Heat Wheel Could Cut Data Center Cooling Bills

32 comments

A relatively new approach to data center cooling known as a “heat wheel” is gaining momentum, and likely to gain a higher profile from an upcoming demonstration of the technology. The heat wheel – also known as a rotary heat exchanger or Kyoto Cooling – is a refinement of existing approaches that take advantage of outside air to improve cooling efficiency and reduce data center power bills. 

Proponents of the heat wheel say it improves upon air-side economization (free cooling), the use of outside air to cool servers in the data center. Rather than introducing exterior air directly into the server room, the heat wheel briefly mixes the outside air and exhaust air to creates an “air-to-air” heat exchanger. 

“The inside heat from the IT room is still removed via the heat wheel, but there is minimal air transfer between the ambient and the computer room,” explains Uptime Technology BV of the Netherlands. “This system has all the benefits of Airside Economizing, without the exposures of airside economizing like contamination and humidity control.”

Heat wheels have been used for many years in industrial air conditioning, but never in data centers. Like air-side economization, heat wheels could produce significant energy savings by reducting the need to use power-hungry chillers for air conditioning. 

Uptime Technology has filed a patent application with the European Patent Office on the technology. In 2007 it partnered with Chatsworth Products Inc., joining the heat wheel with Chatsworth’s hot-air containment to produce Kyoto Cooling. The technology has been implemented by European telecom firm KPN in one of its CyberCenter data centers in Amersfoort, The Netherlands. KPN uses the system at outside temperatures of up to 72 degrees (22 C), and it effectively cools exhaust air of up to 98 degrees.    

“In the Netherlands it is possible to cool the CyberCenter for 354 days a year with KyotoCooling, without needing any additional air-conditioning,” said Max Alias, program manager CyberCenter Services at KPN. “That provides enormous energy savings. Overall power consumption in the CyberCenter is reduced by 20 to 30 percent. The power consumption for cooling is reduced by more than 70 percent. So it’s not only green, it’s also saving money.”

The Amersfoort CyberCenter will be the site of a series of seminars December 1-4 in which Kyoto Cooling will be demonstrated. Representatives of UpTime Technology, Chatsworth and KPN will present, along with Bob Sullivan of the Uptime Institute in the US. 

Sullivan is known for originating the hot aisle-cold aisle configuration used in most modern data centers.  The Uptime Institute has examined the potential of the the heat wheel in its discussions of data center energy efficiency at several of its events, most notably its design charette in 2007.

While Kyoto Cooling has its proponents, its use is limited to climates with temperature profiles that would allow the heat wheel to be used for a substantial part of the year. The geographic range for heat wheel usage is likely to closely resemble regions favorable for air-side economization.

About the Author

Rich Miller is the founder and editor at large of Data Center Knowledge, and has been reporting on the data center sector since 2000. He has tracked the growing impact of high-density computing on the power and cooling of data centers, and the resulting push for improved energy efficiency in these facilities.

Add Your Comments

  • (will not be published)

32 Comments

  1. Bob L

    Um. Not really all that innovative, let alone something worth patenting. I have a EVR (Energy Recovery Ventilator) in my home that uses a wheel..... Interesting that the concept is finally being applied to a Data Center.

  2. Hi Bob. Thanks for the comment. As the story notes, heat wheels aren't new. But you're absolutely right in that it's interesting that it hasn't been implemented in data centers before.

    I think it's a sign of the times in the industry. When energy bills were less onerous, data center operators had less incentive to make the design and ducting changes to implement something like this. Now companies are scrambling for ways to reduce power usage, and chillers are a big chunk of that.

  3. Lenny

    Useful but not that useful in a modern data centres as rack densities climb. My throughts are to target water side economisers and close coupled cooling. Combined this is will a a far bigger impact that the heat wheel.

  4. Alex Long

    A raw site I am working with is contiguous to a canal. Can anyone tell me if heat exchangers have been used to cool data centers using a body of water as the thermal mass? The canal is about a mile long. Another concept is the application of an aeration system, in the canal, to disipate heat in the summer. Thank you. Alex Long, along@ccim.net

  5. A. Shore

    Also worth noting is Google's pressuring Intel to manufacture chips that have a higher heat tolerance than current designs do. Next disc drives?

  6. Ken

    The problem with the canal is how much energy you're considering dumping into the canal and whether the change in temperature is going to affect anything already living there adversely. If there's no indigenous life in the canal that there likely isn't a problem but if there's something there that someone else values then you've got a bigger problem that the technical issue of transferring heat to the water. Make sure you don't need to submit some environmental impact statement before you commit to putting too much work into the actual heat pump.

  7. Tony Toews

    Alex, it's my understanding that a number of large buildings on the Toronto, Canada waterfront are cooled with water from Lake Ontario.

  8. Matt Rosin

    I believe Cornell University in Ithaca, NY is connected to Lake Cayuga. Far above Cayuga's waters, a closed heat exchange system is linked to the depths of the lake. Of course the winters are also quite brrrr! chilly.

  9. James Smith

    Alex, google "water source heat pump". This paper (http://geoheat.oit.edu/pdf/tp15.pdf) has a nice overview.

  10. Dale Thompson

    Air/air heat exchange is one of the dumbest things that the data center is saddled with. I see how we got there and there is a huge infrastructure to support it but any 1st year engineering student would flunk there thermal course if they designed a data center like the "pros" do now. Hot Aisle, Cold Aisle, CRAC units is absurd. My question; how powerful do you want Liebert to be? There is zero incentive for the supply base to change if we follow each other over the cliff and continue to support this madness. Sun, HP, IBM all have very efficient liquid cooled thermal solutions but will not introduce them because the customer base wants to fit it into the hot aisle/cold aisle paradigm. The P575 is a good first step, we should support these efforts or nothing will change.

  11. What about high humidity? I live in a climate that regularly reaches 100% humidity, and usually lingers around the 80% range. It seems something like this sort of thing wouldn't help us at all to cool our little bitty 25kw data center on the island where I work.

  12. Jimmy Weaver

    The energy wheel is actually a desiccant wheel. It requires some energy to regenerate the wheel side that actually removes a certain amount of heat energy to lower the incoming latent air temp to lower the humidity level that is briefly dicussed. The desiccant moisture ABSORBENT chemical gives up a given amount of BTUS that it ABSORBORED when it removed the moisture of the incoming air and this heat is discharged when the desiccant is recharged. This entire cycle does require energy in too work and the maintenace is high. Your chasing dollars if you are using outside air and calling it free cooing to cool servers by using a desiccant wheel to lower the incoming wet bulb temperature.

  13. Robb J

    True, heat wheels have been around a while (over 25 years) but never with the complex controls or the "Plumbing" if you will, that make up the Kyoto wheel. It is NOT air to air heat exchange. There is absolutely no mixing of the air streams at all. It has the benefits of Air-side economization without the challenges of humidity and contamination. Knowing that the wheel has the capacity to cool up to 1mW of blade servers and consuming not more than 30-40kW to do it (Outside conditions influence this number) I believe it makes it the most energy efficient cooling technology currently known and deployable. Of course this will include efficiencies gained from the power cubed law by running multiple wheels at a time. Does anyone else know of a solution that has a COP approaching 50 or better? How many hours a year can it be used? Well that depends on your set point and your data center architecture but I can say that I have seen data in the US allowing for the wheel to be used over 90% of the year. I suppose we will wait and see as the actual data is collected from real data centers using the wheel. Until then, I would love to discuss it with someone. I think that reducing energy by over 90% compared to mechanical cooling is NOT a dumb idea.

  14. Robb J

    Regarding humidity, you would have to be a little more specific about your absolute humidity. 100% is relative to your temperature of course but I can tell from the context of your message that you are also in a warmer climate where the pounds per square foot of moisture must be high. There may be some humidity controls needed with your make-up air requirements but once the humidity is dialed in, you won't have to worry about it that much since there is NO mixing of outside and inside air and with the new ASHRAE requirements, dialing in humidity will be easier and less costly. Use of the wheel is not all that limited to cold climates. Hot and humid climates will certainly have the most difficult challenge when it comes to any economization strategy but using the Kyoto wheel will give you the most gains in most climates. I would say the major selling point on this type of economization will be the size of the data center. You will see the best ROI's with data centers that are over 300kW IT loads. I hope this helps you all to understand Kyoto a little better.

  15. It’s remarkable to read that people are dusting off this archaic methodology to achieve energy efficiency let alone promoting its application to a state-of-the-art Data Center. Aside from climatic conditions and not just temperature and humidity, there’s no mention of the problems associated with using these devices in highly polluted areas. The downsides of Heat Wheels aren’t well defined in this article, which should be addressed. The physical dimensions required for the device and the associated ductwork is problematic especially in an existing building. A Rotary Wheel Air to Air Heat Recovery Wheel generating about 28,000 CFM (equivalent to about two Liebert 30 Tons air conditioning units) is10 ft wide, 10 ft high and 18 inches in depth. Furthermore, the Heat Wheel needs to be incorporated in a water proof enclosure along with the circulating fans, motors, drive assembly, filters and controls; this needs to be about 10 ft high by 10 ft wide by 12 ft deep. Additionally, there’s the ductwork. The ductwork necessary to conduct the air to and from the conditioned space to the air heat recovery device should be not less than 50 X 40 inches (one supply and one return), and must be insulated. Furthermore, the ductwork to the transition at the inlet and outlet of the heat recovery wheel/enclosure needs to be 120” X 60”. It stands to reason for small Data Center applications requiring from (500 to 5000 CFM) capacity, located in proximity to an exterior building wall, in the right geographic location could apply this method of energy recovery. But doesn’t seem practical for a significant size or high-density facility. This of course, is unless the entire building is designed and constructed to accommodate this method of energy recovery. One cannot consider the use of such a system without taking into consideration the enormous amount of space and the cost of such space required to install such a product-- which keep in mind still requires a full mechanical system and generator back up--for the days when the system doesn’t work. It’s unlikely this system would lend itself to existing data centers or retrofits. Additionally, as listed in the ASHRAE Systems and Equipment Handbook in chapter 42 entitled "Air to Air Energy Recovery.” The Rotary Wheel Type: The air volume range is 50 to 70,000 CFM Face velocity 500 to 1000 fpm Pressure drop 0.4 to 0.7 wg Cross leakage 1 to 10% Heat Device effectiveness rang from 50 to 80% Considering the volume in the example above is only 28,000 CFM, one can imagine the size of the units and associated ductwork if this were a 70,000 CFM requirement. Ever since the Rotary Air to Air Wheel was introduced, their remained a problem in maintaining Sensible Heat transfer capacity due to the accumilation of airborne contamination forming a film on the media dramatically reducing the effectiveness of the system. The use of air filters on both sides of the Air Wheel will not solve the problem. The problem is the outside air is typically contaminated with all types of pollutants. Most of the sources are from Furnace and Internal Combustion sources. When the pollutants are carried in the atmosphere, the outside air to the "Air to Air Heat Exchanger Wheel Media," they adhere to the grounded media and form a film that in time insulates the media, interfering with the transfer of energy. This changes the percentage of effectiveness of the media, from when it was clean to a lower percent of efficiency. This has always been a significant downside to the concept making this a high maintenance product. The article goes on to say, ““This system has all the benefits of Airside Economizing, without the exposures of airside economizing like contamination and humidity control.” This seems to be an outreaching claim that would be good to understand how they intend to accomplish this. The TCO and OPEX isn’t mentioned in this article, however, since it’s not stated perhaps it's very high, and has a very long ROI. What’s the real cost and assurance of the system working? Do you want to rely on a system with a historically poor track record to cool, or supplementally cool your data center. Overall, there’s no such thing as “Free Cooling,” there’s costs related to owning and operating such systems and such costs can’t be overlooked. Optimally, if one were to save energy, there’s clearly better ways.

  16. Robb J

    One thing you might consider is to going to www.Kyotocooling.com and learning more about the wheel and perhaps even speaking to the folks who have engineered it. I would hope you would find it impressive to learn that not only is it constructed differently than the wheel technology you might be use to but its efficiencies are much greater than what you will read about in a book. Just to name a few benefits: less than 1% leakage, cools the return air to within <5 degrees of the outside ambient air (great heat transfer percentage), high COP (38 is documented), and in a Tier 4 x 3mW data center considering the construction cost savings, etc. the ROI could be immediate. It is obvious that this is a major upgrade to an old idea/existing technology and that's part of what makes it great. It is clear that the KyotoCooling wheel is not well understood by some folks in this group and the article didn't exactly hit the nail on the head explaining it but I think the point is simple: there could be an amazing opportunity here to significantly reduce energy and save a lot of money. Not everyone will be early adopters but some people are already taking that leap. Let’s avoid using unrelated data to make a decision and at least see what the data centers that are using it report. I too was skeptical at first and did my research on heat wheels etc but as I learned more about the Kyoto wheel, I became a "fan" (if you can't tell) and I encourage the skeptics to do your homework on the Kyoto wheel and not Heat Recovery Wheels in general (I went down that road already) because at minimum, your numbers do not reflect what the wheel is providing and it is worth checking in to. You might find your new favorite economizer. We all know Free Cooling is not "Free." That term has been around for ages as a way to say that Mother Nature to do the cooling for us. When we use the term "Free Cooling" I believe most people familiar with cooling might say to themselves "Compressor Free Cooling". Of course there are still fans, pumps, etc but the cost savings (including OPEX) from this type of economization vs. running your mechanical cooling plant is enormous and most often a faster ROI than a water-side or air-side economizer (comparing apples to apple on capacity). One last question from me: If you could cool a 538kW heat load when it is 56 F outside with a return air temp from the DC at 117 F which cools to the supply set point of 67.5 F as it passes through the wheel and in to the DC while consuming only a total of 14kW (all components in the cooling cell combined), would you consider that an efficient technology? Who wouldn’t? These are real numbers from a data center in the Netherlands and it works with much higher outside temperatures. Regarding mechanical cooling provisions: The wheel gradually steps towards mechanical cooling as the outside temp increases over the set point (80.6 if you like) and approaches the return air temperature where you would be using the integrated mechanical cooling. Again, it is fascinating how this technology has already addressed and conquered the concerns with heat wheels that are being communicated here. Ok I don't want to give everything away. Please go look into this technology and seek to understand it before passing judgment. Our energy grid, our atmosphere, and most of our pocketbooks need you to do your due diligence. Have a wonderful week everyone.

  17. The M-Cycle for cooling of data centers through the Coolerado Cooler offers up to a 90% power reduction. Recently, the U.S. Department of Energy (DOE) published the FEMP review (DOE/GO-102007-2325): “Coolerado Cooler Helps to Save Cooling Energy and Dollars” (Shortcut to: http://www1.eere.energy.gov/femp/pdfs/tir_coolerado.pdf http://www.achrnews.com/Articles/Article_Rotation/BNP_GUID_9-5-2006_A_10000000000000463073) Coolerado air conditioners can be found in markets around the world- in Japan, Europe, and South America, Singapore as well as in the USA from Washington to Florida. The Coolerado cooler utilizes the Maisotsenko Cycle, which has transitioned from the conceptual stage to an energy saving commercially available product. The Coolerado Cooler offers up to a 90% power reduction for air conditioning, and it can be efficient utilized for data centers, electronics cooling and microclimate systems. The Coolerado Cooler is unique apparatus for the expansion of cooling demands in emergency situations. Our cooler falls into a new category of an “ultra” class cooler because of our extreme energy efficiency and ability to cool air below the wet bulb temperatures without a compressor and can offer your company a significant reduction in the power demand for your data centers. The National Renewable Energy Laboratory (NREL) is the nation's primary laboratory for renewable energy and it recommends the Maisotsenko Cycle, which “significantly reduces electric demand for any cooling applications”. The Maisotsenko cycle (M-Cycle) has broken through the thermo-dynamic barrier, cooling air or any fluids below wet bulb approaching the dew point temperature. This far-reaching achievement will bring energy efficiency gains unimaginable before now to the air-cooling, liquid cooling and power production industries (see our U.S. Patents No 6,497,107; 6,581,402; 6,705,096; 6,776,001; 6,779,351; 6,854,278; 6,948,558; 7,007,453; 7,197,887; 7,228,669; etc.). Today the M-Cycle assists Federal agencies reach their energy-use reduction goals and it has been successfully tested and researched for cooling applications by NREL (FEMP), SMUD, Delphi, PG&E, Sanwa (Japan), etc. Since then, this product received wide recognition from all over the world: Coolerado Cooler won the International 2004 R&D 100 award, the US Green Builder 2006 Top Ten Product award, the 2007 Sustainable Business Silver Medal of Honor award, and the 2007 Sustainable Business Silver Medal of Honor award, the History Channel and Invent Now Award 2007, and just recently, the BUILDING PRODUCTS Top 100 Winner for 2008. Coolerado air conditioners can be found in markets around the world- in Japan, Europe, and South America, Singapore as well as in the USA from Washington to Florida. Realizing the M-Cycle through our apparatuses it is possible to reduce of the temperature of any coolant from outside condition (for example, 80°C or 175°F) approaching the incoming air's dew point temperature (for example, 12°C or 55°F). So our apparatuses are like refrigerant machines, which produce the cold air (or liquid), but cheaper, more compact, less consumption of energy (five times) and more reliable. After producing the cold coolant through the Coolerado Coolers, it is necessary to distribute it for rejection heat from existing racks of data centers containing computing units. The M-Cycle combines heat exchange and evaporative cooling in a unique regenerative indirect evaporative cooling process that results in product flow temperatures approaching the incoming air’s dew point temperature. Thermodynamically, the M-Cycle’s working air (air rejecting heat) is pre-cooled before passing through the heat rejection water evaporating area, so the difference between the enthalpy of the working air at it’s dew point temperature and the same working air saturated at a higher temperature is used to provide cooling capacity to reject heat. But only the M-Cycle has additional unique properties, which particularly important for data centers, electronics cooling and microclimate systems: 1) The M-Cycle’s cooling capacity increases (instead of decreases for all existing cooling cycles), when the temperature of the fluid being cooled increases. 2) The M-Cycle, producing cooling capacity to reject heat through the cold product flow, simultaneously produces the working air flow saturated at a higher temperature, which has huge absolute humidity. This saturated working air can be used (if it necessary) to keep for computer room right level of relative humidity within the 40% to 55% range. In this case we don’t need to install a humidifier. Please allow me a brief introduction of our companies. Coolerado LLC and Idalex Inc. are emerging technology firms. We have patented the M-cycle for many practical applications. The world leader in automotive thermal technology Delphi Corp. has since license the manufacturing right to produce our advanced thermodynamic heat and mass exchanger, and has begun to mass-produce this product (Coolerado Cooler) for air conditioning applications. For more information about the M-Cycle, please visit our web sites at www.coolerado.com and www.idalex.com Dr., Prof. Valeriy Maisotsenko Chief Scientist, Idalex Technologies Inc. www.idalex.com vm@idalex.com

  18. Cfulton

    Robb your comments are bang on. This is not our grandfathers heat wheel design. Application of what anyone thinks they know about heat wheels needs to be suspended until they take the time to understand this design. I have seen this first hand and taken it through it's paces with clients and top notch engineering firms in due diligence. This is the real deal. Take a look, if you are serious about energy savings and open to new ideas. I have stood with some of the best in the industry as they shook their heads in amazement having arrived as doubters. They left convinced. Your information and view on the matter is accurate.

  19. We have been doing this for a while now in Australia to take advantage of low ambient conditions at night to provide economy cooling in sealed data centers and telephone exchanges. Our company produce a unique and internationally patented plate heat exchanger that combines both cross and counter flow and has eff up to 80% in both a total and a sensible only model, our units a substantially smaller than wheel based units. On the sensible model we have 0% cross contamination due to the design, this is then used on a closed loop as mentioned in the article, the humidity in the room is not changed as you are not introducing any moisture. We make a small in ceiling unit that does 150 l/s right up to a 10000 l/s unit. If your interested in more info have a look at our website air-change.com

  20. Cfulton

    I just realized after looking at Alex's post that it was perhaps not clear that these KyotoCells are high capacity. They are available in 300kw and 600kw designs at a 12C Delta T and that wheel output doubles as that increases to 24C with density rise in IT gear. The cells carry full capacity DX as well or they can use existing water chilled plant. Now these cells can be aggregated to multi-megawatt designs. This is not just theory it's a fundamental part of the control system. One sophisticated enough to allow adhoc loading of the DC floor and deal with Delta T differentials. One that is totally failsafe and can self adjusting. In the test site they dial up 600kw from standing start and the system adapts to the load in minutes. Not many people drop .6 Mwatt on the floor in 15 minutes but its nice to see a cooling solution just deal with it. So size does matter if you are a large DC and these kinds of systems are for mission critical 600kw and up sites mostly. The control system supports this aggregation with balanced load and n+x redundancy for whatever tier level. So no plate exchanger will do that easily or has the capacity of one of these without struggling with some issues that will arise. Inside Air volume on the 600kw KyotoCell runs to 5M Cubic Ft/hr 150,000 cubic meters or 41,667 L/s. That's a reasonable size to give pause to most plate proponents. This would all be theory if they did not have an advance control design that solved a bunch of problems along the way to make this intelligent, failsafe, plug and play and elegant. It's a solution and it works, and it delivers PUE sub 1.2 designs without opening the windows for large DC's that want best in class quality.For N+2 designs assuming high efficiency UPS (rotary coupled) in New York an annual aggregate PUE based on 10 year bin data of 1.18 is doable with this system. Denver can achieve annual average PUE 1.17 designs. Dallas can achieve annual average PUE of 1.27. San Diego 1.17, Jacksonville 1.25. These PUE's hold for partial loads too. No outrageous issues at quarter or half loads.

  21. Very impressed to hear that the majority of the users are considering this technology. We have been manufacturing heat wheels since 1985. We can provide capacities from 500 to 150,000 CFM. With a level of relibility unequaled in the industry. Zero Failures. Thermotech Enterprises.

  22. sk lee

    I need your data for heat wheel Pls send me e-mail Thankyou

  23. DCtech

    KyotoCooling received patent protection from the US Patent office last year for it's unique design. This patent comes on the heals of other international patent grants. US PATENT 7,753,766 B2 was granted July 13 2010. The design is indeed novel. The approach is protected under law.

  24. ORias

    I'm bringing this back from the dead. I've been doing a bit of reading on the various air side economization technologies designed specifically for data center application and would be interested to hear some opinions on Schneider Electric's EcoBreeze vs Kyoto.

  25. Maryoto

    Our building has 2 units Exchanger, like Fin Exchanger is very dirty so that air circulation is not maksimal.I want to get what it costs to service Fin Exchanger per unit? and technicians is there for it which is located in Indonesia? Thank you.

  26. KyotoCooling now has full Domestic US Manufacturing, Design & Engineering and Support. KyotoCooling announced today an Exclusive License Agreement with Air Enterprises Ohio, a best in Class Air Handler Manufacturer and Owner of Thermotech the Leading US Heat Wheel Manufacturer. Simultaneously KyotoCooling BV announced acquisition news by Cloudsite Development that provides for Accelerated EMEA Growth.