You are on page 1of 6

THERMAL DESIGN OF A LAPTOP COMPUTER BY A

PERSONAL COMPUTER

Masaru Ishizuka

Toshiba Research & DevelopmentCenter


Toshiba Corporation
1,KomukaiToshiba-cho,Saiwai-ku,Kawasaki210 Japan

ABSTRACT Fluid Nodal Point Network

It has become important to set up an analytical model in order to The flow field in a cabinet, shown in Fig.1, was considered. It
apply computer simulation to the thermal design of electronic was assumed that the cabinet flow fields are obtained by flow
equipment. This paper reports on an approach applied to the thermal visualization or a multi-purpose thermo-fluid analysis code, as
design of a laptop computer. It is a simple analysis using a lumped shown in Fig.2. Next, flow paths were divided by arbitrary nodal
model which includes experimental values, such as the fluid points. These nodal points represent the pressure at nearby points.
resistance coefficient and heat transfer coefficients. Simulation was Several nodal points were selected, and the paths between them
carried out using a personal computer. The proposed method is were linked by flow resistance. The flow resistances were
considered to serve thermal designers satisfactorily. composed of heat dissipating components. Nodal point selection
was free in this study, where a network of lumped parameters for
INTRODUCTION air flow, similar to those for heat flow, was considered, as shown
in Fig.2. The nodal points represent the pressure field. The linked
It has been a very important problem to remove the heat dissipated lines between the nodal points indicate flow paths, including the
by electronic equipment and this problem is related to equipment given flow resistance values. With the pressure difference,AP, air
reliability. This has led to many studies of electronic equipment velocity in the flow path, U, and the air density, p, the fluid
cooling[1,2,3,4,5,6] However, there are not many reports on the resistance coefficient, K, is determined so as to satisfy the
numerical thermal analysis technology using computers for following.
designing practical electronic equipment[7]. It has now become
necessary to apply numerical simulation using a computer to the
thermal design of electronic equipment to save time and speed up
the work. However, many difficult problems including the lack of 0Outlet
computer capacity are left unsolved, so it is necessary to form a ,Fan
simple model for the flow and thermal fields before it becomes
possible to do a general thermo-fluid analysis. In this paper, an Power
approach to the thermal design of electronic equipment is presented. supply
This is a simple analysis using a lumped model and including
experimental values, such as fluid resistance coefficients and heat
transfer coefficients. The approach was applied to the design of a
laptop computer using a personal computer. ,FDD

SIMPLE APPROACH
As mentioned above, a complete thermo-fluid analysis for
electronic equipment design appears impossible today, even using
t t 1
Inlet
a supercomputer. However, a simple approach based on
experimental values such as the fluid resistance coefficient and the
heat transfer coefficient might be considered, using a personal FIGURE 1.Flow field in a cabinet
computer.

CH3096-5/92/0000- 0184 $1 .OO 91992 IEEE 184 1992 InterSocietyConference on Thennal Phenomena
P : Pressure V : Flow rate where, Cp is the specific heat and W, and W2 are the dissipation
T : Temperature 0 : Heat flow Outlet values along the flow path between the nodal points Tl and T2 and
F.P. : Flow path 0 a@@ between the nodal points T2 and T3, respectively. The relationship
Fan is deduced from all the temperature points. The equations are solved
for given inlet and outlet conditions. The air temperature values for
the cooling air are obtained from these results. Then the surface
temperatures are obtained using the thermal network method shown
in FigS. The heat flow Q1 among the total heat generated by the
component are assumed to be exhausted to the air. The heat transfer

p3

O Q 8
Inlet
Vl p2
FIGURE 2. Nodal point network

AP= Kp u2/2

The pressure field network shown in Fig.3, was considered. This FIGURE 3. Pressure field
is one simple example of a network. Each point corresponds to a
different pressure value. There is a certain amount of air flow
between any two points. From mass conservation and from the
definition of the fluid resistance coefficient, the following four w2
equations are obtained,neglecting air density changes.

At the same time, the volume flow rate, V, can be related to the FIGURE 4. Air temperature field
flow velocity, U, using the flow path area, A, as follows,

u=VIA (6)
T
b
The results are a set of non-linear equations for pressure and air
velocity. These equations can be solved to determine the flow field
if the K values are given. When the equations are solved, the flow
distribution and pressure drop are obtained.

Temuerature Nodal Point Network -G


Next, the cooling air temperature field will be considered using
Fig.4. Considering the enthalpy in and out of the nodal point T,,
the enthalpy equation is written as follows, Tl T3

pCp (V1T1+V2T2)+W1+W2'PCp ( v1 +'2lT3 (7) FIGURE 5. Thermal network elements

185 1992 InterSocktyConference on Thermal Phenomena


coefficient, a,can be estimated using the U values from Eq.(8). The For example, RI=l/(aA) from Eq.(8). Figure 6 shows a flow
following expression is obtained, where AT is the temperaturerise, chart which expresses the above mentioned procedure.
W is the heat generated by the component, and the surface area is
A, LAPTOP COMPUTER THERMAL DESIGN

m D COmDUter SaUCnUle Model

Therefore, the surface temperature Ts is obtained as follows. A laptop computer was selected as an example of equipment cooled
by forced air. Figure 7 shows the construction of a laptop
Ts=T+ATs (9) computer.Figure 8 shows an internal view of the computer.

Thus, equation formulation is very simple in the thermal network


method. In Fig.5, the heat flow conservation equation at the T,
point and Ohm’slaw for the electric field between the temprature
points are as follows;

Air inlet

Air

(in this figure, Ql=Qd


FIGURE 7. Laptop computer construction
where R indicates thermal resistance.

Designing Electronic Equipment


1. Casing Configuration
2. PCB Arrangement
3. Generated Heat Distribution
4. Designed Temperature Td
5. Total Flow Rate
6. Fan Selection

Take Flow Path as Heat Flow


~
Predict Flow Pattern Using Flow Visualization
Solve Equations for Air
Temperature Tc

I Obtain Surface Temperature Ts L


I I for Equipment I Give Fluid Resistance Coeff,K and Cross Section
Area A for Each Flow Path
c
Solve Nonlinear Equations for Flow Rate V and
4 Yes Pressure P
I DiSDlW TemDerature Field I +
Obtain Velocity U for Each Flow Path from U +/A
c
Calculate Re Number
c
4 Evaluate Heat Transfer Coeff.~
1

FIGURE 6. Calculation flow chart

186 1992 InterSociety Conference on Thermal Phenomena


Cooling air enters from three vents in the main body, the front inlet this low Reynolds number region, it is difficult to measure air
and the left and right hand side inlets. Heated air is exhausted from velocity and pressure loss values using normal methods. Another
the rear by a fan. In this study .the plasma display is was taken into method of obtaining flow resistance values in natural air convection
consideration, since it is not related to the cooling problem. was considered by Ishizuka,et.al[9,10]. The required K values are
those for individual resistances in the cabinet. rather than the values
Design Prowl- for the entire cabinet. Without such detailed data the flow field
network cannot be drawn.
Figure 9 shows the flow pattem and nodal point network whose
Panem was obtained by flow visualization using white smoke from
Paraffin heated on a very thin wire. Figure 10 shows the calculated
and expenmental results. The straight lines in Fig.10 are As one of the most important data items, the coefficients were
experimental results and simulated results fitted to the experimental obtained for wire nettings at an air intake or outlet. Since the
results. The simulated results were obtained by considering flow
resistances which were not previously determined as parameters.
Flow resistance Values were initially assigned to each flow path
using the basic values from pertinent literature[8.9,10], and were
adjusted to fit the experimental results. Since the temperature at
nodal point 14 was higher than the others in this figure, the heat-
dissipating component arrangement was adjusted to obtain a
uniform temperature distribution. That is, the dissipating
components near nodal point 14 were moved into low temperature
regions. The dotted lines in this figure show the calculation results.
A uniform temperature dismbution was calculated using a personal
computer. The results were used when next version of the laptop
computer was designed. The improvement was be achieved by
small changes in the circuit design.However,more experimental
data are needed to develop this network approach.

W W RESISTANCE FOR LOW REYNOLDS NUMBER

For thermal design, flow resistance values for low Reynolds


number, such as natural air convection, are needed, because a very
slow flow exists even in a forced air convection cooling cabinet. In

Fa n
1 Outlet

Inlet - Inlet
FIGURE 9. Nodal points in a flow field

601 1

Nodal point number


t t t
Inlet

FIGURE 10. Application of a simulation


FIGURE 8. View of laptop computer

187 1992 InterSociety Conference on Thermal Phenomena


pressure drop due to these resistance values is significant to the
entire flow field, the data regarding flow resistance coefficients W
, eri netting
might affect the calculated flow field. A ventilation mode1,shown in t
Fig. 11, is considered. Assuming a uniform temperature distribution
and a one-dimensional steady state flow, the two expressions below
can be written, one for the overall energy balance and the other for
the balance between fluid resistance and buoyancy force.

Q = p Cp A uAT (14)

where, Q is dissipated power, p is air density, Cp is specific heat


of the air at constant pressure, U is air flow velocity, A is the duct FIGURE 11. Ventilation model
cross section area,T is temperature rise, g is acceleration due to
gravity, h is the distance between the wire netting and the heater,
and K, is the flow resistance coefficient for the total system. This relationship has one important feature. The coefficient $ can
Subscript, a , expresses atmosphere. Since the pressure change in be evaluated by determining only the dissipated power and
the system is small, the expression was obtained assuming a perfect temperature rise. In addition, these two quantities are relatively easy
gas . to measure. By taking advantage of this, it is possible to evaluate
the flow resistance coefficient for a wire. netting without measuring
the pressure drop. Next, $ is divided into two parts, KOand K,
where KOis the coefficient for the system without a wire netting.
The K value can easily be obtained by using this relation and by
and the equation for Kt was obtained as follows;
making the same measurements as for K, but without a wire
netting. The net coefficient, K, for the wire netting can be obtained
by simply subtracting K, from K,, as follows,

K=K, -K,

(Re=0.4-95)

0.2

.O

FIGURE 12. Relationshipamong resistance coefficient K,


Reynolds number Re, and porosity coefficient p

188 1992 InterSociety Conference on Thermal Phenomena

~
-
p density,kg/m3

Subscript
Figure 12 shows some of the results obtained. In this figure, the
K data evaluated versus a group of two parameters: Reynolds
a atmosphere
number Re and porosity coefficient ,a,was plotted. If the two
in inlet
variables were grouped in this way, all the data correlated well. j junction
The following expression can be drawn from the best fit to the
o outlet
measured data:
s surface
t total
K=28(Rep2/(1-P)) 0 without a resistance plate

Uncertainty analysis by taking into account the two major REFERENCES


uncertainties in the measurements were conducted. One was for
temperature rise, which was estimated to be 0.5 Kelvin. The other 1. Bar-Cohen, A.,Kraus A.D.and Davidson, S.F., "Thermal
was for heat input, where a 2 96 heat leak was presumed. The Frontiers in the Design and Packaging of Microelectronic
maximum total uncertainty obtained was 13 %. Finally, the Equipment," Mechanical Engineeringgp.53-59(1983).
author's data was compared with data reported by Macphale[lO].
These data fell nearly on the author's line. Efforts are still necessary 2. Kraus,A.D. and Bar-Cohen,A.,Themal Analysis and Control of
to be continued to collect more data with greater reliability. Electronic Equipment, Hemisphere .New York(1983).

CONCLUSION 3. Nakayama,W.," Thermal Management of Electronic Equipment:


A review of technology and research topics,Applied Mechanics
Numerical studies have been implemented to examine an effective Reviews,"vol.39, no.12,pp.1847-1868(1986).
computer simulation method of thermal design for electronic
equipment. In this respect, an approach was applied to the thermal 4. Aung,W.,Cooling Technology for Electronic Equipment,
design of electronic equipment. This approach was a simple Hemisphere New York( 1988).
analysis, using a lumped model including experimental values such
as the fluid resistance coefficient and heat transfer coefficients. This 5. Hwang,G.J.,Transport Phenomena in Thermal Control,
was used for a laptop computer using a personal computer. The Hemisphere, New York( 1989).
proposed method will be a useful tool in the thermal design of
electronic equipment. 6. Peterson,G.P.and Ortega,A.,Thermal Control of Electronic
Equipment and Devices in Advances in Heat Transfer, V01.20,
NOMENCLATURE Academic Press,pp.181-299(19).

A flow path cross area[m21 7. Fukuoka,Y. and Ishizuka,M.,"An Application of the Thermal
CP specific heat[JAcgKl Network Method to the Thermal Analysis of Multi-Chip
D duct diameterlml Packages",Japanese J. of Applied Physics,Vol.28. No.9, pp.1578-
d wire diameter [ml 1585(1989).
K flow resistance coefficient
H duct height[m] 8. Fried,E. and Idelchik,I.E., Flow Resistance: A Design for
P pressure[Pa] Engineers, Hemisphere New York( 1989).
Q heat flow ratem
Re Reynolds number=udh) 9. Ishizuka,M.,Miyazaki,Y. and Sasaki,T.,"Air Resistance
S surface area[m2] Coefficients for Perforated Plates in Free Convection".ASME J.of
T temperam[Kl Heat Transfer.Vol.lO!I,pp.54O-543( 1987).
U velocity[m/sl
10. Ishizuka,M.,"Air flow Resistance of Wire, Nettings in Natural
V flow rate[m3/s1
Convection",ASME J.of Fluids Engineering, Vol. 109 ,pp.389-
W dissipated heat values [wl
393(1987).
AT temperatm increase[a
AP pressure difference[Pa] 11. Macphail,D.C.,"Experiments on Turning Vanes at an
Expansion," Brit, Aero, Res, Coun, Rep. and Memo., No.1876
Greek Symbols (1939).
a heat transfer coefficient[W/m2K]
p porosity coefficient
v kinematic viscosity[m2/s]
189 1992 InterSocietyConference on Thermal Phenomena

~ -

You might also like