Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
CITATIONS
DOWNLOADS
VIEWS
36
907
354
1 AUTHOR:
Esmail M. A. Mokheimer
King Fahd University of Petroleum and
66 PUBLICATIONS 361 CITATIONS
SEE PROFILE
Mechanical Engineering Department, King Fahd University of Petroleum and Minerals, P.O. Box: 279, Dhahran 31261, Saudi Arabia
Received 13 June 2001; received in revised form 18 January 2002
Abstract
Performance of annular ns of dierent proles subject to locally variable heat transfer coecient is investigated in
this paper. The performance of the n expressed in terms of n eciency as a function of the ambient and n geometry
parameters has been presented in the literature in the form of curves known as the n-eciency curves for dierent
types of ns. These curves, that are essential in any heat transfer textbook, have been obtained based on constant
convection heat transfer coecient. However, for cases in which the heat transfer from the n is dominated by natural
convection, the analysis of n performance based on locally variable heat transfer coecient would be of primer importance. The local heat transfer coecient as a function of the local temperature has been obtained using the available
correlations of natural convection for plates. Results have been obtained and presented in a series of n-eciency
curves for annular ns of rectangular, constant heat ow area, triangular, concave parabolic and convex parabolic
proles for a wide range of radius ratios and the dimensionless parameter m based on the locally variable heat transfer
coecient. The deviation between the n eciency calculated based on constant heat transfer coecient, reported in the
literature, and that presently calculated based on variable heat transfer coecient, has been estimated and presented for
all n proles with dierent radius ratios. 2002 Elsevier Science Ltd. All rights reserved.
1. Introduction
Annular ns nd numerous applications in compact
heat exchangers, in specialized installations of singleand double-pipe heat exchangers, in electrical apparatus
in which generated heat must be eciently dissipated, on
cylinders of air cooled internal-combustion engines, etc.
In a conventional heat exchanger heat is transferred
from one uid to another through a metallic wall. The
rate of heat transfer is directly proportional to the extent
of the wall surface, the heat transfer coecient and to
the temperature dierence between one uid and the
adjacent surface. If thin strips (ns) of metals are attached to the basic surface, extending into one uid, the
total surface for heat transfer is thereby increased. It
might be expected that the rate of heat transfer per unit
of the base surface area would increase in direct proportion. However, the average surface temperature of
these strips (ns), by virtue of temperature gradient
through them, tends to decrease approaching the temperature of the surrounding uid so the eective temperature dierence is decreased and the net increase of
heat transfer would not be in direct proportion to the
increase of the surface area and may be considerably less
than that would be anticipated on the basis of the increase of surface area alone. The use of ns in one side
of a wall separating two heat-exchanging uids is exploited most if the ns are attached to or made an integral part of that face on which the thermal resistivity
is greatest. In such a case the ns serve the purpose of
articially increasing the surface transmittance.
The ratio of the actual heat transfer from the n
surface to that, that would transfer if the whole n surface were at the same temperature as the base is commonly called as the n eciency. Harper and Brown [1],
in connection with air-cooled aircraft engines, investigated straight ns of constant thickness, wedge-shaped
0017-9310/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 1 7 - 9 3 1 0 ( 0 2 ) 0 0 0 7 8 - 9
3632
E.M.A. Mokheimer / International Journal of Heat and Mass Transfer 45 (2002) 36313642
Nomenclature
A
g
h
i
ks
kf
L
m
Nu
N
P
Ra
R
r
Greek symbols
a
thermal diusivity m2 =s
b
volumetric coecient of thermal expansion
K1
h
dimensionless temperature, h T Tf =
Tb Tf
m
kinematic viscosity of the ambient uid
m2 =s
Subscripts
b
condition or variable specied at the base of
the n
f
ambient uid property
r
local value of the variable at r
o
condition or variable specied at the edge of
the n
s
solid, n material property
u
for upper surface
l
for lower surface
E.M.A. Mokheimer / International Journal of Heat and Mass Transfer 45 (2002) 36313642
3633
5.
6.
7.
8.
9.
10.
3634
E.M.A. Mokheimer / International Journal of Heat and Mass Transfer 45 (2002) 36313642
Fig. 1. (a) General annular n prole. (b) Comparison of the solution based on Eq. (3), for ds 6 dr and solution based on Eq. (4),
- - - - - - for ds dr.
where n is the prole index; n 0 represents the constant thickness n which has a rectangular prole.
n 1=2 corresponds to the convex parabolic n prole
while n 1 describes the triangular n prole with
straight surfaces. The value of n 2 gives the concave
parabolic prole. All the n proles considered in the
present study start with a thickness yb at the base.
The triangular, convex parabolic and concave parabolic
proles have tips at their ends (i.e, y 0 at r ro ) while
the rectangular has a constant thickness along the n.
The annular ns with constant area for heat ow have a
hyperbolic prole. For such a prole, the thickness of
the n varies with the radius such that y r constant,
and the prole can be expressed as
Rb
yr yb
R
the hyperbolic n has a sharp edge at innity, but in
practice, it is cut o at a distance ro from the axis of
symmetry. The general partial dierential equation governing the steady heat transfer from all ns can be
written as
d
dT
ks A r
dr As hu hl r T T1 0;
dr
dr
where ks is the n material thermal conductivity, Ar
2pryr is the cross-sectional area perpendicular to the heat
ow, and As is the local surface area at that section,
As 2pr ds for annular ns. The above equation can be
written for an annular n in polar coordinates as
d
dT
ks 2pryr
dr 2pr dshu hl r T T1 0;
dr
dr
where ds is the arc length and the subscripts u and l
mean the upper and lower surfaces for the case of horizontal ns and the subscript r means the local value at
radius r. This equation can be rewritten as
d
dT
ds hu hl r
T T1 0;
ryr
r
ks
dr
dr
dr
which can be rearranged and written as
yr
d2 T
dr2
0:
yr dyr
r
dr
dT
dr
hu hl r
ks
ds
T T1
dr
E.M.A. Mokheimer / International Journal of Heat and Mass Transfer 45 (2002) 36313642
T T1
dr2
r yr dr dr
dr
ks yr
0:
ds
Nur kf
;
r
104 6 Ra 6 107 ;
105 6 Ra 6 1010 ;
where
gbhr3
:
Rar
ma
The governing equation, Eq. (1), can be rewritten for a
general prole with index n in a dimensionless form as
follows:
d2 h
1
n
dh
m2
ds
h
n
dR2
R Ro R dR
Ro R dr
0;
where
s
hu hl
mL
:
k s yb
The arc length, ds, in the above equation can be calculated approximately from the following equation:
dyr
2
2
!1=2
dr2
dr
dyr
2dr
3635
2
!1=2
1
hence
ds
dr
dyr
2dr
!1=2
2
1
dR2
R Ro R dR
1=2
2
m2
nyb
n1
R
R
1
h
o
Ro Rn
2L
0:
n
dR2
R Ro R dR
Ro R
For annular n with hyperbolic prole the governing
equation will be:
2
d2 h
1 Rb dh
m
h 0:
5
dR2
R R2 dR
Rb =R
These Eqs. (4) and (5), will be solved for thermal
boundary conditions of having the base kept at constant and uniform temperature and the n tip is kept
thermally insulated. The above non-linear ordinary differential equations have been converted to algebraic
equations using the nite dierence techniques. The nal
nite dierence form of Eqs. (4) and (5) can be written as
follows:
hi1 hi1 1
hi1 hi1
n
Ri R0 R
2DR
DR2
i
hi
6
2
2
RomRi n
DR2
3636
E.M.A. Mokheimer / International Journal of Heat and Mass Transfer 45 (2002) 36313642
hi1 hi1 1 Rb
hi1 hi1
Ri R2
2DR
DR2
i
hi
:
7
2
m2
Rb =Ri
DR2
Subject to boundary conditions: at R Rb ; h 1:0 and
at R Ro , dh=dR 0 .
3. Results and discussions
The dimensionless forms of the governing equations
(4) for variable prole ns and (5) for the constant area
prole n include two dimensionless controlling parameters in addition to the index n which denes the n
prole. These two parameters are namely the dimensionless variable m and the n radius ratio which is
implicitly inherited in evaluating the value of Ro in
Eq. (4), Ro ro =ro rb ro =rb =ro =rb 1 and the
value of Rb in Eq. (5), Rb rb =ro rb 1=ro =rb 1,
where ro and rb are the outer and base radii of the n,
respectively. So, the n performance will be expressed
in the form of curves that give the n eciency as a
function of these two dimensionless controlling parameters m and ro =rb .
The nite dierence equations presented have been
tested for the eect of mesh size on the accuracy of the
solution. The numerical solution for an annular n with
concave parabolic prole and radius ratio 2 has been
obtained via numerical meshes of 5, 10, 15 and 20 grid
points. The numerical solution for this case showed independence on the grid size for mesh with grid points of
15 and above. The dierence between the n eciency
that is obtained numerically via a grid of 15 points with
respect to that obtained via a grid of 20 points was
0.012%. So, a grid of 15 points has been adopted
through out the work.
The eect of approximating the incremental arc
length to the incremental length on the radial direction
has been investigated for the concave n prole since it
has a surface prole of a large slope. The solution has
been obtained for a concave n with radius ratio 2 using
Eq. (3) that takes the incremental arc length on the solution and Eq. (4) that takes the approximated radial
increment instead of the incremental arc length. The
comparison of the two solutions is presented in Fig. 1(b)
for the above particular case. Fig. 1(b) shows that this
approximation ds dr has almost no eect on the
accuracy of the solution. So, Eq. (4) has been used to get
the solutions for all cases considered and Eq. (5) has
been used to obtain the solution for the annular n with
constant area of heat ow. Moreover, the present numerical scheme, the solution algorithm and the solution
computer code have been rst bench marked via pro-
E.M.A. Mokheimer / International Journal of Heat and Mass Transfer 45 (2002) 36313642
3637
Fig. 2. Fin eciency with dimensionless parameter m for annular n with rectangular prole with variable heat transfer coecient,
- - - - - - analytical solution for ratio 1, constant heat transfer coecient.
3638
E.M.A. Mokheimer / International Journal of Heat and Mass Transfer 45 (2002) 36313642
Fig. 3. Fin eciency with dimensionless parameter m for annular n with constant heat ow area prole with variable heat transfer
coecient, - - - - - - analytical solution for ratio 1, constant heat transfer coecient.
Fig. 4. Fin eciency with dimensionless parameter m for annular n with triangular prole with variable heat transfer coecient.
E.M.A. Mokheimer / International Journal of Heat and Mass Transfer 45 (2002) 36313642
3639
Fig. 5. Fin eciency with dimensionless parameter m for annular n with concave parabolic prole with variable heat transfer coecient.
Fig. 6. Fin eciency with dimensionless parameter m for annular n with convex prole with variable heat transfer coecient.
3640
E.M.A. Mokheimer / International Journal of Heat and Mass Transfer 45 (2002) 36313642
Table 1
Comparison of the n eciency for annular ns with rectangular prole
Prole
Gardner [8]
Radius
ratio
1
1.5
2
3
4
5
m1
Radius
ratio
1
1.5
2
3
4
5
Radius
ratio
1
1.5
2
3
4
5
Radius
ratio
1
1.5
2
3
4
5
Radius
ratio
1
1.5
2
3
4
5
a
0.7615
0.7231a
0.6920
0.6420
0.6105
0.5846a
Present
0.7792
0.7484
0.7243
0.6883
0.6622
0.6419
Dierence (%)
Prole
Gardner [8]
m1
2.274
3.384
4.460
6.731
7.802
8.922
Radius
ratio
1
1.5
2
3
4
5
7.130
9.365
12.069
17.319
16.119
18.564
Radius
ratio
1
1.5
2
3
4
5
12.585
16.450
18.263
22.129
23.985
23.258
Radius
ratio
1
1.5
2
3
4
5
18.008
20.869
24.633
27.631
31.932
30.502
Radius
ratio
1
1.5
2
3
4
5
23.719
27.698
32.265
35.469
39.167
38.416
Radius
ratio
1
1.5
2
3
4
5
m2
0.4820
0.4308a
0.3915
0.3320
0.3115
0.2846a
0.5190
0.4753
0.4452
0.4015
0.3714
0.3495
m3
0.3310
0.2846a
0.2560
0.2142
0.1895
0.1769a
0.3787
0.3406
0.3132
0.2751
0.2493
0.2305
m4
0.2498
0.2154a
0.1873
0.1560
0.1316
0.1231a
0.3050
0.2722
0.2485
0.2156
0.1934
0.1771
m5
0.2000
0.1693a
0.1445
0.1189
0.1000
0.0923a
0.2622
0.2341
0.2133
0.1843
0.1644
0.1499
Table 2
Comparison of the n eciency for annular ns with constant
area prole
Ullmann and Kalman [9] and many heat transfer textbooks with the exception that Gardner [8] considered
radius ratio range of 14 only. It is worth mentioning
here that radius ratio of 1 represents cases when the
radius of curvature of the annular n approaches innity
which is practically the straight n. So, the results for all
n proles for radius ratio of 1 are practically the results
of a straight n of the pertinent prole. For the two
cases of annular n with rectangular prole and constant heat ow area prole with radius ratio 1, the
0.7615
0.7000a
0.6526
0.5684
0.5053
0.4492a
Present
Dierence (%)
0.7792
0.7256
0.6743
0.5846
0.5124
0.4547
2.274
3.522
3.214
2.766
1.393
1.218
0.5190
0.4568
0.4070
0.3353
0.2865
0.2513
7.130
10.408
10.776
12.110
13.646
14.308
0.3787
0.3304
0.2944
0.2445
0.2115
0.1880
12.585
16.186
17.752
21.219
23.862
26.331
0.3050
0.2675
0.2403
0.2030
0.1782
0.1604
18.008
21.776
27.729
31.020
36.816
37.651
0.2622
0.2324
0.2108
0.1811
0.1613
0.1469
23.719
29.168
37.585
38.984
44.528
49.220
m2
0.4820
0.4092a
0.3632
0.2947
0.2474
0.2154a
m3
0.3310
0.2769a
0.2421
0.1926
0.1611
0.1385a
m4
0.2498
0.2092a
0.1737
0.1400
0.1126
0.1000a
m5
0.2000
0.1646a
0.1316
0.1105
0.0895
0.0746a
results are typically those for a straight n with rectangular prole. So, the analytical solution for these two
cases is the same.
Moreover the n eciency calculated using constant
heat transfer coecient along the n (as given by
Gardner [8], Ullmann and Kalman [9] and most of the
heat transfer textbooks) has been compared with the
eciency calculated through the present work based on
the variable heat transfer coecient along the n as
function of the temperature, for selected values of the
E.M.A. Mokheimer / International Journal of Heat and Mass Transfer 45 (2002) 36313642
Table 3
Comparison of the n eciency for annular ns with triangular
prole
Prole
Ullmann and
Kalman [9]
Radius
ratio
1
1.5
2
3
4
5
m1
Radius
ratio
1
1.5
2
3
4
5
Radius
ratio
1
1.5
2
3
4
5
Radius
ratio
1
1.5
2
3
4
5
Radius
ratio
1
1.5
2
3
4
5
0.6615
0.6230
0.5769
0.5138
Present
0.7236
0.6900
0.6647
0.6279
0.6019
0.5823
Dierence
(%)
0.3769
0.3462
0.3000
0.2500
0.4753
0.4349
0.4055
0.3647
0.3367
0.3161
Ullmann and
Kalman [9]
m1
4.131
6.274
8.119
11.761
Radius
ratio
1
1.5
2
3
4
5
13.336
14.634
17.740
20.908
Radius
ratio
1
1.5
2
3
4
5
18.942
19.644
26.213
29.550
Radius
ratio
1
1.5
2
3
4
5
27.763
33.743
31.826
34.824
Radius
ratio
1
1.5
2
3
4
5
31.074
31.902
38.097
44.474
Radius
ratio
1
1.5
2
3
4
5
m3
0.2615
0.2385
0.1923
0.1538
0.3581
0.3227
0.2968
0.2606
0.2362
0.2183
m4
0.1923
0.1615
0.1446
0.1138
0.2965
0.2662
0.2438
0.2121
0.1905
0.1746
m5
0.1615
0.1462
0.1154
0.0846
0.2607
0.2344
0.2146
0.1864
0.1669
0.1524
Table 4
Comparison of the n eciency for annular ns with concave
parabolic prole
Prole
m2
3641
0. 5692
0.5431
0.4923
0. 4385
Present
Dierence
(%)
0.6537
0.6176
0.5909
0.5530
0.5271
0.5079
7.832
8.096
10.972
13.665
0.4430
0.4049
0.3768
0.3373
0.3108
0.2913
10.707
16.298
20.185
23.421
0.3468
0.3139
0.2894
0.2545
0.2310
0.2135
19.147
25.587
29.347
31.543
0.2943
0.2663
0.2451
0.2146
0.1934
0.1776
27.218
32.210
39.065
41.099
0.2630
0.2389
0.2203
0.1930
0.1613
0.1592
35.609
40.635
40.229
46.869
m2
0. 3652
0.3154
0.2692
0. 2231
m3
0. 2538
0.2154
0.1800
0. 1462
m4
0. 1938
0.1662
0.1308
0. 1046
m5
0. 1538
0.1308
0.1154
0. 0846
3642
E.M.A. Mokheimer / International Journal of Heat and Mass Transfer 45 (2002) 36313642
4. Conclusion
Heat transfer from annular ns subject to locally
variable heat transfer coecient has been studied. The
local heat transfer coecient as a function of the local
temperature has been obtained using the available correlations of natural convection for plates. The results
showed that the assumption of constant heat transfer
coecient along the n in such cases leads to a signicant underestimation of the n eciency. The deviation
between the n eciency calculated based on constant
heat transfer coecient and that calculated based on
variable heat transfer coecient increases with both the
dimensionless parameter m and the radius ratio of the
n. The use of the present results by the designers of heat
transfer equipment that involve annular ns subject to
natural convection heat transfer mode would result in a
considerable reduction in the extended surface area and
hence a signicant reduction in the weight and size of the
heat transfer equipment.
Acknowledgements
The author would like to extend his thanks to King
Fahd University of Petroleum and Minerals for the
support of this article as well as Prof. H.Z. Barakat due
to his valuable discussions during this work.
References
[1] D.R. Harper, W.B. Brown, Mathematical Equations for
Heat Conduction in the Fins of Air-Cooled Engines,
National Advisory Committee for Aeronautics, Report
no. 158, 1922.
[2] E. Schmidt, Die Warmeubertrgung durch Rippen, Zeit.
V.D.I. 70 (1926) 885889, see pp. 947951.
[3] R. Focke, Die Nadel als Kuhelemente, Forsch. Geb.
Ingenieurw. 13 (1942) 3442.