lable at ScienceDirect

Applied Thermal Engineering 49 (2012) 118e123

Contents lists avai

Applied Thermal Engineering

journal homepage: www.elsevier .com/locate/apthermeng

Investigation of heat transfer and fluid flow behavior between straightand inclined fins in tall duct

M.D. Islam a,*, K. Oyakawa b, I. Kubo a

aDepartment of Mechanical Engineering, The Petroleum Institute, PO. Box 2533, Abu Dhabi, United Arab EmiratesbUniversity of the Ryukyus, Okinawa, 903-0213, Japan

a r t i c l e i n f o

Article history:Received 30 October 2010Accepted 25 July 2011Available online 7 August 2011

Keywords:Tall ductRectangular finsFlow visualizationLongitudinal vortexHeat transfer enhancement

* Corresponding author. Tel.: þ971 2 607 5230; faxE-mail addresses: dislam@pi.ac.ae, mdislam02@ya

1359-4311/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.applthermaleng.2011.07.044

a b s t r a c t

This paper describes the flow behavior and flow patterns developed by straight and inclined fins andtheir affect on heat transfer characteristics. A detailed experimental investigation of the heat transfer andflow characteristics of finned surfaces was conducted for airflow (Re ¼ 3824e12,747) in a tall ductcorresponding to 200 mm height. In this experiment short rectangular fins were attached in 7 � 7 arraysto a heating surface and exposed to airflow. T-type thermocouples and an infrared camera witha 160 � 120-point IneSb sensor were used to measure the wall temperature and to get the detailed heattransfer coefficient over the endwall and fin base. A thermal image and the iso-heat transfer coefficientcontour give a complete picture of the heat transfer characteristics of the endwall surface. Smoke flowvisualization reveals the longitudinal vortex generated by the inclined fins which significantly enhancesthe heat transfer in the inter-fin region and the fin surfaces. From the time averaged velocity profiles andspanwise velocity distributions a non-continuous curve caused by the detachment or reattachment ofthe flow was observed which confirms the existence of a longitudinal vortex in the case of inclined fins.But the horseshoe vortex appeared for both straight and inclined fins. The Nusselt number shows thatheat transfer enhancement at a factor of more than three times than the finless duct is achieved for theinclined fins whereas straight fins could reach upto an enhancement of more than two times.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Thermal management, or temperature control, is a greatconcern for heat generating bodies, for example, power plants,nuclear reactors, engines, motors, heat exchangers, process indus-tries, generators, electronic equipment, and digital computers, evenin space craft. If the heat generated in a machine is not removed ata sufficient rate, some problems, including breakdown, can takeplace in the machine due to overheating. This type of problem canonly be overcome by a more efficient heat transfer. Fins are widelyused as the primary means of heat exchange in devices. The needfor more efficient cooling techniques in devices has recentlyprompted study into heat transfer and flow characteristics ofvarious configurations of finned surfaces. The pin fin is a typicalconfiguration that is often used to cool the trailing edge region ofturbine blades; the internal passage of turbine blades can be verynarrow, so the choice of cooling scheme is limited. Recently, aninclined rectangular fin attached to the endwall was found to be aneffective vortex generator for heat transfer enhancement, wherein

: þ971 2 607 5200hoo.com (M.D. Islam).

All rights reserved.

a longitudinal vortex is produced with intensity far downstream. Itis expected that the heat transfer from the endwall and the finsurface can be improved and hence we have identified thisconfiguration as being very promising. Heat transfer enhancementin arrays of rectangular blocks has been investigated by manyresearchers. Sparrow et al. [1,2] studied the heat transfer andpressure drop characteristics of arrays of rectangular modulescommonly encountered in electronic equipment. In their experi-ment, heat transfer enhancements exceeding a factor of two wereachieved by the use of multiple fences like barriers, with the interbarrier spacing and the barrier height being varied parametricallyalong with the Reynolds number. Igarashi [3] studied heat transferfrom a square prismwith different inclination angles and observedreattachment flow for inclination angles of 14�e35�. Turk andJunkhan [4] measured the spanwise heat transfer downstream ofa rectangular fin mounted on a flat plate. Oyakawa et al. [5] studiedthe heat transfer of plate setting rectangular fins with an angle of20� and showed that this configuration can enhance heat transfer.El-Saed et al. [6] investigated heat transfer and fluid flow withrectangular fin arrays and found that the mean Nusselt numberincreases with an increasing Reynolds number, inter-fin space andfin thickness, but they did not examine the endwall heat transfer.

M.D. Islam et al. / Applied Thermal Engineering 49 (2012) 118e123 119

Bilen and Yapici [7] investigated heat transfer enhancement froma surface fitted with rectangular blocks at different inclinationangles and found that the maximum heat transfer was obtained atan angle of 45� while the effect of the inclination angle is small forlarger than 22.5�, though the local heat transfer characteristics andflow behavior were not reported. Besir et al. [8] conducted anexperimental investigation using particle image velocimetry of theflow structure in the flow through a rectangular duct and con-taining a circular cylinder with axis normal to the flow. They ob-tained the time averaged velocity vectors map, patterns ofstreamlines, and corresponding vorticity contours using instanta-neous images. Fu and Rockwell [9] experimentally investigated thehorseshoe vortex system and wake flow regions around a verticalcylinder mounted on a flat plate in shallow-water flow, examiningthe development of horseshoe vortex systems and the convectionof these vortices further downstreamwhich interact with sheddingvortices close to the base plate surface. Yakut et al. [10] experi-mentally investigated the effects of heights and widths of hexag-onal fins in streamwise and spanwise distances between fins andflow velocity on their thermal resistance and pressure drop char-acteristics. Li and Chen [11] used infrared thermography techniqueto investigate thermal performance of plate-fin heat sinks underconfined impinging jet conditions and discussed the influence ofReynolds number and geometry.

In forced convection, flow behavior has greater influence onheat transfer characteristics. So it is important to know the flowbehavior with heat transfer. Thus in this experiment we haveinvestigated the flow behavior and flow structures for both straight

Fig. 1. Experimental apparatus for heat transfer measurement (a) Infrar

and inclined fins and observed in detail how the flow affects theheat transfer in tall ducts.

2. Experimental apparatus and procedure

In this experiment mainly two types of experimental setup wereused, one for the heat transfer experiment and the other for smokeflow visualization as well as for the velocity boundary layer. Exper-iments were conducted in a long tall rectangular duct of 230 mmspanwidth, 784 mm length and of 200 mm height. The first row offins was located 200 mm away from the duct entrance. The experi-mental apparatus with rectangular fins is shown in Fig. 1. The finswere made of aluminum and were rectangular with dimensions20mm long, 5mm thick and 10mmheight. Thefinswere set in linesand rows of seven each. Fin spacing in streamwise and spanwisedirectionwasfixed to40mmand20mmrespectively. The inclinationangle, a of the fins to the flow direction was kept 0� and 20�. Main-streamvelocity at the entrance of the duct as well as local velocity inspanwise directions was measured with a pitot tube and a digitalmicromanometerwith an accuracy of�0.1mmH2O. Tomeasure thelocal heat transfer coefficients, the upperwall of theductwas formedwith a Bakelite plate of thickness 10 mm and a 30 mm, and thickstainless steel foil was attached to the inner side of the duct wall asa heating surface. Then, a window (80 mm � 160 mm) was built atthe upper ductwall to see the back side of thefinned heating surface.A uniform heat flux was formed at the heating surface by supplyinga direct current source. During the experiment, the temperaturedifferencebetween themainstreamand theheating surfacewaskept

ed image technique and (b) Types of fins (All dimensions in mm).

a

M.D. Islam et al. / Applied Thermal Engineering 49 (2012) 118e123120

constant and the heat flux was varied. The infrared image techniquewas applied in order to observe the distributions of the temperatureand the heat transfer coefficients on the overall surface including thefin base and endwall. Infrared images were taken using an infraredcamera with an indium-antimony (IneSb) sensor, which canmeasure the temperature at 160 � 120 points with a resolution of0.025 �C for a black body. In this case, the temperaturewasmeasuredat the back of the heating surface through the window which wascovered by polyvinyledene film with a transmissivity for infraredenergy nearly unity. The back of the heating surface in the windowarea was painted black and the whole experimental apparatus wascovered by black curtain to ensure completely dark surroundings.The mainstream temperature was calculated by averaging thetemperatures of the two T-type thermocouples positioned at themiddle of the duct height as shown in Fig. 1. Detailed information ofthermocouple locations on the heating surface as well as the Pitottube locations forflowmeasurements are shown in Fig. 2. To preventheat loss from the heating surface attached to the bakelite plate, theoutside of the upper duct (bakelite plate) wall was insulated bya layer of foamwith a thickness of 60mmexcept for thewindowareathrough which the infrared image was to be taken. A centrifugal(NIKO TPH) blower was used to supply the air in the duct and theblower specificationwas:flow rate 45m3/min, pressure 350mmAq.,power 5.5 kW and 2830 rpm. The velocity profile at the inlet of theduct was uniform with a thin boundary layer and the turbulenceintensitywas 3%ofmainflowvelocity at duct inlet. The experimentalapparatus for the smoke flow visualization consisted of a duct madeof transparent acrylic sheet, metal halide light, high speed digitalvideo camera (Phantom v7.1) and a smoke generator.

3. Data reduction and uncertainty analysis

The total heat generated from the heating surface is distributedinto the heat transferred by convection to the flowing air, and heatlosses through the insulation and the window. Heat losses from theback of the heating surface except for the windowmay be assumedto be very little as the bakelite plate was insulated by 60 mm thickinsulator. It was found that the heat loss from window was 1.3% ofthe heat flux supplied to the heating surface. So its value is so smallcompared to the heat input value that it can be neglected. Conse-quently the total heat transferred by convection to the flowing airequals to the heat flux supplied to the heating surface.

The heat transfer coefficients,

h ¼ _q=ðtw � tNÞ (1)

Fin PositionThermocouples position,

0.07 mm

Heating surface

Velocity measurement points Fin Polished aluminum

Pitot tube a

b

Fig. 2. Measurement locations in a duct in streamwise direction for (a) velocity and (b)temperature profile.

were obtained at a representative region, where tw is the walltemperature and tN the is the mainstream temperature.

The pressure was measured by a digital micro manometer ofaccuracy � 0.01 mm of H2O. The temperature was measured byboth the thermocouples using a data logger and the infraredcamera. The percentage relative uncertainty in the measuredtemperature for the thermocouples and the infrared camera were�0.25% and�1.3% respectively. The percentage relative uncertaintyin the measured electric power input was �1.4%. The percentagerelative uncertainty in the compound variables was found to be�0.34% both for the velocity and the Reynolds number and �2.7%both for the heat transfer coefficient and Nusselt number.

4. Results and discussions

4.1. Flow pattern analysis

In order to get general information about the flow patternaround the rectangular fin, time averaged velocity profiles and thevelocity profile between fins in spanwise direction, for both thestraight fin (a ¼ 0�) and the inclined fin (a ¼ 20�) are measuredusing pitot tube. Fig. 3 shows the time averaged velocity profilesjust behind the fin ends of the 1st, 2nd, 3rd and 4th rows fromupstream in order. The mainstream velocity is maintained atU ¼ 15 m/s. In the case of straight fins, the velocity is increasedwith the distance from the wall surface, and its thickness ofboundary becomes larger at more downstream rows, and achievesover 15 mm of height just behind the 4th row. Near the wall, theprofiles show some non-continuous curve caused by the detach-ment or reattachment of the flow. On the other hand, the velocityprofile for the inclined fins is complex one, especially when thevelocity just behind the 2nd row increases as it separates from the

b

Fig. 3. Time averaged velocity profile for (a) straight fin (a ¼ 0 deg) and (b) inclined fin(a ¼ 20 deg).

a

b

Fig. 5. Spanwise velocity profile at Y ¼ 10 mm for (a) straight fin (a ¼ 0 deg) and (b)inclined fin (a ¼ 20 deg).

M.D. Islam et al. / Applied Thermal Engineering 49 (2012) 118e123 121

wall, attains its maximum, and then decreases. After hittingminimum values, the velocity profile is developing. The velocityprofiles at the 3rd and 4th row resemble the pattern of the flowseparated from the wall; therefore its boundary layer thicknessbecomes over 20 mm. The inclination angle affects the flowbehavior and causes to form many vortices; consequently, heattransfer is enhanced.

Figs. 4e6 shows the velocity distributions in spanwise direc-tions. In Fig. 4(a), near the wall, the velocity at wake flow of the firstrow fin is very low and then increases in spanwise direction andkeeps an almost constant velocity in the region between fins wherethe flows do not interfere with each other. Behind the 2nd, 3rd and4th row fin, the velocity profile shows a concave pattern at thecenter position between fins and its value is larger for2nd row finscompared to others. This is because the velocity decreases as itapproaches downstream due to the development of boundary layerof fluid in the normal towall surface. On the other hand, for inclinedfins, the velocity is also low: it is the same as straight fins in thewake flow region, however it is rapidly increased at z ¼ 5 mm dueto intensive shear layer separation from the first row fin, and itapparently decreases at range of z¼ 15e20 mmwhich correspondsto a region of the back of the fin. At 10 mm (shown in Fig. 5) and14 mm (shown in Fig. 6) from the wall, the profiles show similarpatterns regardless of streamwise positions. Just behind of fin, theend dead zone of flow appears, and the velocity is very low in thisregion. At Y ¼ 5 mm from the wall, since the measurement point isat half height of fin, the velocity profile is deflected by the mainvortex flow over the inclined fin, acting as the main vortex.AtY ¼ 10 mm, the measurement for 1st row still shows the formationof a longitudinal vortex. The flow developed by inclined fins at andabove z ¼ 14 mm shows similar behavior as that of the straight finobserved earlier with some differences.

a

b

Fig. 4. Spanwise velocity profile at Y ¼ 5 mm for (a) straight fin (a ¼ 0 deg) and (b)inclined fin (a ¼ 20 deg).

4.2. Flow visualization

In order to observe the flow behavior more clearly, specificallyto identify the vortex formation by the straight and inclined fins,the flow was visualized by a metal halide beam light using smoke

25

20

15

10

5

0

z [

mm

]

20 15 10 5 0 u [ m/s ]

1 st row

2 nd

row

3 rd

row

4 th

row U = 15 m/s

25

20

15

10

5

0

z [

mm

]

2015 10 5 0 u [ m/s ]

1 st row

2 nd

row

3 rd

row

4 th

row U = 15 m/s

a

b

Fig. 6. Spanwise velocity profile at Y ¼ 14 mm for (a) straight fin (a ¼ 0 deg) and (b)inclined fin (a ¼ 20 deg).

150

100

50

0

h x [

W/m

2 K]

20151050-5-10X/L

centerlineReL = 2.54 x 104

1.27 x 104

6.37 x 103

Fig. 8. Local heat transfer coefficient distributions of a smooth surface in duct.

M.D. Islam et al. / Applied Thermal Engineering 49 (2012) 118e123122

supplied at the entrance of the duct as a tracer. The flow stagnatesat the front of the fins and horseshoe vortexes are formed for bothcases of straight and inclined fins as shown in Fig. 7. But, formationof a longitudinal vortex appears only in the case of inclined fins.The main longitudinal vortex is formed by the side edge of theinclined fins which then travels to the inter-fin region and isreattached to the endwall just front of the following fin andsweeps thereby. This causes the heat transfer enhancement fromthe endwall.

4.3. Detailed heat transfer analysis

To check the accuracy of the heat transfer measurements, thelocal heat transfer coefficients on the smooth duct surface weremeasured. Fig. 8 presents the streamwise distributions of the localheat transfer coefficient along the endwall from the duct entrancefor U¼ 5, 10 and 20 m/s. The distributions gradually decrease in thestreamwise direction. Since the duct height was large, both thehydrodynamic and thermal boundary layers were still developing,even at the test section. Due to the rough surface at the duct inlet,acting as a tripping surface, the flow was turbulent. The localNusselt number, based on hx and the distance, X from entranceagrees with the equation

Nux ¼ 0:0296 Re0:8x Pr0:6 (2)

for a turbulent boundary layer on a flat plate, as shown by the solidlines in Fig. 8. In the figure, the open and solid symbols designatedistributions along the centerline and the midline, respectively.Both sets of data are equal regardless of spanwise position, indi-cating the two dimensionality of the heating surface. The datashows good agreement with Eq. (2). Fig. 9 shows the thermal imageobserved by infrared camera TVS 8000. This figure indicates howthe flow patterns, especially longitudinal vortex, horseshoe andmain flow, affect heat transfer characteristics on the endwall. Asa typical example, detailed heat transfer distributions around therepresentative fin rows of straight and inclined fins are shown inFig. 9(a) and (b). Here the thermal images are shown for the samescale. In general, for both cases of straight and inclined fins, the

Fig. 7. Smoke flow visualization with (a) straight fins and (b) inclined fins.

infrared images and contour show that the higher heat transferregions are at the fin base and adjacent to the fins. This is because ofthe extended surface effect. But comparatively larger and higherheat transfer regions and contour appear due to the inclined fins.This is because flow touches the fin surfaces strongly and so theextended surface effect is higher. Longitudinal vortexes generatedby the inclined fins touch the inter-fin region as well as fin surfacesand thereby enhance heat transfer. Horseshoe vortexes formed atthe front of the fins for both cases, and so the higher heat transfer isalso achieved in that region. Behind the fins, higher heat transferregions also appeared as the corner edges create more turbulence.The same phenomenon is also observed among the successive finrows. No significant hot spots are visible in any of the thermalimages.

4.4. Area-averaged heat transfer coefficient

The area-averaged heat transfer coefficients at the endwall andfor the overall surfaces of the representative 3rd, 4th and 5th row

Fig. 9. Thermal image and contour of the representative fin region of the endwall with(a) straight and (b) inclined fins.

89

10

2

3

4

5

678

Nu

2 3 4 5 6 7 8 9

1042

Re

straight, endwallstraight, overall inclined, endwall inclined, overall

Fig. 10. Relation between Nusselt number and Reynolds number for both straight andinclined fins.

M.D. Islam et al. / Applied Thermal Engineering 49 (2012) 118e123 123

fins were measured from infrared images with the followingequation:

hoverall ¼Afin

Aoverallhfin þ Aendwall

Aoverallhendwall (3)

Here, Aoverall is the overall surface area, and Afin and Aendwall arethe fin base and the endwall area respectively. Average Nusseltnumbers are calculated based on fin length and the average heattransfer coefficient. Since the Nusselt number is based on theaverage heat transfer coefficient of the projected surface area, it willreflect the effect of variation on the surface area, and flow behaviorformed by straight and inclined fins. Fig. 10 shows the character-istics of the area-averaged Nusselt number versus Reynoldsnumber for the overall surface area and for the endwall for bothtypes of straight and inclined fins. The Nusselt number increaseswith the Reynolds number for both cases. The Nusselt number forthe endwall surface as well as for the overall surface area for theinclined fins is apparently larger than that of straight fins. This isexpected because of the large flow interactions and turbulencecaused by the longitudinal vortex in case of inclined fins. It is alsoshown that inclined fins achieved a significant heat transferenhancement at a factor of more than three times than the finlessduct whereas straight fins can reach to an enhancement of morethan two times.

5. Conclusions

In this investigation flow behavior and its affect on heat transfercharacteristics were examined with straight and inclined fins ina tall duct case. After experiments we can summarize as follows:

� Development of hydrodynamic boundary layers are clearlyobserved from the time averaged velocity profile and spanwisevelocity distributions. In addition to general information, somenon-continuous curves caused by the detachment or reat-tachment of the flow are also noticed, which indicates theexistence of longitudinal vortex.

� The formation of a longitudinal vortex by the inclined fins isvisible in smoke flow visualization that enhances heat transferfrom the inter-fin region and fin surface. This longitudinalvortex was not present in the case of straight fins. But thehorseshoe vortex is observed for both straight and inclined fins,which causes heat transfer augmentation from front of the finsand near the fin base.

� Thermal images and the iso-heat transfer coefficient contoursgive a complete picture of the heat transfer characteristics ofthe endwall surface.

� The NusselteReynolds relationship shows that heat transferenhancement achieved in both of the cases while a significantenhancement at a factor of more than three times than thefinless duct is achieved by the inclined fins.

Acknowledgments

This research work is supported by the HEIWA NakajimaFoundation of Tokyo, Japan and the Petroleum Institute of AbuDhabi, UAE. This support is gratefully acknowledged.

Nomenclature

hx localheat transfer coefficient, W/m2Kh area average heat transfer coefficient, W/m2KL fin length, 20 mmNu area-averaged Nusselt number (hL=l)Nux local Nusselt number (hx/l)Re Reynolds number (UL/n)U mainstream velocity, m/su local air velocity, m/sX* streamwise coordinate (X*¼ 0 at the center of rectangular

fin of third row)Z* spanwise coordinate (Z* ¼ 0 at the center of rectangular

fin of third row)X streamwise coordinate (X ¼ 0 at the center of rectangular

fin of first row)Z spanwise coordinate (Z¼ 0 at the center of rectangular fin

of first row).Hd Duct height 200 mm

Greek symbols:n kinematic viscosity of air, m2/sl thermal conductivity, W/mKa inclination angle, degree

References

[1] E.M. Sparrow, J.E. Niethammer, A. Chaboki, Heat transfer and pressure dropcharacteristics of arrays of rectangular modules, Int. J. Heat Mass Transfer 25(1982) 961e973.

[2] E.M. Sparrow, S.B. Vemuri, D.S. Kadle, Enhanced and local heat transfer,pressure drop and flow visualization for arrays of block-like electroniccomponents, Int. J. Heat Mass Transfer 26 (1983) 689e699.

[3] T. Igarashi, Local heat transfer from a square prism to an air stream, Int. J. HeatMass Transfer 29 (5) (1983) 777e784.

[4] A.Y. Turk, G.H. Junkhan, in: C.I. Tien, et al. (Eds.), Heat Transfer EnhancementDownstream of Vortex Generators on a Flat Plate, Heat Transfer, Hemisphere,Washington 6, 1986, pp. 2903e2908.

[5] K. Oyakawa, Y. Furukawa, T. Taira, I. Senaha, Effect of vortex generators onheat transfer enhancement in a duct, Proc. Exp. Heat Transfer Fluid Mech.Thermodyn. (1993) 633e640 Honolulu, Hawaii.

[6] S.A. El-Saed, S.M. Mohamed, A.M. Abdel-Latif, A.E. Abouda, Investigation ofturbulent heat transfer and fluid flow in longitudinal rectangular fin-arrays ofdifferent geometries and shrouded fin array, Exp. Therm. Fluid Sci. 26 (2002)879e900.

[7] K. Bilen, S. Yapici, Heat transfer from a surface fitted with rectangular blocks atdifferent orientation angle, J. Heat Mass Transfer 38 (2002) 649e655.

[8] S. Besir, O. Nurhan, G. Cahit, Horseshoe vortex studies in the passage of a modelplate-fin-and-tube heat exchanger, Int. J. Heat Fluid Flow 29 (2008) 340e351.

[9] H. Fu, D. Rockwell, Shallow flow past a cylinder: transition phenomena at lowReynolds number, J. Fluid Mech. 540 (2005) 75e97.

[10] K. Yakut, N. Alemdaroglu, I. Kotcioglu, C. Celik, Experimental investigation ofthermal resistance of a heat sink with hexagonal fins, Appl. Therm. Eng. 26(2006) 2262e2271.

[11] H.-Y. Li, K.-Y. Chen, Thermal performance of plate-fin heat sinks underconfined impinging jet conditions, Int. J. Heat Mass Transfer 50 (2007)1963e1970.