The present invention relates to cooling of electrical and electronic components, and more particularly, to the cooling of variable heat loads in electrical, electronic and optical components by pumped two phase loops.
Electrical, electronic and optical components (e.g. microprocessors, IGBT's, power semiconductors etc.) are most often cooled by air-cooled heat sinks with extended surfaces, directly attached to the surface to be cooled. A fan or blower moves air across the heat sink fins, removing the heat generated by the component. With increasing power densities, miniaturization of components, and shrinking of packaging, it is sometimes not possible to adequately cool electrical and electronic components with heat sinks and forced air flows. When this occurs, other methods must be employed to remove heat from the components.
One method for removing heat from components when direct air-cooling is not possible uses a single-phase fluid which is pumped to a cold plate. The cold plate typically has a serpentine tube attached to a flat metal plate. The component to be cooled is thermally attached to the flat plate and a pumped single-phase fluid flowing through the tube removes the heat generated by the component.
There are many types of cold plate designs, some of which involve machined grooves instead of tubing to carry the fluid. However all cold plate designs operate similarly by using the sensible heating of the fluid to remove heat. The heated fluid then flows to a remotely located air-cooled heat exchanger where ambient air or another fluid cools the cold plate fluid before it returns to the pump and begins the cycle again. This method of using the sensible heating of a fluid to remove heat from electrical, electronic or optical components is limited by the thermal capacity of the single phase flowing fluid. For a given fluid to remove more heat either its temperature must increase or more fluid must be pumped. This creates high temperatures and/or large flow rates to cool high power microelectronic devices. High temperatures may damage the electrical, electronic or optical devices and large flow rates require pumps with large motors which consume parasitic electrical power and limit the application of the cooling system. Large flow rates may also cause erosion of the metal in the cold plate due to high fluid velocities.
Another method for removing heat from components when air-cooling is not feasible uses heat pipes or heat pipe assemblies to transfer heat from the source to a location where it can be more easily dissipated. Heat pipes are sealed devices which use a condensable fluid to move heat from one location to another. Fluid transfer is accomplished by capillary pumping of the liquid phase using a wick structure. One end of the heat pipe (the evaporator) is located where the heat is generated in the component and the other end (the condenser) is located where the heat is to be dissipated; often the condenser end is in contact with extended surfaces such as fins to help remove heat to the ambient air. This method of removing heat is limited by the ability of the wick structure to transport fluid to the evaporator. At high thermal fluxes a condition known as “dry out” occurs where the wick structure cannot transport enough fluid to the evaporator and the temperature of the device will increase perhaps causing damage to the device. Heat pipes are also sensitive to orientation with respect to gravity, an evaporator which is oriented in an upwards direction has less capacity for removing heat than one which is oriented downwards where the fluid transport is aided by gravity in addition to the capillary action of the wick structure. Finally heat pipes cannot transport heat over long distances to remote dissipaters due once again to capillary pumping limitations.
Yet another method which is employed when direct air-cooling is not practical uses the well-known vapor compression refrigeration cycle. In this case, the cold plate is the evaporator of the cycle. A compressor raises the temperature and pressure of the vapor, leaving the evaporator to a level such that an air-cooled condenser can be used to condense the vapor to its liquid state and be fed back to the cold plate for further evaporation and cooling. This method has the advantage of high isothermal heat transfer rates and the ability to move heat considerable distances. However, this method suffers from some major disadvantages which limit its practical application in cooling electrical and electronic devices. First, there is the power consumption of the compressor. In high thermal load applications the electric power required by the compressor can be significant and exceed the available power for the application. Another problem concerns operation of the evaporator (cold plate) below ambient temperature. In this case, poorly insulated surfaces may be below the dew point of the ambient air, causing condensation of liquid water and creating the opportunity for short circuits and hazards to people. Vapor compression refrigeration cycles are designed so as not to return any liquid refrigerant to the compressor which may cause physical damage to the compressor and shorten its life by diluting its lubricating oil. In cooling electrical, electronic and optical devices, the thermal load can be highly variable, causing unevaporated refrigerant to exit the cold plate and enter the compressor. This can cause damage and shorten the life of the compressor. This is yet another disadvantage of vapor compression cooling of components.
Existing methods of cooling heat generating devices using a pumped liquid two phase cooling system are disclosed in commonly assigned U.S. Pat. Nos. 6,519,955 and 6,679,081, totally incorporated herein by reference. However, in cooling electrical, electronic and optical devices, often the heat load to be removed changes rapidly almost to the point of being instantaneous. An example is cooling microprocessors where the change from an idle state to burst or full power occurs in much less than a fraction of a second. The same is true in cooling diode lasers where nearly instantaneous changes in heat output can occur. In prior art cooling devices a number of techniques are used to address this rapid change in thermal load. The simplest method is also the most inefficient, that is to operate the fan in the air cooled heat sink or the pump in a single phase cold plate loop to the maximum required flowrate at all times. This is wasteful of energy and can cause premature failure of the fan or pump since they must run at full capacity all the time. Improvements to this brute force approach call for adding variable speed controls to fans and pumps. This adds cost and complexity to the cooling system and sensors are required to tell when more cooling is required.
The situation with vapor compression cycle cooling is even more difficult because the compressor can be damaged by unevaporated liquid due to sudden changes in thermal load. In this case, a suction accumulator should be used to protect the compressor and a sensor at the exit of the evaporator needs to sense when all of the liquid refrigerant has been evaporated or not, at the exit of the evaporator. Then changes to the speed of the compressor must be made to match the load to the compressor output. These variable speed compressors, suction accumulators and associated controls are expensive and complex. Heat pipes are passive devices and can only change the rate at which they remove heat as a function of their inherent capillary liquid pumping rate.
It is seen then that there exists a continuing need for a system and method for cooling variable heat loads in electrical, electronic and optical components.
This need is met by the cooling system and method of the present invention wherein isothermal cooling is provided to large arrays of hard disk drives through the use of a pumped refrigerant loop. The present invention relates to cooling electronic components, using a system and method for controlling the cooling of variable heat loads in heat generating devices related to that disclosed in U.S. patent application Ser. No. 12/0023,970 filed Dec. 19, 2007, and totally incorporated herein by reference.
This invention allows for the cooling of variable heat loads in electrical, electronic and optical components by pumped two phase loops without the high pumping rates required by single phase pumped loops sized to handle the same loads. Also, when compared to heat pipes, dry out is avoided by using this method which will protect the components from damage due to excess heat.
In accordance with one aspect of the present invention, an improved cooling system and method address a hard drive generating heat and required to be cooled. Cooling is achieved by evaporator surfaces in thermal contact with the hard drive. A vaporizable refrigerant is circulated by a liquid refrigerant pump to the evaporator surface, whereby the refrigerant is at least partially evaporated by the heat generated by the hard drive, creating a vapor. A condenser condenses the vapor, creating a condensed liquid rejecting the heat collected from the hard drive. Finally, the liquid exits the condenser and enters a reservoir containing the vaporizable refrigerant, allowing the reservoir to supply liquid refrigerant to the liquid refrigerant pump.
Accordingly, it is an object of the present invention to provide cooling to electrical and electronic components. It is a further object of the present invention to provide such cooling to components by pumped two phase loops without the high pumping rates required by single phase pumped loops sized to handle the same loads. It is an advantage of the present invention to avoid the need to use vapor compression cooling when systems require isothermal cooling of components under varying loads. Still another advantage of the present invention is achieving isothermal heat removal no matter how dense the drive array.
Other objects and advantages of the invention will be apparent from the following description, the accompanying drawings and the appended claims.
Arrays of hard disk drives are packaged for use as mass data storage appliances. The arrays of hard drives are contained in cabinets or enclosures and operate as a system. These arrays of hard drives are used in data centers and other places where computers are located to store large amounts of digital information such as financial records, personnel records, customer information and the like. These storage devices use forced air to remove the heat dissipated by the hard drives when they are operating. Fans or blowers are used to force air through the spaces between the hard drives arrayed in racks in the storage device enclosure. The air heats up as it moves through the hard drive arrays and is exhausted out of the cabinet or enclosure.
Operation of hard drives in any environment is subject to certain temperature limitations, that is, there are maximum temperatures for all the components which comprise the hard drive. When the drive exceeds these temperatures, the drive may become unreliable or even fail to operate. When large numbers of air cooled drives are located in an enclosure or cabinet, the air experiences sensible heating as it moves through the drive array. That is, the air heats up. At some point, the air is no longer able to keep the drives cool enough to insure that the drive remains below the maximum temperature specified by the manufacturer. This air heating limits the number of drives that can be contained in a given cabinet or enclosure.
There is a desire to increase the number of drives in a given cabinet or rack in order to reduce the amount of valuable floor space taken up by mass storage devices as the need for storing digital data has increased substantially. This has led to putting the drives closer and closer together. A consequence of this higher drive density, even when air heating is not at its limit for the hard disk drive array, is an increase in fan power. More and more energy is used by the fans and blowers to cool drive arrays to force air through smaller and smaller spaces between the drives. This adds to the operating cost of drive arrays. Manufacturers have nearly reached the limit of the number of drives which can be packaged in a given volume due to air heating and power consumed by fans and blowers.
The present invention addresses this specific problem and need. Referring now to
Continuing with
At point C on the pressure enthalpy diagram 10, the refrigerant mixture leaves the evaporator(s) and proceeds to the condenser entrance, represented by point D on diagram 10. The connection between the evaporator exit and the condenser entrance is represented by line CD, the line from point C to point D. For some low pressure drop cases, line CD as represented on the pressure enthalpy diagram may be so short as to make points C and D essentially the same point. For illustration purposes, but not to be considered as limiting the scope of the invention, diagram 10 shows line CD with a pressure drop.
The two phase refrigerant mixture enters the condenser at point D and begins to condense, or reject heat, causing the state of the refrigerant mixture to change to a more liquid phase and a less vapor phase. This is also a reduction in the vapor quality within the saturation dome 12. At point E′ (E prime) the vapor has been completely condensed and only a saturated liquid phase is present in the condenser. As more heat is removed from the liquid phase in the condenser, the liquid becomes sub cooled from point E′ to point E to the left of the saturation dome 12.
In
In
In
In alternative embodiments of the present invention, the refrigerant working fluid can be moved from point C to point E using a vapor liquid separator with a condenser. Also, the evaporator as represented by the line from point B to point C may be a single evaporator as described or may by multiple evaporators in series or parallel or a combination of series and parallel flow arrangements. Likewise, the condenser may be a single condenser or multiple condensers rejecting heat to air or another fluid, as necessary. Finally, single or multiple pumps may be used without departing from the intent of the invention. Hence, those skilled in the art will recognize that the scope and purpose of the invention can be achieved with multiple configurations, without changing its essence.
The present invention requires that the circulation rate of refrigerant in the cooling cycle, represented by following the path on the pressure enthalpy diagram 10 represented by state points ABCDEF, be set, so that point C never reaches the saturated vapor line of the saturation dome. That is, point C is always a two phase mixture leaving the evaporator(s). The saturation dome starts at saturated liquid point 14 and extends to saturated vapor point 20 to include all liquid vapor mixtures in between the saturated liquid and the saturated vapor. Furthermore, point C is allowed to move within the saturation dome so that the exit quality of the two phase mixture leaving the evaporator(s) changes with the heat load being removed by the evaporator(s). In this way, rapid changes in heat load are removed from the hard disk drive component(s) in contact with the evaporator(s) without having to change the circulation rate of refrigerant in the cycle. Only the exit quality of the vapor leaving the evaporator at state point C changes. That is, the circulation rate of refrigerant in the cooling cycle is set higher than the maximum required to evaporate all of the refrigerant at the highest design heat load for the system. At no condition will the hard disk drive array heat load evaporate all of the refrigerant and leave no liquid refrigerant in the evaporator(s).
A pumped refrigerant loop, in accordance with the present invention for cooling an array of hard drives, is illustrated in
In accordance with the present invention, the condenser may comprise any suitable condenser such as a liquid cooled condenser, an air cooled condenser, or an evaporative condenser. Furthermore, the liquid refrigerant pump may comprise any suitable pump such as, but not limited to, a hermetic liquid pump. The refrigerant may be any suitable refrigerant, such as R-134a refrigerant.
Specifically, application of the present invention avoids the need to use vapor compression cooling when systems require isothermal cooling of components under varying loads. The use of a pumped refrigerant loop to cool arrays of hard drives has a number of advantages over the prior art. Pumping liquid refrigerant and allowing it to evaporate (two phase heat transfer) when it removes heat is a more efficient method of heat transfer than the single phase heat transfer of blowing air through an array of disk drives 30 and letting it heat up. Not only can heat be removed from the disk drive 30 effectively, it can be transported to a location where it can be dissipated most efficiently. The two phase mixture of liquid and vapor refrigerant can be easily moved to a remote condenser 26 where a small fan can condense the vapor, thus removing the heat from the pumped refrigerant loop. This represents a significant energy savings over using just air to cool drive arrays.
Having described the invention in detail and by reference to the preferred embodiment thereof, it will be apparent that other modifications and variations are possible without departing from the scope of the invention defined in the appended claims.
This is a regularly filed application, based on provisional application Ser. No. 60/966,120, filed Aug. 24, 2007.
Number | Date | Country | |
---|---|---|---|
60966120 | Aug 2007 | US |