U.S. patent application number 17/720090 was filed with the patent office on 2022-08-04 for computer cooling system and method of use.
The applicant listed for this patent is Chilldyne, Inc.. Invention is credited to Steve Harrington.
Application Number | 20220248567 17/720090 |
Document ID | / |
Family ID | 1000006259512 |
Filed Date | 2022-08-04 |
United States Patent
Application |
20220248567 |
Kind Code |
A1 |
Harrington; Steve |
August 4, 2022 |
Computer Cooling System And Method of Use
Abstract
A reliable, leak-tolerant liquid cooling system with a backup
air-cooling system for computers is provided. The system may use a
vacuum pump and a liquid pump and/or an air compressor in
combination to provide negative fluid pressure so that liquid does
not leak out of the system near electrical components.
Alternatively, the system can use a single vacuum pump and a valve
assembly to circulate coolant. The system distributes flow and
pressure with a series of pressure regulating valves so that an
array of computers can be serviced by a single cooling system. The
system provides both air and liquid cooling so that if the liquid
cooling system does not provide adequate cooling, the air cooling
system will be automatically activated. The heat may be removed
from the building efficiently with a cooling tower. A connector
system is provided to automatically evacuate the liquid from the
heat exchangers before they are disconnected. Various turbulators
are also provided, as well as a system and method for optimizing
the heat transfer characteristics of a heat exchanger to minimize
total energy requirements.
Inventors: |
Harrington; Steve; (Cardiff,
CA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Chilldyne, Inc. |
Carlsbad |
CA |
US |
|
|
Family ID: |
1000006259512 |
Appl. No.: |
17/720090 |
Filed: |
April 13, 2022 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
15782034 |
Oct 12, 2017 |
11317535 |
|
|
17720090 |
|
|
|
|
14685524 |
Apr 13, 2015 |
|
|
|
15782034 |
|
|
|
|
13410558 |
Mar 2, 2012 |
9010141 |
|
|
14685524 |
|
|
|
|
61595989 |
Feb 7, 2012 |
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
F28D 1/0246 20130101;
H05K 7/20763 20130101; H05K 7/20772 20130101; F28D 3/00 20130101;
H05K 7/20272 20130101; H01L 23/473 20130101; Y10T 29/4935 20150115;
F28F 1/28 20130101; F28F 7/02 20130101; H05K 7/20281 20130101; B23P
15/26 20130101; F28D 15/0233 20130101; H01L 2924/0002 20130101 |
International
Class: |
H05K 7/20 20060101
H05K007/20; F28D 1/02 20060101 F28D001/02; F28D 3/00 20060101
F28D003/00; F28D 15/02 20060101 F28D015/02; F28F 1/28 20060101
F28F001/28; F28F 7/02 20060101 F28F007/02; H01L 23/473 20060101
H01L023/473; B23P 15/26 20060101 B23P015/26 |
Claims
1. A turbulator for use in a liquid flow passageway of a
coolant-containing heat exchanger that is adapted to transfer heat
from an electrical component to the coolant, the passageway having
a cross-sectional area, the turbulator comprising: a core that is
substantially concentric to the passageway, the core having a
cross-sectional area; and a ridge structure connected to the core,
the ridge structure radiating away from the core, and the ridge
structure defining a flow path, wherein the flow path has a length
more than twice the largest dimension of the passageway; wherein
the cross-sectional area of the core is at least 20% of the cross
sectional area of the passageway.
2. The turbulator of claim 1, wherein the ridge structure is
adapted to allow leakage of coolant over the ridge structure
sufficient to induce swirling of coolant, wherein the swirling of
coolant is substantially perpendicular to flow path.
3. The turbulator of claim 1 wherein the flow path defines a shape
selected from a group consisting a helix, a conical helix, a
rectangular cross-section helix, a round cross-section helix, a
rectangular cross-section single-entry helix, a rectangular
cross-section double-entry helix, a round cross-section
single-entry helix and a round cross-section double-entry
helix.
4. The turbulator of claim 1 wherein the turbulator is adapted to
direct a jet of coolant against a surface proximate the electrical
component.
5. The turbulator of claim 1 wherein the turbulator and passageway
define a second turbulence-inducing liquid flow path when the
turbulator is placed in the passageway, the second
turbulence-inducing liquid flow path being at least 25% shorter
than the first flow path.
6. The turbulator of claim 5 wherein the second flow path causes
swirling of the liquid in the first flow path.
7. A liquid cooling system for cooling an electrical component,
comprising: a coolant containing heat exchanger adapted to transfer
heat from the electrical component to the liquid, the heat exchange
comprising coolant flow passageway having a cross-sectional area; a
turbulator disposed of in the passageway comprising: a core that is
substantially concentric to the passageway, the core having a
cross-sectional area; and a ridge structure connected to the core,
the ridge structure radiating away from the core, and the ridge
structure defining a flow path, wherein the flow path has a length
more than twice the largest dimension of the passageway; wherein
the cross-sectional area of the core is at least 20% of the cross
sectional area of the passageway.
8. The system of claim 7, the heat exchanger further comprising a
base plate thermally coupled to the component and a plurality of
fins extending from the base plate.
9. The system of claim 7, the heat exchanger further comprising a
base plate thermally coupled to the device; a heat pipe thermally
coupled to the base plate; and the heat pipe thermally coupled to a
plurality of fins.
10. The system of claim 7, wherein the ridge structure is adapted
to allow leakage of coolant over the ridge structure sufficient to
induce swirling of coolant, wherein the swirling of coolant is
substantially perpendicular to flow path.
11. The system of claim 7, wherein the flow path defines a shape
selected from a group consisting a helix, a conical helix, a
rectangular cross-section helix, a round cross-section helix, a
rectangular cross-section single-entry helix, a rectangular
cross-section double-entry helix, a round cross-section
single-entry helix and a round cross-section double-entry
helix.
12. The system of claim 7, further comprising: a vacuum pump
adapted to propel the coolant through the passageway at less than
ambient pressure.
13. The system of claim 12, further comprising: a pressure sensor
in fluid communication with the heat exchanger and adapted to take
a pressure reading of the coolant; and a controller connected to
the pressure sensor adapted to signal an alert if the pressure
reading is outside a normal operable range.
14. The system of claim 12, further comprising: a pressure sensor
in fluid communication with the heat exchanger and adapted to take
a pressure reading of the coolant; a valve in fluid communication
with the heat exchanger; and a controller connected to the pressure
sensor and the valve, the controller adapted to open the valve to
allow the flow of coolant into the heat exchanger when the pressure
reading is within a normal operable range.
15. The system of claim 7, further comprising: a vacuum pump
adapted to remove the coolant from the heat exchanger when the heat
exchanger is removed from the system.
16. A method of minimizing the energy needed to cool
heat-generating components inside a cabinet having a higher than
ambient temperature, comprising the steps of: providing a heat
exchanger comprising: a thermally conductive base adapted to
thermally couple to the heat-generating components; a plurality of
thermally conductive fins extending outward from the base; and one
or more coolant pathways thermally coupled to the base and the
fins; balancing the thermal load of the heat generating components
and the ambient air inside the cabinet by positioning the one or
more coolant pathways relative to the base and the fins; thermally
coupling the heat exchanger to the heat generating components; and
providing a source of coolant to the one or more coolant
pathways.
17. The method of claim 16, further comprising the steps of:
providing a fan and locating the fan so that it causes air to flow
across one or more of the fins; and balancing the thermal load of
the heat generating components and the ambient air inside the
cabinet by: positioning the one or more coolant pathways relative
to the base and the fins in further view of the heat transfer
effect of the fan; and adjusting the speed of the fan.
18. The method of claim 16, wherein the one or more pathways is a
coolant flow passageway having a cross-sectional area; a turbulator
disposed of in the passageway comprising: a core that is
substantially concentric to the passageway, the core having a
cross-sectional area; and a ridge structure connected to the core,
the ridge structure radiating away from the core, and the ridge
structure defining a flow path, wherein the flow path has a length
more than twice the largest dimension of the passageway; wherein
the cross-sectional area of the core is at least 20% of the cross
sectional area of the passageway.
19. The method of claim 18, wherein the ridge structure is adapted
to allow leakage of coolant over the ridge structure sufficient to
induce swirling of coolant, wherein the swirling of coolant is
substantially perpendicular to flow path.
20. The system of claim 18, wherein the flow path defines a shape
selected from a group consisting a helix, a conical helix, a
rectangular cross-section helix, a round cross-section helix, a
rectangular cross-section single-entry helix, a rectangular
cross-section double-entry helix, a round cross-section
single-entry helix and a round cross-section double-entry helix.
Description
1.0 CLAIM OF PRIORITY
[0001] The present application claims priority as a continuation of
U.S. patent application Ser. No. 15/782034 filed on Oct. 12, 2017,
which is a continuation of U.S. patent application Ser. No.
14/685524 filed on Apr. 13, 2015, which is a continuation of U.S.
patent application Ser. No. 13/410558 filed on Mar. 2, 2012, now
U.S. Pat. No. 9,010,141 issued on Apr. 21, 2015, which is a
non-provisional of U.S. Patent Application Ser. No. 61/595989 filed
on Feb. 7, 2012. The full disclosure of each of these references is
herein incorporated by reference.
[0002] The present application is also related to U.S. Patent
Application Ser. No. 61/451214 filed on Mar. 10, 2011, U.S. patent
application Ser. No. 13/308208 filed on Nov. 30, 2011, U.S. Patent
Application Ser. No. 61/422564 filed on Dec. 13, 2010, and U.S.
patent application Ser. No. 12/762898 filed on Apr. 19, 2010. The
full disclosure of each of these references is herein incorporated
by reference.
2.0 TECHNICAL FIELD
[0003] The present invention relates to systems and methods for
cooling computer systems.
3.0 BACKGROUND
[0004] Arrays of electronic computers, such as are found in data
centers, generate a great deal of heat. An example Central
Processing Unit of a computer ("CPU") generates over 100 watts of
heat and has a maximum case temperature of about 60 C. An example
rack of 88 CPUs may generate 9 kW of heat. The outdoor temperature
at a hot urban location might be 45 C, so even in hot environments
heat can still theoretically flow away from the higher temperature
computer and toward the lower temperature outside environment.
Accordingly, no refrigeration of computers should be required,
theoretically. Nonetheless, the standard way to keep data centers
cool is to use expensive and relatively inefficient
vapor-compression refrigeration systems at least part of the time.
These conventional cooling or "air conditioning" systems often use
more power that the computers themselves, all of which is
discharged to the environment as waste heat. These systems use air
as the heat transfer medium, and it is due to the low heat capacity
and low thermal conductivity of air that refrigeration must be used
to remove the heat generated by multiple air heat exchangers.
Removing heat generated by heat exchangers is also referred to as
overcoming the thermal resistance of the heat exchangers. Some
operators use evaporation of cooling liquid to cool cooling
liquid-to-air heat exchangers that cool computers, and this is more
thermally efficient than refrigeration, but the computers run
hotter, reducing their reliability, decreasing their efficiency and
making the data center uncomfortable for human occupants. Water is
used as the cooling liquid or coolant throughout this disclosure,
but it will be known to those in art that other coolants may be
used. The cooling liquid may consist essentially of water,
including tap water, or may comprise one or more perfluorocarbons
or avionics cooling liquids. The cooling liquid may flow over a
plated surface.
[0005] Water has approximately 4000 times more heat capacity than
air of the same volume, so water is a theoretically ideal heat
transfer agent for direct heat transfer from heat generating
components. Other cooling liquids offer similar performance. Liquid
cooling is recognized as a thermally efficient way to cool computer
CPUs due to their high concentration of power and heat generation
in a small space, but the rest of a computer's electronics generate
heat at a lower rate and temperature, so air-cooling is appropriate
for much of the associated hardware. Current systems may use liquid
cooling to move the heat from the CPU to a radiator mounted close
to the CPU, or they may use an air-to-liquid heat exchanger to
remove heat from the computer enclosure and heat up liquid in the
heat exchangers. These systems suffer from the high thermal
resistance and bulkiness of air-to-liquid or liquid-to-air heat
exchangers. Other systems use a chilled cooling liquid loop to cool
the computer, but these systems require complex and expensive
connectors and plumbing to connect the server to the building
cooling liquid supply while ensuring that no leaks occur, which may
be devastating in or near a computer. Accordingly, operators of
server systems are rightly concerned about leaks and reliability of
cooling liquid-cooled computers. Furthermore, chillers require a
large amount of power. Additionally, for operation in a data
center, servers, particularly blade servers, need to be compact.
Therefore, what is needed is a compact cooling solution adaptable
for up to a large number of computers, one that combines and
balances air-cooling capacity for low-intensity heat sources with
cooling liquid-cooling capacity for high-intensity heat sources
while using a minimum amount of cooling liquid flow, and one that
is reliable, leak-free and low in power consumption.
4.0 SUMMARY
[0006] The present system addresses these issues and more by
providing in various example embodiments an efficient and compact
heat exchanger for a CPU utilizing liquid under negative pressure
to minimize chances of leakage, with an air-cooling backup system.
Also provided is a cooling solution that integrates with an
air-cooled heat sink for backup and utilizes only the minimum
amount of water necessary to provide adequate cooling for each
heat-generating element. Various embodiments further provide
systems and methods to cool the CPU, the server and the data center
with liquid in an optimal manner, by cooling the CPU to reduce
leakage current, removing heat from the data center by means of the
air cooled portion of the CPU heat exchanger, and utilizing an
outdoor evaporative cooling system or a dry cooler with a part-time
evaporative cooling system that eliminates the need for a chiller
in the liquid cooling system. Additionally, provided is a system
and method for disconnecting and reconnecting liquid-cooled heat
exchangers without losing any water. Heat exchangers employing
efficiency-increasing turbulators are also provided.
[0007] Provided in various embodiments is a system for cooling one
or more electrical devices inside a building, comprising: one or
more liquid coolant-containing heat exchangers thermally coupled to
one or more electrical devices and each having a liquid input port
and a liquid output port and containing liquid coolant at below
atmospheric pressure; a liquid coolant-containing chamber in fluid
communication with the liquid output port of the heat exchanger(s),
the chamber containing liquid coolant and gas at a pressure at
least as low as the pressure of the liquid coolant in the heat
exchanger(s); a vacuum pump in vacuum communication with the gas in
the chamber; a fluid pump with a fluid intake port in fluid
communication with the liquid coolant in the chamber, and a fluid
output port in fluid communication with liquid coolant in an
evaporative cooler operating at substantially atmospheric pressure
and located at least partially outside the building; the
evaporative cooler being in fluid communication with the liquid
input port of the heat exchanger(s); wherein the fluid pump in
combination with the vacuum pump causes the liquid coolant to flow
from the chamber through the evaporative cooler and the heat
exchanger(s) and back to the chamber. Alternatively, the optional
evaporative cooler or other external cooling means can be in a
separate loop, not in fluid communication with the
electronics-mounted heat exchanger system, which may transfer heat
to the external cooling loop via an additional water-to-water
(liquid-to-liquid) or other heat exchanger.
[0008] Also provided in various embodiments is a system for cooling
at least one electrical device inside a building, comprising: one
or more liquid coolant-containing heat exchangers thermally coupled
to a first electrical device and having a liquid input port and a
liquid output port and containing liquid coolant at below
atmospheric pressure; a system of first and second chambers
comprising: a first liquid coolant-containing chamber in one-way
fluid communication with the liquid output port of the heat
exchanger, the first chamber containing liquid coolant and gas; a
second liquid coolant-containing chamber in one-way fluid
communication with the liquid output port of the heat exchanger,
the second chamber containing liquid coolant and gas; a vacuum pump
in switchable vacuum communication with the gas in the first and
second chambers; a source of higher pressure air in switchable
pressure communication with the gas in the first and second
chambers; the liquid coolant in the first and second chambers in
one-way fluid communication with liquid coolant in an evaporative
cooler operating at substantially atmospheric pressure and located
at least partially outside the building; the evaporative cooler in
fluid communication with the liquid input port of the heat
exchanger; wherein the vacuum pump and the higher pressure air
source coordinates with the system to serially pressurize and
depressurize the first and second chambers and thereby cause the
liquid coolant to flow substantially steadily from heat exchanger
through the system of first and second chambers to the evaporative
cooler and back to the heat exchanger. Once again, the optional
evaporative cooler or other external cooling means can be in a
separate loop not in fluid communication with the
electronics-mounted heat exchanger system, which may transfer heat
to the external cooling loop via an additional water-to-water or
other heat exchanger.
[0009] In any of the systems the liquid coolant-containing heat
exchangers may comprise one or more turbulators, and may also be
thermally coupled to the atmosphere adjacent the electrical device,
where a fan may urge circulation of the atmosphere adjacent to the
liquid coolant-containing heat exchangers. A vacuum accumulator may
be in fluid communication with and between the evaporative cooler
and the heat exchangers. The turbulator may be located in a heat
exchanger tube and configured to force the liquid coolant to flow
in a path having a length more than twice the largest dimension of
the heat exchanger tube, or may be configured to reduce the
cross-sectional area of the flow path of the liquid coolant to less
than 50% of the cross-sectional area of the heat exchanger tube.
The turbulator may define a conical helix flow path for the liquid
coolant, may direct a jet of liquid coolant against a surface
proximate to one of the electrical devices, may define a
rectangular cross-section helical liquid coolant flow path, a round
cross-section helical liquid coolant flow path, a rectangular
cross-section single-entry helical liquid coolant flow path, a
rectangular cross-section double-entry helical liquid coolant flow
path, a round cross-section single-entry helical liquid coolant
flow path, a round cross-section single-entry helical liquid
coolant flow path, or a round cross-section double-entry helical
liquid coolant flow path. It may also define a helical path in
which the direction of the helix reverses periodically, for example
from left-handed to right-handed. For purposes of this aspect of
the disclosure, a square cross-section is considered a special case
of a rectangular cross-section, i.e., one where the sides are the
same length. Systems are provided wherein a portion of the liquid
coolant flows axially over the outer surface of the turbulator,
thereby causing swirl and turbulence in the flow path and
increasing the heat transfer effectiveness of the turbulator.
[0010] Also provided are systems comprising: a connector releasably
connecting the liquid coolant-containing heat exchanger to the
chamber, the connector adapted to release the liquid
coolant-containing heat exchanger from the chamber only when
substantially all of the liquid coolant has been evacuated out of
the heat exchanger. For example, provided are: a supply valve in
removable fluid communication with the liquid input port of the
heat exchanger; a return valve in removable fluid communication
with the liquid output port of the heat exchanger; wherein the
supply valve is actuatable to open the liquid input port of the
heat exchanger to atmospheric pressure air that is at a higher
pressure than the water inside the heat exchanger and thereby to
evacuate the water from inside the heat exchanger; the supply valve
and return valve are constructed to close and disconnect the heat
exchanger from the system after the water is evacuated from inside
the heat exchanger. A passive latching system is also provided. The
latching system may include a mechanical delay in order to prevent
premature disconnection.
[0011] Provided in various systems is a liquid level sensor located
in the chamber and providing an output based on the level of the
liquid in the chamber, the fluid pump being adapted to operate in
response to the output of the fluid level sensor. In other
embodiments, provided are fluid level sensors located in both the
first and second chambers and providing first and second outputs,
respectively, based on the respective levels of the liquid in the
chambers, the vacuum pump and the pressure pump each being adapted
to operate in response to one or both of the first and second
outputs and to maintain the fluid levels in the chambers within
predetermined ranges.
[0012] Systems may further comprise a vacuum regulator in vacuum
communication with the vacuum pump and adapted to maintain a
pressure in at least a portion of the system less than atmospheric
pressure. Also provided may be a filter in fluid communication with
the liquid coolant-containing heat exchanger and adapted to prevent
debris from entering the liquid coolant-containing heat exchanger
or valves. Additionally provided is a pressure regulator in fluid
communication with the liquid coolant-containing heat exchanger,
the pressure regulator adapted to provide a constant pressure
differential across the liquid coolant-containing heat exchanger. A
dome-loaded, spring-biased regulator, as is known in the art, may
accomplish this.
[0013] A method is provided of modifying a non-liquid-cooled
electrical device heat exchanger with fins extending from a base to
become liquid-cooled, comprising the steps of: removing at least a
portion of one or more of the fins and thereby making accessible a
portion of the base; and affixing liquid cooling tubing having an
input port and an output port to at least a portion of the exposed
base.
[0014] A method is also provided of disconnecting a heat exchanger
from a system for cooling at least one electrical device, as
described herein, where the method comprises: providing such a
system, and actuating the supply valve and opening the liquid input
port of the heat exchanger to atmospheric pressure air that is at a
higher pressure than the cooling liquid inside the heat exchanger;
evacuating the cooling liquid from inside the heat exchanger;
closing the supply valve and the return valve; and disconnecting
the heat exchanger from the system after the cooling liquid is
evacuated from inside the heat exchanger. This method may also
apply to systems with a plurality of heat exchangers.
[0015] Further provided is a method of minimizing the energy needed
to cool heat-generating electronics inside a cabinet having a
higher than ambient temperature, comprising the steps of: providing
a heat exchanger comprising: a thermally conductive base adapted to
thermally couple to the heat-generating electronics; a plurality of
thermally conductive fins extending outward from the base; and one
or more cooling liquid pathways thermally coupled to the base and
the fins; balancing the thermal load of the heat generating
electronics and the ambient air inside the cabinet by positioning
the one or more cooling liquid pathways relative to the base and
the fins; thermally coupling the heat exchanger to the heat
generating electronics; and providing a source of cooling liquid to
the one or more cooling liquid pathways. This method may also
comprise the steps of: providing a fan and locating the fan so that
it causes air to flow across one or more of the fins; and balancing
the thermal load of the heat generating electronics and the ambient
air inside the cabinet by: positioning the one or more cooling
liquid pathways relative to the base and the fins in further view
of the heat transfer effect of the fan; and adjusting the speed of
the fan.
[0016] Also provided is a system that uses one vacuum pump to
circulate coolant under negative pressure. The system includes a
pump connected to a vacuum line such that the pump creates a
pressure of less than atmospheric on the vacuum line. The vacuum
line, along with a pressurized line, is connected to a valve
assembly, and that assembly is connected to a first and second
fluid chamber. A coolant circuit is provided that allows coolant to
circulate through the first and second chambers, through a primary
heat exchanger and through an electrical device heat exchanger. The
circulation is accomplished through a controller that operates the
valve assembly. The circuit may also have a reservoir, various
pressure and temperature sensors, and other valves and nozzles to
optimize the system. The controller operates the valve assembly by
substantially alternating between (a) actuating the valve assembly
to create a higher pressure in the first coolant chamber relative
to the second coolant chamber, thus emptying coolant from the first
coolant chamber and drawing coolant into the second coolant
chamber; and (b) actuating the valve assembly to create a higher
pressure in the second coolant chamber relative to the first
coolant chamber, thus emptying coolant from the second coolant
chamber and drawing coolant into the first coolant chamber. The
system may also optionally have a coolant recovery device so as to
minimize the maintenance of the system.
[0017] Other aspects of the invention are disclosed herein as
discussed in the following Drawings and Detailed Description.
5.0 BRIEF DESCRIPTION OF THE DRAWINGS
[0018] The invention can be better understood with reference to the
following figures. The components within the figures are not
necessarily to scale, emphasis instead being placed on clearly
illustrating example aspects of the invention. In the figures, like
reference numerals designate corresponding parts throughout the
different views and/or embodiments. It will be understood that
certain components and details may not appear in the figures to
assist in more clearly describing the invention.
[0019] FIG. 1 is a diagram of a vacuum-pumped liquid cooling system
according to various example embodiments.
[0020] FIG. 2 is a top plan view of an example air and cooling
liquid cooled heat exchanger incorporating a turbulator.
[0021] FIG. 3A is a partial section view of the example air and
cooling liquid cooled heat exchanger with a turbulator of FIG.
2.
[0022] FIG. 3B is a partial section view of the example air and
cooling liquid cooled heat exchanger with a turbulator of FIG.
2.
[0023] FIG. 4 is a perspective view of the example turbulator of
FIGS. 2, 3A and 3B.
[0024] FIG. 5 is a diagram showing an example cooling liquid
clearing disconnect system in normal operation.
[0025] FIG. 6 is a diagram showing the example cooling liquid
clearing disconnect system of FIG. 5 during the disconnect
process.
[0026] FIG. 7 is a diagram showing the example cooling liquid
clearing disconnect system of FIG. 5 in a disconnected state.
[0027] FIG. 8 is a diagram of a vacuum-pumped liquid cooling system
according to various example embodiments.
[0028] FIG. 8A is a diagram of a vacuum-pumped liquid cooling
system according to various example embodiments.
[0029] FIG. 8B is a diagram of a vacuum-pumped liquid cooling
system according to various example embodiments.
[0030] FIG. 9A is a sectional view of a vacuum accumulator used to
prevent drops of cooling liquid from leaving the system when it is
disconnected, shown in a low-vacuum condition.
[0031] FIG. 9B is a sectional view of the vacuum accumulator of
FIG. 9A, shown in a high-vacuum condition.
[0032] FIG. 10 is a perspective view of a turbulator assembly
comprising a single-entry flow passage turbulator having a
rectangular cross-section and positioned inside a flow channel,
partially cut-away.
[0033] FIG. 10A is a top plan view of the turbulator of FIG.
10.
[0034] FIG. 10B is a side elevation view of the turbulator of FIG.
10.
[0035] FIG. 10C is a top plan view of a turbulator with a
rectangular cross-section and a double-entry flow passage.
[0036] FIG. 10D is a top plan view of a turbulator with a circular
cross-section and a single-entry flow passage.
[0037] FIG. 10E is a perspective view of a turbulator with a
circular cross-section and a double-entry flow passage.
[0038] FIG. 10F illustrates cross-sectional views of a
turbulator.
[0039] FIG. 10G illustrates cross-sectional views of a
turbulator
[0040] FIG. 10H illustrates a turbulator traveling through a heat
exchanger.
[0041] FIG. 10I illustrates a turbulator traveling through a heat
exchanger.
[0042] FIG. 11 is a perspective exploded view of an example air and
cooling liquid cooled heat exchanger with turbulators positioned
near the primary heat source.
[0043] FIG. 12 is a perspective view of an example air heat
exchanger retrofitted to become an air and cooling liquid cooled
heat exchanger.
[0044] FIG. 13 is a side elevation view of an example air and
cooling liquid cooled heat exchanger with a turbulator positioned
further from the primary heat source.
[0045] FIG. 14 is a perspective view of an example air and cooling
liquid cooled heat exchanger with turbulators positioned in the
fins.
[0046] FIG. 15 is a diagram showing heat flow relationships in an
example server environment that uses a liquid and air cooled heat
exchanger.
[0047] FIG. 16 is a sectional view of a side elevation of a valve
according to various example embodiments.
[0048] FIG. 17A is a heat flow diagram depicting the heat flow in
an example system using only air cooling.
[0049] FIG. 17B is a heat flow diagram depicting the heat flow in
an example system using liquid cooling and air cooling.
[0050] FIG. 18 is a diagram of a vacuum-pumped liquid cooling
system according to various example embodiments.
[0051] FIG. 19A is a sectional view of a side elevation of an
example valve in an example cooling liquid clearing disconnect
system in normal operation.
[0052] FIG. 19B is a sectional view of a side elevation of an
example valve in the example cooling liquid clearing disconnect
system of FIG. 19A during the disconnect process.
[0053] FIG. 19C is a sectional view of a side elevation of an
example valve in the example cooling liquid clearing disconnect
system of FIG. 19A in a disconnected state.
[0054] FIG. 20 is a chart of temperature data resulting from tests
of example computer cooling systems according to various example
embodiments.
[0055] FIG. 21 is a chart of power consumption data resulting from
tests of example computer cooling systems according to various
example embodiments.
[0056] FIG. 22 is a schematic of a single vacuum pump cooling
system according to an example embodiment.
[0057] FIG. 23A is a top view of a of a single vacuum pump cooling
system according to an example embodiment.
[0058] FIG. 23B is an isometric view of a of a single vacuum pump
cooling system according to an example embodiment.
[0059] FIG. 24 is a schematic and top view of a single vacuum pump
cooling system according to an example embodiment when the main
chamber is emptying.
[0060] FIG. 25 is a schematic and top view of a single vacuum pump
cooling system according to an example embodiment when the
auxiliary chamber is emptying.
[0061] FIG. 26 is a schematic of a single vacuum pump and coolant
recovery system according to an example embodiment.
[0062] FIG. 27A is a schematic of a valve that may be used on the
coolant supply side according to an example embodiment.
[0063] FIG. 27B is a schematic of a valve that may be used on the
coolant return side according to an example embodiment.
6.0 DETAILED DESCRIPTION OF EXAMPLE EMBODIMENTS
[0064] Following is a non-limiting written description of example
embodiments illustrating various aspects of the invention. These
examples are provided to enable a person of ordinary skill in the
art to practice the full scope of the invention without having to
engage in an undue amount of experimentation. As will be apparent
to persons skilled in the art, further modifications and
adaptations can be made without departing from the spirit and scope
of the invention, which is limited only by the claims.
6.1 Example Negative Pressure System Designs
[0065] Referring to the example liquid cooling system 100 shown in
FIG. 1, the system 100 provides liquid cooling under negative
pressure for an array of computers or other heat generating devices
with liquid heat exchangers 1 with a minimal flow rate and a
minimal volume of cooling liquid in order to provide cooling in an
efficient and reliable manner. The system 100 may be the same as
that disclosed in U.S. Pat. Pub. No. 2011/0253347 A1 to Harrington,
published Oct. 20, 2011, the full disclosure of which is
incorporated herein by reference. In certain embodiments, the
system 100 includes a cooling tower 11, which may be outdoors, to
cool the cooling liquid 12, a cooling liquid distribution system 4,
5 to supply cooling liquid to multiple CPUs, high performance heat
exchangers 1 to remove heat from said CPUs with a minimum flow rate
and pressure drop, a vacuum pump 8 to suck cooling liquid 12
through said CPUs heat exchangers and to remove any excess air that
may enter the system 100. Water can be used for the cooling liquid
12 due to its low viscosity and high heat capacity. Alternatively,
perfluorocarbons, avionics cooling liquids or any other suitable
fluids may be used. In addition the system 100 may include an
air-cooled heat exchanger (see, e.g., air-cooled heat exchanger 21
in FIG. 2) attached to each CPU to remove the heat in the event
that the liquid cooling system 100 is not operating. The fan that
is typically connected with a CPU heat exchanger (not shown) may
also be used to cool the interior of the computer by transferring
heat from the air inside the computer to the cooling liquid so that
other components within the server enclosure may be cooled with or
without the use of external air flow. An air-to-liquid heat
exchanger may also be used to remove any excess heat from the
portions of the server not cooled by the liquid cooled heat
exchanger.
[0066] In the example embodiment shown in FIG. 1, a supply of
cooling liquid 12 is maintained at a low temperature by the
evaporation of the cooling liquid as it flows out of nozzle 13. The
humid air flows out due to fan 14 in cooling tower 11. Due to the
low pressure in the chamber 6, the cooling liquid flows through a
filter 9, and through a check valve 18 and a supply pipe 5, through
a pressure regulator 3, through another check valve 16 with a
cracking pressure of approximately 1 in Hg, through a vacuum
accumulator 17 and then through a fluid connector 2, to the
computer, server, or server rack with internal heat exchanger 1.
The cooling liquid 12 then receives heat from the internal
electronic components in the computer, such as the CPU, and flows
out through the connector to an extraction pipe 4 and then to the
chamber 6. A vacuum is maintained within the chamber 6 by the
vacuum pump 8. The vacuum pump 8 could be a piston type with a
Teflon or similar seal, which has a long lifetime, or it could be a
linear pump or a diaphragm pump or any other suitable pump. A
liquid ring pump is particularly suitable for this application, in
that it pumps moist air well. The vacuum pump 8 may be compatible
with the humidity and any chemical used to prevent corrosion or
biofilm growth. A float valve 51 may be used to keep cooling liquid
12 out of the inlet of the vacuum pump 8, as shown with respect to
vacuum pump 53 in FIG. 8. The vacuum pump 8 may be controlled by a
pressure sensor 15 to maintain an absolute pressure that is above
the vapor pressure of the cooling liquid 12 in its heated state, to
keep the cooling liquid 12 in its liquid phase. The chamber 6 may
include a level sensor 7 and regulator such that if a certain level
is exceeded, the liquid pump 10 speeds up, thereby pumping cooling
liquid 12 out of the chamber 6 and into the cooling tower 11. This
may provide a constant pressure differential to multiple heat
sources 1. The cooling tower 11 will require makeup cooling liquid
to replace cooling liquid that is evaporated, as is known in the
art of evaporative coolers generally. The optional cooling tower 11
to cool down the cooling liquid 12 may use convection and
evaporation in order to reduce the temperature of the cooling
liquid 12 to the local wet bulb temperature or whatever temperature
is required by the CPUs, which is typically less than 30 C.
[0067] The cooled cooling liquid 12 is preferably moved through the
heat exchanger 1 under a pressure that is less than the local
atmospheric pressure. In certain embodiments the entire system 100
runs at a low absolute pressure, so that any leaks are of air into
the system 100, rather than cooling liquid 12 out of the system
100. One potential issue with cooling liquid-cooled negative
pressure systems is that at low absolute pressures, cooling liquid
may boil. For example, at 50 C, water boils at 4 in Hg absolute, so
the pressure in water-based systems cannot get that low.
Accordingly, this limits the potential pressure drop available to
each heat exchanger 1 to the difference between the vapor pressure
of the warmest cooling liquid 12 within the system 100 and the
local absolute atmospheric pressures. Maximum pressure drops
available for each heat exchanger 1 are thus substantially less
than one atmosphere. The remainder of the available pressure drop
must be used for plumbing to and from the heat exchangers 1 and the
pump 10, including head loss, elevation changes, and increases in
flow resistance due to fouling.
[0068] The plumbing 4, 5, etc. to and from the computer/server/CPU
heat exchangers 1 may be designed for unusually low pressure drop,
so as to keep the total pressure drop of the system 100 within the
aforesaid limits. This may be accomplished in certain embodiments
by using, for example, simple surgical tubing or similar light-duty
material with large-radius bends and low-pressure-drop fittings,
which would not work with conventional high-pressure systems.
Conventional high-pressure systems typically use heavier-duty
plumbing with sharp bends and large pressure-drop interfaces, which
combine to create systems having too much overall pressure drop to
work as described herein.
[0069] Alternatively, the plumbing 5, etc. to the
computer/server/CPU heat exchangers may be high pressure plumbing
supplied by an additional pump (not shown), with a pressure
regulator 3 to reduce the pressure to below atmospheric as the
cooling liquid 12 gets close to the electronics. For the return
plumbing 4, etc., larger pipes may be required for the flow of air
and cooling liquid, as air will be introduced to the system as
computers/servers are removed or replaced. Local air removal
systems (not shown) may be used in order to prevent the return
plumbing 4, etc., from getting too large. Such systems may use
local vacuum pumps, plumbing to a central vacuum pump, or float
actuated drain valves and multiple compartments, as in U.S. Pat.
No. 4,967,832 to Porter, published November 6, 1990, the full
disclosure of which is incorporated herein by reference.
[0070] Each server or computer with a liquid heat exchanger 1 may
have an inlet pressure regulator 3 and an outlet pressure regulator
(not shown) in order to maintain a desired pressure drop across the
CPU heat exchanger 1. Each CPU may have a temperature sensor (not
shown), and an increase in temperature over the inlet cooling
liquid temperature may indicate a problem with the heat exchanger
1. A temperature sensor, such as a thermistor, may be used to
measure the inlet cooling liquid temperature. Flow meters, such as
a rotameter, thermal mass flow sensor or turbine meter with a
digital readout (not shown), may also be used to monitor the flow.
The filter 9 may be used after the cooling tower 11 and before the
heat exchanger 1 to prevent clogging of the passages in the heat
exchanger 1. Chemical additives may be used to prevent fouling of
the heat exchanger 1 with biological films and to prevent
corrosion. The internal passages of the heat exchanger 1 may be
plated or anodized to prevent corrosion.
[0071] The cooling liquid chamber 6 is preferably at a lower
pressure than that of the heat exchanger on the device being cooled
1. This can be accomplished by keeping the chamber 6 at a lower
elevation than the heat exchanger 1 or by means of a check valve
with a given cracking pressure or a pressure regulator (see, e.g.,
check valves 38 and 49 in FIG. 8). This will provide negative
pressure at the CPU heat exchanger 1 by means of a gravity head.
Example cooling liquid distribution systems 100 may provide the
cooling liquid 12 at a pressure of approximately -2 in Hg to the
computer/server/CPU heat exchangers 1. This may be accomplished by
means of the design of the system 100, or by placing a
pressure-regulating valve 3 at the server or rack level. The
plumbing from the fluid supply chamber 6 to the computer/server/CPU
heat exchangers 1 may require an additional pump (not shown) in the
feed line 5 if the computer/server/CPU heat exchanger 1 is at a
significantly higher elevation than the cooling tower 11, such as
if it is on a higher floor than the cooling tower 11. Such a supply
pump's speed may be controlled so that the pressure at the
computer/server/CPU heat exchangers 1 is at the correct value.
[0072] For the fluid pump 10, a seal-less centrifugal pump with a
magnetic drive may be used, as well as a solenoid pump with an
internal fluidic check valve, such as described in U.S. Pat. No.
1,329,559 to Tesla, published on Feb. 3, 1920, the full disclosure
of which is incorporated herein by reference. In addition, a system
may be required to prime the pump 10, as is known in the art of
pumps. For example, this may be accomplished by turning off the
liquid pump 10 and allowing fluid 12 to flow back through the pump
10. A flow actuated shuttle valve in the pump output (not shown)
may be at a default off position allowing the vacuum pump to suck
fluid into the chamber 6. Once the liquid pump 10 is primed and the
level sensor 7 is activated, the liquid pump 10 may then turn on
and pump the fluid out of the chamber 6 and into the cooling tower
11. A pump 10 with a low net positive suction head (NPSH) is
preferred, so that the cooling liquid does not cavitate at the
inlet of the pump 10. The fluid pumps 10 and vacuum pumps 8 for the
system 100 may be selected to be reliable and have a long life.
They also may provide a steady pressure on the suction side, and a
low pressure on the outlet, in order to deliver flow to the cooling
tower 11. One example design for maximum operational life would be
to use a dual chamber pump such as described in, for instance, U.S.
Pat. No. 7,611,333 B1 to Harrington, published on Nov. 3, 2009, the
full disclosure of which is incorporated herein by reference, due
to the very low NPSH required and due to its ability to reject
bubbles from the inlet flow. Such a pump, when driven by a vacuum
pump and an air compressor, may provide a very low inlet pressure
and an independent output pressure. This type of pump may be fitted
with additional backup vacuum pumps and compressors (not shown)
connected with check valves so that any single point failure would
not cause a system-wide failure. In addition, the check valves and
pressurization and vacuum valves and controls may include redundant
units (not shown). A condenser and automatic drain system may be
required to capture any coolant vapor and droplets, which may be
pumped out by said vacuum pump.
[0073] Although a computer or server or server rack with a liquid
heat exchanger 1 is described, systems such as system 100 maybe
used to cool any electronic component. Although water is described
in various embodiments, any coolant 12 may be used instead of or in
addition to water. Although the system 100 is described as using
cooling liquid 12 for evaporation and for cooling, a
liquid-to-liquid heat exchanger may be used to transfer heat from
an evaporator 11 to a closed system (not shown) so that any coolant
12 may be used to interact with the hot components such as CPUs,
such as a non-corrosive or non-conductive coolant. This may be used
in the case of evaporative coolers 11 that use salt water or
reclaimed water, for example. In this way, the coolant used for the
computer heat exchangers 1 may be separate from the cooling used
for other systems. Then the heat can be transferred from one system
to another using, for instance, a plate type heat exchanger in a
separate cooling loop. For low temperature operation, as in
Northern latitudes, a radiator (not shown), fan 14 and glycol
system may be used to reject the heat while preventing freezing of
the coolant 12. A mister system can evaporatively pre-cool the air
going into the radiators (dry coolers) for use during occasional
hot days. Since CPUs can get up to 60 C, cooling liquid 12 can be
heated to 50 C and still be used to cool the CPUs. The cooling
liquid used for cooling the computers may be kept at a temperature
higher than the dew point of the air in the data center to prevent
condensation on the plumbing or the heat exchangers.
[0074] Referring now to the example liquid cooling system 800 shown
in FIG. 8A, the centrifugal pump 8 and chamber 6 of FIG. 1 has been
replaced by a multiple chamber pump which acts as a vacuum pump,
chamber, cooling liquid/air separator and pressure pump. In example
liquid cooling system 800, the system may use a plurality of
chambers, such as a main chamber 6 and an auxiliary chamber 56. The
operation of example system 800 is as follows: the cooling liquid
12 flows under suction in to the chamber 6 from the extraction pipe
4 through check valve 49. The pressure in the chamber 6 is
maintained at a low level by vacuum pump 8, which is connected to
the chamber by valve 44. A vacuum chamber, 55 may be used to
provide a steadier suction. A vacuum chamber 55 may likewise be
located at each server rack 1, and it may have a float-actuated
water release to allow for the release of any accumulation of
water. Such local air release systems may require local vacuum
pumps 8 or connection to a central vacuum system (not shown).
[0075] The cooling liquid flows into the chamber 6 until the level
sensor 41 indicates that the chamber 6 is nearly full. Then the
valve 34 opens, connecting the vacuum pump 8 with the auxiliary
chamber 56 and lowering the pressure of auxiliary chamber 56 so
that cooling liquid may flow into it from the extraction pipe 4
through check valve 38. Once the flow of cooling liquid is
established into both chambers 6 and 56, valve 44 shuts and valve
43 opens, connecting chamber 6 with the pressure pump 53 and
thereby pressurizing the main chamber 6 so that cooling liquid
flows through check valve 48 and into the cooling tower 11. Then
the level in chamber 6 reaches a low level, as indicated from level
senor 42, at which time the valve 43 shuts. Then the valve 44
opens, and flow is again established under suction into the main
chamber 6, at which time the auxiliary chamber vacuum valve 34 is
shut and the valve 33 is opened connecting chamber 56 with the
pressure pump 53 and forcing cooling liquid out of chamber 56
through check valve 39 until the level in the chamber 56 reaches
the low-level sensor 32. Under normal operation, the level sensor
31 would not be activated because the system is designed so that
the flow out of the chambers 6, 56 is higher than the flow into the
chambers 6, 56, so that the auxiliary chamber 56 is never
completely full, thereby allowing for the flow through the heat
exchangers 1 to be steady while the flow to the cooling tower 11 is
intermittent. Accordingly, the level sensor 31 can be used to
indicate if there is a system failure. The pressure and vacuum
levels can be monitored by the pressure pump 53 and the vacuum pump
8 using the pressure sensors 54 and 15. The entire system can be
controlled by a computer or by a logic circuit or any other
suitable means. Floats 51 may be used to sense the levels in the
chambers 6, 56 and reduce evaporation of the cooling liquid 12 in
the chambers 6, 56.
[0076] Referring to FIG. 8B, the system 800' may be substantially
the same as the system 800 in FIG. 8A, except the system 800' may
further include a test valve 61 and a purge valve 62 and a pressure
sensor (not shown). Test valve 61 and purge valve 62 and the
pressure sensor may be used to test the system 800' for leaks and
to purge air out of the system 800'. Temperature sensors (not
shown) may also be added to the plumbing at locations 4 and 5 to
provide data for determining the flow rate of heat removed by the
system 800'. The duration of time that the vacuum pump 8 is on can
be used to determine the rate of air flow in the system 800' and
thereby the presence of an air leak can be inferred, for instance
when the vacuum pump 8 runs longer or more often than normal. The
pressure at the vacuum pump inlet may also be used to determine the
amount of air flowing through the system, or an air flow sensor may
be used. The operator can be alerted if excessive air is finding
its way into the system. The entire pumping and monitoring system
800' can output data in real time to populate a web page or other
output (not shown) that displays various parameters regarding the
system in real time, such as, for example, heat pumped, air leak
rate, coolant resistivity, pH or TDS, and the like. The pressure
and level in the pump chambers can also be reported. The current to
the air pump 53 and vacuum pump 8 can be measured and monitored to
determine if either one is malfunctioning or wearing out. The
plumbing 4, 5 from the pump system to the racks of
computers/servers with liquid heat exchangers 1 can be connected
with quick connect fittings such as, for instance, those available
from the John Guest Corp., so that racks and servers 1 may be
easily reconfigured.
[0077] Referring to FIG. 18, a system 1800 may be provided
incorporating any or all of the features from systems 100, 800, or
800', except system 1800 demonstrates the option of using a closed
and/or sealed liquid pumping system 800'' to re-circulate liquid
through liquid-cooled computers/servers/server racks 1 without
exposing that liquid to the open atmosphere (and resulting
contaminants) of an external cooling source such as a cooling tower
11. This may be accomplished by, for instance, providing a
liquid-to-liquid heat exchanger 1890 that transfers heat from the
liquid used in the liquid-cooled computers/servers/server racks 1
to a separate liquid 12 that is cooled externally, for instance by
a cooling tower 11, as shown in FIG. 18. In the example embodiment
shown in FIG. 18, cooled liquid 12 pumped from the cooling tower 11
enters the exchanger 1890 at a first cooled position 1891, and
travels through the exchanger 1890 while picking up heat from the
hot liquid leaving the computers 1 until that now-heated liquid 12
exits the exchanger 1890 at a second heated position 1892, after
which it returns to the cooling tower 11 to be cooled. At the same
time, separate heated liquid leaving the computers 1 enters the
exchanger 1890 at a first heated position 1893, and travels through
the exchanger 1890 while dissipating, losing, or otherwise
transferring heat to the cool liquid from the cooling tower 11
until that now-cooled liquid exits the exchanger 1890 at a second
cooled position 1894, after which it returns to the pumping system
800'', having never mixed with the liquid 12 that flows through
cooling tower 11. Systems such as system 1800 may advantageously
use a clean, controlled liquid to circulate through the computers
1, while using a less expensive liquid such as gray water or sea
water in the cooling tower, which needs to be supplemented
regularly to make up for evaporation losses.
[0078] Also shown in system 1800 is a flow sensor 1830. The flow
sensor 1830 may include a self-heated thermistor or RTD, such that
if the liquid coolant stops flowing, or the coolant is too hot, the
fan 1840 is turned on to high speed. This could be accomplished by
flowing a known current through a thermistor such that in still
coolant, and under 25 C ambient conditions, the thermistor
temperature rises to 35 C. A comparator circuit could detect the
voltage decrease associated with the temperature rise, and a MOSFET
could be switched on to control the speed of the fan 1840. Under
air cooling conditions, the power to fan 1840 would typically be on
all the time, but under liquid-cooled conditions, the power to the
fan 1840 could be pulse width modulated at 10-500 Hz to slow down
the fan 1840 but not allow it to stop. The controller for the fan
1840 is represented by unit 1850. These features are applicable to
any of the present systems.
[0079] Referring to FIGS. 9A and 9B, an example vacuum accumulator
17 is shown in cross-section, having a liquid inlet 61 and liquid
outlet 63. The vacuum accumulator 17 comprises a flexible diaphragm
62 which may be flat or nearly flat in state 900 when no pressure
differential exists between inside and outside the accumulator 17,
as in FIG. 9A. When a vacuum or pressure less than the external
atmosphere is provided by the system inside accumulator 17, as in
state 900', the flexible diaphragm 62 is displaced inwardly toward
the liquid and holds a steady position, as shown in FIG. 9B. If the
CPU heat exchanger 1 is disconnected from the rest of the system
100, 800, 800', 1800, etc., then the check valve 16 shuts and the
diaphragm 62 springs back into the flat position 900 as in FIG. 9A.
This tends to suck cooling liquid towards the accumulator 17 and
away from the fluid connector 2, prevent dripping of liquid out of
the systems 100, 800, 800', 1800, etc.
[0080] Any leakage in the system may be detected by monitoring the
cycle time of a pump 8 used to remove air from the systems 100,
800, 800', 1800, etc. If the pump 8 is cycling on too often, then a
leak is indicated. The leak may be discovered by pulling a vacuum
on each heat exchanger 1 and measuring the decrease in vacuum over
time. A simple hand-operated vacuum pump may be used for this type
of testing.
[0081] Systems 100, 800, 800', 1800, etc. may use a pump with a
chamber (not shown) to supply fluid to all the heat exchangers 1.
During a shutdown procedure, the pump may evacuate the system;
purge it with air and store the fluid until such time as the liquid
cooling system is reactivated. During a reactivation procedure, the
pump control system may apply a vacuum or a pressure to the system;
check to see if the fluid system loses vacuum or pressure and then
start pumping again, based on the rate of change of the system
pressure.
6.2 Example Dry-Disconnect Systems
[0082] FIG. 5 provides a diagram of an example coolant clearing
system in normal operation 500, depicting the cooling liquid
flowing through a supply valve 71 and then through a heat
exchanger, 21, and then out through a return valve 72, all at less
than atmospheric pressure. In this configuration the valves 71, 72
are both open to flow of cooling liquid and are sealed from the
higher-pressure outside air.
[0083] FIG. 6 shows a diagram of the cooling liquid clearing system
of FIG. 5 during the disconnect process 600. Before disconnecting
the fluid supply and extraction lines (not shown), the valve 71 is
opened to outside air, which allows higher-pressure outside air to
flow into the valve 71 and into heat exchanger 21, shown
schematically. The valve 71 may be connected to a latch (not shown)
that prevents the fluid lines from being removed until the valve 71
is depressed or otherwise actuated to allow entry of air. The latch
can be configured to remain in a latched position, so valve 71
remains actuated to allow entry of air until the connector (not
shown) is reinserted into the computer.
[0084] FIG. 7 shows a diagram of the cooling liquid clearing system
of FIG. 5 upon completion of the disconnect process 700, when the
heat exchanger 21 is disconnected from the liquid cooling system
100. Upon completion of the disconnect process 700, the supply
valve 71 is unactuated to seal the valve 71 from outside air so
that air does not flow into the cooling system 100. A return valve
72 is likewise unactuated to seal the valve 72 from outside air so
that air does not flow into the cooling system 100. Return valve 72
may be unactuated by a pin or latch (not shown) so that it shuts
off when the heat exchanger 21 is disconnected from the liquid
cooling system 100. The connector may be designed to prevent the
disconnection of the heat exchanger 21 from the liquid cooling
system 100 until all the liquid is removed from the heat exchanger
21. Such a disconnection prevention feature could be activated by
the change in sub-atmospheric pressure present in the suction in
the return line as the return line changes from being filled with
cooling liquid to being filled with air. For example, the pressure
drop across the heat exchanger 21 would be less, as the heat
exchanger 21 changes from being filled with cooling liquid to being
filled with air. This change in pressure drop could be calibrated
to trigger the connector to allow disconnection of the heat
exchanger 21 from the liquid cooling system 100 when the heat
exchanger 21 changes from being filled with cooling liquid to being
filled with air. This draining process may be helped by the
following connector arrangement. To detach the connector in one
embodiment, the operator depresses a button (not shown) that
operates a three-way valve 71 that cuts off inlet cooling liquid
flow and vents to allow air into the system 100. Negative pressure
on the return side of the connector holds the connector in until
air reaches the outlet. At this point, the negative pressure in the
system is diminished due to the much lower delta pressure of air
flowing through the heat exchanger and then the connector may be
easily removed. Removal of the connector seals the outlet so that
air does not continue to flow into the cooling system return flow
path. The button stays depressed, thereby sealing off the inlet. To
attach the connector, the operator would insert the coupling, which
would connect the return path, and the button would automatically
release, which would allow the supply flow to reach the components
1. This system may also be actuated with a twist instead of a
button push, or by any other means of activation. Example
connectors adaptable for use with the present system are described
in U.S. Pat. No. 7,602,609 B2 to Spearing et al., published as
application US 2008/0298019 A1 on Dec. 4, 2008, the full disclosure
of which is incorporated herein by reference. The connector may
utilize a sacrificial metal, such as zinc or utilize electrical
potential to prevent corrosion inside the CPU heat exchanger 1.
Using tap water that has a slight alkaline content for the cooling
liquid 12 may reduce the corrosion rate for copper and brass heat
exchangers 1.
[0085] For example, the computers/servers with liquid heat
exchangers 1 may be connected to the pumping system using a
connector 1600 such as that shown in FIG. 16, which prevents the
user from disconnecting the server until the server is purged of
cooling liquid. This connector 1600 may be used in conjunction with
vacuum pumping systems 100, 800, 800', etc. The connector design
1600 in FIG. 16 achieves this in a two-step process. First, the
user or another mechanism depresses the button 1610 which closes
off the supply line 1620 to the server 1 and allows air to flow in
through port 1640 into the system 100, 800, 800', etc. At that
point, the top spool valve 1650 will have moved downward (toward
the bottom of the page in FIG. 16), but the bottom spool valve 1660
will not have moved yet, because its movement will be resisted by a
hydraulic lock created by liquid still present in the bottom
chamber 1670 below the spool valve 1660, which liquid will take a
short period of time to be sucked out. The leak rate from bottom
chamber 1670 is selected such that the second spool valve 1660 does
not move until enough time has passed to ensure that the server 1
is purged of liquid 12. Thus, the spool valve is a mechanical
device that creates a delay in releasing the connection, during
which time the fluid can be evacuated avoiding a leak.
[0086] Then, once the fluid 12 is evacuated from the bottom chamber
1670 to a predetermined level, a larger leak opens up, the bottom
spool valve 1660 drops all the way to the bottom of bottom chamber
1670, and the valve 1600 is closed or sealed from both the supply
1620 and return 1630 lines. The valve 1600 may be latched in the
closed position until it is reconnected to a server 1, at which
point both spools 1650, 1660 rise and the supply and return lines
1620, 1630 flow freely and the bottom chamber 1670 is refilled. The
valve 160 may also be held in the intermediate position (i.e., with
top spool valve 1650 closed while bottom spool valve 1660 remains
open) by the negative pressure which will be present until the
server 1 is purged of liquid 12. For example, a spring-loaded
diaphragm or piston (not shown) could hold the valve in the
intermediate position until the negative pressure was reduced, as
it would be once the server 1 was completely vented of liquid. The
valve 1600 may also be triggered by pressure differences created
with an orifice or venturi, in which case differences would be
higher when flowing liquid than when flowing gas, as is known in
the art of fluid mechanics.
[0087] FIGS. 19A, 19B and 19C illustrate an example connector valve
1901 as discussed above, further comprising an example latching
system 1905, 1915. The valve 1901 is shown in operation in a
latched open position 1900, in a latched intermediate position
1900', and in a closed unlatched position 1900''. Such a connector
1901 will allow air to enter the computer/server with liquid
cooling 1910 through a port 1920 as the valve 1901 is pushed down
into the intermediate position 1900'. If the server 1910 has a
minimum volume of cooling liquid 12, the server 1910 may be purged
of cooling liquid 12 in less than one second while in the
intermediate position 1900'. Once the cooling liquid 12 is purged
from the server 1910, it is also purged from the bottom chamber
1940 below the lower valve 1930. Once the cooling liquid 12 is
purged from the bottom chamber 1940 below the lower valve 1930, the
lower valve 1930 moves to the bottom of the bottom chamber 1940 and
the valve 1901 moves to the closed position 1900'', thereby closing
the air port 1920 as well as the plumbing 4, 5 for the cooling
liquid 12. The movement of the lower valve 1930 to the bottom of
the lower chamber 1940 also moves downward a connected latching
mechanism 1905 that thereby disengages a corresponding latching
mechanism 1915 that is connected with the server 1910. The
disengagement of latching mechanisms 1905, 1915 allows the
connector valve 1901 and plumbing 4, 5 connected thereto to be
removed from the server 1910 without leakage of cooling liquid 12,
for instance if component repair or replacement is required. A
small amount of air may be pulled into the system during this
process, but it will be automatically evacuated and pumped out by
the vacuum pump(s), e.g., vacuum pump 8.
[0088] Each computer or server or server rack with a liquid heat
exchanger 1 may be connected with the present dry disconnect system
that allows for the automatic draining of the heat exchanger 1 as
described above. Such connectors may include supply and return
flows.
[0089] Supply and return flows may be coaxial, in order to allow
for a small interconnect. The system is preferably designed to
remove all of the cooling liquid from inside each heat exchanger
subsystem 1 such as a CPU, server or server rack during the
disconnection process. For example, if the heat exchanger 1
contains one cc of cooling liquid 12, and the flow rate is 150
cc/minute of cooling liquid, then it will take less than 1 second
to drain the cooling liquid out of the computer or server or server
rack with a liquid heat exchanger 1. As the cooling liquid 12 is
replaced by air, the flow resistance of the heat exchanger
decreases, so the process may happen in less than 0.5 seconds.
6.3 Example Turbulator Designs
[0090] Referring to FIG. 2, an example air and cooling liquid heat
exchanger 200 may comprise a cooling liquid cooling portion 210,
which includes inlet tube 22 and outlet tube 23 to provide cooling
liquid (not shown) to a turbulator 400 (shown in more detail in
FIGS. 3A, 3B and 4, its top surface 20 being visible in FIG. 2),
and a metal heat spreader 24 that is in thermal contact with the
electronic device 1820 (shown in FIG. 18) on one side and is in
thermal contact with the cooling liquid on the other side. A series
of fins 21 are provided in thermal contact with flowing air in the
event that the liquid cooling system is not operational. A fan 1840
(shown in FIG. 18) would typically be used in proximity to the fins
21 to provide cooling air. A turbulator 400 fits inside the metal
heat spreader 24 and reduces the amount of cooling liquid needed to
cool the device and increases the velocity and turbulence level in
the cooling liquid. In this example air and cooling liquid heat
exchanger 200, the cooling liquid inlet 22 may be adapted to
provide a point of jet impingement cooling closest to the heat
source, for instance near surface 20, as best seen in FIGS. 2, 3A,
3B and 4, to flow the cooling liquid in a helical path 25 through
the turbulator 400 to the outlet tube 23. In some cases a portion
of the cooling liquid flow may flow over the helical flow passages
25 through a clearance space between the turbulator 400 and the
metal heat spreader 24 as best shown in section view 300. This
"leakage" of cooling liquid flow over the edges of helical flow
passages 25 may enhance heat transfer by causing turbulence and
swirl within the helical flow passages 25.
[0091] FIG. 3A shows a partial cross-sectional side elevation view
300 of the air and liquid heat exchanger 200 shown from the top in
FIG. 2. The turbulator 400 can be seen installed in FIG. 3B in the
heat spreader 24, and providing a narrow helical path or passage 25
for the cooling liquid. The CPU is not shown in this view; it would
normally be attached to the bottom or lower portion of the heat
spreader 24 as shown in cross-sectional side elevation view 300. In
other embodiments the CPU or other heat source could be located
proximate to the upper portion of the heat spreader 24, for
instance near surface 20. FIG. 4 provides an isometric view of the
turbulator 400, which shows the helical flow path 25 more
clearly.
[0092] With reference to FIG. 11, a liquid-cooled heat exchanger
1100 is preferably mounted to a CPU (1820, shown in FIG. 18) and
may comprise one or more passages 1130 with turbulators 1001 to
increase the velocity and turbulence of the cooling liquid 12 near
the heat transfer surface 1111. The turbulator 1001 may also be
designed to minimize the volume of cooling liquid 12 contained
within the heat exchanger 1100 so that the cooling liquid 12 may be
quickly cleared for repairs. The CPU 1820 typically includes an
air-cooled heat exchanger with fins 1810 and a fan 1840 located
nearby to provide air-cooling. The fan 1840 may be controlled by
the temperature of the CPU 1820 so that as it gets hotter, the fan
speed increases. The flow rate of the cooling liquid 12 may be
determined by the acceptable temperature rise of the liquid and the
power dissipated by the CPU 1820. For an example CPU that generates
100 watts, a stream of cooling liquid at 150 cc/minute may result
in a temperature rise of approximately 10 C. The temperature
differential from the CPU case to the cooling liquid should be of
the same order as the temperature rise. The heat exchanger 1100 in
that example may be selected to have a pressure drop of
approximately 4 in Hg so that the system 100 will work properly on
a hot day in a high altitude location, where the difference between
the local atmospheric pressure and the vapor pressure of the hot
cooling liquid may be only about 8 in Hg.
[0093] The heat exchanger 1 may incorporate a helical flow pattern
for the cooling liquid 12 to put a long path into a short passage
to increase heat transfer. This helical flow passage may have
multiple starts and paths, as shown in FIG. 10E, so as to allow for
increased flow in a small passage. This may also be accomplished by
placing a threaded rod, such as shown in FIG. 10D, in a metal tube
so that the flow must take a long path through the heat exchanger
at a high velocity. This has the added benefit of reducing the
volume of cooling liquid in the heat exchanger 1, thereby reducing
the amount of cooling liquid 12 that needs to be cleared to service
the heat exchanger 1. Alternatively, a rod with a tortuous path in
relief may be used to displace fluid in the center part of the
passage and thereby increase the cooling liquid flow and
turbulence, as shown in FIG. 10.
[0094] The rod and cylinder may be square, cylindrical, conical,
triangular, hexagonal, or any other appropriate shape. The rod or
other turbulator structure may be designed so that some of the
cooling liquid 12 flows over the edge 1004 of flow passages 1005 in
an axial direction, for instance directly from a proximal end 1002
to a distal end 1003 of the turbulator 1001 shown in FIG. 10A. This
axial flow may interact with the helical flow in channels 1005,
1006 to provide swirl or turbulence in the heat transfer passages
in order to increase heat transfer. This is shown in FIG. 10G and
discussed further below. In addition, the axial flow will reduce
the flow resistance/pressure drop of the heat exchanger 1. This
arrangement may be particularly useful in situations where the flow
of cooling liquid 12 would otherwise be laminar or nearly laminar.
In some installations, a flat plate heat exchanger may be used. For
high power dissipation systems, or for additional reliability,
multiple parallel turbulators may be used.
[0095] For example, referring to the embodiment shown in FIGS. 10,
10A and 10B, a turbulator system 1000 may include a turbulator 1001
with ridges 1004 and troughs 1005 defining a flow passage between
the turbulator 1001 and the interior of a hollow body or tube 1010,
for instance a helical flow passage, that forces cooling liquid 12
flowing from a first end 1002 to a second end 1003 of the
turbulator 1001 to flow diagonally across one face 1020 of the
interior of the hollow body 1010, and then across to the other side
1030 of said passage, where the flow goes diagonally across and
then back to the previous side 1020, and then repeats this helical
flow pattern from a proximal end 1040 of the hollow body 1010 to a
distal end 1050 of the hollow body 1010. The flow passage may be
fed with a fitting 1060, which may include a hose barb. The flow
passage may likewise be drained with a fitting 1070, which may
include a hose barb.
[0096] Referring to the example embodiment shown in FIG. 10C, an
alternate turbulator 1001' forms a double-entry helical flow path.
This rectangular cross-section design allow for more flow at a
given pressure than the rectangular cross-section design in FIGS.
10, 10A and 10B, in that it defines two parallel flow paths. The
use of two paths, instead of one larger path, increases the
velocity of the fluid and tends to make the device resistant to
clogging. Also, the dual path reduces the tendency of the flow to
short circuit over the top of ridges 1004, thus maintaining the
flow in thermal contact with the heat exchange tube 1010 and
increasing cooling efficiency.
[0097] In the example embodiment shown in FIG. 10D, a circular
cross-section turbulator 1001'' is provided for use inside of a
corresponding circular cross-section tube (not shown). This may be
easily constructed in certain embodiments by placing a threaded rod
1001'' in a tube with a close tolerance. This type of design lends
itself to use in some of the embodiment described below, in which
the liquid flow path is embedded in the fins of a heat exchanger in
order to reduce the thermal resistance to the air.
[0098] FIG. 10E illustrates yet another type of turbulator 1001'''
with a circular cross-section. Like the rectangular cross-section
embodiment shown in FIG. 10c, the circular cross-section turbulator
1001''' in FIG. 10E forms a double-entry helical flow path. To
illustrate these paths, FIG. 10E has lighter shading 1005 that
illustrates one flow path, and darker shading 1006 illustrating the
independent second flow path. This design provides more flow at a
given pressure than the design in FIGS. 10, 10A and 10B, in that it
uses two parallel flow paths. The use of two paths, instead of one
larger path, increases the velocity of the fluid and tends to make
the device resistant to clogging. Also, the dual path and circular
cross-section reduces the tendency of the flow to short circuit
over the top of ridges 1004, thus maintaining the flow in thermal
contact with the heat exchange tube 1010 and increasing cooling
efficiency.
[0099] The design of turbulators shown in FIGS. 10A, 10B, 10C, and
10E all have a core that is concentric to the passageway in which
the turbulator is installed. Radiating away from the core are fins
or ridges that create the channels in which the coolant flows. FIG.
10F illustrates a cross-section that is perpendicular to the
longitudinal axis of the turbulator 1001'''', while FIG. 10G
illustrates a cross-section that is parallel to the longitudinal
axis of the turbulator 1001''''. The turbulator core is labeled
1070 and the fins/ridges 1072. The core effectively reduces the
cross-sectional area of the passageway and forces the coolant
through the turbulator at a higher pressure. While this obstruction
of the turbulator causes an increase in the pressure drop, it has
the benefit of causing the coolant to flow in a highly turbulent
fashion which increases the heat exchange with the coolant. The
cross-sectional area of the core relative to the passageway may be
greater than 20%, but more preferably at least 40 percent. Further,
the turbulator may be designed such that the fins/ridges
intentionally allow flow or leakage from one channel to an adjacent
channel. While at first blush this may seem to reduce efficiency,
it actually causes the coolant to experience even more turbulence
by creating swirls that are perpendicular to the flow in the
helical channel, which increases the heat transfer to the coolant.
FIG. 10H illustrates a turbulator 1001'''' traveling through a heat
sink 1074, with FIG. 10I showing an enlarged view of the turbulator
1001''''. The turbulator 1001'''' has helical channels 1075 (shaded
dark gray) and 1076 (shaded light gray) that are adjacent to each
other. Because ridge 1078 is designed to allow leakage, a swirl
1080 is created that is substantially perpendicular to the helical
channel flow shown by arrow 1082. The central core of the
turbulator may consist of baffles, which reduce the flow velocity
instead of solid material. This achieves the goal of increased heat
transfer, but it adds unnecessary fluid to the system.
6.4 Example Heat Sink Designs
[0100] Referring to FIGS. 11 and 18, in the embodiment shown in
FIG. 11 heat exchanger tubes 1010 are soldered into slots 1130 in
the base plate 1110 of the heat sink 1100, thereby reducing the
thermal resistance from the CPU 1820 (located adjacent surface
1111) to the liquid. The turbulators 1001 enhance the heat transfer
from the liquid 12 to the base of the heat sink 1111 and to the top
of the CPU 1820.
[0101] Referring to FIG. 12, a fluid supply fitting 1210, heat
exchanger tube 1010 and fluid return fitting 1230 are added to the
heat sink 1200 so that the cooling system can be connected to a
fluid cooling system without affecting the mechanical attachment of
the heat sink 1200 to the CPU 1820 or circuit board. The path of
the heat exchange tube(s) 1010 is shown by dashed line 1220. Heat
sink 1200 can be created from an existing non-liquid heat sink
without changing the footprint of the heat sink by removing a few
fins 120 and adding one or more heat exchanger tubes 1010 with
fluid connections 1210, 1230.
[0102] Referring to FIG. 13, in this example embodiment the heat
exchanger tube 1320 is positioned on the top surface 1330 of the
base plate 1110 of the heat sink 1300. In this configuration, the
thermal resistance from the CPU 1820 (located adjacent to surface
1111) to the liquid coolant 12 (running through tube 1310) is
greater than in the design shown in FIG. 11. At the same time, the
thermal resistance from the liquid to the air in heat sink 1300 is
reduced compared to heat sink 1100.
[0103] Turning to FIG. 14, in heat sink 1400 the heat exchanger
tubes 1420, 1440 are placed in the fins 1120, still further away
from the base plate 1110 than in heat sink 1300. This further
increases the thermal resistance from the CPU 1820 (located
adjacent to surface 1111) to the liquid coolant 12 (running through
tubes 1420, 1430, 1440 and 1450), and further reduces the thermal
resistance from the liquid coolant to the air. The various example
heat sink designs 1100, 1200, 1300 and 1400 demonstrate that the
distance from the bottom 1111 of the base plate 1110 to the tubes
1010, 1130, 1320, 1420, 1440 may be adjusted in order to adjust and
balance the thermal resistance from the liquid coolant 12 to the
air (through fins 1120) and from the CPU 1820 (adjacent surface
1111) to the liquid coolant 12.
6.5 Design Optimization
[0104] A thermodynamic model of these competing thermal resistances
is shown in FIG. 15. In model 1500, the relationship of the thermal
resistance from the CPU to the liquid and the air, and from the air
to the liquid, is illustrated. By means of the prior embodiments,
the thermal resistance from the heat sink 1810 to air, the CPU 1820
to the heat sink 1810, the heat sink 1810 to the liquid 12 and the
air to the liquid 12 may be adjusted and optimized to minimize
overall total power consumption, including that of the entire data
center. For example, increasing the number or area of the fins
1120, may decrease the thermal resistance from the heat sink 1810
to the air. The thermal resistance from the air to the liquid 12
may be decreased by placing the liquid heat exchanger tubes 1420,
1440 closer to the center of the fins 1120. For instance, an
example air cooled heat exchanger may have a thermal resistance of
0.15 C/watt. The liquid cooled heat exchanger may have a thermal
resistance of 0.05 C/watt. By adjusting the position of the cooling
liquid-cooled heat exchanger within the assembly the thermal
resistance from the air to the cooling liquid and the CPU 1820 may
be suitably controlled so as to provide optimal cooling for the air
in the data center and the CPU chip. In some cases, multiple
passages may be used to cool both the fins and the processor. Heat
pipes and any other thermal structures may also be used to control
the flow of heat in connection with the present systems, as will be
apparent to persons of skill in the art upon reviewing this
disclosure.
[0105] The fan 1840 that is typically connected to the CPU heat
exchanger may also be used to cool the interior of the computer by
transferring heat from the air inside the computer to the cooling
liquid 12 so that other components within the server enclosure may
be cooled with or without the use of external air flow--i.e., the
computer may be sealed. The speed of the fan 1840 may be adjusted
to remove additional heat from the air inside the server enclosure
of the data center as required to minimize the overall power
consumption of the data center. The overall power consumption
versus fan speed may be determined based on the power consumption
of the air conditioning system versus temperature in the data
center and the power consumption of the CPU 1820 versus its
temperature. The CPU 1820 uses additional power depending on the
temperature of the processor due to leakage currents, with the
leakage currents increasing exponentially with the processor at the
higher temperature range. For example, CMOS-based processors use
more energy as the temperature of the processor goes up, due to
leakage currents. Also, the air conditioning system of the data
center uses additional power depending on the temperature of the
data center and the building heat removal requirements. This
increase is generally linear; with higher temperatures requiring
proportionally higher air conditioning power. By controlling and
selecting the optimal speed of the CPU fan 1840, the flow rate of
liquid 12 through the heat exchanger 1, and the position of the
liquid heat exchanger tubes 1010, 1130, 1320, 1420, 1440 in the
overall assembly consisting of a base 1110 and fins 1120, the
overall power required for the data center can be decreased.
Examples of these relative flows of heat between the various
components are depicted by the wavy arrows in FIGS. 17A and 17B,
and may be analyzed and optimized using an electrical analog, as
shown in example heat flow diagram 1500 in FIG. 15.
[0106] With further reference to FIGS. 17A and 17B, it can be seen
that the air conditioning system of the data center would use
additional power when the computers/servers 1 use air cooling only,
rather than air cooling plus liquid coolant that is routed outdoors
to cool.
[0107] This heat load difference is represented by the comparison
of the larger heat load 1710 that must be removed from the air of
the data center in system 1700 with the smaller heat load 1720 that
must be removed from the air of the data center in system 1700'.
The difference is represented by heat load 1730 that is removed by
the liquid, and is preferably routed outdoors to cool as shown in
the foregoing embodiments. A heat sink 1810 for a system that uses
an internal liquid cooling passage could be designed to remove heat
from other items in the server, such as hard drives, memory chips,
and any other heat-generating electronics, as shown in FIGS. 17A
and 17B. In these designs, the heat sink 1810 may be selected to be
oversized for the CPU 1820, but this will reduce the cooling load
on the air conditioning system in the data center by transferring
to the liquid coolant not only heat from the CPU 1820, but also
heat from the other nearby heat-generating electronics.
6.6 Example Test Results
[0108] An example heat exchanger design started with an existing
air-cooled system. In order to provide the best cooling with
minimum volume and input power, a spiral cooling channel with a
Reynolds number just above the laminar limit was used. This is
believed to provide the best cooling with a reasonably sized
channel that can pass contamination.
[0109] For example, if a 140-watt CPU is to be cooled with water,
and an 18 degrees F. (10 degrees C.) temperature rise can be
accepted, then a flow rate of 220 cc/minute would be needed, based
on the heat capacity and mass flow rate of water. Next, rocket
science was employed to develop a nozzle cooling system, which in
rocket science is done with an array of tubes that cool the nozzle
and preheat the fuel on the way to the combustion chamber. The goal
there is to adjust the length and diameter of the parallel tube
array to get the optimum cooling for a given flow rate. In the
present case, the water outlet temperature and the heat sink
temperature are desired to be within 1 degrees C. of each other. So
a fluid path was selected with a Reynolds number slightly higher
than 2100, so that the flow was turbulent, but the pressure drop
was not too high. In this example two helical flow passages were
used, 0.055 inch (1.4 mm) in diameter. This system was analyzed
using empirical heat transfer equations for flow in a tube, modeled
using computation flow dynamics (CFD), and tested with a Xeon
processor running stress software. The thermal resistance heat sink
to water, based on the temperature of the water into the heat sink,
was 0.04 watt/degrees C. with 230 cc/minute flow rate per CPU. A
similar heat sink design with coolant passages in the base is shown
in FIG. 11.
[0110] The test heat sink worked exactly as modeled, but when the
flow was increased, it was discovered that it could actually remove
heat from the entire system. A stack of three DL380 servers was run
at idle power levels in an insulated box, and the heat sinks were
able to remove all the heat (700 watts) from the computers. In this
case the ambient air was 107 degrees F. (42 degrees C.), and the
coolant inlet was at 76 degrees F. (22 degrees C.).
[0111] Additionally, a test was conducted with a 2 kW rack of
servers in an office environment at 75 degrees F. (24 degrees C.)
ambient. The servers were either air-cooled or water-cooled using
an outdoor miniature cooling tower with water at 65 degrees F. (18
degrees C.). The temperature data is shown in FIG. 20. When the
liquid cooling was turned off the HVAC system was not able to keep
up, so the door was opened slightly to keep the temperature
relatively constant. A set of 7 Servers (3.times. HP Proliant DL380
G4 2.times.3.4 GHz and 4 Verari 2U 2.times. Opteron 245) consumed 2
kW using air-cooling while running a processor stress test program
(2 instances of BurnK7). With liquid cooling, and slowed-down fans,
the power was reduced to 1.8 kW with 1 kW of heat extracted using
the liquid cooling system. In addition, the average processor
temperature decreased 25 F (14 C). The hard drives warmed slightly
with liquid cooling due to the reduced airflow, but unlike
processors, they last longer at warmer temperatures.
[0112] The RAM temperatures were lower with liquid cooling because
the RAM chips were located downstream of the heat exchanger.
Assuming a typical data center power distribution of 56% Servers,
30% HVAC, 5% UPS and 6% other, the total power required for the
original air cooled system would be 3.6 kW (Server Power divided by
0.56). Using liquid cooling allows 1 kW to bypass the HVAC system
and go directly outdoors, saving HVAC power. And this has a
multiplying energy savings effect, since it takes more than 1 kW of
energy for an HVAC system to remove 1 kW of heat. It also saved 10%
of the server input power due to lower fan power and because the
processors required less power at lower temperatures. The liquid
pump and cooling tower fan used only 50 watts. This reduced the
overall power consumption based on typical data center power
distribution to 2.9 kW, a total power reduction of approximately
20%. The power reduction is diagrammed in FIG. 21. This experiment
was done using a miniature cooling tower that was only 52%
efficient which lowered the water temperature down to (65 degrees
F.) 18 degrees C. in a (50 degrees F.) 11 degrees C. wet bulb
environment. A commercial-grade cooling tower with 75% efficiency
would be able to reduce the temperature of the cooling water to (59
degrees F.) 15 degrees C. Assuming that the heat removed is
proportional to the difference between the ambient and the cooling
water inlet, the more efficient cooling tower would boost the heat
removal by 50%, leading to a predicted total overall power savings
of 25%.
[0113] Accordingly, the combination of an air-cooled heat sink
modified for redundant liquid cooling, a negative pressure system
to prevent leaks, and a connector that automatically purges the
coolant adds up to a system that offers a path from the current
air-cooled technology to the liquid cooled data center of the
future, without having to modify the building. The present
liquid-cooled and air-cooled heat sink system reverses the
thermodynamics of traditional systems so that the heat sink removes
heat from the CPUs and the server interior and the data center in
general in order to reduce the HVAC loads and fan power by a large
margin.
6.7 Single Vacuum Pump Cooling System
[0114] In previously described embodiments, a separate vacuum pump
and circulation pump are used to circulate the coolant throughout
the system at negative pressure. The embodiment shown in FIG. 22
contains one vacuum pump that both circulates the coolant and
creates the negative pressure. The benefit to a single vacuum pump
system is that it is less prone to failure and it uses less energy
to operate.
[0115] Turning in detail to FIG. 22, a single vacuum cooling system
2200 contains a single vacuum pump that creates a vacuum line 2202
and the system further includes a pressurized line 2204, i.e., a
line that is at higher pressure than the vacuum line 2202. The
vacuum pump is illustrated in FIG. 26 and discussed below. The
vacuum line 2202 is connected to both a main chamber 2206 and an
auxiliary chamber 2208, with a valve 2210 regulating the vacuum
line 2202 to the main chamber 2206 and valve 2212 regulating the
vacuum line 2202 to the auxiliary chamber 2208. The pressure line
2204 is connected to both a main chamber 2206 and an auxiliary
chamber 2208, with a valve 2214 regulating the pressure line 2204
to the main chamber 2206 and valve 2216 regulating the pressure
line 2216 to the auxiliary chamber 2208. Valves 2210, 2212, 2214
and 2216 make up a valve assembly, and that assembly is controlled
by the controller 2232. As described below with reference to FIGS.
23 and 24, switching the valves 2210, 2212, 2214 and 2216 will
cause the coolant to circulate throughout the system under negative
pressure.
[0116] In one embodiment, both the main chamber 2206 and the
auxiliary chamber 2208 are connected to the reservoir 2218, such
that the coolant can travel in one direction from the main/aux
chamber to the reservoir, the one direction travel being
accomplished by the use of check valves 2220. The reservoir 2218 is
where the coolant is drawn from for circulation to the electronic
equipment, shown as servers 2222, and the removal of heat from that
equipment through the use of an electronic equipment heat exchanger
2223. The reservoir 2218 connects to the primary heat exchanger
2224 (this can be a liquid-liquid exchanger or an air-liquid
exchanger) reducing the temperature of the coolant prior to
circulating the coolant via cold manifold 2226 to the servers 2222,
and returning the heated coolant via hot manifold 2228 back to the
main and auxiliary chambers (2206 and 2208). The coolant from the
hot manifold 2228 travels only in one direction to the main and
auxiliary chambers (2206 and 2208), the one direction travel being
accomplished by the use of check valves 2230. The travel of the
coolant throughout the system 2200 is also referred to herein as
the coolant circuit.
[0117] Alternatively, the main chamber 2206 and the auxiliary
chamber 2208 can be connected directly to the primary heat
exchanger 2224, completely obviating the need for the reservoir
2218. The reservoir 2218, however, is helpful in equalizing the
negative pressure through the system 2200, such that the flow of
coolant is more constant and less pulsating. Also, the reservoir
2218 allows the system 2200 to hold more coolant, minimizing the
possibility that the system 2200 will run dry.
[0118] The system 2200 may also have redundant valves and pumps to
reduce the chance of shutdown to a negligible level. One such
redundancy system may have two vacuum pumps running at 50%
capacity, such that if one fails, the other ramps up to cover the
load. This redundancy also imbues the system 2100 with enough
vacuum capacity to work with one server completely open to air.
[0119] The system 2200 may also have several sensors, filters and
structures to help optimize its performance. For example, the
reservoirs 2218 may include level sensor to make certain that there
is sufficient cooling liquid in the system to meet the demands of
the electronic equipment. Filters may be placed throughout the
system to remove debris that could interfere with the valves and
negatively affect performance. A set of temperature (2240) and
pressure (2242) sensors may be placed on the cool manifold and a
set on the hot manifold to detect the temperature and pressure
difference of the coolant. All the information from these sensors
may be fed to the controller 2232. If for example, the system
detects insufficient coolant, the system may open the fill valve
2234 to add more coolant and alert the system operator that the
coolant level was low. If the pressure sensors detect an abnormal
pressure drop, this could signal that there is a leak in the system
and the system would alert the operator. Because the system
operates under negative pressure, the leak would not expose the
computer equipment to the cooling liquid, but rather would
introduce air into the system and potentially reduce the efficiency
of the system in cooling the computer equipment. To reduce the
ability of a leak to compromise the cooling efficiency of the
system a novel set of valves and nozzles are used on the hot and
cold manifolds, and discussed in greater detail with reference to
FIGS. 27A and 27B. The system 2200 may also have a condensation
return line 2250 connected to the main or auxiliary chambers, for
use in the capture of coolant as described in reference to FIG.
26.
[0120] Other valves may be used to further optimize the system. For
example, test valve 2236 may be used when the system is first
turned on. Test valve 2236 should remain closed until the system
detects at the various pressure sensors that the appropriate amount
of negative pressure has been reached and maintained. This prevents
the system from being activated with leaks present and prevents
coolant from circulating to the electronic equipment under
atmospheric or near atmospheric pressure, such that a leak would
actually cause coolant to spill. Purge valve 2238 may be used to
purge the system of coolant when the system is turned off. Again,
this prevents coolant from remaining in the electronic equipment
plumbing under atmospheric or near atmospheric pressure, such that
a leak would actually cause coolant to spill.
[0121] The components of the system 2200 encompassed by the box
2244 may be sufficiently small to be installed as a rack mount
device in a traditional server tower. Further, those components may
be placed on a tray such that any leaks that may occur in the
rack-mounted unit would be captured by the tray and would not
impact any of the server equipment.
[0122] FIGS. 23A and 23B illustrate a top and isometric view of an
actual rack-mountable system 2300 that would be encompassed by a
box 2244. The principal components of the system include the main
chamber 2206 and the auxiliary chamber 2208 that are in one-way
fluid communication with the reservoir 2218. Valves 2210 and 2212
are connected to the vacuum line. Valves 2214 and 2216 are
connected to the pressure line. The liquid-liquid heat exchanger
224 is connected to the reservoir 2218. Arrows 2246 illustrate the
movement of the coolant into and out of the rack-mountable system
2300. A tray 2248 may be placed under the system 2300 to capture
any liquid that might escape, thus preventing damage to any
equipment that is in the same server rack as this rack-mountable
system 2300. The rack-mountable system 2300 could cool up to 10 kW
of servers in up to 10 racks. This would allow the costs of the
system to be spread out the cost over a number of servers. It would
be apparent that the teaching of this disclosure can be used to
cool even larger server farms.
[0123] The operation of the system will now be described with
reference to FIGS. 24 and 25. In FIG. 24, the main chamber 2206 is
emptying into the reservoir 2218, as shown by arrow 2405, while the
auxiliary chamber 2208 is filling with water returning from the
computer equipment as shown by arrow 2410. This circulation is
accomplished by opening valve 2114 (which is pressurized) and
simultaneously opening valve 2212 (which is under vacuum). This
creates a difference in pressure between the main chamber 2206 and
the auxiliary chamber 2208 of about 10 to 15 in Hg, circulating the
coolant though the system. Once the main chamber 2206 has emptied
sufficiently, then it must be filled and the auxiliary emptied to
continue the circulation. This operation is shown in FIG. 25. In
FIG. 25, the auxiliary chamber 2208 is emptying into the reservoir
2218, as shown by arrow 2505, while the main chamber 2206 is
filling with water returning from the computer equipment as shown
by arrow 2510. This circulation is accomplished by opening the
valve 2216 (which is pressurized) and simultaneously opening the
valve 2210 (which is under vacuum). This creates a difference in
pressure between the main chamber 2206 and the auxiliary chamber
2208 of about 10 to 15 in Hg, circulating the coolant though the
system. The functions illustrated in FIGS. 24 and 25 are
alternatively performed creating a circulation of the coolant using
a single vacuum pump. To optimize the system, there may be slight
overlap between these alternatives, but for the majority of time,
when one chamber is filling the other is emptying. The control of
the valves is accomplished by way of the controller 2232.
[0124] FIG. 26 illustrates the single vacuum pump 2602 used in
system 2200. The pump compressor 2602 may actually include two or
more vacuum pumps to provide redundancy to the system, which would
reduce the chance of shutdown to a negligible level. For example
having two vacuum pumps 2602 sized to run at 50% capacity, would
allow one vacuum pump to take the entire load should the other
fail. The vacuum pump is connected to the vacuum line 2202 and its
exhaust is outputted to the pressurized line 2204. The vacuum pump
creates the pressure differential that causes the coolant to
circulate through the system 2200. Prior to the vacuum pump 2602 a
coolant recovery device 2608 may be placed. Here the coolant
recovery device 2608 is an air/water separator, but other coolant
recovery devices include, and are not limited to, mufflers and
thermoelectric devices that condense any moisture out of the air.
In fact, a thermoelectric device may be used to condense moisture
out of the atmosphere in order to make up for any coolant loss. As
the vacuum line sucks air out of the main and auxiliary chambers,
the air will be at or near 100% relative humidity and if that
moisture in the air is not captured, then the system will require
frequent coolant addition. This can then become an annoying
maintenance issue. By routing the vacuum line 2202 through the
coolant recovery device 2608, moisture can be removed from the
humid air in the vacuum line 2202. That coolant drops to the bottom
of the device 2608. The device 2608 is connected to main chamber
2206 via the condensation return line 2250, such that when the main
chamber 2206 is under vacuum, there is a pressure differential of
about 10 to 15 in Hg pushing the recovered coolant to the main
chamber 2206. A check valve 2610 prevents the coolant from
traveling in the wrong direction within the condensation return
line 2250. Of course, a pump could also be used to pump the coolant
back to the system. For example, a piston pump, gear pump or
peristaltic pump would be suitable.
[0125] After the vacuum pump 2602, a second coolant recovery device
2604 may be placed. Here the second coolant recovery device 2604 is
a muffler/condenser. Although not shown, the pressurized line 2204
may be vented to atmospheric pressure. When vented in this fashion,
the system 2200 still operates, and the pressurized line 2204 would
be at atmospheric pressure and would be pressurized as compared to
the vacuum line 2202. As the vacuum pump 2602 evacuates the air, it
will still have some humidity, and if that moisture in the air is
not captured, then the system will require frequent coolant
addition. The second coolant recovery device 2604 condenses the
coolant out of the air, allowing the coolant to collect at the
bottom of the device 2604. Here the coolant recovery device 2604 is
a muffler, but other coolant recovery devices include, and are not
limited to, air/water separators and thermoelectric devices that
condense any moisture out of the air. In fact, a thermoelectric
device may be used to condense moisture out of the atmosphere in
order to make up for any coolant loss. The device 2606 is connected
to the main chamber 2206 via the condensation return line 2250,
such that when the main chamber 2206 is under vacuum, there is a
pressure differential of about 25 to 30 in Hg. A float valve 2606
may be used, such that the valve 2606 is closed until a sufficient
amount of coolant has collected at the bottom of the device 2604.
Once the valve 2606 opens, the pressure differential pushes the
recovered coolant to the main chamber 2206. Of course, a pump could
also be used to pump the coolant back to the system. For example, a
piston pump, gear pump or peristaltic pump would be suitable.
6.8 Flow Control in the Event of a Single Point Gross Leak
[0126] In the event that one of the servers has a damaged liquid
cooling system and is leaking air into the system through a
completely broken coolant conduit, the rest of the system should
still operate, provided that the leakage rate into both sides of
the liquid cooling system is controlled.
[0127] Referring to FIGS. 27A and 27B, structures that can control
the leakage rate are described. FIG. 27A illustrate a valve 2705 on
the supply side (i.e., the valve 2705 is upstream of the electronic
equipment as defined by the direction of coolant flow shown by
arrow 2710), and FIG. 27B illustrates a valve 2715 on the return
side (i.e., the valve 2715 is downstream of the electronic
equipment, as defined by the direction of coolant flow shown by
arrow 2720). In one embodiment these valves may be placed near the
hot and cold manifold 2228, 2226) as shown in FIG. 22. On the
supply side (valve 2705), the coolant may flow through a check
valve, such as the one described by Tesla or a fluid diode, or a
poppet and seat type of check valve with a built in leak. A groove
in the seat can provide the necessary leak. The opening in the seat
is shown at 2725, which allows a leak across the valve 2705 shown
as arrow 2730. The reverse flow through the valve 2705 must be such
that any coolant in the line upstream from the valve 2705 can be
sucked back into the cold manifold 2226 and not leak out on the
electronic equipment 2222. The leak, however, must not be so severe
so as to adversely affect the other servers in the loop by
introducing too much air into the system.
[0128] On the return side (valve 2715), the air flow must not be so
excessive as to reduce the pump effectiveness substantially. One
way to achieve this goal is to use a Venturi, which has a low
pressure drop when flowing coolant at the nominal flow rate, and
limits the flow of air into the system to that which can flow
through the minimum diameter of the Venturi at the speed of sound.
For example, a Venturi with a throat of 0.05 inches, will have a
pressure drop of approximately 1 in Hg at 300 cc/min coolant flow
rate, and it will flow approximately 1/2 standard cubic feet per
minute of air at 20 in Hg vacuum. A large-scale system designed to
cool 100 kW of servers may flow approximately 35 gal/min of
coolant. A completely open sever line would therefore represent
approximately 17% of the overall volume flow rate, and the system
could still efficiency cool the electronic equipment.
6.9 Leak Detection
[0129] Leak detection can also be included in the systems
previously described. Detecting leaks is important because it can
lower the efficiency of the system. To detect a leak of air into
the system, the flow rate of air and coolant back into the system
should be measured. The flow rate of coolant may be measured by
measuring the time it takes to fill one of the chambers (i.e.,
2206, 2208) because the volume of those chambers is already known.
Placing a level sensor in the chamber (see FIG. 22, part 2260)
which sends a signal to the controller 2232 would allow the
controller 2232 to calculate the coolant flow throughout the
system. Alternatively, a flowmeter may be placed on the coolant
line to measure the flow directly.
[0130] The flow rate of air may be measured by a flowmeter (see
FIG. 22, part 2262) in the vacuum line 2202, that outputs a signal
to the controller 2232. This flowmeter 2262 may be, but is not
limited to, a hot wire, laminar flow element, orifice, or venturi.
Under normal operation, when the main chamber 2206 is filling under
suction, there will be a period when the air in the main chamber
2206 is being pumped down to the correct negative pressure level.
After that period the negative pressure level will stabilize, and
the flow rate of air should be equal to the flow rate of coolant.
As is known in the art, air may dissolve in water or other
coolants, and the maximum amount that can be dissolved as a
function of temperature and pressure is well known. For example at
80 F, up to 1.6% air may be present in water. Therefore, if an air
leak exists, it may be detected by measuring the air flow rate out
of a pump chamber (i.e., 2206, 2208), comparing it to the coolant
flow rate into the pump chamber, and if the air flow rate is
excessive this would indicate an air leak and an alarm would
optionally be activated to alert the operator that a leak is
present. This comparison and triggering of the alert may be
performed by the controller 2232.
[0131] For example, if the vacuum pump has a displacement of 0.05
liters per revolution, and it spins at 1500 RPM, then it should
flow 75 liters per minute. There are some losses and leaks within
the vacuum pump, but they are repeatable and known for a given
pump. Therefore given the RPM of the pump, and the pressure at the
inlet, then the mass flow rate can be determined using the ideal
gas law. The flow rate can also be measured at the vacuum pump
outlet by means of a thermal mass flowmeter, or a venturi, orifice
or other flow meter. The flow meter should be calibrated for humid
air. The temperature and pressure of the water entering and leaving
the cooling system is known, and this can be used to predict the
amount of air that is dissolved in the water. It can also be used
to predict the amount of water vapor in the gas above the water.
The maximum amount of air that can be evolved from the water is the
difference between the amount that could be dissolved at the
temperature and pressure of the water leaving the cooling system,
and the amount of air that could be in the water returning to the
system. The controller could use a look up table to calculate the
two amounts and the difference that would be expected at the system
air outlet. If the amount of air in the water returning to the pump
from the servers is excessive, then an alarm could be activated. A
humidity sensor could be used to determine how much of the gas
evolved is water vapor, or testing could be used to determine the
range of outlet humidity. For example, in a typical system with a
single vacuum pump, the maximum vacuum achieved with no air leaks,
and a flow of 1 gal/min of water might be 24 in Hg. If there is a
leak of 0.03 ft.sup.3/min of air into the system, then the maximum
vacuum will be only 20 in Hg, and this would indicate a leak or a
vacuum pump failure.
[0132] The invention has been described in connection with specific
embodiments that illustrate examples of the invention but do not
limit its scope. Various example systems have been shown and
described having various aspects and elements. Unless indicated
otherwise, any feature, aspect or element of any of these systems
may be removed from, added to, combined with or modified by any
other feature, aspect or element of any of the systems. As will be
apparent to persons skilled in the art, modifications and
adaptations to the above-described systems and methods can be made
without departing from the spirit and scope of the invention, which
is defined only by the following claims. Moreover, the applicant
expressly does not intend that the following claims "and the
embodiments in the specification to be strictly coextensive."
Phillips v. AHW Corp., 415 F.3d 1303, 1323 (Fed. Cir. 2005) (en
banc).
* * * * *