BOOKCOMP, Inc. — John Wiley & Sons / Page 1047 / 2nd Proofs / Heat Transfer Handbook / Bejan TREATED SURFACES 1047 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 [1047], (19) Lines: 425 to 434 ——— 5.927pt PgVar ——— Normal Page PgEnds: T E X [1047], (19) potential startup problems (Thome, 1990; Webb, 1994; Bergles, 1998). Furthermore, there are limited data which indicate that the tube bundle performance in a ther- mosyphon reboiler may be different from that of a single tube (Yilmaz et al., 1981). A later study (Jensen et al., 1992), however, suggests that single-tube results may be used to predict tube bundle behavior. Several attempts have been made to model the nucleate boiling mechanisms on structured surfaces and to develop predictive correlations. These analyses have tried to scale and correlate the effects of geometrical properties of the structured surface, or 0 4 8 121620 2 2 5 4 4 3 3 6 7 8 9 10 1 ⌬T sat [K] h [kW/m . K] 2 Teflon in pits Teflon spots on surface Smooth surface Pitted surface Water ( = 1 atm) boiling on stainless steel p Figure 14.8 Enhanced boiling heat transfer characteristics of water at 1 atm from a stainless steel surface coated with Teflon spots. (From Young and Hummel, 1965.) BOOKCOMP, Inc. — John Wiley & Sons / Page 1048 / 2nd Proofs / Heat Transfer Handbook / Bejan 1048 HEAT TRANSFER ENHANCEMENT 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 [1048], (20) Lines: 434 to 443 ——— 0.927pt PgVar ——— Normal Page PgEnds: T E X [1048], (20) the characteristics of nucleating cavities and liquid-replenishing intercavity channels, on the heat transfer. Extended discussions of several different models can be found in Nakayama et al. (1980a,b), Webb (1994), Chien and Webb (1998), Yabe et al. (1999), and Kim and Choi (2001), among others. For less wetting or relatively higher-surface-tension fluids, coatings of nonwetting material (e.g., Teflon) on either the heated surface or its pits and cavities have been found to improve stable nucleation and reduce the required wall superheat (Griffith and Wallis, 1960; Young and Hummel, 1965; Gaertner, 1967; Vachon et al., 1969). Young and Hummel (1965) sprayed a smooth as well as a “pitted” stainless steel sur- face with Teflon to create spots of the no-wetting material on the heated surface and in the pits. This was found to promote nucleate boiling in water, with relatively low wall superheat and three- to four-times-higher heat transfer coefficients, as shown in Fig. 14.8. In a more recent study of boiling of alcohols (methanol, ethanol, and isopropanol) at atmospheric and subatmospheric pressures on a horizontal brass tube coated with polytetrafluoroethylene, Vijaya Vittala et al. (2001) have reported a sig- nificant enhancement in the heat transfer. A typical set of their data for ethanol is presented in Fig. 14.9. Surfaces coated with thin films of low-thermal-conductivity materials have also been shown to enhance pool boiling heat transfer under heater surface temperature-controlled experimental conditions (Zhukov et al., 1975). This 65789234 10 5 10 4 2 5 4 3 6 7 8 9 10 4 10 3 qЉ [W/m ] 2 PTFE coated tube, = 97.60 kN/mp 2 Plain tube, = 97.60 kN/mp 2 PTFE coated tube, = 37.67 kN/mp 2 Plain tube, = 37.67 kN/mp 2 Experimental data (Vijaya Vittala et al., 2001) h [W/m K] 2 Figure 14.9 Enhancement in the boiling heat transfer coefficient for ethanol at different subatmospheric pressures and heat fluxes from PTFE-coated tubes. BOOKCOMP, Inc. — John Wiley & Sons / Page 1049 / 2nd Proofs / Heat Transfer Handbook / Bejan TREATED SURFACES 1049 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 [1049], (21) Lines: 443 to 458 ——— 0.927pt PgVar ——— Normal Page PgEnds: T E X [1049], (21) essentially breaks up film boiling by reducing the fluid-surface interface temperature to promote a more effective transition or nucleate boiling. This has also been found to be the case with scale or oxide coatings, which destabilize film boiling and signif- icantly reduce the quench times of a heated surface (Bergles and Thompson, 1970). 14.2.2 Condensing Vapor space condensation heat transfer can be enhanced primarily by treated surfaces that promote dropwise condensation. The intent here is to prevent surface wetting and break up the condensate film into droplets, as depicted in Fig. 14.10, which lends to better drainage and more effective vapor renewal at the cold heat transfer interface. This technique had been found to enhance the heat transfer by factors of 10 to 100 in comparison with that in filmwise condensation (Bergles, 1998). Nonwetting coatings of several different materials have been used effectively in different investigations (Hannemann, 1977; Tanasawa, 1978; Griffith, 1985). This includes coating the condensing surface with a thin film of an inorganic compound, a thin-film plating of a noble metal, or a film of an organic polymer (Rose et al., 1999; Das et al., 2000). Of these, organic coatings have been used considerably in steam systems, and Marto et al. (1986) and Holden et al. (1987) have given a comprehensive evaluation of their performance in promoting dropwise condensation. However, this technique is useful in promoting dropwise condensation in steam (or other relatively high-surface-tension fluid) condensers, as there seem to be no nonwetting (or “Freon- phobic”) substances that would induce dropwise condensation in refrigerants and other low-surface-tension fluids (Iltscheff, 1971). Glicksman et al. (1973) have shown that by strategically placing strips of Teflon or other nonwetting material in a helical or axial arrangement around the circum- ference of horizontal tubes, the average condensation heat transfer coefficients of steam on horizontal tubes can be improved by 20 to 50%. In such an arrangement, the condensate flow is interrupted near the leading edge of the nonwetting strip, and Figure 14.10 Examples of dropwise condensation of steam: (a) on a plain surface (from Hampson and Ozisik, 1952); (b) on a SAM-coated corrugated tube (from Das et al., 2000). BOOKCOMP, Inc. — John Wiley & Sons / Page 1050 / 2nd Proofs / Heat Transfer Handbook / Bejan 1050 HEAT TRANSFER ENHANCEMENT 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 [1050], (22) Lines: 458 to 468 ——— 5.7pt PgVar ——— Normal Page PgEnds: T E X [1050], (22) the condensate film that reforms at its downstream edge is much thinner. Cary and Mikic (1973) suggest further that Marangoni convection (surface tension–gradient or interfacial temperature–gradient, driven flow) at the liquid–vapor–strip interface may also be a contributing factor inenhancingtheheattransfer. There are, however, several practical issues relating to the application and surface integrity of nonwetting material coating, as discussed by Tanasawa (1978), Marto et al. (1986), and Das et al. (2000). There have been a few attempts to develop theoretical models for dropwise con- densation as well. In this, the studies reported by Le Fevre and Rose (1966), Rose (1988), and Tanaka (1975a,b) are particularly noteworthy, and their brief assessment is provided by Rose et al. (1999). The application of a hydrophobic coating of self- assembled monolayers, formed by a chemisorption of alkylthiols on metallic surfaces, to promote dropwise condensation has been investigated by Das et al. (2000). Experi- ments on steam condensation on coated corrugated tubes with gold and copper–nickel alloy surfaces under atmospheric (101 kPa) and subatmospheric (10 kPa) pressure conditions with wall subcooling of about 16°C and 6°C, respectively, showed that condensation heat transfer coefficients increased by factors of 2.3 to 3.6 compared to those for uncoated tubes. The characteristic condensation behavior is illustrated photographically in Fig. 14.10b. Although technically, this study would fall under the compound enhancement category, the noteworthy and novel technique to effect dropwise condensation is particularly relevant here. In a different scheme to enhance film condensation, Notaro (1979) has described a patent for a tube surface that is intermittently coated with small-diameter metal particles. The objective of this scheme is that condensation would occur over the strategically coated surface, and liquid film would then drain along the uncoated, bare tube surface. With a typical configuration (a 6-m vertical tube with 0.5-mm- diameter particle coatings interspaced over 50% of the tube surface), up to 17 times higher steam condensation coefficients than that in normal film condensation have been reported (Notaro, 1979; Webb, 1984). 14.3 ROUGH SURFACES 14.3.1 Single-Phase Flow One of the earliest and perhaps simplest and yet highly effective techniques is the use of surface roughness in turbulent single-phase flows (Nikuradse, 1933; Dippery and Sabersky, 1963; Webb et al., 1971); in laminar flows, small-scale surface rough- ness tends to have little effect. It essentially disturbs the viscous sublayer in the near- wall turbulent flow structure to promote higher momentum and heat transport (Niku- radse, 1933). Much of the early work focused on “naturally” occurring roughness in commercial tubes. However, as pointed out by Bergles (1998), because such natural roughness is not well defined, artificial or structured roughness is now commonly employed in most applications. Structured roughness can be integral to the surface, or the protuberances can be introduced in the form of wire-coil-type inserts. The former can be produced by machining (e.g., knurling, threading, grooving), forming, BOOKCOMP, Inc. — John Wiley & Sons / Page 1051 / 2nd Proofs / Heat Transfer Handbook / Bejan ROUGH SURFACES 1051 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 [1051], (23) Lines: 468 to 480 ——— 2.95601pt PgVar ——— Normal Page PgEnds: T E X [1051], (23) casting, or welding, and the resulting surface proturberances or grooves can be two- dimensional or discrete three-dimensional in their geometrical arrangement (Webb, 1994; Bergles, 1998). As a result, almost an infinite number of geometric variations are possible for structured roughness, and this is also reflected in the more than 700 studies, and still counting (Bergles et al., 1995; Bhatnagar and Manglik, 2002) that have been reported in the literature. Rough surfaces have been employed to enhance heat transfer in single-phase flows both inside tubes and outside tubes, rods, and tube bundles. A collection of studies and patents that deal with their thermal–hydraulic performance as well as fabrica- tion and manufacturing methods is given by Webb et al. (1983), Ravigururajan and Bergles (1986), Webb (1994), Bergles et al. (1995), and Bergles (1998). Some of the two-dimensional structured roughness used inside tubes includes repeated or trans- verse ribs, helical ribs, and wire coil inserts, in a variety of profiles, as illustrated in Fig. 14.11. Of these, tubes with grooves are often referred to as corrugated, roped, indented, fluted, or convoluted tubes, among other names, and are illustrated in Fig. 14.12a. They provide an external rough surface as well and have been used in double- pipe and shell-and-tube bundles to enhance annulus- or shell-side heat transfer. Two representative tubes with three-dimensional roughness, composed of serrated ribs, dimples, and cross-rifled surfaces, are also depicted in Fig. 14.12b and c. Besides these, knurling (to form diamond knurls), discrete slotting, rolling, and particle depo- sition have been investigated as possible candidates for three-dimensional roughness (Durant et al., 1965; Groehn and Scholz, 1976; Achenbach, 1977; Fenner and Ragi, 1979; Menze et al., 1994; Webb, 1994). Over the last five decades, a fairly extensive set of experimental data for heat transfer coefficients and friction factors in rough tubes has been obtained (Ravigu- rurajan and Bergles, 1986; Esen et al., 1994; Bergles, 1998). Several attempts have also been made to develop predictive equations for the turbulent flow regime. One of the first such efforts was the momentum heat transfer analogy solution proposed by Dippery and Sabersky (1963) for sand-grain roughness. Subsequently, Webb et al. (1971, 1972) devised another analogy-based correlation for heat transfer coefficients for different fluids (0.7 ≤ Pr ≤ 38) and tubes with transverse repeated-rib roughness (α = 90°, 0.01 ≤ e/d ≤ 0.04, 10 ≤ p/e ≤ 15; Fig. 14.11). Withers (1980a,b), sim- ilarly, extended this correlating technique to commercial single- and multiple-helix internally ridged tubes. A detailed discussion of these solutions and their prediction efficacy is given by Webb (1994). In a more recent experimental study on turbulent flows of water and oil in spirally corrugated tubes, Dong et al. (2001) have developed a new set of analogy-based friction factor and Nusselt number correlations. Adopting an empirical approach, combined with a statistical analysis of a fairly large database for heat transfer coefficients and friction factors for the various types of roughness shown in Fig. 14.11, Ravigururajan and Bergles (1996) have proposed the following Nusselt number and Fanning friction factor correlations: Nu = Nu o 1 + 2.64Re 0.036 e d 0.212 p d −0.21 α 90 0.29 · Pr −0.024 7 1/7 (14.10) BOOKCOMP, Inc. — John Wiley & Sons / Page 1052 / 2nd Proofs / Heat Transfer Handbook / Bejan 1052 HEAT TRANSFER ENHANCEMENT 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 [1052], (24) Lines: 480 to 502 ——— 3.29807pt PgVar ——— Normal Page PgEnds: T E X [1052], (24) f = f o 1 + 29.1Re a1 e d a2 p d a3 α 90 a4 1 + 2.94 sin β n 15/16 16/15 (14.11a) where a1 = 0.67 − 0.06 p d − 0.49 α 90 a2 = 1.37 − 0.157 p d a3 =−1.66 × 10 −6 Re − 0.33 α 90 a4 = 4.59 + 4.11 × 10 −6 Re − 0.15 p d (14.11b) Figure 14.11 Different type of structured roughness and their profile shapes considered by Ravigururajan and Bergles (1986, 1996). BOOKCOMP, Inc. — John Wiley & Sons / Page 1053 / 2nd Proofs / Heat Transfer Handbook / Bejan ROUGH SURFACES 1053 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 [1053], (25) Lines: 502 to 517 ——— 0.12354pt PgVar ——— Normal Page PgEnds: T E X [1053], (25) and the respective smooth tube Nu o and f o performances are given by Nu o = Re · Pr(f o /2) 1 + 12.7 √ f o /2(Pr 2/3 − 1) (14.12) f o = (1.58 ln Re − 3.28) −2 (14.13) The prediction reliability and accuracy of eqs. (14.10) and (14.11) have been shown to be very good compared with more than 1800 experimental data points, representing the full range of roughness types and profiles depicted in Fig. 14.11 (Ravigurura- jan and Bergles, 1986; Bergles, 1998), and some typical predictions are graphed in Fig. 14.2b. A novel technique to provide variable-roughness “on demand” inside tubes has recently been proposed and tested by Champagne and Bergles (2001). The concept involves using a wire-coil insert made of a shape-memory alloy (SMA) that alters its geometry in response to changes in temperature (Bergles and Champagne, 1999). With a fixed roughness height (e/d), the wire-coil insert changes from a compressed shape, which occupies a small fraction of the tube length, to an expanded shape that has the desired or “trained” roughness pitch (p/d) and helix pitch (α/90°) upon being heated. In essence, the SMA insert can be designed to provide the required roughness to meet “active” changes in process conditions, which are usually reflected by the variations in the tube wall temperature and enhancement objectives. A schematic rendering of this concept is given in Fig. 14.13. In a proof-of-concept experimental Figure 14.12 Tubes with two- and three-dimensional structured roughness: (a) corrugated tubes (courtesy of Wolverine Tube, Inc.); (b) serrated ribs on outer surface and staggered dimples on inner surface (courtesy of Sumitomo Light Metal Industries); and (c) Cross-rifled inner surface (from Nakamura and Tanaka, 1973). BOOKCOMP, Inc. — John Wiley & Sons / Page 1054 / 2nd Proofs / Heat Transfer Handbook / Bejan 1054 HEAT TRANSFER ENHANCEMENT 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 [1054], (26) Lines: 517 to 527 ——— 1.49751pt PgVar ——— Normal Page PgEnds: T E X [1054], (26) test, Champagne and Bergles (2001) have further shown that depending on the flow rate, a typical SMA (NiTi) wire coil insert can produce 30 to 64% increases in the heat transfer coefficients in single-phase turbulent flows. Flows in annuli with an inner rough surface, commonly encountered in a double- pipe heat exchanger with an inner tube that is convoluted or corrugated, or has some other structured roughness on its outer surface, have also been investigated in the literature (Kemeny and Cyphers, 1961; Brauer, 1961; Dalle Donne and Meyer, 1977; Hudina, 1979; Garimella and Christensen, 1995a,b; Salim et al., 1999; Kang and Christensen, 2000). It an early study with helical grooves and helical protuberances, the experimental data of Kemeny and Cyphers (1961) suggest enhancement factors of 1.25 to 2.0 for different flow rates and roughness (Bergles, 1998). Surface protrusions had a relatively higher performance than grooves. Some of the more recent work with water and propylene glycol flows in annuli with fluted or corrugated tubes (Garimella and Christensen, 1995a,b; Salim et al., 1999; Kang and Christensen, 2000) indicate up to four times higher Nusselt numbers, with up to 10 times higher friction factors in the turbulent flow regime. Two different sets of correlations have also been devised by Garimella and Christensen (1995a,b), and Salim et al. (1999) that are based on their own respective data. No attempt appears to have been made to consolidate and compare the various data sets available in the literature. For high-viscosity fluids (e.g., propylene glycol) and/or large temperature-dependent viscosity variations in the flow field, Kang and Christensen (2000) suggest incorporating a Sieder and Tate (1936) type of viscosity-ratio correction factor in the predictive equation. The heat transfer enhancement in annuli with three-dimensional diamond-knurls type of roughness on the inner heated tube’s outer surface has been investigated by Durant et al. (1965). At fixed pumping power conditions, the heat transfer coeffi- cients for the knurled annuli were found to be up to 75% higher than those for the equivalent smooth annuli. Dalle Donne (1978) provides turbulent flow data for three- dimensional roughness made up of rows of staggered studs. The Stanton numbers were found to increase by factors of 3 and 4; the associated friction factors were 8 to 12 times higher. For their application in gas-cooled reactors, the grouping St 3 /f is often used as a figure of merit to interpret the performance data. This translates into enhancement ratios of up to 2.3 for the three-dimensional roughness. Similarly, annuli data have been “transformed” for application to the rod-bundle geometry (Hall, 1962; Lewis, 1974; Hudina, 1979; Webb, 1994). Some work has also been reported Figure 14.13 Conceptual representation of “roughness on demand” provided by a shape- memory alloy wire coil insert. (From Champagne and Bergles, 2001.) BOOKCOMP, Inc. — John Wiley & Sons / Page 1055 / 2nd Proofs / Heat Transfer Handbook / Bejan ROUGH SURFACES 1055 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 [1055], (27) Lines: 527 to 540 ——— 0.0pt PgVar ——— Normal Page * PgEnds: Eject [1055], (27) on crossflow over tube bundles that have roughness on the outer surface, particularly for gas-cooled reactors and shell-and-tube heat exchangers (Bergles, 1998). Much of this work is based on the extended experimental investigations with cross flows over a single rough cylinder. The case of pyramid-shaped roughness elements has been considered by Achenbach (1977) and Zhukauskas et al. (1978), with air and water flows, respectively, and up to 150% increase in the Nusselt number has been reported. Structured roughness, usually in the form of ribs, in both two-dimensional and discrete three-dimensional arrangements, has also been applied to plate-type and rect- angular channels (Han, 1988; Han et al., 1991; Liou and Hwang, 1992; Prakash and Zerkle, 1995; Ekkad and Han, 1997; Olsson and Sund ´ en, 1998; Sund ´ en, 1999; Gao and Sund ´ en, 2001). Much of this work is directed toward compact heat exchang- ers and gas turbine cooling and regeneration systems. A variety of arrangements for rectangular ribs attached to two opposite heated walls of rectangular (or plate) channels have been considered (Han et al., 1991; Zhang et al., 1994; Olsson and Sund ´ en, 1998). With ribs angled at 60°, some examples of the geometrical schemes investigated experimentally by Olsson and Sund ´ en (1998) and Sund ´ en (1999) are depicted in Fig. 14.14. Their relative enhanced performance, as represented by the volume goodness factor or heat transfer coefficient versus pumping power expended per unit heat transfer area [the development of this figure of merit is given in Kays and London (1984) and Shah and London (1978)], is also presented in this figure. Of the five configurations shown here, the last one, with short rib segments, essentially constitutes three-dimensional structured roughness. An interesting variation in the form of wedge- and delta-shaped ribs, which tend to produce longitudinal vortices in the near-wall region of the flow field, have been suggested by Han et al. (1993). Finally, as noted by Bergles (1998), surface roughness is usually not considered in applications with free or natural convection. In this thermal-fluid-transport regime, the buoyancy-induced fluid flow velocities are generally very low and are not easily disrupted by small-scale surface roughness so as to produce flow separation and recirculation (the primary mechanisms for roughness-induced enhancement). Bergles et al. (1979) reviewed the limited data available in the literature up to that time, which are for air, water, and oil with machined or formed roughness on the surface exposed to free convection. Where up to 100% increases in heat transfer have been reported in an air system, it is questionable if radiation effects have been accounted for adequately; in liquids, on the other hand, the enhancement has been found to be considerably small. 14.3.2 Boiling The application of rough-structured surfaces in pool boiling situations has been clas- sified under treated surfaces (Bergles, 1998), and their performance was discussed earlier. Their use in forced convection or flow boiling is primarily addressed here. The earliest works on the effects of surface roughness in forced-convection boiling were the investigation of improvements in subcooled CHF by knurling or threading the heated surface (Durant and Mishak, 1959; Durant et al., 1965). When compared BOOKCOMP, Inc. — John Wiley & Sons / Page 1056 / 2nd Proofs / Heat Transfer Handbook / Bejan 1056 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 [1056], (28) Lines: 540 to 549 ——— * 528.0pt PgVar ——— Normal Page * PgEnds: PageBreak [1056], (28) Figure 14.14 Enhanced performance of rectangular channels with structured roughness reported by Sund ´ en (1999): (a) different rib orientations; (b) variation of heat transfer coefficient with specific pumping power (volume goodness factor performance). . Hummel, 1965.) BOOKCOMP, Inc. — John Wiley & Sons / Page 1048 / 2nd Proofs / Heat Transfer Handbook / Bejan 1048 HEAT TRANSFER ENHANCEMENT 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 [1048],. boiling heat transfer coefficient for ethanol at different subatmospheric pressures and heat fluxes from PTFE-coated tubes. BOOKCOMP, Inc. — John Wiley & Sons / Page 1049 / 2nd Proofs / Heat Transfer. al., 2000). BOOKCOMP, Inc. — John Wiley & Sons / Page 1050 / 2nd Proofs / Heat Transfer Handbook / Bejan 1050 HEAT TRANSFER ENHANCEMENT 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 [1050],