6 CONVECTIVE HEAT TRANSFER IN CROSS FLOW A. Zukauska: Academy of Sciences of the Lithuanian SSR, Vilnius, USSR 6.1. Specitic Features of Cross Flow Fluid Dynamics and Heat Transfer 6.2. Heat Transter for Single Tubes and Bodies 62.1 Fluid Dynamics over a Single Tube and a Sphere 622 Drag on a Cylinder and a Sphere 623 Local Heat Transfer for a Cylinder and a Sphere 6.24 Average Heat Transfer for a Cylinder and a Sphere 625 Factors influencing Heat Transfer 626 Heat Transfer trom Tubes of Different Shapes 627 Heat Transter rom Tubes in Narrow Channels 62.8 The Etfect of Yaw Angie on the Heat Transfer from a Gylinder 6.3 Heat Transter and Pressure Drop for Smooth-Tube Bund 6.3.1 Fluid Dynamics in Smooth-Tube Bundles 6.32. Drag on Smooth-Tube Bundles 63.3. Drag on Yawed-Tube Bundles 63.4 Local Heat Transfer from a Tube in a Bundle 6.35 Average Heat Transfer from a Tube in a Bundle 6.3.6 The Effect of Yaw Angle on the Heat Transfer 6.4 Heat Transter and Pressure Drop for Rough-Tube Bundles 6.4.1 Fluid Dynamics in Rough-Tube Bundles 6.42 Drag on Rough-Tube Bundles 6.43 Heat Transfer for Rough-Tube Bundles 6.5 Heat Transfer and Pressure Drop for Finned-Tube Bundles 65.1 Finned-Tube Bundles 6.5.2 Fluid Dynamics in Finned-Tube Bundles 6.5.3 Drag on Finned-Tube Bundles 65.4 Heat Transfer for Finned-Tube Bundles Nomenclature References(CONVECTIVE HEAT TRANSFER IN CROSS FLOW 6.1 SPECIFIC FEATURES OF CROSS FLOW FLUID DYNAMICS AND HEAT TRANSFER ‘The fundamentals of heat transfer by forced convection are presented in Chaps. 1 and 2. We deal here with heat transfer from tubes of different cross sections and bodies of other geometries, which are applied in modern heat exchangers. Heat transfer is closely related to fluid dynamics. That is why heat transfer is considered simultaneously with fluid dynamics. Fluid dynamics and heat transfer around curvilinear bodies are complex processes, mainly dependent on the fluid type and the Reynolds number Re=— (61) where u and d are the reference (or characteristic) velocity and diameters, respectively, and » the kinematic viscosity. In real fluids, due to their viscosity, a laminar boundary layer is formed on the front part of a body, its thickness increasing downstream. It also involves a longitudinal pressure gradient caused by the curved surface. From a two-dimensional equation of ‘momentum (see Chap. 1) dP lar fe eh oh pak” pay with the boundary condition = » = 0 when y = 0. A relation between shear stress derivative at y = 0 and the pressure gradient is (3),.°% «> ‘The shear stress at the wall for a laminar boundary layer is determined by the ‘near-wall velocity gradient aw oil S) (64) A decrease or a zero value of the shear stress sis reflected by a corresponding change in the velocity gradient 4u/@y. In the classical theory of low separation, the ‘boundary-layer separation point is assumed to be at (u/8y), 0 = 0. From Eq. (6.4), ‘we see that this isthe point where the shear stress acquires a2er0 value, t, = 0, so that the boundary layer can separate from the surface (see Fig 6.1). The separation point is followed by an inverse flow, where the velocity vectors of near-wall uid masses are in ‘opposite directions. The inverse layer contacts the boundary layer and curls up in a vortex, which begins to rotate. Each of the above phenomena of fuid dynamics is reflected in the local heat transfer. For a heat flax to appear, there must be a temperature gradient in the flow. ‘The local heat transfer coeficient is defined by —- (3)... (65)FEATURES OF CROSS FLOW FLUID DYNAMICS AND HEAT TRANSFER Separation point ero sy vetocty Tine Figure 6.1. Velocity distribution on a curvilinear surface near the separation point With a laminar boundary layer developing on the front part of a curvilinear body in ‘ross flow, the heat transfer coefficient decreases with increasing thickness, and is lower in fluids of lower heat conduction. Heat transfer increases in the rear recirculation region, downstream from the boundary-layer separation ‘Heat transfer from tubes and other bodies in cross flow is determined by the stream velocity, turbulence level, physical properties of the fluid, thermal load, heat flux direction, geometry of the body, and some other factors. The most general dimension- less description is (6.6) To predict heat transfer by Eq. (6.6), js usually written in the following way: Nu=cRe"Pr" (61) where the Nusselt number Nu = hd,/k, the Prandtl number Pr= yc,/k, and the Reynolds number Re = uyd,/?. uid physical properties inthe heat transfer equation are described by the Prandtl number, which is about 0.7 for most gases. Therefore, from experimental data on heat transfer in air or in other gases, we have Nu= eRe" (68) “The process of heat transfer involves @ change of temperature, and consequently, variable fluid physical properties. Thus to account for the effect of fuid physical properties in heat transfer implies the prediction of the effect of their change on a temperature variation across the boundary layer. We encounter problem of choosing the so-called characteristic temperature, or reference temperature, for the physical properties. Different approaches to evaluating temperature dependence of fluid physical proper- ties are currently in use. We suggest the bulk mean temperature of the fluid as the reference temperature for moderate temperature gradients. This approach is simple and sufficiently accurate for practical purposes.6-4 CONVECTIVE HEAT TRANSFER IN CROSS FLOW ‘A marked influence of fluid physical properties on the heat transfer in viscous fluids is related to their dependence on the heat flux direction and on the temperature gradient, Experimental data, with reference to the bulk mean temperature, show that hheat transfer coefficients are higher for wall-to-luid heat transfer (fuid heating) than for fluid-to-wall heat transfer (fluid cooling). The difference increases with the tempera- ture gradient. To account for the heat flux direction, when fluid physical properties refer to the fluid bulk mean temperature, we introduce ratio Pr/Pr, with a proper power index p, where Pr, stands for Pr evaluated at the wall temperature. Thus we atrive at the following relation for the heat transfer from bodies in cross flow of viscous fluids: Nu nerre( 2) 9 where fluid properties refer to the bulk mean temperature, For gases, Pris constant and Pr/Pr, = 1. Both the fluid dynamics and heat transfer are also influenced by free-stream turbulence, geometry, surface roughness, etc. AA large number of heat exchangers employ tube bundles or other arrays. Both the fluid dynamics and heat transfer over bundles are again different from those over single tubes, because of the additional influence of the neighboring tubes. Still other types of flow are induced by the application of smooth or rough-surface or finned tubes. 6.2 HEAT TRANSFER FOR SINGLE TUBES AND BODIES 6.2.1. Fluid Dynamics over a Single Tube and a Sphere The velocity distribution over a circular cylinder in eross flow of an ideal fluid (no boundary layer) is deseribed by = sasinel+ (2)' (619 Here 7, and 7, are outside radius of the cylinder and the radial distance from its axis, respectively. The velocity is larger at smaller distances, and on the surface itself itis um 2u,sing (6.11) We now relate Eq, (6.11) to the Bernoulli equation P+ 4pu2, = constant (6.12) and find a functional relation between the velocity of the flow and the pressure coefficient = 1 4sintg (613) Thus from Fig. 6.2, the pressure coefficient for potential flow has two maxima at @ = 0° = 360° and $= 180°, and two minima in the medial cross section. The pressure coefficient is distributed symmetsically én i¢zal fuids.HEAT TRANSFER FOR SINGLE TUBES AND BODIES 65 0 o 120 180 240 300 360 4.08 Figure 62. Circumferential distribution ofthe pressure coefient over a eylinder in ross ow. “The distribution of the pressure coefficient is a reflection of the flow phenomena. On a cylinder in areal uid, their interaction with the viscous force gives rise ta laminar boundary layer, which is formed on the front part and whose thickness increases downstream, The main determining parameters of this layer are the Reynolds number Re and the turbulence level Tu. In Fig. 6.2, the curve for the subcritical flow represents the circumferential distribution of the pressure coefficient on a cylinder in the medium range of Re (Re < 2 10°), The kink in the curve at a point ¢ ~ 80° corresponds to the boundary-layer separation and to the formation of a complex vortical flow in the ‘With an increase of Re (Re > 2 10°), the flow enters the critical regime, and the corresponding distribution of Cp is shown in Fig. 62. The laminar boundary layer Separates on the front part point S, forms a separation bubble, and later reattaches at point B. Reattachment is followed by a turbulent boundary layer, which withstands the Jncteased pressure gradient and finally separates at @ = 140°. In the supercritical regime, at Re > 0.6 X 10°, the laminar-turbulent transition ‘occurs in a nonseparated boundary layer, and the transition point is shifted upstream ts shown in Fig. 62. The separation of the turbulent boundary layer occurs at @ between 120° and 140°, “The location of a laminar-turbulent transition in the boundary layer depends both oon the Reynolds number and on the turbulence level. Figure 63 shows the dynamic behavior of the laminar-turbulent transition as @ function of Re and Tu, according to measurements of fluid dynamic and thermal parametefs [1,2], For example, the transition from the laminar boundary layer initiates fat @ = 80° for Re = 43 x 10° and Tu = 7%. Oa a sphere, the distribution of pressure and velocity is analogous to that on @ circular cylinder (3) 6.2.2 Drag on a Cylinder and a Sphere “The total drag is generated by the friction force Fj and pressure force F, acting on a body in cross low. At very low Re, a cylinder and a sphere are streamlined, and their drag consists mainly of friction. Figure 6.4 presents the local skin friction coefficient ¢y Gna circular cylinder for various Re [2,4], We see that Z increases from 2er0 at geo° to a maximum value at ¢ = 60°, and again diminishes. The point which Judicate zero friction coefficient represents the boundary-layer separation. With a further increase of Re, the contribution of inertial forces begins to grow, so that in a highly vortical flow, skin friction drag constitutes just a few percent of the total drag100 Tues % 7 oa) és 6 _ y 2 oar 2 68 10 2 te Figure 63 Locations ofthe sprain point of te laminar boundary lyr for Re <2 10 sd ofthe anina-turbuent boundary layer tanson point foe Reo 2 8 108 tox Re and Te Oe 6 > > He gz 2 Re= 9.7108 [2} 4sx10 (a o 30 Cy 0 120 150 dee Figure 6.4. Friction coefiient ofa cylinder for variable Re.[HEAT TRANSFER FOR SINGLE TUBES AND BODIES 6+7 100 10 ‘subertcal [ Gyinder | Supercritical 19] Soher| oa wt 18 Re Figure 65. Total drag cefcieat ofa cylinder and a sphere for variable Re (1), Regions of Re {1 laminar, @) suberitical, (3) evtieal and (4) supereiial, IO? ‘A dimensionless expression for the total drag coefficient is 5+F - th , Co pueda on “The drag coefficient for a circular cylinder and a sphere is shown in Fig. 65 as @ function of Re. The drag coefficient decreases significantly with increasing Reynolds number in the low range of Re (region 1), due mostly to the contribution of the skin fhetion, In the subcritical flow regime (region 2), Cp changes insignificantly with Re. In the eritical flow regime (region 3), the total drag coefficient sharply decreases with Re. ‘This is due to a much narrower wake caused by the turbulent boundary layer and a downstream shift ofits separation, In the supercritical flow regime (region 4), the total drag increases again because the streamlines are displaced by the thicker turbulent boundary layer. Figure 6.5 includes a curve of the drag coeficent of a sphere, which is similar to that of a circular cylinder. “An increase in the turbulence level leads to an carlier onset of the critical flow regime with corresponding changes of the total drag coefficient as shown in Fig. 66. cylinder in the critical low regime (2,4). Here the surface roughness causes a higher drag coefficient and an earlier onset of the critical flow regime. 6.2.3 Local Heat Transfer for a Cylinder and a Sphere ‘The variable fluid flow over a cylinder in cross flow gives rise to similar variations of The Toeal heat transfer. On the front part, up to the separation of a laminar boundary layer, the heat transfer can be determined by either approximate or exact analytical6-8 CONVECTIVE HEAT TRANSFER IN CROSS FLOW 20 10 os. Cp 06. oa 03 3 4 6 8 10 2 4 6 8 x Re Figure 6.6, Total drag cocficient ofa cylinder for variable Tu and Re (1) 20 T efdy=0,T0= 1% (2) 10 e/d,=10(4) ee 08 i . efdy= 8X10", T= 1% (2) 06 oa Tee er ene 2 4 6 8 10 Re Figure 6.7. The effet of surface roughness ofa cylinder on its pressure drag coefficient techniques [5,6] In the lower range of Re (Fig. 68, Re = 500), the heat transfer on the front part of a cylinder is at its maximum. It gradually decreases with the development of a laminar boundary layer. At higher values of Re (Fig 68, Re = 104), heat transfer sradually increases downstream of the laminar boundary-layer separation and. is ‘mainly determined experimentally (1) Two heat transfer minima (Fig. 68, Re = 2 x 10*) are observed in the eritical Now regime, The fist is at the separation of a laminar boundary layer (# about 84°), where a separation bubble appears. The second minimum is atthe separation ofa turbulent boundary layer. Jn the supercritical flow regime, the first heat transfer minimum (Fig, 68, Re = 2 x 10°) corresponds to the laminar-turbulent transition in the boundary layer (@ ~ 30°)HEAT TRANSFER FOR SINGLE TUBES AND BODIES 69 Nu Re “05 p02 0 30 oo 30 720 750 780 4.00 Figure 6.8. Local heat transfer from a eylinder for variable Re 1} ‘The second minimum is at the separation of a turbulent boundary layer. Although the results of Fig. 68 are for Tu = 1%, we note again that the location of the laminar- turbulent transition depends on Re and Tu. Tn the supercritical flow regime, an inerease of turbulence level causes an upstream shift of the first heat transfer minimum, so that the region of the laminar boundary layer is narrowed. As a result total heat transfer from the front part is considerably augmented. Effects of higher turbulence levels ate in general insignificant in the rear part in the subcritical and critical (not shown in Fig. 68), or supercritical flow regime (chown in Fig. 6.9). A higher turbulence level causes heat transfer augmentation on the front part of a cylinder (2. ‘Heat transfer can be augmented by different influences on the laminar boundary layer, where the thermal resistance isthe highest. A higher level of turbulence augments the heat transfer through an external influence on the boundary layer. Internal influences are equally effective. Thus surface elements on a rough heat transfer surface turbulize the laminar boundary layer or even destroy it. Special studies suggest that a 60 to 80% augmentation of the heat transfer can be achieved with an optimal surface roughness (2,7. With an increase of Pr, the thermal resistance concentrates in the ‘iseous sublayer, so that heat transfer augmentation can be achieved by lower surface Uements. The additional turbulization of the boundary layer by surface elements is Similar to the effects of higher turbulence levels. With higher surface elements, heat iransfer augmentation is accompanied by an earlier onset of the critical low regime. ‘Local heat transfer on a sphere is similar to that on a circular cylinder {8}6-10 CONVECTIVE HEAT TRANSFER IN CROSS FLOW 28 w 26 Re=7.7%:108 % 22 & 18 12% 1a 10 ° 30 cy cy 120 150180 deg Figure 6.9. Local heat transfer from a cylinder to air in the supercritical flow regime for variable Twn} 6.2.4 Average Heat Transfer for a Cylinder and a Sphere ‘The average Nusselt number for cross flow around a cylinder depends upon the Reynolds and Prandtl numbers. It is shown in Fig. 6.10 for air and water heating with constant fluid properties. From Eq. (6.9) Nu Kah were Rem on Pe) cm {At low values of Re, the slope of the curve, which corresponds to the exponent m of Re, varies from Oto. Inthe higher subesal range of Ret inexeone to 6 tn the critical flow regime (Re 10°), the behavior of the heat transfer is difficult to define, but in the supereiial fow rege the exponent of Re inaeats 8 08 he the exponent m of Re varies from 0.4 to 0.8 [1]. ‘The exponent n of Pris equal to 037 in the subcritical flow regime, and 04 inthe supercritical ow regime (0.375 shown in Fig, 610, but 04 mee eee) ‘With physical properies evaluated atthe bulk mean temperature, the eet on the heat transfer for heating and cooling is satisfactorily approximated by (Pr/Pr, )?, Fer ‘For constant Muid properties, (Pr/Pr,)? = 1 and n= 0.37 for the ordinate of Fig. 6.10HEAT TRANSFER FOR SINGLE TUBES AND BODIES 6+ 11 2 SZ by 10} | S108 103 10" 10 10 Re igure 610, Average heat rans for ow normal oa etindr for Hult heating {12} wallto-uid heat transfer, the value of p is about 0.25% for fluid-to-wall heat transfer wa Sor A mean value of p = 0:25 can be assumed for bork Heat flux directions at x prerate temperature diferences. The data in Fig, 610 are satisfactorily described by Ea. (69). For the heat transfer from a eylinder {9 ero flow {1,2}, we recommend a. (6°? mutes for different ranges of Re as shown in Table 6 Here Re = wod./¥s seve "pe Muld properties are evaluated at the bulk mean tempers se feat ranafer to and from a. sphere is similar to that for cylinder and is determined mainly by Re. Heat transfer for a sphere is described by the following fequations for Re < 7 x 10° vuczvcanes vomsnenee( TE” (149 For 4x 105 < Re <5 x 10° and Pr = 0.71, Nu (495.9 + aRe + bRe? + Re?)Pr?* (6.160) where a= 5.167 x 10"4, b 0.288 x 10°, and ¢ = -3:58 x 10°". “TABLE 6.1, Host Transter Correlations for a Cylinder In Cross Flow TABLE 6.1. He Trane ee ee aange of ‘Recommended Correlation? Range of Re Nu = OTRO PONE E/Pra 109-4 x 10 Nu = 0.52 Re™PrO"(Pr/ Pra)” 4x 10-10 Nu = 0.26Re"*Pr°°"CPr/Pr)” 102-2 x 10° Nu = 0023 Re™*Pr°*(Pe/Pre)” 2x 108-10" Tp 2 025 for id eating and 7 = 020 for Ha cooling12 CONVECTIVE HEAT TRANSFER IN CROSS FLOW 6.2.5 Factors Influencing Heat Transfer Gr= sBATL jo? mn the case of mixed convection, the relation becomes Nu = f(Re, Gr, Pr, ®) (6.18) (Na 035)' = (02460 + 041604)" (05ReIS)) —( 19) yuh am increase of Re, heat transfer approaches asymptotically the forced convec: tion valu, and Eq, (6.19) for ar Now reduces to (13) Nu = 0.35 +0.5Re?S (6.20) Heat transfer from rough-surface cylinders hasbeen studied by Achenbach [4,7] in aie, and by Zukauskas et al. 2] in water (Fig. 6.11), In air, the average heat transfer {rom ‘Blinder: could be ugmented with an increase of relative roughens e/d,, where e is the oushness height. Similar to the effect of turbulence iene’ higher surface TeeenReS cases the onset ofthe critical ow regime at lower Re’ Ia woe liquids, Lbetmal resistance is concentrated inthe viscous sublayer. For fees heat transfer in such flows, lower surface elements are preferableHEAT TRANSFER FOR SINGLE TUBES AND BODIES 6+ 13 4 a T A we SA Zz : 6 4 2 10 & & ab £ a ae ‘mm Tu.% Ref . j ow 7 O 4 £ oo 1 Oy y 1 oy soo 4 oas (al 2 6 8 10 2 a 6 8 We 2 Re Figure 6.11. Average heat transfer from a rough surface cylinder for different Tu, The data from Ref. 7 ae for air; those from Ref. 2 are for water. 6.2.6 Heat Transfer from Tubes of Different Shapes Cross flow heat exchangers employ tubes of circular, elliptical, and other cross-section geometries. Different geometries introduce additional complications in the heat transfer prediction techniques. Zukauskas and Ziugida [2] proposed the following relation for average heat transfer from elliptical and circular tubes: seg =ormgsno( 2) ey i a ae oe ees sence eh tener aes a Noy, = ¢+ JNU}, tam + NF (6.22) where Ly is the stream length, or length along the surface, ¢.g. Ly = #dq/4 for ¢ = 06414 CONVECTIVE HEAT TRANSFER IN CROSS FLOW to 90° along the circular-cylinder surface. According to Gnielinski (14), ¢ = 0.3. The value of Nu, 4x has to be calculated from equations for a laminar boundary layer, and Nu, sy from turbulent boundary-layer correlations on smooth fat plates. 6.2.7. Heat Transfer From Tubes in Narrow Channels ‘eat transfer from a tube bounded by two flat walls depends on the ratio of the tube diameter (dp) to the channel height or spacing between the two plates (e,), Heat transfer studies [1,2] at Re from 10° to 10° showed that at constant Re and Pr, heat transfer increases with y = dy/e, and depends on the blockage factor. The experimental results are satisfactorily approximated by the relations in Table 6.1, but the reference mean velocity up in the minimum space in Re must be replaced by a reduced velocity value ug = (1 ~ y*)up. Then for the subcritical flow regime, the Nu correlation from Table 6.1 is modified as Nu = 026{(1 ~ ¥2)Re]*Pe" y (623) and for the supercritical low regime, Ea) : (624) Nu = 0.023[(1 - w Re) Peel If the free-stream velocity u,. is known, the reference velocity tig in Re, according to Perkins and Leppert [13], can be found from the following equation: van me all (628) 6.2.8 The Effect of Yaw Angle on the Heat Transter from a Cylinder ‘The average heat transfer from a single tube decreases with a decrease in the yaw angle, and is particularly low at small yaw angles. The effect is similar to tube bundles as shown in Fig, 6.22 (Sec. 6.3.3). A vast body of results on the heat transfer from yawed cylinders has been generalized by Morgan [5] 6.3 HEAT TRANSFER AND PRESSURE DROP FOR SMOOTH-TUBE BUNDLES 6.3.1 Fluid Dynamics in Smooth-Tube Bundles ‘A circular tube bundle is one of the most common heat transfer surfaces, particularly in shell-and-tube exchangers. Fluid flow is ideally normal to the tubes (= 90°), but some tubes operate at diferent yaw angles 8 to the flows ‘The most common tube arrays are staggered and inline (Fig, 6.12), although other arrangements are possible. Bundles are described by the ratio of the transverse (X¢ = X,/d,), longitudinal (X* = X//d,), or diagonal (X= X,/d,) pitch to the tube diameter. Inside a bundle, the flow converges in the intertube spaces and forms a highly turbulent flow over the inner tubes, The recirculation region in the rear of an inner tubeHEAT TRANSFER AND PRESSURE DROP FOR SMOOTH-TUBE BUNDLES 6+ 15 a i= dy x t ¢ leas s =o88 BBs e80 = O50 ae oo @ » @ Figure 6.12. Most common types of tube bundle arrangements: (a) ine, (b)-Ce) staggered ‘Menimom intertube spacing at setion I between two tubes js smaller than in a single tube. The situation is governed by the relative pitches and the bundle geometry. The more compact a bundle is, the larger is the difference from the Single-tube situation. Some differences depend on the number of longitudinal rows because of the inlet-outlet effects. Fluid flow inside a staggered bundle may be compared to a periodically narrowing and widening channel ‘Flow inside inline bundles approaches that in straight channels, and the mean velocity distribution in the minimum intertube space of a transverse row is highly influenced by the relative pitches. ‘The leading tubes induce a vortical low and a variable velocity distribution around the inner tubes. At low Re, the inside flow is predominantly laminar with large vortices in the recirculation regions. Their effect on the front parts of inner tubes is eliminated by viscous forces and by negative pressure gradients. Laminar boundary layers are sl formed on the inner tubes which separate and form recirculation region in the rear. ‘This pattern may be called a predominantly laminar flow. It is observed at Re 1000 ‘Significant changes are introduced at higher values of Re. The intertube flow becomes vortical and highly turbulent, On inner tubes, in spite of high turbulence, laminar boundary layers are still observed. "A nepative pressure gradient on the front part of an inner tube causes an accelera- tion of the flow, The boundary layer is thin and changes but little with the distance from the front stagnation point, ‘Both the intensity of turbulence and its generation in the intertube spaces are governed by the bundle geometry and Re, With shorter transverse pitches, the veloity fluctuations become more intensive Te turbulence Tevel of the main flow can influence fluid dynamics only over the first and second rows. A tube bundle acts as a turbulizing grid, and establishes Specine level of turbulence. Highly turbulent transient lows are observed on inner tubes in the intertube spaces. Ta set bundles, steady-state flow begins on the third row; On an inner tube, the istibution of pressure and velocity is very different from that on a single tube In Fig Gaye sce circumferential distbution of the pressure coefficient over the ‘ending sone and over the fourth row of an infine and a staggered bundle with 0 and Xj = 2.0 at Re = 10,800.6-16 CONVECTIVE HEAT TRANSFER IN CROSS FLOW 15 —F imine toe bunaie FP Stoazered tube bundle First tube ow os Fourth tbe row asa i -10) ° 30 60 90 120150 180 6.008 Figure 6.13. Pressure coeficient on a tube inthe fist and fourth tube rows of an inline bundle and a staggered bundle; air asthe test id [1 ‘The pressure coefficient of a tube in a bundle is given by Pooo > Py tous Ga1- (6.26) There u is the mean velocity in the minimum intertube space. It is obvious from Fig. 6.13 that in the subcritical flow regime,’ fluid flow phenomenon over tubes of the leading row is very similar to that over a single tube. But on an infer tube of Staggered bundle, higher pressure coeficients than on a single tube precede the Separation. On an inner tube of an inline bundle, a maximum pressure cosficient is observed at $= 40", at the point of attack. This means that there are two points of attack and two pressure maxima on a tube inside an inline bundle On a tube inside a staggered bundle, in a way similar to a singe tube, the stream is split atthe front stagnation point, and a laminar boundary layer begins o develop. A certain influence of Reis also evident. But unlike a single tube, based on heat traveler ‘sults, the boundary layer on a tube inside a staggered bundle separates at ¢ 190" ‘The separation is preceded by a laminar-turbulent transition, where the transition point is dependent on Re. At high values of Re, the boundary-layer separation fluctuate in the range of ¢ = (150 4 5)° [1] In symmetric staggered bundles of increased compactness (such as in Fig. 614), C, becomes a function of Re as early as $ = 20° for XP = 1.25 tube inside the {The suberitical, crea, and supererical egimes fora tube bundle are defined the same way a8 fora single tube, as shown in Fig 6,HEAT TRANSFER AND PRESSURE DROP FOR SMOOTH-TUBE BUNDLES 6-17 T T | “ BaP > PD 125x125 | oa LLY TT ° o 120) 180 ode igure 6.14, Pressure coeficient on a tube in te inne ows of staggered bundle in ait sabe bundle. For @ fom 30° 10 55% the caves have galt ak dus OF OO tue a om the preeling bes, The stb of Con 6 at of atubeis separate [om te ret staggered bundle 19 syne line buns 2 ta as ee 1a) nthe sabe ate of es a of Cis A A AS ean tow in Fig, 615s minimum value is ot @ Toe or. hase Sf tbe, Cy zene, buts eat conta Ae Pe The Te the rar ar Ot anr row of aline bundles become similar to these fr staggered cavers en the supercritical flow regime is established nes, when th Pty dstbuton i abe Dundes, he ec of eames eis pore. The rte of vloty growth i interte PaCS ‘mainly shoul ever s ansere pitch, and it intease rata marowes FO ermine by th pe minimum itera space of travers We FY [OF 8 Cane mean serve pitch and a constant flow rate is given bY xe My > Me Ye = (627) and the mean velocity in free flow area at an angle @ from he font stagnation point (see Fig. 6.12) is given by x (6.28) He Mee sing : -The velocity distributions inthe intertube spaces are notably different in staggered ‘and inline bundles. inline tubes of staggered bundles, the point of attack is located at the stagnation point, and the maximum velocity oFcurs the minimum intertube spaces‘18 CONVECTIVE HEAT TRANSFER IN CROSS FLOW I x 105 Re=6x 10% F > 10K 1.25% 1.25, os \ / / 06 [i i } }-————} Re = 2x 195. NY ea os i Roi Re=6x 108. J Wy U7 fs . ee ae 4.40 laure 6.18. Presurecocficient on ate in iner rows of inline bundle in aie (1, are tid dynamics is mainly determined by the location ofthe minimum free flow area Gi ther # = 90 or # & 90°. In the tear part of an more tube, the velocity sharply decreases, and even acquires a negative value in the recirculation region as shown in Fig. 6.16. Over inner tubes of inline bundles, the point of attack ies at 4 = 40°; the velocity iat ts minimum there and increases downstream. Facthe: circumferential variations oF local velocities are similar to those in staggered bundling 6.3.2 Drag on Smooth-Tube Bundles dn ube bundles in cross ow, the total dag also consists of friction and pressure (or crema) drag. The drag on a tube ina bundle is dcet st by its skin friction coeficient cy and pressure drag coefficient c.- The skin theta’ Coctlicient is defined as 2n, Pus y (6.29) where +. is the shear stress at the wall of a tubeHEAT TRANSFER AND PRESSURE DROP FOR SMOOTH-TUBE BUNDLES 6°19 Fist ow ith ow Taal i a = | =g2 ae é i e : Dt ee ie = nt vt inet) et PMaggered bundle atthe distance »/d, = 0019 [1 “The local shear stress 1, is determined as a produet of fluid viscosity and veloc tradient atthe wall, Thus for known velocity distributions in the ‘boundary layer, the aeaerst s.(@) may be determined for a region from the front stagnation Paint TANS the Year stagnation point or the separation point. The total friction drag is Evaluated by integrating 1,() over the circumference. “The pressure drag coeficient is defined as 2, o ert (630) where R= BAY Roose, a) ad,L A= (6.32) Fy is the longitudinal component of the pressure fore, 4 is the WBE Se section Fexpendiular to the flow, 1 isthe tube length, 4 is the ede surtane clement between resreretrating lines onthe tbe, drawn through the centers of two measured a8. Iwo Bement measured areas, P, isthe pressure on the (th measured asa vet We aoe eeeance, and @, is the angle measured from the stagnation point fo the e=nier of the area 4 around the circumference. res 4 akin fiction coefficients of thee staggered bundles in the lower, medi and higher ranges of Re are presented in Fig 617. As already mentioned = 9 2 the points of attack and of separation, With an increase of @y the Yas ‘ofc; increases and Point of anim at from 50 to 80°, depending upon Re and the bundle Bowe ‘Rhter the separation, the value of ¢ becomes negative, though it never reaches large absolute values.6-20 CONVECTIVE HEAT TRANSFER IN CROSS FLOW 02 o12 Re 107 oo16 105 4 008 125% 125 108 7 0008 oot g ‘0.000 ° 201i 60/9 90's 1206 -ate0 ea a0 6.deg Figure 6.17. Friction coeficient over a tube in ast taggered bundle for variable X* > Xi* and Re; aviation ol asthe test fu (1) Frat inner tubes of inline and staggered 2.0 x 20 bundles, the pressure drag Fe rasans Deins to decrease sharply from Re = 10° (Fig. 6.18) and beosmes sable st Re from 3 x 10” to 10* as well as at Re from 10° to 10°. Selow Re, the values of cand cy are of the same order. At higher Re, the vale of for commen cantly that of cy. For example, at Re = 3 x 10°,thevatio o/e, = 0.1 for staggered bundles, and becomes even less at higher Re {Ine combined friction and pressure drag over tube bundles in cross ow constitute nei hydraulic drag, Consequently, the hydraulic drag depends on the comfigann ng ‘bundle. The hydraulic drag is also proportional othe numberof longitudigal moe sn 16 7 me Ata} 20x20 Experimental points 5 3 © Stoegered omer 4 Inline LT [seep os sh Sq || + Inline a UT 108 108 108 108 10° Re Figure 6-18. Pressure drag coeicieat in inner rows of staggered and intne bundles 1}HEAT TRANSFER AND PRESSURE DROP FOR SMOOTH-TUBE BUNDLES 621 and for small numbers of tube rows the contribution of the inputoupus foes of aie eeerny is significant, They may be significant ina short bundle ( sal ‘umber vine ce The total drag in a heat exchanger also depends on the physical properties of the fluid. vpertthe basis of the above considerations, the pressure drop over & bundle is implicitly given by AP =f( sor Xrs Xin dor BoB P)+ (6.33) a dimensionless form by xX, ovnifte 2) 0 wnere the Euler number Eu = 2.4P/(pujn) and is defined per tube row. Here Ho the mean velocity in the minimum intertube spacing The revults are usualy correlated in the following form for enginesring WS Eu= eRe" (6.238) -The design data for the hydraulic dag over banks of tubes ae useful fox cosines: sng. Such experimental results are presented in Figs, 6.19 and 62) rom Zukauskas tra [16} and from recent experiments with different Muids carried out at the Institute ee sland Technical Problems of Energetics, Academy of Sens of te oO rsial eR Closedcform equations for the results of Figs. 619 and 62) ste arsed in (3), At high values of Re, the data coincide with the reals Of nati, Fee Ey Hammeke etal (17) and Niggeschmit (18), and at lower values of Re with the results of Bergelin (291, ith the feet qverage Eu pet tube row for multirow infne bundles are presen in Fig fig a a funtion of Re and X¢ (~ 2). For other whe pies (7 9 Fig. 619 a ior in fist determined from the inset of Fig, 6.19, and then B/x fom ‘pe main figure to find Eu for a specified inline tube bundle main Ba erage Eo pe tube row for multiow staggered we bundles With eT ey and = (13 /2)%] ae presented in Fig, 620 a fctons oR tnd. X2, For other tube pitches [X; + 0/3 /2) Xi) the, correct factor x is first an sf the inset of te igure, and used in Eu/x obtained from the min Hebe ronnyen X, to find Ea for a specie staggered tube bundle, Eu values determined from Figs. 6.19 and 6.20 have an uncertainty within £108. The pressure drop of a multirow bundle is then given by spo(E)e ted a ee 39 Here & is the hydraulic drag coefficient ei ae a Eu for the whole bundle increases (at a decreasing, 0) with the snuabey of tube rows, An average value of Eu fr one row is found by dividing the total10 snoues pur “aye ae spimy 1s) 9yp sotouesed © se yo Teurpo uo ayn sey Yup wee 241 {91 6 < 20 spa aut Jo apuND Hop ogmapsy oy He aed ne ay 901 OT #00 eT zor {or € ose re oz (-Loat=2xy Lost v I vo 20 10 fz - roto I 20 vo ; " g0l ‘ 9 and is obtained from Fig. 6.19 or 6.20, The charts in Figs. 6.19 and 6.20 refer to isothermal conditions. They also apply to nonisothermal flows if the fluid physical properties are evaluated at the bulk mean temperature and a correction is applied to account for variable fuid properties for liquids as follows: Ba, = Bo)” (6.38) Here Eu, is the Euler number for both heating and cooling, Eu the Euler number for isothermal conditions, and 1, and ye are the liquid dynamic viscosities at the wall temperature and at the bulk mean temperature, respectively. For Re > 10°, p = 0. For Re < 10%, = -0.0018Re + 0.28 — for liquid heating (639) P= ~0.0026Re + 0.43 for liquid cooling (6.40)HEAT TRANSFER AND PRESSURE DROP FOR SMOOTH-TUBE BUNDLES 6+ 25 10 os + Staggered % osf—, Inline > a oa 03. Pr a a ee pce Figure 622. Coreton factor c forthe hydraulic dag of staggered and inline Dundes 0} Many other investigations have been reported worldwide, presenting Eu or tts variants for various tube bundles, The most notable is by Gaddis and Gnielinski (20) Who suggested curves and formulas for the pressure drop for inline and staggered tube bundles. 6.3.3. Drag on Yawed-Tube Bundles ‘The drag on a yawed-tube bundle depends, as in crossflow eases, on the bundle seorhetry and pitches, as well as on the Reynolds number, the yaw angle, and some Biher factors, Figures 6.19 and 6.20 apply to obtain their drag, but with a correction factor for the effect of the yaw angle B. Based on the experimental results, the drag decreases with decreasing fat constant flow rates. The decrease of drag for 8 < 90° is related to an altered flow in the bundle. ‘The hydraulic drag on a tube consists of the friction drag and pressure drag, and any change of ft involves a change of the ratio between the two components, as well as of the flow phenomena, “To determine the hydraulic drag on a yawed-tube bundle (f < 90°), a correction factor Evy oe (6.41) 8 Bupa ust be introduced for the yaw-angle effect. It is a ratio of Euler numbers defined in Eq, (6.41) for a constant flow rate and is presented in Fig, 6.22 6.3.4 Local Heat Transfer from a Tube in a Bundle ‘The general laws governing the distribution of local heat transfer from a tubs fy tae swe the same as those for a single tube. The circumferential distribution of the ant ransfer is governed by flow phenomena, which in turn depend on the bundle trometry; eB, two poins of attack, and consequently two heat transfer maxima, existCONVECTIVE HEAT TRANSFER IN CROSS FLOW 13 =] 12 Intine tube ‘bundle 10 Staggered tube bundle 08 al? 06. Single tube 04 02 030 608020150 Ta0 6. deg Te 629, Local heat rans roma single tube, a tube in an nner row ofa taggeed bundle, Aan cen a imme tow of an inline bundle for Kx x"= 22 and Re = 4% IO" Heh Air i the test uid On 2 tube inside an inline bundle. The flow over a tube inside a staggered bundle is fend ilar to that in a single tube, but with higher turbulence in the surrounding fluid. A laminar boundary layer on the front part, formed by a split stream fer ae inline bundle, exists only at the lower values of Re. Tocal heat transfer from a tube inside an inline bundle and a staggered bundle is fompared with that from a single tube in Fig. 6.23. Inside a staggered bundle, heat transfer from the front part is higher than that on a single tube, because of the impingement of the two upstream fluid streams on the front part and because of higher {urbulence. On the remaining part, heat transfer is higher because of higher turbulence Inside an inline bundle, maximum heat transfer is observed at @ ~ 50°. which is the Point of attack of a stream coming from the preceding intertube space. Because of the lateral point of attack, which is reflected in a sharp heat transfer maximum, a laminar boundary layer begins its development not at ¢ = 0°, but at @ from 30 to SO" At higher angular distances, the heat transfer decreases with growth of the laminar boundary layer thickness, Figure 6.24 shows peripheral local heat transfer coefficients for tubes in an inline bundle. Tubes of the second row and onwards are under the influence of the leading row; therefore their heat transfer is different. A steady state in the inline bundles is stablished from the fourth row onwards. Local heat transfer is similar for inline bundles of different relative pitches (1.25 < X* < 2.0), with the boundary-layer sep. aration at ¢ = 120°, Now we consider peripheral local heat transfer coefficients for tubes in a staggered bundle. As shown in Fig. 6.25, the trend is similar to the single-tube case fee the teading row. A tube in the second row is under a siteam which comes from the leading row, so that the heat transfer from its front partis strongly increased, The flow is turbulized in the two leading rows, and this is immediately reflected in the heat transfer, The circumferential heat transfer from a tube in the thied sow inHEAT TRANSFER AND PRESSURE DROP FOR SMOOTH-TUBE BUNDLES 90 20 < 0 2 Fistrow ~ H Second ow A 2 0 fener on) 2 10 oC 6, dea Figure 6.24. Local heat transfer coefiients from a tube inthe lading row, the second row, and [miner row of an inline XX" = 20 X 20 bundle at Re = 1.4 x 10* (1,16). Airis the test ‘uid higher than that from a tube in the second row. Over the subsequent rows, the steady-state behavior is the same as over the third row. Heat transfer from the rear parts of inner tubes is always higher than that from the firstrow tubes, because of higher turbulence. Tn the critical flow regime, heat transfer increases at # = 120°, the point of the taminar-turbulent transition. At ¢ = 150°, boundary-layer separation is observed in 0 %0 Inner om < © } : 2 Second) 2 0 Fest om 20 10 oC] 6.dee Figure 6.25. Local heat transfer coefficients from a tube inthe leading row, the second row, and 3 Xf = 20 X 20 bundle at Re = 1.4 X 104 (1,16), Air fan inner row (mn > 4) ofa staggered is the test id6-28 CONVECTIVE HEAT TRANSFER IN CROSS FLOW ‘most configurations. In the superritical flow regime (for Re > 2 x 10°) in staggered bundles, the first heat transfer minimum corresponds to the laminar-turbulent transi- tion, which begins to shift upstream and reaches = 25° at Re = 10°, The second heat transfer minimum, which corresponds to the separation ofa turbulent boundary layer, is practically table in staggered bundles, and remains at ¢ ~ 150° even at higher Re 6.3.5 Average Heat Transfer from a Tube in a Bundle ‘The average heat transfer from bundles of smooth tubes is generally determined by Ea, (6.9). Studies {1, 16,21] in the ranges of Pr from 1 to 10,000 and Re from 1 to 2 x 10° Suggest a constant exponent of Pr, which is n = 0.36 for the subcritical low regime and n = 0.40 for the supercritical flow regime at low Prandtl numbers (0.2 < Pr = 1). The exponent of the ratio Pr/Pr, can safely be approximated with a constant value of p= 025, ‘The effect of the bundle arrangement on the average heat transfer varies with Re, In the low range of Re (+s 50), the heat transfer from a tube in the leading row actually ‘coincides with the single-tube and inner-tube values. In the higher range of Re (> 50), ‘heat transfer rates from inner tubes are higher than those from the leading row because of increased intertube turbulence due to the leading rows acting as turbulizers. In most bundles, steady-state heat transfer is established from the third or the fourth row on. A comparison with heat transfer coefficients from the leading rows and from the inner ‘ows illustrates heat transfer augmentation in bundles due to higher turbulence. As a rule, heat transfer from inner tubes increases with decreasing X,. This is in ‘agreement with observations on single tubes mounted at different distances from a ‘urbulizer grid, An exception is presented by inline bundles of short X), where heat ‘transfer may be reduced even more with further reduction in X,. This phenomenon is ‘elated to a less intensive recirculation in the rear, especially for low or medium values of X, Because of the higher turbulence, under the influence of a variable tube longitudinal pitch, heat transfer from the inner tube may be from 30 to 100% higher than from the leading row. In most cases, heat transfer from the second row is lower than from the further rows, except for inline bundles at low Re. Figure 6.26 shows the heat transfer from a staggered and an inline 2.0 x 2.0 bundle. ‘At Re < 10°, the heat transfer from a tube in the leading row of the inline bundle is about 25% higher than that from the inner tubes, Any inner tube is in the wake of a receding one, and its heat transfer is decreased accordingly by a lower velocity in the wake—the so-called “shadow” effect, In contrast to the inline bundle, heat transfer in the staggered case is lower in the leading row than in the inner rows. For a 2.0 x 2.0 bundle at Re ~ 30, the difference reaches 7%, but increases to 35% at higher Re. This is due to different flow phenomena in the two configurations In the range of predominantly laminar fluid flow, the exponent of the Reynolds ‘umber varies widely with Re. At Re < 10?, we have m = 0.33 to 0.4, but m= 0.5 at higher Re (-s 10°), and reaches 0.6 or even 0.63 with the increase of Re for the predominantly turbulent fluid flow in the subcritical zone (Re < 2 x 10°). For inline bundles at higher Re, the heat transfer from inner rows exceeds that of the first row, as for staggered bundles, Figure 6.27 presents the average heat transfer from the first and inner rows of a 1.5 1.25 inline bundle, and from the inner rows of 2.0 x 1.25 inline bundle. At Re <2X 10%, the exponent m of Re is between 0.6 and 065, but at higher Re it