You are on page 1of 5

The Eighteenth International Symposium on Transport Phenomena, 27-30 August, 2007, Daejeon, KOREA

DRAG OF A SINGLE NON-CIRCULAR CYLINDER


A. Nouri-Borujerdi 1, A. M. Lavasani2
1

School of Mechanical Engineering Sharif University of Technology, Tehran, Iran, Emali: anouri@sharif.edu 2 Islamic Azad University Sience & Reasurch Branch, Tehran, Iran Prasad et al. [5] reported the heat transfer and pressure drop from an aerofoil in cross flow. Their aerofoil test section was the NACA-0024 and they concluded that this shape gives higher values of Stanton number to pressure drag coefficient, St/CD, compared with a circular tube. Badr et al. [6] simulated numerically the unsteady flow over an elliptic tube at different orientations and obtained form drag and skin friction with the major axis parallel to the flow direction. The total drag coefficient was obtained 0.8 at Re=3700 and 0.9 at Re=700 for an oval with a length ratio of minor-to-major axis equal to 0.6. The form drag was 80 - 90% of the total drag. Compared with a circular tube, the drag coefficient was reduced between 10 - 20%. The drag force decreased as the oval tube was made more slender, i.e., the shape factor was decreased. Matos et al. [7] studied numerically and experimentally three-dimensional optimization of staggered finned circular and elliptic tubes in forced convection. They have reported that the optimal elliptic arrangement exhibits a heat transfer gain of up to 19% compared with the optimal circular tube arrangement. The results illustrate that the heat transfer gain and the relative total mass reduction of up to 32% show that the elliptical arrangement has a potential to deliver considerably higher global performance and lower cost. Zhihua et al. [8] investigated numerically the heat transfer enhancement of finned tubes using shaped polymer tubes. The approach was illustrated for three streamlined shapes with fins of lenticular and oval profiles. They highlighted the effects of the geometry and the Biot number on the tube efficiency and heat transfer enhancement. Convective heat transfer is enhanced for the oval-shaped tube when 2 103 < Re < 2 10 4 and Bi<0.3. The potential benefit of reduced form drag remains. Very recently, Bouris et al. [9] designed an intensified heat exchanger for reduction of fouling rates. They proposed a tube cross-section that contained a parabolic shape in front and a semi-circular one at rear. They carried out experiments and numerical simulations on the novel tubes bundle heat exchanger for studying the thermal, hydraulic and fouling characteristics. They attained higher heat transfer levels with a 75% lower deposition rate and 40% lower pressure drop. In recent papers a very few cam-shaped tubes are used in heat exchangers and their capabilities have not been studied for a single tube or tubes bundle. The purpose of the present work is to investigate the flow characteristics over a cam-shaped cylinder under different attack angles. The attack angle is varied from 0 to 360 and the variation of the heat transfer rate and pressure drag coefficient features are discussed in

ABSTRACT Pressure drag coefficient and heat transfer are experimentally investigated around a single cam-shaped cylinder in cross flow under different attack angles. The testing fluid is air and the Reynolds number is in the range of 1.5 104 < Re eq < 4.8 104 . The range of attack angle is 0< < 360. The pressure drag coefficient features are clarified in relation to the flow behaviors around the cylinder. The results show that the pressure drag coefficient has a maximum value at about =90 and 270over the whole range of the Reynolds number. It is found that the pressure drag coefficient of a slender cam-shaped cylinder is more dominant at the right attack angle. Furthermore, the trend of mean Nusselt number against attack angle has a wavy shape but with more smooth relative to the drag coefficient behavior. The effects of the cam-shaped cylinder sizes upon the pressure drag coefficient are discussed.

INTRODUCTION Heat exchangers with high performance for saving and making effective use of energy are a very important subject. Among many types of heat exchangers, those have circular and noncircular tubes have been used in many industries. The circular tubes are almost exclusively used in the construction of heat exchangers, primarily because of the ease of manufacture. In contrast to the circular tubes which cause severe separation and large wake regions to produce high pressure drops, noncircular tubes of streamlined shapes offer very low hydraulic resistance and consequently require less pumping power. Merker and Hanke [1] measured the heat transfer and pressure drop along the shell-side of tubes bank having oval-shaped tubes with a transverse tube pitch-tominor axis ratio of 2. The total drag was reduced 95% compared with an array of circular tubes. Ota et al. [2, 3] measured heat transfer and flow of air behavior around an elliptic tube with axes ratios of 1:2 and 1:3. In this work the Reynolds number changed between 8000<Re<79000 with the attack angle 0 < 90. They found that the maximum mean heat transfer coefficient occurred at <90 over the whole range of the Reynolds number. They also concluded that the minimum mean heat transfer rate was higher than that of a circular tube. Drag coefficient was reduced about 30 50%. Ruth [4] measured the pressure drop and heat transfer rate for a lenticular tubes bundle at 1000< Re<5104 with a transverse tube pitch-to-diameter ratio of 2, the drag was reduced by 70% compared with an array of circular tubes.

2104

The Eighteenth International Symposium on Transport Phenomena, 27-30 August, 2007, Daejeon, KOREA

relation to flow behavior around the cylinder.

APPARATUS AND MEASUREMENT Test section was a single cam-shaped cylinder located at a distance of 10 cm in front of an open wind tunnel outlet with diameter 24 cm, (Fig. 1). The test cylinder was mounted horizontally between walls of a frame and perpendicular to the flow direction. The cross section profile of the cylinder comprised some parts of two circles with two line segments tangent to them. The test cylinder was made of a commercial copper plate with 0.3 mm thickness and a length of L=12 cm. Three test cylinders were used to investigate the effect of tube dimensions on flow characteristics. These three cylinders had identical diameters and equal to d=1.2 cm and D=2.2 cm but, with three different distances between their centers, l=1.1, 2.9 and 6.6 cm, (Fig. 2). Their equivalent diameters were 2.47, 3.57 and 5.91 cm respectively. The angle of attack varied in the range of 0 360 and denotes the inclination angle between the major axis with the direction of the free stream velocity. The angles were measured with positive sign in the clockwise direction and with negative sign in the opposite direction. A pitot static tube measured the free stream velocity in front of the frame cross section. The positions of the pitot tube were predetermined by dividing the frontal area into 25 equal sections. The air velocity varied from 12 to 22 m/s by controlling a variable speed motor. To verify the uniformity of the free stream velocity, the upstream velocity was measured at different locations. The results showed nearly uniform velocities for all ranges of velocity. The tube surface was covered with 20 holes (1 mm in diameter) drilled to measure the static pressure on the tube surface by a dial manometer. They were radially positioned at the middle length of the tube at 360/20=18 deg. intervals from the leading edge to the tail. In order to make more clear correlations between the heat transfer and flow characteristics, another cam- shaped tube was also used to measure the heat transfer. This tube was almost the same as the one used before for pressure drag calculation. To heat up the tube, an electric heating element supplied hot water and a control valve regulated the hot water at the tube inlet. A pump circulated hot water between a tank and the tube. Water temperature was measured at the inlet and outlet of the tube using type-k thermocouple wires. A glass tube flow meter measured the flow rate with one percent uncertainty in full-scale flow. A steady state condition reached between 10 to 30 minutes, depending on the ambient temperature and free stream velocity before data collection started. At lowest velocity the quantity of Gr / Re 2 = 1.29 10 4 showed that the effect of natural convection could be neglected. The length scale in the Reynolds and Grashof numbers was based on an equivalent circular tube whose circumferential length was equal to that of the camshaped tube. To measure the surface temperature 8 thermocouples were embedded and cemented on the surface around the tube where their positions are shown in Fig. 1 from T1 to T8. Two other thermocouples were

used to measure the axial temperature. Then the mean surface temperature at the middle length and film temperature for fluid properties [10] were calculated. The rate of heat transfer from the tube to the air stream around the tube was obtained by measuring the mass flow rate, inlet and outlet water temperatures through the tube as: q = m w C p,w (Twi Two ). The average pressure drag coefficient, CD, was also obtained by the pressure coefficient C P over the tube surface as follows:

CD =

1 Deq

C
i =1

20

p ,i

Cos i Si

(1)

where i is angle between normal to the tube surface and the free flow direction, (see Fig. 2). Si represents a differential element of the tube perimeter belong to each hole. The angle in the figure is a constant and depends on the dimensions of cam-shaped tube. For instance, the values of for the tubes are 117, 99.92 and 94.3 deg. respectively. Furthermore, the pressure coefficient needed in Eq. (1) was calculated as CP ,i = 0.5( Pi P ) / U 2 , where Pi is the static pressure measured by a dial manometer at the location of the holes drilled on the tube surface. P and U are the pressure and velocity of air free stream.

RESULTS AND DISCUSSION


To assessment the accuracy of the experimental data, first a single circular tube with diameter of 2.47 cm and length of 12 cm was tested before testing the cam-shaped tubes and taking the data process. This circular tube had the same circumferential length as the cam-shaped tube with d=1.2 cm, D=2.2 cm and l=1.1 cm. The experimental results showed a pressure drag coefficient of 0.73 < C D < 0.75 in the range of 1.9 10 < Re < 3.5 10 . Whereas, the pressure drag coefficient of the circular tube by the results of White [11] is about C D = 0.74 that indicates a discrepancy 1-2 percent with the experimental result. Fig.2 represents the pressure coefficient distribution against S / Deq on the upper and lower parts of the cam-shaped tube
4 4

surface for l/D=0.5, = 0, 30, 60, 90, 120, 180 degree, and

Reeq = 2.7 104.

The

dimensionless

abscissa

S/ Deq indicates the distance from the leading edge of the

tube surface. The positive and negative values on the abscissa indicate the measured distance along the upper and lower parts of the tube surface respectively. The results show that the trend of the pressure coefficient is nearly similar to the one produced by the perfect-fluid theory around a circular tube; like the solid curve for = 30 deg. When the attack angle increases the maximum pressure coefficient shifts towards the negative values of the abscissa. In other words, the lower part becomes more effective near the leading edge in comparison with the upper part. This is probably due to wake formation on the upper region. Fig. 3 illustrates the pressure coefficient distribution around the previous tube but for different attack angles,

2105

The Eighteenth International Symposium on Transport Phenomena, 27-30 August, 2007, Daejeon, KOREA

= 180, 210, 240, 270, 300, 330, and 360 degree. In this case, the pressure coefficient becomes more dominant on the upper part and shifts towards the positive values of the abscissa in comparison with the lower part region. Figs. 4a-c depict the effect of l/D on pressure coefficient over the tube surface for three attack angles of = 0, 120 and 270. The free stream velocity is the same for three angles and is equal to U =15 m/s. The trend of curves in Fig. 4a is similar for three values of l/D=0.5, 1.3 and 3 and discrepancy among them is limited in the range of -0.5< S/ Deq <0.5. This treatment will intensively continue in
Figs. 4b-c for 120 and 270 respectively. The results indicate that the effect of attack angle is more effective in slender tubes. Fig. 5 exhibits variation of pressure drag coefficient against

occurs at = 180 degree like the pressure drag coefficient in Fig. 5. Integration of Nu Cam / Nu Cir over

0 360 indicates that the ratio of the mean Nusselt number becomes larger than 1 at a critical Reynolds number while this ratio may be equal or even less than 1 when the Reynolds number is below the critical one. This is probably due to a large wake formation below the critical Reynolds numbers over a large region of the cam-shaped tube.

for l/D=0.5 in the range of 2 10 < Re < 3.3 10 . The shape of all curves is almost the same and repeated o every other 180 degree so that the overall relation between the variables can be expressed as:
4 4

CD = 0.0069 6 0.1307 5 + 0.9308 4 3.0564 + 4.5128 2.3366 + 0.76


3 2

(2)

where the attack angle is in radian, 0 2 . The maximum and minimum values of drag coefficient are about CD=0.9 and CD=0.4 respectively and their minimum occurrence are at three different attack angles. The reason could be explained by external flow configuration which involves the cam-shaped tube in cross flow where the tube is normal to the flow direction at = 180 or is under attack angles of = 30 or 330. Consider the fluid mechanics of the situation = 180 in cross flow. Under 2 104 <Re<3.3 104 the laminar boundary layer is formed over the most part of the tube surface and separation is delayed thereby, reducing the extent of wake region and the magnitude of the form drag. In other words, the pressure differential in the flow direction resulting from formation of the wake is low compared to the boundary layer surface shear stress (friction drag) and causes a low pressure drag coefficient. Fig.6 illustrates St / C D versus attack angle. In this figure the thermal hydraulic performance of the cam-shaped tube for three Reynolds numbers and at any attack angle is shown. It is clear the curve is symmetric relative to = 180 degree and that St / C D for the cam-shaped tube is function of attack angle where its value is higher at results indicate that the difference performance between the maximum ( and 9) is about 46 percent. Reynolds number, a circular tube diameter equal to D eq = 2.47 cm , 180and 330. The of thermal hydraulic and minimum values In the range of the with the equivalent has St / C D =0.77,

CONCLUSIONS The pressure drag coefficient and Nusselt number around a cam-shaped cylinder are experimentally investigated. The Reynolds number range examined is between 1.5 104 < < 4.8 104 and angle of attack is varied from 0 to 360 The pressure drag coefficient versus attack angle for all Reynolds number studied is almost the same and repeated every other 180 so that the overall relation between variables is expressed by Eq. (2). The maximum value occurs at = 90 and 270 and minimum at = 180 .The maximum value of the Nusselt number occurs at 90 and 270 over the whole range of Reynolds numbers studied and minimum value at = 180. In order to compare the available pressure drag coefficient and Nusselt number values of cam-shaped and circular cross-section cylinders with the same circumferential length, a Reynolds number based on the equivalent tube diameter has been defined. These comparisons have shown that the cam-shaped cylinders give higher values of St / C D relative to the circular crosssection for the most range of attack angles.

NOMENCLATURE Bi Biot number, hDeq / k


CD pressure drag coefficient

CP static pressure coefficient d small diameter D large diameter 3 Gr Grashof number, g TD eq /


h heat transfer coefficient k thermal conductivity l distance between centers L cylinder length N Nusselt number, hD / k u P pressure Re Reynolds number, U D /

whereas the cam shaped tube with the same equivalent diameter has higher values of St / C D except at = 90, 120 and 2700. Fig. 7 shows the cam-shaped tube Nusselt number to the counterpart value in a circular tube with the same equivalent diameter in the range of 1.6 104 < Reeq < 2.6 104 . The maximum value is about 1.1 and occurs at angles of = 90 and 270 degree and the minimum value is 0.85 and

S streamline coordinate St Stanton number, h / C P U


T temperature U velocity x, y cartesian coordinate

Greek Letters attack angle volume expansion coefficient

2106

The Eighteenth International Symposium on Transport Phenomena, 27-30 August, 2007, Daejeon, KOREA

difference density kinematic viscosity hole angle

from Tubes, Int. J. Heat Mass Transfer, 23, pp. 463-469.

Superscripts mean

Subscripts cam cam shaped tube cir circular eq equivalent i hole number, inlet o outlet water w free stream

REFERENCES [1] Merker, G.P., and Hanke, H., 1986, Heat transfer and pressure drop along the shell-side of tube banks having oval-shaped tubes, Int. J. Heat Mass Transfer, 29 (12), pp. 19031909. [2] Ota, T., Aiba, S., Tsuruta, T., and Kaga, M., 1983, Forced Convection Heat Transfer from an Elliptic Tube, Bull. JSME, 26(212), pp. 262 267. [3] Ota, T., and Nishiyama, H., 1984, Heat Transfer and Flow around an Elliptic Tube, Int. J. Heat Mass Transfer, 27(10), pp. 17711779. [4] Ruth, E. K., 1983, Experiments on a Cross flow Heat Exchanger with Tubes of Lenticular Shape, ASME J. Heat Transfer, 105, pp. 571575. [5] Prasad, B.V.S.S.S., Tawfek, A.A., and Rao, V.R.M., 1992, Heat Transfer from Aerofoils in Cross-Flow, International Communications in Heat and Mass Transfer, 19, pp. 870-890. [6] Badr, H. M., Dennis, S. C. R., and Kocabiyik, S., 2001, Numerical Simulation f the Unsteady Flow Over an Elliptic Tube at Different Orientations, Int. J. Numerical Methods Fluids, 37(8), pp. 905931. [7] Matos, R.S., Laursen, T.A., Vargas, J.V.C. and Bejan, A., 2004, Three-dimensional optimization of staggered finned circular and elliptic tubes in forced convection, International Journal of Thermal Sciences, 43, pp. 477487. [8] Zhihua Li, Jane H. Davidson and Susan C. Mantell, 2004, Heat Transfer Enhancement Using Shaped Polymer Tubes: Fin Analysis, Journal of Heat Transfer, 126, pp. 211-218. [9] Bouris, D., Konstantinidis, E., Balabani, S., Castiglia, D. and Bergeles, G., 2005, Design of a novel, intensified heat exchanger for reduced fouling rates, International Journal of Heat and Mass Transfer, 48, pp. 38173832. [10] Netcati Ozisik, M., Heat transfer, 1985, McGraw-Hill, New York. [11] White, F. M., Fluid Mechanics, 2005, McGraw-Hill, New York. [12] Quarmby, A. and Al-Fakhri, A. A. M, 1980, Effect of Finite Length on Forced Convection Heat Transfer

Fig.1. Diagram of experimental apparatus with the camshaped tube cross section

Fig. 2 Pressure coefficient along the cam-shaped tube surface for 0 180 deg.

2107

The Eighteenth International Symposium on Transport Phenomena, 27-30 August, 2007, Daejeon, KOREA

(c) = 270 deg. Fig. 3 Pressure coefficient along the cam-shape tube surface for 180 360 deg. Fig.4. Pressure coefficient along the surface of cam shaped tube for different l / D

(a) = 0 deg.

Fig. 5 Pressure drag coefficient along the cam-shaped tube for different Reynolds numbers

(b) = 120 deg

Fig. 6 Stanton number to drag coefficient ratio for different Reynolds numbers

2108

You might also like