Professional Documents
Culture Documents
110 / Vol. 121, FEBRUARY 1999 Copyright © 1999 by ASME Transactions of the ASME
Nomenclature
Ac --= channel flow cross section area, P\,P2 == Reynolds number exponents X. = wavelength of surface corrugation,
m2 A P == pressure drop, N/m 2 Fig. 1(b), mm
A-- = actual or effective heat transfer Pr == Prandtl number, =\xcplk fi = dynamic viscosity, kg/m-s
surface area, m2 Q--= heat load, W p = fluid density, kg/m 3
b--= corrugation depth, mm Re == Reynolds number, =mdJNAcjj, 4> = ratio of effective to projected
C- = heat capacity rate, =mcp, W/K t == plate thickness, mm surface area of plate corrugations
c„-= specific heat, J/kg-K Arlm == log mean temperature differ-
ence, Eq. (2), K Subscripts
d„- = equivalent diameter, =2b, mm 2
E--= frictional pumping power, W/m T--= temperature, K b = at bulk fluid temperature
f --= Fanning friction factor, Eq. (11) U--= overall heat transfer coefficient, c = cooling water stream
h~-= heat transfer coefficient, W/m 2 - Eq. (6),W/m 2 -K ch = pertaining to chevron plate
K h = hot process fluid stream
k--= thermal conductivity, W/m-K Greek Symbols i= inlet fluid condition
L- = plate length betweerupoFts, m P == corrugation inclination angle o = outlet fluid condition
m -= mass flow-rateTEg/s , i relative to flow direction pp = pertaining to flat parallel-plate
N--= number of interplate channels 7 z= corrugation profile aspect ratio, channel
Nu == Nusselt number, =hdjk =2bl\ w = at plate wall conditions
NTU == number of transfer units, Eq. (5) £ == thermal effectiveness, Eq. (4)
PROCESS PROCESS
LIQUID LIQUID
ELECTRIC
POWER
INPUT
-0
MOYNO PUMP
ROTAMETER
COOLING WATER IN
c&o-
TO
CONOENSATE
MEASUREMENT
kW/K. Furthermore, even though only 12 to 24 plates were the plate. However, in order to calculate hh from Eq. (7), the
used in these experiments, the influence of small number of cooling water side heat transfer coefficient hc is needed. This
thermal plates on NTU, as suggested by Shah and Focke (1988) was determined from calibration equations that were devised
was insignificant (less than 0.5 percent). This was so because on the basis of extended experiments employing the modified
C*, (e- C*), and (C*-NTU) values were relatively small in Wilson plot technique, as described in the next subsection. The
all data sets; the plate effects are significant only when the average wall temperature of the plate was determined iteratively
converse is true (Manglik, 1999). The overall heat transfer from the heat and resistance balance
coefficient was thus determined from
g = hhA(T„ - Tw)„ (kwA/t)(Tw,h - Tw,c)
t/=(C m i n NTU)/A (6)
= hcA{Tw - Tb)c. (8)
and the heat transfer coefficient for the process fluid stream
calculated from the thermal resistance summation From the measured overall pressure drop across the PHE in
the process stream, the core pressure drop can be obtained from
(IMA*) = {IIUA) - (t/Akms) - (1/Ahc). (7)
Note that the heat transfer area A is the same on both sides of A/?c, Apn, A/7port - Appi (9)
Here, based on an empirical equation (Shah and Focke, 1988;
Kays and London, 1984; Kays, 1950) and the mean port veloc-
3.0 1 ' ' 1 i i ' i i ity, port losses were estimated by
Chevron plates (304 SS) t/e - 5.08 mm
A p= 30° L = 0.392 m Appon= \.5(pV2poJ2). (10)
O /?= 60° AAi^ A 1 = 0.6 mm
2.5 ^ ^ w = 0.163 m The pressure loss in small pipe sections at the inlet and outlet
^ <i=1.29 of the PHE (Appipl!) was estimated on the basis of smooth tube
/A
°\ y= 0.556 friction factor and pipe flow velocity. Thus, for the shear loss
/A
M, 2.0 OA within the corrugated passages, the Fanning friction factor was
calculated as
°\
h. ©A /= pdeApcoJ[2L(m/NAc)2 (11)
•a 1.5 U oA
These measurements were made for isothermal conditions, with
r oA fluid properties calculated at the mean flow temperature.
r oA
1.0 oA Modified Wilson Plot. In order to determine the process
oA. fluid heat transfer coefficient from Eq. (7), the cooling water
oA resistance (l/hcA) needs to be known. For predicting hc, cali-
0.5
' i i ft = 1.27[l+sin{^(l-2.25V4.5}]
oA,
oA.
oA
bration equations were established for each of the three chevron
plate arrangements from data obtained by the modified Wilson
plot technique (Shah, 1990; Manglik and Bergles, 1991). This
modification of the classical Wilson plot method (Wilson,
,A^° 1 1 1 , 1 , 1 I . I . I °JV,
0.0
0 1 2 3 4 5 6 7 8 9 1915) essentially incorporates and accounts for variable fluid
Corrugation wavelength, X [mm] property effects. Very briefly, on the basis of a power-law type
relationship between Nu and Re, Pr, and (fjjfiy,), the resistance
Fig. 3 Geometrical characteristics of plate surface corrugations balance of Eq. (7) can be expressed as
Nu
Re < 2000, (L/de) = 52.17
(13)
p. 0.023 Re 0 8 Pr1'3(fi/fj,w)0M
Re > 4000.
The heat transfer enhancement, with two to five times higher
Nu c = 0.135Re c °- 7 Pr" 3 (^/^„ Nu than the flat-plate channel, is clearly evident. Higher Nusselt
Re c >1000 numbers are obtained with increasing (5, which reflects the in-
creased intensity of swirl flows generated by the larger 0 chev-
ron plates. In Fig. 5, the data are also compared with the results
of Okada et al. (1972), Focke et al. (1985), and Thonon et al.
0 1 2 3 (1995). The present data are seen to be in good agreement with
07 3
the results of Okada et al. for all three /3 arrangements. However,
[Reoj p r l/3 (^^fi^Ak/d,)],, I [Re ' Pr" (p/fiy,)™' (Aklde)\c
the results of Focke et al., that are based on mass transfer
Fig. 4 Typical modified Wilson plot results for the calibration of cooling
experiments with simulated scaled-up plate surfaces, are more
water side heat transfer coefficient for p = 30-deg chevron plate than two times higher for all cases. The Thonon et al. data, on
the other hand, are in fair agreement for /3 = 30 deg, but those
for (3 = 45-deg and 60-deg plates are considerably higher.
These disagreements can perhaps be attributed to, among
-Re"Pr"3(-^ some other factors, the geometric differences in the chevron
de \ fj,w C'„ plate surface corrugations employed in various studies (Man-
glik, 1996). These include different shapes of corrugation pro-
files (Ding and Manglik, 1996), depth b and wavelength X. of
k
Re«Pr1/3(^- Re'Pr1" the furrows (Okada et al., 1972; Gaiser and Kottke, 1990; Ding
C[ de \fJ-y,
and Manglik, 1996), plate surface enlargement factor 4> (ratio
of effective corrugated surface area to its projected area), and
(12) flow distribution channel configurations. For example, <fi =
1.464 in the scaled up surfaces employed by Focke et al. (1985)
By acquiring data over an extended range of both cooling in their mass transfer studies, as compared with 4> = 1-29 in
water and test fluid flow rates, as per the methodology described the present case. The deeper furrows in the former would tend
in detail by Shah (1990), and Manglik and Bergles (1991), to induce greater swirl mixing, and hence have higher Nu.
an iterative multiple linear-regression analysis yields the lead Whereas Okada et al. (1972) have employed plates with <f> =
coefficients (C/,, C') and Reynolds exponent (a). Typical re- 1.294, and, as such, there is good agreement between their
sults for P = 30 deg/30 deg plates are presented in Fig. 4. results and present data. The plate surface geometry features
Similar results for the other two chevron plate arrangements (/3 have not been reported by Thonon et al. (1995); in fact, few
= 60 deg/60 deg and 30 deg/60 deg) can be found in Muley
and Manglik (1995, 1998). It may be noted, however, that the
cooling water side Nu equation devised in this manner is only
1 1—i i i i i 1 1 — - — i 1 — i -
a "calibration" equation for the present test facility, valid only Experimental Data (p = 1/3, (*= 1.29, y= 0.556)
over the limited range of Wilson plot data, and should not be
A p= 30730°
considered as a general correlation. In all subsequent experi- O ^=30760° (45° avg.)
ments, the cooling water side conditions were always main- O jS= 60760°
tained within the range of Wilson plot data. Okada etal. (1972), (,u/,uw) = 1, p = 0.4
Focke etal. (1985), (jul^) = 1, p = 0.5
Experimental Uncertainty. For the measurements and de- 102 Thonon etal. (1995), (/i/// w )=l
rived results presented in this paper, the experimental uncertain-
ties were calculated according to the procedure outlined by
Moffat (1988). The maximum errors in the primary measure- ^
ments of m, Ap, Ar l m , and Q were ±4.1 percent, ±6.0 percent,
±4.2 percent, and ±7.8 percent, respectively, after accounting
for errors in both the process and colling water flow streams.
Based on this precision and the propagation of error analysis, I
the maximum uncertainties in Re,/, and Nu were ±4.2 percent,
±7.2 percent, and ±10.4 percent, respectively. Also, as men-
tioned earlier, for majority of the heat transfer data the energy
balance was within ±5.0 percent. Extended details of the error
and uncertainty analyses are given in Muley and Manglik
(1998).
( i I
l i — i — i — i — i — i — i — i — r
While the chevron plate surface corrugations promote higher O Present Experiment
heat transfer coefficients (Fig. 5), there is a higher pressure V Tovazhnyanski etal. (1980)
D Fockeetal. (1985)
drop penalty as well (Fig. 6 ) . The friction factors are 13 to 44 ! A Heavner etal. (1993)
times higher than those for the flat-plate channel. As would be O Thononetal. (1995)
expected, the swirling fluid motion in the interplate channel 0.8 - ,.---'
&$&?'' - o
_ o /3= 30760° (45°)
a o /?= 60760°
- Predictions,Eq. (16)
JM
£s&s'
/ 2 "- v
. *
A Okada etal. (1972)
Okada etal. (1972) '
\P- 60°, n = 0, p - 0.4
+ - - ; +15% ..•••-' A _
o\
\e
o . Flat-plate channel „, „ - - ' ^^^^^^ _----"'_
o
0 : V :
- -
1
"3.
7
/ .±^<^'''' ^
=e 10- 1
-}
1.2 1.3 1.4 1.5
- rX-'' -> 0% -
1 p/> i i
Fig. 10(a)
i i • • i - i i
: • Present study ,
, 1 , , - + Tovazhnyanski et al. (1980 1/3 = 30°, 45°, 60°
i i i i
- o Focke etal. (1985) > o :
102 10-* - * Heavner et al. (1993), p= 34°, 45°, 67.5°
Re [0.728 + 0.0543 sin{(ff/?/45) + 3.7(]
2 -
Fig. 8 Comparison of predictions and experimental data for the effect o. " /Ac 2 (/?)Re p 2 | / , ) j = [5.474-19.02^ + 18.93<*2 -53410 3 ] +15% :
of chevron angle p on Nu in turbulent flows (Re a 1000)
S
u "'-'15% ;
the Nu and /data, respectively, are presented in Figs. 8 and 9,
and the excellent agreement between the two (within ±10 per-
s; i ; Flat- plate channel . - ^'^--'-"~ - ' ' ' O
cent and ±5 percent error bands, respectively) is clearly evident.
Besides the corrugation inclination angle /3, the area enlarge- O ;
ment factor 0 also has an influence, and higher heat transfer ''••L
and pressure drop are obtained with increasing 0; deeper corru- • . *
gations increase the effective surface area as well as promote 1.1 1.2 1.3 1.4 1.5
greater swirl mixing. To extend the applicability of Eqs. (16)
and (17), which are limited to 0 = 1.29, the present results
Fig. 10(b)
are correlated with those given in the literature for different 0
values. This is depicted in Fig. 10, where the normalized Nu Fig. 10 Correlation of the effect of surface enlargement factor <f> in
and / d a t a for the three chevron plate arrangements are graphed turbulent flows (Re a 1000): (a) Nusselt number and (b) isothermal
friction factor
along with the results of Okada et al. (1972), Tovazhnyanski
et al. (1980), Focke et al. (1985), and Heavner et al. (1993).
As seen from this figure, the effects of 0 can be represented by cubic polynomials in the range 1 < 0 =s 1.5, with 0 = 1
for a flat-plate channel. Consequently, Nu and isothermal / c a n
be predicted by the following correlations, which include the
effects of /3 and 0:
Nu = [0.2668 - 0.006967/5 + 7.244 X l O ^ 2 ]
X [20.78 - 50.940 + 41.160 2 - 1O.5103]
X R e [°- 7 2 8 + 0 - 0 5 4 3 s i n KT/5/45 ) + 3.7 I P r " 3 ( u / W )014 (18)
and
/ = [2.917 - 0.1277/3 + 2.016 X 1 0 ^ 2 ]
X [5.474 - 19.020 + 18.930 2 - 5.3410 3 ]
X P e - | O . 2 + 0.0577sin|(7r/)/45)+2.11
(19)
3
Equations (18) and (19) are generally valid for Re > 10 , 30
deg < P < 60 deg, and 1 < 0 s 1.5,4 with sinusoidal shape
chevron corrugations, and the predictions represent much of the
available data reasonably well. Additional data, however, would
be desirable to further establish their validity.
To quantify the heat transfer enhancement in heat exchangers
for different applications, several figures of merit have been
0.2 0.3 0.4 proposed in the literature (Webb, 1994; Bergles, 1985; Kays
R e -[0.2 + 0.0577sin{(7r/?/45) + 2.11]
4
While manufacturing plates with 0 «* 1.5 may be very difficult because of
Fig. 9 Comparison of predictions and experimental data for the effect tooling and fixtures limitations, this limit reflects the values of (ft = 1.412 and
of chevron angle ji on isothermal f in turbulent flows (Re 2= 1000) 1.464, respectively, for the Okada et al. (1972) and Focke et al. (1985) data.
Acknowledgments
0.5 Support for this study was given by the Procter & Gamble
10J 10" Co., and the National Science Foundation (Grant #CTS-
Re,ch 9502128).
Fig. 11 Effect of chevron angle ji and surface enlargement </> on heat
transfer enhancement in a PHE for fixed surface geometry and constant References
pumping power constraints
Bergles, A. E., 1985, "Techniques to Augment Heat Transfer," Handbook of
Heat Transfer Applications, 2nd Ed., W. M. Roshenow et al., eds., McGraw-Hill,
New York, Chapter 3.
and London, 1984). Relative to an equivalent parallel-plate Ding, J., and Manglik, R. M., 1996, "Analytical Solutions for Laminar Fully
pack, the enhanced thermal performance of chevron plates, Developed Flows in Double-Sine Shaped Ducts," Heat and Mass Transfer, Vol.
31, pp. 269-277.
based on the criterion of equal pumping power (E) and fixed Focke, W. W., Zachariades, J., and Olivier, I., 1985, "The Effect of the Corru-
surface geometry, can be expressed as gation Inclination Angle on the Thermohydraulic Performance of Plate Heat Ex-
changers," International Journal of Heat and Mass Transfer, Vol. 28, pp. 1469-
[Qck/Qpp] = [Nurf,/Nuch\L,cl„NA,EAT
c„] (20) 1479.
Gasier, G., and Kottke, V., 1990, "Effects of Corrugation Parameters on Local
where the constraint of equal pumping power and Integral Heat Transfer in Plate Heat Exchangers and Regenerators," Heat
Transfer 1990, Vol. 5, G. Hetsroni, ed., Hemisphere, New York, pp. 85-90.
(/Re 3 ) c , 1 = (/Re 3 )„ ) (21) Heavner, R. L., Kumar, H„ and Wanniarachchi, A. S., 1993, "Performance of
an Industrial Plate Heat Exchanger: Effect of Chevron Angle," AIChE Symposium
provides the relationship between the Reynolds numbers for the Series, Vol. 89, No. 295, pp. 262-267.
two channels. Because chevron plates have higher friction fac- Kakac, S., Shah, R. K., and Aung, W., 1987, Handbook of Single-Phase Con-
vective Heat Transfer, John Wiley and Sons, New York.
tors compared to the parallel-plate channel, Eq. (21) requires Kays, W. M., 1950, "Loss Coefficients for Abrupt Changes in Flow Cross
that, for a given Re,,,,, Re ;v must be higher. Figure 11 shows Section with Low Reynolds Number Flow in Single and Multiple Tube Systems,''
the enhancement of heat transfer in chevron plates according Transactions of ASME, Vol. 72, pp. 1067-1074.
to the criterion of Eq. (20). Here, the Nusselt number data for Kays, W. M„ and London, A. L„ 1984, Compact Heat Exchangers, 3rd Ed.,
McGraw-Hill, New York.
chevron plates have been modified to account for enlargement Manglik, R. M., 1996, "Plate Heat Exchangers for Process Industry Applica-
factor cj> (reduced to a projected area basis), so that the chevron- tions: Enhanced Thermal-Hydraulic Characteristics of Chevron Plates," Process,
plate and parallel-plate channels have the same heat transfer Enhanced and Multiphase Heat Transfer, R. M. Manglik and A. D. Kraus, eds.,
area. It is seen that the heat transfer is enhanced by 1.1 to Begell House, New York, pp. 267-276.
Manglik, R. M„ 1999, ' 'On the Influence of Number of Thermal Plates on the
2.8 times that in an equivalent parallel-plate channel at equal Performance of Plate Heat Exchangers," International Communications in Heat
pumping power. The enhancement is also seen to increase con- and Mass Transfer, submitted for publication.
siderably with increasing chevron angle (3 and surface enlarge- Manglik, R. M., and Bergles, A. E., 1991, "Heat Transfer Enhancement of
ment <]>. Typically with Rerf, = 1500 and 4> = 1-29, the heatIntube Flows in Process Heat Exchangers by means of Twisted-Tape Inserts,"
Report No. HTL-18, Heat Transfer Laboratory, Rensselaer Polytechnic Institute,
transfer rates in chevron plates are 1.2, 1.4, and 1.6 times higher Troy, NY.
in f3 = 30 deg, 45 deg, and 60 deg chevron plates, respectively, Manglik, R. M., and Bergles, A. E., 1998, "Numerical Modeling and Analysis
than that in parallel-plate channels. However, the heat transfer of Laminar Flow Heat Transfer in Non-Circular Compact Channels," Computer
enhancement diminishes with increasing RecA. For instance, Simulations in Compact Heat Exchangers, B. Sunden and M, Faghri, eds., Compu-
with P = 30 deg/30 deg and <f> = 1.29, enhanced heat transfer tational Mechanics Publications, Southampton, UK, Chapter 2.
Manglik, R. M., and Muley, A., 1993, "Heat Transfer and Pressure Drop
is obtained only when Rec/, < 10 4 . Also, based on the present Characteristics of Plate-and-Frame Heat Exchangers: A Literature Review," Re-
results, there appears to be no significant advantage in using a port No. TFL-Int-1, Thermal-Fluids Laboratory, University of Cincinnati, Cincin-
mixed-plate (P = 30 deg/60 deg) arrangement over a symmet- nati, OH, Sept.
ric (p = 45 deg/45 deg) arrangement in turbulent flows. This Marriott, J., 1977, "Performance of an Alfaflex Plate Heat Exchanger." Chemi-
observation is reflected in the results previously reported by calMoffatt,
Engineering Progress, Vol. 73, No. 2, pp. 73-78.
R. J., 1988, "Describing the Uncertainties in Experimental Results,"
Heavner et al. (1993) and Muley and Manglik (1997) as well. Experimental Thermal and Fluid Science, Vol. 1, pp. 3-17.
Muley, A., and Manglik, R. M., 1995, "Experimental Investigation of Heat
Transfer Enhancement in a PHE with (3 = 60° Chevron Plates," Heat and Mass
Conclusions Transfer 95, Tata McGraw-Hill, New Delhi, pp. 737-744.
Muley, A., and Manglik, R. M., 1997, "Enhanced Heat Transfer Characteristics
Experimental data for Nu and isothermal / in a single-pass of Single-Phase Flows in a Plate Heat Exchangers with Mixed Chevron Plates,"
PHE with three different chevron plate arrangements (symmet- Journal of Enhanced Heat Transfer, Vol. 4, pp. 187-201.
ric P = 30 deg/30 deg and 60 deg/60 deg, and mixed p = 30 Muley, A., and Manglik, R. M., 1998, "Heat Transfer and Pressure Drop in
deg/60 deg) are presented. They represent cooling of hot water Plate Heat Exchangers," Thermal-Fluids Laboratory Report No. TFL-2, Depart-
ment of Mechanical, Industrial and Nuclear Engineering, University of Cincinnati,
(2 < Pr < 6) flows in the turbulent regime (600 < Re < 10"). Cincinnati, OH.
Compared to flows in an equivalent flat-plate channel, up to Okada, K„ Ono, M., Tomimura, T„ Okuma, T„ Konno, H., and Ohtani, S.,
five times higher Nu is obtained; however, the concomitant 1972, "Design and Heat Transfer Characteristics of a New Plate Heat Ex-
isothermal / i s 44 times higher. Along with the fluid flow param- changer," Heat Transfer-Japanese Research, Vol. 1, No. 1, pp. 90-95.
eters Re, Pr, and (fjj/iw), this performance is found to be ticsSavostin, A. F„ and Tikhonov, A. M., 1970, "Investigation of the Characteris-
of the Plate Heating Surfaces," Thermal Engineering, Vol. 17, No. 9, pp.
strongly influenced by the chevron angle j3 and surface enlarge- 75-78.