Professional Documents
Culture Documents
www.elsevier.com/locate/ijhmt
Abstract
A simple mathematical model following the suggestion of Smithberg and Landis has been created to predict the heat
transfer coecients for the case of a fully developed turbulent ow in a spirally corrugated tube combined with a
twisted tape insert. The heat transfer can be predicted from the combined eects of the axial and the tangential
boundary layer ows coupled with an additional vortex mixing eect near the wall through the solution of the
corresponding momentun and energy transfer equations. The wall roughness has an eect simultaneously on the
axial velocity, secondary uid motion and the resulting swirl mixing. The model reects the inuence of the wall
roughness and the twisted tape on the thermal resistances of the helicoidal core ow, twisting boundary layer ow and
the viscous sublayer near the wall. The calculated heat transfer coecients have been compared to 544 experimental
points obtained from 57 tubes tested. Four hundred thirty-eight points (80.5%) have a relative dierence of less than
15% and 106 points (19.5%) have a relative dierence between (1520)%.
2003 Elsevier Ltd. All rights reserved.
Keywords: Compound heat transfer; Corrugated tubes with twisted tape inserts; Heat transfer coecient; Single-phase turbulent ow
Nomenclature
(b) the transport processes are time independent; The heat transfer coecients Sta;t and Stv can be de-
(c) the turbulent ow is both hydrodynamically and scribed as follows:
thermally stabilized; (i) The heat transfer attributable to axial turbulent
(d) the axial conduction and viscous dissipation in the tube owSta Nua =ReH Pr. This contribution is ob-
uid are neglected. tainable through the evaluation of the thermal resistance
of the axial turbulent tube ow. The latter can be de-
The physical model used in the friction factor anal- composed into two components:
ysis [10] also serves as a basis for the heat transfer pre-
dictions and the complexity of the results is a function of (a) the thermal resistance of the core region, 2e=Di <
the boundary layer details and the Prandtl number ef- g 6 1, Stc1 ,
fects considered. The total heat transfer in the absence of (b) the thermal resistance of the uid nearest to the wall
tape n eects will therefore be due to the additive eects and surrounding the roughness elements in the re-
of the tube wall boundary layer ow and the energy gion 0 6 g 6 2e=Di , including the thermal resistance
interchange caused by vortex mixing. An investigation of the viscous sublayer and buer zone, Stw1 .
of Migai [11] revealed that the thermal resistance dis-
tribution of the uid in the dierent parts of the Having this in mind one can describe the thermal
boundary layer is quite ununiform and has a very strong resistance of the axial tube ow in the form of a Stanton
Prandtl number dependence. For example, for Re 104 number as
and Pr 0:7 the thermal resistances of the laminar
sublayer, buer zone and turbulent core are respectively qcp Um
Sta1 Tw Tm Stw1 Stc1 : 3
32.3%, 52.0% and 15% of the total thermal resistance; qa
for Re 104 and Pr 10:0these quantities are 74.6%,
22.0% and 3.4%, respectively. It is very remarkable that The thermal resistance of the turbulent core, Stc1 , can be
the bigger part of the total thermal resistance for obtained through the value of the Lyons integral [13]
Pr 0:7 is distributed in the buer zone and turbulent Nuc for prescribed value of ReH as
core instead of in the laminar sublayer. On the other
R 2
hand, it is well known that when Re increases the Z g
1
1g
1 gU g dg
thickness of the laminar sublayer gradually decreases Nu1 dg
c 2 4
and it has to be expected that the thermal resistance of 0 1 g 1 PrPri emm
the sublayer will be smaller than the pure axial ow due
to the superposed vortex motion. When wall rough-
ness and vortex uid motion are combined this will where the distributions of U g and em =mg are already
cause additional decrease in the thickness of the laminar available. The turbulent Prandtl number Prt is dened
sublayer. From the above it follows that for gases [14] as
(Pr 0:7) the thermal resistance of the laminar sublayer
B
can be neglected without any substantial error whereas Prt 0:909 ; g 0; 5
for liquids (Pr > 1) it cannot be ignored and has to be 26
taken into account. This idea has been very successfully 1 expgr0 =26
implemented to predict the heat transfer coecients for Prt 0:909 ; 0 < g 6 gw ; 6
1 expgr0 =B
turbulent ow in a smooth pipe with a twisted tape in-
sert [12]. Therefore, the model for the heat transfer Prt 0:909; gw < g 6 1; 7
process has to reect both the eect of the presence of X
5
the wall roughness and twisted tape insert and the B Pr0:5 Ck log10 Prk1 ; 0:02 < Pr < 15; 8
change in the thermal resistance of the viscous sublayer. 1
points [69]. Ninety three points (81.6%) showed a rel- gv to be 1.5gD but no smaller than 0:5e=Di ; gv
ative dierence of less than 10%; for 18 points (15.8%) 1:5gD P 0:5e=Di . Introducing the Stanton number,
this dierence is (1015)% and for the remaining only Z gD
qv 1 gD T TS
three points the dierence is greater than 15%. Stv g dg
qcp Um Tw Tm gD H =Di 0 Tw Tm
It has been assumed that when corrugated tubes are Z gv
combined with twisted tape inserts the thermal resis- 1 T TS
1 g dg
tance of the wall layer calculated from Eq. (11) should H =Di gD Tw Tm
decrease. In this regard, the thermal resistance calcu-
lated from Eq. (10) has been restricted not to exceed that or
one calculated from Eq. (11). Z gD
1 gD T TS
(ii) The heat transfer attributable to axial and tan- Stv g dg
gD H =Di 0 Tw Tm
gential tube wall owsSta;t . To evaluate the contribu-
Z gv
tion due to the axial and tangential tube wall ows a 1 T TS
1 g dg; 17
Reynolds analogy is used [2] assuming that an eective H =Di gD Tw Tm
shear stress sa;t can be dened which is dependant on the
total velocity vector at the edge of the buer zone where
V. Zimparov / International Journal of Heat and Mass Transfer 47 (2004) 385393 389
p
f =2 The model discussed here comprises the following
Tw Tm ; cases concerning the calculation of the heat transfer
St
T TS Tw TS Tw T DTS DT : coecient for a fully developed turbulent ow in a pipe:
The temperature dierences DTS and DT can be ob- (a) Dg 0; H =Di 0 and Stw1 0the model calcu-
tained from integration of the usual heat ux relation lates the friction factor and heat transfer coecient
dT for a smooth pipe [16];
q a eH qcp ; 18 (b) Dg > 0, H =Di 0 and Stw1 > 0the model calcu-
dy
lates the friction factor and heat transfer coecient
in dimensionless form for a corrugated pipe [15];
(c) Dg 0, H =Di > 0 and Stw1 > 0the model calculates
q 1 PrPrt emm oT the friction factor and heat transfer coecient for a
: 19 smooth pipe combined with a twisted tape insert [12];
qw r0 Pr og
(d) Dg > 0, H =Di > 0 and Stw1 > 0the model calcu-
Thus, lates the friction factor and heat transfer coecient
Z gv for a corrugated tube combined with a twisted tape
q=qw
DTS Tw TS dg; 20 insert.
1 1 em
0
r0 Pr
Pr1t m
Z g
q=qw
DT dg: 21 3. Results and discussion
1 1 em
0
r0 Pr
Pr1t m
The heat transfer coecients (transformed as Stan-
Using the linear distribution for q=qw , Eq. (17) yields ton numbers) have been calculated using Eqs. (2)(22)
0 and compared with 544 experimental points obtained
Z
1 gD St @ g2D gv 1 g from 57 corrugated tubes combined with twisted tape
Stv p dg
gD H =Di f =2 2 0 1 1 1 em inserts. The values of the geometrical parameters of the
r0 Pr Pr m
0 1 t1 corrugated tubes are presented in Table 1. Two hundred
Z gD Z gv
@ 1 g sixty-eight points (49.3%) have a relative dierence of
dgAg dgA
0 0 1 1
1 em less than 10%, for 170 points this dierence is 10
Pr Pr m
r0
0 t
15%, and for the remaining 106 points the dierence is
1520%. The numerical results for the heat transfer
St 1
p @gv gD 1 gv gD coecients (transformed as Nusselt numbers) together
H =Di f =2 2
with the experimental data of the tubes studied (corru-
Z gv
1 g gated tubes combined with twisted tape inserts) are
dg
0 1 1
Pr1t emm presented in Figs. 213. Taking into account the ex-
r0 Pr
0 1 1 perimental error in the measurements, this agreement
Z gv Z g should be considered as fairly acceptable. Because of the
@ 1 g
dgA1 gdgA: lack of other experimental data of this kind (for corru-
1 1 1 em
gD 0
r Pr Prt m
0 gated tubes with twisted tape inserts) an attempt to
22 verify the model has been made with the study of Bergles
Table 1
Values of the characteristic parameters of the corrugated tubes
Tube Di (mm) e (mm) p (mm) b () e=Di p=e b
3010 13.90 0.312 5.76 82.4 0.0224 18.46 0.916
3020 12.44 0.515 4.48 83.4 0.0414 8.70 0.927
3040 13.39 0.497 5.77 82.2 0.0371 11.61 0.913
3050 13.15 0.593 5.06 83.0 0.0451 8.53 0.922
3070 13.66 0.622 8.12 79.3 0.0456 13.05 0.881
4020 13.53 0.507 4.55 72.2 0.0375 8.97 0.802
4030 13.73 0.781 5.82 68.0 0.0569 7.45 0.755
4040 13.68 0.557 5.97 67.4 0.0407 10.73 0.749
4050 13.38 0.581 5.08 70.1 0.0434 8.74 0.779
2010 13.68 0.315 6.67 90.0 0.0230 21.17 1.000
2040 13.65 0.440 6.01 90.0 0.0322 13.66 1.000
2070 13.59 0.464 8.55 90.0 0.0341 18.43 1.000
390 V. Zimparov / International Journal of Heat and Mass Transfer 47 (2004) 385393
et al. [17] where heat transfer coecients for straight and predict the heat transfer coecient in a rough tube
swirl ow in rough tubes with twisted tape inserts were (without twisted tape insert). In this case, the g-function
reported, Fig. 14. The authors suggested an equivalent has been determined through the experimental data of
sand grain roughness of eS =Di 0:0050.015. Giving [17] in the form
dierent values of eS =Di for dierent Reynolds numbers ge 0:246
Pr0:6 :
S ; Pr 5:9668eS 23
the friction factors have been calculated through the
model discussed in [10]. Since the roughness was sug- The calculated values of the heat transfer coecient
gested as one-dimensional (equivalent sand grain for the rough tube acting alone are very close to the
roughness) the model presented in [18] has been used to experimental data (relative dierence of less than 5%)
V. Zimparov / International Journal of Heat and Mass Transfer 47 (2004) 385393 391
Fig. 11. Nusselt number vs. Reynolds number. Comparison Fig. 13. Nusselt number vs. Reynolds number. Comparison
between computed and experimental results (2010 H =Di 0; between computed and experimental results (2070 H =Di 0;
2011 H =Di 15:35; 2012 H =Di 12:28; 2013 H =Di 2071 H =Di 15:45; 2072 H =Di 12:36; 2073 H =Di
7:75; 2014 H =Di 5:85). 7:80; 2074 H=Di 5:89).
(2) The model reects the inuence of the wall rough- ence is (1520)%. To verify the validity of the
ness and the twisted tape on the thermal resistances model for another type wall roughness the calcu-
of the helicoidal core ow, twisting boundary layer lated heat transfer coecients for turbulent ow in
ow and the viscous sublayer near the wall. rough tubes (sand grain roughness) with twisted tape
(3) The calculated heat transfer coecients have been inserts have been compared with the experimental
compared with 544 experimental points obtained data of Bergles et al. [17]. The agreement between
from 57 tubes tested. Four hundred thirty-eight predicted and experimental data is fairly good.
points (80.5%) have a relative dierence of less than (4) The model comprises the possibilities to predict the
15% and for the remaining 106 points this dier- heat transfer coecients for the cases of a turbulent
V. Zimparov / International Journal of Heat and Mass Transfer 47 (2004) 385393 393