²) with an overall system thermal resistance of 2 °C/W. The heat spreader thermal resistance was 0.75 °C/W.

Figure 11. Cooling system component temperatures with (a) HiK™ plate and (b) PHP as heat spreader

Figure 12. Instantaneous temperature profile of electronics cooling system

Figure 11 shows cooling system component temperatures with HiK™ plate and PHP as heat spreaders in the cooling system. In the case of HiK™ plate as heat spreader, the maximum recorded temperature at 42.4 W heat input was 43 °C. While, the temperature drop between the center and the edge was small, the temperature drop from the heat spreader edge to the chassis rail was large. On the other hand, with PHP as the heat spreader, the mean oscillation temperature was closer to 44 °C at around 43.4 W. The temperature drop between the center to the edge was higher than the HiK™ plate. However, the temperature drop between the edge and the chassis was smaller than the HiK™ plate. In this test, no noticeable flow-induced vibration of PHP was observed. Potential reasons for unnoticeable vibration could be (1) the mass ratio between the fluid and solid is small enough so that the oscillation of slugs in the mini-channels will not induce overall device vibration (2) the channel are symmetrically distributed in the plate[i].

The cooling system with all three heat spreaders was compared with instantaneous temperature plots at 21 W of heat input at the film heater. The results are shown in Figure 12. The aluminum plate heat spreader had the steepest temperature gradient from the center to edge at about 7.8 °C. This is the component thermal resistance which is limited by the thermal conductivity of the material. The component thermal resistance was the highest of the three heat spreaders at 0.75 °C/W. On the other hand, the temperature drop from the center to the edge of the HiK™ plate was the lowest at only 0.5 °C, while for PHP, it was around 2.1 °C. The isothermal profile comparison from the heat spreader center to the edge is shown in Figure 13.

Figure 13. Heat spreader temperature distribution during testing

However, it is interesting to note that the edge to the chassis rail temperature drop for the PHP was the lowest of all the three heat spreaders. For both aluminum plate and the HiK™ plate, the temperature drop was 8.7 °C and 6.6 °C, respectively, while, for the PHP, it was only 4.2 °C.

Figure 14. Gravity independence test of (a) HiK™ plate; and (b) PHP

A set of tests with horizontal and vertical orientation was performed to determine the influence of gravity on the performance of the HiK™ plate cooling system and the PHP cooling system. The test results are shown in Figure 14 at about 40 W heater power. At the start of the test, the HiK™ plate was first placed in the usual vertical configuration. The temperature drop from the center to the plane edge (same plane) was 0.4 °C. Then, the system was placed in a horizontal configuration. No change in temperature drop from the center to the edge was observed. After this, the system was returned to vertical orientation. HiK™ plate heat spreading performance was found to be independent of gravity. PHP heat spreader system, on the other hand, was first tested in a horizontal configuration. The temperature drop from the center to the edge was 4.7 °C. Then the system was switched to a vertical orientation, in which case, the temperature drop mildly changed to 4.3 °C. The system then was returned to horizontal orientation. The temperature drop from the center to the edge was again about 4.9 °C. The change in PHP component thermal resistance was negligible, so it can be surmised that the PHP performance was also unaffected by gravity. The performance of both, the HiK™ plate and PHP heat spreader cooling system is thus independent of gravity.

 I.     Finite Element Analysis

FEA was performed to analyze the heat transfer paths of the three heat spreaders from the center to the card edges, where ICE-Lok^® interfaces with a chassis. The results are shown in Figure 15. In this FEA, the applied heater power was 21 W at the center of the heat spreaders, and the convective heat transfer coefficient of 500 W/m²-K at 20 °C reference temperature was applied at the card-stepped edges. The temperature drop across the heat spreader edge to the stepped plane was highest in the case of aluminum plate, followed by heat spreader and lowest in the case of PHP.

Figure 15. The temperature profile of heat spreaders to analyze temperature distribution from the edge to the stepped plane

This reason can be explained as follows:

• As the CAD model shows (Figure 3), the two-phase PHP channels extend all the way to the stepped edge of the heat spreader where the ICE-Lok^® interfaces with the chassis. The two-phase channels have higher thermal conductance compared to the solid conduction. The heat pipes of the HiK™ plate only extend until the edge of the heat source plane, and not on the stepped edge. As a result, the heat transfer along the stepped plane is by base plate conduction in the case of the HiK™ plate.
• Comparing the aluminum plate and the HiK™ plate: the two-dimensional conduction plane heat transfer originates near the stepped edge for the HiK™ plate, while for the baseline, heat transfer is by conduction only from the center to the stepped edge.

So, it is evident that two-phase channels of the PHP reduce the thermal resistance across the stepped edge plane to the chassis rails. The system thermal resistances with aluminum plate, HiK™ plate and PHP as heat spreaders were 2 °C/W, 1.1 °C/W, and 1 °C/W, respectively.

 I.     Conclusions

ACT developed and tested two-phase-based HiK™ plate and PHP as advanced heat spreaders for 3U card electronics cooling. HiK™ plate is a copper-water heat pipe embedded aluminum plate. In the PHP, propylene was used as the working fluid due to its suitability in working temperature and higher merit number over other common working fluids. The unique feature of the PHP is that the capillary channels can accommodate the sharp edges like that of the ICE-Lok^® interface plane. Trade study of both HiK™ plate and the PHP showed heat handling capability up to 80 W and higher. Experimental results showed that both HiK™ plate and PHP can reduce the system thermal resistance by up to 50% from over 2 °C/W to around 1 °C/W in comparison to the aluminum plate (baseline). This can be attributed to the superior heat transfer characteristics of the two-phase working fluid in both the HiK™ plate (water) and the PHP. Table 1 summarizes the performance of the two heat spreaders.

The performance of the PHP heat spreader developed in this program is similar to that of the HiK™ plate but is lighter than the other two options. This is because the milled-out solid materials that form the channels is replaced by the saturated two-phase working fluid, which is less dense. The advantage of the PHP over the aluminum plate is that the two-phase channels extend onto the stepped edges, which improves the thermal conductance across the heat spreader-ICE-Lok^® interface.

Acknowledgments

This project is sponsored by NASA Goddard Space Center under an SBIR Phase I program (Contract# 80NSSC21C0211). We would like to thank Jeffrey Didion (technical monitor) for his support and valuable input to the program. NASA AMPS program managers (Jeffrey Csank and Brent Gardner) also provide insightful comments on this project. We would like to acknowledge Greg Hoeschele for his support on the development of the HiK™ plate. We would also like to acknowledge Justin Boyer, Jonathan Murray, and Philip Texter, who have significant technical contributions to prototype development and testing.

References

[1] A.J. Colozza and B.G. Gardner, Advanced Modular Power System Electronics Enclosure Thermal Testing, NASA/TM- 2019-220011, 2019.

[1] J. Zuo, W. Anderson, R. Bonner, Advanced Thermal Management Technologies for High Power Density Automotive Equipment, Proc. of National Defense Industrial Association Ground Vehicle Power and Energy Workshop, 2008.

[1]ACT’s HiK^TM Plate product page : https://www.1-act.com/when-to-use-heat-pipes-hik-plates-vapor-chambers-and-conduction-cooling/hik-plates/?hilite=Hi-K+plate

[1] H. Ma, Oscillating Heat pipes. New York: Springer, 2015.

[1] H. Yang, S. Khandekar, M. Groll, Operational limit of closed loop pulsating heat pipes, Applied Thermal Engineering, 28 (2008).

[1] V. Ayel, M. Slobodeniuk, R. Bertossi, C. Romestant, and Y. Bertin, Flat plate pulsating heat pipes: A review on the thermohydraulic principles, thermal performances and open issues, Applied Thermal Engineering, 197 (2021).

[1] A. Slippey, W.G. Anderson, M.C. Ellis, C. Hose, J. Schmidt, J. Weyant, Thermal Management Technologies for Embedded Cooling Applications, 2018 17^th IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems (iTherm), IEEE, 2018

[1] Advanced Modular Power Systems (AMPS) Project Mechanical Design Drawing Package,” NASA Glenn Research Center, 2020

[1] J. Kim, S.J. Kim, Experimental investigation on working fluid selection in a micro pulsating heat pipe, Energy Conversion and Management, 205 (2020).

[1] B.L. Drolen and C.D. Smoot, Performance Limits of Oscillating Heat Pipes: Theory and Validation, Journal of Thermophysics and Heat Transfer, 31 (2017).

[1]  S.M. Smith, M.J. Rohodes, B.S. Taft and E.V. Ardelean, “Experimental Investigation of Flow-Induced Vibrations of an Oscillating Heat Pipe” 46th AIAA Thermophysics Conference, Washington, D.C., 13–17 June 2016