Patent Application
20130205822
The invention relates to a system and a method for cooling a computing system, in particular for cooling a server farm.
Computing systems, in particular server farms comprising a large number of racks, generate large amounts of heat during operation. For example, a server farm typically has a pure heat output of several kilowatts. In order to discharge these large amounts of heat, generally, air conditioners are used which are very energy intensive in operation.
There are approaches known from practice which attempt to use the waste heat of a computing system to heat a building. However, often such an approach is not feasible for lack of heatable building surfaces in the proximity, and depending on the climate zone and season it is not suitable to provide for adequate cooling of an existing computing system in an building. Moreover, in the summer heating energy for heating buildings is often not required.
It is estimated that the energy needed for server farms will increase up to 100 GWh or more in the next few years, with up to 40% of this energy being attributed to cooling alone.
Conventional computing systems are usually cooled through the air conditioning of the room, and the individual computers release heat to the ambient air via a fan.
But there are also more recent approaches in which the racks of a computing system are cooled using a liquid, wherein the cooling air of the racks transfers its heat energy to a liquid to be cooled outside the rack via a heat exchanger integrated in the rack or arranged adjacent to the rack.
Another approach is to directly discharge the heat energy from the processors using a liquid cooling circuit. A direct discharge of heat energy is to be understood as a configuration in which liquid-cooled heat sinks are in direct contact with the processors. This method, though it has the advantage that a majority of the generated heat can be discharged from a small local volume, has not yet been implemented in practice, at least not on an industrial scale, possibly due to the fact that the technical difficulties associated with liquid cooling of processors, such as adequately ensuring tightness, are still in no reasonable relation to the benefits.
In view of the ever increasing computing power in less and less space it can be assumed that the cooling load and associated energy consumption will also increase.
Therefore, an object of the invention is to reduce the energy requirements of a conventional cooling system for computing systems.
The object of the invention is already achieved by a system for cooling a computing system, by a computing system, and by a method for cooling a computing system according to any of the independent claims.
The invention relates to a system for cooling a computing system, which comprises a refrigeration machine. A refrigeration machine commonly refers to a device which is used to produce cold, i.e. a temperature that is lower than the ambient temperature.
The invention especially relates to compression-type refrigeration machines, i.e. refrigeration machines having a mechanical compressor by means of which the coolant is liquefied and can subsequently evaporate in the cold section of the refrigeration machine, whereby it cools down and produces the cooling effect.
However, the invention also relates to any other types of refrigeration machines, in particular sorption refrigeration machines such as adsorption or absorption refrigeration machines, in particular refrigeration machines operating on the principle of absorptive dehumidification, also commonly referred to as a desiccant cooling system (DCS), refrigeration machines operating on the magnetocaloric effect, refrigeration machines operating with Peltier elements, geothermal refrigeration machines, steam jet refrigeration machines, refrigeration machines operating on the Joule-Thomson effect, and/or refrigeration machines operating on the principle of evaporative cooling.
According to the invention, the computing system comprises at least a first and a second cooling circuit, wherein the first cooling circuit is operable via a liquid and/or via heat conduction. That means, the first cooling circuit is not an air-based cooling system. Rather, the cooling effect is accomplished using a liquid such as water, or by heat conduction whereby the heat is directly discharged from the heat-generating components via components having a good thermal conductivity. Heat conduction also refers to the use of heat pipes which remove the heat more quickly due to a condensation and evaporation process. Further, according to the invention, at least the second cooling circuit which usually is an air-based cooling circuit, is connected to a cold section of the refrigeration machine.
The invention in particular suggests to cool components which are operated at high temperature, in particular the processor, using a liquid-based or heat conduction based cooling circuit. Due to the high temperatures, in particular a temperature above 50° C., at which these components can be operated, it is often possible to discharge the produced heat to the outside without the use of a refrigeration machine, or to further use it as useful heat for heating purposes and hot water preparation.
The processor cooling circuit in the sense of the invention may not only include the main processors of the computing system, rather the processor cooling circuit may also include additional processors and electronic devices such as memory circuits, hard disks, chip sets, power components of the power supply, which in turn are included in different components of the computing system such as in server racks, telecommunications equipment, power supplies, data storages, and other components of the computing system.
Hence, the computing system in the sense of the invention not only comprises the servers but also other power supply components, especially power adaptors and emergency power supplies, communication modules, data storages, etc.
The second cooling circuit which is typically configured as an air-based cooling circuit and is connected to the cold section of a refrigeration machine, now only needs to remove the energy which cannot be discharged (or is not discharged) through the first cooling circuit which is operated at a much higher temperature. The second cooling circuit typically operates at a feed flow temperature which does not substantially exceed 20° C., in particular at a maximum of 20° C. Specifically, this cooling circuit may be configured as a closed system to which the components of the computing system are connected. Since, now, the amount of heat to be dissipated over the comparatively inefficiently operating refrigeration machine is low, the cooling energy required for the system can be reduced considerably.
The invention thus relates to a system for cooling a computing system having a plurality of cooling circuits, in particular at least two cooling circuits, wherein by distributing the total thermal energy to be discharged to a plurality of cooling circuits a distribution is effected to different temperatures of these individual cooling circuits, which permits, based on the respective temperature of this thermal energy, to cool a fraction or several fractions of the total thermal energy to be discharged in a particularly efficient manner, or to supply it for a further use.
In one embodiment of the invention, a return flow of the first cooling circuit can be connected with both a heat exchanger and the cold section of the refrigeration machine. For example, the heat exchanger may be mounted to the outside of a building. However, a heat exchanger in the sense of the invention also refers to a reuse of the cooling fluid, for example for generating useful heat. Due to the fact that the first cooling circuit, in particular the processor cooling circuit, can be connected both to a heat exchanger, in particular an outside heat exchanger, and to the cold section of the refrigeration machine, it is possible to selectively distribute the amount of heat which is to be discharged through the refrigeration machine and which is to be discharged through the heat exchanger, in particular via a directional valve. So in order to cool the first cooling circuit the refrigeration machine has only to be used if, for example due to an elevated outside temperature, cooling via an externally arranged heat exchanger is no longer possible. Thus, the energy-intensive use of the refrigeration machine is reduced to a minimum, while the system enables to provide reliable cooling even in case of very high outside temperatures.
In one embodiment of the invention, the cooling fluid may be passed through nearby heat exchangers which are connected to the printed circuit boards and so cool the printed circuit boards and/or the devices thermally coupled with the printed circuit board. Also, the cooling fluid may be passed through the printed circuit boards themselves.
One embodiment of the invention comprises at least three cooling circuits, one cooling circuit thereof being operated by air and the other two cooling circuits being operated by means of a liquid or heat conduction, and at least one cooling circuit of the other cooling circuits, i.e. the liquid-based cooling circuits, can be connected both to an external heat exchanger and to a cold section of the refrigeration machine. The so defined system effectively operates as a three-stage system. Specifically, it is intended to provide three cooling circuits with different feed flow temperatures.
For example, the processors of the computing system may be cooled by a first cooling circuit having the highest feed flow temperature. This cooling circuit usually does not required the support of a refrigeration machine, rather it is possible, at least in temperate climates, to remove the heat to the outside via an external heat exchanger. Alternatively, it is possible to use the high temperature level for other purposes, for example to heat buildings, or to produce hot water or electricity. It will be understood that it may yet be useful to configure the system in such a way that the fluid of this cooling circuit may likewise be fed to the cold section of the refrigeration machine in order to ensure reliable cooling even at extremely high outside temperatures.
Another liquid-based cooling circuit is operated at a feed flow temperature which is between the feed flow temperature of the above mentioned cooling circuit and the feed flow temperature of a further, in particular air-based, cooling circuit. Another group of heat-generating components which are cooled using a lower feed flow temperature than the components of the first cooling circuit can be connected to this cooling circuit. These may be hard disks and storage devices, for example.
This second cooling circuit whenever possible uses an external heat exchanger so that the use of the refrigeration machine can be dispensed with. So depending on the climate zone it is possible, at least in the winter months, to operate even this other cooling circuit without using a refrigeration machine. In case of elevated outside temperatures, on the other hand, recourse is made to the refrigeration machine by selectively distributing the coolant.
Another cooling circuit is usually air-based and operates at a lower feed flow temperature than the two abovementioned cooling circuits. This cooling circuit for example cools the air in the racks or even the air in the room in which the computing system is installed. Since for this purpose, generally, temperatures of 20° C. or below are needed, this generally requires the use of a refrigeration machine. However, because of the heat discharge over the two other cooling circuits it is possible to considerably reduce the use of the refrigeration machine. It will be appreciated that depending on the size and configuration of the system, more other cooling circuits may be provided at intermediate temperatures in order to optimize the system so that as much heat as possible may be discharged without using refrigeration machines.
Furthermore, as suggested according to another embodiment of the invention, the system may comprise means for selectively distributing the cooling fluid within the computing system.
In particular, it is suggested to distribute the cooling capacity within the computing system in function of the workload thereof. According to one embodiment of the invention, the system for cooling the computing system may be connected with the computing system itself, to optimize the distribution of coolant. It is conceivable, for example, that at least some individual servers of the computing system report, via an interface, the respective work load and/or the respective temperature to control electronics of the cooling system, so that the cooling system is controlled in function of the work load, in particular regarding the local distribution of different processing powers or waste heat generation within servers, racks, components of the computing center and within the computing center itself.
One advantage of such a control system is, inter alia, that additional cooling capacity is already requested at a very early stage, namely immediately upon an increase of utilization of the computing system. In this way, for example in cooling systems including a cold storage, the required capacity of thermal storages for storing cold may be reduced by having the refrigeration machine already switched on as soon as a need for additional cooling energy is foreseeable due to the utilization of the computing system (and not only when the temperature in the computing system has already risen) and thus reducing the time to be bridged by a cold storage between the demand of cooling energy and provision thereof by the refrigeration machine (the refrigeration machine usually requires a few seconds or minutes from its activation to provide the cooling energy). Smaller thermal buffers permit more compact configurations, which may be of advantage especially in case of cooling modules integrated in the computing system, e.g. in racks, or connected to the racks.
Furthermore, it is conceivable in case of an increase of computing power to increase the amount of coolant supplied and/or to reduce the temperature thereof. In particular it is also conceivable in case of increasing computing power to operate at least one coolant circuit at a lower temperature.
Moreover, in contrast to simpler control systems which are controlled for example based on the return flow temperature, it is possible in case of a lower work load to operate the computing system at a significantly reduced power without the risk that in case of a sudden increase of the work load overheating of individual components occurs due to the low cooling capacity supplied.
A temperature monitoring program may be installed on the individual computers to control the cooling system, which runs in the background and reports increasing cooling requirements to the controller of the cooling system. As an interface in the context of the invention an already existing LAN port may be used, for example.
Also conceivable is remote monitoring and/or control of the cooling system via a network, in particular based on the internet.
In one embodiment of the invention, both the first cooling circuit and the hot section of the refrigeration machine each include a heat exchanger, which heat exchangers are thermally coupled via another heat exchanger, so that the waste heat from the two cooling circuits may be discharged collectively. This embodiment of the invention is based on the realization that the hot section of a refrigeration machine, in particular a compression-type refrigeration machine, through which generally the heat extracted from the system is discharged to the outside, and a processor cooling circuit may be operated at similar feed flow temperatures. If now the two cooling circuits are coupled via heat exchangers, the generated heat may be discharged to the outside using a single heat exchanger. An advantage thereof is that only one external port has to be provided. This is particularly advantageous in conjunction with the modular configuration described herein, or with the embodiment in which the refrigeration machine is integrated in the server or directly connected to the server.
In one embodiment of the invention, the system comprises at least redundantly configured pumps for distributing the cooling fluid and/or a redundantly configured refrigeration machine. At least in larger server farms, even in case of a failure of individual components of the cooling system permanent continued supply of the cooling fluid has to be ensured for a long time, for which purpose buffering storages are generally not sufficient.
For emergency cooling, an additional conventional refrigeration compressor or a supply of cold tap water into the system may be provided, for example.
Furthermore, the cooling circuits themselves may be configured redundantly, so that for example in case of a loss of coolant in a cooling circuit the heat may be removed via a cooling circuit redundant thereto.
In one embodiment of the invention, the system may be integrated into an existing air conditioning system of a building and/or into a hot water supply system and/or into an electricity supply system. For example, it is conceivable that the process heat which results from a refrigeration machine is used at least partly to heat the building and/or to provide hot water. It is also conceivable to use the process heat to generate electricity, for example using Peltier elements. The term process heat refers to any kind of energy removed from the cooling system, also referred to as recooling.
However, it is also conceivable, as suggested according to another embodiment of the invention, to use a refrigeration machine which is operated using heat as driving energy. This is in particular the case with sorption refrigeration machines. In such an embodiment of the invention it is possible to connect the first cooling circuit which operates at a higher feed flow temperature to the hot section of the refrigeration machine so as to provide driving energy. The second cooling circuit may then be connected to the cold section of the refrigeration machine to cool the air in the servers, for example. Especially advantageously, the fluid discharged from the hot section of the refrigeration machine is passed through an external heat exchanger. This is because this fluid usually still has a sufficiently high temperature so that heat can be discharged externally to the outside without the use of refrigeration machines (depending on the particular, for example climate-related, outside temperatures).
In a preferred embodiment of the invention, the system has a modular configuration and comprises at least one cooling module in which at least the refrigeration machine and an electronic controller are arranged. It is in particular suggested to provide a module which comprises a housing with ports, and to which, in addition to a power supply and optionally the connection of the computing system via an interface, the feed and return conduits of the cooling circuit of the computing system can be connected. Furthermore, the module preferably comprises ports for an external heat exchanger through which process heat from the refrigeration machine is discharged.
The controller of the system for cooling a computing system is also integrated in the module.
The modules are preferably sized according to the system dimensions of components of the computing center, e.g. as a 19″ system for rack components.
Ports for electric power supply, for communications and/or for cooling conduits are preferably configured so as to be automatically connected when the module is inserted. In this way, a faster installation is realized for maintenance and replacement purposes.
Moreover, in a preferred embodiment of the invention, the module has at least one independent emergency power supply which at least ensures the operation of the pumps which supply the cooling fluid to the computing system, even in the event of a power outage. It is conceivable to additionally connect at least one control electronics of the cooling system to the emergency power supply. For simpler control is also possible to drive the pumps such that they continue to run in case the control electronics is switched off.
Alternatively, the emergency power supply of the cooling system may be ensured by an emergency power supply of the computing system, in particular by an uninterruptible power supply.
Computing systems generally have an uninterruptible power supply. Such uninterruptible power supplies of computing systems are usually also cooled. Therefore, it is intended to use the cooling system also for the uninterruptible power supply. An uninterruptible power supply generally comprises at least accumulators which can bridge momentary interruptions of the mains voltage. The uninterruptible power supply usually starts up within a few milliseconds, so that even short term voltage disturbances are compensated for.
A computing system usually also has telecommunications devices, such as modules for connection to a telecommunications network. It will be appreciated that the cooling system according to the invention, if necessary, also ensures cooling of these telecommunications modules.
In another embodiment of the invention, a liquid of the second cooling circuit, after having passed through a cold section of the refrigeration machine, may be fed through a heat exchanger to cool the air prevailing in the computing system. That is, as already described above, the second cooling circuit is air-based, and the air is cooled down by a heat exchanger which is for example integrated in a rack of the computing system.
In this embodiment of the invention, the liquid after having passed through the heat exchanger is fed to the first cooling circuit. Therefore, this is an embodiment in which the two cooling circuits are connected in series, so that the cooling fluid first supplies cold to the second cooling circuit which is operated at a low feed flow temperature, and is then fed to the first cooling circuit, in particular the processor cooling circuit, at a higher temperature.
In one embodiment of the invention, the refrigeration machine is integrated in a rack or in a server. This particularly permits refrigeration machines to be accommodated in the system in decentralized manner. Also, this may permit a server to cool itself, for example. In this case each cooling circuit may be optimized to the specific individual device. A port for a cooling circuit in the sense of the invention refers to any type of interface through which heat energy can be transferred.
In one embodiment of the invention, refrigeration machines are provided integrated in or immediately adjacent to a server, the refrigeration machines being configured as a module.
Preferably, each module comprises heat exchangers, controllers, pumps, interfaces, and the refrigeration machine itself. However, providing just the refrigeration machine as a module is also conceivable.
It is also possible, as suggested according to another embodiment of the invention, that the refrigeration machine (preferably implemented as a module) is arranged immediately adjacent to the server or rack. For example, the refrigeration machine may be disposed above or below a server or rack. So no additional footprint is required. It is also possible that the cooling module is arranged laterally, also, one refrigeration module may supply several components such as racks with cooling energy.
As is known, heat exchangers and fans for generating an internal air circulation for cooling a rack, for example, such as those of a rack cooling system, may be arranged both within or adjacent to a rack, for example. So in another embodiment of the cooling module, an arrangement of the cooling module in such a unit for internally cooling for example a rack is possible.
In one embodiment of the invention, the refrigeration machine is integrated in a server, in particular a blade server.
According to one embodiment, the refrigeration machine is configured as a module that is insertable into a server, in particular as a plug-in module. This embodiment of the invention may be used for conventional blade servers, for example.
Integration into the server or the computing system, or an arrangement directly adjacent thereto allows for short cable lengths and transfer paths for the coolant (liquid, heat conduction, air), thereby reducing thermal losses of the cold transfer, furthermore reducing the energy required for the cold transfer (e.g. the pump power), and hence increasing efficiency. Furthermore, integration or adjacent arrangement of the cooling modules allows a modular configuration of the computing center with respect to the cooling system; the components (e.g. racks) each comprise a separate cooling system tailored to the component. Thus, the computing center can be extended without the need to extend a central cooling system or to expand the cooling capacity thereof (except for the discharge of process heat). Furthermore, a computing center is conceivable that does not require a central cooling system (except for the discharge of process heat). Another advantage of integrated or adjacent cooling modules, depending on the embodiment, is a reduction of external ports for the cooling circuits when several cooling circuits are already combined in the cooling modules. So, for example, only one fluid port is needed as an external port to discharge process heat. Otherwise, all components can be integrated in the server or the rack.
A system for cooling a computing system may be configured such that non-adjacent and non-integrated cooling modules or refrigeration machines and adjacent or integrated refrigeration machines or cooling modules are used in an optimum combination in terms of investment costs and operating costs.
In one embodiment of the invention, the system for cooling a computing system comprises a system for detecting a loss of coolant and for initiating an emergency shutdown in a loss of coolant event, which may also be configured as a module.
In particular, the means for emergency shutdown are configured as a power supply interrupter.
For example, in a liquid-based cooling system which is preferably operated at a positive pressure for pressure detecting purposes, it is conceivable to centrally detect based on the pressure if a fluid leak exists.
Then, in the event of a drop of pressure, for example the entire computing system or portions of the computing system and/or the pumps for circulating the coolant may be switched off, so that the components are disconnected from power. This ensures that at worst components are damaged which come directly into contact with the cooling water, and that other damage to components due to the electric conductivity of the cooling water is avoided by disconnecting the power supply of the components.
Furthermore, it is conceivable that the emergency shutdown comprises a pump which in the event a loss of fluid is detected generates a negative pressure in the coolant system. For example, the pump may pump out the liquid into a designated reservoir or into drains. Due to the resulting negative pressure no or only little additional water will leak, so that the damage in the system remains localized.
In another embodiment of the invention, means for emergency shutdown are integrated in a rack of the computing system.
For example, each rack may comprise means for detecting a loss of coolant, in particular a humidity sensor. In case of a liquid leakage, the power supply of the rack is disconnected. It is likewise possible to provide the rack with automatically closing valves, so that it is separated from the coolant circuit. An advantage of this embodiment of the invention is that in this way not the entire computing system fails, and at the same time leakage of larger amounts of fluid from a rack is prevented.
Furthermore, the module for detecting a loss of coolant may be a component of a cooling module.
In one embodiment, the invention provides for a system for cooling a computing system that comprises a plurality of cooling circuits, wherein one cooling circuit is coolable without using the refrigeration machine and another cooling circuit is coolable using the refrigeration machine. In particular, it is suggested to discharge the waste heat of a first cooling circuit having a higher feed flow temperature and higher return temperature without the use of a refrigeration machine by using the outside air for recooling, or by using the heat as useful heat, for example for heating purposes and for hot water supply.
Alternatively, or in combination therewith, at least two cooling circuits are coupled thermally and so are combined to form one cooling circuit.
In particular, it is suggested to connect at least two cooling circuits with a lower and a higher feed flow temperature, respectively, in series. So, for example, the return flow of a rack cooling circuit may be used to cool power components and processors.
In one embodiment of the invention, process heat is dischargeable through the combined cooling circuit. In particular, the return flow of the last cooling circuit which has the highest temperature is supplied to a heat exchanger.
In one embodiment of the invention, the system for cooling a computing system comprises a plurality of cooling circuits, wherein at least in one cooling circuit a bypass is provided, by which the volume flow in the module to be cooled and connected to the cooling circuit may be increased by a partial recirculation of the coolant without increasing the total volume flow of the coolant in the cooling circuit. So it is possible to keep the total volume flow substantially constant, and to direct a portion of the volume flow to detour components to be cooled, via a bypass. If now additional cooling is required, the flow rate in the bypass may be reduced, whereby the flow rate along the components to be cooled increases.
However, vice versa it is likewise possible to reduce the volume flow in the module connected to the cooling circuit without reducing the total volume flow in the cooling circuit.
In one embodiment of the invention, a liquid is used as the coolant which has an electrical conductivity of less than 10*10−6 S/m. Specifically, pure water or a water-glycol mixture may be used. In this way, electrical damage to the components of the computing system and the risk of electric shock to operating personnel are reduced.
In one embodiment of the invention, components of the refrigeration machine, in particular a compressor, may be cooled using at least one of the cooling circuits. Thus, the refrigeration machine is integrated in simple manner into the cooling of the system and is in particular cooled by a liquid.
In particular, by connecting the compressor or the motor of the compressor to one of the liquid-based cooling circuits, the waste heat of the compressor or of the compressor motor (motor heat) may be further used for example as thermal energy (for example for building heating or for generating electric energy). Moreover, this may avoid the need for generating the cooling power to cool the compressor of the refrigeration machine by the refrigeration machine itself (or by another refrigeration machine), by removing the motor heat via another cooling circuit which can be cooled directly in the recooling without any other refrigeration machine, for example because of its high temperature. In this way, the energy required to cool the computing center is reduced.
In another preferred embodiment of the invention, a refrigeration machine is used which comprises a compressor that has a soft start circuit.
A soft start circuit reduces the inrush current and thus reduces the torque of the motor in the start-up phase. In addition to a softer start, the service life of the motor may be significantly increased in this way.
The invention provides for a substantially thermally neutral computing system which merely releases a heating power of less than 20%, preferably less than 10% of the power consumption of the computing system as thermal energy into a room in which the computing system is installed. Therefore, in many cases cooling of the room by an energy-consuming refrigeration machine can be dispensed with.
Thus, the computing system may possibly also be installed in office areas, etc.
For reducing the heat transfer to the environment, a housing of the computing system may be insulated, in particular the housing walls may have a heat transfer coefficient of k<3 W/m2K, preferably k<1 W/m2K.
For this purpose, the housing walls may comprise a thermally insulating material, such as hard foam. It is also conceivable to apply insulating material to the inner and/or outer surface of the housing walls.
Furthermore, the housing walls of the computing system may be coolable to approximately ambient temperature using fluid lines which are connected to a cooling circuit. To this end, the walls may be of a double-walled construction or may include cooling coils. By cooling the walls, unwanted release of heat into the environment can be prevented, even with a temperature in the housing above room temperature.
The invention further relates to a computing system and in particular to one embedded in a system as described above for cooling a computing system. As far as details of the computing system and cooling system thereof have been described above, reference can be made to the corresponding features concerning the computing system, without the computing system necessarily being a part of the system described above.
The computing system comprises a housing in which the components of the computing system, in particular processors, memory, hard discs, etc. are arranged.
According to the invention, the computing system comprises at least a first and a second cooling circuit, wherein the first cooling circuit permits to cool processors and power components of the computing system using a liquid and/or by heat conduction, and wherein the second cooling circuit comprises a heat exchanger arranged in the housing.
By the heat exchanger which in particular can be cooled by a liquid, the inside of the housing is cooled by a cooling circuit, and in this way thermal energy is discharged, which is not removed through the first cooling circuit. The heat exchanger may for example comprise a cooling coil arranged in the rack, or channels in the housing.
Preferably, the computing system includes fluid ports for both the first and the second cooling circuit.
While due to the high temperature the major part of the thermal energy may be discharged through the first cooling circuit by free cooling, it is also possible, other than with the system for cooling a computing system described above, to dispense with the use of a refrigeration machine for the remaining thermal energy discharged through the second cooling circuit, and to also operate this cooling circuit via free cooling.
In this way, the computing system may be configured to be thermally neutral.
The housing is preferably formed as a rack.
The invention further relates to a cooling module, in particular for a system for cooling a computing system as described above.
The cooling module comprises a port for a first cooling circuit, in particular a cooling circuit which permits to cool processors and power components of a computing system using a liquid. Moreover, the cooling module comprises another port for a further cooling circuit. This further cooling circuit permits to cool for example housings or servers of a computing system, at a lower temperature. Furthermore, the cooling module comprises a port for discharging process heat. Via this port recooling is accomplished such that thermal energy is removed from the cooling system.
In a preferred embodiment of the invention, the cooling module comprises a refrigeration machine. The refrigeration machine is in particular intended to provide a sufficiently low temperature for the second cooling circuit which is operated with a lower feed flow temperature.
In one embodiment of the invention, the cooling module is configured as a plug-in module for a server, in particular a blade server.
To this end, the cooling module includes mechanical means to be inserted into a slot. The cooling module has a standard size which occupies one or more slots of a server.
The invention further relates to a housing of a computing system, which is in particular configured as a rack. The housing includes a heat exchanger arranged in the housing, and a fluid port connected to the heat exchanger. Also, the housing walls may be formed as a heat exchanger.
The heat exchanger permits to cool the interior of the housing.
Furthermore, the housing comprises another fluid port to which modules, in particular plug-in modules or power components, may be connected.
The invention further relates to a computing module, which is configured as a plug-in module for a rack. A computing module may comprise processors, for example, but also hard disks, telecommunications electronics, etc.
According to the invention, the rack comprises a fluid port through which processors and power components of the computing module may be supplied with a cooling fluid.
The computing module may comprise a further fluid port for supplying cooling fluid which in particular cools the housing of the computing module, for example using an integrated heat exchanger. However, it is also conceivable to accomplish cooling of the housing merely based on air, so that the housing in which the computing module is disposed is cooled down by a heat exchanger arranged in the housing.
The invention further relates to a module for detecting a leak, in particular for a system for cooling a computing system as described above. The module comprises means for detecting a leak in the cooling system, a controller, and means for shutting down a computing system, at least partially.
Preferably, the module for detecting a leak is adapted to determine the location or size of the leak based on measured parameters such as the pressure in the fluid system, moisture sensors, etc., and then selectively shuts down the computing system or performs an emergency shutdown by interrupting the power supply, in function of the location and severity.
The module for detecting a leak may be incorporated in another component such as the cooling module described above. Likewise, it is conceivable for the module itself to be configured as a plug-in module for a server.
The invention further relates to a method for cooling a computing system, in particular using a system for cooling a computing system as described above.
The computing system comprises at least a first and a second cooling circuit, wherein the first cooling circuit is operated at a higher temperature than the second cooling circuit and by means of a liquid and/or by heat conduction. Specifically, the feed flow temperatures of the two cooling circuits differ by at least 20° C., preferably by at least 30° C.
Furthermore, at least the second cooling circuit is operated through a cold section of a refrigeration machine.
The invention permits to reduce the cooling power generated by the refrigeration machine to a minimum, since the first cooling circuit which for example is configured as a processor cooling circuit as defined above is operated at such a high feed flow temperature that the heat can be removed without the use of a refrigeration machine, at least for the majority of the time.
In one embodiment of the invention, the return flow of the first cooling circuit is temporarily connected both to a heat exchanger and to the cold section of the refrigeration machine, in particular by means of a directional valve.
Therefore, waste heat from the first cooling circuit is only fed to the refrigeration machine if an external discharge thereof, for example via a heat exchanger, is not possible, for example due to high ambient temperatures.
The first cooling circuit preferably provides for cooling processors and/or power components of the computing system, whereas the second cooling circuit preferably cools the racks of the computing system and/or the room in which the latter is arranged.
A heat exchanger also refers to providing the heat of the hot section of the refrigeration machine and/or the heat of the first cooling circuit for useful heat in particular for room heating and/or water preparation.
Shown is a server with two cooling circuits.
The first cooling circuit is a liquid-based cooling circuit and comprises a feed flow line 2 and a return flow line 3, through which liquid-cooled components may be cooled, such as processors and other power components.
Furthermore, the system 1 for cooling a computing system comprises a second cooling circuit in form of an air cooling, comprising a fluid inlet 4, and an outlet 5. This second cooling circuit is coupled with a refrigeration machine (not shown).
The second cooling circuit serves to cool components 7 which are not connected to the liquid-based cooling circuit.
The first cooling circuit is coupled to a heat exchanger, via feed flow 2 and return flow 3, and the heat generated thereby may be used as useful heat for the building. The first cooling circuit may be operated at a higher temperature, for example the target temperature may be 50° C. at the feed flow and 60° C. at the return flow. Due to the high possible feed flow temperature, a refrigeration machine is not necessarily required for cooling.
The second cooling circuit comprising the air cooling, by contrast, is coupled with a refrigeration machine (not shown), since it has to be operated with a lower temperature, for example the temperature is not more than 20° C. at the inlet and 35° C. at the outlet.
However, since much of the energy to be discharged as heat can be removed via the first liquid-based cooling circuit, there are significant energy savings resulting in the system for cooling a computing system.
The saved energy is calculated from the amount of energy discharged via the first cooling circuit divided by the efficiency, or coefficient of performance (COP), of the refrigeration machine.
Since refrigeration machines usually work with poor efficiency, energy savings are considerable.
Referring to
Refrigeration machine 8 comprises a coolant circuit 13 which may be considered as the refrigeration machine's internal coolant circuit. The coolant in evaporator 9 expands, thereby becoming gaseous and causing a temperature decrease. Evaporator 9 forms the cold section of the refrigeration machine. Via a compressor 10, the coolant is fed through the cooling circuit 13 to a condenser. Through an increase of pressure the coolant liquefies and can release waste heat at the condenser to extract energy from the system. Condenser 11 forms the hot section of the refrigeration machine 8. Via expansion valve 12, the coolant is again fed to the evaporator, and thus a closed circuit is formed.
a schematically illustrates a refrigeration machine, in which the internal coolant circuit 13 is connected, via an internal heat exchanger 59, to coolant ports outside the refrigeration machine.
a thus illustrate the possibility of configuring a cooling circuit according to the invention such that it includes, instead of a cooling liquid (for example water), the coolant of the refrigeration machine, wherein cooling is accomplished directly through the evaporator of the refrigeration machine. In this manner, the size of the refrigeration machine can be reduced, which may be important in particular for refrigeration machines integrated in servers, for example.
Referring to
Hot section 19, likewise, comprises an inlet 17 and an outlet 18. The hot section, for example, may have a feed flow temperature of 50° C., whereas the return flow temperature of the cold section is 15° C., for example.
The system 1 for cooling a computing system comprises a first cooling circuit 21.
The first cooling circuit is a liquid-based cooling circuit which serves to cool processors and power components arranged in rack 20.
Through the first cooling circuit 21, heat is fed to the environment via heat exchanger 23. It will be understood that this heat may be used as useful heat, or to generate electric energy.
Furthermore, the system 1 for cooling a computing system comprises a second cooling circuit 22. Second cooling circuit 22 comprises a heat exchanger 24 built into the rack 20 or connected to the rack 20, which serves to cool the air in rack 20 and in a rack-internal air circuit. Cooling circuit 22 is connected to a refrigeration machine 8.
The feed flow temperature of cooling circuit 22 is substantially lower than that of cooling circuit 21. Therefore, the use of refrigeration machine 8 which is in particular configured as a compression-type refrigeration machine is necessary, unless free cooling can be used, as mentioned above.
Waste heat, also referred to as process heat, is discharged to the outside by heat exchanger 25 through the hot section of refrigeration machine 8.
In contrast to the exemplary embodiment illustrated in
A first cooling circuit 21 provides liquid-based cooling, which cools the processors and power components of the computing system 30.
Via port 29 the heat may be provided as useful heat (for example for building heating, hot water supply, or for generating electric energy), or may be discharged to the outside.
The second cooling circuit 22 comprises a heat exchanger 24 preferably arranged in the rack of computing system 30, by which the air in the rack is cooled. Due to the small amount of heat to be discharged, tap water may be used as a cooling medium, for example. It will be understood that it is also conceivable, for example, to preheat the tap water for hot water supply (for example by heat exchangers—not shown), so that the energy extracted from the second cooling circuit may be used, which only results in a return flow temperature of for example below 30° C.
In this embodiment, the cooling fluid cooled by refrigeration machine 8 is first supplied to heat exchanger 24 which cools the air in the rack.
The so already heated coolant fluid is then passed into the first cooling circuit 21 and cools the processors and power components.
In this manner, the cooling circuits are connected in series, and the cooling liquid, for example provided by a refrigeration machine, first passes through the cooling circuit with the lower temperature level and then through the cooling circuit with the higher temperature level. It will be appreciated that more than two cooling circuits can be connected in series in this way, for example the cooling circuits 21, 22, and 38 of server 37 shown in
With reference to the drawings of
In the exemplary embodiment of a system 1 for cooling a computing system shown in
The fluid extracted from the return flow of the first cooling circuit 21 is first passed via a heat exchanger 25 and then fed into the return flow of the warm section of refrigeration machine 8.
This embodiment may also be referred to as a sequential cooling circuit.
In this embodiment, an intermediate heat exchanger 31 is provided. Coupled to heat exchanger 31 is both the first cooling circuit 21 for cooling the processor as well as a cooling circuit 32 which forms the cooling circuit of the hot section of the refrigeration machine. Heat exchanger 31 thermally combines these cooling circuits and couples them to heat exchanger 25 arranged outside.
An advantage of this embodiment of the invention is that thus only two ports are required for connecting an external heat exchanger 25. Because of a maximum temperature difference of 20° C., preferably 10° C., in the first cooling circuit 21 and in cooling circuit 32 of the refrigeration machine this is possible in a particularly simple manner.
With reference to
Referring to
The system 1 for cooling a computing system comprises a first group of heat generating components 34 which are connected to a first cooling circuit 21 which is liquid-based.
A second group of heat generating components 35 which is likewise arranged in server 37 is also equipped with a liquid-based cooling circuit. This additional cooling circuit will be referred to as a third cooling circuit 38 below.
A third group of heat generating components 36 is formed by heat generating components which are not connected to a liquid-based cooling circuit.
These components 36 are cooled through air cooling by a second cooling circuit 22 which comprises a heat exchanger arranged in the rack.
Furthermore, a refrigeration machine 8 is provided having a cold section 16 by which at least the second cooling circuit 22 is cooled. The hot section of refrigeration machine 8 is connected to an external heat exchanger.
Moreover, there is yet another external heat exchanger 23 provided, through which waste heat can be removed to the outside.
Now, the sense of this system is that three cooling circuits are provided that work with different feed flow temperatures. The air-cooled components of the third group of heat generating components 36 require the lowest feed flow temperature. The processors and power components assigned to the first group of heat generating components 34 are cooled with the highest feed flow temperature, in particular with a feed flow temperature of about 50° C.
Therefore, it is usually possible to largely or entirely dispense with the use of a refrigeration machine, at least for the first group of heat generating components 34, and to cool them through external heat exchanger 23.
The second group of heat generating components 35 is cooled with a feed flow temperature which is between that of the first cooling circuit 21 and that of the second cooling circuit 22.
Using valves 33, the cooling fluid of the third cooling circuit 38 may now be selectively distributed to the heat exchanger 23 and the cold section 16 of refrigeration machine 8.
Depending on the cooling power required and the current outside temperature, it is now possible to only have recourse to the refrigeration machine 8 for cooling the third cooling circuit 38 if necessary, for example due to high outside temperatures.
It will be appreciated that, in similar manner, the first cooling circuit may be distributed selectively to heat exchanger 23 and to the cold section of refrigeration machine 8 (not illustrated), in function of the required cooling power and the existing outside temperature.
Thus,
It can be seen that both the first cooling circuit 21 and the third cooling circuit 38 are connected such, by means of the valves, that these cooling circuits are connected to heat exchanger 23.
Therefore, only the second cooling circuit 22 has to be supplied through the refrigeration machine 8.
With reference to
On top of
This could be both a temperature profile of a day as well as a temperature profile of the average temperature in a year.
Below the temperature graph, it is indicated when the refrigeration machine has to be used. Periods in which the refrigeration machine has to be used are marked by vertical lines, while periods during which cooling may be accomplished through an external heat exchanger are indicated by oblique lines.
It can be seen that the first cooling circuit may be operated without using the refrigeration machine for the entire time.
In contrast, the second cooling circuit, i.e. the cooling circuit of the three cooling circuits which is operated with the lowest feed flow temperature, however, has to be operated using the refrigeration machine for a considerable period of time; only at night for example, and/or only in the winter the use of the refrigeration machine may be dispensed with.
The additional third cooling circuit with a feed flow temperature between the feed flow temperatures of the first and second cooling circuits further improves the efficiency of the system. This cooling circuit needs to be operated through the refrigeration machine only at a temperature above 30° C.
a shows, by way of example, a system for cooling a computing system. Server 37 and the three cooling circuits 21, 22, and 38 for the components 34, 36, and 35 of server 37 have been described in conjunction with
However, in contrast to
In this example, the three cooling circuits of the server are connected in series, first cooling circuit 22 with a feed temperature of 15° C. and an outlet temperature of 20° C. This cooling circuit 22 is connected to cooling circuit 38, with a feed temperature of 20° C. and an outlet temperature of 40° C. This cooling circuit 38 is in turn connected to cooling circuit 21, with a feed temperature of 40° C. and an outlet temperature of 60° C. Thus, the connection in series of these three circuits as a whole results in a feed temperature of 15° C. (inlet of circuit 22), and an outlet temperature of 60° C. (outlet of circuit 21). Heat exchanger 25, in this example, provides an outlet temperature of the cooling fluid of 20° C. This cooling fluid is passed to a heat exchanger 56 for free cooling and cools the outlet temperature of the cooling fluid of cooling circuit 21 from 60° C. to 60° C.-ΔT, before the latter is passed to the inlet of the cold section 16 of refrigeration machine 8. Thus, the cooling power that has to be provided by refrigeration machine 8 is reduced.
Process heat is discharged through the hot section of refrigeration machine 8 and by heat exchanger 25.
Processors and power components are connected with a first cooling circuit 21, and the heat therefrom is discharged to the outside through heat exchanger 23.
Here, recourse is made to the sequential cooling described above, in which the return flow of the hot section of refrigeration machine 8 is coupled to the first cooling circuit 21.
That means, the cooling fluid is first passed from the return flow of the hot section of refrigeration machine 8 via the processors and power components.
Then, energy is extracted from the system using an external heat exchanger 25, and the cooling fluid is returned to the hot section of refrigeration machine 8.
In particular, the cooling system is configured modularly to provide for a simple adaptation to the racks or other components of the computing center.
The possible components of a cooling module are shown in the organization chart illustrated in
Especially the refrigeration machine is not forcibly a part of the cooling module, it may be arranged outside the cooling modules or only in one cooling module serving a plurality of racks, the latter comprising another or each comprising another cooling module including the other components.
The cooling system is in particular configured modularly in order to provide for a simple adaptation to the racks or other components of the computing center.
It is also possible that a cooling module comprises a subset of the components shown. The components may be configured as a cooling module, or non-modularly. Therefore, it will be understood that the system may consist, as far as technically feasible, of any combination of the following components.
Specifically, the individual components are defined as follows:
Cooling module 39 comprises a housing 45 with a refrigeration machine 8 and a controller 40 by which the cooling module is controlled.
Furthermore, cooling module 39 comprises a port 43 for processor cooling or for supplying a first cooling circuit.
Also, a port 44 is provided to which the rack cooling circuit may be connected to provide a second cooling circuit.
As described in a previous embodiment, an intermediate heat exchanger 31 is provided, which allows to combine process heat from refrigeration machine 8 and heat from the first cooling circuit to be discharged via port 41.
Cooling module 39 comprises an own internal heat exchanger 46 to cool the cooling module. The air flow 47 is indicated by arrows. This cooling circuit cools the waste heat of the cooling module itself. This waste heat is generated by the components of the cooling module (for example by the motor of the compression refrigeration machine, or by the controller). Due to the internal cooling system of the cooling module, the cooling module is thermally neutral to the outside.
The cooling module may likewise be cooled using existing rack cooling means, for example an air circulation existing in the rack.
Furthermore, the cooling module comprises leak detection means 42 (a module for detecting a loss of coolant), as described in
It will be understood that the cooling module 39 may comprise additional components, such as electronic interfaces and other cooling ports, mechanical connections, for example to be inserted into a 19″ rack system, etc.
a shows an exemplary embodiment of a cooling module 39 which may be attached for example at or in a server or rack (not shown). Here, in contrast to
Assuming a feed temperature at port 41 (recooling) of <20° C., for example, as illustrated, and a feed temperature at port 44 (rack cooling) of 20° C., for example, the coolant may already be pre-cooled at the inlet of port 44, by heat exchanger 56, before being passed to refrigeration machine 8. Therefore, the coolant does not has to be cooled from 20° C. to 15° C., i.e. by 5 K, by the refrigeration machine, but by 5K-ΔT. Thus, the cooling power to be provided by the refrigeration machine is reduced, and accordingly the energy consumption for cooling.
It will be understood that the principle of free cooling is not only applicable in the cooling module, as illustrated, but in the entire cooling system according to the invention. Moreover, the principle of free cooling may be applied in combination with at least two, preferably at least three cooling circuits, as illustrated in
Referring now to
Via a first port (43 in
In order to cool the air inside rack 20, a further cooling circuit 22 is provided, which is connected to the second port of the cooling module (44 in
Using this heat exchanger 24, the air within the rack is cooled. Cooling circuit 22 is connected to the cold section of the refrigeration machine integrated in the cooling module 39 (8 in
Due to the internal cooling of the rack, the rack may be designed to be thermally neutral to the outside. Since the cooling module is also thermally neutral to the outside, as described in
a shows a rack with a cooling module 39, as in
Referring to
The cooling module comprises a controller which is in particular responsible for controlling the pumps and valves and for controlling temperature and humidity sensors. Using this controller and the appropriate pumps and valves, the coolant is controlled, which for example flows to a heat exchanger or to a refrigeration machine, etc. Therefore, the controller is connected with all components which are supplied by the cooling module, for example via a network connection.
Moreover, the controller is connected to a leak controller including a moisture or pressure sensor, by means of which the pumps and/or the voltage can be switched off, if necessary.
Furthermore, the cooling module is connected, via a network connection, with the computing system and with the individual sub-systems, such as individual racks, means for power supply and telecommunications.
Referring to
In this embodiment, cooling module 39 is positioned above the server 20 and is in thermal communication with the server 20, as in
The cooling module 39 comprises the refrigeration machine, a controller, valves, pumps and sensors, a heat exchanger through which the heat of a first cooling circuit and the process heat from a refrigeration machine are combined and can be discharged through port 41.
Furthermore, the cooling module comprises a leak controller and an internal cooling circuit.
A particular advantage thereof is that with this modular configuration only the process heat has to be discharged to the outside via port 41.
Referring to
The system comprises a refrigeration machine 8 integrated in the housing of server 48, in particular a compression-type refrigeration machine.
The cold section of compression refrigeration machine 8 supplies a cold liquid to a heat exchanger 24 arranged in the server, fans 50 generate an air flow in server 48, which is cooled in heat exchanger 24. The temperature may be maintained at about room temperature. Furthermore, instead of the fans, other means for producing a fluid motion may be used, for example means based on the principle of electro-hydrodynamics (not shown).
Through port 41 (feed and return flow), process heat of the refrigeration machine 8 is discharged to the outside.
Furthermore, a first group of heat generating components 34 is coupled with a processor cooling circuit, via port 49 (feed and return flow). Through this processor cooling circuit, a large part of the energy is discharged without the use of refrigeration machine 8.
Another group of heat generating components 36 is not coupled to a processor cooling circuit but is cooled by the cold air in the housing of server 48.
Referring to
Cooling module 39 is configured in correspondence with the modular configuration of the blade server and is likewise plugged-in. In this exemplary embodiment, it occupies two slots of the blade server.
A refrigeration machine 8 provides cold cooling fluid which is supplied to an internal heat exchanger 24 to cool the interior of the housing of blade server 51.
Process heat from the refrigeration machine may be removed through the hot section and port 41.
Furthermore, modules 52 are provided with a processor cooling circuit.
The fluid of the processor cooling circuit does not need to be passed through refrigeration machine 8 but may be discharged through port 44. It is also conceivable to direct the fluid through the hot section of refrigeration machine 8, as with the above-described sequential cooling, or to thermally combine the processor cooling circuit with the discharge of process heat, by means of an intermediate heat exchanger.
In this way only one port is needed for discharging the process heat.
The blade servers 51 comprise only one port for removing process heat.
Otherwise, as illustrated herein by way of example, the blade servers include, inter alia, a plugged-in cooling module as illustrated in
a illustrates a second, air-based cooling circuit for blade servers, in which the air flow 62 is directed through heat exchanger 24. The blade server is designed such that this air flow 62 forms a closed air circulation within the blade server, so that the blade server is or can be thermally neutral to the outside (except for the liquid-based removal of the process heat).
Referring to
It will be understood that instead of the two racks a plurality of further racks may be added.
a shows an embodiment similar to that illustrated in
As can be seen from the legend, a rack may also be understood as another, similar component of the computing system, for example a telecommunications device or a power supply device.
A server may likewise be understood as another module such as a hard disk module, etc.
Furthermore, components of the computing system as well as components of the system for cooling a computing system according to the invention may likewise be accommodated in a container (not shown).
Referring to
Shown is a fluid carrying conduit 54.
The fluid carrying conduit is surrounded by two electrodes 55, 56. If now water penetrates into the region between electrodes 55 and 56, both the capacity and the conductivity between the electrodes changes.
Using an appropriate controller, a leak can be deduced from the conductivity and/or from the capacity between the electrodes.
A similar system may be configured as a sheet structure, as shown in
In this manner, the electrodes may be used as a part of the housing or may be placed at the bottom of a rack or server, for example.
Again, a leak may be deduced based on conductivity and/or capacity.
Referring to
This module comprises means for detecting a loss of coolant and means for initiating an emergency stop.
As illustrated herein, the system may comprise a separate controller which has communication interfaces, interfaces for reading sensors, and for triggering actions as will be described below.
A loss of coolant may be detected based on coolant pressure monitoring (for example in a cooling system that is operated at a positive pressure), or using sensors which can detect liquids (capacitively or resistively, see
An advantage of using sensors which operate independently of the electrical conductivity of the coolant (for example, pressure sensors) is that this permits to use coolants having a comparatively low conductivity (for example below 2*10−8 S/m).
Preferably, a system including a plurality of means and sensors as described above is distributed in and near the cooling module, the servers, racks, other components of the computing center such as power supplies, and connecting lines for the coolant. In this way, in the event of a leak the location of the leak can be determined.
The means of initiating an emergency shutdown may include a communications interface through which components of the computing center and/or responsible personnel is informed about a loss of coolant. Furthermore, the means for emergency shutdown may in particular comprise means for interrupting the power supply of the concerned component (or components) of the computing center (e.g. for the rack), and interfaces for controlling or shutting down pumps and valves, by means of which the emergency shutdown as described below may be effected.
Furthermore, using a leak controller and an associated control system it is possible to determine which interventions must be taken to protect the system (shutdown, partial shutdown, controlled or immediate shutdown).
For this purpose, the leak controller is connected to a switch of the main power supply for a server or a rack. Furthermore, the controller has a communications interface in order to generate a leak message, visually and acoustically, for example on the computing system or on a higher-level control and monitoring system of the computing center, and/or to control other modules, or to co-ordinate a controlled shutdown.
Furthermore, the controller comprises a direct interface for controlling pumps and valves.
Depending on the size, location, and severity of the leak, a system may be shut down in controlled manner, for example, or in the event of an emergency, may be abruptly disconnected from the power supply and shut down.
For example, a situation may arise in which, though coolant is leaking, this does not present an immediate risk of damage to the computing system or its components yet. In this case, the computing system may be shut down in controlled manner and turned off, so that the running applications are closed and data is saved. Optionally, the applications and/or data may be relocated to other computing systems or components thereof which are not affected by the loss of coolant. With such controlled shutdown it can be ensured that an interruption of coolant flow does not result in a local overheating in the components connected to the cooling circuit.
The system for detecting loss of coolant may receive a command for emergency stop through the communication interface. With this emergency stop, the entire computing system or portions of the computing system may be disconnected from power supply, for example, and/or the pumps for circulating the coolant may be switched off. In this way, possible damage to components of the computing system or computing center caused by cooling water can be avoided or reduced. Further, it is conceivable that the emergency shutdown involves a pump by which, in case a loss of fluid is detected, a negative pressure is generated in the coolant system. For example, the pump may pump out the liquid into a designated reservoir or into drains. Due to the resulting negative pressure, no or only little additional water will leak, so that the damage in the system will remain localized. Furthermore, the cooling fluid conduits may be closed using valves, whereby further liquid can be prevented from leaking from the system.
There may also arise a situation in which an immediate danger of damage to the computing system cannot be excluded. In this case, disconnection of the power supply and the pumps may be effected immediately, without previously shutting down the computing system in controlled manner and without closing the running applications and securing the data.
In another embodiment of the invention, means for emergency shutdown are integrated in or adapted to a rack or other component of the computing system.
Especially in the case where the cooling system is configured as an integrated or adapted module, the shutdown of the cooling module and of the component of the computing system or computer center made be effected locally, if necessary, without affecting other components of the computing system or computing center.
The procedure of shutting down and switching off is illustrated in the flow charts of
As soon as a leak is detected in a rack, the computing center will be informed via an electronic interface.
The computing center will then distribute applications that run in the section affected by the leakage to other parts of the system which are not affected by the leakage. Also, the data is backed up.
Subsequently, the affected system is shut down and then separated from the power supply.
In case of an emergency shutdown, for example due to a major release of water, the power supply for a rack is disconnected immediately (immediate shutdown), as shown in
Computing center 55 comprises a plurality of racks 20 which in turn comprise individual modules 52, such as servers, hard disk units, etc.
In this embodiment, modules 52 are coupled by liquid-based cooling to a first cooling circuit 21, through which heat is discharged to the outside via heat exchanger 23.
A second cooling circuit 22 with a lower feed flow temperature, which cools the components of the rack by an internal air circulation, is supplied from a refrigeration machine 8. For this purpose, a heat exchanger is provided within the racks 20.
With respect to the individual racks, the first cooling circuit 21 is combined and the second cooling circuit 22 is combined. This may be implemented through connection of heat exchangers 24 to cooling circuit 22 by connecting them in parallel to cooling circuit 22. It is also conceivable for the heat exchangers 24 to be connected in succession, so that the cooling fluid flows from one heat exchanger to the next (not shown).
Using another external heat exchanger 25, process heat from the hot section of the refrigeration machine is discharged.
Unless otherwise stated, computing center 55 is similar to that of the exemplary embodiment illustrated in
In contrast to
This is possible, because the cooling circuits have a similar temperature.
An advantage of this embodiment of the invention is that consequently only one port has to be provided for discharging waste heat through heat exchanger 25.
In contrast to
The waste heat sections of the refrigeration machines are combined.
The advantage of this embodiment is that for extracting energy from the system, the cooling modules only have to be connected to cooling circuit 53 through which heat is discharged to the outside via heat exchanger 25. It will be understood that this heat may be used as useful heat.
Furthermore, the bypass and the so allowed increase of the amount of coolant flowing through the server permit to achieve a more homogeneous temperature distribution among all the components connected to the processor cooling circuit.
In case of an operating state, for example, in which only one component out of a plurality of components connected to the processor cooling circuit of a server generates much heat energy to be dissipated, and the other components very little, overheating of this component may be prevented by increasing the flow rate of the coolant in the individual server without influencing the coolant flow rate of the overall system.
In this way, the system may adapt to changing computational loads or operating conditions by controlling the cooling fluid in the bypass.
The bypass and the amount of cooling liquid flowing through the bypass may be adjusted by means of controllable valves and controllable pumps. Controlling (not shown) may be accomplished by the server or from outside the server, and temperature sensors (not shown) may also be involved.
It is also possible to provide a bypass across an entire rack instead for individual servers (not shown), for example across heat exchangers 24 of the second cooling circuit 22 or another system of the computing center (for example a power supply). The operation thereof corresponds to that of a bypass across a server.
It will be understood that the bypass is also applicable to more cooling circuits, for example cooling circuits 21, 22, and 38 as illustrated in
A cooling system in the sense of the invention may consist of a number of cooling circuits which may be connected together in different ways. As for example illustrated in
A bypass may also be provided across an entire rack, for example across heat exchangers 24 of the second cooling circuit 22, or across another system of the computing center (for example a power supply) instead for individual servers (not shown). The operation thereof corresponds to that of a bypass across a server.
As another difference to
Computing center 55 comprises a plurality of racks 20 which in turn comprise individual modules 52, such as servers, hard disk units, etc.
In this embodiment, modules 52 are in thermal communication with a first cooling circuit 21, via liquid-based cooling.
A second cooling circuit 22 with a lower feed flow temperature, which cools the components of the rack in an internal air circulation, is supplied by a refrigeration machine 8. For this purpose, a heat exchanger is provided within rack 20. This cooling circuit is only shown in phantom here, the cooling circuit is not shown.
With respect to the individual racks, at least the first cooling circuit 21 is combined.
The exemplary embodiment comprises a thermoelectric generator, or Peltier element 66. The first cooling circuit with a feed flow temperature T1 first extends to a heat exchanger (hot side) 67 of an element for generating electric energy, which is in thermal communication with one side of thermoelectric generator 66. Another heat exchanger (cold side) 68 of the element for generating electric energy is in thermal communication with the other side of thermoelectric generator 66, the return flow temperature of this heat exchanger (cold side) 68 being T2. The heat exchanger (cold side) 68 is in thermal communication, through a cooling circuit, with recooling heat exchanger 25.
Thus, the temperature difference at thermoelectric generator 66 is ΔT=T1−T2. In this manner, electric energy is generated, which in this embodiment is fed as recycled energy 70 via an inverter 69 into the power supply, so that the electric energy required for the power supply of the computing center 65 is reduced by the recycled energy 70, assuming that an approximately constant amount of energy is provided to power the components of the computing center 64.
Thus, thermal energy may be converted into electric energy which is supplied to the computing center, thereby reducing the amount of electric energy required for the power supply of a computing center.
Moreover, the amount of thermal energy to be discharged by recooling heat exchanger 25 is reduced, which results in reduced operating costs for the heat exchanger.
It will be understood that other physical effects for generating electric energy from thermal energy or based on a temperature difference ΔT may likewise be used; for example, instead of thermoelectric generator 66, a mechanical generator based on the Carnot cycle may be used, for example, an ORC (organic rankine cycle) machine which drives an electric generator to produce energy. Also, a Stirling engine may be used. Moreover, a thermo-magnetic generator may be used.
Since in some physical processes for converting thermal energy into a different form of energy, the efficiency thereof is proportional to the temperature difference ΔT provided (for example in the Carnot cycle), the system according to the invention for cooling a computing system which provides a first cooling circuit with a high temperature permits or at least better promotes a conversion of thermal energy from a computing center into electric energy.
The invention enables to considerably reduce the power consumption required to cool a computing system.
| Number | Date | Country | Kind |
|---|---|---|---|
| 10 2010 026 297.8 | Jul 2010 | DE | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/EP11/00617 | 2/10/2011 | WO | 00 | 4/1/2013 |
