10

Special Applications for Tough Cooling Jobs

10.1 NEW TECHNOLOGY-APPROACH

WITH CAUTION

Design engineers are always seeking better methods for packaging high-powerdissipating electronic equipment. As the power dissipations continue to increase, standard conduction and forced-air convection techniques no longer provide adequate cooling for sophisticated electronic systems. The reliability of the electronic system will suffer if high temperatures are permitted to develop. This is extremely important in mass transportation systems such as airplanes, ships, and trains. The failure of a critical electronic control element in these systems can lead to extensive property damage and the loss of many lives. When standard cooling methods are no longer adequate, exotic new methods are often utilized. New cooling methods should be approached with some caution, because they often bring in new technology, which may contain hidden dangers and new failure mechanisms that are not well known or easily recognized. New materials, together with new manufacturing processes and new technology, will always lead to new problems.

10.2

HEAT PIPES

A heat pipe is a hollow tube type of enclosed structure, containing a fluid that transfers large quantities of heat when it evaporates and a wick that brings the fluid back to its starting point when it condenses. This entire process is accomplished with no outside power, no mechanical moving parts, and no noise. The design is extremely simple and very efficient, since it can transfer heat hundreds of times better than any solid metal conductor [62-651.

363

364

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

The birth of the heat pipe dates to about 1942. Very little attention was paid to this device until about 1963, when it was suggested for spacecraft applications. Since then, many different heat pipe applications have found their way into cooking food, heating homes, cooling motorcycle engines, and recovering heat in industrial exhaust systems. Heat pipes can be made in many different shapes and sizes. The most common shape is the hollow cylinder or tube. They are often made in flat shapes, with S turns, and in spirals. A typical heat pipe consists of a sealed tube that has been partially evacuated, so that its internal pressure is below the standard atmosphere of 14.7 psia. The inside walls of the tube are usually lined with a capillary wick structure and a small amount of a fluid, which will vaporize. When heat is applied at one end of the tube, the fluid within the pipe vaporizes or boils. This generates a force that drives the vapor to the opposite end of the tube, where the heat is removed. Removing the heat forces the vapor to condense, and the wick draws the fluid back to the starting point, where the process is repeated [66-671. There is a small pressure drop between the heating (or evaporating) end and the cooling (or condensing) end of the pipe. Therefore, the boiling and condensing cycle takes place over a very narrow temperature band. As a result, there is only a small temperature difference between the heat source and the heat sink. A 24 in (60.9 cm) long pipe, 0.50 in (1.27 cm) in diameter can pump 220 watts of heat at 212F (lOO"C), with about a 2.4"F (1.3"C) temperature difference along the length of the pipe. A solid copper bar for the same length and power, but with a 100F (55.5 "C) temperature difference along its length, would require a cross-sectional area of about 10.8 in2 (69.7 cm2), and would weigh about 75 lb (34,050 g), compared to the heat pipe weight of about 0.75 lb (340 g). The wick is probably the most critical part of the heat pipe design. It determines the capillary sucking action available for drawing the condensed liquid back to the evaporator end. The porosity and the continuity of the internal passages determine the fluid resistance along the wick. The wick must have sufficient capacity to supply liquid to the heat input end. An inadequate fluid return will result in the drying up of the wick at the heat input end, which results in a breakdown of the wick operation [68, 691. Capillary action of the wick in the heat pipe permits it to operate in any orientation in a gravity or acceleration field. A typical wick on earth, which has an acceleration of 1.OG, must be capable of raising the fluid against gravity and with a small internal resistance to flow at the same time. A small pore size is required to draw the liquid up to a high level in a capillary tube. However, a small pore size increases the internal flow resistance. A balance must therefore be made to provide good fluid flow over long distances in the heat pipe wick design, depending upon the orientation in a gravity or acceleration field. A typical section through the length of a heat pipe is shown in Figure 10.1. Heat pipe wicks are often made of porous ceramic or woven stainless steel wire mesh. Sometimes the wick is an integral part of the tube housing, formed by extruding small grooves along the inner surface of the wall structure, as shown in Figure 10.2.

10.3 DEGRADED PERFORMANCE IN HEAT PIPES

365

end

IWick

--

section Liauid return

- -

end

Sealed
end

. . . . .

in

transport

out

Figure 70.1 Section through the length of a typical heat pipe.

Extruded grooves

Figure 10.2 Heat pipe wick designs.

10.3 DEGRADED PERFORMANCE IN HEAT PIPES

Heat pipes appear to have the ability to solve many thermal problems involving high heat densities or the transport of heat over long distances. However, it should be pointed out that many of the old heat pipe manufacturers have gone out of the business, and only a handful1 are left. This points out how difficult it is to make good heat pipes, with good quality control, which will give good performance for a long period of time, and which can be sold at a good price. One of the biggest problems with heat pipes is that many of them often exhibit degraded performance after they have been in operation for about three to six months. This degraded performance occurs slowly, so it may not be noticed in the operating hardware. The operating temperatures of the equipment will increase over a period of time, until malfunctions and failures occur. Since thermocouples are not usually mounted in operating hardware, the gradual temperature increase over a long time span usually goes unnoticed. The single greatest cause of this degraded performance appears to be contamination that affects the vapor pressure within the heat pipe. Contamination often results from the type of operation that is used to seal the ends of the tube after it is assembled. Many sealing methods can be used effectively if the processes are properly controlled, the parts are adequately cleaned, and the assembly takes place in a clean room. Electron beam welding is reported to be one of the best known methods for sealing heat pipes today. This technique appears to provide a good

366

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

seal with very little contamination. Many other sealing methods are still used, because they are less expensive. If the processes are properly controlled, a good seal can be obtained. Other sources of outgassing are often found in the wick. Since the wick must utilize many small orifices to pump the fluid from the condenser to the evaporator, this section can easily contain trapped gasses that may affect the vapor pressure. Wicks must therefore be manufactured and cleaned very carefully before they are installed in the heat pipe. A clean room should be used to assemble heat pipes, to minimize possible contamination. The maximum airborne particle contamination should be about 100,000 particles per cubic foot, when the particle size is around 0.5 pm. When the particle size is around 5.0 pm, the maximum airborne particle contamination should be limited to about 700 particles per cubic foot. Burn-in tests should be performed on all heat pipes, for at least 100 hr. The bum-in should be performed with 100% of the required design load but in an ambient temperature at least 20% higher than the highest specified operating temperature, to ensure reliable operation. The cleaning process is extremely important. One heat pipe manufacturer reported an attempt to reduce manufacturing costs by using the extruded grooves in place of the porous material for the wick, as shown in Figure 10.2. The fluid that was used in the extrusion process was very difficult to remove completely, and the resulting Contamination affected the internal vapor pressure. This reduced the performance of the heat pipe after several months of operation. Since the particular program required very precise temperature controls for very long time periods, it was decided that the lower cost extruded grooves should not be used for this application.

10.4

TYPICAL HEAT PIPE PERFORMANCE

A heat pipe can work in any orientation, but its performance may be degraded when it is forced to work against gravity. This condition occurs when the heat input end (evaporator end) is higher than the heat output end (condenser end). In this position, the fluid in the wick is forced to move up the pipe, against the direction of gravity. Figure 10.3 shows approximately how the performance of a 48 in (122 cm) long water heat pipe with a coarse, medium, and fine wick changes with the angle of orientation (0). The coarse wick is capable of handling much more power while it is operating in a horizontal position. However, when the condenser end is angled down slightly, its pumping capacity is sharply reduced. The fine wick cannot pump as much heat in the horizontal position. However, its pumping capacity does not drop off as fast when the condenser end is angled down. The typical performance that can be expected from various heat pipes is shown in Table 10.1. A wide variety of fluids can be used with heat pipes. A few of these fluids and their typical operating ranges are given in Table 10.2.

10.4 TYPICAL HEAT PIPE PERFORMANCE

EvaDorator end

367

'I

4 20 " "

'
I
10

Condenser end

01

I 20

30

40

I 50

60

70

80

90

Angle, 8

Figure 10.3 Changes of the heat pipe capability when the condensor end is below the evaporator end. TABLE 10.1 Typical Heat Pipe Performance

Outside Diameter (in)

Length (in)
6 12 18 6 12 18 6 12 18

Power (watts)
300 175 150 500 3 75 350 700 5 75 550

a
8

TABLE 10.2 Operating Temperatures for Several Heat Pipe Fluids

Fluid Ammonia Sulfur dioxide Water FC 43 Mercury Cesium Sodium Lithium

Operating Temperature Range ( O F )

-50- 125 15-110 40-450 250-430 400-820 750- 1830 920-2200

1500-3000

368
10.5

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

HEAT PIPE APPLICATIONS

Heat pipes are generally used to transport heat from the source directly to the sink without any external power. Heat pipes can eliminate hot spots and can accept heat that has a high power density. Most heat pipes are external to the device they are cooling. Some new applications are available where the heat pipe is an integral part of the case on an electronic component. A typical application may involve a component or a subassembly, such as a PCB, which has a high power dissipation and a relatively long heat flow path to the heat sink. The resulting temperature rise from the electronic components to the heat sink is excessive, so forced-air cooling and then liquid cooling are examined. When there is no room for fans and ducts, and when the size, weight, and price of a liquid system is evaluated, the price of one or two small heat pipes may look more attractive. Many compact airborne electronic systems utilize air-cooled cold plates for the sidewalls of the chassis. Plug-in PCBs are then used to support the electronic components. These are cooled by conducting the heat, along metal strips laminated to the PCB, to the sidewalls of the chassis, as shown in Figure 10.4. Considering a typical ATR size chassis, the cross-sectional dimensions will be approximately as shown in Figure 10.4. A plug-in PCB with an aluminum core might be capable of dissipating about 15 watts, based upon a maximum component surface temperature of 212F (l0OOC) and a cooling air temperature of 131F (55 "C) flowing through the sidewall cold plates. What happens when the same size PCB has a power dissipation of 70 watts? A

Wire harness

Plug in connector

'\Bottom

cover

Figure 10.4 Cross section through a chassis with cold plate sidewalls.

10.5 HEAT PIPE APPLICATIONS

369

metal heat sink core on the PCB would have to be so large and heavy that it is simply unacceptable. A hollow core air-cooled PCB, similar to those shown in Figure 6.21, might be used with a multiple-fin heat exchanger at the center. However, even this type of construction is only capable of dissipating about 50 to 55 watts in the same environment. A liquid cooling system could be used, and it would do an excellent job. However, there would be a large increase in the size and weight of the system. Heat pipes are ideal for the conditions described above. They can be added to the back surface of the PCB to sharply increase the heat transferred from the center of the PCB to the edges. A high temperature rise may still occur at the interface of the PCB with the chassis cold plate, unless a high interface pressure device such as a wedge clamp, shown in Figure 3.23d, is used. Aluminum or copper heat sinks should be used under the heat pipes on the PCB to improve the heat flow to the heat pipes, as shown in Figure 10.5. Heat pipes have been made with 90" bends to improve the heat transfer from circuit boards that do not plug in. When cold plate sidewalls are used in a chassis, the 90" bend permits the heat pipe to carry the heat directly from the PCB to the sidewall of the chassis, as shown in Figure 10.6. This sharply reduces the temperature rise across the interface from the PCB to the chassis. Sometimes cooling fins must be extended to improve the cooling by increasing the effective surface area. The heat transfer efficiency of a long fin can be improved substantially by using heat pipes along the length of the fin, as shown in Figure

Component mounting surface

>
'

Flat heat pipes in alurninusTn[CB heat

Plug in connector

Figure 10.5 Flat heat pipes in an aluminum composite plug-in PCB.

370

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

Components on printed circuits

Chassis cold plate side wall

r"

I 0 0 I I 1.Q 00 I

Aluminum heat sink and heat pipe with 90" bend

Figure 70.6 Heat pipe with a 90 bend for cooling circuit boards.

10.7. The temperature gradient along the fin will be quite small with the heat pipe, producing an isothermal fin. The temperature difference between the fin and the ambient air will be increased, so that the heat transfer from the fin is also increased. Cooling fins can be added to the condenser end of the heat pipe to improve natural and forced convection cooling. The heat pipe can be bent, if necessary, to reach hot components located in remote areas of a chassis or console. Figure 10.8 shows a tall console with finned heat pipes that extend through the rear panel to provide the required cooling. Experimental models have been made with small heat pipes that are fabricated into the structure of the electronic component itself. Power transistors have been built with a porous dielectric wick that is placed in contact with the active chip

Figure 70.7

Using heat pipes in long cooling fins to improve cooling efficiency.

10.6 DIRECT AND INDIRECT LIQUID COOLING

High power electronic component

371

Finned heat pipes

Tall electronic console

Rear panel

Figure 10.8 Section through a tall console showing heat pipes extending through the rear panel to improve cooling.

substrate and the inside surfaces of a standard TO type of container. The unit is evacuated, the wick is saturated with a suitable fluid, and the assembly is hermetically sealed. The evaporation process of the heat pipe takes place on the transistor substrate, and the condensation takes place on the cooler surfaces of the transistor walls. Capillary action in the wick brings the fluid back to the substrate, where the evaporation process is repeated.
10.6 DIRECT AND INDIRECT LIQUID COOLING

Liquid cooling, in general, is much more effective for removing heat than air cooling. Therefore, when high power densities are involved, liquid cooling may be the only practical method for maintaining reasonable component temperatures. Liquid cooling systems are basically classified as direct or indirect. In a direct cooling system, the liquid is in direct contact with the electronic components, which permits the coolant to pick up the heat and carry it away. In an indirect system, the liquid coolant does not come into direct contact with the components. Instead, heat is transferred from the hot component to some intermediate system and then to the liquid. The intermediate system conveys the heat to the liquid by conduction, convection, or radiation. Typical examples of intermediate systems are heat exchangers and fans [ 6 ] . Direct liquid cooling systems usually have the electronic components completely immersed in a fluid that has no effect upon the electrical operation of the system. Heat is transferred directly to the fluid by conduction and convection, with very little radiation. Sometimes a low pressure pump is added to circulate the fluid

372

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

through the electronics to increase the effective cooling. Sometimes the pump is used to spray the liquid coolant directly on the electronic components to carry away the heat. Indirect liquid cooling systems usually shift the heat transfer problem from the electronic components to another, more remote location. Since the heat must be dumped somewhere, it is often convenient to pump the coolant fluid to a remote air-to-liquid heat exchanger. Large fans can then be used to drive cooling air through the heat exchanger, which will cool the fluid and heat the air. The cooler fluid is pumped back to the electronics to pick up more heat, and the hot air is exhausted to the surrounding ambient. indirect cooling systems are often used in missiles and satellites that must operate in the hard vacuum of outer space. The cooling liquid is usually circulated through cold plates, to pick up heat from components mounted on the cold plates. The liquid is then pumped to remote space radiators that are located on the surface of the spacecraft, facing away from the sun toward deep space, where the sink temperature is absolute zero (-460"R or -273K). Heat from the liquid is picked up by the space radiator and dumped into space. The cooler liquid is pumped back to the electronics to pick up more heat. The U.S.Navy makes extensive use of indirect cooling to remove heat from electronic systems. Electronic components are often mounted on cold plates that use fresh distilled deionized water as the coolant. Forced convection cooling with fresh water is much more effective than is forced convection cooling with air. Typical heat transfer coefficients for fresh water can easily be as much as 100 times greater than typical heat transfer coefficients for air. In addition, water has a specific heat that is more than four times greater than air. Water can therefore absorb four times as much heat as air for the same temperature rise and weight flow. The fresh water is cooled by circulating it through a remote heat exchanger, which has fresh water on one side and saltwater (seawater) on the opposite side. After the fresh water has been cooled, it is returned to the electronic cold plate. Saltwater is used to cool the fresh water. The salt water, which has picked up the heat from the fresh water, is dumped back into the sea. Special care must be used to prevent the fresh water from freezing during the winter, or when operating in cold climates.

10.7 FORCED-LIQUID COOLING SYSTEMS

High-power-dissipating electronic systems often make use of forced-liquid cooling techniques to control hot spot temperatures. One very common type of cooling device is the cold plate, which provides cooling by conduction and by forced-liquid convection. Electronic components are mounted on a metal plate, through which a cooling liquid is circulated, to carry away the heat. The electronic components are fastened directly to the cold plate, which is made of aluminum or copper for high heat conduction. This provides a good heat conduction path from the components to the cooling liquid in the cold plate [ 6 , 70, 711.

10.7 FORCED-LIQUID COOLING SYSTEMS

373

The methods for analyzing the thermal characteristics of a liquid cooling system are very similar to the methods shown in Chapter 6 for analyzing forced-air cooling for electronics. Many of the equations are the same, with only the values for such parameters as density, viscosity, thermal conductivity, and specific heat changing. A pump replaces the fan in a liquid-cooled system. To select a suitable pump, it is necessary to be able to calculate the total pressure loss through any given system. Three major pressure drops are usually evaluated: (1) friction, which is determined by the liquid velocity and surface roughness within the pipe; (2) difference in elevation; and (3) fitting losses due to elbows, tees, and transitions. These fittings are usually the major source of the pressure drop through an electronic liquid cooling system. As in the air-cooled systems, an exact calculation of the pressure drop is very difficult to obtain. Therefore, it is convenient to use approximate methods that are well documented. One of the most common methods makes use of an equivalent pipe length for various fittings such as elbows and tees. Since the pressure drop through straight pipes with various degrees of roughness is well documented, the calculations are simplified. To obtain a common basis for calculating the equivalent pipe length for various fittings, the effective length is expressed as the number of diameters of a round pipe. In this way, the equivalent length of many fittings can be determined. Table 10.3 shows the effective length of pipe for several fittings expressed in terms of their equivalent diameters. The values are for turbulent flow conditions, but they can also be used to approximate the pressure drop for laminar flow systems. Pressure losses will also occur at the pipe inlet, depending upon the geometry

TABLE 10.3 Effective Length for Various Pipe Fittings (3, 6,

241

Type of Fitting
45" elbow 90" elbow miter, zero radius 90" elbow, 1 .O diameter radius 90" elbow, 1.5 diameter radius 90" elbow, 2.0 diameter radius 180" bend, 1.5 diameter radius 180" bend, 4-8 diameter radius

Effective Length (Number of Diameters)

15 60 32 26 20 50 10 300 7 40

Globe valve, fully open Gate valve, fully open Gate valve, f closed Gate valve, 4 closed Gate valve, 3 closed Coupling union

200
800 0

374

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

TABLE 10.4 Effective Length for Various Pipe Entrances

-

___Effective Length (Number of Diameters)

Type of Opening

Description

?y
\ Qkuz

a m

Square edge entry

20

7 r
\

Well-rounded entry

& L

j .

J 7

Protruding entry (Borda's mouthpiece)

40

-T
Point Pointc

Tee elbow entering run a t point A

60

Tee elbow entering branch at point B Branch

90

of the inlet. Some typical losses are shown in Table 10.4. The entry losses are shown as an effective pipe length, which is expressed in terms of the equivalent diameter of a round pipe [3, 6, 241.

10.8 PUMPS FOR LIQUID-COOLED SYSTEMS

The pump must be capable of circulating the proper amount of coolant through the system against the total pressure drop through all the fittings, the heat exchanger, and the cold plate. The electric motor driving the pump will usually be matched to the pump so that the motor cannot be overloaded. Many different types of pumps are available, such as gear, reciprocating, and centrifugal. Gear type pumps do not become airbound very easily, as do centrifugal pumps. However, some centrifugal pumps have been developed to operate

10.11 SIMPLE LlOUlD COOLING SYSTEM

375

without rotating shaft seals, so that a long-term leakproof system can be obtained [721. A number of different safety devices are available to turn off the pump if the flow of the liquid is blocked or if the fluid leaks out. Protective devices are also available for preventing damage due to excessive pressure in a liquid cooling system.

10.9 STORAGE AND EXPANSION TANK

Totally enclosed cooling systems are often used to cool electronic equipment. Therefore, some provisions must be made to accommodate the thermal expansion of the fluid as the temperature increases. Also, air must be removed from the coolant and some type of cushion should be provided to reduce pressure surges in the system. This is all accomplished with an expansion tank, which also acts as the storage tank for the liquid coolant.

10.10 LIQUID COOLANTS

Water is probably the best cooling liquid available in terms of density, viscosity, thermal conductivity, and specific heat. For good long-term operation, distilled water that has been deionized should be used. When the temperature is expected to drop below freezing or when the surface temperature is expected to exceed 212F (lOO"C), ethylene glycol should be added to the water. Ethylene glycol has a lower thermal conductivity than water, so its addition will degrade the thermal performance of the water. However, when both aluminum and copper are being used in the cooling circuit, the addition of ethylene glycol will prevent corrosion of the passages. Many other liquids are available for cooling electronic systems, such as silicones, but they have a much lower thermal conductivity than water. Coolanol 45, a silicone ester made by Monsanto Chemical Company, is a very effective coolant that has a useful cooling range from -58F (-50C) to 392F (200C). Its viscosity is substantially higher than that of water. This may increase the pressure drop, which may require the use of a larger pump motor in a system where the water is replaced by Coolanol [72]. A liquid that has approximately the same viscosity as water is FC 75, a fluorochemical made by the 3M Company. This liquid has a specific heat about one fourth that of water. Tables 10.5 and 10.5A show the thermal properties of several different coolant fluids.

10.1 1 SIMPLE LIQUID COOLING SYSTEM

A simple liquid cooling system can be designed by combining the various items described in the previous sections. A simplified schematic of an elementary liquid cooling system is shown in Figure 10.9.

TABLE 10.5 Properties of Various Cooling Fluids: English Units

Water 77F (25C) Viscosity (Ibift hr) Density (lb/ft3) Thermal conductivity (Btuihr ft "F) Specific heat (Btuilb OF) Boiling Freezing
Pour Point.

FC 75 77F (25C) 3.49 109.8 0.037 140F (60C) 2.05 103.6 0.035

Coolanol 25 77F (25C) 10.89 56.19 0.076 140F (60C) 5.93 54.63 0.074

Coolanol 45 77F (25C) 43.32 55.87 0.078 140F (60C) 16.84 54.31 0.075

140F (60C) 1.14 61.40 0.378

2.17 62.2 1 0.353

0.998 0.998 212F (100C) 32F (0C)

0.248 0.263 214F (101C) -171F (-113C)

0.45 0.48 355F (179C) -125F (-87"C)*

0.45 0.49 355F (179C) -88F (-6TC)*

TABLE 10.5A

Properties of Various Cooling Fluids: Metric Units

Water 25C (77F) Viscosity (poise) (g/cm sec) Density (g/cm3) Thermal conductivity (cal/sec cm "C) Specific heat (cal/g "C) Boiling Freezing
'Pour Point

FC 75 25C (77F) 0,01441 1.757 0.000153 60C (140F) 0.00847 1.658 0.000144

Coolanol 25 25C (7PF) 0.045 0.900 0.000314 60C (140F) 0.0245 0.875 0.000306

Coolanol 45 25C (77F) 0.179 0.895 0.000322 60"C (140F) 0.0696 0.870 0.000311

60C (140F) 0.00470 0.982 0.00156

0.00896 0.995 0.00146

0.998 0.998 100C (212F) 0C (32F)

0.248 0.263 101C (214F) - 113C ( - 171F)

0.45 0.48 179C (355F) -87C (-125"F)*

0.45 0.49 179C (355F) -67C (-88"F)*

10.72 MOUNTING COMPONENTS FOR INDIRECT LIQUID COOLING

377

I\

Expansion and storage tank

\I

H h

Electronic equipment

10.12 MOUNTING COMPONENTS FOR INDIRECT LIQUID COOLING

Electronic components must always be mounted so that there is a low thermal resistance heat flow path to the heat sink. This is true for any electronic system that must provide reliable operation for steady state conditions. This may not be necessary for transient conditions, where the power is not on long enough for the temperature to stabilize. Components mounted on liquid-cooled cold plates should be provided with flat smooth surfaces, to improve the transfer of heat across the mounting interface. Power devices such as transistors, resistors, diodes, and transformers can be fastened to the cold plate with bolts that are capable of applying high forces. Bolted components can be removed easily if they have to be replaced. Silicone grease can be applied at the interface of the component to the cold plate to further reduce the thermal resistance on high-power-dissipating devices. Silicone grease should not be used anywhere on a cold plate if there are cemented joints in the area, Silicone grease tends to migrate, so that it will contaminate the cemented surfaces and reduce the bond strength. In a vibration or shock environment, the cemented interface can fracture at a very low stress level, producing a catastrophic failure. For very-high-power-dissipating components, it is often necessary to reduce the interface temperature rise to an absolute minimum. Under these circumstances, the component may be soldered directly to the cold plate mounting surface. This is done by first pretinning both surfaces with a low temperature solder. The solder is then reflowed using a high temperature for a short period of time to prevent heat from soaking into the component and damaging it. A heat sink may be used on the body of the component to pull away some of the heat during the reflow process, to reduce the possibility of damage. When a large number of similar high-power components must be cooled, it is convenient to mount them on a liquid-cooled cold plate, as shown in Figure 10.10,

378

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

Cold plate High power component

Additional methods for mounting high-power components are discussed in Sections 4.4 through 4.8. Plug-in PCBs are often cooled with the use of liquid-cooled cold plates, which are part of the side walls on an electronic chassis. Aluminum or copper heat sinks are used on the PCB to conduct the heat from the components to the sidewall cold plates, as shown in Figure 10.11. This technique is very similar to the air-cooled cold plates described in Sections 6.23 through 6.26, except that the cooling fluid is a liquid instead of air.

10.13

BASIC FORCED-LIQUID FLOW RELATIONS

The standard fluid flow equations can be used to solve heat transfer problems for liquids as well as for air. Most of the airflow equations can also be used for liquid flow, with some minor modifications. The symbols used for airflow are the same as the symbols used for liquid flow, except that the values are different for liquids and gases.

High power comp

Liquid cooled si

Aluminum core PCB

Figure 10.17

Liquid-cooled chassis with conduction-cooled PCB.

10.13 BASIC FORCED-LIQUID FLOW RELATIONS

379

The two parameters most often evaluated in forced-air cooling, as well as in forced-liquid cooling, are the temperature rise and the pressure drop. The temperature rise is measured in O F for English units and in "C for metric units. The pressure drop in air-cooled systems is measured in inches or centimeters of water. However, with liquid cooled units the pressure drops are much higher, so that they are normally expressed in lb/in2 or in g/cm2. The temperature rise in a liquid-cooled system can be determined from Eq. 6.7. The symbols remain the same as those defined by that equation, except that everything now relates to the flow of the liquid instead of the air flow. Consistent sets of units must be used for English and metric systems.
A t = - (ref. Eq. 6.7) WC, where Q = power dissipation W = liquid coolant flow rate C = specific heat of the liquid coolant ,
A convection film will develop in liquid cooled systems as well as in air-cooled systems. This film clings to the heat transfer surface and restricts the flow of heat. The forced convection coefficient for liquid cooling is also shown as h, and its characteristics are defined in Eqs. 3.53 and 5.7 for English units and metric units. The temperature rise across the forced convection film then becomes: At = hcA

(ref. Eqs. 3.53 and 5.7)

where Q h, A

= = =

power dissipation forced convection coefficient surface area

The value of the liquid forced convection coefficient can be determined from Eq. 6.9. The symbols are the same, except that they now must reflect the values of the liquid instead of the air.
h, = JC,G h, = JC,G

(F,) (g)
/ \

-213

(ref. Eq. 6.9)

where J = Colbum factor C = specific heat , G = weight flow velocity K = thermal conductivity p = viscosity The Reynolds number, which is shown in Eq. 6.11 must be obtained before the Colburn J factor can be determined. The symbols are the same as for air; only the values change.

380

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

GD N R = -VDP - CL
CL

(ref. Eq. 6.11)

For laminar flow conditions through smooth tubes with Reynolds numbers less than about 2000, the Colburn J factor can be determined from Eq. 10.1. J =

1.6
(~~)0.666

(10.1)

3 ' ) ,

The forced convection coefficient h, has relatively little change in the laminar flow region, with low weight flow velocity ( G ) values. In this range the forced convection coefficient is really determined by the hydraulic diameter, as shown in Figure 10.12. The forced convection coefficient increases rapidly as the hydraulic diameter decreases. When the weight flow velocity increases in the turbulent flow range, there is a corresponding increase in the convection coefficient [ 6 ] . For turbulent flow conditions through smooth pipes, where the Reynolds number is greater than about 7000, the J factor can be determined from Eq. 10.2 [3, 151. (10.2)

2
800
U
O

Weight velocity, G (g/wc cm2) 20 40 6 810

60 80100

Tube diameter

600

4i

400-

0.125

0.317

- 0.010
- 0.008
I
.t"

rn

10.006
I I

104

I 6

I I 8105

I 2

I 4

I 8 106

Weight velocity, C (Ib/hr ft')

Figure 10.12 Forced convection coefficient for water in smooth pipes. (Ref: General Electric
Heat Transfer Data Book, 1975)

10.73 BASIC FORCED-LIQUID FLOW RELATIONS

387

The relation above applies for low viscosity liquids where the Prandtl number (C,plK) is between values of about 1.5 to 20. The weight flow velocity G through pipes and ducts is determined from Eq. 6.12. The symbols for liquids are the same as those for air.
W G =A

(ref. Eq. 6.12)

where W = weight flow of liquid A = cross section flow area Pressure drop relations for liquid flow through smooth pipes and ducts are determined from the same Darcy flow equations for air. When the pressure loss is to be expressed in the height of a column of water, Eq. 6.82 is used. The symbols for liquids are the same as those for air. The equation can be used for laminar or turbulent flow conditions.

HL = 4f

( ) :
NR

(ref. Eq. 6.82)

For laminar flow conditions in round tubes, the Fanning friction factor is used in Eq. 6.82, as shown in Table 6.7. 16 f =(ref. Table 6.7)

Sometimes it is more convenient to write the head loss using the Hagen-Poiseulle friction factor. The head loss relation is then shown by Eq. 10.3. (10.3) For laminar flow in round tubes, the Hagen-Poiseulle friction factor is then used, as shown in Eq. 10.4.*

f = -64
NR

(10.4)

Note that the friction constant in front of both Eq. 6.82 and Eq. 10.3 will have a value of 64. For turbulent flow conditions with Reynolds numbers up to lOO,OOO, the friction factor to be used with Eq. 10.3 is shown in Eq. 10.5 [15].
*Do not become confused with the Darcy equationfor laminarjlow in roundpipes, shown in Eq. 6.82. In the Darcy equation, the Fanning friction factor is used where f = 161NR.

382

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

f=-

0.316
(Nk)0.25

(10.5)

For turbulent flow conditions with Reynolds numbers up to 300,000, the friction factor to be used with Eq. 10.3 is shown in Eq. 10.6.

(10.6)
Ducts with circular cross sections are not always used in liquid cooling systems. Therefore, it is convenient to use a hydraulic diameter for evaluating liquid systems. The hydraulic diameter is defined as shown in Eq. 10.7.

D = H

4 x cross-sectional area wetted perimeter

(10.7)

For a circular cross section, this is simply the pipe diameter.

D = H

4( R D '14) =D RD

pipe diameter
X

For a rectangular cross section with inside dimensions a diameter is

b, the hydraulic

D H ---=-4ab 2a+2b

2ab a + b

Sometimes the hydraulic radius is required for some flow problems. The hydraulic radius is defined in Eq. 10.8.

R = H

cross-sectional area wetted perimeter

(10.8)

For a circular cross section, the hydraulic radius is one fourth of the diameter of the pipe.

RH

--=-

7rDL/4 XD

D 4

(10.9)

10.14 SAMPLE PROBLEM-TRANSISTORS WATER-COOLED COLD PLATE

ON A

A water-cooled cold plate supports 16 stud mounted transistors, which dissipate 37.5 watts each, for a total power dissipation of 600 watts. The coolant flow rate is 1.0 gal/min (62.8 glsec) with an inlet temperature of 95F (35"C), flowing

10.14 SAMPLE PROBLEM-TRANSISTORS

ON A WATER-COOLED COLD PLATE

383

_-

28 stud mounted transistors

Figure 10.73 Water-cooled cold plate with 16 transistors.

through a tube that has a 0.312 in (0.792 cm) inside diameter, as shown in Figure 10.13. The maximum allowable component mounting surface temperature is 160F (71 "C). Determine the component surface temperature when silicone grease is used at the transistor mounting interface. Also determine the pressure drop through the system resulting from a flow rate of 1 gal/min.

SOLUTION: COMPONENT SURFACE TEMPERATURE

The component mounting surface temperature is determined first. This is obtained by calculating the temperature rise along individual segments along the heat flow path from the component to the coolant as follows:
Ar, = temperature rise across the transistor mounting interface from the transistor case to the heat sink surface, using silicone grease at the interface A r2 = temperature rise through the aluminum cold plate from the transistor to the coolant tubing At3 = temperature rise across the liquid coolant convection film from the walls of the tube to the coolant At4 = temperature rise of the coolant as it flows through the cold plate picking up heat from the transistors

The physical properties of water are shown in Table 10.6 for English units and in Table 10.6A for metric units.

384

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

TABLE 10.6

Properties of Water: English Unlts

t
(OF)

c,

CpCL
K

(g)(&)
1.009 1.005 1.002
1 .Ooo

(&)
0.327 0.332 0.338 0.344 0.349 0.355 0.360 0.364 0.368 0.372 0.375 0.378 0.381 0.384 0.386 0.389 0.390 0.392

(dimensionless)

32 40 50 60 70 80 90 100 I10 120 130 140 150 160 170 180

190

200

0.998 0.998 0.997 0.997 0.997 0.997 0.998 0.998 0.999 1.Ooo 1 .001 1.002 1.003 I .004

4.33 3.75 3.17 2.71 2.37 2.08 1.85 1.65 1.49 1.36 1.24 1.14 1.04 0.97 0.90 0.84 0.79 0.74

13.4 11.3 9.4 7.9 6.8 5.8 5.1 4.5 4.0 3.6 3.3 3.0 2.7 2.5 2.3 2.2 2.1 1.9

Starting with A t , the temperature rise across the transistor interface is obtained from Table 3 . 3 , using thermal grease at the mounting interface of the a-28 stud mounted transistor. "C A t , = 0.30 -(37.5 watts) watt
=

11.2"C (20.2"F)

(10.10)

Symmetry is used to determine the temperature rise through the aluminum cold plate from one transistor to the coolant tube, as shown in Figure 10.14. Conversions for English units and metric units are made with Tables 1.1 through 1.11, The heat is conducted through the aluminum plate, so that the conduction heat transfer equation is used.
At - 2-KA

QL

(ref. Eq. 3.2)

10.74 SAMPLE PROBLEM-TRANSISTORS

ON A WATER-COOLED COLD PLATE

385

TABLE 10.6A

Properties of Water: Metric Units

(T)
0 4.4 10.0 15.6 21.1 26.7 32.2 37.8 43.3 48.9 54.4 60.0 65.6 71.1 76.7 82.2 87.8 93.3

(s) (A)
CL

K
cal (sec cm T )
0.00135 0.00137 0.00139 0.00142 0.00144 0.00147 0.00149 0.00150 0.00152 0.00154 0.00155 0.00156 0.00157 0.00158 0.00159 0.00161 0.00161 0.00162

Q !
Y n

(dimensionless)
13.4 11.3 9.4 7.9 6.8 5.8 5.1 4.5 4.0 3.6 3.3 3.0 2.7 2.5 2.3 2.2 2.1 1.9

1.009 1.005 1.002 1.OOo 0.998 0.998 0.997 0.997 0.997 0.997 0.998 0.998 0.999 1.000 1.001 1.002 1.003
1.004

0.0179 0.0155 0.0131 0.01 12 0.0098 0.0086 0.0076 0.0068

0.0061

0.0056 0.005 1 0.0047 0.0043 0.0040 0.0037 0.0035 0.0033 0.003 1

Given Q = 37.5 watts = 128 Btu/hr = 8.96 cal/sec L = 1.0 in = 0.0833 ft = 2.54 cm K = 100 Btu/hr ft "F = 0.413 cal/sec cm "C

In English units:
At2 =

(128)(0.0833 ft) = 15.4"F (lOO)(O.O0694)

(10.11)

In metric units:
At2 =

(8.96)(2.54 cm) = 8.5"C (0.413)(6.45)

(10.1 la)

The forced convection coefficient across the liquid film in the coolant tube cannot be obtained until it has been determined if the liquid flow is laminar or turbulent. Therefore, the Reynolds number must be determined.

386

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

Transistor

37.5 watts

. -

c
1.0 in

_- -_______

Figure 10.14 Heat transfer path from transistor to coolant tube.

GD
NR

cl

(ref. Eq. 6.11)

Given W = 1.0 gallmin = 8.3 lb/min = 62.8 g/sec A = (~/4)(0.312)* 0.0764 in2 = 0.000531 ft2 = 0.493 cm2 = W (8.3 lb/min)(60 min/hr) lb G = - = = 9.38 x 105A 0.000531 ft2 hr ft2 g G = 62.8 g/sec = 127.38 0.493 cm2 sec cm2 D = 0.312 in = 0.026 ft = 0.792 cm p = 1.75 lb/ft hr = 0.00722 g/cm sec [(ref. Table 10.6) at 95F (35"C)I
In English units:
N,q
=

(9.38 x 105)(0.026ft) = 1.39 X lo4 (dimensionless) 1.75

(10.12)

In metric units:
NR

(127.38)(0.792) = 1.39 0.00722

lo4 (dimensionless)

(10.12a)

Since the Reynolds number is well above 3000, the coolant flow is turbulent. The forced convection coefficient for the coolant is determined from Eq. 6.9. The Colburn J factor for the turbulent flow is determined from Eq. 10.2.

Given h, = JC,G

(ref. Eq. 6.9)

NR = 1.39 X'104 (ref. Eq. 10.12)

10.14 SAMPLE PROBLEM-TRANSISTORS

ON A WATER-COOLED COLD PLATE

387

J = o,2 (NR)

0.025 (ref. Eq. 10.2)

-K

"'- [dimensionless; ref. Table 10.6 at 95F (35"C)I 4.8

h, = (0.00371)(0.997)(9.38 x 105)(4.8)-0.666
h,
=

0.025 = 0.00371 (dimensionless) (1.39 x 104)0.2 Cp = 0.997 Btu/lb "F = 0.997 cal/g "C [ref. Table 10.6 at 95F (35"C)I G = 9.38 x lo5 lb/hr ft2 = 127.38 g/sec cm2
J =

Substitute into Eq. 6.9 for English units.

1220

Btu hr ft2"F

(10.13)

Substitute into Eq. 6.9 for metric units.

h,

(0.00371)(0.997)(127.38)(4.8)-0~666

h, = 0.166

cal sec cm20C

(10.13a)

The temperature rise across the liquid coolant film in the tube is determined from Eq. 5.7.
At3 = - (ref. Eq. 5.7) hcA Given Q h, A A A
= 600 watts = 2047.8 Btu/hr = 143.4 cal/sec = 1220 Btu/hr ft2 "F = 0.166 cal/sec cm2 "C
= ?r

DL (inside surface area of 0.312 in diameter tube)

= n(0.312)(16 in length) = 15.7 in2 = 0.109 ft2 = 101.2 cm2

Substitute into Eq. 5.7 for English units.

At -

2047.8 Btu/hr = 15.4"F (1220)(0.109 ft2)

"( 10.14)

Substitute into Eq. 5.7 for metric units.

At3 =

143.4 calhec = 8.5"C (0.166)( 101.2 cm2)

(10.14a)

388

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

The temperature rise of the coolant as it flows through the cold plate is determined from Eq. 6.7. (ref. Eq. 6.7) WCP Given Q = 600 watts = 2047.8 Btu/hr = 143.3 cal/sec W = 1.0 gal/min = 8.3 lb/min = 498 lb/hr = 62.8 g/sec Cp = 0.997 Btu/lb "F = 0.997 cal/g "C (ref. Table 10.6)
At4 = -

Substitute into Eq. 6.7 for English units.

At4 =

2047.8 Btu/hr (498)(0.997)

4.1"F

(10.15)

Substitute into Eq. 6.7 for metric units.

At -

143.4 cal/sec

- (62.8)(0.997)

2.3"C

(10.15a)

The surface temperature of the transistor is determined by adding all of the temperature rises along the heat flow path to the inlet temperature of the coolant. In English units:

+ A t , + At2 + At3 + At4 t, = 95 + 20.2 + 15.4 + 15.4 + 4.1 = 150.1"F

t, = 95

(10.16)

In metric units:
t, = 35

+ 11.2 + 8.5 + 8.5 + 2.3 = 65.5"C

(10.16a)

Since the transistor surface temperature is less than 160F (71"C), the design is satisfactory.

SOLUTION: PRESSURE DROP

The pressure drop through the system is determined from Eq. 10.3. The flow through the cold plate is turbulent, as shown by Eq. 10.12. Therefore, the friction factor shown in Eq. 10.5 is used for the pressure drop calculations.

H, = f

(L, (") D 2g

(ref. Eq. 10.3)

10.14 SAMPLE PROBLEM-TRANSISTORS

ON A WATER-COOLED COLD PLATE

389

Given NR = 1.39 x lo4 (ref. Eq. 10.12) (dimensionless) 0.316 f=(ref. Eq. 10.5)(friction factor) (NR)0.25 0.316 o,25 = 0.0291 (dimensionless) = (1.39 x 10 ) D = 0.312 in = 0.026 ft = 0.792 cm (diameter) g = 32.2 ft/sec2 = 980 cm/sec2 (gravity) W = 8.3 lb/min = 0.138 lb/sec = 62.8 g/sec (fluid flow) p = 62.4 lb/ft3 = 1.0 g/cm3 (density of water)
A =

(0.312)2 = 0.0764 in2 = 0.000531 ft2 = 0.493 cm2 (area) 4 W V = - (velocity of water in pipe)
PA

In English units:
V =

0.138 lb/sec (62.4)(0.000531ft2)

ft 4.16 sec

In metric units:
V =

62.8 g/sec cm = 127.4 (1.0)(0.493 cm2) sec

The length (L) the coolant flow path is obtained from the length of the straight of tubing plus the pipe fittings expressed in terms of equivalent pipe diameters, as shown in Table 10.3. The pipe diameter is 0.312 in (0.792 cm); see Table 10.7. Substitute into Eq. 10.3 for English units.

HL = (0.0291)

(=)

(4.16 ft/sec)2 = 1.15 ft H20 (10.17) 0.312 in (2)(32.2 ft/sec2)

TABLE 10.7 Diameter and Length of Various Fittlngs

Number of Fitting Type Diameters in 6.3 6.2 6.2 180" bend, Length = ~ ( 2 . 0 ) 90" elbow, 2 diameters radius 20 90" elbow, 2 diameters radius 20 Straight pipe, 1+2 6 in 8 8 2 Total equivalent pipe length

Length cm 15.9 15.7 15.7 68.6 115.9

+ + +

27.0 45.7

390

SPECIAL APPLICATIONS FOR TOUGH COOLING JOBS

Substitute into Eq. 10.3 for metric units.

HL = (0.0291)

0
115.9 cm 0.792 cm

(127.4 cm/sec)2 = 35.2 cm H 2 0 (2)(980 cm/sec2)

(10.17a)

The head loss can also be expressed as a pressure loss by comparing the value with the standard atmosphere. This is where a standard atmosphere will support a column of water 34 ft high, so that 14.7 lb/in2 equals 34 ft of water. 1.15 lb A P = -(14.7) = 0.50 3 34 in

(10.18)

For metric units, a standard atmosphere is 76 cm of mercury, which represents about 1034 cm of water. This is then equivalent to a pressure of 1034 g/cm2. 35.2 g A P = -(1034) = 35.2 7 1034 cm
(10.18a)

Figure 10.15 shows a multiple-bay cabinet assembly for naval ships and submarines with engine instruments and an engine throttle control.

Figure 10.15 A mu/tip/e.bay cabinet assembly for naval ships and submarines with engine
instruments and an engine throttle control. (Courtesy of Litton Systems Inc.).

10.15 SOLID STATE THERMOELECTRIC COOLING

391

10.15 SOLID STATE THERMOELECTRIC COOLING

Solid state thermoelectrics are being used for cooling in a wide variety of applications from refrigerators to inertial navigation systems. These devices are available in many different sizes, shapes, operating currents, voltages, and pumping capacities, and they operate on direct current. When direct current is applied, it produces a hot face and a cold face. Electronic components that must be cooled are mounted on the cold side. Heat must be removed from the hot side to prevent its transfer to the cold side of the device. The hot side can be attached to a heat exchanger to improve the heat removal, which improves the performance of the cold side. The system operation can be reversed by simply switching the positive and negative electrical connections, which will switch the hot and cold faces [73]. A single-stage thermoelectric device can achieve a temperature differential of about 65C across the hot and cold faces. This can be increased by stacking one module on the top of another module, which is known as staging or cascading. When two modules are stacked together, they can achieve temperature differences of about 85C across their hot and cold faces. Thermoelectric modules are usually packaged between ceramic plates to provide electrical insulation along with good thermal conduction and a high mechanical strength. Since ceramic is usually very brittle, special care must be used in the mounting or the clamping methods to prevent the ceramic from cracking. A soft interface material, such as indium foil, is often used at the clamping interfaces to distribute the loads more uniformly to reduce high forces and stresses. Sometimes silicone grease is used to reduce the thermal resistance across mounting interfaces, and sometimes a soft RTV adhesive is used for the same reason. Electrical power is required to operate a thermoelectric cooling device. This represents a heat load which must be removed. Any device that is mounted on the thermoelectric cooler will also dissipate heat, which must be removed. Therefore, there are two heat loads that must be considered in the application of a thermoelectric cooler; the electrical input power (or I 2 R), and the heat from the device being cooled. The total heat load is the sum of these two heat sources. There are many cases where the local ambient temperature is too high to provide adequate cooling for temperature-sensitive devices. Refrigeration systems or thermoelectric cooling devices are then required. Where a small size, light weight, low noise, and low cost are important, thermoelectric systems can be used very effectively. Special manufacturing processes and some types of test equipment often require very precise temperature controls. These requirements can often be met with thermoelectric devices.