experimental investigation of impingement heat transfer on a flat and dimpled stuttgart2011

8/4/2019 Experimental Investigation of Impingement Heat Transfer on a Flat and Dimpled Stuttgart2011

1/13

Experimental investigation of impingement heat transfer on a flat and dimpled

plate with different crossflow schemes

Yunfei Xing *, Bernhard Weigand

Institut fr Thermodynamik der Luft- und Raumfahrt (ITLR), Stuttgart University, Pfaffenwaldring 31, 70569 Stuttgart, Germany

a r t i c l e i n f o

Article history:Received 13 October 2009

Received in revised form 13 April 2010

Accepted 27 April 2010

Available online 2 June 2010

Keywords:

Impingement cooling

TLC

Heat transfer

Dimple

a b s t r a c t

A nine-by-nine jet array impinging on a flat and dimpled plate at Reynolds numbers from 15,000 to35,000 has been studied by the transient liquid crystal method. The distance between the impingement

plate and target plate is adjusted to be 3, 4 and 5 jet diameters. Three jet-induced crossflow schemes,

referred as minimum, medium and maximum crossflow correspondingly, have been measured. The local

air jet temperature is measured at several positions on the impingement plate to account for an appro-

priate reference temperature of the heat transfer coefficient. The heat transfer results of the dimpled

plate are compared with those of the flat plate. The best heat transfer performance is obtained with

the minimum crossflow and narrow jet-to-plate spacing no matter on a flat or dimpled plate. The pres-

ence of dimples on the target plate produce higher heat transfer coefficients than the flat plate for max-

imum and minimum crossflow.

2010 Elsevier Ltd. All rights reserved.

1. Introduction

The last 20 years have seen a large improvement in gas turbine

technology, due to mainly an increase in turbine pressure ratio and

turbine inlet temperature. The effect of firing temperature is very

important: for every 100 F (55.5 C) increase in temperature, the

work output increases approximately 10% and gives about 1

1.5% increase in thermal efficiency [1]. In practice, high turbine in-

let temperatures have been achieved because of the growth of

materials technology, new coatings and new cooling schemes. Film

cooling methods are mainly considered in the past for the cooling

of the combustor liner. But now the modern dry low emission

(DLE) combustor is required to produce low NOx emissions. The

control of the NOx problem requires to minimize film cooling and

dilution air. These combustors are typically cooled by enhanced

backside convective heat transfer. The liner has a double-wall

structure and impingement cooling is often used to keep the cool-ing effectiveness high.

Numerous investigations on flow and heattransfer characteristic

of multiple jet impingement have been published in order to tailor

the impingement hole shape, size and locations to attain both a suf-

ficiently high average heat transfer coefficient andthe uniformity in

the surface distribution to avoid local hot or cold spots. Han and

Goldstein [2] published in 2001 a review of jet impingement heat

transfer of a single and multiple jets in gas turbine systems. The

variation of local Nusselt number with jet Reynolds number,

jet-to-plate spacing and interactions between multiple jets have

been discussed.

There have been a number of attempts to complement jet

impingement with other enhancing techniques such as crossflow,

ribs and turbulators in order to effective heat transfer with low

pressure loss. Dimple arrays are firstly an attractive method for

internal cooling channels, since they produce time varying multi-

ple vortex pairs which augment local Nusselt number distributions

downstream of the dimple. Now dimpled plate has become into

the consideration due to its potential in heat transfer augmenta-

tion, light weight, low pressure penalty and low maintenance [3].

The impingement on a dimpled plate has not yet been well

apprehended because of numerous complications. Gau and Chung

[4] measured slot jet impingement on concave and convex surfaces

by varying the Reynolds number from 6,000 to 350,000 and the

slot to plate spacing from 2 to 16. They found that the Nusselt

number increases with increasing surface curvature for impinge-ment on a concave surface. Azad et al. [5] measured an array of in-

line air jets impinging on dimpled target plates using a transient

liquid crystal technique with three different spent air crossflow

orientations, jet Reynolds numbers ranging from 4,850 to 18,300

and a jet-to-plate spacing H/d of 3. The dimple diameter is equal

to the jet diameter (Dd/d = 1) and the dimple depth td/Dd = 0.5.

The results showed that the Nusselt numbers for a dimpled and

a flat plate are about the same. Ekkad and Kontribitz [6] investi-

gated the effect of jet impingement on a target plate with a dimple

pattern for Reynolds numbers varying from 4,800 to 14,800 and

the jet-to-plate spacing H/d = 3. The dimples diameter of Dd/d = 2

and two different dimple depths of td/Dd = 0.1 and 0.2 have been

0017-9310/$ - see front matter 2010 Elsevier Ltd. All rights reserved.doi:10.1016/j.ijheatmasstransfer.2010.05.006

* Corresponding author. Tel.: +49 711 685 62444; fax: +49 711 685 62317.

E-mail address: [email protected] (Y. Xing).

International Journal of Heat and Mass Transfer 53 (2010) 38743886

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / i j h m t
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2010.05.006mailto:[email protected]://www.sciencedirect.com/science/journal/00179310http://www.elsevier.com/locate/ijhmthttp://www.elsevier.com/locate/ijhmthttp://www.sciencedirect.com/science/journal/00179310mailto:[email protected]://dx.doi.org/10.1016/j.ijheatmasstransfer.2010.05.006


2/13

investigated. The results showed that the presence of dimples on

the target plate produces lower heat transfer coefficients than

the flat plate. Kanokjaruvijit and Martinez-Botas [3,710] made a

series of investigations on the impingement heat transfer on dim-

pled plate with different crossflow. The jet Reynolds number was

in the range of 500015,000, and jet-to-plate spacings H/d from 1

to 12 were used. The effect of dimple geometry was considered

with two different dimple configurations: hemispherical and

cusped elliptical with the same wetted area and they found that

both dimples showed similarity in heat transfer results. The inves-

tigated dimples parameters are dimple depths (td/Dd) of 0.15, 0.25and 0.29, and dimple diameter (Dd/d) of 0.866, 1.732, 2 and 4. The

shallow dimples (td/Dd = 0.15) improved heat transfer significantly

by 70% at H/d = 2 compared to that of the flat plate, while this value

was 30% for the deep ones (td/Dd = 0.25). The improvement also oc-

curred for the moderate and lower Dd/d. They found that the total

pressure is independent of target plate geometry when H/dP 2.

The levels of the total pressure loss of the dimpled plates are not

different from those of the flat plate under the same setup condi-

tions. Woei et al. [11,12] made investigations on the heat transfer

for an impinging jet array onto two enhanced plates using concave

and convex dimples with effusion for the Reynolds number varying

from 5,000 to 15,000 and a jet-to-plate spacing H/d from 0.5 to 11.

The investigated dimple parameters are dimple depths (td/Dd) of

0.3, and dimple diameter (Dd/d) of 3.5. The results showed thatthe heat transfer performance with convex dimples is better than

their counterparts with concave dimples without effusion. An in-

crease of H/d reduces heat transfer differences between the effu-

sion and non-effusion results for both concave and convex-

dimpled surfaces.

Theobjective of thecurrent study is to investigate theheat trans-

fer and pressure loss values for the impingement on a flat and dim-

pled plate. Although the results in the literature mentioned above

provide many insideinto the heat transfer performance on the dim-

pled plate,their jetarrays andflow arrangementsare quite different

from the present case. The two most important of these unexplored

areas are higher jet Reynolds numbers and calculation method of

heat transfer value inside the dimples. Here various geometric para-

medics such as Reynolds number, jet-to-plate spacing (H/d) andcrossflow schemes are chosen to explore the possibility to enhance

the heat transfer. All results from the dimpled plate are also com-

pared to those from the flat plate.

2. Experimental setup

2.1. Test section

Fig. 1 shows a sketch of the experimental setup. A vacuum

pump system is used to generate the desired air flow in the test

channel. The air enters the channel under atmospheric conditionsvia a filter and a heater. The heater consisted of several meshes

made out of stainless steel and is able to heat the air within less

than 0.3 s from ambient temperature up to 100 C. Downstream

of the heater the air enters the inlet plenum and then the impinge-

ment model. This model is equipped with thermocouples and pres-

sure taps for the measurement of the heat transfer and pressure

loss. It consists of an impingement plate, a target plate and side

rims with effusion outlet holes, as shown in Fig. 2. The spent air

flows through the outlet holes on the exit rim to the outlet plenum.

The target plate is made out of perspex, because it has low thermal

conductivity and allows optical access, needed for the heat transfer

measurements. The target plate is observed from the outside of the

outlet plenum with two CCD video cameras. The model is symmet-

rical, therefore the target plate is only observed for half of the mod-el by two cameras from left to right side.

There area total of 81 impingement holes for theinline impinge-

ment plate. The ratios of jet-to-jet spacing in both directions on the

impingement plate are the same (X/d = Y/d = 5). Because of temper-

ature gradient of the inlet flow from center to corner, it is quite nec-

essary to install many thermocouples on the impingement plate to

certain thelocalreference temperature which is neededfor the heat

transfer evaluation. Fig. 3 shows the inline impingement pattern

used in the scope of the present work and the positions of the ther-

mocouples used for the data evaluation. Because the impingement

plate is symmetrical, only half of it is presented. These thermocou-

ples are placed directly in the center of the impinging hole at the

jet exits. The thermocouples are placed directly at the jet exits.

Grooves are milled into the wall of the impingement plate and thethermocouples along with their wires are glued into these grooves.

Nomenclature

A open area of exit rims (m2)c specific heat (J/(kg K))Cd discharge coefficient ()d impingement jet diameter, target plate thickness (m)Dd dimple diameter (m)

H heat transfer coefficient (W/(m2 K))h jet-to-plate spacing (m)k thermal conductivity (W/(mK))L target plate length (m)_m mass flow (kg/s)

n number of impinging jets ()N discrete interval ()Nu Nusselt number, based on nozzle diameter ()Nu spanwise averaged Nusselt number in x direction ()Nu area averaged Nusselt number ()p pressure loss (Pa)Pr Prandtl number ()Re Reynolds number, based on jet diameter ()R dimple radius (m)T temperature (C)

t time (s)

td dimple depth (m)x, y coordinate (m)X jet hole spacing in x direction (m)Y jet hole spacing in y direction (m)

Greek symbolsq density (kg/m3)H temperature ratio ()

Subscripts0 initial conditionB bulkcurv curved surfaced dimpledimpled dimpled plateflat flat platei indexI impingement plateO exit rimW wall

Y. Xing, B. Weigand / International Journal of Heat and Mass Transfer 53 (2010) 38743886 3875


3/13

A detailed view on the positioning of thermocouple is provided in

Fig.4. There areonly15 impingement holes placedwiththermocou-

ple and the diameter of the thermocouple is only 1/10 of that of theimpingement hole. Therefore, the influence of the thermocouples is

considerable negligible. Only Threeratios of the jet-to-plate spacing

(H/d = 3, 4 and 5) as well as three crossflow schemes (arranged by

changing the exit rim) are investigated as shown in Fig. 5. Three jet

Reynolds numbers in the experiments are 15,000, 25,000 and

35,000.

The dimples are manufactured in a staggered array as shown in

Fig. 6. The impinging positions are set to be onto the dimples. The

dimple diameter is Dd/d = 1.8 and dimple depth is td/Dd = 0.15.

2.2. Transient measurement technique

A transient method using thermochromic liquid crystals (TLC) is

applied for the measurement of the heat transfer [13]. The tran-sient measurement is based on a boundary condition of the third

kind for the one-dimensional solution of Fouriers equation with

a semi-infinite wall. Narrow bandwidth liquid crystals from Hall-

crest type (with an indication temperature of 31 C) are used inthe present work. To use TLCs for a temperature sensing on sur-

faces the exact relation between reflected wavelength and surface

temperature has to be identified. It is necessary to calibrate the

TLCs before they are applied in an experiment. The calibration unit

provides a 1D heat conduction enabling to easily calibrate the used

liquid crystal coatings. The maximum G-value is recognized by

using the maximum intensity method. For a detailed assessment

as to calibration process, the reader is referred to [14]. The crystals

are sprayed directly onto the target plate and covered with a coat-

ing of black paint to provide a uniform background for the image

acquisition. The target plate is observed from the outside with

two CCD video cameras. Unlike the flat plate, the dimpled plate

should be divided into three regions for the evaluation process,

which are a flat surface, a curved surface and the edge, as shownin Fig. 7.

Fig. 1. Sketch of the experimental setup.

Fig. 2. The impingement model and the axial positions of the static pressure taps.

3876 Y. Xing, B. Weigand / International Journal of Heat and Mass Transfer 53 (2010) 38743886


4/13

2.2.1. Flat surface

Initially, the target plate is at a uniform temperature (T0) all

over the surface and is suddenly exposed to the impinging air. Each

pixel location will reach the prescribed temperature (Tw) depend-

ing on the local heat transfer coefficient (h). The local heat transfer

coefficient over a surface coated with liquid crystals can be ob-

tained by using a 1D semi-infinite solid assumption for the test

surface. As shown by Wagner et al. [15], this assumption can be

considered valid as long as the condition at atd20 1

4is fulfilled. Here,

a is the thermal diffusivity of the material. With typical materialproperties of Plexiglas, the target plate thickness of d = 20 mm,

and a maximum testing time of t = 90 s the semi-infinite wall

assumption was fulfilled. So in the present study, all the measure-

ments were conducted in less than 90 s.

The 1D transient conduction equation is

ko2T

ox2 qc

oT

ot: 1

The boundary conditions are

t 0; T T0;

x 0; koT

ox h Tw TB ;

x ! 1; T T0:

Solving Eq. (1) with the boundary conditions, the dimensionless

temperature ratio at the convective boundary surface (at x = 0) is

obtained:

H TW T0TB T0

1 exp h2 tkqc

erfc h

ffiffiffiffiffiffiffiffit

kqc

s !: 2

This equation can be solved numerically to obtain the real heat

transfer coefficient for a measured wall temperature TW and the

time tby which this temperature is reached (indicated by the color

change of the liquid crystals). Eq. (2) is only valid for an ideal tem-

perature step within the flow. But in reality, the thermocouples re-

cord a time-dependant temperature evolution. To overcome this

problem, the temperature data are divided into a series of small

discrete intervals (0.2 s) as Fig. 8. Within these intervals the tem-

perature evolution can be considered to be a temperature step.

Thus Eq. (2) can be extended for a temperature evolution according

to the Duhamel principle [16]:

Fig. 3. The inline impingement pattern and positions of thermocouples.

Fig. 4. View of the detailed position of thermocouple.

Fig. 5. The crossflow schemes.



5/13

Fig. 6. The view and geometry of the dimpled plate.

Fig. 7. The different regions of the dimpled plate for the data analysis.

Fig. 8. Measured temperature evolution of a thermocouple.



6/13

TW T0 XNi1

H t ti TB;i TB;i1

; 3

where TB,i is the bulk-temperature at one specific time ti.

Solving Eq. (3) by using Newtons method, the heat transfercoefficient is determined iteratively. For a detailed assessment,

the reader is also referred to [14]. The temporal evolution of all

thermocouples is interpolated both spatially and temporally and

afterwards the film data and the bulk-temperature field in the

present study are evaluated by the program ProTeIn [14], which

has been developed at the ITLR.

2.2.2. Curved surface

Buttsworth and Jones [17] presented an approximate solution

of the 1D heat conduction equation for surface curvature:

H TW TBTB T0

1

1 nk2hcurvR

1 exp hcurv nk

2R 2

t

qcpk

!"

erfc hcurv nk

2R

ffiffiffiffiffiffiffiffiffiffiffit

qcpk

s !#: 4

Here, n = 1 for a cylinder and n = 2 for a sphere, the minis sign is

used for convex surfaces, and the plus sign is applied for concave

surfaces.

Fig. 9. The different regions of the dimpled plate for evaluation process.

Fig. 10. Comparison to literature for the maximum crossflow scheme.



7/13

For an infinite radius R, Eq. (4) would reduce to Eq. (3).

A Labview program is used to calculate the heat transfer coeffi-

cient with Eq. (4) inside the dimples by using the time file and tem-

perature field file which have been obtained by the program

ProTeIn.

Kotulla el al. [18] evaluated heat transfer coefficients on surface

with curvature with Eq. (2) and (4) and they found that using Eq.

(2) for surface with curvature can lead to an error up to 10%.

2.2.3. Edges

There should be heat exchanges between the flat surface and

curved surface, so the edge should not be evaluated as one dimen-

sional. Kanokjaruvijit and Martinez-Botas [9] made a comparisonof the averaged heat transfer coefficient with and without edges,

but they simply use a same edge area for each dimple. The results

showed that the heat transfer values illustrate no significant reduc-

tion when edges are taken out of consideration.

In the present study, we only use a one-dimensional method to

evaluate the heat transfer values on the target plate. A mask is used

to district the data on the flat surface or the dimples, as illustrated

in Fig. 9. The heat transfer coefficients on the flat surface are eval-

uated by using Eq. (2). For the dimples, Eq. (4) is used. A combina-

tion of the two data fields lead to the final local heat transfer

distribution.

2.2.4. Pressure loss

The pressure drop is measured by static pressure taps along thewall of the test section. The axial positions are sketched in Fig. 2.

Fig. 11. Spanwise averaged Nusselt number ratios on the flat and dimpled plate for jet-to-plate spacing 4, maximum crossflow and different Reynolds number.



8/13

The discharge coefficient Cd is generally used to evaluate the

pressure loss of flow systems. It is defined as

Cd _m

_mideal: 5

In the present work, the discharge coefficients of the impingement

plate and exit rims are defined for an incompressible flow as

Cd;I _m

q pnd2

4

ffiffiffiffiffiffiffiffiffiffiffiq

2DpI

r; 6

Cd;O _m

qA

ffiffiffiffiffiffiffiffiffiffiffiffiq

2DpO

r; 7

where DpI 1

2 p1 p2 1

4 p3 p4 p5 p6, and DpO 1

4 p3p4 p5 p6

12p7 p8.

2.3. Measurement uncertainties

The approach used for the measurement error analysis here isbased on the description by Kline and McClintock [19]. The accu-

racy of the measured heat transfer coefficient bases mainly on

the accuracy of the thermocouples, the calibration of the liquid

crystals and the time detection.

The accuracy of the Reynolds number depends on the accuracy

of the volume flow measured by the vortex meter, and on the range

of the static pressure sensors. The resulting uncertainty for the jet

Reynolds number is below 2.5%. For narrow band TLC, the transi-

tional temperature range is 1 C, and the typical uncertainty in

measuring this temperature is approximately 0.1 C [20]. TB and

T0 are measured with thermocouples. A thermocouple calibration

procedure showed that the error on temperature measurement is

below 0.2 C.

It should be noted, that the measurement uncertainties varywith the adiabatic wall temperature and therefore are different

Fig. 12. Local Nusselt number distribution on the flat and dimpled plate for jet-to-plate spacing 3, a Reynolds number of 35,000 and different crossflow scheme.



9/13

at every position at the target plate. Especially in the stagnation

point under the impinging jets, the lateral heat conduction will

influence the results. According to a method by Kingsley-Rowe

et al. [21] the maximum error of the heat transfer coefficient due

to lateral heat conduction has been calculated to be below 2%,

when the dimensionless temperature ratio H is in the range of

0:3 6 H 6 0:7 .

The total measurement uncertainty for the heat transfer coeffi-

cient is below 9% for all experiments which have been carried out

in this study. The accuracy of the pressure measurements depends

on the range of the pressure sensors. The maximum error of the

pressure measurements is below 2.5%.

3. Results and discussion

3.1. Area averaged Nusselt number on a flat plate

Before beginning with the heat transfer results of jet impinge-

ment for the dimpled plate, a baselinecaseon the flat plate is exam-

ined and compared with the literature as Metzger et al. [22],

Florschuetz et al. [23], Son et al. [24], El-Gabry and Kaninski [25]and Parket al. [26]. Fig. 10 shows thecomparison of thepresent data

and the published results. All experiments were conducted for the

case of maximum crossflow, where the flow exits in only one direc-

tion. A good agreement is found. Although the impinging plate

parameters aredifferent from each case, the dependence of thearea

averaged Nusselt number on the jet Reynolds number is in general

common. As expected, the heat transfer is higher at the lower jet-

to-plate spacing and higher Reynolds numbers.

3.2. Influence of Reynolds number

The jet Reynolds number has similar effect on impingement

heat transfer for all three flow orientations. An increase in jet Rey-

nolds number increases the local heat transfer coefficient through-

out the target plate. While in order to normalize the heat transfer

value, the Nusselt number has been referred to Re0.8Pr1/3.

We can see that the spanwise averaged Nusselt number ratios

are nearly the same for different Reynolds numbers on the flat

plate and on the dimpled plate, as shown in Fig. 11. It means that

the heat transfer characteristics can be scaled with Re0.8 typical for

a turbulent boundary layer. The heat transfer coefficient is slightly

higher with higher Reynolds number near the impinging zone.Note that as the jet-induced crossflow develops, the peak positions

Fig. 13. Spanwise averaged Nusselt number on the flat and dimpled plate for jet-to-plate spacing 3, a Reynolds number of 35,000 and different crossflow schemes.



10/13

on the flat plate can be seen to shift slightly downstream as the in-

creased crossflow displaces the jets. For the dimpled plate, the

peak position shifts because of the crossflow and the disturbances

by dimples. Hence, we can see that the shift distance on the dim-

pled plate is larger than that on the flat plate.

For the flat plate, the heat transfer coefficients increase slightly

as the jet-induced crossflow increases at first. The highest value is

reached in the stagnation point at the jet row 4, while afterwards

as the increased crossflow, the jets could not reach the plate easilybut mix with the crossflow first. We could see there is a significant

downstream degradation in heat transfer due to maximum jet-in-

duced crossflow in the last two jet rows. For the dimpled plate, the

peak values are more homogenous comparing to the value on the

flat plate. In the region near the exit rims, the heat transfer coeffi-

cients are higher than those on the flat plate.

3.3. Influence of crossflow schemes

Fig. 12(a) shows the local Nusselt number distributions on thetarget plate for a Reynolds number of 35,000, jet-to-plate spacing

Fig. 14. Spanwise averaged Nusselt number on the flat and dimpled target plate for different jet-to-plate spacing.



11/13

H/d = 3 and maximum crossflow, where the spent air exhausts only

in one direction. Because of the symmetry, only one jet row has

been analyzed. For the impingement on the flat plate, the positions

of the impingement jets are clearly visible in the heat transfer pat-

tern on the target plate. After the jet impinges on the target plate,

the heat transfer rates are very high but decrease quickly towards

the side. It indicates that the heat transfer performance in the

impinging zone increases first with increasing the jet-induced

crossflow, and it decrease suddenly in the last two jets due to

the influence of crossflow confining the jet and reducing its cover-

age. For the dimpled plate, the positions of the dimples which lo-

cated between the impinging jets are quite clear in the heat

transfer pattern because the values inside are lower compared to

the surrounding data. The dimples which are located just at the

jet impinging positions appear to produce little effect on the jet

impingement for initial rows of jets. Further downstream as cross-

flow develops, the jets are pushed towards the edges or even

downstream of dimples which can augment local Nusselt number

distributions, especially downstream of the dimple. With the pres-

ence of the dimples, the spent air separates in a higher degree than

that caused by jet impingement on a flat plate. When the velocity

of the spent flow consequently increases, it helps to relieve the

heat transfer degradation that occurs mentioned before.

Fig. 12(b) shows the local heat transfer distribution for a Rey-

nolds number of 35,000, jet-to-plate spacing H/d = 3 and medium

crossflow, where the outlet holes are arranged in two opposite

directions. The heat transfer performance in the impinging zone in-

creases with an increasing of jet-induced crossflow on the flat

plate. For the dimpled plate, Nu values are generally higher over

the central portion of the target plate than near the two exhaust

Fig. 15. Area averaged Nusselt number on the dimpled plate.

Fig. 16. Normalized area averaged Nusselt numbers for different crossflow schemes and different arrangements.



12/13

openings because the dimples can only augment local Nusselt

number distributions downstream of the dimple as the jet-induced

crossflow increases. The heat transfer coefficients around the edgesof the dimples are higher due to a thinner boundary layer around

the edges, and it is in agreement with that of Azad et al. [5].

Fig. 12(c) shows the local Nusselt number distributions on the

target plate for a Reynolds number of 35, 000, jet-to-plate spacing

H/d = 3 and minimum crossflow. Because of the symmetry, only

theup-rightquarter of thetarget plate is tracedto take heat transfer

evaluations. We can seethat there is strongenhancement causedby

thedimplesunder thejets,while thedimples between thejets cause

the local minima value. The jets impinge on the dimples with the

idea of employing the edges of dimples to thin the boundary layer,

and enhance the heattransfer.For maximum and medium crossflow

schemes,the edges areonly helpfulin thedownstream side. For min-

imum crossflow, we can see the heat transfer enhancement caused

by the edges appears all round of dimples. So it causes much higherenhancement.

For a quantitativecomparison,we compare the spanwise averaged

valuewith different crossflow schemes asshownin Fig.13.Fortheflat

plate, the value for maximum crossflow is lower compared to the

other schemes,especiallyin theregion near theexit rims. Forthe dim-

pled plate, theheat transfer values for each jet roware more homog-

enous on the whole plate no matter of the crossflow scheme. The

dimplesmakethe influence of thecrossflow lower.And theminimum

crossflow reach the highest heat transfer performance.

3.4. Influence of jet-to-plate spacing

The jet-to-plate spacing strongly affects the heat transfer per-formance on both the flat and dimpled plate. For the flat plate,

the narrow H/d lead to higher heat transfer. So does the dimpled

plate as shown in Fig. 14 that H/d = 3 results in the highest heat

transfer value. For the maximum crossflow schemes, the difference

between jet-to-plate spacing (H/d) 3 and 4 occurs only in the

downstream region on the flat plate. While on the dimpled plate,

the jet-to-plate spacing H/d = 3 gets always higher heat transfer va-

lue on the whole plate. A sudden increase on the dimpled plate is

reached by the jet-to-plate spacing 3 in the last four jet rows. For

the medium crossflow schemes, the heat transfer values of jet-

to-plate spacing 3 and 4 are quite similar no matter on the flat or

dimpled plate. While jet-to-plate spacing H/d = 5 always gets the

lowest heat transfer coefficient. For minimum crossflow, the heat

transfer coefficient is nearly the same with different jet-to-platespacings on the flat plate. However a significant improvement is

found at the narrow jet-to-plate spacing H/d = 3 on the dimpled

plate.

3.5. Influence of dimples

Fig. 15 shows the area averaged Nusselt number on the dimpled

target plate for a Reynolds number of 35,000 and for different

arrangements. It can be seen that the area averaged Nusselt num-

ber decreases with an increasing jet-to-plate spacing for any cross-

flow scheme. The heat transfer exchange is always better for the

minimum crossflow for different jet-to-plate spacings.

Fig. 16 shows the normalized area averaged Nusselt numbers

for different crossflow schemes. The heat transfer enhancement ra-

tio increases with increasing Reynolds numbers. The narrow jet-to-

plates spacing H/d = 3 results in the highest heat transfer enhance-

ment for different crossflow schemes. For maximum crossflow

scheme, the application of dimples leads to an increase in the heat

transfer coefficient by 6.2%. For medium crossflow scheme, the

present of dimples do not help, but reduce the heat transfer by

10%. The highest enhancement, caused by the minimum crossflow

scheme for the jet-to-plate spacing H/d = 3, is 12.3% for higher Rey-

nolds number Re = 35,000. Note the surface area of dimpled target

plate is 26.4% higher compared to that of the flat target plate. This

increased area will further increase the total heat transfer.

Impingement onto dimpled plate performs best for the mini-

mum crossflow scheme and narrow jet-to-plate spacing due to

the fully usage of the dimple edge to make the boundary layer

thinner and a higher crossflow velocity. Jet impingement on a dim-

pled plate can be thought as a coupled effect of jet impingement

and channel flow caused by the spent air. For maximum crossflow

scheme, heat transfer of the channel flow part is much enhancedbecause of the dimples, so the heat transfer performance in the

downstream part is even better than that of upstream part. The

enhancement ratios decrease with increasing jet-to-plate spacings.

For medium crossflow scheme, the heat transfer values are re-

duced by the dimples because of the less crossflow and one exit

opening direction. The recirculation flow which occurs inside the

dimples could not escape fast enough from each dimple without

the powerful channel flow.

3.6. Pressure loss

Fig. 17 shows the dischargecoefficients of the impingement plate

and exit rims for different arrangements. We can see that the dis-charge coefficients of the impingement plate are nearly the same

Fig. 17. Discharge coefficients of the impingement plate and exit rims for different arrangements.



13/13

on the flat and dimpled plate and they are nearly independent of

Reynolds numbers, jet-to-plate spacing and crossflow schemes.

The discharge coefficients of the exit rims for the dimpled plate

are slightly lower than those for the flat plate. The discharge coef-

ficients of the exit rims are also nearly independent of Reynolds

numbers. The discharge coefficients of the exit rims for the flat

plate decrease with increasing the jet-to-plate spacings no matter

for the flat or dimpled plate. For minimum crossflow, the dischargecoefficient of the exit rims for the dimpled plate is lower than that

for the flat plate. It might be the fully usage of the edge of dimple.

However for the medium crossflow and maximum crossflow

schemes, the values are similar compared to those for the flat

and dimpled plate. This is thought due to the vortices produced

form the dimples do not significantly cause the difference from

the case when the flat plate is employed.

4. Conclusions

The heat transfer characteristics in an inline impingement mod-

el with high Reynolds number on a flat and dimpled plate are

investigated in the present work. Nusselt number distributions

have been measured on the target plate using a transient liquid

crystal technique. The jet exit temperature are interpolated both

spatially and temporally to determine the reference temperature

for the calculation of the heat transfer performance. The jet-to-

plate spacing (H/d = 3) is better than the others either on the flat

or dimpled target plate. The heat transfer performance on the dim-

pled plate is always better for the minimum crossflow for different

jet-to-plate spacing. The heat transfer enhancement ratio increases

with increasing Reynolds numbers. The narrow jet-to-plates spac-

ing H/d = 3 results in the highest heat transfer enhancement for dif-

ferent crossflow schemes. The highest enhancement ratio is up to

12.3%. The discharge coefficients of the impingement plate and exit

rims are similar for different arrangements. Some difference occurs

for the discharge coefficients of the exit rims for minimum

crossflow.

References

[1] M.P. Boyce, Gas Turbine Engineering Handbook, second ed., Gulf Publishing,USA, 2001.

[2] B. Han, R.J. Goldstein, Jet-impingement heat transfer in gas turbine systems,Ann. NY Acad. Sci. 934 (2001) 147161.

[3] K. Kanokjaruvijit, R.F. Martinez-Botas, Jet impingement on a dimpled surfacewithdifferent crossflowschemes, Int.J. Heat MassTransfer 48 (2005) 161170.

[4] C. Gau,C.M. Chung,Surfacecurvature effect on slot-air-jetimpingementcoolingflow and heat transfer process, ASME J. Heat Transfer 113 (1991) 858864.

[5] G.S. Azad, Y. Huang, J.C. Han, Impingement heat transfer on dimpled surfacesusing a transient liquid crystal technique, J. Thermophys. Heat Transfer 14(2000) 186193.

[6] S.V. Ekkad, D. Kontrovitz, Jet impingement heat transfer on dimpled targetsurfaces, Int. J. Heat Fluid Flow 23 (2002) 2228.

[7] K. Kanokjaruvijit, R.F. Martinez-Botas, Jet impingement onto a dimpled surfacewith different crossflow schemes, in: Proceedings of the International GasTurbine Congress, Tokyo, IGTC2003, Ts-074, 2003.

[8] K. Kanokjaruvijit, R.F. Martinez-Botas, Heat transfer and pressure investigationof dimple impingement, in: Proceedings of GT2005 ASME Turbo Expo 2005,

GT2005-68823.[9] K. Kanokjaruvijit, R.F. Martinez-Botas, Parametric effects on heat transfer of

impingement on dimpled surface, ASME J. Turbomach. 127 (2005) 287296.[10] K. Kanokjaruvijit, R.F. Martinez-Botas, Heat transfer and pressure investigation

of dimple impingement, ASME J. Turbomach. 130 (2008) 011003-1011003-11.

[11] C.S. Woei, J.Y. Jena, C.S. Fei, Heat transfer of impinging jet-array over convex-dimpled surface, Int. J. Heat Mass Transfer 49 (2006) 30453059.

[12] C.S. Woei, H.F. Liou, Heat transfer of impinging jet-array onto concave- andconvex-dimpled surfaces with effusion, Int. J. Heat Mass Transfer 52 (2009)44844499.

[13] P.T. Ireland, T.V. Jones, Liquid crystal measurements of heat transfer andsurface shear stress, Meas. Sci. Technol. 11 (7) (2000) 969986.

[14] R. Poser, J. von Wolfersdorf, E. Lutum, Advanced evaluation of transient heattransfer experiments using thermochromic liquid crystals, in: SeventhEuropean Conference on Turbomachinery, Athens, Greece, 2007, pp. 793801.

[15] G. Wagner, M. Kotulla, P. Ott, B. Weigand, J. von Wolfersdorf, The transientliquid crystal technique: influence of surface curvature and finite wallthickness, in: Proceedings of ASME Turbo Expo 2004, GT2004-53553, Vienna,Austria.

[16] W.M. Kays, M.E. Crawford, B. Weigand, Convective Heat and Mass Transfer,fourth ed., McGraw-Hill International Editions, NY, 2004.

[17] D.R. Buttsworth, T.V. Jones, Radial conduction effects in transient heat transferexperiments, Aeronaut. J. 101 (1997) 209212.

[18] M. Kotulla, G. Wagner, P. Ott, B. Weigand, J. von Wolfersdorf, The transientliquid crystal technique: influence of surface curvature and finite wallthickness, ASME J. Turbomach. 127 (2005) 175182.

[19] S.J. Kline, F.A. McClintock, Describing uncertainties in single-sampleexperiments, J. Mech. Eng. (1953) 38.

[20] Y. Yan, J.M. Owen, Uncertainties in transient heat transfer measurements withliquid crystal, Int. J. Heat Fluid Flow 23 (2002) 2935.

[21] J.R. Kingsley-Rowe, G.D. Lock, J.M. Owen, Transient heat transfermeasurements using thermochromic liquid crystal: lateral-conduction error,Int. J. Heat Fluid Flow 26 (2005) 256263.

[22] D.E. Metzger, L.W. Florschuetz, D.I. Takeuchi, R.D. Behee, R.A. Berry, Heattransfer characteristics for inline and staggered arrays of circular jets withcrossflow of spent air, ASME J. Heat Transfer 101 (1979) 526531.

[23] L.W. Florschuetz, C.R. Truman, D.E. Metzger, Streamwise flow and heat transferdistributions for jet array impingement with crossflow, ASME J. Heat Transfer103 (1981) 337342.

[24] C. Son, D. Gillespie, P.T. Ireland, Heat transfer and flow characteristics of anengine representative impingement cooling system, in: Proceedings of ASMETurbo Expo 2000, GT2000-219.

[25] L.A. El-Gabry, D.A. Kaninski, Experimental investigation of local heat transferdistribution on smooth and roughened surfaces under an array of angledimpinging jets, ASME J. Heat Transfer 127 (2005) 532544.

[26] J. Park, M. Goodro, P. Ligrani, M. Fox, H.K. Moon, Separate effects on Machnumber and Reynolds number on jet array impingement heat transfer, ASME J.Turbomach. 129 (2007) 269280.


experimental investigation of impingement heat transfer on a flat and dimpled stuttgart2011

Documents