GAS COOLED CONDENSERS FOR LOOP HEAT PIPE LIKE ENCLOSURE COOLING
A cooling device includes an enclosure housing, a primary cooling system including a loop heat pipe like (LHPL) device that rejects heat to an external coolant. The LHPL device including an evaporator module, condenser module, vapor line, liquid return path including a liquid return line, one or more compensation chambers, optional inline storage chambers, a working fluid having a liquid and vapor phase and an optional pump in either the vapor or the liquid return line. The evaporator module including a porous wick that transfers heat from the evaporator shell to the wick including escape channels that help to generate the delta P that drives the working fluid about the cooling loop at the same time helping to prevent the switch from nucleate to transition boiling that takes place on flat surfaces and which makes it possible for an LHPL device to cool heat loads with hot spots releasing more than 100 Watts per square
This application is a continuation in part application of 13/863,379 filed Apr. 15, 2013, 13/863,379 is a continuation in part application of 13/068,029 filed May 2, 2011 13/068,029 is a continuation in part of application 12/103,695 filed Apr. 15, 2008 12/103,695 claims priority to 60/923,588 filed Apr. 16, 2007, all disclosures of which are hereby incorporated by reference in their entireties.
TECHNICAL FIELDThe disclosure relates generally to devices and methods for cooling electronics housed in enclosures and specifically to loop heat pipe like devices.
BACKGROUNDA way that may be employed to efficiently cool servers found in data centers as well as workstations that contain electronics housed in enclosures which employ two phase heat transfer devices we call LHPLs that employ a working fluid loop in which the evaporator and condenser may be separated by significant distances to cool devices that reject a significant percentage of the heat being rejected within an enclosure that may employ a combination of other methods to cool the remaining heat loads more easily cooled that reject heat to liquids or gases or both. Loop Heat Pipe Like (LHPLs) combines the features of Loop Heat Pipes, Capillary Pumped Loops, Advanced Loop Heat Pipes and similar devices.
The cooling of electronics housed in enclosures has for many years been dominated by methods that were much more concerned about getting the job done than the energy it took to get the job done. Methods for improving the efficiency of electronic cooling using passive heat transfer such as heat pipes have been available since at least the Manhattan project, yet have only become inexpensive with the advent of commodity CPUs that rejected 40 or more Watts. These standard heat pipes are typically used to distribute heat from intense electronic sources that employ heat spreaders to reduce the intensity which then use heat pipes to distribute rejected heat to a set of fins. The main purpose of the heat pipes was to reduce the thermal resistance of the thermal conduction path between the heat spreader and the midpoint of the fins. In the case of air cooled LHPLs, they often end-up doing the same thing, except that they make it possible to move the heat to locations where there is little room for large fins to room in enclosures where there is much more room for a large set of fins, or to a location along one of the walls of the enclosure making it possible to remove the heat using less powerful fans and rejecting the heat outside of the enclosure before it may recirculate heating the air coming into the enclosure. There are other advantages to using LHPL devices that include removing heat from very intense sources that may not be effectively cooled by ordinary heat pipes as well as the ability to replace air with a liquid coolant, which may reduce the thermal resistance of air cooled devices by a factor of 3 or more. This disclosure employs passive closed loop heat transfer devices that may dramatically improve not only the energy efficiency of electronic cooling but also makes it possible to cool devices that reject 500 or more Watts mounted on densely packed printed circuit boards and in the case of a data center reduce the energy required to cool it by 80% or more.
The devices at the heart of this disclosure are Loop Heat Pipes, Capillary Pumped Loops and derivatives of Loop Heat Pipes that may include devices like pumps in the condenser lines that reduce the temperature of the devices being cooled. We lump all these devices together into a category we define as Loop Heat Pipe Like (LHPL).
In general the LHPLs employed provide the best energy efficiency of any electronic cooling device ever invented. Not only are they passive, but their ability to reject heat to new locations is often measured in meters while employing very condenser pipes (often less than 3 mm) that make it possible for them to move heat out of congested spaces to condensers that may reject heat to condensers that feature large heat transfer areas rejected to coolants such as air and water. And it is this ability to transport heat to new locations in an enclosure and then pass it through a reasonably long pipe to a condenser which employs a large heat transfer area that makes it possible to efficiently transfer the heat being rejected by tiny hot spots in semiconductors to secondary coolants that remove the heat to cooling loops that ultimately reject waste heat to the outside world. It is also this ability to distribute the primary heat load over large areas that makes it possible to create very efficient counter flow heat exchangers that retain the quality of the heat and make the total heat transfer process so efficient. It is the low total thermal resistances of these devices that make it possible to produce LHPLs whose overall thermal resistance is only 0.15.degree. C./Watt and that have a heat transfer coefficient of 0.15.degree. C./(Wcm.sup.2). In the case of a 100 Watt CPU whose LHP condenser was cooled by water at 30.degree. C. the output from the condenser turned out to be 47.degree. C. with the CPU running at a heat spreader temperature of 59.degree. C. Translated into practical terms, employed in the 1 U rack mount chassis that dominate modern data centers, this device makes it possible to remove the heat directly from a server housed in a rack cabinet and move it directly back to the cooling tower in the data center. In the process, the noisy fans in the racks that produce many points of failure and may consume as much as 30% of the energy being used by servers along with the main CRAC unit blowers and water chillers that consume 35% of the total power employed by the data center end-up getting cut out of the cooling process. The temperature we chose to cool these LHPs with, 30.degree. C. was chosen based on ASHRAE tables, and the performance of commercially available evaporative cooling towers. This temperature turns out to be the temperature of the coolant that this type of cooling tower will produce running in Atlanta Ga. on the hottest and most humid day of the year. A quick comparison of the power consumption at institutions like Lawrence Livermore National Labs suggests that:
TABLE 1—Electronics 50%, Water Chiller 25%, Air Blower 10%, 1U fans 9%, UPS 5% Lighting and 1% changes to this:
TABLE 2- Electronics 83%, 1U fans 1.6%, Cooling Tower Pump and fan 5%, UPS 8.3% and Lighting 1.6%.
Which is to say the total power consumed by the data center goes down by 40%.
The same energy benefits that accrue to data center cooling also accrue to the general cooling of all electronic enclosures that are air cooled, but to a lesser extent, for the simple reason that air is a much poorer heat transport medium than the chilled water that gets employed to move heat from servers that are cooled with it back to the water chiller or in the best case cooling case we have run up against, the data center cooling tower. At the head of the list of benefits in addition to reduced energy costs are huge reductions in noise, the elimination of heat arriving at the walls of enclosures that may be so hot that it is almost possible to get burned touching them, the frequent failure of rotating cooling components including fans and pumps (in the case of pumped liquid cooling) which now occur so often that the systems that employ them have to mount them so that they may be easily swapped out without turning the machine off along with the ability to reject heat loads from devices that produced 500 or more Watts and to cool efficient devices such as CPUs and GPUs mounted in laptops where improved energy efficiency may improve battery life.
To appreciate the benefits of employing LHPLs to cool electronic enclosures, including air and water cooled electronic devices used to do everything from control the operation of space vehicles to reject heat to the cooling towers of data centers, it is first necessary to recite the goals of this disclosure, which were to efficiently cool electronic enclosures in which semiconductor devices that rejected large quantities of heat (greater than 50 Watts) mounted on densely packed PCBs along with devices that shared the same enclosures that rejected the balance of the heat but did not provide a dense source of heat. In the case of air cooled enclosures housed in rack mounted chassis, we wanted to make the first goal achievable while at the same time improving the quality of heat being rejected to the data centers CRAC system. In cases where liquid cooling, including chilled water was available on the data center floor, our goal was to reject high quality heat all the way back to the data center cooling tower, on a year round basis in most localities in the world. To achieve these goals, we employed LHPLs, some of whose other outstanding properties in addition to the fact that they are passive devices, turns out to be that the eliminate most of the electric motors, fans, blowers, compressors and other rotating devices found in servers and throughout the data center that end-up making noise, contribute to frequent server failures and cost money to maintain and operate.
These goals don't get met without skepticism from prior art and other technologies, so we will now address the advantages of our approach in detail, while at the same time laying out the critical items that need to be overcome to reach our goals.
There are a large number of technologies that have been recently investigated whose main purpose has been to improve the heat transfer capabilities of devices that may be used to cool semiconductor devices that reject large quantities of heat. LHPLs continue to remain as good as or better than these other devices. We will examine just a couple of the higher end sensibly cooled heat exchange technologies: microchannels and jet impingement. Microchannels require a liquid under pressure, usually pump driven and drive the liquid across a channel which extracts heat from the processors heat spreader. The contact areas that may be achieved with these devices is less than the wick areas provided by LHPLs which means to provide equivalent cooling, they need to make up for the fact that LHPLs absorb a factor of 100 as much heat per gram of coolant than they typically do. As a consequence, they end-up leaving the region of the device at higher velocities and at much colder temperatures. It is also very difficult for them to provide uniform cooling across the entire heat spreader for the simple reason that they do not uniformly expose the heat spreader to a uniform flow. Jet impingement, on the other hand does expose the surface to a more uniform flow, but because of a characteristics of the way in which jets interact with surfaces along with the fact that the heated water has to be quickly removed from the region of contact, a fair amount of mixing goes on, again reducing the temperature of the resulting effluent. The heat transfer coefficient of the LHPs employed in our experiments was 0.15.degree. C./Wcm.sup.2. This state of the art performance makes it possible to cool semiconductor dies whose are is 1 inch squared and reject as much as a kilowatt. It is possible that jet impingement may be able to cool devices that reject more power, simply because of the energy that they may eject into the flow employing pumps. But, for now at least, what we have just demonstrated is that for all of the semiconductor devices that are available or likely to become available, this technology not only may reject as much energy as the competitors, but do it without requiring additional energy and at the same time producing effluents whose heat quality is excellent.
The critical role that LHPLs perform in the removal of heat from hot semiconductors, is they make it possible to remove large quantities of it, using small devices that may be packed into small locations and while at the same time providing rejection distances that make it possible to locate large efficient condensers that may employ counter-flow designs at locations in the electronic enclosure where that heat may be exchanged with either air or water. That being said, the next most important feature of the technology that this disclosure brings to the table is methods that make it possible to maintain the quality of that heat as long as possible, whether it be exchanger with air or a chilled liquid. This is a crucial part of the design approach to the heat transfer problem that we have taken.
The method we will use to greatly improve the overall coefficient of performance (COP) of the data centers cooling is by eliminating the majority of the motors typically employed to cool a data center. For this feat to be realized, it becomes important to maintain the quality of the heat being rejected by the rack cabinets that the data center may use to contain its server units.
One of the big problems in energy conservation is underestimating the important role that the quality of the heat being rejected to the final cooling device in major thermodynamic systems plays in the overall cost of buying and operating such systems. Reducing the quality of the heat too much in the case of coal fired powerplants results in the sulfuric acid condensing out in them so fast that they have to be frequently replaced. In the case of a clean large multi-megawatt fuel cell power plant, extracting too much energy from the exhaust flow ends-up driving the cost of the energy and the fan required to cool the plant up to the point where the savings get lost. The implication for data center cooling is, keep the quality of the heat up, unless you want to spend a lot of money rejecting it at the cooling tower. In existing systems, the cost of rejecting it at the cooling tower consumes 25% of the cost of running the data center, i.e., running a water chiller.
The naive approach to the use of passive heat devices suggests that like the extra cooling loops that currently consume close to 35% of the energy required to run a data center, simply stringing a series of these devices in a row, ought to be able to solve the cooling problem, without even using a cooling tower. As it turns out, a sequence of such devices will operate less efficiently than a single large Loop Heat Pipe, whose condenser line moves the heat the same distance, simply because a sequence of these devices will end-up losing energy at each point of contact that connects the devices. And, since the effective driving range of the LHPs used to cool the semiconductor devices we are working with is several meters, the bottom line is that unless the cooling tower you are planning to use is in the immediate vicinity of the server you are cooling, stringing passive device together does not buy very much, but does just like the sequence of cooling loops currently employed, does dramatically degrade the heat you are attempting to reject. So, the ground rule for employing LHPLs employed to cool semiconductors turns out to be, exchange the energy with another secondary coolant, preferable one in the liquid state, as quickly as possible, if your goal is to use that coolant to drive a cooling tower directly, or to employ that heat in a cogeneration scheme or if it is simply to return air to a CRAC units heat exchanger at the highest possible temperature, thereby improving the efficiency of even an air cooled data center.
Having rejected the heat from the primary heat load in our electronic enclosure with the highest possible quality to either an air or in the case of a liquid, most likely chilled water, our goal now becomes to move it to the outside world with the smallest loss in energy. However, while doing that, we also need to consider how our LHPL primary heat removal solution interacts with the rest of the devices we use to gather up heat from the enclosure.
In data centers in which the average rack cabinet only consumed 5 KW, the fans on the rear of rack cabinets were a convenient way to help cool the contents. However, their main function at today's power levels of 20+ KW is mostly to hide the unsightly cables that drape the servers contained in the cabinet. A significant portion of the air being drawn through a typical rack cabinet ends-up being drawn around the stack of server chassis within and often the asymmetric flows within the cabinet may result in eddies that circle back to the front of the cabinet near the top, heat up the top servers by as much as 15 degrees F. To get around that problem fans may be added to the top of the rack cabinet and baffles inserted between the servers and the side panels. A better way to employ such fans, is simply to insert a duct in the cabinet that may be used to gather up all the air from the rack mounted chassis and exhaust it out the rear of the cabinet by connecting it the fans on the rear door or out the top using fans mounted on the top panel or possibly to the CRAC units return air flow ducting. To make sure that this duct does what it is intended to do, a mechanism has been provided in the disclosure to seal the chassis to the duct and at the same time make sure that in the event that a chassis is not installed the duct does not leak. Furthermore, to help solve the problem of potential leaks in situations where direct chilled water is being employed within the rack cabinets, the duct may be used to contain the chilled water manifolds that serve the rack mount chassis. Finally, to make it possible for the air being removed from the rack to be reused without having to make the long trip back to the CRAC units blower, simply inserting chilled water air heat exchangers in the exit path from the rack mount chassis to the duct, makes it possible to eject the air from the cabinet at the ambient air temperature of the room. This strategy has a number of benefits of other approaches to the cooling of high power rack mount chassis that employ water cooled air heat exchangers within the rack cabinet. Besides taking up much less space in the rack cabinet, and making it possible to employ distributed heat exchangers whose total area is much larger than the ones employed by other solutions, it also reduces the total high speed fetch that the air has to make. And in the process, the amount of energy that gets injected into air flow ends-up being minimized.
Reducing the energy employed moving air is one of our overall goals. When we have to do it, our goal is to move the air the smallest distance at the smallest possible speed that gets the cooling job done. The reason for this is quite simple, energy losses due to drag scale as the velocity of the air cubed multiplied by the distance it travels. Keeping the velocity and distance down, makes an enormous difference in the energy consumed by the fans driving the serves, rack cabinets and the data center itself. The technology we employ cuts down on these losses three different ways. First, when exchanging energy between air and either the primary coolant being chilled in a condenser, or a chilled liquid that is cooling it, employing finned condensers that have large areas, which our technology enables by doing things like moving the heat being rejected out of tight spaces, ends-up reducing the velocity of the air required. Next, by moving air the smallest distances possible, which we make possible by picking up secondary heat sources in 1 U sealed rack mount chassis (which reduces the distance and velocity needed) or by cutting down on the distance that air needs to flow at high velocities when it is being cooled by a negative pressure duct, or by completely eliminating the need for the air to travel back to the data center's air heat exchanger, we make large reductions in the amount of energy that needs to employed by air fans and blowers. This strategy plays an important role in our energy conservation effort, and is embodied in both our sealed chassis and sealed duct designs.
The final energy reduction principle that needs to be taken into account that out embodiment improves is water condensation. In some data centers, as much as 40% of the energy being employed by water chillers gets used to remove (by condensation) water vapor from the cooling flow which then, apparently needs to get added back into the flow to keep IT people wandering through the data center happy. It turns out that there no longer is and ESD requirement on the minimum air content of the air being employed in data centers, which basically means that keeping the relative humidity below the point where condensation occurs in the equipment may now be achieved by simply making sure the dew point of the air in the systems being cooled is less than the temperature of the liquid coolant being employed to cool systems, saving roughly 10% of the energy employed to cool some data centers, especially those in humid localities.
Our sealed chassis embodiments make this possible by keeping the dew point of the air inside of the rack mount chassis below the temperature of the coldest liquid coolant employed. This is simply accomplished in an embodiment in which we pass slightly pressurized air through a cold trap that removes excess water from it before slowly bleeding it into the “sealed” chassis, that are allowed to slowly leak air back to the ambient, at a rate that guarantees that the average air content of the chassis remains dry enough to avoid condensing if and when it comes into contact with chilled surfaces.
When it comes to cooling air cooled enclosures in general, LHPLs make it possible to make great strides in efficient uniformly distributed air cooling, by the simple act of placing the LHPL condenser at the point in the chassis where the air flow is normally exhausted out of the chassis. In the two enclosures we have studied, 1 U rack mount chassis and desktop cooled chassis, the fans that are employed on the exterior surfaces of these chassis have provided high enough flow rates to in the case of a 1 U chassis only require a single blower (already employed to pull air out of the chassis) to cool a pair of 120 Watt processors (it normally takes four to eight 1 U fans to accomplish the same task) and a single 120 mm fan running at just 1800 RPM to cool a 500 Watt CPU sitting in either the PCIe bus of the system. In all of the chassis we have examined, including the 4 U chassis employed to cool four to eight Opteron multi-core processors, the existing fans on the rear wall of the chassis that we have examined have more than enough cooling fans to make it possible to cool all of the processors, without the need for CPU fans. Which is to say, all of the chassis tested, when their CPUs were cooled using LHPLs, could get by without the need for CPU cooling fans. Not only that, the CPUs that were being cooling in situations like the 4P/8P chassis, normally require very high air flow rates even with cooling fans that fit into 2 U tall spaces simply because the CPUs in the front row end-up heating the air used to cool the rear row of processors. This problem goes away with LHPLs, making it possible to actually reduce the air flow rates on the rear wall while at the same time eliminating the four to eight fans typically used to cool processors. And, while we may not claim that air cooling does as good a job as water cooling, we have gone about as far as you may go with air cooling to maintain the quality of the heat being rejected. In addition to providing sealed ducts, more uniform distribution of cooling air across the chassis and the reduction of the ambient temperature within the chassis, we have also introduced LHPL condenser designs which employ counter-flow cooling, which results in increased exit flow air temperatures which in turn end-up improving the efficiency of an air cooled data center's water chiller.
When it comes to liquid cooling, the embodiments provided make it possible to employ LHPL cooling with condensers that are either directly or indirectly cooled with chilled liquids including water, safely. A new method for interfacing all closed loop passive heat transfer devices to chilled liquids has been introduced which employs a cold plate along with what we call a cold spreader (that is thermally attached to the LHPL working fluid's condenser lines) that comes into contact with the cold plate when a rack mount chassis gets installed inside of a rack cabinet. This interface, while not quite as efficient as the directly cooled interface we are about to describe, in certain situations, like blade and COTS Single Board Computer (SBC) situations, makes it possible to cool these devices as well, without using the quick disconnects that direct chilled liquids require. To improve the quality of the heat being rejected by these split condensers, a counter-flow version is also embodied and examples are provided of how to employ the cold plates that are a component of a split condenser to also cool air that is either circulating within a sealed chassis or being passed through a chassis that is being evacuated either by internal fans or a negative pressure air duct.
The most efficient cooling that we believe may be obtained using LHPLs comes when they are cooled directly with chilled liquids housed in a sealed enclosure in which the remainder of the components within the enclosure are being cooled by either liquid cooled cold plates, air that is circulating about a chassis that includes cold plates that cool it and the PCBs in the chassis and that is driven by low energy fans or blowers or air that is being circulated through the sealed chassis that passes through a chilled liquid air heat exchanger that may be a part of a component that includes the LHPL's condenser. The condenser design that we created that did the best job of producing high temperature effluent employed counter-flow heat exchange and used a chilled water jacket that was made of a material that does not readily conduct heat in addition to employing a helical wire that was thermally attached to the serpentine shaped condenser pipe, forcing the liquid to take a longer path and simultaneously increasing turbulent flow.
The final claim in the disclosure is for a data center cooled with the afore mentioned devices in which the servers in the data center room is directly attached to the cooling tower, eliminating the need for air ducting, special insulation in the walls of the data center (to keep humid air out), the need for an air blower and finally the water chiller employed by the air blower, in localities in the United States, when on the hottest most humid days of the year, an evaporative cooling tower will return water to the data center room that is at least 30 C, which is to say for most locations as hot and humid as locations like Atlanta Ga., 365 days of the year.
DEFINITIONS USED
Loop Heat Pipe (LHP)
A Passive two phase heat transfer device that consists of an evaporator that contains a compensation chamber and a wick with escape channels on one side that receives heat from the device being cooled which causes a working fluid to change phase on the inside of the escape channel walls producing vapor phase coolant that leaves the evaporator passing through a vapor line to a condenser where the vapor passes through a condensation channel that may be cooled by a gas or liquid causing it to change phase back to a liquid and dumping the rejected heat into the cooling gas or liquid before it returns to the evaporator through a liquid return line that may pass through a compensation chamber located within the evaporator between the evaporator's inlet and the side of the wick facing the evaporator inlet. The device being cooled is typically thermally mounted to the evaporator shell which encloses the wick in the region of the wick that contains the escape channels.
Loop Heat Pipe Like (LHPL)
A device that may contain any of the ingredients of a Loop Heat Pipe or Capillary Pumped Loop as well as an additional derivative feature such as a pump at any point along its condenser path designed to either increase the working pressure of the working fluid leaving the pump or extending the length of the lines. An LHPL may contain one or more compensation chambers that may lay within the evaporator, attached to the evaporator, inline in the liquid return line or attached to the liquid return line at any point along it.
Capillary Pumped Loop (CPL)
A device that contains all of the ingredients of a Loop Heat Pipe but in which the compensation chamber is no longer situated within the evaporator between the inlet and the wick, but at some point along the liquid return line.
LHP Compensation Chamber
A volume situated between the liquid return line inlet of an LHPL evaporator module and the evaporator wick which distributes returning liquid to the wick and whose volume is large enough to impact the performance characteristics of the LHPL. See description of
Internal Compensation Chamber (ICC)
Identical to LHP Compensation Chamber, a volume placed along the liquid return path of an LHPL which is contained by the region within the evaporator module where returning liquid is received.
An Undeclared Compensation Chamber (UCC)
An ICC whose volume is too small to impact the performance of the LHPL. This may arise in situations where there is not enough room to place a large enough volume within the evaporator to compensate for the length of the vapor and liquid lines which prevents it from impacting the performance of the LHPL. See description of
Liquid Return Path
The path that liquid returning from the condenser to the wick within the evaporator module takes which includes the liquid return line that connects the condensation channel exit with the evaporator inlet and the region between the evaporator module inlet and the exposed surface of the wick inside of the evaporator module body.
External Compensation Chamber(ECC)
A volume connected to the liquid return path which extends from the condenser outlet to the evaporator wick that connects a volume to either the liquid return line or to the evaporator shell using a small length of tubing. This is the form of CC that normally is associated with Capillary Pumped Loops (CPLs). See description of
An INLINE Compensation Chamber (ILCC)
A volume inserted into the liquid return line that is primarily used to compensate for the differences in the length of the lines used to interface evaporators and their condensers cooling a system that has several primary heat rejecting components. See description of
Standard Heat Pipe (HP)
A sealed tube that is lined with a wick that contains a working fluid that receives heat at one end called the evaporator, causing the working fluid in the wick to change phase and enter the empty center of the tube which allows the vapor produced by the heat to flow to the other end called the condenser where a coolant receives the heat and causes the working fluid to change back to a liquid which then gets carried by the wick lining the walls back to the evaporator end of the sealed tube.
Two Phase Passive Heat Transfer Cooling Devices
LHPs, LHPLs, CPLs, HP, Thermo siphon and any similar device
Two phase Heat Transfer Cooling Devices
Any device that is a member of the Two Phase Passive Heat Transfer class of devices plus any device that transfers heat between an evaporator and a condenser that employs a pump to affect the motion of the working fluid.
Primary Heat Load and Components
The devices within an enclosure whose cooling is facilitated by LHPL devices.
Secondary Heat Load and Components
The devices within an enclosure that are not being cooled using an LHPL device.
Thermally Attached
Techniques which connect a pair of thermally conducting devices together which improve heat transfer between the two such as soldering, the employment of heat conducting epoxies and thermal transfer products that use solids and pastes along with the use of clamping pressure and surfaces that are polished flat.
Heat Sink
A finned device made of a thermally conducting material such as copper which typically has solid “base plate” on one side to which a set of fins have been thermally attached, the base plate acting as a heat spreader that conducts incoming heat from the device being cooled to the fins which typically reject the heat passed to them to the air flowing through the fins.
Laminar Flow Disruptor
An obstacle placed along the wall or center line of a conduit carrying a gas or liquid which causes the boundary layer flow along a wall of the conduit to mix with the flow at the center of the conduit.
PCB
Printed Circuit Boards (PCB) are currently multi-layer devices consisting of insulating plastic materials that have circuits etched on both of their sides prior to being glued together using copper Vias that connect the etch on their surfaces together providing the “nets” that connect components soldered to the outer layers of the boards to each other. PCBs often employ extensive areas of copper plating that are used to provide the power and ground planes that are required to enable high speed circuits to function. A 12 layer PCB will often employ 6 of these copper ground planes, providing a mechanism for gathering secondary heat from components soldered to the board. These copper planes often play a crucial role in the cooling of the components that receive power and grand from them, by virtue of the fact that high power devices like CPUs may consume over 100 Amps at voltages of 1.2 volts (which is how said CPUs may end-up rejecting over 100 Watts.) while often employ 1,000 or more pins that make connections between their BGAs (ball grid arrays) and the PCB which are made using said copper vias, half of which are used to provide the silicone die with power and ground. The Vias that connect these pins to the internal power and ground copper end-up playing an important role in the cooling many high power circuits including MOSFETs, which makes it possible to collect rejected secondary heat by simply placing a cold beneath a PCB.
Recirculation “an act or instance of circulating, moving in a circle”, which in the case of air flowing over a reentry vehicle occurs in the rear of the vehicle which is where wake chemistry occurs in the heated air that collects there, in the case of a device rejecting heat to air that is sitting either within an enclosure or within a room that contains an enclosure, ends-up raising the temperature of the cooling air that is used to cool the device being cooled. In the case of electric devices sitting inside of an enclosure or inside of an enclosure within an room that has cooling air entering it that in turn enters the enclosure, recirculation occurs when some percentage of the air that carries off the heat that has been rejected by the device to cooling air succeeds in finding its way back to the point where the cooling air either enters the enclosure or mixes with the ambient air within the enclosure that mixes with the cooling air entering the enclosure or is ejected from the enclosure in a manner that allows it to mix with the cooling air entering the room the enclosure sits in. The way that recirculation is defeated is by gathering the preheated air leaving the device being cooled and immediately exhausting it out of the enclosure containing the device being and if the enclosure is in a room surrounded by cooling air using a device outside of the enclosure that prevents the preheated air from mixing with the cooling air entering the enclosure. Recirculation at both the enclosure and the room level both end-up increasing the energy required devices within enclosures for a sequence of reasons, the first being the ambient air within the enclosure gets heated by the mixing which requires the velocity of air moving over the devices being cooled to rise, that in turn requires the energy expended running fans to rise, raising the velocity of the air leaving a typical enclosure in which the fans are located at the front or rear of the enclosure that ends-up increasing the volume of air that the heat being rejected resides in that causes the temperature of the air leaving the enclosure and arriving at the HVAC system providing cooling air that raises the flow rate that the HVAC blowers need to provide as well as the energy consumed by the HVAC water chillers whose real goal is to ensure that the water arriving at a cooling tower is high enough to guarantee that the heat may be exchanged with the outside air on hot days, requiring the water chiller to expend more energy raising the temperature of this water when the air being delivered to it is at a lower temperature due to its increased velocity within the enclosure and the room that contains the enclosure.
SUMMARYMethods were disclosed which make it possible to employ Loop Heat Pipe, Capillary Pumped Loops, Advanced Loop Heat Pipes (that are not passive and employ pumps to augment their ability to cool devices) and other similar devices that employ an evaporator module and a condenser module connected by a pair of lines, one of which carries the vapor from an evaporator to a condenser where it is cooled either by a gas or a liquid before being transported back to the evaporator by a second line, the evaporator typically using escape channels to reduce the resistance to flow to generate a delta pressure or P across the evaporator, being used to typically cool electronic devices within an enclosure, in which the design and location of the condenser module may make a large reduction in the energy expended to cool devices within the enclosure and any external device such as an HVAC system or cooling tower to reject heat to the outside world.
The resulting methods made dramatic reductions in the amount of energy employed to cool electronic components housed in electronic enclosures while at the same time making dramatic improvements in other operating characteristics, including reliability, the amount of heat being reject to the outside world, the amount of noise produced, the size of the power supplies needed to power units, the cost to build and operate data centers and last but not least, the ability to cool very hot electronic devices housed in electronic enclosures that are densely packed and employ silicon dies that reject 100 or more Watts per square cm.
The embodiments included designs for LHPL condensers, including air and water cooled condensers that employed counter-flow techniques, LHPL CPU heat spreaders, sealed chassis and sealed air ducts, methods for controlling the vapor content of air within sealed chassis, methods for connecting chilled liquid sources to condensers including split condensers that eliminate the need for quick disconnects and quick disconnects that are shielded from the chassis being cooled by a duct connected to a blower that sucks cooling air through chassis eliminating the need for small fans running at high speeds that are much less efficient. In addition secondary cooling systems are described in which a single external coolant is used to cool the LHPL devices in which either the external coolant is used to cool the secondary components directly or indirectly employing a different secondary coolant that accumulates a secondary heat load and transfers it to the external heat load using devices that include standard heat pipes and metallic plates that move heat small distances and in the case of standard heat pipes several inches that reject heat to the secondary coolant using devices like heat exchangers or wherein the heat is rejected directly to the secondary coolant from the secondary component using convection.
The methods included embodiments that make it possible to cool the majority of the data centers operating in the United States without having to employ either air blowers or air chillers 365 days of the year, reducing the acquisition costs significantly while at the same time reducing the energy consumed by 40% or more.
The drawings are illustrative embodiments. The drawings are not necessarily to scale and certain features may be removed, exaggerated, moved, or partially sectioned for clearer illustration. The embodiments illustrated herein are not intended to limit or restrict the claims.
FIG.45 is a perspective view of an embodiment in cooling device employing an LHPL to air cool a primary heat rejecting device in which a portion of the secondary cooling device is liquid cooled employing a liquid secondary coolant that is directed by a pump to run in a loop that connects several liquid cooled heat exchangers that are thermally connected to secondary heat rejecting components and which reject their heat to the gaseous primary coolant employing either a finned heat exchanger or a thermal transfer device to conduct the heat to a nearby LHPL evaporator module that in turn is air cooled rejecting its heat to the gaseous primary coolant as well.
Referring in
Contained within the 1U rack mount chassis 100 is a pair of CPUs that are mounted on a PC server motherboard 107 that are being cooled by a pair of LHPs in this embodiment one of which is an embodiment of a high pressure LHP that employs a cylindrical shaped evaporator module and is employed with high pressure working fluids such as Ammonia, whose evaporator 101 is thermally attached to the CPU being cooled and whose working fluid is being cooled and condensed from a gas back into a liquid by the a condenser 104 using a cool gaseous flow that typically employs but is not restricted to air being pulled through the condenser fins by an “exit blower” 105 that is one of the devices normally employed to pull air through 1U chassis. The device being used to pull air from a chassis may also be a 1U fan or any other form of air moving device, even devices placed external to the 1U chassis. This embodiment illustrates how the existing fan that comes with a 1U chassis may also be used to cool a pair of CPUs using a condenser whose details are discussed below. This figure also calls out the condenser line 103 that returns the condensate back to the evaporator and the condenser line 102 that transports vapor from the evaporator to the condenser. The sharp angle shown in the drawing where the condenser line 102 attaches to the evaporator body 101, in the devices we used, was rounded. The second CPU is being cooled by a Copper Water LHP whose evaporator module 101a that feeds a vapor line 102a that transports the water vapor to its condenser 104a returning liquid water in line 103a. Future figures will detail the differences between both of the condenser designs, 104 and 104a detailed here. The particular air cooled condenser designs are especially good. In an ideal LHPL half of the thermal resistance is contributed by the evaporator and the other half by the condenser. However, this efficiency level may only be achieved by liquid cooled condensers. In air cooled condenser the heat has to travel a significant distance through a good conducting metal (typically copper) before it arrives at the midpoint of the fin that rejects the heat to air or a gas flowing between the fins. As a consequence, the thermal resistance of an air cooled will typically be a factor of 3 or more greater than a liquid cooled one, so it becomes important to minimize the thermal resistance of the condenser fins. This is accomplished by reducing the thermal resistance between the condenser tubing and the fins and to reduce the distance the heat has to travel to get to the fin midpoint. The designs shown both have especially low thermal resistance and are detailed below along with another trick used to improve their performance. This figure points out the differences between evaporator modules intended to handle working fluids that at the temperatures that most of the devices we plan to cool are either at high pressure like Ammonia which require cylindrical evaporators to contain the working fluid while working fluids that run in the vicinity of one atmosphere like water (which typically will be at pressures of half an atmosphere) may be handled by flat oval evaporators, providing of course that both types of evaporators are excellently sealed and do not permit even trace quantities of other chemicals to enter the device over the lifetime of the device. Also called out in this figure are the power supply 106, the rear wall of the 1 U chassis, 108, and the exhaust port employed by the blower, 109.
In this embodiment a single rear mounted exhaust device (blower) 105 is used to cool both processors eliminating the typical need for four and eight “1U” CPU cooling fans that are normally needed to get the job done when heat sinks whose fin areas are a factor of two smaller that are mounted on top of the CPUs are used to cool the same CPUs. A pair of baffles, 105a and 105b are shown, the first coupling the blower 105 to a pair of condenser fin blocks each of which cool one of the two LHPs. The second baffle, 105b, couples the heated air leaving the blower to the rear exit port of the 1U chassis, allowing the heated air to leave the chassis without recirculating within the chassis heating up the ambient air within the chassis, which if allowed, will end-up reducing the delta T across the cooling fins which would increase the cooling air flow rate needed to get the same amount of cooling. In a typical installation, the chassis and blower would be arranged so that the blower was mounted up against the rear wall of the chassis eliminating the need for a second baffle. The reduction in components, points of failure, noise and the need for electrical power is made possible by the passive nature of the loop heat pipes employed along with the use of a condenser whose finned heat exchanger has a much larger area than the fins typically used to cool these CPUs when they are mounted on a base plate that sits on top of the CPUs. It is this increased fin area that dramatically reduces the need for high speed air to cool the CPUs. In the case of the Ammonia LHP, the rear region of the 1 U chassis is being accessed using LHP condenser tubing who's OD may be as small as 2.5 mm while the lines in the Copper Water device turn out to range from 4 to 5 mm. One of the primary benefits of this approach is the elimination of recirculation currents within the enclosure that mix the heated exhaust with the incoming cold air raising the ambient temperature of the air that cools the CPUs and other devices which in turn requires that the speed of the exhaust fans be increased to get the same amount of cooling for all of the devices held by the enclosure. In this implementation, the rotary cooling device used to cool the primary and secondary heat sources in many cases is one or more devices mounted at the rear or within the enclosure, and in this case these devices have been consolidated into a single device, providing an additional benefit to the approach.
The condenser 104 in FIG.1 is the first of many embodiments of a two/three piece condenser design in which the bottom half of the condenser remains with the enclosure. This design employs modified CPU “heat sinks” which normally employ a thick copper base plate which acts like a heat spreader and which transfers the heat between the CPU it is attached to and the fins attached to the base plate. In this design a pair of these CPU heat sinks is employed. The top exchanger is modified so that it may be attached to the enclosure. Employing a pair of inexpensive commodity CPU heat sinks doubles the total fin area of a single CPU heat sink. When the CPU needs to be replaced, the top heat sink used to create the condenser is unbolted and removed. The condenser line typically takes a serpentine path. In the two piece version an identical path is hogged out of the base plates of the top and bottom heat sinks providing an excellent thermal interface. Alternatively, a pair of heat spreaders may be soldered to the condenser line providing a flat surface that conducts heat to the base plates that sandwich these spreaders. The spreaders add weight and reduce the thermal resistance of the design. This type of condenser may be seen in
A processor 902 or other device (not necessarily restricted to an electric heat source device) is thermally attached to an evaporator 904. A hot gas phase working fluid 906 exits the evaporator and travels to a condenser 908 via a first line 910. After the gas phase working fluid 906 moves through the condenser 908 it changes from a gas to a liquid and then returns to the evaporator 904 as a liquid using a second line 912. The condenser 908 produces the phase change by cooling the gas phase working fluid which results in the deposition of the heat carried by the working fluid into whatever coolant is used to cool the condenser. The split condenser shown is one of three basic designs that is encompassed by our first patent provisional application and was one of the types broken out by the USPTO as a separate invention. It is water cooled, and while many of the figures in this application continue to describe these devices, we are not detailing the liquid cooled portion of this split condenser, as we are primarily interested in exploring the other unique features of our cooling approach. The heat is transferred between the working fluid and another liquid coming into contact with a cold plate which is in turn in contact with the portion of the condenser that has the working fluid passing through it. We are presenting this figure here for the purpose of elucidating the combinations of features found in LHPL devices which we have defined as having features commonly found in Loop Heat Pipes, Capillary Pumped Loops as well as a device called an Advanced Loop Heat Pipe that employs a pump along with a condenser that may be broken into two pieces. This figure does not restrict LHPLs to devices that only employ water cooled split condensers: the condenser may also employ the other two cooling mechanisms claimed in the patent applications: forced convection whose coolant is either a liquid or gas. A plurality of compensation chambers 914 may be included in what the claims describe as the “liquid return path,” which the independent claims define as the path that the liquid takes between the condenser outlet and the evaporator module wick. There are four different types of compensation chambers that an LHPL may have, one being an unclaimed compensation chamber, which turns out to be an internal compensation chamber whose size is too small to impact LHPL performance. This comes about because of the fact that some internal volume within the evaporator module is required to distribute the returning working fluid that enters the evaporator module through an inlet port that is smaller than the cross sectional area of the evaporator. As a consequence, the returning liquid needs to pass through a section of the evaporator where the liquid may expand out to cover the entire wick that it needs to come into contact with to flood the wick and this region by definition also forms a volume. The use of compensation chambers to differentiate the different types of LHPLs is in fact a rather strange situation, as the primary purpose of the compensation chamber is to modify the performance curve of both LHPs and CPLs as the devices approach full power and they carry out this mission using a volume, which in one case sits external to the liquid return path (CPL) or internal to the evaporator module (LHP). To this group we have added a third form, which is inline within the liquid return line. The second provisional application includes the concept of changing the volume of a liquid return line by either placing a CC in the line or possible adding volume by placing a serpentine in the line. In the embodiment illustrated, the evaporator 904 includes a mechanical pump 920, whose purpose is to reduce the pressure in the gas side of the evaporator module 904 and we recently added a pump to the liquid return path not seen here whose purpose is to drive returning liquids long distances, which is of as much use as a pump on the vapor side whose purpose is to decrease the evaporation temperature within the evaporator module.
Typically, Loop Heat Pipes are differentiated from Capillary Pumped Loops by the location of the compensation chambers. More than one location and purpose for these devices was envisaged by the author when he wrote the first and second provisional applications. The compensation chamber may provide liquid to the evaporator when the device starts up, provide a location for water to accumulate outside of the condenser as the power being rejected increases allowing the boundary between the gas and liquid phase working fluid inside of the condensation channel to move towards the channel's exit wherein the available area within the condensation channel that converts gas to liquid increases improving the condensation process and helping in situations where there is not enough volume within the evaporator to provide an adequately large compensation chamber (due to the length of the vapor and liquid return lines) in evaporators whose length is not free to grow and finally making it possible to use inline compensation chambers that allow a single evaporator module design to be used with LHPLs whose evaporators and condensers are separated by a range of distances.
In
In the case of a blade or SBC card mounted in a COTS or Blade chassis, an air path is normally provided for cooling which directs the air across the board in either a vertical or horizontal direction. Normally, COTS style SBCs are housed in special chassis. The solutions described herein could be used in situations where multiple rows of blades are cooled by air that is directed vertically through a plurality of them.
The ideal location for an LHP air-cooled condenser in this embodiment is at the location where the air normally exits the blade.
The embodiment seen in
The plurality of condenser tubes seen in
The lower 3D rendering exposes the evaporator shell 940 whose end caps 941 and 942 are also called out and which are called out in greater detail in
The chassis bottom half 1000, has cooling air inlet holes 1001 situated in its front. Items of interest include four sets of DIMM modules one of which 1002 is pointed out. Items 1003 and 1013 are Inline Liquideous Working Fluid Storage Chambers 1013 that have been inserted in the liquid return lines and would normally be used to tune the lengths, adding capacity to the final CC being done by either building more CC capacity into the evaporator or adding a volume attached to the liquid return line at any point using a tube that may also be inserted to the rear of the evaporator module itself. There are four flat oblong low pressure evaporators 1004 the first of which is called out. 1005 points out one of four LHP vapor lines while 1006 points out one of four optional DIMM module cooling fans, which may or may not be needed, depending on the type and quantity of memory installed in the two motherboards used, neither of which may easily be separated in this drawing, but sit beneath the electronic components being cooled with air that include the DIMM modules, CPUs (which sit beneath the four evaporators) as well as secondary components like MOSFETs. 1007 points to one of the four liquid return lines while 1008 points to one of the four sets of lower condenser fin sets which sit beneath the serpentine vapor lines that are fed by the evaporators. 1012 points out the upper most set of fins which clamp to the lower ones using studs that may easily be seen in the two sets of lower cooling fins. 1009 points out the hot air exhaust ports in the chassis through which the hot air was pushed using three blowers 1010. This is followed by the removable power supply 1011, a pair of hard disks 1016 and a PCIe communication card 1014. The blowers that provide the greatest portion of the cooling of this design consume less than half the power of the typical fans used to cool a similar 1U chassis. The reduction in power is a direct result of the fact that the large fin area reduces the flow rate required to cool the processors the fact that the blowers run at just 6,000 RPM which dramatically improves reliability over fans and the fact that the air used to cool CPUs does not end-up heating DIMM modules, which causes a typical 1U chassis to have to further increase its cooling air flow rate. In
To create the LHPL condenser seen in
In addition to the flat oval evaporator
The condenser at the bottom of
The form of the liquid cooled condenser exchanger may become critical when it is desired to be especially energy efficient. The liquid jacket may be as simple as a pipe that encloses the condenser line however a pipe that conducts heat ends-up reducing the temperature of the coolant leaving the condenser and this may make a big difference on the ability for example to provide free cooling in hot climates, something that water cooled LHPLs make possible. When the goal switches from simply condensing the primary vapor to condensing it while at the same time producing the hottest external coolant, then other things become important as well. These include preventing heat leaks across the condenser while at the same time making it possible for the smallest secondary coolant flows to be fully taken advantage of. All three criterion may be met in LHPL condensers making LHPLs probably the best devices available for not only efficiently transferring energy off hot running semiconductor devices but also producing secondary coolant flows whose temperatures are as hot as possibly may be obtained.
The ability of the LHP evaporator to accomplish the three critical tasks starts with their ability to drive primary coolants through long lengths of condenser tubing making it possible to employ large heat contact areas between the primary and secondary working fluids. Making the condenser jacket out of a thermal insulator solves the second problem. The third problem, making it possible to get good thermal conduction between the primary and secondary coolants even when the secondary coolant velocity has been reduced, is made possible in our case by the use of a condensation channel that has a laminar flow disruptor placed on either side of the condensation channel. When the disruptor is made of a thermally conducting material it may also end-up increasing the effective area of the channel on either side. The use of a material which adds area needs to be weighed against the fact that this material may also reduce the thermal conductivity of the wall by increasing its effective thickness.
In general, split condensers are not as efficient as the direct liquid cooled condenser design that we terminate this disclosure with. When working with them, one becomes aware of the fact that raising the temperature of the output flow is a non-trivial exercise. What makes it difficult to raise the temperature of the output flow from a water-cooled condenser, be it direct or indirectly (i.e. split condenser) cooled is the fact that for any design that employs heavy copper plates to exchange heat there are thermal conduction shorts across the plates used for the cold spreader and cold plate which enable heat to move from the hot to the cold side of the split condenser. Such thermal shorts are no different than the thermal short that occurs within an air cooled enclosure in which the heated air mixes with the ambient air, reducing the temperature of the effluent leaving the chassis and thereby reducing the COP of any cooling system employed to provide either chilled water or air out of the hot side of the condenser and exchanges it with the cold side, reducing the delta T across the condenser and reducing the temperature of the effluent. The only way to eliminate this problem is to isolate the hot and cold sides of the condenser from each other. The design in
The design that is shown in detail in
The upper portion of the figure is an end on view of the blades inserted into the chassis. The end view provides a detailed description of the cooling arrangement. A pair of cold plates, 400, each cooled by a liquid stream 403 is employed to cool the components on the blade employing a split condenser. The blade PCBs 402 slide into the enclosure using “card edge guides” 414 that are designed to both guide and hold the PCB as it slides into place. Between each of the card edge guides used by adjacent blade PCBs we have inserted a “U channel” 401, at both the top and bottom of the blade. These U channels provide the metallic component that defines the cold spreader and are in thermal contact with the cold plates, LHP condenser lines 412 and DIMM module heat pipes, 413. The blades are inserted into the electronic enclosures with their U channel cold spreaders 401 making light contact with the cold plates 400 that they will become thermally attached to after the board is completely inserted into the enclosure. For the channels 401 to make good thermal contact with the cold plates, 400, it is necessary to apply pressure to them.
In this kind of situation there are few possibilities for creating this pressure. The method we have chosen to provide the clamping pressure, which is certainly not unique but does demonstrate a possible approach, is to employ a scissors jack with arms 408. The jack is engaged using a knob 406 that comes out the front of the blade and uses a shaft 407 and threaded nuts 417 to squeeze the two arms 408 of the scissor together, creating a vertical compression stress on the springs 409 that in turn is applied to the channels 401. The squeezing action is effected using a pair of threaded nuts, 417, one of which is called out and is opposed by a second in each jack. As the threaded section of the shaft 416 turns the two nuts 417 get pulled together putting opposing forces on the springs, 409, one of which is being used to exert pressure on each of the cold spreaders 401. The sleeve 419 rides up and down on a vertical shaft not called out which attaches at its end to a cold spreader with a female attachment 418 that holds it snuggly to the channel.
The cold plates 400 are obvious in the side view and we mark the cold (404) and hot (405) liquid coolant channels that feed them at the top as well as the LHP 410 and its condenser lines 412 that get used to cool the processors which sit beneath them. The memory region of the card 413 is also called out as well as the device used to cool the DIMM modules, which employs an ordinary heat pipe 413 whose condenser is also thermally attached to a cold spreader. The balance of the components on the board that dissipate energy may be easily cooled using a copper mask that fits over them and is attached to them using a thermal interface material that is in turn attached to either the LHP or the heat pipe used to cool the DIMM modules. Or alternatively, the ground planes of the PCB may be thermally attached to the cold spreaders using a flexible sheet of copper made of shim stock that is soldered to the PCB and gets inserted between the cold spreader and the cold plate.
For this blade cooling solution to work, the thermal resistance of the split condenser defined by the cold spreader and the cold plates has to be adequate. In situations where the cards are spaced on 1 inch centers and are 18 inches deep, this solution provides 36 square inches of interface area between each of the cold spreaders and their cold plates. Based on earlier experiments this ought to handle at least 500 Watts of power per card. Sixteen of these devices packed into a chassis that is 7 inches (i.e. 4 U tall) will reject 8 KW, and in a 40 U tall rack cabinet this results in a cooling system that will handle 80 KW. This cooling solution also turns out to be much more efficient than any blade cooling solution that employs air and direct water cooling, both of which require additional power. The power densities here are high enough to suggest that the resulting water might even be hot enough to be sent directly to a cooling tower bypassing a systems water chiller. But, since we haven't built one of these units, we feel it is premature to make this claim, especially considering the fact that the split condenser here does not employ counter-flow principles like the unit we are about to describe.
1) The use of energy efficient passive devices of any type, including LHPs, LHPLs and ordinary heat pipes to remove the primary heat load from the devices rejecting it and to deliver this heat load to one or more efficient direct liquid cooled heat exchangers.
2) A method for removing the secondary heat load and transferring this heat load directly to a chilled liquid, that may employ one or more methods including combinations thereof, including direct liquid cooled closed loop air circulation, passive connections between the secondary components and a chilled liquid cold source such as a cold plate, chilled liquid cold plates in contact with PCBs (see section in definitions which discusses how internal layers of PCBs are employed to gather secondary heat loads) enclosed by the enclosure and circulation of air within the chassis in a closed loop that does not involve the use of a chilled liquid air heat exchanger. The last method was specifically included in the list of secondary methods to allow distributed cold plates within the chassis to directly cool the PCBs that reject most of the secondary heat (i.e. heat other than the primary loads) including motherboards and power supplies while at the same time employing circulating air within the chassis to gather up the remainder of this heat and allow it to be inexpensively exchanged with the PCBs being cooled by cold plates, which just happen to contain a lot of copper distributed over large surface areas.
The specific embodiment in
Beneath the LHP counter flow heat exchangers may be seen the fins 209 of a liquid cooled air heat exchanger. The liquid that feeds the air heat exchanger in this embodiment just happens to come from a distribution block 226 that surrounds and contains the heat exchangers and includes a liquid cooled base plate that is thermally attached to the fins. The liquid that cools the air heat exchanger does not have to be shared with the heat exchangers and the precise order in which the liquids that cool the LHPs and the air get applied, is up to the engineer designing the system and the inclusion of other methods, which such as a method for bleeding air at a particular relative humidity into the rack mount chassis. Another pair of lines could have also been employed to supply the liquid assisted air heat exchanger or two of the four lines in
This method uses a similar technique described in the embodiment in
What is different about the method is its use of direct liquid cooling of the primary LHPs or LHPLs along with the fact that it may combine direct liquid cooled air heat exchangers, cold plates and air circulating that rejects heat picked up from secondary heat loads and passes it to the motherboard itself in situations where the motherboard is cooled by a cold plate, which ends-up converting the large area of the mother board into a heat exchanger without fins. In the embodiment displayed in
The intent here is to seal the chassis, using a positive pressure technique that slowly bleeds dry air into the chassis when the temperature of condensing surfaces within the chassis are below the relative humidity of the air in the room. In poorly managed data centers, up to 10% of all energy gets devoted to condensing and then humidifying the airflow. This is done to make the data center more convenient for humans and to reduce ESD. It turns out that in well designed and grounded circuits, such as those employed within rack mounted chassis, the ESD requirement has now been eliminated. However, care still has to be taken not to condense water vapor out within the chassis.
The two things that make this method so extraordinarily energy efficient are the very low thermal resistance of the LHP primary heat load cooling mechanism combined with an efficient as possible approach to recovering the secondary heat load. The latter has been greatly improved over other methods that combine direct chilled liquid cooling of the primary load (which is two orders of magnitude less efficient than the passive techniques employed here) with circulating air, by minimizing the energy required to circulate the air and also by employing optional PCB cold plates where possible to minimize the amount of energy employed to circulate air. Consider for a moment, the fact that the distance that the air inside of this enclosure needs to travel to release its heat is hundreds of times shorter than the distance air leaving an enclosure needs to travel before it arrives at a chilled water heat exchanger where it dumps the rejected heat it is carrying to chilled water.
To further isolate any liquid cooling hazard from the enclosures electronic components, the LHPL liquid cooled condensers 1026 are housed in a separate water tight section 1036 of the chassis. In this embodiment we have inserted an optional vertical air channel 1047 between the 1036 and the main portion of the chassis 1035 which houses the server electronics. This channel makes it possible to create chassis in which air is used to remove the secondary heat loads employing a negative pressure rotary device that sits on the top of the rack cabinet, and exhausts heated air to another device. To help reduce energy consumption further, this embodiment includes a cold plate 1043 that circulates cold water throughout the server section. To eliminate external heat exchangers which cool heated air, we have inserted an optional liquid cooled heat exchanger 1046 that may be used to help restore air leaving the 1U chassis to its inlet temperature that is in thermal contact with the cold plate 1048. In an airtight version of this 1U chassis, this heat exchanger would be coupled with one or more low speed cooling fans to remove heat from ambient air within the chassis. This chassis has a pair of PCIe GPU cards 1027 which each is cooled by an LHPL whose evaporator 1031 may be seen on the left hand side device. The orientations of the GPUs are different, hiding the evaporator for the right hand GPU in
The liquid cooled cold plate 1043 seen in
Two cases are considered here, one in which the condenser is air cooled (on the top of
To operate on very hot days, cooling towers vendors specify for a particular outside air temperature what temperature the external coolant they are cooling needs to hit. For example, on the hottest day of the year in Atlanta Georgia a cooling tower vendor specified that it could chill the liquid leaving the tower to 30 C if the external coolant it was tasked to cool arrived at its inlet at 35 C. This 5C “delta T” is typical of heat transfer devices that do not consume excessive amounts of energy to operate their cooling fans. The power required to move an external coolant to and from a cooling tower and run its cooling fan is a small percentage of the power required to run a water chiller. In the case of an air cooled LHPL, the object becomes simply to maximize the temperature of the coolant leaving the LHPL while at the same time providing enough cooling to guarantee that the device being cooled does not exceed its maximum temperature.
In both cases, the external coolant exit flow temperature is raised by reducing the flow rate of the external coolant. The temperature of the device being cooled is typically monitored by a diode inside of it, in the case of a CPU or GPU being cooled by an LHPL, this temperature may be obtained from programs that run on these devices. However not all semiconductors have such devices embedded within them. The two circuits provide two different methods for obtaining die temperatures. In both circuits a temperature sensor who's net labeled TS #1 produces a variable used by the microcontroller called Tevaporator_out. The temperature sensor is an device that may be implemented using different devices including thermocouples and thermistors is mounted either in or on the side of the thin heat spreader that conducts heat between the device being cooled and the LHPL evaporator. In the upper circuit used to cool an air cooled LHPL, a temperature sensor who's net is also called out as TS #1 Tevaporator_out is attached to the vapor line several centimeters from the point where the vapor line leaves the evaporator. The temperature measured by TS#3=TCond-out at the start of the liquid return line will typically be half way between the temperature of the device being cooled and the outlet temperature of the external coolant leaving the condenser. For every LHPL, this difference in temperature may be used to produce a look up table which will provide a measure of the total power being rejected by the LHPL which along with the temperature of the evaporator may be used to compute the die temperature by essentially adding to the evaporator temperature the temperature lost in the heat spreader which may be computed by multiplying the measured thermal resistance between the heat spreader and the die by the current heat flow being rejected.
Both circuits employ an microcontroller that in its operation is very similar to the standard IPMI devices that are now used to monitor and control motherboards. The method that the microcontroller uses to communicate with the outside world is also, in this case taking the output of a UART and employing an RS-232 to RS-485 converter to drive an RS-485 signal pair (shown as a single line, as are all the other circuit pairs in these schematic overviews). Often, the embedded CPU employs a GigE interface (instead of RS-485) and usually carries out a number of tasks, not shown, such as monitoring voltages, reading and controlling other cooling fans as well as monitoring reset and on/off switches and generating these signals remotely. In this case the 8051 embedded CPU contains the interfaces needed to both read A/D lines (i.e. temperature sensors and voltages) as well as the I/O control lines needed to read and control pumps, fans, tach lines, limit switches and other devices.
In addition to maximizing external coolant temperature this control circuit may provide other beneficial services. After a rapid shutdown, it is possible to put some LHPLs into a dry state in which there is no working fluid in the evaporator, resulting in what is called dry out. By actively controlling the temperatures about the cooling loop during shut down and start up, it becomes possible to fine tune the operational parameters of these devices to eliminate them in solutions that require them. Both circuits show a heater element positioned in the vicinity of a compensation chamber (CC) located in the liquid return line (i.e. for controlling a CPL). The position chosen is and could have been anywhere about the loop where liquid phase working fluid accumulates when the LHP is cold, including the condenser. A thermoelectric (TEC) cooler is also shown positioned above the CC located within an evaporator (the case where the device is classified as an LHP) which may be used to draw working fluid to the evaporator during shut down. The TEC cooler may also be used to improve cooling at low power levels by preventing working fluid in the LHP CC from heating up due to shell conduction. In addition two or more temperature sensors could be positioned in the condenser to monitor the location of the point where the boundary between working fluid liquid and vapor sits: monitoring this location helps to maximize condenser performance, something that may be aided by changing the external coolant flow rate. In this case temperature sensor TS #2=TCond_out is situated at the point where the condenser tube leaves the condenser while TS #3=TCond_near_out is situated at a position on the condenser tube a small distance upstream of TS #2T=TCond_out. At low power settings increased condenser performance may be achieved by reducing the flow rate of the external coolant which will let the working fluid meniscus retreat towards the condenser's exit. As the meniscus within the condensation channel approaches the exit point it pays to increase heat transfer between the working fluid and the external coolant by increasing the amount of cooling which also keeps the meniscus at the exit point of the channel. This type of control is enabled by TS#2 and TS#3. Ideally, one would like the entire condensation channel to participate, as this maximizes the amount of condensation going on within the condenser, and produces the highest external coolant outlet temperature. The circuit also provides control lines for monitoring and controlling the speed of the cooling device, which in the case of a liquid coolant is most likely to be some form of valve that restricts coolant flow while in the case of an air cooled condenser it will be a fan speed control. The two lines which typically are included in IPMI microcontrollers that control motherboard reset and power supply on/off are also shown. GPUs are a good example of devices that typically run at a single cooling speed and do not employ IPMI control devices.
For these monitoring and control objectives to be met software needs to be written that simultaneously monitors working fluid and external coolant temperatures that gets employed as inputs to control cooling fans and pumps as well as initiating CC cooling or heating during shut down and low power operation.
The first action at the head of the loop is to read the input registers, which like the write registers are contained in an I/O block. The I/O blocks that are dotted contain constants that never change value, i.e. global constants used in calculations. The actual values stored for variables like Tevaporator_out do not have to be the last value read in. LHPL devices may experience instabilities that cause their temperatures to oscillate several degrees. The period of these oscillations is typically varied from 30 to 120 or more seconds. Just below the start point on the cycle path we have placed a WAIT statement whose value may be set and whose job is to wait some time before reading the next set of register values. The actual value of things like temperatures that are passed into the program implementing the flow chart does not need to be the current value. A more reasonable value is a running average taken over N cycles that is designed to average out fluctuations. This will avoid the control circuits to spend too much time chasing phantom changes in LHPL operation. Regarding the flow chart itself, in situations where decisions need to be made using several input parameters, we bind them together into a common inlet channel using a sold dot to indicate that all of the I/O parameters or constants enter the decision box through a single location. There is one decision box that might be a little confusing. Within this decision statement is the text, “if inputs true then try to reduce cooling.” The input to the top of this box is the “NO” from the last box, which we interpret to mean that Tdie<Tdie_max If this is true, and the result of the box to its left which is where the flow actually went is also true (which compares the external coolant temperature with the desired temperature) we are in a position to attempt to reduce cooling, otherwise we go to cycle and repeat the entire loop.
The first job that the flow chart approaches is the most troublesome case, which would be an LHPL cooling device that was not adequately cooling a CPU or GPU. While we may in some instances simply read the temperature of the device, in situations where we may not, such as a GPU, we need to compute that temperature. There are several methods that we could take, for example measuring the energy flowing into the external coolant by measuring the flow rate, inlet temperature and outlet temperature. We have instead chosen building a table whose input parameter is the vapor temperature leaving the LHPL evaporator. For any particular LHPL, the power being rejected may be determined from the evaporator temperature (except in the case of an LHPL that is in a peculiar orientation which we discuss below). A pair of look up tables get used to compute TDie as well as a the expected temperature of working fluid entering the liquid return line TCond_out and the computed die temperature TDie. This last value is the expected temperature of the silicon die we are cooling. It is employed in the first decision box to determine if the die temperature is greater or equal to the maximum die temperature allowed, which normally would be a temperature slightly below the maximum temperature at a point where the lifetime of the semiconductor die is not impacted. If this condition is met, we have a problem and now have to check if we may increase cooling by checking if either the pump (i.e. valve controlling a flow) or the fan being used to cool the LHPL condenser may have its speed increased. If it has reached Max speed we have a problem that may only be resolved by asking the device which controls the speed of the device being cooled, to reduce its speed, by presumably reducing its clock frequency. Max_speed is derived from the input variable Current Speed. It is either computed, or possibly is set by limit switches in the case of a valve being controlled by a servo circuit. In the case where we have hit both limits, we have no choice but to send a message to the machine controlling the server being cooled, to reduce frequency. In most motherboards used today, this actually happens automatically, using built in circuitry that uses a similar scheme. If we have not reached Max Speed, we send a message or a signal to the device controlling our cooling device to increase the flow rate.
If the die temperature was below TDie, we are in a position to increase the temperature of the external coolant by reducing the cooling rate. However, before we do that, we need to check to make sure that what is coming out of the condenser is a liquid and not vapor. To do that we again go to a look up table that contains the value of TCond_out_comp. As long as the Condenser outlet temperature TCond_out is less than TCond_out_comp, we still have liquid leaving the working fluid exit of the condenser, and it is safe to increase cooling further. The entire flow chart is, in any real device the sensors used to collect information about the operating conditions of the LHPL are likely to be different than those we have chosen. In addition, in situations where the performance curve of the device being cooled is a multi-valued function (has more than two or more temperatures corresponding to different operating point temperatures) a more detailed analysis of the total heat load rejected to the external coolant will be needed in conjunction with a measurement of working fluid vapor temperature.
In situations where scale builds up in condenser modules and miraculously does not in the passageways typically made of the same materials used in heat exchangers, scaling may be a problem, especially in evaporative cooling towers where evaporation leaves minerals behind. Dry cooling towers in which the water passed to LHPL condenser modules may be kept clean by using a closed loop system (the water does not participation which is how minerals that do not leave with evaporation end-up being left behind unless water is continuously let out of the system to keep the concentration of scale down while at the same time adding biocides that keep bacteria from becoming an issue. There are two types of water coolers that may solve this sort of problem the first retains a closed loop, pumping the coolant to the roof and passing it through heat exchangers that exchange the heat being rejected to the outside air without employing evaporative cooling. When this is not adequate, a hybrid cooler may be employed that is essentially a dry cooler in which water is sprayed on the heat exchanger of a dry cooling causing evaporation to occur.
One of the basic components of this series of patent applications is the cooling of secondary components. In many instances, the amount of heat rejected by DIMM modules may approach the power rejected by the processors that they feed with data, making it even possible to consider cooling them with LHPLs. For the most part, however, the amount of heat that they reject is about 40% of the heat being rejected by CPUs, so going to these extremes is not necessary. However, the amount of heat that they reject may be large enough to make the cooling of some motherboards housed in 1U enclosures almost impossible, so demonstrating how DIMM modules may be cooled turns out to be a crucial issue in the use of LHPL cooling. Several of our embodiments used devices including ordinary heat pipes and cold plates to cool DIMM modules.
In two of the LHPL embodiments presented a more effective approach to cooling the DIMM modules was undertaken. The circle 839 represents either a tube carrying coolant that turns the plate 830 into a liquid cooled cold plate or a standard heat pipe. In the case of a tube carrying coolant, the heat would end-up being directly to the secondary coolant in the system. In the case of a standard heat pipes, the heat could be dumped either to the secondary coolant, be it air using a forced convection heat exchanger of a liquid cooled cold plate. Another alternative that we explore is simply terminating the standard heat pipe on the heat receiving end of an LHPL evaporator which is also being used to cool something else, like a CPU.
This embodiment includes a PCIe card 906 plugged into the computer's “IO channel” whose rear chassis outlet holes 911 may be seen at the left hand side of the chassis in the rendering and which is similar to the card being cooled by an LHPL in
A study of Thermo siphons used to cool a computer was conducted by Hewlett-Packard (HP) that demonstrates the orientation problem. This study employed a “microchannel” like evaporator design that has additional problems that the LHPL evaporators we employ do not, including instabilities caused by bubbles that cause them to boil over at high heat loads. Simply comparing the results of the HP study with ours we discovered that even with small distances separating the evaporator and condenser their performance was at a distinct disadvantage their thermal resistance of 0.41 C/W was roughly twice as large as ours cooling a similar load sitting at three times the distance. However their device turns off when tipped or turned upside down while our device may work in all orientations at peak power, which is the reason that HP has rejected the use of a thermo siphon in its commercial computer products.
CPLs and LHPs (Loop Heat Pipes) were invented roughly 12 years after the first heat pipe with a separate liquid return line, at the same time in the United States and Russia for use in space vehicles. They work in all orientations and may be distinguished from each other by the location of their Compensation Chamber (CC): in CPLs the CCs are located in the liquid return lines while LHPs have CCs located within their evaporators. We use the term LHPL (Loop Heat Pipe Like) in this document when discussing devices which have CCs in either or both their evaporators and liquid return lines in addition to devices derived from them that include a vacuum pump in the vapor line. CCs play a crucial role in the start-up characteristics of all passive heat transfer devices and in the case of LHPs enable an important feature called auto-regulation, in which the working fluid moves out of the condenser into the CC as the heat being rejected increases. This effectively results in an increase in the volume of the condensation channel as the power goes up, increasing the condensation rate in the condenser which in turn enables the rejection of increased heat loads. Without a CC in the evaporator you lose this important performance feature which is unique to an LHP. A number of the mislabeled prior art devices do not include CCs guaranteeing poor performance at high power. The disadvantage of putting a CC in the liquid return line is a more complicated start up that may require that the CC be heated to move the liquid stored in the liquid return line into the evaporator to avoid dry out. We solve this problem below for situations where a CC in the liquid return line might prove useful by embedding a computer in our cooling system which we employ for regulating the operation of the device as conditions change, which includes attaching low power refrigeration devices such as a TEC (thermoelectric cooler) to the CC.
A major benefit of our approach to LHPL design is the performance of our evaporators whose high volume vapor output may make it possible to dramatically reduce the number of fans required to cool a 1U chassis by distributing the rejected heat using a set of condenser fins with large area. The high pressure vapor we produce makes it possible to reject heat over a large set of fins without the use of ordinary heat pipes, which are often used to improve the performance of a set of heat exchanger fins by moving heat away from the entry point in a heat spreader out to the fins. LHPs eliminate this need and at the same time guarantee that the thermal resistance between the working fluid and the fins themselves is minimized, which reduces the thermal resistance of air cooled condensers. This performance feature takes advantage of the LHP classic evaporator wick design which employs escape channels in the condensation zone to minimize pressure losses in the wick. We frequently found devices labeled CPL that was missing either a CC or escape channels or both. One such device a condenser that depended on gravity to return liquid to the evaporator (i.e. it was a thermosiphon). A peer reviewed publication for this device cooling a 1U chassis enabled us to compute its thermal resistance which turned out to be roughly a factor of two worse than our equivalent design. The crucial LHPL parameter that needs to be reduced to improve energy efficiency is thermal resistance. The bottom line on any invention is its commercial viability. Variations on the basic LHP or CPL designs that have been proven to work but do not result in improved cooling performance or which turn off when tipped or turned upside down are by definition, not commercially viable.
Another drawback of some heat pipe based Tower CPU coolers is the fact that the rejected heat is not carried to the periphery of the tower chassis, and is free to move about raising the ambient temperature within the chassis forcing the exhaust fans mounted on the periphery to pull more air through to cool everything. By contrast, the GPU LHPLs whose performance we highlight in
Another major problem for LHPs turns out to be heat conduction through the evaporator shell into the CC (which is inside the LHP evaporator) at low power. When the heat being rejected by the evaporator moves backwards into the CC by conduction through the evaporator shell, it heats up the working fluid entering from the liquid return line. At high powers this heating effect is handled by the forward motion of the working fluid through the CC. To improve low power performance we have added to this disclosure cooling devices that extract heat from the evaporator shell in the vicinity of the CC that include finned heat exchangers and heat conducting strips. The net benefit of these cooling devices may be as much as a 10 degree reduction in the temperature of the semiconductor device being cooled. Other devices could easily be employed to cool CCs, tiny coolers with embedded blowers, ordinary heat pipes that reject heat to nearby cool objects, liquid cooled cold plates and thermoelectric (TEC) coolers. TEC coolers could also be used to reposition working fluid in the CC at shut down or start up to avoid dry out. Their use for this purpose presupposes the existence of a control device which turns them on at start up or shut down.
For many years the emphasis in the LHPL prior art arena has focused on creating evaporators that may reject higher and higher heat loads. We recently demonstrated a device that may reject 1000 Watts/cm.up.2. Our prior art includes evaporator that employ evaporator wicks with vapor escape channels that reduce the pressure losses associated with the vapor leaving the wick and entering the vapor line. One consequence of this approach is evaporators that may sustain a larger pressure drop across their wicks than devices using other geometries. The wicks themselves are typically made from sintered metal particles that are chemically compatible with the working fluid. A new evaporator was disclosed herein that makes it possible to inject heat into a flat evaporator on two sides at the same time. Except for this new form of evaporator, the design features of the wicks we employ are covered by prior art (U.S. Pat. No. 6,892,799).
One of the problems encountered in LHPL designs, is gathering up the heat from both the primary and secondary devices being cooled. In GPU designs in particular, a metal heat spreader is often attached to the GPU itself as well as the nearby components that cools these secondary heat rejecting devices. The heat is typically transported to a set of long parallel fins using as many ordinary heat pipes capable of being placed above the GPU itself. The upper limit to this approach becomes the contact area above the GPU that is free to accept ordinary heat pipes. It only takes a few heat pipes to consume all of this area and the distance that the heat has to be transported often limits the effectiveness of this approach by requiring that larger heat pipes be used to reject heat to portions of the fins that are farthest from the GPU. The space limitation that limits ordinary heat pipes in this application is not a problem for LHPs, whose evaporators may reject 500 or more Watts per square cm of heat spreader area. The remaining problem includes gathering the heat from the nearby secondary heat sources and conducting that heat to the LHPL's evaporator. A unique design we have come up with is a flat oblong evaporator that may absorb rejected heat on two sides at the same time, making it possible to cool the secondary components with either a heat spreader or ordinary heat pipe attached to one side while the other side cools the GPU itself. This evaporator could also be used to remove heat from a pair of semiconductor components at the same time, each of which has been located on either side of its flat sides.
The use of LHPLs to cool GPUs located on PCIe cards along with their use to cool other types of add in cards found in COTS (commodity off the shelf) computers including computer motherboards was already disclosed. We added new figures that demonstrate how this cooling may be employed in a typical air cooled desktop PC tower chassis that may also be laid on its side and employed as a desktop or workstation chassis. The embodiment here described is not limited to either PCs or to a particular type of add in card, nor to the geometry. It could also be used to cool any hot semiconductor mounted inside of an enclosure employing an air cooled condenser that is mounted to a cooling fan situated on the exterior surface or for that matter within an enclosure if cooling efficiency is not an issue, but heat transfer performance is.
Our use of the term LHPL includes LHPs, CPLs and ALHPs: devices which employ a pump in the condenser line that heretofore was intended to extend the distance between an evaporator and its condenser. In this disclosure we add another use for pumps in LHPLs, and that is to reduce the operating temperature of the evaporator by reducing the vapor pressure in the evaporator and vapor line. While a miniature LHP has been developed that may reject almost 1,000 Watts per cm squared, one of the problems with this device is that it reaches this transfer rate at an evaporator temperature of 120.degree.C. In the case of a very high performance multi-core semiconductor device, lowering this temperature by adding a vacuum pump in the vapor line that reduced the vapor pressure at the exit point of the wick evaporator produces a hybrid cooling solution that provides both high levels of heat rejection while at the same time preserving some of the passive benefits of LHPL technology. Such an ALHP is actually a hybrid cooling cycle half way between a refrigerator and a Loop Heat Pipe.
Another problem that this disclosure provides a solution for is the use of a single evaporator design employed to cool LHPLs whose condensers and evaporators are separated by different distances. The volume of the lines used to connect components together impacts things like the size of the CC. For a single evaporator to work with designs whose distances vary, balancing line length using either a serpentine shaped liquid return line or by placing a volume in the liquid return line (which converts an LHP into a CPL complicating start up issues) provides a solution that enables the same evaporator to be used in different locations.
To improve the overall thermal resistance of an enclosure cooling solution the crucial issue for us became the design and location of LHPL condensers. Ultimately, it is the transitions that the rejected heat has to make as it passes across the metal barriers separating the working fluids that make the largest contributions to thermal resistance. Improvements to the condensation channel design include reducing the thermal resistance between the condensation channel itself and the devices used to cool the channel and carry off the heat as well as the location of the condenser. In the case of both air and liquid cooled condensers, placing the condensation channel between a pair of cooling devices, including either a pair of fins or a pair of liquid cooled cold plates, results in the best channel performance by minimizing the distance that the heat has to flow through metal to reach the external coolant.
In the case of a serpentine shaped tubular condensation channel, that means placing the tube at the center of the fins. One of the problems with these designs in particular is the intensity of the heat being rejected by the condensation channel, which may be significantly greater in a two phase device than in single phase heat transfer devices. To help distribute this intense heat in the case of an air cooled condenser the solution includes increasing the amount of metal at the base of a fin set. The designs already presented include finned Aluminum condensers where the metal that made contact with the condenser tubing was beefed up and made to wrap around the tubing. A similar design was added to this disclosure in which inexpensive commodity copper CPU heat sinks had channels machined into their base areas making it possible to surround the condensation channel with the base plates of these condensers. In situations where the pressure loss of a serpentine tubular shaped condensation channel presents a problem, a tubular condenser that employs a manifold to distribute and collect the working fluid may be employed to drive a set of fins.
This disclosure includes planar condensation channels that are both air and liquid cooled and which are also placed at the center of the condenser, making it possible to maximize the heat being extracted from the channel. Studies made of these channels at the authors laboratory show that the liquid condenses out along the edges of the channel. To minimize the pressure loss of the channel which helps reduce thermal resistance by reducing vapor pressure losses between the evaporator and the point in the condensation channel where liquid condensate appears, the channel must be kept clear of obstacles. In the ideal case, the heat flows out through both the top and bottom of the channel to either a set of air cooled fins or a flowing liquid on both sides, whose thermal conductivity has been increased, as the low liquid flow rate needs some form of disruptor to bring it into full contact with the condensation channel walls. In situations where the geometry requires that the condensation channel inlet and outlet are on the same side of the channel, the length of the condensation channel may be increased by simply inserting a barrier down the middle that produces a U shaped channel which enables the condensation channel to have its inlet and outlet on the same side.
The original disclosure included a liquid cooled serpentine condenser channel whose performance was enhanced using the techniques mentioned in the prior paragraph that also eliminated heat conducting thermal shorts between the external coolant inlet and outlet and which also employed a counter-flow geometry to maximize the temperature of the external coolant at the condenser outlet, thereby continuing the reduction in thermal resistance. To further improve the performance of this device, a disruptor made of a wire was wound around the serpentine shaped condensation channel before it was enclosed in a liquid cooling jacket. Where possible, these methods should be applied to all liquid cooled condensers and most of them were incorporated in the two new designs. One of the problems with serpentine shaped condensation channels is that they take up a lot of space. In this disclosure the two new compact liquid cooled condensers combine the properties of both the serpentine shaped condensers already disclosed with the properties of the planar air cooled condensers. The first design employs a cylindrical condensation channel while the second employs a planar channel. On both the inside and outside walls of the cylindrical design we provide another pair of cylindrical channels that carry the external liquid coolant. A similar strategy is employed in the planar design to cool the central condensation channel.
An air cooled planar condenser in which both the top and bottom of the condensation channel was simultaneously cooled by a set of fins was employed to cool a pair of CPUs that were each cooled by LHP evaporators that shared the single high performance miniature condenser. Two types of condenser channels were tested: U shape channels in which the working fluid enters and leaves on the same side, and a design in which the flow passes in a straight line. Prior art for the use of a similar planar condenser that only cooled a single side of a condensation channel was found: the performance of this device when compared with our devices would have been limited by both the evaporator and condensation channel design. However a further reading of that patent reveals that what is actually claimed is not any form of LHPL, but rather an ordinary heat pipe that employs a wick filled liquid return line first patented in 1969 (U.S. Pat. No. 3,543,839). This device is missing the secret ingredient of both LHP's and CPL's, which is an evaporator that produces a high speed vapor flow capable of carrying the working fluid about the loop with a minimal loss in pressure. As a consequence, the vapor flow velocity in the condenser is low, making it possible to insert a disruptor into the working fluid to improve heat transfer, something not needed in either an LHP or CPL condensation channel, but which is useful in the external coolant channel when the external coolant is a liquid. Our condenser design provides the exceptional performance required to justify its manufacture for products which need to reject large amounts of heat in small spaces and it includes the use of CC cooling to improve performance accomplished with a carbon strip to remove heat from the CC and conduct it to the body of the air cooled blower.
The lowest thermal resistance electronic enclosure cooling device yet discovered employed liquid cooled LHPs in conjunction with secondary cooling devices that rejected the heat that they collected to the same external liquid coolant. This disclosure provides an embodiment that elucidates features discussed or pictured in our initial specification which employs flat planar condensers disclosed in the U.S. Provisional Application 60/923,588 filed Apr. 4, 2007 that serve the same function of the liquid cooled serpentine condensers used in
Another issue that needs to be taken into consideration is how do we manage two different liquid coolant streams that get combined at some point into a single external coolant stream. There are several possible approaches. In many systems, the primary heat load is 60% of the total IT load while the secondary is 40%. Eliminating the fans typically needed, changes this to a 70%/30% ratio. In the situation that we investigated in which we cooled a pair of 100 Watt primary loads, given a source of 30.degree.C external liquid coolant, we had no problem meeting the return temperature of 35.degree.C required by a cooling tower in Atlanta Georgia running at the hottest day of the year. In fact, we could return coolant whose temperature was 45.degree.C. This delta T was three times the required amount and provides the headroom required to use the liquid coolant we have preheated cooling secondary components to then reject the heat being released by the LHPL condensers. Alternatively, we could have mixed the two heated liquid coolant streams together, and by reducing the flow rate to the LHPL condenser increased the delta T produced by it, making it possible to increase the mixed outlet temperature to the desired 35 degree. C level.
The liquid cooled 1U chassis embodiment shown in
Although the steps of the method of assembling the device illustrated herein may be listed in an order, the steps may be performed in differing orders or combined such that one operation may perform multiple steps. Furthermore, a step or steps may be initiated before another step or steps are completed, or a step or steps may be initiated and completed after initiation and before completion of (during the performance of) other steps.
The preceding description has been presented only to illustrate and describe embodiments of the methods and systems of the present invention. It is not intended to be exhaustive or to limit the invention to any precise form disclosed. It will be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the scope of the invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from the essential scope. Therefore, it is intended that the invention not be limited to the particular embodiment disclosed as the best mode contemplated for carrying out this invention, but that the invention will include all embodiments falling within the scope of the claims. The invention may be practiced otherwise than is specifically explained and illustrated without departing from its spirit or scope. The scope of the invention is limited solely by the following claims.
Claims
1. A cooling device to cool a plurality of heat rejecting components and a plurality of other components, comprising:
- an enclosure housing enclosing the heat rejecting components and the other components;
- an external heat rejection device including a gaseous external coolant that transfers a primary heat load from a primary cooling system and a secondary heat load from a secondary cooling system to an environment outside of the enclosure housing, the enclosure housing includes an enclosure entry port and an enclosure exit port that allows the gaseous external coolant to pass into and out of the enclosure housing;
- a first rotary cooling device sitting within or without the enclosure housing that causes the gaseous external coolant to circulate within the enclosure housing and to pass through the enclosure housing entering it through the enclosure entry port and leaving through the enclosure exit port exchanging the primary heat load with the gaseous external coolant using the primary cooling system and the secondary heat load with the gaseous external coolant with using the secondary cooling system, the primary cooling system includes a loop heat pipe like or LHPL device, the primary cooling system cooling a primary heat rejecting component that produces the primary heat load, wherein the primary heat rejecting component is one of the heat rejecting components, the LHPL device includes: an evaporator module; a condenser module; a vapor line; a liquid return path; and a working fluid having a liquid phase and a vapor phase, wherein the primary heat load produced by the primary heat rejecting component being cooled causes the liquid phase working fluid within the evaporator module to change from the liquid phase to the vapor phase, the vapor phase leaves the evaporator module passing through the vapor line and into the condenser module where the working fluid releases the primary heat absorbed in the evaporator module and returns to the liquid phase, the liquid phase then leaves the condenser module passing through the liquid return path and the liquid phase working fluid returns to the evaporator module, the evaporator module includes: a component evaporator heat spreader; an evaporator body; and an evaporator component clamp, wherein the component evaporator heat spreader is clamped to the primary heat rejecting component providing thermal contact to transfer the primary heat load produced by the primary heat rejecting component being cooled to the evaporator body by reducing the thermal resistance between the primary heat rejecting component and the evaporator body, the evaporator body includes: an evaporator outer shell; a working fluid inlet port; a final compensation chamber; a working fluid exit port; and an evaporator wick having a plurality of vapor escape channels, wherein the evaporator body receives the liquid phase working fluid through the working fluid inlet port where the liquid phase working fluid enters a space between the evaporator outer shell, the working fluid inlet port and an evaporator wick working fluid entrance surface forming the final compensation chamber before flooding the evaporator wick working fluid entrance surface and then passing into and through the evaporator wick by capillary action where the working fluid absorbs the primary heat load being rejected by one of the primary heat rejecting components causing the liquid phase working fluid to boil and change to a vapor phase working fluid that includes the absorption of the heat of evaporation of the working fluid resulting in the vapor phase working fluid carrying off the primary heat produced by one of the primary heat rejecting components as it flows out of the evaporator wick using the vapor escape channels that provide a low flow resistance path for the vapor phase working fluid to leave the evaporator wick before passing out of the evaporator module through the working fluid exit port, the condenser module includes: a condensation channel; a condensation channel wall; a condensation channel working fluid inlet; a condensation channel working fluid exit; a condensation channel to external coolant thermal interface that includes: a condensation channel wall to cooling object thermal interface; an external coolant passageway; a second rotary cooling device located within or without the enclosure housing that pulls or pushes the gaseous external coolant through the external coolant passageway; and a plurality of cooling objects within the gaseous external coolant passageway whose surface the gaseous external coolant flows over thereby removing the heat they reject to the gaseous external coolant, wherein the gaseous external coolant whose velocity is controlled by an external rotary cooling device is directed into the gaseous external coolant passageway where it comes into thermal contact with the surfaces of the cooling objects placed within the gaseous external coolant passageway where they receive the heat from the condensation channel wall to cooling object thermal interface that they pass to the gaseous external coolant that comes in contact with their surfaces and which the gaseous external coolant transports into or out of the enclosure housing, the heat originating within the condensation channel whose condensation channel wall receives the heat from the vapor phase working fluid that enters the condensation channel through the condensation channel inlet where it comes into contact with the condensation channel wall that employs the condensation channel to cooling object thermal interface that is part of the condensation channel to gaseous external coolant thermal interface to first remove the sensible heat from the vapor phase working fluid causing the vapor phase working fluid to condense the resulting phase change releasing the heat of condensation that with the sensible heat includes all of the heat that was rejected by the primary heat load in thermal contact with the evaporator module at which point the liquid phase working fluid exits the condensation and the condenser module using the condensation channel working fluid exit that connects to the liquid return path, the liquid return path includes: a wick entrance surface near the evaporator module's working fluid inlet port; a liquid return tube that connects the condensation channel working fluid exit and the wick entrance surface within the evaporator body; an electric heating device, the liquid phase returning to the evaporator working fluid inlet port through the liquid return tube where it encounters a space within the evaporator body between the working fluid inlet port of the evaporator body and the evaporator wick entrance surface at the end of the evaporator wick near where the liquid return tube is attached to the evaporator body providing an incoming liquid phase working fluid with access to the wick entrance surface that is flooded by the liquid phase working fluid entering the evaporator wick, the space between the evaporator wick entrance surface and the evaporator shell in the vicinity of the working fluid inlet port forming a storage volume for the liquid phase working fluid within the liquid return path called a default compensation chamber that sits at the point where the liquid phase working fluid enters the evaporator body flooding the evaporator wick; a plurality of optional attached compensation chambers that connect to the liquid return path using a condensation channel attachment tube or CC attachment tube that connects the optional attached compensation chamber to the liquid return tube anywhere along the liquid return tube or to the evaporator body between the working fluid inlet port and the wick entrance surface which requires the electric heating device during the LHPL device start-up thereby adding storage volume to a default compensation chamber; an optional compensation chamber within the evaporator body between the working fluid inlet port and the wick entrance surface created by increasing the distance between the working fluid inlet port and the wick entrance surface thereby adding to the storage volume of the default compensation chamber; and a plurality of optional inline liquid return path chambers located anywhere along the liquid return path within the liquid return tube that increase the volume of the liquid return path, wherein the liquid phase working fluid passes from the condensation channel working fluid exit into the liquid return tube and through the evaporator module's working fluid inlet arriving at the wick entrance surface of the evaporator wick within the evaporator body, and the optional compensation chamber within the evaporator body or the default compensation chamber within the evaporator body as well as the optional attached compensation chamber connected to the liquid return path or the evaporator body using the CC attachment tube making it possible to employ compensation chambers to adjust the point within the condenser module's condensation channel where a boundary forms between the vapor phase working fluid and the liquid phase working fluid, whereby the LHPL device's heat rejection performance is improved by moving this boundary as close as possible to the point where the liquid phase working fluid exits the condensation channel through the condensation channel working fluid exit and thereby increasing the condensation wall area available for exchanging heat with the working fluid within the condensation channel, where the working fluid in the vapor phase comes into contact with the condensation channel wall and the optional inline liquid return path chambers located within the liquid return tube providing the liquid phase working fluid with a storage volume making it possible for a single sized evaporator module to employ a plurality of the liquid return paths of different lengths that contain the same volume of the liquid phase working fluid thereby making it possible for the LHPLs whose distance between their evaporator and condenser modules varies being manufactured using a single volume working fluid for all of the LHPLs, the secondary cooling system includes: the secondary coolant being the same gaseous external coolant used to cool the LHPL device that now circulates through the enclosure housing cooling a secondary heat rejecting component, wherein the secondary heat rejecting component is one of the other components; an optional finned heat exchanger in thermal contact with the secondary heat rejecting component rejecting its heat to the gaseous external coolant; an optional large thermally conductive surface that is in thermal contact with the secondary heat rejecting component over which the gaseous external coolant flows and whose surface area is large enough given the velocity of the gaseous external coolant passing over the surface to reject the secondary component's heat in thermal contact with the gaseous external coolant; an ordinary heat pipe in thermal contact with a first LHPL evaporator module whose condenser module is cooled using the gaseous external coolant, the LHPL device primary task being to cool the primary heat rejecting component where the opposing end of the ordinary heat pipe is in thermal contact with a secondary heat rejecting component thereby transferring this secondary heat load to the LHPL device which then rejects it to the gaseous external coolant; and an optional second rotary cooling device that directs the gaseous external coolant to flow across the optional finned heat exchanger, the optional finned heat exchanger and the optional large thermally conductive surface except for the optional second rotary cooling device all of which are located within the enclosure housing, the optional second rotary cooling device being placed within or without the enclosure housing and being powered by a source of electricity or another form of power, wherein the secondary heat produced by the secondary heat rejecting component is released directly to the gaseous external coolant flowing over the optional finned heat exchangers or the optional large thermally conductive surface all of which employ forced convection to deliver their heat to the gaseous external coolant while they are also in thermal contact with the secondary heat rejecting component or to the ordinary heat pipe that is connected to a second LHPL evaporator module that also employs forced convection to deliver its heat to the gaseous external coolant, wherein all of the secondary heat rejecting components reject their heat to the gaseous external coolant along with the primary heat rejecting component cooled by the LHPL device that also employs forced convection to pass its heat to the same gaseous external coolant which then either circulates within the enclosure housing before passing out of the enclosure exit port or exits directly out of the enclosure housing after leaving the LHPL device through its condenser module passageway that is connected to the enclosure exit port.
2. The cooling device according to claim 1, wherein the primary heat rejecting component situated within the enclosure housing further comprises a component mounting device that the primary heat rejecting device is mounted to while providing a mechanism to hold an LHPL evaporator module in thermal contact with the primary heat rejecting component.
3. The cooling device according to claim 2, wherein a LHPL condenser module is attached to the enclosure or the component mounting device or some other structure within the enclosure.
4. The cooling device according to claim 2, wherein a primary electric rotary cooling device is mounted to the condenser module or to the enclosure in a position to force gaseous coolant to flow over the condenser fins in the process removing the primary heat from the primary heat rejecting component being cooled by the LHPL cooling device.
5. The cooling device according to claim 2, further comprising a plethora of cooling fins optionally made of a heat conducting metal, wherein the condenser module is moved to one or more locations within the enclosure where there is more free space than is found in either the vicinity or directly above the primary heat rejecting component making it possible to employ the condenser module whose fin area is greater than the fin area available to an air cooled heat sink or finned heat exchanger that is mounted directly to the primary heat rejecting component, thereby the larger fin area making it possible to reduce the velocity of the external coolant passing over the fins of an LHPL cooling system's condenser module whose area is often a factor of 2 greater than the area of the fins of either a heat sink with fins or the air cooled finned heat exchanger mounted directly to the primary heat rejecting component.
6. The cooling device according to claim 5, wherein the larger fin area of the condenser module reducing the velocity of the air passing over the fins being inversely proportional to the increase in fin area made possible by the use of the LHPL device in turn reducing the rotational speed of the primary electric rotary cooling device whose rotational speed is proportional to the air velocity further reducing the energy required to run the rotational cooling device whose power approaches being proportional to the rotations per minute or RPMs of the device cubed, resulting in a factor of two increase in the fins that are producing up to a factor of 8 reduction in the power required to run an electric rotary cooling device while at the same time making a dramatic improvement in the reliability of the electric rotary cooling device used to cool the primary cooling system and reducing operating costs related to fan failures along with loud noise produced by small electric rotary devices running up to 20,000 RPMs.
7. The cooling device according to claim 2, further comprising the environment outside of the enclosure housing being an HVAC cooling system that returns chilled air to the enclosure, an enclosure wall with a hole that allows air to pass between the interior and exterior of the enclosure along with mounting the condenser module or the primary electric rotary cooling device to the enclosure wall near the hole.
8. The cooling device according to claim 7, wherein the primary heat rejecting component situated within the enclosure mounted to the component mounting device creates thermal contact between the LHPL evaporator module and the primary heat rejecting component.
9. The cooling device according to claim 8, wherein the LHPL condenser module is attached to either side of the enclosure wall near the hole enabling the primary electric rotary cooling device to pass air through the condenser module enabling air to be removed from the enclosure's interior.
10. The cooling device according to claim 9, wherein the condenser module situated on either side of the hole in the enclosure wall lets the ambient air within the enclosure flow out of the enclosure to convectively cool the condenser module sitting on either side of the enclosure wall resulting in a reduction of the temperature of the ambient air within the enclosure that would not have occurred if the condenser had been mounted in front of an air inlet hole or within the enclosure in a manner that would not have allowed the heated air exhausting from it being directed out of the enclosure, instead mixing with the ambient air within the enclosure allowing recirculation to take place within the enclosure resulting in pre-heated air making more than one pass through the condenser module so that its exhaust is not directed out of the enclosure, thereby the mounting of the condenser module is so that the air being exhausted from it leaves the enclosure without mixing with the ambient air within the enclosure making an improvement in the energy efficiency of the primary and secondary cooling systems by reducing the temperature of the ambient coolant used by both the primary and secondary cooling systems which in turn reduce the required velocity of the air flowing over secondary components as well as the fins used in air cooled finned heat exchangers and an air cooled LHPL cooling device condenser module that in turn reduces the energy required to run the primary electric rotary cooling devices that provide the convective heat transfer, the increase in the volumetric heat content of the air leaving the enclosure through all of its vents improving the energy efficiency of the HVAC cooling system used to cool the heat rejecting components in the enclosure is placed in a geographic location that requires the HVAC cooling system to deliver the primary and secondary rejected heats to the outside air or to an external heat sink such as a body of water or the earth.
11. The cooling device according to claim 2, further comprising a passageway that brings air from the outside of the enclosure housing to the LHPL condenser module within the enclosure and a passageway that removes air exhausted by an air cooled condenser and delivers it to the air outside of the enclosure without pre-heating the air within the enclosure.
12. The cooling device according to claim 11, wherein the condenser module of the LHPL cooling device is situated outside of the wall of the enclosure housing or is provided cool air by the passageway that has not been pre-heated by components within the enclosure and a second passageway that delivers the exhaust of the condenser module to the outside air without pre-heating the ambient air within the enclosure.
13. The cooling device according to claim 11, wherein the component and the evaporator module of the LHPL cooling is located within the enclosure the temperature of the air used to cool the LHPL's condenser module having not been raised by the pre-heated air within the enclosure making it possible for the LHPL cooling device whose condenser module does not receive pre-heated air to increase the amount of heat that the LHPL cooling device rejects making it possible to cool primary heat rejecting components whose rejected heat is related to their throughput allowing primary heat rejecting components that deliver their rejected heat to air cooled LHPL cooling devices to increase their operating frequencies and computational throughput while not impacting the secondary cooling system if the rejected primary heat does not pre-heat the ambient air within the enclosure.
14. The cooling device according to claim 2, further comprising a component mounting device that is a circuit board, wherein the primary heat rejecting component and the secondary heat rejecting component are situated within the enclosure and the primary and secondary heat rejecting components are mounted to the circuit board providing them with electric power often containing layers that conduct both electricity to components mounted on the circuit board while conducting heat away from the components that reject their heat to copper layers of the circuit board using pins that connect the components to the copper layers and copper via that conduct heat between the internal copper layers and the top and bottom layers of the circuit board which in the case of circuit boards with large areas make it possible to convectively extract heat from the circuit board by flowing air over the circuit board's surfaces or to mount air cooled finned heat exchangers to the exposed copper layers to further increase the surface area of the circuit board delivering rejected heat to the secondary coolant.
15. The cooling device according to claim 1, wherein the primary and secondary rotary electric cooling devices further comprise a conduit placed adjacent to one or more enclosure housings to which mating sealed holes are placed in both the enclosures and the conduit, the enclosure coming together with conduit using an air tight seal between the holes forming an air passageway between the enclosure and conduit and employing the rotary cooling device that need not be powered by electricity being mounted to the conduit so as to exhaust air out of the conduit reducing the air pressure within the conduit and the enclosure below the pressure outside the enclosure.
16. The cooling device according to claim 15, wherein a pressure gradient being established between the conduit and the air outside of the enclosure causes air to be sucked into the enclosure through an entrance hole in the wall of the enclosure, the air flowing into the entrance hole through the enclosure with a velocity capable of convective cooling of primary and secondary heat rejecting components, the air being exhausted out to the passageway and the conduit before leaving for an external HVAC system in the case where the outside air is not cold enough to cool the components within the enclosure, thereby making it possible to cool the heat rejecting components within the enclosure without the need for primary electric rotary cooling devices replacing them with large rotary cooling devices whose energy efficiency is much higher than the small primary electric rotary cooling devices typically employed within enclosure housings and completely eliminating the need for small primary electric rotary cooling devices that also eliminates the noise that they generate, the conduit collecting all of the heated air being released to the conduit that normally is mixed with the cold air within the rooms that contain air cooled electrically powered enclosure housings that raises the temperature of the air entering the enclosure to convectively cool components causing the velocity required to get the same amount of cooling to rise which spreads the external heat out over a larger volume of cooling air that leaves the enclosure, reducing the temperature of the air leaving the enclosure ultimately raising the energy put into accelerating air throughout the system and reducing the energy efficiency of the HVAC system used to provide air to cool the room that the enclosure sits in, which except for the case when the outside air is cold enough to cool the enclosure rarely persists for more than four months of the year.
17. The cooling device according to claim 16, further comprising the environment outside of the enclosure housing being an HVAC cooling system that returns chilled air to the enclosure housing being a rack mounted chassis with a vent hole in the front panel of the chassis that allows cooling air to leave the chassis through an exhaust hole in the its rear allowing the heat cooling air to flow into an air tight passageway that connects the exhaust hole in the rear of the chassis with a conduit inlet that is cut into a rack mounted conduit that is mounted to the rack cabinet so will remain stationary in the rear of a rack cabinet, the conduit inlet forming the end of the air passageway that starts with the exhaust hole of the rack mounted chassis employing an airtight seal that guarantees when the rack mount chassis is fully inserted into the rack cabinet so that it butts-up against the rack mounted conduit so air cannot leak into the passageway from the outside, the rack mounted conduit employing a conduit vacuum seal that closes when the rack mounted chassis is removed from the rack cabinet preventing air to leak into the conduit when the rack mounted chassis is not installed, the vacuum seal being broken when the rack mounted chassis is inserted into the rack cabinet coming into contact with the mechanism that controls the conduit vacuum seal.
18. The cooling device according to claim 17, wherein an air flow path is established between the front panel vent hole of the rack mount chassis and the rack mount conduit that employs the rotary cooling device to exhaust air from the conduit thereby lowering the pressure in the conduit establishing a pressure gradient between the front panel vent hole and the rotary cooling device that sucks air into the chassis causing air to flow through the rack mount chassis with a velocity high enough to cool by convection the primary cooling and secondary cooling systems delivering the heat rejected by the primary and secondary heat rejecting components to the external coolant being air, thereby further improving energy efficiency by eliminating the need for small electric rotary cooling devices within the rack mount chassis and rack cabinet rear door fans, both of which being small are less energy efficient than the large rotary cooling device capable of being mounted on an air conduit or an air duct that remove air from either the conduit or the duct, the rack cabinet rear door fans creating air that leaves them at high velocity and is free to force its way around ducts used to capture the heat coming out of the rear of rack cabinets in a hot isle of a data center allowing hot air leaving the rack cabinet to recirculate within the data center room mixing with the air being produced by the HVAC cooling system that cools air directed into the room returning to the front of the rack mount chassis after mixing with the hot air being directed around the ducting designed to capture the hot air then mixing with the incoming cooling air causing it to rise in temperature before it enters the front panel vent of the chassis and being hotter than it would have been if a rack mounted conduit had been employed causing the electric rotary cooling devices and rear door mounted fans in a rack cabinet cooling system that does not employ a rack cabinet conduit to increase speed to compensate for the higher temperature of the external coolant that needs to be raised when removing a constant amount of heat from the heat rejecting components within the enclosure and in addition to consuming extra power also causing volumetric heat content of the air leaving the enclosure to fall resulting in the HVAC cooling system to employ more energy in its water chiller to reject the heat to the outside air and finally by eliminating the internal electric rotary cooling devices as well as the rear door fans completely eliminating noise.
19. The cooling device according to claim 1, wherein the gas cooled condenser module further comprises one or more heat sinks and a serpentine shaped condensation channel, wherein the serpentine condensation channel is thermally attached to either one or more heat sinks resulting in a condenser composed of a single heat sink whose base plate is thermally attached to the condensation channel or a condenser composed of two or more heat sinks whose base plates are both thermally attached to a condensation channel which is sandwiched between the base plates making it possible to flow the heat between the condensation channel and the base plates of the heat sinks thermally attached to it.
20. The cooling device according to claim 19, wherein the gas cooled condenser module further comprises a multitude of fins each containing a plurality of extruded holes whose spacing matches that of a group of condensation channels that are part of a network formed by attaching them to a pair of manifolds one of which receives working fluid vapor from the vapor line that connects it to the evaporator and distributes the vapor to the condensation channels attached to it while the second manifold situated on an opposing side of the network receives working fluid liquid from the condensation channels attached to it and returns it to the evaporator using the working fluid path, wherein the working fluid path is created that distributes incoming vapor to the group of condensation channels which have been thermally attached to the multitude of fins by the extruded holes in the fins whose spacing makes it possible to stack the fins over the condensation channels creating a thermal transfer mechanism between the condensation channel and the fins that enables the transfer of the heat of condensation from the condensation channels to the fins from which the heat of condensation gets passed to the air flowing down the length of the fins, resulting in the heat being rejected to the air while simultaneously returning the working fluid to its liquid phase before it leaves the condenser by flowing into an opposing manifold, which is connected to the liquid return path.
Type: Application
Filed: Feb 17, 2016
Publication Date: Aug 17, 2017
Inventor: Stephen Fried (Kingston, MA)
Application Number: 15/046,251