You are on page 1of 6

Analytical and numerical analysis of heat transfer from conical spine extended surface

with hexagonal cross-section


Nicolas Aguilera and Juan Cano
Department of Chemical and Environmental Engineering,
Universidad Nacional de Colombia, Bogotá, Colombia
(Dated: July 6, 2020)
Heat transfer was quantified in a conical profile, hexagonal cross-section pin array applied to
the dissipation of heat generated by an Intel R Core R i3 8100 microprocessor. In addition, the
efficiency curve of the spines was estimated, along with the efficiency of the spine array; the latter
was compared with the efficiency of other arrays.

I. MATHEMATICAL MODEL ductivity is the same in all directions, and it re-


mains constant.
A. Statement of the problem
3. The convective heat transfer coefficient on the faces
of the fin is constant and uniform over the entire
Consider a conical spine extended surface with hexag- surface of the fin.
onal cross section of base width δb and length b is pro-
truding from a wall at a constant temperature Tb . This 4. The temperature of the medium surrounding the
extended surface is made of a material of thermal con- fin is uniform (T∞ ).
ductivity constant property.
5. The fin thickness is small, compared with its height
and length, so that temperature gradients across
the fin thickness and heat transfer from the edges
of the fin may be neglected.
6. The temperature at the base of the fin is uniform
(Tb ).
7. There is no contact resistance where the base of the
fin joins the prime surface.
8. There are no heat sources within the fin itself.
9. The heat transferred through the tip of the fin is
negligible compared with the heat leaving its lateral
surface.
FIG. 1: Conical spine with hexagonal cross section.
10. Heat transfer to or from the fin is proportional to
the temperature excess between the fin and the sur-
Furthermore, this extended surface is subjected with a rounding medium.
Dirichlet, Neumann’s and Robin boundary condition of
temperature at its base and tip. In addition, convection
heat transfer to the environment that is maintained at a C. Governing Equations
constant temperature T∞ is imposed along its surface as
shown in figure 1.
Applying the energy balance on the differential element
∆x of the extended surface, figure 1, as follows:
B. Assumptions
Qx+∆x − Qx − Qconv = 0
The analysis of the fin geometry provided in this pa-
per employ the assumptions proposed by Murray (1938) Dividing the equation above by ∆x and applying the
and Gardner (1945). These limiting assumptions, which limit as ∆x approach to zero:
are almost always referred to as the Murray–Gardner as-
sumptions, are:
 
Q |x+∆x − Q |x
lim − h̄Px (T − T∞ ) = 0
∆x→0 ∆x
1. The heat flow in the fin and its temperatures re-
main constant with time.
d
2. The fin material is homogeneous, its thermal con- [Qx ] − h̄Px (T − T∞ ) = 0
dx
2

From Fourier’s Law: δb


Px = 6 x (3)
b
Substituting
√ equations 2 and 3 in equation 1 and di-
 
d dT
k Ax − h̄Px (T − T∞ ) = 0 3 3 δb2
dx dx viding by , yields:
2 b2
o

d2 T dT 2 h̄ 2 3
x2 + 2 x − b x(T − T∞ ) = 0 (4)
d2 T dAx dT h̄ dx2 dx k δb 3
Ax 2 + − Px (T − T∞ ) = 0 (1)
dx dx dx k
applying the change of variable θ = T − T∞ into equa-
Equation 1 is the governing equation for the system tion 4, yields:
with the following boundary conditions:
d2 θ dθ
x2 + 2x − M 2 xθ = 0 (5)
1. at x = 0, T = f inite solution dx2 dx
Where
2. at x = b, T = Tb
√ !1/2
2 3 2
D. Analytical Solution
M= bm
3

To solve equation 1 it is necessary to define the cross and


section area and perimeter of convective area. According
to figure 1, the area Ax is the area of a regular hexagon  1/2
circumscribed about a circumference of radius δx , this 2 h̄
m=
radius varies with x direction. The cross section area is k δb
defined by
Comparing equation 5 with the generalized Bessel
equation
Ax = 3 a δx
 
where a is the apothem, δx is the side of the regular d dθ
xp + a xj + b xk θ = 0

hexagon. The next step is to define δx and the apothem dx dx
in terms of the width. By geometry, the apothem can be
written as The solution to equation 5 has the form

√ h    i
δx 3 θ = xβ/α C1 In ω x1/α + C2 Kn ω x1/α
a= π = 2 δx
2 tan
6 In this case,
According to figure 1 the side of the polygon can be
defined as
p=2 j=1

δb
δx = x a = −M 2 b=k=0
b
Thus, the cross section area is because b = 0, equation 5 is reducible to Bessel’s equa-
tion where evaluation of the constants α, β, ω and n
√ proceeds as follows:
3 3 δb2 2
Ax = x (2)
2 b2
2 2
α= = =2
The perimeter of the convective area is function of the 2−p+j 2−2+1
variable x and is defined as

1−p 1−2
Px = 6 δx β= = = −1
2−p+j 1
3

√ p Z x=b Z x=b
ω = −i α a = i2 −M 2 = 2 M δb
S= Px dx = 6 xdx = 3bδb
x=0 x=0 b

q q The ideal heat flow is obtained from this surface with


2 2 operation at the base temperature excess:
(1 − p) + 4b (1 − 2) + 0
n= = =1
2−p+j 1
The general solution is therefore Qid = h̄(3bδb )θb
√ !1/2
2 3
1  √  √  Then the efficiency with M = 2
bm is
θ = √ C1 I1 2M x + C2 K1 2M x (6) 3
x
√ √
Where C1 and C2 are arbitrary constants to be evalu- k 3 3 δb2 θb M I2 (2M b)
√ √
ated from the boundary conditions in Section I C. How- Qb 2 b I1 (2M b)
ever, it can be noted immediately that in order to have η= =
Qid h̄(3bδb )θb
a finite temperature
√ excess
√ at x = 0, C2 must equal zero
because K1 (2M x) / x is unbounded at x = 0. Hence, or
only C1 need to be evaluated. This evaluation is carried
out at x = b, and when the result is substituted into √ 


equation 6, the particular solution for the temperature k 3 δb M 2 I 2M b
excess is obtained: η= √  √ 
2 h̄ b3/2 b I1 2M b

 1/2 √
b I1 (2M x)
θ(x) = θb  √  (7)
x I1 2M b √ !
(2)30.75 2
√ I2 mb
which reduces, as it should, to θb at x = b. 30.25 2 3
η= √ (9)
The heat flow through the base can be obtained via
!
mb (2)30.75 2
differentiation of equation 7, evaluation at x = b, and I1 mb
substitution into 3



Qb = kA
dx x=b
II. ANALYSIS AND RESULTS
Differentiation of equation 7 goes as follows:
A. Temperature Distribution profile and Heat Map
√ √
dθ θb bM I2 (2M x)
= √ The values of the variables used in this section are
dx x I1 (2M b)
presented in table I, according to the considerations de-
this expression evaluated at x = b, yields: scribed in section I A.

√ TABLE I: Parameter Values.


dθ θb M I2 (2M b)
= √ √
dx x=b b I1 (2M b) Parameter Value Units
√ b 0.03 m
with the area at the base of the spine A(b) = 3 3δb2 /2,
δb 0.002 m
the heat flow through the base is
h̄ 40 Wm−2 K−1
√ √ k 160 Wm−1 K−1
k 3 3 δb2 θb M I2 (2M b) T∞ 16 ◦
C
Qb = √ √ (8) ◦
2 b I1 (2M b) Tb 80 C

The surface area of the conical spine is the integral


of the perimeter function evaluated between the limits The temperature distribution profile can be plotted
x = 0 and x = b: using equation 7. This behavior is shown in figure 2.
4

√ √
2 I2 (2 2mb)
η= √ (12)
(mb)I1 (2 2mb)

4√

√ I1 2mb
3 2 3
η= (13)
4√
 
4
(mb)I0 2mb
3
For equations 11 - 13, m is defined as
FIG. 2: Temperature distribution from the conical spine
1/2
with hexagonal cross section in x direction.

2 h̄
m=
k δb
and the values are taken from table I. In figure 4 values
Heat transfer can be obtained from Fourier’s Law and of η as a function of mb have been plotted.
differentiating equation 7 as follows:

√ √
k 3 3 δb2 θb M I2 (2M x)
Q= x √ (10)
2 b3/2 I1 (2M b)

FIG. 4: Efficiency chart for different pin fin profile


geometry

FIG. 3: Heat Map From figure 4 it can be seen that the fin efficiency is
higher for the geometry adopted. Typically at mb = 1.0
ηf in = 85%, 64%, 60% and 58% for the conical hexagonal
geometry, conical profile, rectangular profile and convex
parabolic profile respectively. The high efficiency of the
B. Efficiency Chart, overall efficiency and pin fin in comparison to the other profile, is due to the
effectiveness
low P/Ac ratio. This is opposite to that for high effec-
tiveness. Since the basic purpose of the installation of
In this section a comparative process was made be- fins on a surface is to enhance the heat transfer rate, the
tween several profile geometries for pin fins. The profiles fin effectiveness is a more meaningful parameter for as-
used were: sessing the usefulness of the fins. The fin effectiveness
can be related to the fin efficiency as follows:
• Constant cross section (rectangular profile)
• Conical Qf in = ηf in h̄ Af θb
• Convex parabolic Effectiveness of a fin must be always greater than 1.
Mathematically, it can be expressed as
Using the data shown in table I, with equation 9 and
the efficiency equations for the profiles. Equations 11, 12
and 13 are respectively the efficiency equations for the ηf in h̄ Af θb
εf in =
profiles presented above: h̄ Ac θb

tanh mb Af 3bδb
η= (11) εf in = ηf in = ηf in √ 2
mb Ac 3 3δb /2
5

2b
εf in = ηf in √ (14)
3δb
It means that the fin effectiveness equals the fin effi-
ciency multiplied by the ratio of fin surface and cross-
sectional areas. In figure 5 is shown how varies the effec-
tiveness in the conical hexagonal base pin fin.

FIG. 6: Top view of rectangular profile spines.

FIG. 5: Effectiveness chart for a conical hexagonal base


pin fin.

III. DESIGN CHOICE FOR THE SPINE

This cross section was chosen for the spines, due


to the ease of covering areas with less material, since, FIG. 7: Top view of conical profile spines.
mathematically, the regular hexagon is the flat figure
that provides the greatest coverage of a surface with
the smallest perimeter. This was demonstrated by the placed. However, in the figure 6, in the conical profile
mathematician Thomas Callister Hales in 1999, solving spines, the faces of the spines are in free contact with
the honeycomb conjecture. the airflow lines, so that more spines can be grouped
together on one surface.
In addition, the conical profile was chosen because, if
a rectangular profile were used, the faces of the spines
It can be said that the choice of these characteristics for
would overlap each other, reducing the convection area
the spines was quite right, because it made the search for
of the fin array, and therefore a spacing between spines
the previously mentioned analytical solution much easier.
would be necessary, while with the conical profile spines,
the target surface could be covered without generating
overlaps.
IV. APPLICATION TO A SPINES
ARRANGEMENT
As can be seen in the figure 5, in the rectangular profile
spines, the airflow lines (arrows with white tips) do To define the efficiency of a hexagonal array of spines,
not have contact with the overlapping faces, so the a fairly simple application was chosen: the Intel R
convection area is reduced. If this problem is scaled Core R i3 8100 processor’s heat sink spine array; the
to an arrangement of spines covering an entire surface, latter has a surface area of 0.00375 × 0.00375m
convection would only occur on some faces of the
outermost spines, as all other spines would be in contact In principle, the heatsink has a base of
with others. In that case, fewer spines would need to be 0.045 × 0.045 × 0.002m; on this base the spines
6

are positioned, side δb = 0.002m, and height b = 0.03m. Finally, the values in the overall efficiency equation are
It is possible to place a total of 175 spines in the area of replaced, giving a value of:
the heatsink base, but the air flow would not circulate
efficiently between the spines, so a total of 131 spines
was chosen. ηG = 0.974 ≈ 0.97

Based on the efficiency curve shown in figure 4, an


approximate value can be estimated. In this case, for the An interesting factor that emerged when calculating
calculation of the mb value, the values shown in Table 1 the overall efficiency of the arrangement with different
were taken, which give a value of mb ≈ 0.47. This value amounts of spines can be noted here. Even if the number
corresponds to an efficiency η ≈ 0.97. of spines was increased or decreased, or the dimensions
of the spine were changed, the overall efficiency was
always very close to the spine efficiency.
With this value the overall efficiency ηG can be
estimated, given by the following equation:

Nspine Aspine V. CONCLUSION


ηG = 1 + (η − 1) (15)
At
From the above results, the following conclusions can
be drawn:
Where Nspine is the number of spines in the array, - The hexagonal area spine and conical profile is quite
Aspine is the surface area of a spine, and At is the total effective, due to the perimeter/convective area ratio,
area, which is given by: which presents a lower value for this geometry, compared
to other more common geometries.
- The implementation of this type of fins in small scale
At = Aspine,t + An heat transfer applications is promising, since it allows
the optimization of heatsink raw material, preserving
their effectiveness.
Where Aspine,t is the surface area of all the spines of -Caution should be taken with the arrangement of
the array and An is the ”naked” area of the base (area spines in certain applications and arrays, since, despite
where there are no spines). Clearing the equation leads the optimization of available space, a large number of
to: spines can generate less efficient convective heat transfer.

At = Nspine (Aspine − Ahx ) + Ab


VI. REFERENCES

Where Ahx is the area of the hexagon of the base of each [1]. Incropera F.P., DeWitt D.P. (1996).Fundamentals
spine and Ab is the base area. By replacing the values of heat transfer. 4th Edition. PRENTICE HALL. Mex-
in the above equation, we get a value of At ≈ 53.79cm2 . ico. p.112

You might also like