fe , K Ata for per ment surface kept at a constant wall temperature boundary condition for the swirling (SIJ), multi-channel (MCIJ) and conventional and mass transfer in a number of industrial applications. Air jets have been frequently used at the gas turbine work- used to locate isotherms on a heated surface. By adjusting transfer coefficients for confined impinging air jets were investigated by Colucci and Viskanta [4]. A thermochro- circular and elliptic jet arrays using a liquid crystal tech- nique. Both continuous and broken V-shaped-rib configu- rations with different exit flow orientations were considered circular and elliptic jet. The best heat transfer performance was obtained with a surface with 45� v-shaped ribs. In addition, the surface with continuous ribs provided a better * Corresponding author. Tel.: +90 442 231 4864; fax: +90 442 236 0957. E-mail address:
[email protected] (K. Bilen). Experimental Thermal and Fluid Sci ing at the high temperature, tempering of glass, drying of paper, textile and film industries, thermal treatments of the metals, cooling of turbine blades and electronic compo- nent [1,2]. In the experimental study carried out by Gold- stein and Timmers [3], a visualization technique was used to measure the heat transfer coefficient distribution on a flat plate on which either a single jet or an array of jets impinges. Liquid crystals coated on a mylar sheet were matic liquid-crystal technique was used to visualize and record isotherms on a uniformly heated impingement sur- face. The effects of Reynolds number, nozzle-to-plate spac- ing, and nozzle geometry on the local heat-transfer coefficients were reported and compared with similar experiments for unconfined jets. Yen and Mei [5] and Yan et al. [6] examined the detailed heat transfer coefficient distributions over a ribbed surface under impingement of impinging jet (CIJ) using liquid crystal technique. The swirling jet assembly consisted of a housing tube and a solid swirl generator insert which had four narrow slots machined on its surface. The swirl angle, h, was set as 0�, 22.5�, 41�, 50� to change the direction and strength of the swirl in the air flow exiting the housing tube. The local Nusselt numbers of the MCIJ (h = 0�) were generally much higher than those of CIJ and SIJs. As the swirl angle increased, the radial uniformity of the heat transfer was seen compared to MCIJ and SIJ; the best results were for h = 50� and the jet-to-surface distance of H/D = 14. The location of the distance of the maximum heat transfer for the swirl angles of h = 41� and 50� was shifted away from the stagnation point in a radial distance of nearly r/D = 2.5. Increasing Rey- nolds number for same swirler angle increased the heat transfer rate on the entire surface, and increased saddle shape heat transfer dis- tribution on the surface, but had no significant effect on the position of the individual impingement regions, but increased saddle shape heat transfer distribution on the surface. The lower Reynolds number (Re = 10000) and the highest H/D = 14 gave much more uniform local and average heat transfer distribution on the surface, but decreased their values on the entire surface. � 2007 Elsevier Inc. All rights reserved. Keywords: Swirl flow; Jet impingement; Heat transfer enhancement; Convective heat transfer 1. Introduction Impinging jets have been widely used to increase heat the surface heat flux, contours of constant heat transfer coefficient were obtained on the impinged surface. Effects of hyperbolic nozzle geometry on the local heat- Visualization of heat trans K. Bakirci Department of Mechanical Engineering, Received 8 November 2006; received in revised Abstract The objective of the experimental study was to visualize the tem 0894-1777/$ - see front matter � 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.expthermflusci.2007.03.004 r for impinging swirl flow . Bilen * turk University, 25040 Erzurum, Turkey m 12 February 2007; accepted 12 March 2007 ature distribution and evaluate heat transfer rate on the impinge- www.elsevier.com/locate/etfs ence 32 (2007) 182–191 rma impingement heat transfer than with broken ribs. In another study done by Yan et al. [7], detailed heat transfer characteristics on a flat surface under arrays of impinging elliptic jets were measured by a transient liquid crystal tech- nique. The elliptic jet holes of five different aspect ratios, AR = 4, 2, 1, 0.25, and 0.5, jet Reynolds numbers Re = 1500, 3000, and 4500, and three exit flow conditions were investigated. Lee et al. [8] carried out an experimental study to investigate the heat transfer characteristics by an elliptic jet impinging upon the flat plate surface using a thermochromic liquid crystal. For H/D from 4 to 10 and for all Reynolds numbers, the local Nusselt number decreased monotonically with increasing radial distance. Nomenclature D diameter of nozzle (m) H/D distance of dimensionless jet-to-surface X/D dimensionless distance from stagnation point r/D dimensionless radial distance from stagnation point in x-direction h angle of swirl generator (�) L thickness of plate (m) LL length of nozzle (m) LLh length of the swirl generator (m) k thermal conductivity (W m�1 K�1) Q heat transfer rate (W) h local heat transfer coefficient (W m�2 K�1) A area of heat transfer (m2) V mean air velocity in the nozzle (m s�1) Re Reynolds number Nu Nusselt number K. Bakirci, K. Bilen / Experimental The Nusselt number for the elliptic jet in the impingement region was larger than that for a circular jet. This was attributed by the large entrainment rate and large scale coherent structure of the elliptic jet. One of the important parameters affecting impinging jet heat transfer is the flow condition at the nozzle exit. Swirl flow at the nozzle exit would alter jet-spreading rate. It is, therefore, important to understand the swirling effect on the flows and heat transfer so that distinctions between favorable and undesirable effects of swirl to many flow and heat transfer processes could be made. In the swirling jet, the degrees of jet growth, entrainment of ambient air, and jet decay are affected by the degree of swirl. The studies on a single and multi-channel jets have been carried out widely in the literature, whereas not enough swirling jet impingement studies were encountered. Ward and Mahmood [9] investigated to rates of the mass and heat transfer of the swirling air jets with four lon- gitudinal narrow slots perpendicularly impinging on a plane, and they determined the mass transfer data by using the thin-film naphthalene sublimation technique. The mea- sured radial distribution of local Nusselt number was slightly more uniform than that of a conventional imping- ing jet, but its values were significantly lower, particularly in the vicinity of the stagnation point. Huang and El-Genk [10] conducted experiments of heat transfer and flow obser- vation to investigate and compare the performance of swirling and multi-channel impinging jets with that of a conventional impinging jet (CIJ). It was reported that the local and surface average Nusselt numbers of MCIJ were generally much higher than those of CIJ. SIJs demon- strated large increases in both Nusselt numbers and signif- icant enhancement in radial uniformity of heat transfer compared to MCIJ and CIJ. Arzutug et al. [11] carried out an experimental study on the comparison of mass transfer distribution on a plate subjected to impingement m kinematics viscosity (m2 s�1) T temperature (�C) dt thickness of liquid crystal package and plate Subscripts cond conduction conv convection rad radiation lat lateral p plate a air or jet LC liquid crystal f film net net avg average l and Fluid Science 32 (2007) 182–191 183 by a CIJ, MCIJ. The local mass transfer coefficient was measured by the electrochemical limiting diffusion current technique. It was reported that the values of the mean mass transfer coefficients over the surface for CIJ and MCIJ were relatively close to each other, being slightly higher for MCIJ. Lee et al. [12] studied the effect of the swirl jet with eight narrow channels upon the local and average heat transfer distribution. For small jet-to-surface distance, H/D = 2, the average Nusselt numbers of the swirling jet flows were larger than those without swirling flow for all swirl angles, but for H/D > 10, the effect of the swirling jet flows was rarely seen. A few papers have been published to investigate heat transfer characteristics with a swirling impinging jet. So, to better understand the heat transfer performance of this swirl generator, additional heat transfer visualization experiments are needed for both CIJs and SIJs. For swirl- ing jets impingement, very little work has been reported on the complete mapping of temperature and heat transfer on the impingement surface using liquid crystal technique. Most work involved in the literature use constant heat flux boundary conditions at the wall and constant swirl gener- ator height. The present study differs from the ones in the literature by using constant wall temperature boundary condition and constant helical rotation of the air passage in all the swirl generators. The swirlers were so designed that the helical air path in all swirler were turned only once around the swirl generator rod (360�). This paper reports the results of heat transfer and temperature visualization experiments which are performed to investigate and com- pare the heat transfer performance and mapping of the temperature distribution for the swirling jet on the surface with those of a CIJ having the same diameter at the same conditions. The performance of these jets is evaluated in terms of the measured increases in the values of the local tank and the pressure regulator. An air filter was installed plate. The dimension of the impingement plate was 450 mm · 450 mm in area and it had a thickness of 5 mm. A smooth aluminum pipe of 15 mm inner diameter and 50 diameters lengths (=50D) was employed as a nozzle. The back surface of the impingement surface was kept at a constant temperature by circulating water past the sur- face with circulator equipment having a thermostat. 2.1. Design of nozzle and swirlers Swirl jet assembly consisted of a housing aluminum tube 184 K. Bakirci, K. Bilen / Experimental Thermal and Fluid Science 32 (2007) 182–191 between the air tank and flow meter to clean the air from oil and other impurities. Following the flow meter the air flow was directed to the impingement surface passing through a rubber hose of 2 m length and then to the nozzle. Due to its low thermal conductivity as well as its strength against the outward bending of the plate at high temperature, a glass plate was chosen as an impingement 1213 14 3 4 5 6 7 8 9 10 11 15 21 Fig. 1. Schematic diagram of the experimental setup. 1. Air compressor; 2. Air tank; 3. Open-close valve; 4. Pressure regulator; 5. Heat exchanger; 6. Air filter; 7. Flow meter; 8. Jet pipe; 9. Impingement plate; 10. Pipe and surface average Nusselt numbers and the uniformity of their radial distributions on the impinged surface. In the heat transfer experiments the effects of (a) swirl angle, h, for 0�, 22.5�, 41� and 50�; (b) jet-to-surface distances, H/D, for 6, 8, 10 and 14 and (c) air Reynolds number, Re, for 10000, 20000 and 30000 were investigated by using liquid crystal. 2. Experimental details The experimental setup is shown in Fig. 1. The air rate for the system was obtained from a compressor and was stored in an air tank, and supplied to the system through a pressure regulator, condenser and a flow meter, respec- tively. A pressure regulator was installed at the exit of the air tank so as to control the exit air pressure. Following this, a valve was employed to adjust the flow rate of the jet air. In order to reduce temperature variation of the sup- plied air and bring the jet flow temperature to the ambient temperature, a heat exchanger attached between the air plumbing; 11. Constant temperature bath; 12. Water vessel; 13. Pump; 14. Video camera; 15. Television. (length of 50D) and a solid swirl generator insert which had four narrow flow channels machined on its surface (Fig. 2). The swirl angle was varied to change the direction and strength of the swirl in the air flow exiting the housing tube (nozzle). Four short helical swirl generators were manufactured from steel rod of 15 mm diameter with swirl angels of h = 0�, 22.5�, 41� and 50� to the pipe axis. When h is 0�, flow channels were vertical and jet is referred to as multi-channel impinging jets (MCIJ). When 90� > h > 0� the jet was referred to as swirling impinging jet (SIJ). A conventional impinging jet (CIJ) had no inserts inside the housing tube. The swirl generator was placed inside the end of the tube, and only its conical tip was outside of the tube as shown in Fig. 3a. They were manufactured by machining four channels on a rod, leaving a hub with 2.5 mm diameter. To avoid a sudden expansion after the swirler the hub was extended conically tapering to a point. The swirlers were so designed in such a way that the heli- cal air path in all swirlers were turned only once around its rod (360�), so it changes the swirl generator heights for all cases, which are equal to 0.123cosh length in m; thus the heights of swirls were chosen to be LLh = 0.123cosh in m (Fig. 2). The difference between the swirlers of the present work and those of Huang and El-Genk [10] is the fraction of the cross-sectional area of the helical channels; the chan- nels occupy approximately 60% of the cross-sectional area of the swirler, reaching a hub of 5 mm in the centre in the present work while they occupy a small part of the cylindri- cal rod, approximately 15% of the total cross-sectional area, in the work of Huang and El-Genk [10]. Further- more, they used swirler at fixed height, but the present swir- Fig. 2. Swirl generators with different angles. disturbance effect on the fluid flow. Therefore the micro- rma capsules of the liquid crystal were diluted by water and then, were applied on the surface to obtain a film of �50 lm after coating with a special black dye (�25 lm thickness) produced for a better liquid crystal application. In the experiments, the colours on the scaled impingement surface after reaching the steady-state condition were recorded with a video camera, and the corresponding tem- lers have ones fixed helical air path rotation for all swirler so that the height of swirler was changed with swirl angle. In the case of fixed swirl height, all swirl generators with different swirl angles have the different rotation rates on the swirl rod, so each one gives different swirl effects to the fluid passing through it, causing more pressure drops. 2.2. Calibration of thermosensitive liquid crystals Thermosensitive cholesteric liquid crystal was used to measure the temperature distributions on the heated sur- face. After a thin layer of black dye, the liquid crystal must be applied to the surface as smooth and thin (0.075 mm) as possible to avoid the resistance to the heat transfer and the L Tp TLC Ta, h z k Convection θ LLθ Swirl generator Nozzle Fig. 3. (a) Swirl generator placed inside the jet tube (b) boundary conditions at the thickness of the plate. K. Bakirci, K. Bilen / Experimental The peratures and locations were then determined to evaluate heat transfer coefficients. The relationship between the color and temperature of TLCs was found by calibration experiment in a constant temperature bath. The liquid crys- tal was applied on a copper plate of high conductivity to obtain the uniform temperature distribution over the sur- face. The colour change of the liquid crystal was deter- mined adjusting the temperature of the bath at a given value. The values were determined in two ways, by heating and cooling. In the water bath, the change in colour appears to be uniform over the whole area of the coated liquid crystal. The red and green colour of liquid crystal used here had a narrow range of 1.10 and 1.25 �C, respec- tively, and the colour of red (from black to red) and green (from red to green) start to display at 35.6 �C and 36.7 �C, respectively. In the experiments, the colour of red–green line, which is 36.7 �C, was used. Temperature resolution of the red–green colour line was better than 0.1 �C. The cal- ibration was done before and after test runs and no change was observed. 3. Calculation of heat transfer coefficient Local heat transfer rate could be obtained from the tem- peratures of the front and back plate surfaces and the jet air using liquid crystal observations. The heat transfer rate between the hot water and the back plate surface is equal to the sum of the conduction heat rate (Qcond in z-direction and Qlat in vertical direction, Fig. 3b) through the plate thickness at steady state. It is expected that the lateral con- duction in the plate have no significant effect on the local surface temperature response. Because of the thermal insu- lation and thin plate thickness, the heat losses from the lateral conduction are neglected [8,13]. Thus this heat trans- fer rate is also equal to the sum of the heat losses from the front plate surface to the environment by convection and radiation. The energy balance for the plate becomes [13]: Qcond ¼ Qconv þ Qrad ð1Þ where Qconv and Qrad express the heat losses from the impingement plate by convection and radiation, respec- tively, and Qconv is the conductive heat through the plate thickness. The local surface temperature for the target plate exposed to the air is thus represented by the classical one- dimensional heat conduction. The expression for heat transfer coefficient can be written from Eq. (1) as follows [13,15]: h ¼ 1ðT LC � T aÞ ktðT LC � T PÞ dt � erðT 4LC � T 4PÞ � � ð2Þ where kt is experimentally measured mean conductivity of the liquid crystal package and plate, dt is the thickness of liquid crystal package and plate (�5.075 mm), TLC is the temperature of the surface (liquid crystal red–green iso- therms temperature, TLC = 36.7 �C). TP is the temperature of plate from water side and Ta is the temperature of air or jet. The experimental results are presented in terms of a lo- cal Nusselt number. The Nusselt numbers were given as Nu ¼ hD=ka ð3Þ and Reynolds number as Re ¼ VD m ð4Þ where D is the diameter of the nozzle (aluminum pipe), ka is the conductivity of air at film temperature (TLC + Ta)/2. Reynolds number was evaluated at the inlet condition be- fore the insert (swirl generator). V is the mean air velocity calculated from the flow rate measured by the flow meter. The surface was heated from behind by a water bath, and the heat flux through the plate was varied by adjusting the water temperature. With fixed temperature difference between air and liquid crystal isotherms, different heat l and Fluid Science 32 (2007) 182–191 185 transfer coefficient contours determined by varying the heat flux values (varying the plate temperature of the water side). The contours of the isotherms recorded by 3000M are used to obtain the heat transfer coefficient distribution on the impinging surface. As the temperature of the circu- impingement surface. The following subsections present and discuss the results of the heat transfer experiments. These experiments investigated the effects of swirl angle (h = 0–50�), jet-to-surface distance (H/D = 6–14), Rey- nolds number (Re = 10000–30000) on values and the radial uniformity of both the local Nusselt number and the stagnation point heat transfer. All the measurements were taken at the steady-state conditions, that is, when the color of the liquid crystal coating on the impinged sur- face remained unchanged. 4.1. Constant heat transfer contours Figs. 4a–8a contain photographs of the liquid crystal Table 1 Typical non-dimensional interval for the relevant variables Variables Uncertainty (%) Kinematic viscosity of air, m ±2.9 Thermal conductivity of air, ka ±3.0 Thermal conductivity of plate and liquid crystal package, kt ±4.6 The thickness of liquid crystal package and plate, dt ±3.5 Air volume rate of flow ±2.0 Liquid crystal temperature, TLC ±0.27 The temperature of air or jet, Ta ±0.55 The temperature of the plate from water side, TP ±0.3 186 K. Bakirci, K. Bilen / Experimental Thermal and Fluid Science 32 (2007) 182–191 lating water changes, the position of the color isotherm (red to green line, which is 36.7 ± 0.15) also moves on the plate surface at state conditions. This allows an easy observation and complete mapping of the heat transfer coefficient over the entire surface. Since the temperature differences are small (in the order of approximately 12 �C), the resulting heat transfer coefficients are indepen- dent of the level of heat flux used. By using the estimation method of Moffat [14], the max- imum uncertainties of the investigated non-dimensional parameters are as follows: Re, 3.7%; Nu, 6.8%. The individ- ual contributions to the uncertainties of the non-dimen- sional parameters for each of the measured physical properties are summarized in Table 1. 4. Results and discussion The location and the color of the liquid crystal were used to calculate the local heat transfer coefficient on the Panasonic video camera are not directly equivalent to con- stant heat transfer coefficient. They are determined after taking into account heat losses from the surface. Ten to 16 isotherms (each corresponding to a different heat flux) CIJ H/D=8 Re=20 000 Y /D -6 0 6 X/D 0 8 -2 -4 6 2 4 -6 -8 Fig. 4. For CIJ, H/D = 8 and Re = 20000 (a) map of constant tem while Figs. 4b–8b present the contours of constant heat transfer rate for the jet flows with and without swirl on the impingement surface. For the jet flows of CIJ, MCIJ and h = 22.5�, Figs. 4b–6b show that at L/D = 8, the con- stant heat transfer rate increases as the stagnation point is approached and the constant heat transfer contours are cir- cular in shape over the entire surface, except for the case for h = 22.5� showing not a perfect circle. Figs. 7 and 8 exhibit that with increasing swirler angle, every channel (four channels in swirl generator) tends to behave like an individual jet, forming each a separate impingement regions on the impinging surface. Further more, when swirl angle increased, the maximum heat transfer point of the individual regions for h = 41� and 50� does not occur at the stagnation point (r/D = 0) and shifts away in radial direction from the stagnation point (Figs. 7 and 8). Here after, in axial direction the radial distance from the stagna- tion point will be shown as r/D instead of X/D. 4.2. Stagnation point heat transfer Fig. 9 shows that the stagnation point heat transfer (at r/D = 0) has a strong dependence on the swirl number. It is seen from the figure that for different H/D values, the highest Nu0 at the stagnation point is achieved at H/D = 6 for all jet flow cases, while the highest stagnation X/D -10 -8 -6 -4 -2 0 2 4 6 8 10 Nu=35.245.567.277.1 Re=20 000 H/D=8 CIJ perature contours (b) contours of the constant Nusselt number. 8 6 4 rma K. Bakirci, K. Bilen / Experimental The point heat transfer rate (Nu0) is achieved for h = 0� at all H/D values. On the other hand for SIJs of h = 22.5�, 41� and 50�, the stagnation point Nusselt numbers are smaller than without swirl jet flow (CIJ), and the Nu0 values of both h = 0� (MCIJ) and CIJ at H/D = 8 are nearly same, there is only a little difference between them. It is also seen θ=0° H/D=8 Re=20 000 Y /D -2.5 0 2.5 X/D 0 -2 -4 2 -6 -8 Fig. 5. For MCIJ (h = 0�), H/D = 8 and Re = 20000 (a) map of constan θ=22.5° H/D=8 Re=20 000 0 8 -2 -4 6 2 4 -6 -8 -2.9 0 2.9 X/D Fig. 6. For h = 22.5�, H/D = 8 and Re = 20000 (a) map of constant te Y /D θ=41° H/D=8 Re=20 000 -8.5 0 8.5 X/D 0 8 -2 -4 6 2 4 -6 -8 Fig. 7. For h = 41�, H/D = 8 and Re = 20000 (a) map of constant te Re=20 000 H/D=8 =0°θ l and Fluid Science 32 (2007) 182–191 187 from the figure that the differences between the Nu0 values of h = 41� and h = 50� are very small for the jet flows at the jet-to-surface distance range of 6 6 H/D 6 14. In compar- ison of Nu0 for air jet flows to that of CIJ, Nu0 is 8.2% higher for MCIJ than for CIJ, and is 9.8%, 66.7% and 68.7% lower for 22.5�, 41� and 50� than for CIJ at -10 -8 -6 -4 -2 0 2 4 6 8 10 Nu=33.851.565.986.2 X/D t temperature contours (b) contours of the constant Nusselt number. -10 -8 -6 -4 -2 0 2 4 6 8 10 Nu=28.739.451.680.6 X/D Re=20 000 H/D=8 θ=22.5° mperature contours (b) contours of the constant Nusselt number. X/D -10 -8 -6 -4 -2 0 2 4 6 8 10 Nu=28.433.942.162.6 Re=20 000 H/D=8 θ=41° 42.1 mperature contours (b) contours of the constant Nusselt number. Y /D 0 8 -2 -4 6 2 4 -6 -8 t te rma θ=50° H/D=8 Re=20 000 -6.7 0 6.7 X/D Fig. 8. For h = 50�, H/D = 8 and Re = 20000 (a) map of constan 120 160 o CIJ MJCI (0˚) SJI (22.5˚) Re= 20 000 188 K. Bakirci, K. Bilen / Experimental The H/D = 6, respectively. When changing the H/D value from H/D = 6 to 8 at same conditions, Nu0 becomes 10.8 higher than that of without swirl (CIJ), and the Nu0 value is 19%, 68.9% and 74.1% lower for 22.5�, 41� and 50� than for CIJ, respectively. The Nu0 values for SIJ’s (22.5�, 41� and 50�) are smaller than that for CIJ. The decrease in Nu0 with an increasing swirl number for the same H/D may be asso- ciated with the corresponding reduction in the jet arrival velocity at the impinging surface. Besides, when the swirl- ing jet travels further down from the nozzle exit, axial flux of the tangential momentum becomes weaker due to a strong mixing of the spreading jet, resulting in decrease in Nu0. 4.3. Effect of nozzle and swirlers Fig. 10 shows the swirl effect on the radial Nusselt num- ber distributions at constant values of both H/D = 8 and Re = 20000. The values of both Nu0 and maximum heat transfer (Numax) for jet flows occur at the stagnation point (geometrical impingement point), but for swirl angles of 41� and 50�, the location of maximum Nusselt number (Numax) is shifted away from the stagnation point (centre of the jet) to the radial position, near r/D = 2.5 for both 0 40 80 4 6 8 10 12 14 16 18 H/D N u SJI (41˚) SJI (50˚) Fig. 9. Stagnation point heat transfer for jet flows with and without swirl. -10 -8 -6 -4 -2 0 2 4 6 8 10 Nu=22.834.243.658.9 X/D Re=20 000 H/D=8 θ=50° 43.6 mperature contours (b) contours of the constant Nusselt number. 80 120 160 N u CIJ MCIJ (0°) SIJ (22.5°) SIJ (41°) SIJ (50°) Re=20 000 H/D=8 l and Fluid Science 32 (2007) 182–191 h = 41� and 50�. Similar behavior for swirling jet flows was observed by Lee et al. [12], who used swirl generator with eight grooved channels instead of four grooved chan- nels used in the present work. This displacement of Numax location may be due to either the blockage at the centre of the swirl generator for larger swirl angle and/or the high spreading rate for high swirl angles (h = 41� and 50�). The highest Numax for h = 41 is 6% larger than that of h = 0� at the radial distance r/D = 2.5. The radial Nusselt numbers are larger for h = 41� and 50� than for h = 22.5� in the region 2 6 r/D 6 2.5 and the effect of the swirling motion of the jet flow is rarely seen in the wall region cor- responding to r/DP 3, but Numax for all cases decrease with an increasing swirl angle. For h = 0�, the local Nusselt numbers are higher than those without swirl generator in the entire region. This may be due to the fact that for h = 0�, the jet possesses the multiple jet characteristic. To quantify the radial uniformity of heat transfer on impinged surface by SIJs, for H/D = 8 and Re = 20000, Fig. 9, when a horizontal line representing Num = [(Nu0 + Numax)/2] was drawn across, all mean deviation of Nu from Num for CIJ and SIJs (h = 0�, 22.5�, 41� and 50�) in the region of r/D < 7 were within 40%, 38%, 30%, 29% and 24%, respectively. It is concluded that the best uniformity is achieved at the largest swirl angle of h = 50�. 0 40 -8 -6 -4 -2 0 2 4 6 8 r/D Fig. 10. Effect of swirl angle on the local Nusselt number distributions, Re = 20000, H/D = 8. 4.4. Effect of Reynolds number Fig. 11a and b show the local Nusselt number distribu- tion at the jet-to-surface distance of H/D = 8 for two differ- ent Reynolds numbers (Re = 10000 and Re = 30000). The increasing Reynolds number is much more significant upon the heat transfer, but has nearly no effect on the shift of the location of the maximum heat transfer as seen in Fig. 11a and b. Maximum heat transfer also first occurs for h = 0�, and then CIJ h = 22.5� and finally 50�, respectively for both Re = 10000 and 30000. It is seen from the figures that, when the Reynolds number is changed from 10000 to 30000, similar behavior is obtained for the case of Re = 10000. Nu0 is also 95%, 102%, 79% and 141% increased for MCIJ, CIJ, h = 22.5� and 50�, respectively, and the location of the Numax for h = 50� remains at nearly same location, but its value increases from 50 to 80. In the region r/D > 2.1, the local Nusselt number of h = 50� is the highest for the jet flows with and without swirl. It is seen from Fig. 11b that for Re = 30000 and H/D = 8, the loca- tion of maximum Nusselt Number (Numax) of h = 50� occurs at a radial distance of about r/D = 2.5 from the stagnation point. The results show that the local heat trans- fer coefficient depends on the jet Reynolds number. As expected, the impinging force on the target surface is stron- ger for a higher jet Reynolds number Re. Therefore, larger Nu distributions are found for a higher Re. This confirms The radial shift of the jet impingement location depends K. Bakirci, K. Bilen / Experimental Therma 0 40 80 120 160 -8 -6 -4 -2 0 2 4 6 8 r/D N u CIJ MCIJ SIJ (22.5°) SIJ (50°) Huang and El-Genk [10] Re=10 000 H/D=8 (Re=8100, SIJ (45°), H/D=6) 0 40 80 120 160 -8 -6 -4 -2 0 2 4 6 8 r/D N u CIJ MCIJ SIJ (22.5°) SIJ (50°) Re=30 000 H/D=8 Fig. 11. Effect of Reynolds number and swirl angle on the local Nusselt number distributions (a) Re = 10000, (b) Re = 30000. strongly on the swirl angle due to higher tangential velocity component. The heat transfer performance of swirl angle of 22.5� seems only to be similar trend to that of h = 0�. But, relatively, 41� and 50� swirl angles result in a very uni- form heat transfer enhancement over the entire impinge- ment surface. Comparison of the corresponding results in Fig. 10 indicates that a better uniform heat transfer perfor- mance on the surface is noted for larger swirl angle of 50�, while the location of the maximum heat transfer shifts away from the stagnation point as swirl angle increases. As it can be seen from the Fig. 11a, there is a good agreement between the local Nusselt values of the study and Huang and El-Genk [10] in the literature. 4.5. Effect of jet-to-surface distance In Fig. 12a and b, the Nusselt number distribution at Re = 20000 is shown for different jet-to surface distances (6 6 H/D 6 14). It is seen that for the h = 0�, the heat transfer rate is the highest at the stagnation point in the range of 6 6 H/D 6 14 (Fig. 12a). Therefore, it is con- cluded that the shorter the jet-to surface distance, the stronger the jet edge effect at which a strong shear layer is created with large momentum for all the range investi- gated. Fig. 12a also shows that the local Nusselt numbers for H/D = 6 and h = 0� are a little higher than that of no swirl, and Nu0 values are 0.2, 7.4 and 19.8 lower for H/D = 8, 10 and 14 than that for H/D = 6. The trends for H/D = 10 and 14 are very similar to the case of H/D = 6. However, the differences between with H/D = 6 and 8 are small and the local Nusselt numbers are nearly the same at the wall jet region beyond r/D � 5. It can be said that increasing the jet-to-surface distance decreases the local Nusselt number distribution for the range of parameters investigated. Fig. 12b shows the effect of the H/D on the local Nusselt number for h = 50�. From the figure, it is seen that the largest H/D ratio gives much more uniform heat transfer rate on the impinged surface for h = 50�, which cannot be said for h = 0�. Fig. 12a also shows the comparison of the local Nusselt number obtained by Lee et al. [12] and the present work. It can be said that there is a good agreement between them, considering that their Nusselt number are a little higher than those of the present work (they used Re = 23000 instead of Re = 20000 for the present work). 4.6. Surface average Nusselt number The average Nusselt number in an area of impinged sur- face having a radius, r, was evaluated as follows [10]: the concept that for convective heat transfer, better heat transfer is noted for a case with a higher Reynolds number. As the swirl angle increased, the locations of the heat transfer peaks are significantly shifted from the center. l and Fluid Science 32 (2007) 182–191 189 NuavgðrÞ ¼ QnetD=½AkðT LC � T aÞavg� ð5Þ H/D=10 H/D=14 0 40 80 120 160 -8 -6 -4 -2 0 2 4 6 8 r/D N u a v g(r ) CIJ MCIJ SIJ (22.5°) SIJ (41°) SIJ (50°) Re=20 000 H/D=8 Fig. 13. Effect of swirl angle on average Nusselt number. 0 40 80 120 160 -8 -6 -4 -2 0 2 4 6 8 r/D N u a v g(r ) CIJ MCIJ SIJ (22.5°) SIJ (50°) Re=10 000 H/D=8 rma 0 40 80 -8 -6 -4 -2 0 2 4 6 8 r/D N u Fig. 12. Effect of jet-to-surface distance on the local Nusselt number 0 40 80 120 160 -8 -6 -4 -2 0 2 4 6 8 r/D N u H/D=6 H/D=8 H/D=10 H/D=14 Lee [12] Re=20 000 MCIJ (0°) (Re=23 000, H/D=10) 120 160 H/D=6 H/D=8Re=20 000 SIJ (50°) 190 K. Bakirci, K. Bilen / Experimental The where Qnet is the net heat transfer rate calculated by taking into account heat losses from the surface, and (TLC(r) � Ta)avg was calculated form the numerical integra- tion of measured temperature as ðT LCðrÞ � T aÞavg ¼ 1 2pr2 � �Z r 0 Z 2p 0 ðT LCðr0Þ � T aÞr0 d/dr0 ð6Þ Fig. 13 shows that for both constant values of Re 20000 and H/D = 8, swirl angle affected the values of Nuavg(r) for SIJs. MCIJ (h = 0�) gave the best heat transfer perfor- mance in terms of higher Nuavg(r) values in the entire sur- face, while SIJ induced the second much higher Nuavg than those for all SIJs (h = 22.5�, 41� and 50�). For SIJs with h = 41� and 50� Nuavg(r) gave more uniformity on all sur- face. In the area r/D > 3, Fig. 14a, Nuavg(r) started getting higher values for h = 50� than for 22.5� at Re = 10000. Close to stagnation point, r/D < 4, however, SIJ with h = 22.5� gave higher Nuavg(r) values than with h = 500, similar behavior was obtained by Huang and El-Genk [10]. Fig. 14a indicated at low Reynolds number, Re = 10000, radial distributions of Nuavg(r) for SIJ with h = 50� were more flat. As Re increased, however, Nuavg(r) became less uniform, exhibiting a saddle shaped radial dis- tribution (Fig. 14b). The radial location of peak Nu in the saddle-shaped radial distribution was almost independent distributions (a) for MCIJ (h = 0�), (b) h = 50�. l and Fluid Science 32 (2007) 182–191 of Re, but increased as the swirl angle increased; it was r/D = 0 and 3 for h = 0� and 50�, respectively (Fig. 14c). 0 40 80 120 160 -8 -6 -4 -2 0 2 4 6 8 r/D N u a v g(r ) CIJ MCIJ SIJ (22.5°) SIJ (50°) Re=30 000 H/D=8 0 40 80 120 160 -8 -6 -4 -2 0 2 4 6 8 r/D N u a v g(r ) Re=10 000 Re=20 000 Re=30 000SIJ (50°) H/D=8 Fig. 14. Effect of Reynolds number and swirl angle on average Nusselt number (a) Re = 10000, (b) Re = 30000, (c) h = 50�. K. Bakirci, K. Bilen / Experimental Thermal and Fluid Science 32 (2007) 182–191 191 The effect of jet-to-surface distances of H/D = 6, 8, 10 and 14 on Nuavg values is exhibited in Fig. 15. The maxi- mum Nuavg values were obtained at H/D = 6. As H/D increased, Nuavg values decreased but gave more radial uni- formity on the entire surface. The best uniformity of Nuavg(r) values was obtained at H/D = 14 with decreased its value. 5. Conclusions Flow visualization and heat transfer experiments were performed to investigate and compare the heat transfer performance of new swirling jet designs with that of a con- ventional impinging jet having the same diameter at the same conditions. The performance of these jets was evalu- ated in terms of the increases in the local and surface aver- age Nusselt numbers and the improved radial uniformity of their radial distributions on the impinged surface. Heat transfer experiments demonstrated that as the swirl angle increased, every channel tended to behave like an individual jet, forming each a separate impingement regions. SIJs with h = 41� and 50� produced saddle-shaped radial distributions of the local Nusselt numbers, which were more pronounced at high Reynolds numbers and/or 0 40 80 120 160 -8 -6 -4 -2 0 2 4 6 8 r/D N u a v g(r ) H/D=6 H/D=8 H/D=10 H/D=14 Re=20 000 SIJ (50°) Fig. 15. Effect of jet-to-surface distance on the average Nusselt number for h = 50� and Re = 20000. small jet spacing. The uniformity of the surface distribution was strictly depending upon the swirl angle. When swirl angle increased, the locations of the heat transfer peaks were significantly shifted from the stagnation point. The radial shift of the jet impingement location depended strongly on the swirl angle due to higher tangential velocity component. The local and average Nusselt number values decreased as swirl angle increased, and the highest swirl angle of h = 50� gave more uniform heat transfer rate on the surface. The double-peaks in the saddle-shaped radial distribution of Nu and Nuavg(r) became less notable as Re decreased. Increasing Reynolds number at same swirl angle increased heat transfer rate on the surface but had no sig- nificant effect on the position of the individual impinge- ment regions, however increased saddle shape heat transfer distribution. And lower Reynolds number gave much more uniform local heat transfer distribution on the surface. The radial uniformity of Nu and Nuavg(r) of SIJs depended on values of the swirl angle, jet spacing, and air Reynolds number. In general, the radial uniformity of these Nusselt numbers improved as the swirl angle and/ or jet spacing was increased. Acknowledgement The authors would like to thank to Ataturk University Research Foundation for the financial support provided for this research project. References [1] K. Bakirci, An experimental investigation of heat transfer character- istic of swirling impinging jet, Master thesis, Ataturk University, Erzurum, Turkey, 1998. [2] K. Bilen, An experimental investigation of heat transfer characteristic of a normal and oblique air jet impinging on heated flat plate, Ph.D. thesis, Karadeniz Technical University, Trabzon, Turkey, 1994. [3] R.J. Goldstein, J.F. Timmers, Visualization of heat transfer from arrays of impinging jets, Int. J. Heat Mass Transfer 25 (12) (1982) 1857–1868. [4] D.W. Colucci, R. Viskanta, Effect of nozzle geometry on local convective heat transfer to a confined impinging air jet, Exp. Therm. Fluid Sci. 13 (1) (1996) 71–80. [5] W.M. Yan, S.C. Mei, Measurement of detailed heat transfer along rib-roughened surface under arrays of impinging elliptic jets, Int. J. Heat Mass Transfer 49 (1–2) (2006) 159–170. [6] W.M. Yan, S.C. Mei, H.C. Liu, C.Y. Soong, W.J. Yang, Measure- ment of detailed heat transfer on a surface under arrays of impinging elliptic jets by a transient liquid crystal technique, Int. J. Heat Mass Transfer 47 (2004) 5235–5245. [7] W.M. Yan, H.C. Liu, C.Y. Soong, W.J. Yang, Experimental study of impinging heat transfer along rib-roughened walls by using transient liquid crystal technique, Int. J. Heat Mass Transfer 48 (2005) 2420– 2428. [8] S.J. Lee, J.H. Lee, D.H. Lee, Local heat transfer measurements from an elliptic jet impinging on a flat plate using liquid crystal, Int. J. Heat Mass Transfer 37 (6) (1994) 967–976. [9] J. Ward, M. Mahmood, Heat transfer from a turbulent, swirling impinging jet, Proceedings of the Seventh International Heat Transfer Conference, vol. 3, Hemisphere, Washington, DC, 1982, pp. 401–407. [10] L. Huang, M.S. El-Genk, Heat transfer and flow visualization experiments of swirling, multi-channel, and conventional impinging jets, Int. J. Heat Mass Transfer 47 (3) (1998) 583–600. [11] M.E. Arzutug, S. Yapici, M.M. Kocakerim, A comparison of mass transfer between a plate and submerged conventional and multichan- nel impinging jets, Int. Commun. Heat Mass Transfer 32 (6) (2005) 842–854. [12] D.H. Lee, S.Y. Won, Y.T. Kim, Y.S. Chung, Turbulent heat transfer from a flat to a swirling round impinging jet, Int. J. Heat Mass Transfer 45 (2002) 223–227. [13] A.D. Ochoa, W. Baughn, A.R. Byerley, A new technique for dynamic heat transfer measurements and flow visualization using liquid crystal thermography, Int. J. Heat Fluid Flow 26 (2005) 264–275. [14] R.J. Moffat, Describing the uncertainties in experimental results, Exp. Therm. Fluid Sci. 1 (1988) 3–17. [15] O. Covellero, G. Tonda, An experimental investigation of forced convection heat transfer in channels with rib turbulators by means of liquid crystal thermography, Exp. Therm. Fluid Sci. 26 (2002) 115– 121. Visualization of heat transfer for impinging swirl flow Introduction Experimental details Design of nozzle and swirlers Calibration of thermosensitive liquid crystals Calculation of heat transfer coefficient Results and discussion Constant heat transfer contours Stagnation point heat transfer Effect of nozzle and swirlers Effect of Reynolds number Effect of jet-to-surface distance Surface average Nusselt number Conclusions Acknowledgement References