BOOKCOMP, Inc. — John Wiley & Sons / Page 765 / 2nd Proofs / Heat Transfer Handbook / Bejan ENHANCED IN-TUBE CONDENSATION 765 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 [765], (47) Lines: 1291 to 1316 ——— 4.89717pt PgVar ——— Normal Page * PgEnds: Eject [765], (47) typically have 30 to 80 trapezoidal fins that spiral down the tube axis with a helix angle (α) between 6 and 30°. The fin height (e) ranges between 0.1 and 0.25 mm, which is approximately less than 0.04D o . The fin apex angle β varies from 20 to 60° for commercial microfin tubes, where most are close to the middle of this range. Several factors are responsible for the passive enhancement of condensation in microfin tubes. First, the fins provide additional heat transfer area over that of a smooth tube of the same cross-sectional flow area. The fins add in the range of 50% more surface area per tube length. Surface tension is another mechanism that can act to thin the film on the fin tips for very high vapor qualities and low mass velocities. Yang and Webb (1997) have modeled surface tension effects on the fin and vapor shear effects for small-diameter extruded aluminum tubes with microgrooves. Swirl effects due to the riffling of the fins along the tube axis may improve heat transfer for certain operating conditions. Finally, the fins enhance heat transfer as a roughness would in the mixing of the flow at the wall. Currently, the prediction of convective condensation heat transfer in microfin tubes is handicapped by the lack of flow pattern maps for microfin tubes. Consequently, correlations and models that presently exist for the prediction of the condensation heat transfer and pressure drop in microfin tubes do not use them. Cavallini et al. (2000) provides an excellent review of predictive correlations/models for flow condensation in microfin tubes. Cavallini et al. (2000) shows that the models of Cavallini et al. (1993, 1995), Yu and Koyama (1998), and Kedzierski and Goncalves (1999) produce similar results for various tube geometries and mass fluxes. The Kedzierski and Goncalves (1999) model is presented here because it has the simplest form and is based on the largest number of data points. Although the Kedzierski and Goncalves (1999) model is based on 1489 data points, this data set was for the same tube geometry. The hydraulic diameter concept was used to generalize the correlation to other microfin true geometries. The hydraulic diameter of the microfin tube is defined as D h = 4A c cos α N f S (10.80) where S is the perimeter of one fin and channel taken perpendicular to the axis of the fin, N f the number of fins, A c the cross-sectional flow area, and α the helix angle of the fin. Equation (10.80) was used to derive an expression for the hydraulic diameter for the geometry given in Fig. 10.14: D h = πD 2 r − 2N f t b e cos α N f [ b r + 2e/ cos(β/2) ] (10.81) The local Nusselt number (Nu) was calculated based on the actual inner surface area of the tube as Nu = h 2φ D h k l (10.82) BOOKCOMP, Inc. — John Wiley & Sons / Page 766 / 2nd Proofs / Heat Transfer Handbook / Bejan 766 CONDENSATION 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 [766], (48) Lines: 1316 to 1366 ——— 1.6844pt PgVar ——— Normal Page * PgEnds: Eject [766], (48) The convective–condensation Nusselt numbers (Nu) were correlated following the law of corresponding states philosophy presented by Cooper (1984). Cooper sug- gested that the fluid properties that govern nucleate pool boiling can be well rep- resented by a product of the reduced pressure (P r /P c ), the modified acentric factor −log 10 (P r /P c ) , and other dimensionless variables to various powers. The above reduced-pressure terms and several other locally evaluated terms were used to corre- late locally measured convective condensation Nusselt number for the microfin tube: Nu = h 2φ D h k l = 2.256Re β 1 lo · Ja β 2 · Pr β 3 l P r P c β 4 −log 10 P r P c β 5 · Sv β 6 (10.83) where β 1 = 0.303 β 2 =−0.232x q β 3 = 0.393 β 4 =−0.578x 2 q β 5 =−0.474x 2 q β 6 = 2.531x q and the ranges for which the correlation holds are 3500 ≤ Re lo = G T D h µ l ≤ 24,000 0.004 ≤ Ja = c pl (T sat − T w ) λ ≤ 0.16 1.7 ≤ Pr l = c pl µ l k l ≤ 3.6 0.86 ≤ Sv = ν g − ν l ν ≤ 10.3 0.22 ≤ P r P c ≤ 0.62 0.06 ≤ x q ≤ 1.0 where the liquid only Reynolds number (Re lo ), the Jakob number (Ja), the liquid Prandtl number (Pr l ), the reduced pressure (P r /P c ), the dimensionless specific vol- ume (Sv) and the quality (x q ) are all evaluated locally at the saturated condition. The Reynolds number is based on the total mass velocity (G T ). 10.7.2 Microfin Tube Pressure Drop Kedzierski and Goncalves (1999) fitted Fanning friction factor (f r ) data for a 60-fin, 8.91-mm-root-diameter microfin tube to the equation f = 0.00228Re −0.062 lo Φ 0.211 3500 ≤ Re lo = G T D h µ l ≤ 24,000 (10.84) BOOKCOMP, Inc. — John Wiley & Sons / Page 767 / 2nd Proofs / Heat Transfer Handbook / Bejan ENHANCED IN-TUBE CONDENSATION 767 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 [767], (49) Lines: 1366 to 1409 ——— 0.25208pt PgVar ——— Normal Page * PgEnds: Eject [767], (49) The exponent on the two-phase number (Φ = ∆x q λ/∆L g ) given in eq. (10.84), 0.211, is consistent with that given by Pierre (1964): 0.24. However, the exponent on the Reynolds number, −0.062, is very different from that given by Pierre (1964): −0.24. The exponent on the Reynolds number of eq. (10.84) is approximately equiva- lent to −0.06 which is the exponent that one would calculate from the transition zone of the Moody (1944) chart using the fin height (0.2 mm) for the roughness height. The fact that the Reynolds number exponent is consistent with that obtained from the Moody (1944) chart suggests that the fins of a microfin tube act like a roughness to enhance the convective condensation heat transfer. If roughness mixing dominates the enhancement mechanism, neither swirl effects nor surface tension drainage have much influence on the heat transfer. The lack of importance of surface tension and swirl flow may be a consequence of the flow conditions and surface geometry. The corroboration between the present Re exponent and Moody’s (1944) also suggests that the frictional pressure drop of microfin tubes should depend on the fin height/root-tube diameter (e/D i ) ratio. If it is assumed that the fins act purely as a roughness, the Moody (1944) chart can be used to interpolate between the foregoing friction factor equation and Pierre’s (1964) smooth tube friction factor for a given e/D i ratio as follows: f = 0.002275 + 0.00933 exp e/D i −0.003 Re −1/4.16+532(e/D i ) lo Φ 0.211 (10.85) The pressure-drop equation for which eqs. (10.84) and (10.85) are valid is ∆P = f(ν o + ν i ) ∆L D h + (ν o − ν i ) G 2 T (10.86) where ν o and ν i are the specific volumes of the exiting and entering flows, respec- tively, while G T is the total mass velocity of the flow. 10.7.3 Twisted-Tape Inserts Twisted-tape inserts have been used to enhance heat transfer since the nineteenth century. Marine steam boilers were fitted with retarders (twisted tapes) to reduce coal consumption. Royal and Bergles (1978) surveyed horizontal convective condensation with twisted tapes and found improvements in heat transfer coefficients by as much as 30% over empty tube condensation. In addition, the most popular use of twisted tapes in flow boiling is to delay the occurrence of burnout. Figure 10.15 shows a schematic of a twisted-tape insert in a tube. The flow en- hancement of the twisted tape arises primarily from increased flow path length and swirl mixing. Swirl flow and increased path length are also expected to benefit two- phase heat transfer. Manglik and Bergles (1992) suggest that swirl flow effects may be correlated by the swirl parameter Sw: Sw = Re s √ y = Re D √ y 1 + (π/2y) 2 1 − 4δ/πD i (10.87) BOOKCOMP, Inc. — John Wiley & Sons / Page 768 / 2nd Proofs / Heat Transfer Handbook / Bejan 768 CONDENSATION 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 [768], (50) Lines: 1409 to 1447 ——— 0.92316pt PgVar ——— Normal Page PgEnds: T E X [768], (50) Figure 10.15 Cross section of twisted-tape insert in a tube. where Re D is the all-liquid empty-tube Reynolds number based on liquid properties G T D i /µ l , and y is the twist ratio of the tape. The twist ratio is defined as the ratio of the 180° twist pitch normalized by the internal diameter of the tube, D i . The thickness of the tape is δ. The 1/ √ y parameter accounts for convective inertia effects that are ignored by 1/y, which has been used traditionally to correlate swirl flow data. Manglik and Bergles (1992) show that laminar flow heat transfer for three different twist ratios are correlated well with the swirl parameter. Agrawal et al. (1986) also correlated R-12 swirl flow boiling heat transfer coefficients to y −0.5219 for Re D approximately in the range 7000 to 14,000. Consequently, it is inferred that the 1/ √ y parameter is suitable for correlating flow boiling and convective condensation data for low and somewhat high Reynolds numbers. Kedzierski and Kim (1998) correlated 2253 locally measured convective con- densation Nusselt numbers for refrigerants R-12, R-22, R-152a, R-134a, R-290, R-290/R-134a, R-134a/R-600a, R-32/R-134a, and R-32/R-152a with a y = 4.15 twisted tape to the following equation: Nu = 0.00136Sw α 1 · Pr α 2 l · P α 3 r (−log 10 P r ) α 4 · Ja α 5 (10.88) where α 1 = 0.613 +0.647x q α 2 = 0.877 α 3 =−1.735 + 2.362x q α 4 =−2.815 + 4.197x q α 5 =−0.528, and the ranges for which the correlation holds are 2200 ≤ Sw ≤ 18,000 0.14 ≤ P r = P P c ≤ 3.6 2.4 ≤ Pr l = c p,l µ l k l ≤ 3.6 BOOKCOMP, Inc. — John Wiley & Sons / Page 769 / 2nd Proofs / Heat Transfer Handbook / Bejan FILM CONDENSATION ON TUBE BUNDLES 769 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 [769], (51) Lines: 1447 to 1462 ——— 1.55309pt PgVar ——— Normal Page PgEnds: T E X [769], (51) 0.008 ≤ Ja = c p,l (T sat − T w ) λ ≤ 0.1 0.023 ≤ x q ≤ 0.9 Equation (10.88) correlated 95% of all the condensation Nusselt numbers to within approximately ±20% for R-12, R-22, R-152a, R-134a, R-290, R-290/R-134a, R- 134a/R-600a, R-32/R-134a, and R-32/R-152a. The single equation was found to predict single-component and zeotropic Nusselt numbers equally well. This suggests that mass transfer has a negligible effect on condensation heat transfer with twisted- tape inserts for the low relatively temperature glide mixtures R-32/R-134a and R-32/ R-152a. The intention of the swirl parameter is to generalize the correlation to other twist ratios, but this aspect of the correlation has not been validated. 10.8 FILM CONDENSATION ON TUBE BUNDLES In this section we address how to size or rate multitube condensers using the predic- tion methods already discussed in this and earlier chapters. These prediction methods and most published data apply for a single tube. With large condensers, which some- times contain many thousands of tubes, other factors must be considered because they can drastically reduce the performance compared to the single-tube results. The four major issues that adversely affect the condensor performance are (1) maldis- tribution of the shell-side and tube-side fluids, (2) shell-side condensate inundation (condensate from adjacent tubes), (3) saturation-temperature depression due to the vapor pressure drop, and (4) noncondensable gases. Both shell-side and tube-side condensations are addressed. The common char- acteristics of most shell-side condensations are horizontal tubes, single-component shell-side condensation (limited inlet noncondensable gases), and either tube-side forced convection or in-tube evaporation. The vapor generally flows downward in crossflow to the tube bundle. This configuration, called an X shell (Fig. 10.16), is the most common, especially if the shell-side pressure drop can affect the thermal per- Figure 10.16 Longitudinal view of a typical X-shell condenser. BOOKCOMP, Inc. — John Wiley & Sons / Page 770 / 2nd Proofs / Heat Transfer Handbook / Bejan 770 CONDENSATION 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 [770], (52) Lines: 1462 to 1487 ——— 1.82507pt PgVar ——— Normal Page PgEnds: T E X [770], (52) formance negatively because of the saturation temperature depression. For tube-side condensation, the discussion is restricted to the crossflow configuration. This design type is commonly used for all gas-cooled condensers. 10.8.1 X-Shell Condensers (Shell-Side Condensation) Consider first the adverse effects of tube-side flow and temperature maldistributions on X-shell condensers. Next, a discussion of different condenser rating methods deal- ing with the other above-mentioned adverse bundle effects is presented. Some final comments are given regarding good noncondensable gas management and venting practices. Tube-Side Flow and Temperature Maldistribution The major cause of tube- side flow maldistribution is a poor design of the inlet header. Flow variations (maxi- mum/minimum flow rates per tubes) as much about 1.5 were measured by Gotoda and Izumi (1977). They also supplied an equation to estimate the velocity maldistribution: u g,max u g,av 2 = 1 + 1 K t A tube A pipe 2 − A tube A plate 2 (10.89) where u g,max is the maximum velocity; u g,av is the average or mean velocity; A tube is the total inside cross-sectional area or n t (π/4)D 2 i , n being the number of tubes and D i the inside tube diameter; A pipe is the nozzle or pipe cross-sectional area; A plate is the inside area of the tube sheet; and K t is the total loss coefficient. This coefficient is calculated with the equation K t = K ent + K ext + K box + fL D i (10.90) where K ent is the tube entrance loss, K ext is the tube exit loss, K box is the box loss, f is the friction factor, and L is the tube length. Note that the maldistribution, u g,max /u g,av , decreases as the loss coefficient increases. Rabas (1985a) showed that only a 10% reduction of the outside condensing coefficient occurred with a u g,max /u g,av value of 1.66. With a u g,max /u g,av value of 1.22, only a 1% reduction occurred for different lengths, nozzle angles, and condensing vapors. The simple calculation of u g,max /u g,av will quickly determine if tube-side distribution is an important consideration. A nonuniform inlet temperature distribution can exist for some X-shell condenser applications, such as the multistage flash evaporators used in the desalination industry and the multistage condensers sometime used in power plants. For both applications, the overall tube length is divided into stages with partitions perpendicular to the tube axis. Each stage operates at different pressure levels. The outlet temperature distribution of the lower-temperature stage is the inlet temperature of the higher- temperature stage; hence, a nonuniform inlet temperature distribution is obtained. Rabas (1987a) showed that these typical nonuniform temperature distributions have almost no effect on the thermal performance. BOOKCOMP, Inc. — John Wiley & Sons / Page 771 / 2nd Proofs / Heat Transfer Handbook / Bejan FILM CONDENSATION ON TUBE BUNDLES 771 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 [771], (53) Lines: 1487 to 1518 ——— 6.12808pt PgVar ——— Normal Page PgEnds: T E X [771], (53) Condenser Sizing Methods We still need to consider the adverse effects of condensate inundation, noncondensable gas pockets, and the saturation temperature depression on the performance of X-shell condensers. There are three methods that are used to size and/or rate these condensers: (1) vertical row-number correction, (2) bundle-factor method, and (3) pointwise or numerical computer programs. Both the vertical row-number and bundle-factor methods use a single evaluation of the overall heat transfer coefficient using the standard summation of the separate heat transfer resistances. The major uncertainty in the calculation is the average, outside condensing heat transfer coefficient, h o,av . Vertical Row-Number Method The vertical row-number method is commonly used to size and rate small condensers (tube counts less than about 2000) with limited inlet concentrations of noncondensable gases. For small condensers, shell-side mal- distribution and saturation temperature depression normally are not important con- siderations. A correction is made to the outside single-tube condensing coefficient to account only for the condensate loading or inundation for the additional tubes within the bundle: h o,av = f(n t )h o,Nus (10.91) where h o,Nus , the outside condensing coefficient, is predicted with the single-tube Nusselt equation or eq. (10.12) and f(n t ) is the bundle vertical row-number correc- tion, an empirical function of the number of tubes in a vertical row. Some f(n t ), expressions are f(n t ) = 1 n t 0.25 Jakob (1949) (10.92a) 1 n t 0.1667 Kern (1958) (10.92b) 0.6 + 0.4353 1 n t 0.2 Eissenberg (1974) (10.92c) Marto (1984) compared these predictions with a broad range of these results and found that the data fell between the Eissenberg and Jakob (commonly referred to as Nusselt) predictions, the latter being the most conservative. Before using the correction method, an estimate of the shell-side pressure drop is recommended to determine if the saturation temperature depression can be ignored. This estimate can be made based on the following assumptions: 1. The inlet velocity u g,i is uniform and can be calculated with the continuity equation (m g,i = ρ g wLu g,i ), where m g,i is the inlet vapor flow rate, L the tube length between the tube sheets, and w the minimum free-flow width for the top transverse row of the bundle. BOOKCOMP, Inc. — John Wiley & Sons / Page 772 / 2nd Proofs / Heat Transfer Handbook / Bejan 772 CONDENSATION 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 [772], (54) Lines: 1518 to 1552 ——— 0.0pt PgVar ——— Long Page PgEnds: T E X [772], (54) 2. The single-phase crossflow pressure-drop prediction methods and friction- factor correlations applies and the Reynolds number can be calculated based on u g,i /2. 3. The distance of travel is the bundle depth, the distance between the top and bottom tube rows. 4. The velocity varies in a linear manner as the vapor flows through the bundle (a one-third reduction in the pressure drop or ∆P prediction evaluated using u g,i as the velocity value). No concern is needed if the saturation depression, ∆T sat , based on this ∆P value is small with respect to temperature difference between the inlet vapor and coolant tem- peratures. This simplistic approach was found to yield reasonable estimates for most feedwater heaters, some very small power plant condensers, and some condensers in smaller multistage flash evaporators. The other restriction on the use of the row-correction method is that the non- condensable gases will not affect the performance significantly. Rabas and Mueller (1986) showed that noncondensable gases will not have a measurable impact if the mass ratio m g,i /m a,i is greater than 1000, where m g,i is the vapor inlet flow rate and m a,i is the inlet noncondensable flow rate. Bundle Factor Method The bundle factor method is very similar to the vertical row-number correction except that the correct factor now contains some corrections for noncondensable gas and the saturation temperature depression. It is expressed as follows: h o,av = F to h o,Nus (10.93) where h o,Nus is again the condensing coefficient obtained with the single-tube Nusselt equation (10.12). The bundle factor F to is commonly obtained from field test results. Sklover and Grigor’ev (1975) and Sklover (1990) proposed a method to calculate F to that contains corrections for condensation inundation, noncondensable gases, and vapor-velocity effects (the saturation temperature depression). This method has not received wide acceptance. However, many heat exchanger manufacturers have devel- oped proprietary correlations for F to based on field experimental data. These corre- lations are then applied for particular applications with similar design conditions— similar inlet flow rates (vapor, noncondensables, and coolant) and temperature levels. Rabas (1992) suggested that the same F to obtained with a plain tube condenser could be used to evaluate the performance improvement obtained after retubing with en- hanced tubes. Pointwise or Numerical Computer Programs It is apparent that the two overall methods have difficulty capturing the impact of the bundle size and shape on the thermal performance. Pointwise or numerical methods were developed to account for these geometry effects and to incorporate the latest techniques to correct for condensate inundation, noncondensable gas accumulation or pockets, and saturation temperature depression. BOOKCOMP, Inc. — John Wiley & Sons / Page 773 / 2nd Proofs / Heat Transfer Handbook / Bejan FILM CONDENSATION ON TUBE BUNDLES 773 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 [773], (55) Lines: 1552 to 1561 ——— -0.603pt PgVar ——— Long Page PgEnds: T E X [773], (55) Figure 10.17 Tube bundle layout of a typical X-shell condenser. An accurate computer model of an X-shell condenser need not be three- dimensional. Figure 10.17 shows that this single pass X-shell condenser is divided into a number of bays by the vertical tube supports. The three-dimensional nature of the condenser can therefore be modeled by repeated one- or two-dimensional pre- dictions for each bay, with the inlet vapor flow being adjusted to satisfy the common overall pressure-drop constraint, the same pressure drop from the vapor inlet to the vent off-take. A one-dimensional model is a realistic representation when the flow is essentially down or up through the bundle with limited bundle shell-side leakage. For example, a one-dimensional approach is reasonable for the condenser cross section shown in Figure 10.17 but not for many large surface condenser cross sections. There are some situations where a one-dimensional approach is acceptable, that is, when the transverse tube pitch is much less than the longitudinal pitch, as shown in Figure 10.18. The two-dimensional flow field within each bay can be considered as a series of one-dimensional parallel paths with dividing streamlines corresponding to these rays of tubes extending from the bundle exterior to the central core. Other power plant condenser manufacturers employ tube-layout patterns that can only be analyzed with a two-dimensional numerical model for each bay. Computer codes were developed based on one-, two-, and three-dimensional mod- els. Barsness (1963), Kistler and Kassem (1981), and Rabas and Kassem (1985) de- veloped one-dimensional models. The model described by Kistler and Kassem was BOOKCOMP, Inc. — John Wiley & Sons / Page 774 / 2nd Proofs / Heat Transfer Handbook / Bejan 774 CONDENSATION 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 [774], (56) Lines: 1561 to 1567 ——— 0.15701pt PgVar ——— Normal Page PgEnds: T E X [774], (56) Figure 10.18 Tube pattern layout of a large power plant condenser. the basis of an early Heat Transfer Research, Inc. (HTRI) code called CST that was commercially available and is also applicable for other shell types. Later develop- ments are the two- and three-dimensional codes. Some recent publications that dis- cuss these codes are Zhang (1994, 1996), Zhang et al. (1993), Zhang and Zhang (1994), Frisina et al. (1990), and Pouzenc (1990). These and some other codes are proprietary but can be purchased from organizations such as Heat Transfer Research, Inc., and Heat Transfer and Fluid Flow Services (HTFS). In addition to optimizing the tube bundle layout at the design stage, these codes have been very useful in cor- recting some performance problems on older units (Beckett et al., 1983). Because of the proprietary nature, no further discussion is presented. Noncondensable Gas Management and Proper VentingTechniques The adverse effects of noncondensable gases and saturation temperature depression can be minimized with good condenser design practice, which includes proper management of the noncondensable gases and proper venting. Management of the Noncondensable Gas There are three causes of noncon- densable gas pockets in X-shell condensers: (1) radial flow maldistribution within a given bay, (2) longitudinal heat flux maldistribution for bay to bay, and (3) bottle- necking of the mixture in route from the bay exit to the vent. The pockets caused by the radial flow maldistribution can be located anywhere in the bundle cross section and in all the bays. The pockets caused by the longitudinal maldistribution are located around the air-cooling section in the cold-end bays of the condenser. Only the first two causes are discussed. The third is the result of poor design practice that can be corrected with straightforward fluid flow principles. pockets formed by radial flow maldistribution Noncondensable gas pockets formed by a radial flow imbalance are most common with large power plant surface condensers with radial vapor flow into the bundle. A good tube bundle layout should . the overall heat transfer coefficient using the standard summation of the separate heat transfer resistances. The major uncertainty in the calculation is the average, outside condensing heat transfer. convective condensation heat transfer. If roughness mixing dominates the enhancement mechanism, neither swirl effects nor surface tension drainage have much influence on the heat transfer. The lack. that laminar flow heat transfer for three different twist ratios are correlated well with the swirl parameter. Agrawal et al. (1986) also correlated R-12 swirl flow boiling heat transfer coefficients