Pumped Two Phase Cooling
Billy: Welcome and thank you for joining us for today’s webcast, Advanced Thermal Management Solutions: Pumped Two Phase Cooling. Sponsored by Advanced Cooling Technologies and Tech Brief’s Media Group. I’m Billy Hurley, associate editor with Tech Brief’s Media Group, and I’ll be your moderator today. Our webcast will last approximately 30 minutes, and there will be a question and answer period at the end of the presentation. If you have a question, you may submit it at any time during the presentation by entering it in the box at the bottom of your screen. Our presenters will answer as many questions as possible at the conclusion of the presentation. Those questions not addressed during the live event will be answered after the webcast. During this presentation we will also present you with two polling questions, and we invite you to answer those. In order to view the presentation properly. please disable any pop-up blockers you may have on your browser.
At this time, I’d like to introduce our speakers. Pete Ritt is ACT’s vice president of technical services. ACT’s technical services business provides thermal consulting, design, and analysis to a variety of industries, including medical devices. Mr. Ritt is a former RCA Thompson executive where he most recently was manager of strategic programs and business manager for the Lancaster R&D Center. In that role he was responsible for developing technologies, processes, and products for scale up and commercialization, which is common to some of his responsibilities at ACT. He has been granted over 20 patents and holds degrees in chemical engineering from the University of Notre Dame and an MBA from Shippensburg University.
Also on the line for our Q and A, is Dr. Xudong Tang. Dr. Tang, research and development engineer, joined ACT in 2006 after receiving his PhD degree in mechanical engineering at Carnegie-Mellon University. He has served as principal investigator on multiple R&D grants and contracts for the Air Force, Army, Navy, MDA, NASA, and NSF. Dr. Tang has extensive experience in two-phased high heat flux cooling, thin film evaporation, and boiling enhancement. He led the development of two-phase mini-channel heat sinks with microporous coating for laser diodes and high power electronics cooling. His technical expertise also includes vapor compression refrigeration systems and thermal energy storage. So, at this time I’d like to hand the program over to our speaker, Pete Ritt. Pete.
Pete: Thank you Billy. I’m Pete Ritt, along with Dr. Xudong Tang. We are delighted to be with you. For today’s webinar, we’d like to present a potentially new thermal management technology for design engineers to consider when solving challenging, high-heat dissipation problems — pumped two-phase cooling. Two-phase cooling generally means using a cooling fluid where at least some of the fluid is intentionally transformed into vapor upon heating, resulting in a vapor-liquid mixture in the control glue. Because a fluid’s latent heat of vaporization can be two orders of magnitude larger than the sensible heat of that single phase liquid cooling, there can be a significant improvement in heat absorption per unit volume of fluid. Pumped two-phase cooling has been the focus of research in government and academia but has not yet been broadly adopted in the commercial sector, in spite of some very attractive performance features.
In this webinar, we will do an overview of the market outlook for pumped two-phase systems, then discuss advantages and challenges of the technology, compared to pump liquid solutions. We will review the options within pumped two-phase cooling. We’ll then spend some time talking about mini and micro-channel cooling, which appears to be the most likely candidate for broader commercial acceptance and look at how the addition of microporous coatings can be used to address some of the identified performance limitations. We’ll also explore some new concepts that address maintenance issues, like quick disconnects and scalability issues, like multiple parallel evaporators and offer new insights to these challenges, which have been raised with pumped two-phase systems.
For the past few decades, excellent thermal management solutions have been attained using well-established technologies such as air cooling and pump liquid system. However, several emerging thermal management applications would have such large space and power requirements for either an air-cooled or pump liquid system, that they are no longer viable solutions. Some of these applications are pictured here. ITBT and Power Electronics, Data Center Cooling, hybrid battery packs, and new vertical cavity surface emitting lasers. In some projected applications, heat fluxes upwards of 1000 watts per centimeter squared are predicted. Therefore, there is an increasing need for compact, high heat-flux, thermal management system. Pumped two-phase cooling offers a potential solution to address that need.
The two most common thermal management technologies that dissipate kilowatt or higher heat loads are pump liquid or pumped two-phase solutions. Pump liquid systems usually contain a cold plate, which is in close contact with the heat source. There are channels in a cold plate in which a sub-cooled liquid is pumped through, removing the waste heat. The physics for these systems is well understood. Typical heat flux capabilities are in the 10 to 20 watt per centimeter squared range. For example, the cold plate pictured on the center right was designed to dissipate 6 kilowatts within over 12 by 12 inch area, with a heat flux of 15 watts per centimeter squared. With pump liquid systems, the larger the evaporator, the more degraded the isothermality will be. The fluid will pick up heat as it traverses along the tubing path, increasing the delta T from inlet to outlet.
Increasing heat flux requirements can be addressed with a bigger pump, but this requires additional space and power. Pumped two-phase systems offer the potential of significantly higher heat flux capabilities. Most experts estimate that fluxes in the range of 300 to 1000 watts per centimeter squared are feasible. Additionally, well-designed systems that achieve vaporization over the majority of the evaporator surface can provide excellent temperature uniformity as well. The presence of vapor as a heat transfer medium also reduces thermal resistance and provides higher heat transfer coefficient values. A signature benefit of these systems is that flow rate requirements are substantially less than pump liquid systems. A typical rule of thumb is that these vaporization and convective benefits yields a 2-4 times increase in heat removal capabilities, directly correlating to pump size requirements. Optimal performance is achieved when the fluid enters the evaporator close to its boiling point. This does imply that a well-controlled system is needed. Non-optimal systems can be susceptible to flow and temperature instabilities.
There are, however, some advancements that can be added to minimize these instabilities. The system pictured at the bottom right is a pumped two-phase system, which achieved the heat flux of 350 watts per centimeter squared over a 2.6 centimeter squared area. Total heat dissipated is 800 watts. For larger, evaporator areas, these systems can be modularized.
To give a clear sense of the flow rate and pumping power differences between pump-liquid and pumped two-phase systems, the following comparison is provided. In order to dissipate 80 kilowatts of heat, a pump liquid system using PAO as the fluid would require a flow rate of 35 gallons per minute and approximately 5.3 kilowatts of power. Pumped two-phase system using R-245fa would require only 6 gallons a minute and 250 watts of power, which is only 20% of the flow rate and 5% of the power needed by the pump liquid system. A significant difference for sure.
To further reinforce the potential benefits of pumped two-phase cooling for high heat-flux application, here is a chart based on results from [inaudible 00:08:48] et al, comparing the different thermal management technologies. Log scale of heat transfer coefficient is plotted on the x-axis, and delta T is on the y-axis. The most desirable conditions is low delta T, with a high heat transfer coefficient, which is where pumped two-phase systems, boiling in this chart, is situated.
Let’s review pumped two-phase systems in more detail. But first, let’s turn it back to Billy for a polling question.
Billy: Thanks Pete, now it’s time for our first poll question and it should appear on your screen now. The question is, who does your company use to resolve critical thermal management issues. Your choices here are: A) In-house resources, B) External experts, C) A combination of both in-house and external resources, or D) Not sure. You can make your selection now by choosing the appropriate button on your screen. Again, who does your company use to resolve critical thermal management issues, In-house resources, external experts, a combination of both, or not sure. Back to you Pete.
Pete: Thanks Billy. There are two major embodiments of pumped two-phase cooling. Spray or jet impingement cooling and mini or micro channel heat sinks. The first is a pressurized stream, which is sprayed onto the hot surface. The second option is closely analogous to pump-liquid cooling, in that there’s fluid flow across the evaporator, with the flow inlet and outlet. Let’s briefly look at all of these options.
Here are two sketches, one of spray cooling on the left and jet impingement cooling on the right. Both have pressurized fluid spray that form a two-phased film when it contacts the hot surface. In the case of spray cooling, when the fluid leaves the nozzle, droplets are formed which impinge on the evaporator surface. With sufficient heating from the evaporator surface, there is bubble formation in the liquid film and the system operates as a two-phase cooler. The spray enables wide coverage and efficient spreading of liquid, offering good temperature uniformity at low flow rates. Challenges for spray cooling include precisely controlled fluid flow, as well as maintenance and life concerns. There is potential erosion and corrosion from extended exposure of the spray. Additionally, nozzle clogging can occur from impurities.
Jet impingement cooling is similar to spray cooling but with a different nozzle and dispensing mechanism. Here, a continuous flow of fluid directly impacts a focus section of the evaporator. Again, as in spray cooling, with adequate heating bubbles can form and the system can operate as a cooler in two phases. There are two principle sections of fluid flow: The impingement zone where the fluid hits the surface and the flow region where the fluid flows from the impingement zone to adjacent areas. Since the jets concentrate most of the cooling in a specific area, the heat transfer coefficient is higher in these areas. Multiple jets can be implemented to cover larger areas. Because the fluid flow is continuous, unlike droplets in spray cooling, the fluid dynamics is a bit simpler. In addition to increased erosion-corrosion effects, jet impingement requires high flow rates and may also result in increased temperature non-uniformity and increased thermal stress.
The other principle pumped two-phase cooling option is mini and microchannel heat-sink cooling. The difference between the two is the thin pitch in the heat sink. With minichannels typically having a pitch of 2 to 8 mm and microchannels having finer thin pitches, usually less than 1 mm. Here are sketches of the two options. Microchannels offer the advantage of increased surface area but typically require higher pressure drops and are more vulnerable to clogging and corrosion-erosion than minichannels.
Two-phase mini and microchannel heat-sinks can have both steady state and transient pressure instabilities, which can make them vulnerable to both dry-out and critical heat flux overshoot that can cause severe overheating and damage the electronic component. Nevertheless, mini and microchannel coolers have demonstrated the potential to provide the thermal management necessary for emerging high heat flux applications. Heat fluxes between 350 and 800 watts per centimeter squared have been achieved. Larger evaporator areas may be addressed through modular heat-sink systems. Because their potential application areas are so large and the benefit is high, we will now focus on mini and microchannel systems and see what may be done to improve on the major system performance concerns.
A detail schematic of a pumped two-phase cooling loop is shown here. Not all elements are required. Key components include the pump, evaporator, condenser, and the accumulator. The two-phased tank and the pre-heater units are precision control components of this particular lab system. Dirge tank consists of vapor and liquid at saturation. By controlling the tank temperature, the system pressure and in turn boiling temperature of the working fluid is controlled. Since the working fluid exits the condenser at a sub-cooled state, a pre-heater unit is included to adjust the temperature of the liquid close to the saturation temperature before it enters the evaporator. This is important as boiling heat transfer is most effective at saturation. Usually only a small pump, with low flow and pressure head is required, because the heat transfer is mainly due to phase change in a large value of latent heat of vaporization.
Here is a picture of a pumped two-phase lab system. The system was used for cooling laser diodes. The evaporator area was 2.6 centimeter squared and dissipated 800 watts of heat. E-max includes a thermal interface material which has a thermal conductivity of 0.04 to 0.1 degree C per watt. Note the relatively small pump size. To offer another example of comparative pump requirements, in a separate study it was reported that to dissipate one kilowatt of heat, a pump liquid system required a pump flow rate of 46 liters per hour. A pumped two-phase system only required a flow rate of seven liters per hour to dissipate the same heat load.
We previously mentioned that one of the key issues with pumped two-phase systems is temperature and flow instabilities, which is usually caused by erratic fluid boiling in the evaporator. As already mentioned these instabilities can lead to dry up and result in damaging the electronic component. An enhancement that can be implemented to improve stability on the microchannel heat sink is the addition of microporous coatings. These microporous coatings increase the number of nucleation sites in the evaporator and enable bubble formation throughout the heat-sinks’ structure. The porous coating also serves as a liquid supply layer which minimizes the risk of dry up and increases critical heat flux. You can see on the left a representative minichannel heat sink. Seen on the right are examples of microporous coatings, epoxy based on the top, and sintered coatings on the bottom. We will next examine how they perform.
The effect of various coatings on heat transfer performance of flat and minichannel copper heat-sinks was evaluated in water pool boiling tests. Coating were developed in collaboration with Professor Yu at the University of Texas. ABM is aluminum Devcon brushable ceramic and methyl ethyl ketone. TCMC is thermally conductive microporous coating, and HTCMC is high temperature thermally conductive microporous coating. The graph shows delta T saturation, also known as super heat on the x axis, which simply is the device interface temperature minus the fluid temperature. Heat flux is on the y axis. Porous coatings are plotted. What’s most interesting is that the HTCMC is almost a vertical line over a wide range of heat fluxes. This means that with use of coatings, device temperature doesn’t change over a broad range of heat dissipation levels. This is very important in the case of laser applications where temperature stability is critical. The type of coating and the sealing processor are obviously very important details but results clearly show the potential benefit of coatings.
Here’s additional data demonstrating the benefit of coatings on both flat and minichannel heat sinks. Uncoated heat sinks are blue and coated ones are red. The graphs on the left plot heat flux versus heat transfer coefficient. One can see an increase of up to three times improvement in heat transfer coefficient with addition of a microporous coating.
Similarly, on the graphs on the right, the incipient wall of super heat, again the device interface temperature minus the fluent temperature at the onset of boiling, is lower for the coated versus uncoated sample. Note also the relatively flat heat flux in the flat heat-sink case in the graph on the upper right, again demonstrating the ability of pumped two-phase systems to handle a range of heat fluxes without having the device temperature change.
Here is a quick video that demonstrates the benefit of microporous coatings. It shows flow through the pumped two-phase minichannel heat sink system that we have been talking about. In the video, flow is from left to right. The first case, without coatings, you can see large bubble formations on the left hand side at the entrance of the minichannel heat sink. The bubbles effect flow into the heat sink and cause undesirable back pressures, which result in instabilities in the system. The unstable flow causes fluctuation in both temperature and pressure across the heat sink. At one point, the camera will focus on a gauge to show pressure fluctuation.
In the second part of the video, a heat sink with a microporous coating is shown. All the other conditions are the same as in the first case. The microporous coating provides additional nucleation sites, which promotes boiling throughout the evaporator. Note how the system is far more consistent and stable. The formation of large bubbles, seen in the initial case, is no longer present. Let’s have a look.
I think that’s a pretty clear example of the benefits of microporous coatings in pumped two-phase systems. Here are some results for the two heat-sinks we just saw. In both cases, a 12-channel heat sink is used with thermal couples mounted on the heat sink, half a millimeter under the channels, refrigerant was R-245fa, the flow rate was 0.3 liters per minute. Heat flux ranged from 0 to 214 watts per centimeter squared.
The two graphs across the top show the results of temperature and pressure readings for the uncoated minichannel. The lower two graphs show the results under the same conditions with a microporous coating. In the microporous coating test, delta T was reduced from 5.8 to 1.4 degrees C and also yield a comparable drop in pressure fluctuation. These results clearly suggest coatings may be a good solution to address system instability.
Before we move on to quick disconnect, it’s time for our final polling question. Back to you Billy.
Billy: Thanks Pete, again this is our second poll question and it should appear on your screen now. The question is “Are you facing any thermal management issues in the next, A)0-6 months, B) 6-24 months, or C) are you having no immediate issues?” So you can make your selection now by choosing the appropriate button on your screen. Again, the second poll question, “Are you facing any thermal management issues in the next 0-6 months, 6-24 months, or C) are you having no immediate issues?” So we’ll give you a second to answer that question and as you do that, I will hand it back over to Pete Ritt.
Pete: Thanks Billy. Many high-power electronics are intended for mobile use. There is some concern that pumped two-phase systems may not be suitable for field deployment. Specific issues have been raised about the ability of pumped two-phase systems to be used with quick disconnects to permit rapid replacement of the electronics. ACT recently completed testing that showed good system performance before and after disengaging and re-engaging flow with quick disconnects to the evaporator sections. The disconnect reconnect cycle was replicated 20 times, which is a reasonable number to accommodate initial setup and some field replacements. As you can see on the graph on the right, wall superheat variability was small, plus or minus 1.5 degree K, indicating stable performance from the system. Here is a quick video of the test setup and operation of one cycle. First the operator disconnects flow and then re-engages. Take a look at the video.
Note there was minimal loss of refrigerant or contamination with air. It’s not real clear but the liquid level in the system remained the same before and after the quick disconnect cycle. This simple test does demonstrate feasibility of using quick disconnects on pumped two-phase systems.
We have mentioned that thermally managing multiple evaporators with non-uniform heat loading is a well known challenge for pumped two-phase systems. The issue is that application of heat to the evaporator causes vapor generation which in turn causes more resistance to the flow of coolant in that evaporator leading to flow non-uniformity between parallel evaporators. Some recent testing at ACT has shown some promising results in this area. Testing was done with two parallel evaporators initially set at the same condition, where QB over QA equals 1. A diagram of the setup is seen on the upper left. Evaporator A was unchanged during testing, while the heat loading on evaporator B was varied up to a factor of 4. As you can see on the graph on the right, the evaporator temperature on A was unaffected by the large shifts on heat loading on B. And even B, while showing an expected increase in evaporator temperature, did not exhibit run away performance and was well within acceptable operating conditions. This encouraging result is attributed to upscale pressure drop elements seen in the diagram and the relative weak relationship between heat transfer and pooling flow rate, assuming the minimum pressure of flow rate is maintained. We have seen this insensitivity to flow rate in other graphs presented today and along with more compact system size, it is one of the key benefits of pumped two-phase systems.
In the webinar today, we reviewed pumped two-phase cooling, its advantages, operations, challenges, and some new technology being developed to address these challenges. Pumped two-phase cooling offers the potential to address increasing heat dissipation requirements up to 1000 watts per centimeter squared by some estimates, which air and pump liquid cooling would be weight and power challenged to accomplish. Pumped two-phase systems require only a small pump and can be designed to be compact and light weight. We reviewed various pumped two-phase solutions, of which mini and microchannel cooler heat sinks seem to have the broadest market potential. These pumped two-phase mini and microchannel systems have demonstrated high heat flux, 350 watts per centimeter squared in the results shown here but have been hampered by concern of system instability and complexity.
Continuing, microporous coatings on the minichannel heat sinks were shown to improve system stability and temperature uniformity by increasing the number of nucleation sites across the evaporator surface. For maintenance and field deployments, pumped two-phase system stability with repeated engage, disengage of quick disconnects to the evaporator was shown without loss of refrigerant. Some encouraging results showing insensitivity for fluctuating heat loading on a parallel evaporator was provided. So the potential benefit of pumped two-phase cooling remains large. And now with some enhancements such as microporous coatings, broader market acceptance of the technology is more likely.
Thank you for your attention. We hope you have found this information to be useful. Now we’ll turn it back to Billy for some questions.
Billy: Thanks Pete. At this time we’d like to being our Q and A. I’d also like to welcome to the line ACT’s research and development engineer, Dr. Xudong Tang. Dr. Tang thanks for being with us, here’s our first question. “What’s the basic difference between a pump liquid cold plate design and a pumped two-phase cold plate design?”
Xudong: Okay, comparing with a pumped two-phase and the pump liquid, there are two fundamental differences, the first is heat transfer coefficient. We know that in the pumped two-phase because of nuclear boiling, it has a much higher heat transfer coefficient than pump liquid loop. So in order to enhance the age, pump liquid loop usually, they want to increase flow rate which result in bigger pump or using microchannels or something like that to increase the surface area. But, the microchannel also increases the pressure drop on the pump site. It also has filter issues and corrosion and erosion issues. Another fundamental difference is pumped two-phase uses latent heat instead of sensual heat to cool the heat source, though you can achieve very good isothermality over large surface area but for a pump liquid loop, it is very hard to achieve that. And moreover, because we use refrigerants in our pumped two-phase loop, refrigerants, you may ask, we know that the electronics manufacturers are reluctant to use water to cool their electronics because they are afraid of leak but since we use refrigerants, the leak won’t do any damage for the system, so I think that’s a major benefit also for the pumped two-phase refrigerant system. Next question please.
Moderator: Here’s a question. An attendee’s asking for examples for two-phased fluids.
Xudong: Oh, well, we generally use refrigerant 134a or 245fa. It basically depends on the temperature you want to control at. So you need to look at the saturation temperature and of course you want a higher latent heat so that’d be lower flow rate and less pressure drop. Another important property of refrigerant that we look at is the saturation pressure, because you always want to design a pressure wise system instead of vacuum because pressure wise system is a lot easier to maintain. And another important property is liquid to vapor phase density ratio. You want to have this as low as possible because higher density ratio, it’s easier to cause flow instabilities. We also have [inaudible 00:32:17] number to consider in the chemical conductivity, viscosity, even surface tension to evaluate different refrigerant for different applications. Next, please.
Billy: How important is it to control vapor quality within the evaporator section and how is this done?
Xudong: The vapor quality is important, especially for the minichannel inlet and outlet. They have a great, strong impact on the heat transfer coefficient. If the inlet has too much subcooling, you only have a single phase at the entrance of the minichannel [inaudible 00:32:54] low so for high heat flux applications or for tight applications with high temperature control, you’re going to have a temperature gradient that’s no good. But for the exit, people usually want to increase the exit quality to 100%, basically to evaporate all your liquid refrigerant. But for the high heat flux, we generally try to control the exit vapor quality below 50, because some theory or testing says that when the quality is beyond 50%, the heat transfer coefficient goes down. Of course, there are contradicting theories and testing results but for relatively lower heat flux, like heat flux around 100 watts per square centimeter, exit vapor quality is not a big concern. Next, please.
Billy: Here’s another question, “Are there performance differences with microporous coatings and what are key attributes of a good coating?”
Xudong: Right now, we apply microporous coating on aluminum fins surface or on the copper fins surface. For the copper fins surface, we have our unique recipe of microporous coating. We can apply very evenly from coating on the minichannel surface. So, there are many… For the aluminum surface, we use Dr. Yu’s coating. He’s a professor at the University of Texas at Dallas. He uses epoxy-based coating. For coating, of course thickness, the pore size, they are all very important. For thickness, you want to have a very uniform thickness over the fins surface, which is also pretty hard to achieve. Next, please.
Billy: This will have to be our last question due to time, how do these coatings work for extruded aluminum fins with forced airflow through a channel?
Xudong: On the extruded fins surface, we have our special sintered copper powdered coating and we can achieve a pretty uniform coating because of the way… I don’t want to disclose the details how to put the powder on the surface and the sintered but the result is, we have a very uniform coating because if you don’t have a uniform coating, you don’t have those interlinked flow passage inside the coatings. The flow stability cannot be easily controlled and also you cannot have a very high heat transfer coefficient.
Billy: Thanks Dr. Tang, that will conclude today’s webcast. We’re out of time. Again, if we did not get a chance to answer your question our sponsors have them, and we’ll do our best to address them after today’s presentation. So, thanks to everyone for joining us. I’d like to thank our speakers, Pete Ritt and Xudong Tang. Just a reminder, this webcast will be available on demand at www.techbriefs.com for the next 12 months. Have a great day.