0 views

Uploaded by mortezagashti

louvered

- Bao 2003
- Laminar Fountains – What are they_ _ Scuttlebots.pdf
- Chapter 6 Flow Control in S Duct
- 10.1038@s41567-017-0026-3.pdf
- parveez2014
- A Lattice-Boltzmann Simulation Study of the Drag Coefficient of Clusters of Spheres
- Muy Bueno Este Poole_29
- Cv - Orifice Diameter.pdf
- Xuereb_3034.pdf
- FMX Static Mixer
- Younis_cv
- Hydraulics - Series 3.pdf
- Effect of Different Parameters on the Performance of Rib Roughness Solar Duct
- 2003 [Kathleen a. Hinko] Transitions in the Small Gap Limit of Taylor-Couette Flow
- Suppressing Turbulence and Enhancing Liquid Suspension Flow in Pipelines With Electrorheology
- Wind Tunnel Testing
- 1985 - H Kalman - Thicknessofthermalandvelocityboundarylayersonamobi[Retrieved-2016!11!11]
- Aero Mechanical Design
- Boundary.layer.analysis
- Acoustical characterization of perforated facings

You are on page 1of 13

Ralph L. Webb

Paul Trauger

Department of Mechanical Engineering,

The Pennsylvania State University,

University Park,. Pennsylvania

automotive heat exchangers was performed using a dye injection technique with

10:1 scale models. The geometrical parameters, louver pitch, louver angle, and

fin pitch were varied to determine their effect on the flow structure. Tests

covered Reynolds numbers of 400-4000 based on louver pitch. Data are

presented in the form of a dimensionless flow efficiency (defined in terms of the

mean flow angle relative to the louver angle) and Reynolds number. Correlations are developed to predict the flow efficiency as a function of dimensionless

geometrical groups and Reynolds number. The flow structure is also discussed.

INTRODUCTION

Enhanced surface geometries permit reductions in heat exchanger size and weight. Their use is of particular importance

in the automotive industry, where a premium is placed on

space, weight, and cost [1]. For the case of automobile

radiators (Fig. 1), the dominant thermal resistance generally

occurs on the airside, which may account for 80% or more of

the total thermal resistance. Therefore, a reduction in the

air-side thermal resistance will improve the heat transfer

performance of the heat exchanger. This reduction can be

achieved by several means. First, additional heat transfer

surface area can be obtained through the use of extended

surfaces. Second, heat transfer enhancement techniques can

be employed to increase the gas-side heat transfer coefficient.

This is generally accomplished through the use of features on

the extended surface. For gas-side applications, fin geometries fall into two categories: continuous and interrupted

surfaces. Continuous surfaces achieve heat transfer enhancement through the secondary flow patterns introduced by

sudden velocity changes. Examples of continuous surfaces

are wavy and dimpled plate fins. On the other hand, interrupted surfaces (Figs. 2 and 3) achieve heat transfer enhancement by the continuous growth and destruction of laminar

boundary layers on the interrupted portion of the geometry.

Examples of interrupted surfaces are the offset strip fin and

louvered fin.

Louvered fins can be manufactured by high-speed production techniques and as a result are less expensive than other

interrupted flow geometries when produced in large quantities. For this reason, they have found wide use in the

automotive industry. Louvers are generally formed by cutting

the metal and pushing out the cut elements from the plane of

the base metal [2]. Commonly, corrugated fins (shown in

Fig. 1) provide the base for the louvered surface (shown in

Fig. 2); however, there are some applications that require the

use of plate fins. Louvered surface geometries can be made in

a wide variety of shapes depending on fabrication techniques

and consWaints.

the 1950s, it has been only within the past 20 years that

serious attempts have been made to understand the flow

phenomena and performance characteristics of the louvered

fin. Kays and London [3] were the first to report heat transfer

and pressure drop data on louvered fins. However, the geometries of the test samples reported by Kays and London [3]

do not reflect present industrial designs, so the data are of

little present value. To date, the only known open literature

sources for heat transfer and pressure drop data on the

louvered fin are Davenport [4, 5] and Achalchia and Cowell

[6]. Davenport tested single-row, corrugated fin louvered

heat exchangers, and Achaichia and Cowell tested one- and

two-row louvered fin exchangers having round tube and flat

(louvered) fin geometry.

Although experimental data will always be used in the

design of heat exchangers, researcher have begun to use

analytical [7-9] and numerical [10, 11] techniques to predict

the performance characteristics of the louvered surface geometry. These methods can provide detailed understanding of

the enhancement mechanism of the louver geometry, if the

model is correct. However, results of numerical methods may

be misleading because of the assumptions made. This will

occur if the model assumes laminar flow parallel to the louver

or is incapable of predicting flow separations and the effect of

the resulting eddies on the flow structure. The objectives of

this study are to investigate the flow phenomena in the

louvered fin array and to establish the influence of velocity

and geometrical parameters on the flow structure within the

array. This information will be useful in determining the

validity of proposed analytical and numerical solutions. Also,

it is intended that the study will provide insight into improved

fin design.

LITERATURE REVIEW

Beanvals [12] appears to have been the first to conduct flow

visualization experiments on the louvered fin array. In his

1965 study, a smoke flow visualization technique was used

Address correspondenceto Professor Ralph L. Webb, Departmentof Mechanical Engineering, The PennsylvaniaState University, University1Park, PA

16802.

1991 by ElsevierScience Publishing Co., Inc., 655 Avenueof the Americas, New York, NY 10010

0894-1777/91/$3.50

205

COOLANT

FIN

TUBE

Cell

value, since the actual geometric dimensions and velocities

were not documented in the paper. From inspection of the

photographs included in the paper, it appears that the louver

angle (0) was approximately 30* and the louver-to-fin pitch

ratio (Lp/Fp)was approximately 0.80. His figures show that

the main flow was nearly parallel to the louvers for the

velocities tested. Prior to this work, it was speculated that

louvers acted as a surface roughness that enhanced the performance characteristics of the fin by promoting turbulence.

In 1973, Wong and Smith [13] measured heat transfer and

pressure drop data on a 5:1 scale model of a typical louvered

fin array. Comparison of the model data with those of similar

full-scale louvered arrays showed good correlation between

the model and actual cores. It was concluded that the airflow

phenomenon was similar for the full-scale and model cores.

In 1980, Davenport performed flow visualization experiments identical to those of Beauvais and demonstrated that

the flow structure within the louvered array was a function of

Reynolds number (ReLp). At low values of ReLp, the louvers

had only a slight influence on the flow structure. Thus, the

main flow stream did not pass through the louvers. However,

at high values of ReLp the flow became nearly parallel to the

louvers. Davenport speculated that at low air velocities the

developing boundary layers on adjacent louvers became thick

enough to effectively block the passage, resulting in nearly

axial flow through the array. This information is discussed by

Achaichia and Cowell [6].

Figure 4 shows heat transfer data of Achaichia and Cowell

[6] for three louver fin geometries used on plate fins with 11

mm tube pitch. At the hightest ReLp, the data are parallel to

(but lower than) that for laminar boundary flow over a flat

plate (the "flat plate" line). At low ReLp , the data show

characteristics similar to those of laminar duct flow. The

"duct flow" line is for fully developed laminar flow in a

rectangular channel with a 3.3:1 aspect ratio. The aspect ratio

for the finned tubes is between 2.7 and 4. This behavior is

consistent with Davenport's flow observations. Since the heat

transfer performance is closely related to the flow structure,

Bose S u r f a c e

we may infer that two types of flow structure exist within the

louvered, flat fin array:

1. Duct flow, in which the fluid travels axially through the

array, essentially bypassing the louvers

2. Boundary layer flow, in which the fluid travels parallel to

the louvers

Figure 4 shows the Stanton number (St) characteristics for

three samples, numbers 1 and 2, where the louver angles are

29* and 20", respectively. For both samples Fp/Lp = 2.63.

The louver angle affected the Stanton number only in the

transition region between the duct and boundary flow regions.

Achaichia and CoweU [11] also numerically modeled the

flow through the louvered fin array. Figure 5, taken from

their work, shows that as the Reynolds number approaches

large values, the mean flow angle u approaches the louver

angle 0 to within a few degrees. The mean flow angle a is

the integrated average value over the length directional flow

path [11]. Their analysis, which assumes a fully developed,

periodic laminar flow, supports Davenport's hypothesis concernlng the boundary layer development on the louvers at low

values of ReLp. The study shows that the flow structure is a

function of Reynolds number ReLo , the louver-to-fin pitch

ratio Lp/Fp, and the louver angle/9. Figure 5 shows that the

mean flow angle is a strong function of ReLp at low ReLp and

is independent of ReLp at high ReLp. If eddies are shed from

the louvers, because the flow is not parallel to the louvers,

1

I I 111

L = 0.81 mm

0'1 : ~

~

Line

FalLp

~ (deg)

260

29

2.66

2o

4.11

29

z

c

.

. . . . . . .

Duct FLow~

_

6')

0"01

10

SECTION

A--A

F'IN

GEOMETRY

PER

TUBE

ROW

Itlnll

100

I I Ill

1000

R e y n o l d s N u m b e r - ReLp

1.0

0.9

0.8

0.7

C~

~5

0.6

0.s

0.4

0.3

0.2

0.1

,

0.0

10

100

, ,

1000

ReLp

Figure 5. Mean flow angle dependence in a louvered fin array.

(From Achaichia and Cowell [11].)

their analysis will probably not accurately predict the effects

of these eddies on the heat transfer coefficient and friction

factor.

Recently, Howard [14] performed flow visualization experiments on a 10:1 scale model of a two-dimensional louvered

fin array using a dye injection technique. Howard observed

that flow instabilities (i.e., vortex shedding) commenced at a

Reynolds number based on fin thickness of approximately

Re t = 30 (ReLp = 900) and that the instabilities progressed

upstream (from the exit end of the array) as the velocity was

increased. He characterized the flow structure as being either

"efficient" or "inefficient." The flow was considered to be

efficient if it was approximately parallel to the louvers.

L=160

Fp=15

predominantly axial (i.e., did not go through the louvers).

This definition is based on the assumption that an efficient

flow structure should yield a higher heat transfer coefficient

than one having the inefficient flow pattern. Howard's definitions are consistent with the definitions of "duct flow" and

"flat plate" flow shown on Fig. 4. If the flat plate flow line

of Fig. 4 persisted into the low-ReLp region, St would be

considerably higher than the experimental values.

Howard observed a transient region between inefficient and

efficient flow, just as a transition region exists from laminar

to turbulent flow. His study, which was performed for a 20*

louver angle, showed that the transition region occurred at

Lp/Fp = 0.7-0.8. The flow structure was considered to be

efficient if Lp/Fp > 0.8 and inefficient if Lp/Fp < 0.7.

Kajino and Hiramatsu [10] also conducted a flow visualization study of the louvered fin array using dye injection and

hydrogen bubble techniques. Figure 6 shows photographs of

flow through the array for a louver angle of 26* Lp/Fp =

0.67 at low Reynolds number (ReLp = 500). Boundary layers

exist on both the upper and lower surfaces of the louver, and

a laminar wake exists downstream from each louver. Flow

separation was observed on the back side of inlet louvers.

Kajino and Hiramatsu concluded that heat transfer enhancement is due to the thin boundary layers that form at the

leading edge of each louver. However, this will be true only

if the flow passes through the louvers.

Figure 7 shows two louver arrays that have the same

louver pitch but different fin pitches. In the left-hand portion

of Fig. 7, a significant fraction of the flow bypasses the

louvers. This is because the hydraulic resistance of the "duct"

Lp=lO

207

~=26

ReL=500

208

L =160

Lp = 10

e=26'

ReL = 500

flow region is substantially smaller than that for boundary

layer flow across the louvers. When the fin pitch is reduced,

as in the right-hand part of Fig. 7, the hydraulic resistance of

the "duct" is increased, so that most of the flow passes

through the louvers.

EXPERIMENTAL PROGRAM

In order for flow visualization experiments using large-scale

models to accurately simulate the flow in a full-scale array,

both geometric and dynamic similarity must be satisfied. In

addition, the turbulence levels of the model and prototype

flows should be the same. Beauvais made turbulence measurements on large-scale and full-scale geometries with air as

the fluid and found that the two turbulence levels were in fair

agreement.

Previous studies demonstrated that the geometrical parameters that are most likely to influence the flow structure in the

louvered fin array are the louver pitch Lp, louver angle 0, fin

pitch Fp, and fin thickness t. The flow visualization model

geometries selected for testing were 10:1 scale and had

Lp/Fp, Lp/t, and 0 typical of common industrial designs.

Table 1 gives the model geometries. Two louver angles (20*

and 30*) were selected. For both louver angles, the louver

pitch Lp was 15 nun and the fin pitch Fp was variable from

11 to 30 ram. This allowed for testing of 0.49 < Lp/Fp <

1.31.

The louvered arrays selected permitted study of the effect

of louver angle 0 and the louver-to-fin pitch ratio Lp/Fp on

the flow structure within the array. Each model consisted of

10 louvered sections in the spanwise direction. The range of

Reynolds numbers was approximately 400 < ReLp < 4000,

which corresponds to a frontal air velocity range of 2.8 < Ufr

< 28.0 m/s. The louvers were made from 0.635 nun thick

brass sheet were approximately 150 nun high. All louvers

had two tabs at each end (as shown in Fig. 8), which allowed

the louver to be mounted on support plates, thus preventing

the louvers from moving during the experiment. The position

of the holes in the support plates fixed the desired louver

angle. Top and base support plates were made from 3.18 mm

Lexan.

The flow visualization tests were conducted in a closed-circuit, open water channel. The channel was approximately 5

m long, 0.3 m wide, and 0.46 m deep. Figure 9 shows a

schematic plan view of the flow visualization test section.

Water was pumped from a reservoir and through the channel

using one of two pumps for different flow rate ranges. A

metal screen at the upstream end of the channel provided a

uniform water velocity. A bank of drinking straws was placed

15 mm upstream from the model to serve as a flow straightener. The flow rate was measured with two calibrated orifices

connected to a manometer. For a given flow rate, the depth of

the water in the channel was adjusted by raising or lowering a

gate at the downstream end. The height of the water channel

was adjusted to be just above the top support plate. Water

temperature was measured with a thermometer located downstream of the test section. The width of the model varied

depending on which fin pitch was being tested. A contraction

INLET DEFLECTIONLOUVER

Lp

Fp

Model

(deg)

(mm)

(ram)

(mm)

1

2

20.0

30.0

15.0

15.0

0.635

0.635

11-30

11-30

Figure 8.

=ii

.,~

llr

=-

Q,

':,

/-7

BASESUPPORT

PLATE

array tested.

DYE INJECTION

209

FLOW STRAIGHTENER

D = (t/2

section was constructed to narrow the width of the water

channel to that of the model (not shown in Fig. 9).

Flow visualization was performed using a dye injection

technique. The dye was a powder (Best-acid azure blue from

Best Color and Chemical, Inc.) that was mixed with water.

This solution was then mixed with approximately 20% whole

milk. The fat content of the milk is presumed to reduce the

diffusion of the dye filament [15]. Dye was injected into the

flow with a hypodermic needle, which was connected to a dye

reservoir by plastic tubing. The dye reservoir was mounted

1.0 m above the channel test section, which allowed gravity

flow of the dye. The injection rate of the dye was controlled

by a small adjustable clamp placed on the tubing. Typically,

dye was injected in a straw of the flow straightener, so that it

entered at the fin pitch centedine and at one-half the fin

height (75 mm). The maximum boundary layer thickness at

the exit end of the louver array support plates was 38 mm.

The visualization tests were conducted as follows:

1. Place the appropriate model in the channel, and set up the

contraction section, if required.

2. Set the flow rate at the lowest value, and adjust the height

of the water to be just above the top support plate.

3. Introduce the dye continuously, and observe the flow.

Observations were recorded in a notebook, and sketches

of the flow streamlines were made on a schematic diagram

for the specific model geometry.

4. Increase the flow rate incrementally, and note observations until the entire Reynolds number range is covered.

- s)tano

O)

occurs halfway across the first louver. The L e of the first

louver is the same as for the downstream louvers. The

numerator of Eq. (1) was taken from the flow pattern sketches

made during the test runs. The definition of flow efficiency

was selected on the basis of its practical nature. The flow

efficiency is equal to 1 when the flow is parallel to and

through the louvers. It is equal to zero when the flow is axial

through the array (100% duct flow).

EXPERIMENTAL RESULTS

Data were taken for six louver-to-fin pitch ratios ( L p / F p =

0.49, 0.56, 0.66, 0.79, 0.98, and 1.31) and two louver

angles (0 = 20 and 30). Data were taken over a Reynolds

number range of 400 < ReLp < 4000. Figures l l a and l l b

present flow efficiency (for all six pitch ratios) plotted against

Reynolds number for louver angles of 20 and 30 , respectively. As will be seen, the Reynolds number based on louver

pitch was found to be a better correlating parameter than that

1.0

0.9

D

[]

0.8

0.7

[]

- - 1 4 - - 1

.....

[2

0.6

g-

"

[]

Symbol

0

0.5

Lp/Fp

049

0.56

0.66

0.79

0.98

131

z~

0.4

0

0.3

02

0:20

Ol

O0

1000

3OO

data, a Reynolds number (ReLp) and "flow efficiency" (~/)

were calculated. Figure 10 provides a visual definition of the

flow efficiency. The flow efficiency 71 is defined from Fig. 10

ReLp

(a)

as

N

*/ = D

1.0

ideal transverse distance

m-

(1)

,11

[3

.

O A& @O A@

0.9

0.8

&

0 0

Symbol

Lp/Fp

0.49

0.56

0.66

0.79

[]

0.98

1.31

0.7

0.6

~"

0.5

0.4

0.3

0.2

=>

0=30"

0.i

0.0

300

I000

Re

------

IDEAL

ACTUAL

STREAMLINE

STREAMLINE

Lp

Co)

Figure 11. (a) Flow efficiency versus Reynolds number for

0 = 20". Co) Flow efficiency versus Reynolds number for 0 =

30".

based on hydraulic diameter (RED). One can convert the data

using Re o = ReLp[2(Fp - t)/Lp].

The maximum error associated with the ReLR is +--2%, and

for measurement of dimension N in Eq. (1) it was 3 mm.

Using the 3 mm uncertainty of dimension N in Eq. (1), the

greatest uncertainty of 7/is +2.2%.

D I S C U S S I O N OF R E S U L T S

Although great care was exercised in the experiments, the

fact that the flow efficiency data were based on visual observations raises some uncertainty. However, the data were

generally checked for repeatability, and good agreement was

found.

Figures 11 a and 11 b show 7/ versus ReLp for louver angles

of 20* and 30*. Inspection of these figures reveals that the

flow efficiency increases with increasing ReLp. This occurs

up to a particular Re value, which we define as the critical

Reynolds number, Re~.p. Above Re[p, the flow efficiency

becomes independent of Reynolds number for fixed Lp/Fp.

From inspection of Figs. I 1a and 11 b, it appears that the

critical Reynolds number is independent of Lp/Fp for a

fixed louver angle. Comparison of these figures reveals that

the critical Reynolds number decreases slightly with increasing louver angle. For louver angles of 20* and 30", the

critical Reynolds number is approximately 1380 and 1200,

respectively.

As the Reynolds number increases from small to large

values, the flow pattern will transition from the duct flow

pattern (left-hand portion of Fig. 7) to approach a flat plate

flow pattern (right-hand portion of Fig. 7). Note that the

maximum flow efficiency at high ReLp is lss than 1.0 for the

smaller values of Lp/Fp.

The following empirical correlation was developed to predict the critical Reynolds number as a function of the dimensionless louver angle.

Reap = 828(0/90) -0.34

greater fraction of the flow passing in the axial direction. As

Lp/Fp is increased (smaller fin spacing), the axial direction

hydraulic resistance increases, forcing more of the flow in the

louver direction. At the smallest fin spacing (Lz,/Fp = 1.31),

the hydraulic resistance in the louver direction is much

smaller than in the axial direction, so all of the flow will go in

the louver direction.

A comparison of Figs. 11 a and 11 b shows that increasing the

louver angle (for fixed Lp/Fp) increases the flow efficiency.

This holds true up to the critical Reynolds number (Re~p),

after which the flow efficiency is independent of louver angle.

As Lp/Fp increases (for ReLp < Re~p), the effect of louver

angle on flow efficiency decreases. Flow efficiency may vary

by as much as 30% for the two louver angles tested. For

ReLp < Reap and Lp/Fp = 1.31 (small fin spacing), the

effect of louver angle becomes very small. It should be noted

that changing the louver angle from 20* to 30* increases the

projected blockage area by nearly 50%. However, the flow

will incur a larger turning loss as the louver angle is increased.

Prediction o f Flow Efficiency

Inspection of Figs. 11 a and 11 b shows two distinct Reynolds

number regions. Therefore, separate flow efficiency correlations are required for each of these regions. These correlations are given below.

function of Lp/Fp alone. The following empirical correlation predicts the flow efficiency for ReLp > Reap within

___2.5%.

r/ = 0 . 9 5 ( L p / F p )

23

(4)

(3)

where 0 is in degrees.

Effect o f Louver to Fin Pitch Ratio

From inspection of Figs. 11 a and 11 b, it is evident that the

flow efficiency increases with increasing Lp/Fp. The flow is

parallel to the louvers (~/ = 1) only if Lp/Fp >_1.31 (the

highest value tested). For Re > ReLp and 0 -- 20 , 30 , the

flow efficiency varies by nearly 30% over the range of

Lp/Fp tested (0.49 < Lp/Fp < 1.31). For Lp/Fp = 0.49

and 1.31, the asymptotic flow efficiency is 0.78 and 1.0,

respectively.

We propose the following qualitative model to explain the

flow efficiency versus Lp/Fp and ReLp relationship. The

flow must choose whether it will follow the louver direction,

the axial direction, or some intermediate angle between the

two directions. Which direction it goes will depend on the

hydraulic resistance of the flow path in the louver direction

relative to that in the axial direction. As L_/Fp is decreased

with fixed Lp, the fin spacing increases, l~owever, the flow

cross-sectional area between the louvers is unchanged. It

appears that this wider fin spacing reduces the hydraulic

resistance in the axial direction relative to that in the louver

Lp/Fp,

performed to provide the best fit of the data in this region.

The resulting correlation is

7/ = 0.091(ReLp)39(

(5)

_+ 10% with this equation. The correlation is only moderately

successful in predicting the flow efficiency near the critical

Reynolds number. Also, small overprediction, for example,

6 - 8 % , will occur for Lp/Fp = 1.31, resulting in flow efficiencies greater than unity. Should the user predict ~ > 1,

one should use 71 = 1.0.

Since Eqs. (4) and (5) are empirical, we do not recommend

that they be extrapolated beyond the range of the experimental data. The range studied is 400 _< ReLp < 4000, 0.49 _<

Lp/Fp < 1.31, and 20* -< 0 -< 30*.

Figure 12a shows a flow pattern map for 0 = 30 with

and 1.31. The 17 versus ReLp data points were

Lp/Fp = 0.56

1.0

d

0.9

0.8

0.7

//

0.6

/

///

0.5

//

o

Lp/Fp

0.56

1.31

Symbol

0.4

0=30

0.3

0.2

0.i

0.0

300

1000

Rehp

(a)

Lp/Fp = 0.56,

O.

LOW ReLp

e =30 deg.

b.

ReLp> ReLp

/

!

Lp/Fp= 1.31, O =30deg.

c.

LOW ReLp

/ / / /

d.

ReLp> Re'~p

/ / / /

(b)

Figure 12. (a) Flow pattern map for 0.56 < L p / Fn < 1.31. Co)

Defined flow patterns for (a).

taken from Fig. l lb. Figure 12b shows the flow patterns

associated with the symbols a, b, c, d on Fig. 12 a. Sketches

a and b of Fig. 12 b show the flow pattern for small Lp/Fu

at small and high ReLp, respectively. Similarly, sketches c

and d show the corresponding flow patterns for high Lp/Fp

(smaller fin spacing). Purely laminar flow on the louvers and

in the wakes was observed for ReLp < 500-600 at all Lp/Fp

and at both louver angles. However, separated flows were

observed at wide fin pitches (small Lp/Fp), as discussed

below.

Sketch a of Fig. 12 b shows the flow pattern observed for

wide fin spacing at low ReLp. A recirculation zone was

observed at the inlet detection louver and at the first two full

louvers. Eddies were shed at the downstream side of the

louvers. As Re L_ is increased, the flow angle increases and

the extent of th~ separated flow decreases. At the higher

ReLp, the trailing wakes rapidly mix the dye, and it was

difficult to perceive very small separated flow zones. For

ReL~ > 1200, 7/ = 0.82, for which the flow angle is 25

compared to the 30 louver angle. We were unable to discern

flow separations, although trailing vortices may have existed.

For a smaller fin spacing, Lp/F. = 1.31, the flow can

more easily approach the louver ang(e, as shown in sketches

211

~/ = 0.94, for which ot = 28.2 , as compared to the 30*

louver angle. No recirculation zones were observed. The

boundary layer on the louver was laminar, as were the

wakes. The laminar wake appeared to fully dissipate before

the flow reached the downstream louver. At the highest ReLp,

the flow was parallel to the louver, so no recirculation zones

were observed. However, the wake region was not laminar

and showed intense mixing of the dye.

A criterion was not developed to define the transition from

laminar to turbulent wakes. However, Figure 12 a shows that

the low values of ~/at low ReLp initiate flow separations and

thus wake instability. The wake instability does not occur

until higher values of ReLp for a smaller fin spacing (larger

Lp/Fp). The visual observations performed here show that

nonlaminar wakes can occur for ReLp < Re*p.

The numerical solution of Kajino and Hiramatsu [10]

assumed laminar flow on the louvers and in the wakes and

used the boundary layer approximations, which are unable to

predict flow separation. Hence, their model was unable to

account for the separated flow structure and the effect of

eddies observed in the present study.

Achaichia and Cowell [11] numerically solved the elliptic

form of the Navier-Stokes equations for "fully developed

flow, in the periodic sense" to predict the flow results plotted

on Fig. 5. This method is capable of predicting flow separations, but it ignores the possibility of entrance region behavior. It is interesting to compare their predicted results with

the present experimental results. Figure 13 shows the Fig. 5

predictions superimposed on the two data sets used to prepare

Figs. 11 a and 11 b. Figure 13 shows that the predicted results

agree with the asymptotic value, 7/ = 0.9. However, the data

show that the flow efficiency begins its drop at a Reynolds

number approximately 10 times higher than that predicted by

the numerical solution [11]. Our flow visualization showed

stronger flow separation and recirculation zones at the first

few louvers than occurred deeper in the array. The observed

"entrance region" behavior may partially explain the difference between the observed flow and that predicted by

Achaichia and Cowell.

The analytical modelers [7-9] assumed a laminar flow

parallel to the louvers. Flow parallel to the louvers was

observed only for Lp/Fp >_1.31 and ReLp > Reap. As previously described, it appears that the flow angle relative to

1.0

q~

. . . . . . . .

. . . . . . . .

0.9

0.8

0.7

~ / / / / I -

~,4gD*

i/

0.5

//

/

0.3

//

0,2

Legend:

0 Offi20(exp)

----

0.8

effi20O(pre)

effi30(pre)

//

o.1

Lp/Fp=2/3

10

100

1000

ReLp

F i g u r e 13. Comparison of measured flow efficiency with that

predicted by Achaichia and Cowell [11].

212

VARIANT 14. OSF, AND FLAT PLATE VS. REIp

,.ooo

........

~f-louv

........

';-

/f-

~.~

"~

, , r 1.000

o.

efficiency-0.600

~'~_~,.,,.~

(ReL~ ~

0.010

. . . . . . . .

10

"

0.200

~

. . . . . .

100

oo

~

,,I

- - IO UV

=,

0.000

1000

Relp

offset strip fins of the same dimensionless parameters. Curve of

7/versus Re for the louvered fin geometry.

flow path parallel to the louvers, relative to the flow path that

entirely bypasses the louvers. The boundary layer on the

louvers should be laminar for practical operating ranges. For

a given ReLp, a higher value of ~ will exist for larger values

of Lp/Fp because of the higher resistance of the axial flow

path. For a given Lp/Fp, the axial flow path may be viewed

as a rough surface. At low velocity, this flow path may be

"hydraulically smooth." One may speculate that as the flow

speed increases, the friction factor and pressure drop of the

axial flow path substantially increase, forcing more of the

flow in the louver direction.

APPLICATION TO HEAT TRANSFER

DESIGN

An objective of the study was to identify louvered surface

designs that provide improved performance. The results suggest that the highest heat transfer performance should occur

for large values of Lp/Fp. As the value of Lp/Fp is

reduced, a smaller fraction of the flow follows the louvers,

and eddies are shed from the trailing edge of the louvers.

Such eddies should yield a form drag component, which may

increase the pressure drop without significantly adding to the

heat transfer. This possibility was investigated by studying

the heat transfer and friction data of Achaichia and Cowell

[6] and then comparing these data to the corresponding data

for a related fin geometry, the offset strip fin.

Figure 14 shows curves of j and f versus ReLp for three

surface geometries:

1. The louver fin data for core no. 14 of Achalchia and

Cowell [6]

2. The offset strip fin (OSF) geometry for the same basic

dimensions as that of Achaichia and Cowell's core no. 14

3. A constant-temperature flat plate with laminar flow, whose

results are represented by j = 0.664(ReLp) -'5

The OSF (Fig. 3) geometry is sometimes described as a

"parallel louver fin." The air passes parallel to the fins in the

offset strip fin, so the concept of flow efficiency does not

apply to this geometry. Both the louver and OSF geometries

have the same dimensionless geometric variables, L p / F p =

0.507, t/Lp = 0.0455, and aspect ratio Ar = (Fp - t)/H

for the OSF geometry. The j and f versus Re of the offset

strip fin were predicted using the correlation of Manglik and

Bergles [ 16].

Both the louver and OSF fin geometries are intended to

provide heat transfer enhancement by periodic development

and destruction of laminar boundary layers. Their performance is compared on Fig. 14 with that for laminar flow over

a constant-temperature flat plate. Figure 14 shows that the

louver fin yields higher performance than the OSF geometry,

but both geometries fall below that of the flat plate. The slope

of the j factor for the OSF geometry closely approximates

that of the flat plate. However, the slope of the louver fin j

factor is more negative than that of the OSF. Figure 14 shows

that the j factor of the louver fin attains a maximum and then

drops as the ReLp is further reduced, contrary to the performance of the OSF geometry.

Figure 15 shows the ratios flouv/fosf and Jlouv/Josf v e r s u s

ReLp , for the louver (louv) and offset strip fins (OSF). When

the j factor of the louver fin begins to drop (Fig. 14), the

flouv/fosf ratio of the louver fin significantly increases as

Re Lp decreases Clearly, the louver fin experiences a detrimental flow phenomenon that does not occur with the offset

strip fin. We will attempt to show that the reducing flow

efficiency of the louver fin is responsible for the unfavorable

performance of the louver fin. The line labeled "flow efficiency" on Figs. 14 and 15 was calculated using Eqs. (4) and

(5).

Examination of Figs. 14 and 15 allows distinction of the

following performance differences between the louver and

OSF fins:

1. As ReLe is reduced from its highest value to the point

where Jlouv attains its maximum, flour increases faster

than fosf does as Re LP is reduced. This is apparently

because of the enhancement provided by the eddies. Note

that the flow efficiency of the louver fin decreases from

68% a t R e L p = 1000 to 35 % at Re t p = 2 0 0

2. As ReLp IS reduced to 200, an inflection point occurs on

the Jl~v curve. Below ReLp = 200, the ratio Jlo~v/Josf

decreases dramatically, and flour/fosf rapidly increases

(see Fig. 15). The flow efficiency continues to drop for

ReLp < 200, attaining ~/ = 0.18 at ReLp = 50.

Examination of the data for 15 core geometries tested by

Achalchia and Cowell [6] shows that all of the louver fin

j-factor curves display the same characteristic behavior as

that of Fig. 14. These curves show a sigmoidal shape rather

than the approximately linear shape shown by the OSF geometry. Such a sigmoidal curve contains an inflection point. It is

possible that this inflection point indicates the initiation of

significant vortex shedding. As ReLp is reduced, the intensity

of the vortex shedding apparently increases and the flow

efficiency decreases. With further reduction of ReLp, the j

factor begins to fall and the friction factor increases. Below

this ReLp, the louver fin geometry shows poor performance.

As shown by Fig. 15, this drastic behavior is initiated at

ReLp = 200, at which 7/ = 0.35. We define this as the "incipient poor performance condition" and denote the associated Reynolds number by R e ~ .

It is desirable to establish the flow condition at which the

"incipient poor performance" is initiated. We may define

this condition by fitting a straight line to the approximately

linear portion of the j-factor curve between the maximum j

VARIANT14VS.OSFWITHFLOWEFFICIENCY

.......

4.000

3.5003.000"

2.5002.000-

,I

. . . .

,~'low

efficiency

0.7

0.800

0.6

0.600

~"

0.5

KEY:

0.4

~ /

0.400

--

0.2

0.200

........

'.

100

........

1000

, ,

0ffi30*

0.0

0.000

Relp

14.2

Lp=1.0 m m

0.1

10

FPI=

- - FPIffi20O

.... FPI=33.0

0.3

j ~ / / ~ . . ~ , ~

1 .ooo.

0.500

...-'"

0.9

1.000

08

~f-louv

~-osf

~

j-louv

1.5oo-

. . . . . . . .

1.0

213

10

15

20

25

30

Ufr(m//s)

fin) versus ReLp, and flow efficiency for the louvered fin.

1.0 mm and 0 = 30*

Figure

Figure

experimental j factor deviates from this straight line is

indicated by the noint

labeled ReL*.

on Fig " 14 " We have

,K~1found that thts Re L. condition can be correlated with the flow

efficiency. Ten of ~chaichia and Cowell's 15 core geometries

have values of Lp/Fp within the ran,,ge of the present tests.

The value of ~ corresponding to Re L_ was predicted for the

10 geometries by using Eq. (5) Equation (5) applies because

Re~., < R e ~ , as defined, by Eq. (3). The observed values of**

Reto and the predicted values of flow efficiency (~/*) at Reto

are listed in Table 2.

The value of 7/at the "incipient poor performance" point

is in the range 35-45 %. Therefore, it can be concluded that

there is a very close relationship between the flow efficiency

7" and the incipient poor performance condition. It is recommended that the louver fin not be selected to operate at

~/< 045

(4) and (5) were used to prepare Fig. 16, which shows the

flow efficiency versus the air frontal velocity (at 20"C) with

FPI (fins/in.) shown as a parameter. This figure shows that

the flow efficiency for the 14 fins/in, geometry begins to drop

rapidly for air velocities less than 13 m/s. The figure also

indicates that lower flow efficiencies will occur for fewer than

14.2 fins/in. Below 0.45 flow efficiency, there should be

significant deterioration of the louver performance, as suggested in the previous section. This occurs at approximately 4

m / s for the 14.2 fins/in, geometry.

In general, reducing the louver pitch will increase the FPI

required to meet a specific flow efficiency. Conversely, increasing the louver pitch will reduce the required FPI to meet

a specified flow efficiency. Also, reducing the louver pitch

will increase the air frontal velocity required to meet a

specific flow efficiency and vice versa.

RESULTS

When the air velocity is decreased, a condition is reached at

which the performance of specific louver fin arrays will begin

to deteriorate. Equations (4) and (5) can be used to predict

the flow efficiency as a function of the louver geometry,

Ret4,, and Lp/Fp. The application of these equations is

illustrated for an actual louver geometry having a louver pitch

Lp of 1.0 mm (0.040 in.) and 30* louver angle 0. Equations

2. Value of ~/at "Incipient Poor Performance

Condition" for Achaichia and Cowell [6] Data

Table

Core

1

3

4

5

6

7

11

13

14

15

L ~ / F,,

Re~

,1"

0.69

085

0.67

0.69

0.65

0.82

0.51

0.51

0.51

0.51

25.5

255

21.5

28.5

25.5

25.5

300

28.0

22.0

22.0

210

230

200

200

145

145

185

200

195

270

0.43

0.47

0.38

0.39

0.36

0.40

0.38

0.38

0.35

0.40

CONCLUSIONS

1. A flow visualization study of the louvered fin was performed to determine the influence of the louver surface

geometry and Reynolds number on the flow structure.

2. Correlations have been developed (based on the Table 1

geometries) to predict flow efficiency as a function of

louver geometry and Reynolds number.

3. The condition of laminar flow in the wake region is

generally not realized over the Reynolds number range of

practical interest.

4. The study provides insight to understanding the requiremerits for design of more efficient louvered surface geometries.

5. Comparison of the heat transfer and friction of Achaichia

and Cowell [6] with the present correlation for ~ suggests

that the j factor will begin to drop below a log-linear

behavior at Ret. p corresponding to 0.35 < ~ < 0.45. For

Ret. p below this value, the j / f ratio is reduced.

6. Equations (4) and (5) can be used to define the air

velocity, FPI, and louver pitch combinations that will

result in low flow efficiency, and possibly low j and low

j / f performance.

This work was performed under the sponsorship of The International

Copper Association. Mr. Paul Trauger performed the flow visualization

experiments and contributed greatly to the interpretation, of the results

and preparation of this manuscript.

214

NOMENCLATURE

Cp specific heat of fluid, J/(kg K)

D ideal transverse distance traveled by streakline, Fig.

10, m

D h hydraulic diameter of fin array, m

Fp fin pitch, m

FPI fins per inch, in. - i

h heat transfer coefficient, W / ( m 2 s)

H fin height, m

L fin array depth in flow direction, m

Lp louver pitch, m

N actual transverse distance traveled by streakline, Fig.

10, m

Re~ Reynolds number based on hydraulic diameter ( =

UfrD h / ~y~,), dimensionless

ReLp Reynolds number based on louver pitch ( =

u fr L p / a p ) , dimensionless

Re t Reynolds number based on fin thickness ( = Ufrt/au),

dimensionless

Reap critical Reynolds number used in Eq. (3) ( =

UfrL p / a u ) , dimensionless

ReLp point of incipient poor performance, dimensionless

S half-length of inlet deflection louver ( = L p / 2 ) , m

St Stanton number (= h/UfrCp), dimensionless

t fin thickness, m

Ufr frontal velocity, m/s

u velocity accounting for contraction ratio ( = u f r / a ) ,

m/s

Greek Symbols

c~ mean flow angle, deg

flow efficiency, ratio of minimum flow area and frontal

area, dimensionless

0 louver angle, deg

a contraction ratio (minimum flow area-to-frontal area)

of louver array, dimensionless

v kinematic viscosity of fluid, m 2/s

REFERENCES

1. Mori, Y., and Nakayama, W., Recent Advances in Compact Heat

Exchangers in Japan, in Compact Heat Exchangers- History,

Technological Advancement and Mechanical Design Problems,

R. K. Shah, C. F. McDonald, and C. P. Howard, Eds., ASME

Syrup. Vol. HTD-Vol 10, pp. 5-16, ASME, 1980.

2. Shah, R. K., Heat Exchangers, in Handbook of Heat Transfer

Applications, 2nd ed., W. M. Rohsenow, J, P. Hartnett, and E. N.

Ganic, Eds., pp. 4-225-4-227, McGraw-Hill,New York, 1985.

3. Kays, W. M., and London, A. L., Compact Heat Exchangers, 3rd

ed., McGraw-Hill,New York, 1984.

4. Davenport, C. J., Heat Transfer and Flow Friction Characteristics

of Louvered Heat Exchanger Surfaces, in Heat Exchangers: Theory and Practice, J. Taborek, G. F. Hewitt, and N. Afgan, Eds.,

pp. 397-412, Hemisphere, Washington,D.C., 1983.

5. Davenport, C. J., Correlations for Heat Transfer and Flow Friction

Characteristics of Louvered Fin, Heat Transfer- Seattle 1983,

AIChE Symp. Ser., No. 225, Vol. 79, pp. 19-27, 1983.

6. Achalchia,A., and CoweU, T. A., Heat Transfer and Pressure Drop

Characteristics of Flat Tube and Louvered Plate Fin Surfaces, Exp.

Thermal Fluid Sci., 1, 147-157, 1988.

7. Smith, M. C., Gas Pressure Drop of Louvered Fin Heat Exchangers, ASME Paper 68-HT-27, 1968.

8. Smith, M. C., Performance Analysis and Model Experiments for

Louvered Fin Evaporator Core Development, SAE Paper No.

720078, 1978.

9. Howard, P., An Analytical Model for Heat Transfer and Friction

Characteristics of a Multi-LouveredFin Heat Exchanger, Masters

Paper, The PennsylvaniaState University, 1987.

10. Kajino, M., and Hiramatsu, M., Research and Development of

Automotive Heat Exchangers, in Heat Transfer in High Technology and Power Engineering, W. J. Yang and Y. Mori, Eds., pp.

420-432, Hemisphere, Washington,D.C., 1987.

11. Achaichia, A., and Cowell, T. A., A Finite Difference Analysisof

Fully Developed Periodic Laminar Flow in InclinedLouvre Arrays,

Proc. 2nd UK National Heat Transfer Conference, Glasgow,

Vol. 2, pp. 883-888, 1988.

12. Beauvais, F. N., An AerodynamicLook at AutomobileRadiators,

SAE Paper No. 650470, 1965.

13. Wong, L. T., and Smith, M. C., Air-Flow Phenomena in the

Louvered Fin Heat Exchanger, SAE Paper No. 730237, 1973.

14. Howard, P., PreliminaryReport on Flow VisualizationStudies on a

Two-DimensionalModel of a Louvered Fin Heat Exchanger, Penn

State Project Report, 1987.

15. Mueller, T. J., Flow Visualizationby Direct Injection, in Fluid

Mechanics Measurements, R. J. Goldstein, Ed., pp. 352-355,

Hemisphere, Washington,D.C., 1983.

16. Manglik, R. M., and Bergles, A. E., The Thermal-HydraulicDesign of the Rectangular Offset-Strip-FinCompact Heat Exchanger,

in Compact Heat Exchangers: A Festschrift for A. L. London,

R. K. Shah, A. Kraus, and D. E. Metzger, Eds., pp. 123-149,

Hemisphere, Washington,D.C., 1990.

WRITTEN DISCUSSION

Webb has presented the results of an experimental study of the

flow-directing properties of louver arrays. He has defined a

"flow efficiency" parameter as the ratio of the lateral

displacement of the flow over the first bank of louvers relative

to the ideal lateral displacement, that is, the lateral displacement that would have occurred if the flow had become fully

aligned with the louvers immediately upon entry to the array.

Correlating equations relating the experimentally determined

"flow efficiency" to Reynolds number and the array geometric parameters are presented. Webb quotes experimental data

on the performance of louvered plate fins from the literature

[6] and relates his flow efficiency parameter to the flattening of

Stanton number curves that was observed with these surfaces at

low Reynolds numbers. He aims to show how his correlating

equations can be used as predictors of curve flattening. The

paper raises a number of points that are worthy of further

discussion,

Flow Efficiency

The definition given in the paper suffers from the drawback

that the flow-alignment process can be gradual over the first

few louvers. The number of louvers required for fully

developed periodic flow is a function of Reynolds number and

the geometric parameters of the array. For example, Zhang

and Lang [17] and Antoniou et al [18] show that the fully

developed periodic condition can take five louvers to become

established. This means that, strictly speaking, "flow efficiency" as defined by Webb is valid only for an array of seven

louver rows as used in his experiments. The experimental

Stanton number data of Ref. 10 is based on samples for which

the number of louvers in a bank ranged from 10 to 18. As

recognized by Webb, it is the significant extent of this

developing flow region and the use of relatively few louvers in

his model that are principally responsible for the large

differences between his flow efficiency data and those based on

numerical analysis of the fully developed periodic situation

presented in Ref. 14. While use of the fully developed periodic

flow data has the opposite shortcoming--it ignores the developing flow region--we propose here that it is a better predictor

of behavior of these geometries.

Correlating Equations

Webb admits that his correlating equations have limited

accuracy and recommends that they not be extrapolated beyond

the experimental range of Reynolds number, that is, 4004000. However, in their subsequent application to the prediction of curve flattening, the relevant Reynolds numbers range

from 145 to 270. The implications of this are made clearer by

consideration of Fig. 1, in which the experimental data and

correlating equation for a louver angle of 20* are plotted

together.

It can be seen that the equations have limited ability to

predict likely experimental behavior at Reynolds numbers

around 200.

Flattening Prediction

As already pointed out, Webb acknowledges the numerically

determined fully developed periodic "flow efficiency" data

presented in Ref. 14. However, he did not recognize that in

Ref. 10 Achaichia and Cowell used the numerical results in

their correlating equation for the prediction of Stanton number.

Their data, which covered 15 different geometries, are

described by

Ot

(6)

1"0

, .%

-----;,&~

c-"

identifiers

.4 ~

"6

. . . . . . .

"0---

.,..~"-% ~ " j . ~ / . ~

.....

0"~

experiment

- I

200

.----e

, = --;-0_-:--_'

"o-~ / ~ ~ ~ _ ~ . ~ ,

--

curve fit

I

400

1-31

0-56

(~/~9

g = 20"

I

600

Re/ll~

1(100

number

ReLp

(5).

215

Sample

Re~*

7/*

1

2~

3

4

5

6

7

8t

9*

l0 t

11

12 t

13

14

15

210

136

230

200

200

145

145

167

173

194

185

211

200

195

270

0.43

0.29

0.47

0.38

0.39

0.36

0.40

0.30

0.35

0.27

0.38

0.30

0.38

0.35

0.40

c~*/0

0.89

0.80

0.91

0.86

0.90

0.86

0.88

0.82

0.85

0.74

0.87

0.75

0.86

0.82

0.84

c~*/C~mx

0.95

0.92

0.96

0.94

0.94

0.93

0.93

0.94

0.95

0.94

0.95

0.93

0.95

0.94

0.95

* Omitted by Webb.

c~/0 can thus be considered a flow efficiency of similar

nature to that defined by Webb. It is a ratio of angles, whereas

Webb's parameter is a ratio of tangents of angles. The term

was deliberately introduced to the correlation procedure to

accommodate the curve-flattening behavior, and yielded an

equation valid down to a Reynolds number of 75. Achaichia

and Cowell presented the following equation to fit their

numerically obtained flow efficiency results in the region of

interest

ct = 0.936 - 243/ReLp -- 1.76 Fp/L, + 0.9950

0

0

(7)

Equations (6) and (7) demonstrate that this flow efficiency is

a determinant of heat transfer behavior and that it should be

able to predict flattening.

Table 1 lists the "point of incipient poor performance"

ReLp** as identified by Webb from the Stanton number data in

Ref. 10. He omitted five of the samples because they lay

outside his experimental geometric range. All the samples have

been included here, and values of ReLp** are identified where

necessary. The values of "incipient poor performance flow

efficiency" as defined by Webb are given in the third column.

The fourth column lists the values of ct*/0 at which the

curves have started to flatten off, as calculated from Eqs. (6)

and (7). This parameter appears to be a somewhat better

predictor than ~/*. a*/O values cover the range 0.74-0.91 (a

ratio of 1.23) for all 15 samples, while ,/* goes from 0.35 to

0.47 (a ratio of 1.34) for the samples covered by Webb.

Nonetheless, two samples in particular show significant

deviation from the 0.8-0.9 range within which most of the

samples lie. Consideration of the significance of Eqs. (6) and

(7) leads to a better curve-flattening predictor.

For any particular geometry, the mean flow angle approaches a limiting value Ctm~x as Reynolds number is increased. Equations (6) and (7) mean that the Reynolds number

at which the limiting value is reached is also the Reynolds

number at which the straight-line region of the log St/log ReLp

curve begins. The appropriate parameter for the identification

of curve flattening is therefore c~*/C~m~xrather than ot*/0. This

216

Ralph L. Webb

final column of Table 1. [C~m~is calculated by setting ReLp to

infinity in Eq. (7).] The values all lie within the range 0.920.96. The mean is 0.94 with a standard deviation of 0.01.

These values confirm that Eqs. (6) and (7) can be used for

predicting Stanton number curve flattening. Their validity

covers the wide range of Lp/Fp values of the experimental data

in Ref. 10, despite the fact that they extend beyond the range

used in the numerical determination of the fully developed

periodic flow angles. What is more, the method arrives at a

limiting value that matches with common sense; one would

expect to detect flattening once the curve had dropped to less

than 95 % of its extrapolated straight-line value.

It is worth noting that although the narrow range of a*/am~x

values suggest that the parameter is a very precise predictor of

flattening, the curves of t~*/C~ax against Reynolds number

have small slope as they approach unity, and for any geometry

the predicted value of critical Reynolds number will be quite

sensitive to the choice of critical parameter value. On the other

hand, neither can the precise point at which an experimental

curve starts to deviate from a straight line be identified with

great accuracy.

Conclusion

Reynolds number at which a louvered fin begins to demonstrate deterioration in performance due to changes in the flow

deflection characteristics. He has proposed the use of experimental data for developing flow for this purpose. It is

recognized that developing and fully developed flow behavior

of louver arrays are interrelated, and as such, by suitable

manipulation, information on either should be able to predict

Stanton number curve flattening. However, it is suggested here

that fully developed periodic data [14] lead to a better

predictor, principally because of the sounder physical basis of

the equations involved.

REFERENCES

17. Zhang, H., and Lang, X., The Experimental Investigationof Oblique

Angles and Interrupted Plate Lengths for Louvered Fins in Compact

Heat Exchangers, Exp. Thermal Fluid Sci., 2, 100-106, 1989.

18. Antoniou, A. A., Heikal, M. R., and Cowell, T. A., Measurements

of Local Velocity and Turbulence Levels in Arrays of Louvered Plate

Fins, Heat Transfer 1990, 4, 105-110, 1990.

T. A. Cowell and A. Achaichia

Department of Mechanical and Production Engineering

Brighton Polytechnic

Brighton BN2 4GJ, UK

AUTHOR'S REBUTTAL

I am appreciative of the additional material provided by Drs.

Cowell and Achaichia (C & A). The motivations for the

present work were the photographs of Kajino and Hiramatsu

[10], which showed that the flow does not necessarily follow

the louver angle, and the numerical predictions of Cowell and

establish the actual flow pattern in a typical louver array for a

range of geometric parameters and Reynolds numbers. If the

test sections were geometrically scaled to the louver pitch

dimensions used in radiator cores, the airflow depth would be

27 mm for Lp = 1.5 mm, and 18 mm for Lp = 1.0 mm. These

flow-depth dimensions are within those currently used for onerow cores. As stated in the paper, the dye was injected into the

flow-straightener straws upstream from the test section. The

measured lateral deflection of the dye was used to define the

mean flow angle over the water flow distance. I believe that the

empirical correlations developed are reasonably accurate for

the range of test conditions.

C & A use the parameter a/O rather than my "flow

efficiency" 7. These terms are related by the simple expression

tan a

=

=rl tan 0 0

(8)

Either expression is valid to define the characteristics of the

mean flow. The approximation ~ = c~/0 is valid within 3 % for

0 = 20* for 0.2 < 71 < 0.9.

The C & A predicted values are for fully developed periodic

flow. However, the present experimental data are for an array

typical of those used in 1-row radiators. It is not expected that

"fully developed" flow would prevail in such arrays 18-27

mm deep. Hence, we question the applicability of the C & A

predictions to actual 1-row radiators. Further, C & A have not

validated their predicted "fully developed" values by experiments. Figure 10 in the present paper shows substantial

disagreement between their predicted t~/0 and the flow

efficiency measured in the present work.

I was intrigued by the maxiumum value shown by the Staton

number data of C & A and thought that the dropoff of flow

efficiency at low Reynolds numbers might be responsible for

the behavior. Hence, I endeavored to explain why this

behavior exists. The flow efficiency is an indicator of the

fraction of the flow that is parallel to the louvers. A flow

efficiency less than 1 means that part of the flow bypasses the

louvers and flows in the "duct region" between the louvered

passages. It is reasonable to expect that if the flow fraction

passing over the louvers decreases, the j factor should also

decrease. I sought to show that the dropoff of the j factor can

be related to the flow efficiency. I was able to relate the

approximate value of the ReLp associated with the "incipient

poor performance" to the flow efficiency. This is presented in

Table 2.

In their discussion, C & A present their Table 3 as a

proposed alternative to Table 2. The value or*/0 is approximately equal to the flow efficiency, as shown by Eq. (8) above.

However, comparison of Table l columns 3 and 4 show that

the flow efficiency (from the present correlation) is substantially smaller than the values shown in column 4. We conclude

that the predicted values of ct*/O, based on the C & A

numerical solution, does not agree with the results of the

present experiments.

C & A propose that the value tx*/'Vr~ (column 5 of Table 1)

is a better predictor of the "point of incipient poor performance." This means that when the flow angle falls only 5%

below its maximum (limiting) value, the j-factor curve will

drop as ReLpis reduced. This is contrary to the findings of this

work. Consider, for example, core 4 in Table 2. The louver

angle is 21.5 , which is very close to the 20 data plotted in

Fig. 8a. At Lp/Fp = 0.66, one reads ~/m~ = 0.88. The

experimental value of R e ~ (point of incipient poor performance) is 200 (Table 3). Extrapolation of the Lo/Fp data points

in Fig. 8a to ReLp = 200 suggests that 7/will be less than 0.5

(the correlation gives ,/ = 0.38). Using the C & A criterion of

a*/Otm~ = 0.95 would predict the incipient point of poor

performance at ~/ = 0.95 x 0.88 = 0.84, at which ReLp =

217

the j factor begins to drop.

I agree that a more precise correlation than Eq. (5) is desired

for prediction of ~ with ReLp < Re*p. However, for the range

of Lp/Fp shown in Table 2 (0.51-0.82), Eq. (5) does a pretty

good job.

R. L. Webb

- Bao 2003Uploaded bybajkiszon
- Laminar Fountains – What are they_ _ Scuttlebots.pdfUploaded byGia Minh Tieu Tu
- Chapter 6 Flow Control in S DuctUploaded byKadiyam Vijay
- 10.1038@s41567-017-0026-3.pdfUploaded byabbey
- parveez2014Uploaded byaleyhaider
- A Lattice-Boltzmann Simulation Study of the Drag Coefficient of Clusters of SpheresUploaded bykubacr
- Muy Bueno Este Poole_29Uploaded byfrankkqqzz
- Cv - Orifice Diameter.pdfUploaded byManuel
- Xuereb_3034.pdfUploaded byDHAVALESH AM
- FMX Static MixerUploaded byTint Lwin
- Younis_cvUploaded byMounir Ghardaia
- Hydraulics - Series 3.pdfUploaded byNica Remollo
- Effect of Different Parameters on the Performance of Rib Roughness Solar DuctUploaded byIRJET Journal
- 2003 [Kathleen a. Hinko] Transitions in the Small Gap Limit of Taylor-Couette FlowUploaded byTont Ekawuth
- Suppressing Turbulence and Enhancing Liquid Suspension Flow in Pipelines With ElectrorheologyUploaded byboomboomboomasdf2342
- Wind Tunnel TestingUploaded bySumit Malik
- 1985 - H Kalman - Thicknessofthermalandvelocityboundarylayersonamobi[Retrieved-2016!11!11]Uploaded byxyplosis12
- Aero Mechanical DesignUploaded byDavide Pinzi
- Boundary.layer.analysisUploaded bycaltech
- Acoustical characterization of perforated facingsUploaded byGowtham Reddy
- 2619Uploaded bymitra1006
- [] Estimation of Pressure Drop in Gasket Plate Heat ExchangersUploaded byzniperx
- Jones Santer Papadakis Debiasi SciTech2016 SubmittedUploaded byyavercan
- 2 - Flow in Pipes Closed Conduits [Compatibility Mode]Uploaded byEddy Bong
- Guide on How to Develop a Small Hydro Power PlantUploaded byBesim Gülcü
- B.tech. Civil Engineering Full Syllabus IndiaUploaded byAnubhav Garg
- Me2204 Unit Wise QuestionsUploaded byMohanraj Subramani
- Flume Final Report 2012Uploaded byjabbj1
- Leonard Et Al-1995-International Journal for Numerical Methods in FluidsUploaded bybill
- Csu 1294789859werUploaded byZiyaul Haq Khan

- HMRI2222Uploaded bymortezagashti
- HMRI2257676722.pdfUploaded bymortezagashti
- HMRI2257676722.pdfUploaded bymortezagashti
- 8511290.pptUploaded bymortezagashti
- 8511290.pptUploaded bymortezagashti
- Python Operator Overloading - The Python Guru.pdfUploaded bymortezagashti
- Python Operator Overloading - The Python new.pdfUploaded bymortezagashti
- Scan 0021Uploaded bymortezagashti
- 43166Uploaded bymortezagashti
- Python Operator Overloading.pdfUploaded bymortezagashti
- Python Operator Overloading - The Python new.pdfUploaded bymortezagashti
- 3236Uploaded bymortezagashti
- Python Operator Overloading - The Python GuruUploaded bymortezagashti
- constant2017.pdfUploaded bymortezagashti
- HMRI2257676722.pdfUploaded bymortezagashti
- HMRI2257676722.pdfUploaded bymortezagashti
- thagiUploaded bymortezagashti
- نحوه محاسبه تراز در آزمون کنکور سراسریUploaded bymortezagashti
- RajabUploaded bymortezagashti
- BIOM9711_CourseOutline_2016Uploaded bymortezagashti
- نحوه محاسبه تراز در آزمون کنکور سراسریUploaded bymortezagashti
- RamazanUploaded bymortezagashti
- AliUploaded bymortezagashti
- control-farayand(www.prozheha.ir).pdfUploaded bymortezagashti
- 1-s2.0-089417779190065Y-mainUploaded bymortezagashti
- CourseDescription_FlowJEM_COMSOLUploaded bymortezagashti
- ComsolUploaded byWillian Gomez Zabaleta
- ComsolUploaded byWillian Gomez Zabaleta
- ConfTokyo2014Mini_OptimizationUploaded bymortezagashti

- Refrigeration 101Uploaded bykaleab
- residue ugradingUploaded byvasathchem02
- Training Report of Different Block of Haldia RefineryUploaded byneo_868686
- 2018 MULTI v CatalogueUploaded bySaviour Udo
- mecbesyUploaded byAnupam Kulkarni
- Heat Transfer LabUploaded byfifafifa
- Heat Exchanger 2015.pdfUploaded bySanketDhande
- Alfandi CvUploaded byAshraf Zoubi
- Entwurf ACU 3 75 EnUploaded byPlaton Mihai
- Mil Hdbk 270aUploaded byAnshuman Roy
- lec1Uploaded byTommyVercetti
- Condenser Maintenance and OperationUploaded bykalyan1967
- design criteria.docUploaded bynefoussi
- 2000-ST-04Uploaded byAlroman
- Young Touchstone FHF-SSF CatalogueUploaded byangie2702
- performance-evaluation-of-gas-turbine-by-reducing-the-inlet-air-temperatureUploaded byijteee
- Eurosun2014-Proceedings _ ImportentUploaded byAkhileshkumar Pandey
- Cooptimisation of Coc for Cooling Towerscorrected.......Uploaded byarunava001
- Multi V5 Catalog-Catalog_2017Uploaded byCK
- ValvesUploaded byIgor Matos
- Clivet WSAN-EE HeatPumpUploaded byJihane Z
- Problem 4.docxUploaded byFery Reykha Ombing
- Refrigeration and Air Conditioning PrinciplesUploaded byMiroslav Mažar
- CRR20-40_Mp-d.pdfUploaded byMelissa Macias
- Hvac Myths Realities 2017Uploaded byDaniel Michel Andrade
- Report PBL 2 (1)Uploaded bylibbissujessy
- Air PreheaterUploaded bydavoudian
- JeradDLId0335vol008issue002Uploaded byIndra MauLana
- atc-e-marketingUploaded byHarikrishnan
- Vol 7-1-041 055 Azridjal Aziz, DISTRIBUTION OF TWO-PHASE FLOW IN A DISTRIBUTORUploaded byAzridjal Aziz