Home
The type of cooling for electronic high density equipment is undergoing significant changes. Equipment is becoming more compact, performance losses are increasing. Pure air cooling meets its limits more and more often.
Praxis Profiline spoke to Dr. Adam Pawlowski, specialist in climate control at Schroff GmbH, Straubenhardt, about the chances and possibilities of liquid cooling.
Question: For which application requirements respectively in which application areas does liquid cooling offer itself in your opinion?
Everywhere, where the ambient temperature is above 55 °C or where a performance requirement per area or volume is demanded (for instance with CO2-Lasers, welding equipment), water cooling is often the only alternative. With applications where the generated heat loss is not allowed to enter the room, where a very high IP protection is demanded or where the ambient air is so aggressive, that the use of a special cooling unit with components made from specific materials is not possible or not economically viable, water cooling is also a possibility. More and more it is also used in the PC sector for the cooling of high performance processors and in the server area for the cooling of 19" cabinets which are assembled with blade servers for performance losses to 30 kW. One has to imagine that in a data centre 100 or more of such server cabinets can be located. Alone in the TeleCity Data Centre in Frankfurt this amounts to a cooling performance of 3.2 MW. A very high air volume is required (at a guess around 700.000 m3/h at ΔT = 13 K), in order to transport such energy volumes to heat exchangers. Water would be far more efficient as an energy carrier.
Question: In your experience has liquid cooling been accepted in practice yet?
Yes, in my experience liquid cooling is fully established in mechanical engineering applications. In the field of electronics it is progressing very slowly. The reasons that it is not that widespread yet are mostly commercial:
The biggest advantage of liquid cooling lies in its independence from the ambient temperature at the installation site. Whilst with air cooling it has always to be observed, that the cooling air has a lower temperature than the elements it has to cool, liquid cooling functions independently from the room temperature. The water flow temperature for the cooling of components as a rule is always considerably lower than the ambient temperature.
The next advantage results from the fact, that water has a higher specific heat capacity (related to its volume) by a factor of 4000 in comparison to air. Through this for instance the direct cooling of a processor with fluid is clearly more effective than with air. In order to compensate for this disadvantage with air cooling a larger air volume has to be circulated. This in turn will generate an increase noise level, whereby we come to the next two advantages; liquid cooling is considerably less noisy than air cooling and it is more maintenance friendly, as it does not require the exchange of filter mats for example.
Finally cost effectiveness splays a part too. The profitability studies, which well-known manufacturers of air/water heat exchangers (LWWT) have carried out to find out what is more favourable, air cooling or water cooling, have proved, that the TCO (Total Cost of Ownership) of water cooling spread over the entire area of operation "lie quite considerably below those of conventional climate control".
Question: How can total liquid cooling be realised within the cabinet platform?
Total liquid cooling inside a cabinet, without the aid of a fan in the internal cycle, is hard to imagine, as the performance achievement would be fairly poor. Despite of that some water cooled assembly plates are offered on the market, to make a direct assembly of electronic components possible. In addition direct chip cooling is also on offer. The cooling elements are assembled directly on the chips and are connected via flexible water pipes to the water distribution. Schroff presented such a solution to the public with the example of an AcvancedTCA system as early as 2004.
The so-called "spray cooling" is another pure liquid cooling concept. An alternative to the spray cooling is represented by "micro channel" cooling. For some time now almost every research laboratory in the chip industry is working on the direct water cooling of chips through 30 to 100 µm size micro channels.
Compared with this air cooling with water as energy carrier with so-called air/water heat exchangers is part of today's technology. These are made up of a heat exchanger unit and a fan for the circulation of the air inside the cabinet. The cool water is fed in through the chiller (cold water replacement). The air/water heat exchangers can be assembled on the side or on the top. With cabinets the side panel version is preferred in most cases, so that the air can be extracted horizontally from the rear area, be cooled and then blow the air back in at the front.
Praxis Profiline spoke to Dr. Adam Pawlowski, specialist in climate control at Schroff GmbH, Straubenhardt, about the chances and possibilities of liquid cooling.
Question: For which application requirements respectively in which application areas does liquid cooling offer itself in your opinion?
Everywhere, where the ambient temperature is above 55 °C or where a performance requirement per area or volume is demanded (for instance with CO2-Lasers, welding equipment), water cooling is often the only alternative. With applications where the generated heat loss is not allowed to enter the room, where a very high IP protection is demanded or where the ambient air is so aggressive, that the use of a special cooling unit with components made from specific materials is not possible or not economically viable, water cooling is also a possibility. More and more it is also used in the PC sector for the cooling of high performance processors and in the server area for the cooling of 19" cabinets which are assembled with blade servers for performance losses to 30 kW. One has to imagine that in a data centre 100 or more of such server cabinets can be located. Alone in the TeleCity Data Centre in Frankfurt this amounts to a cooling performance of 3.2 MW. A very high air volume is required (at a guess around 700.000 m3/h at ΔT = 13 K), in order to transport such energy volumes to heat exchangers. Water would be far more efficient as an energy carrier.
Question: In your experience has liquid cooling been accepted in practice yet?
Yes, in my experience liquid cooling is fully established in mechanical engineering applications. In the field of electronics it is progressing very slowly. The reasons that it is not that widespread yet are mostly commercial:
- Cold water is not available.
- The acquisition of a chiller (cold water replacement) is not justifiable economically for a small number of users
- Cold water is available, but the temperature is not sufficient for the additional users. The acquisition of an additional chiller or the up-grading of the existing equipment is not economically viable for the small group of users.
- A large enough group of users exist, but these are spread all over the company, so that the laying of water pipes could not be justified.
The biggest advantage of liquid cooling lies in its independence from the ambient temperature at the installation site. Whilst with air cooling it has always to be observed, that the cooling air has a lower temperature than the elements it has to cool, liquid cooling functions independently from the room temperature. The water flow temperature for the cooling of components as a rule is always considerably lower than the ambient temperature.
The next advantage results from the fact, that water has a higher specific heat capacity (related to its volume) by a factor of 4000 in comparison to air. Through this for instance the direct cooling of a processor with fluid is clearly more effective than with air. In order to compensate for this disadvantage with air cooling a larger air volume has to be circulated. This in turn will generate an increase noise level, whereby we come to the next two advantages; liquid cooling is considerably less noisy than air cooling and it is more maintenance friendly, as it does not require the exchange of filter mats for example.
Finally cost effectiveness splays a part too. The profitability studies, which well-known manufacturers of air/water heat exchangers (LWWT) have carried out to find out what is more favourable, air cooling or water cooling, have proved, that the TCO (Total Cost of Ownership) of water cooling spread over the entire area of operation "lie quite considerably below those of conventional climate control".
Question: How can total liquid cooling be realised within the cabinet platform?
Total liquid cooling inside a cabinet, without the aid of a fan in the internal cycle, is hard to imagine, as the performance achievement would be fairly poor. Despite of that some water cooled assembly plates are offered on the market, to make a direct assembly of electronic components possible. In addition direct chip cooling is also on offer. The cooling elements are assembled directly on the chips and are connected via flexible water pipes to the water distribution. Schroff presented such a solution to the public with the example of an AcvancedTCA system as early as 2004.
The so-called "spray cooling" is another pure liquid cooling concept. An alternative to the spray cooling is represented by "micro channel" cooling. For some time now almost every research laboratory in the chip industry is working on the direct water cooling of chips through 30 to 100 µm size micro channels.
Compared with this air cooling with water as energy carrier with so-called air/water heat exchangers is part of today's technology. These are made up of a heat exchanger unit and a fan for the circulation of the air inside the cabinet. The cool water is fed in through the chiller (cold water replacement). The air/water heat exchangers can be assembled on the side or on the top. With cabinets the side panel version is preferred in most cases, so that the air can be extracted horizontally from the rear area, be cooled and then blow the air back in at the front.
Dr. Adam Pawlowski at his desk
Dr. Adam Pawlowski in the Climate Control Laboratory at Schroff
Cabinet cooling: The LHX 20 plug-in unit from Schroff consists of a self-supporting unit, the air/water heat exchanger,
drop collector, six fans, three-way valve, as well as a control unit with user and display elements
drop collector, six fans, three-way valve, as well as a control unit with user and display elements
Schroff hybrid solution:
combined water and air cooling in an AvancedTCA system
combined water and air cooling in an AvancedTCA system
Question: How sensible are hybrid solutions with air and liquid cooling and how would they look?
Hybrid solutions are often found for climate control. Everywhere, where air cooling alone is no longer sufficient, hybrid solutions are chosen. The simplest example is the air/water heat exchanger, where at least two cooling concepts are combined: air cooling in the internal cycle and water cooling for the outer cycle. Strictly speaking the outer cycle should often be seen as a hybrid solution, as the cold water has to be cooled again through a cooling unit. A good example of hybrid cooling is a combination of a cooling unit and a vortex pipe in the outer cycle. Such a principle comes into force, when the ambient temperature is too high for the cooling unit and no other cooling possibility exists. One also talks of a hybrid solution, when a heat pipe is cooled with air in order to increase the cooling performance and efficiency.
Question: How does the idea of spray cooling, as realised by LCEC (Liquid Cooled Embedded Computing), work?
This idea is trend-setting. Schroff is working on an enclosure concept for this. A first sample was shown in June 2006 at the GlobalComm in Chicago.
Spray cooling is based on the fact that during evaporation of a fluid more energy can be taken up (approx. 50 to 60 W/cm2) than during the mere warming of a liquid. With the concept at hand the main part of a board is surrounded by a cassette into which via couplings liquid or gas like fluorocarbon (e.g. Fluorinert FC-72) is supplied to or led away from. The electrical connection of the module board with backplane, boards or front panel is ensured via enclosed connectors. These boards can be cooled with a very minimal air flow. Pumps and liquid reservoirs are integrated into the bottom part of the chassis, the re-cooling takes place in an area separated from the enclosure, so that no further climate control is required. The aim of this concept is to use board, which are five times more efficient than today's boards. Therefore an even more improved version has been envisaged already; "direct liquid jet impingement" is based on liquid moistening of the chip surface and is supposed to be capable of leading away 120 W/cm2. Trials show, that the performance of components will not be limited by its cooling in future, but through the efficiency of the connectors.
Question: With regard to fluid cooling concepts are there any standards or guidelines which may help the user?
A new IEC Standard is developed in order to regulate the most important interfaces, for instance for water cooled 19" control cabinet/electronic enclosures. On one hand the most vital interfaces will be defined with this standard and on the other hand it will help users with the layout and design of the water cooling in control cabinets or electronic enclosures.
The standard divides the water cooling of control cabinets/electronic enclosures into three levels.
Modern climate control in the cabinet area and also for instance in the car industry is still based on R134a as cooling agent. This cooling agent does not destroy the ozone layer (ODP), but does influence the atmosphere through the "greenhouse" effect. However, it would be ideally suited for the cooling of cabinets, as the leaking liquid would immediately evaporate in case of a leakage and shorting of electronic components would be avoided. The third generation server from Engera, Blade Frame EX, is cooled in this way.
An adequate cooling agent without the "greenhouse" effect is not in sight at present, if one ignores the faint-hearted trials with CO2. I can see a large user potential here, particularly in the metal industry, but largely economical considerations hinder many companies to undertake the research.
An ideal would be to use the natural resources of the earth for cooling purposes without destroying the balance of nature. What would be available for us here? Earth core heat and the energy of the sun. There are numerous projects that already harness the earth core heat. In contrast the use of solar energy for cooling purposes is hardly looked at. For instance the use of Peltier elements (electrical/air) is not forced enough. To generate electrical energy solar cells could be used here, the cooling of the other side takes place with water or air.
Hybrid solutions are often found for climate control. Everywhere, where air cooling alone is no longer sufficient, hybrid solutions are chosen. The simplest example is the air/water heat exchanger, where at least two cooling concepts are combined: air cooling in the internal cycle and water cooling for the outer cycle. Strictly speaking the outer cycle should often be seen as a hybrid solution, as the cold water has to be cooled again through a cooling unit. A good example of hybrid cooling is a combination of a cooling unit and a vortex pipe in the outer cycle. Such a principle comes into force, when the ambient temperature is too high for the cooling unit and no other cooling possibility exists. One also talks of a hybrid solution, when a heat pipe is cooled with air in order to increase the cooling performance and efficiency.
Question: How does the idea of spray cooling, as realised by LCEC (Liquid Cooled Embedded Computing), work?
This idea is trend-setting. Schroff is working on an enclosure concept for this. A first sample was shown in June 2006 at the GlobalComm in Chicago.
Spray cooling is based on the fact that during evaporation of a fluid more energy can be taken up (approx. 50 to 60 W/cm2) than during the mere warming of a liquid. With the concept at hand the main part of a board is surrounded by a cassette into which via couplings liquid or gas like fluorocarbon (e.g. Fluorinert FC-72) is supplied to or led away from. The electrical connection of the module board with backplane, boards or front panel is ensured via enclosed connectors. These boards can be cooled with a very minimal air flow. Pumps and liquid reservoirs are integrated into the bottom part of the chassis, the re-cooling takes place in an area separated from the enclosure, so that no further climate control is required. The aim of this concept is to use board, which are five times more efficient than today's boards. Therefore an even more improved version has been envisaged already; "direct liquid jet impingement" is based on liquid moistening of the chip surface and is supposed to be capable of leading away 120 W/cm2. Trials show, that the performance of components will not be limited by its cooling in future, but through the efficiency of the connectors.
Question: With regard to fluid cooling concepts are there any standards or guidelines which may help the user?
A new IEC Standard is developed in order to regulate the most important interfaces, for instance for water cooled 19" control cabinet/electronic enclosures. On one hand the most vital interfaces will be defined with this standard and on the other hand it will help users with the layout and design of the water cooling in control cabinets or electronic enclosures.
The standard divides the water cooling of control cabinets/electronic enclosures into three levels.
- Level I - cabinet platform: cabinet with an air/water heat exchanger. A global view point, whereby the entire control cabinet/electronic enclosure is cooled with an air/water heat exchanger;
- Level II - Shelf (subrack) platform: cabinet with an air/water heat exchanger which is located to cool individual subracks;
- Level III - Board platform: The most profound platform, where the water cooling has an impact directly on the boards or chips.
Modern climate control in the cabinet area and also for instance in the car industry is still based on R134a as cooling agent. This cooling agent does not destroy the ozone layer (ODP), but does influence the atmosphere through the "greenhouse" effect. However, it would be ideally suited for the cooling of cabinets, as the leaking liquid would immediately evaporate in case of a leakage and shorting of electronic components would be avoided. The third generation server from Engera, Blade Frame EX, is cooled in this way.
An adequate cooling agent without the "greenhouse" effect is not in sight at present, if one ignores the faint-hearted trials with CO2. I can see a large user potential here, particularly in the metal industry, but largely economical considerations hinder many companies to undertake the research.
An ideal would be to use the natural resources of the earth for cooling purposes without destroying the balance of nature. What would be available for us here? Earth core heat and the energy of the sun. There are numerous projects that already harness the earth core heat. In contrast the use of solar energy for cooling purposes is hardly looked at. For instance the use of Peltier elements (electrical/air) is not forced enough. To generate electrical energy solar cells could be used here, the cooling of the other side takes place with water or air.
