Flooding During Reflux Condensation of Steam In

FIOOUNG DURING REFLUX CoNDENSATIoN oF STEAM IN AN INCLINED ELLIPTICAL TUBE

P.D. Schoenfeld* D.G. Krogert

Received September 1999; Final version December 1999

In this erperimental inuestigation the pressure drop is Subscripts measured between the headers located at the ends of an a alr in,clined air-cooled elliptical tube in which refl,ur conden- c cross-section sation of steam occurs. The cross-section of the 7 m long d diameter tube has a height of g7 mm (major aris) and a width fl, flooding of 16 mm (minor ads). Steam temperatures are in the f inlet range of 45o C to 65o C. The pressure drop can be pre- / liquid dicted accurately using the Z apke- I(rri g er pressure drop .s steam, superficial model applicable to refl,ur condensers. At a certain steam t tube fl,ou, rate a sudden sharp increase in the pressure drop u vapour occurs. This phenoTne?t,on is lcnowrt, &s fl,ooding. The Dimensionless groups rneasured aapour uelocities at fl,oodin,g agree well with FrH,o = pur?, I [Q, - p,) gU] Superfi cial densimetric the ualues predicted by the Zapke-Krciger fl,ooding corre- vapour Froude number lation. The erperimental results also show that flooding Frn,t = ptr?,/ lQ, - p,) gdnl Superficial densimetric has a detrimental effect on the thermal effectiueness of liquid Froude number Ihe elliptical tube. Rero = puurodHlp, Superficial vapour Reynolds number Nomenclature ZK (p,dro) I pt = Oh-r Dimensionless liquid property A Are" (-2) parameter cpo Specific heat of air (J/kgK) Introduction Hydraulic diameter (m) dp Forced draft air-cooled steam condensers are increasingly e Heat exchanger effectiveness used in power generating plants, especially in arid re- Gravitational acceleration G (-/r') gions. The condensers consist of bundles of finned tubes H Height of duct (rn) arranged in an A-frame configuration above the fans as iJ n Latent heat of vap orrzation (J/kS) shown in Figure 1(a). Dimensionless eonstant, I\ eqn (3) The inclination angle of the finned tubes, which gener- Mass flow m rate (kg/r) ally have a round, elliptical or flat-profile cross-sectional n, Dimensionless constant, €etr (4) geometry, is approximately 60o to the horizontal. In Heat transfer rate A (W) the finned tubes the steam and condensate flow concur- T Temperature ('C) rently downward into the drainage header. To prevent Velocity (m/s) u subcooling of the condensate and the formation of dead Dynamic viscosity (kg/ms) tt zones due to the accumulation of non-condensable gases p Densitv (kg/m3) in the condenser, a secondary reflux condenser or de- o Surface tension (N/-) phlegmator is connected in series with the main con- Inclination angle (degrees) 0 denser. The dephlegmator ensures that there is a net outflow of steam from the bottom of the main condenser. This steam condenses in a reflux mode in the dephlegmator, the steam flow being upwards countercurrent to * Department of Mechanical Engineering, University of Stellen- bosch, Private B.g X1, Matieland, T602 South Africa the condensate flow. Non-condensable gases that may t Department of Mechanical Engineering, University of have leaked into the system are removed at the top of Stellenbosch the dephlegmator by means of an ejector

R & D lournal, 2000, l6( I ) Since the dephleghmator operates in the reflux con- 60o. An air outlet manifold (9) is mounted on top of the densation mode, flooding of the dephlegmator can oc- wooden casing and is connected to a centrifugal fan (10) cur. At a so-called flooding vapour inlet velocity, the which draws cooling air across the finned tube. condensate does not drain freely into the bottom header During an experimental run the air inlet and out- of the dephlegmator, but starts to accumulate in the let temperatures as well as the steam temperatures in tubes. This results in a large pressure drop across the the inlet and outlet headers are measured using cop- dephlegmator headers as well as a decrease in the ther- per constantan thermocouples. Calibrated propeller- mal effectiveness of the dephlegmator. type anemometers (11), situated in each lateral of the Numerous studies have been conducted to investigate air outlet manifold, measure the air volume flow rate the pressure drop and flooding in reflux condensers. over the finned tube while a pressure transducer mea- Banerjee et al.,L Girard k Chang2 and Obinela et al.3 sures the differential pressure between the inlet and top studied reflux condensation of steam in a vertical water- headers of the tube. The steam condensation rate is de- cooled glass tube with the top and bottom header pres- termined using a measuring cylinder (3) that is attached sures held constant duritg an experimental run. Russella to the steam generator. made use of a long air-cooled finned tube inclined at 57o to the horizontal in which steam at atmospheric pressure Experimental results was condensed. The various flow modes were observed through a sight-glass. Reuter & Kroger5 conducted ex- The pressure drop and flooding in the elliptical finned periments in a vertical and inclined water-cooled glass tube were investigated using steam ranging in tempera- tube in which low pressure steam was condensed in the ture from 45oC to 650C. It was found that the header- reflux condensation mode. Bellstedt6 studied reflux con- to-header pressure drop can be predicted accurately us- densation of low pressure steam in an air-cooled finned irg a two-phase pressure drop model applicable to reflux tube inclined at 60o to the horizontal. Groenewald7 con- condensers as was proposed by Zapke k Kriiger.8'e Ac- ducted similar experiments in a flattened finned tube. cording to Zapke k Kriigere the pressure drop is both a In this study the pressure drop and flooding during function of the vapour Reynolds number and the densi- reflux condensation of low pressure steam is investigated metric vapour Froude number. At low vapour flow rates in an air-cooled elliptical finned tube inclined at 60o to the pressure drop is Reynolds number related. How- the horizontal. The experiments were conducted in a ever, &t high vapour flow rates greater liquid-vapour in- steam temperature range from 45o C to 650C. teraction takes place, resulting in wave formation on the surface of the liquid. The vapour Froude number there- Experirnental apparatus fore becomes the governing dimensionless group and the duct height the characteristic dimension. The pressure The experimental apparatus is schematically depicted drop in the elliptical tube as predicted by the Zapke- in Figure l(b). Hot water (2), electrically heated, is Kriiger pressure drop model is compared to experimen- pumped through a shell-and-tube heat exchanger (1) in tal header-to-header pressure drop data as a function of which low pressure steam is generated. The steam is the vapour superficial velocity at the tube entrance as ducted to the steam inlet header (a) which has radial in- shown in Figure 2. Note that it is more meaningful to let and guide vanes that ensure that the steam is vortex plot the pressure drop data in terms of the vapour ve- free on entering the test tube. The test tube (5) is an locity since the pressure drop is a function of both the elliptical finned tube ,7 m long and with an inside height vapour Reynolds number and Froude number. (major axis) and width (minor axis) of g7 mm and 16 As was mentioned above, at high vapour velocities the mm, respectively. The inlet of the tube is square-edged densimetric vapour Froude number becomes the govern- (90').ThecroSS-Sectionalareaofthetube,A"t ing dimensionless group. At a certain vapour velocity, mm2 and the hydraulic diameter, dn = 25.922 mm. A namely the flooding velocity, a sharp increase in the top header (6) is connected to the upper end of the tube. header-to-header pressure drop occurs. The fact that A water-driven vacuum pump (7), connected to the sys- flooding is governed by the vapour Froude number as tem at the outlet header, ir used to obtain subatmo- opposed to the vapour Reynolds number is clear when spheric pressures in the system prior to an experimental considering Figures 3 and 4. They are plots of the pres- run and to remove non-condensable gases that may col- sure drop in the tube versus the vapour Reynolds nurnber lect in the tube during a run. The entire finned tube is and densimetric vapour Froude number, respectively. In mounted in a wooden casing (8), which also forms part Figure 3, the flooding vapour Reynolds number varies of a support frame. The frame is hinged at its lower end, between approximately 13 000 and 20 000 depending on thus making it possible to adjust the inclination angle of the steam temperature. In terms of the Froude number, the tube. In this investigation the tube is at an angle of flooding occurs at approximately 0.47 irrespective of the

R & D lournal, 2000, 16( I ) fo ejector

o0da

Fons

Dephlegmotor

Condensote

Figure 1 (a) Condenser with dephlegmator

Elliptical nrbe

Cross-section A-A

Figure 1 (b) Schematic representation of the experimental apparatus

R & D lournal, 2000, 16( I ) 4000

3500

3000

25W d /tl L< \/ 2000 aE 1500

r000

500

0 40 50 vsv (*ts)

Figure 2 Header-to-header pressure drop in an elliptical reflux condenser tube with a square-edged inlet

4000

3 500

3000

2500 63 t-tn \/ 2000 lr

-a I 500

I 000

500

10000 I 5000 20000

Re'sv

Figure 3 Total pressure drop in the elliptical reflux condenser tube versus the superficial vapour Reynolds number

R & D lournal, 2000, 16(I) steam temperature, &s shown in Figure 4. The slight It should be noted that Fru* is based on the height of variation in the flooding vapour Froude numbers is due the duct while Frarr is based on the hydraulic diameter to the different condensate flow rates at the respective of the duct. steam temperatures at flooding. Note that in Figure 4 As mentioned previously, Reuter & Krogers conducted the flooding vapour velocities, expressed in terms of the reflux condensation experiments in a glass tube with an Froude numbers, have been indicated for the respective inside diameter of 30 mm. Their flooding data and the steam temperatures to clarify visually the definition of experimental flooding data obtained in this investiga- the flooding velocity. tion are plotted in the form of the superficial densimet- Using various liquids and gases, Zapkelo conducted ric vapour Froude number versus a product of the liquid adiabatic two-phase counterflow experiments in flattened densimetric Froude number and the Ohnesorge number tubes of different heights at inclination angles between as shown in Figure 5. 0o and 90o . The tubes had square-edged inlets (90'). Numerous researchers [Banerjee et al.,,r Girard k Flooding was defined as that condition when the gas flow Chang,2 Obinelo et a1.,3 Reuter U Kriigers] have sug- starts to carry the liquid up beyond the liquid feed point. gested that flooding determines maximum heat transfer At that condition a significant rise in pressure drop was in a tube in which all the steam condenses in the reflux observed. Zapke correlated the flooding data in terms mode. of the superficial vapour Froude number expressed as a Since the superficial vapour velocity at the tube en- function of the liquid Froude number and a dimensionless trance is representative of the heat transferred, it is use- number originally proposed by Zapke & Kriiger,8 i.e. ful to express the flooding data in terms of the flooding velocity. Upon rearranging eqn (6), the predicted flood- (p,dno) I pt - oh-r (1) ing velocity at the tube entrance can be expressed as which accounts for the liquid properties. Oh is the Ohne- sorge number. The correlation takes into account the u effect of tube geometry (height and hydraulic diameter) ry 0 .70367 lb, - p,) s Hlo and inclination angle, and is expressed as (7) where H is the duct height for an elliptical tube, or the FrHr, - I{yl exp (-nyr x Fr|;u, x Oh'r:) (2) diameter for a round tube. The right-hand side of eqn (7) is dependent on the fluid properties which are tem- where perature dependent. It is therefore possible to express the flooding vapour velocities in terms of the steam tem- Iiyt - 7.9143 x 10-2 + 4.9705 x 10-3 x 0 (3) peratures. This has been done graphically in Figure 6 where eqn (7) is compared to experimentally determined +1.5183 x 10-4 x 02 1.9852 x 10-6 x 03 - flooding vapour velocities for the elliptical tube studied and in this investigation as well as the 30 mm tube used by n3r -18.149-\.gd\x0 +6.T058 Reu\er &, l(roger.S (4) Under ideal conditions, i... floodittg does not occur x 10-2 x gz - 5.3227 x 10-4 x gz irrespective of the heat transfer, it is possible to calculate the ideal heat transfer rate of an air-cooled reflux with 0 the inclination angle in degrees. In a reflux con- condenser tube using the air mass flow rate and air inlet denser tube the flooding fluid velocities and properties temperature flowing over the tube as follows: are such that the expression in brackets on the right- hand side of eqn (2) is very small. Therefore, by mak- Qia".t - emocoo(7, -T"t) (8) ing use of the Taylor series expansion for an exponential The effectiveness e in eqn (S) is a function of the overall function and neglecting second and higher order terms, heat transfer coefficient [/, which in turn is a function of eqn (2) may be written approximately as the condensation heat transfer coefficient. In the case of heat transfer the condensation heat transfer coef- FrH,o x liyr (1 n1t x Fr|,u, x Ohoo:) Iiyr (5) 'ideal' - = ficient for film condensation is used to calculate U. Once At an angle of inclination of 0 = 600 this equation is flooding occurs the film condensation heat transfer coef- further simplified to ficient is not valid anymore and conditions are no longer 'ideal'. This is the reason for the deviation depicted in FrH,o r 0.49516 (1 - 27.7615 x Frl,6, x Oh'o:) Figure 7. (6) The ideal superficial vapour velocity at the tube en- ry 0.49516 trance corresponding to the ideal heat transfer rate is

R & D lournal, 2000, 16( I ) 3 500

3000

2500 //^ GI 2000 vO. o. 1 500

1000

500

0 I 0. I 0.2 0.3 04 05 Frrtu

Figure 4 Total pressure drop in the elliptical reflux condenser tube versus the superficial densimetric vapour Froude number

0.55 0.53

0.5 1 0.49 0.47 ca E tJr t 0.45 E- 0.43 0.41 0.39 0.37

0.3 5 0.0003 0.0006 0.0009 0.0012 0.0015 0.0018 F'f,f oh!,? "

Figure 5 Flooding superficial vapour Froude number at the tube entrance

R & D lournal, 2000, I6(l) 100

cn 60 (- - 50 tn 40

20 l0

0 55 Tri ("C)

Figure 6 Flooding vapour velocity at the tube entrance as a function of the steam inlet temperature

35 o Ideal heat transfer rate (45 "c) x Experimental data (45 "C) 30 o B Ideal heat transfer rate (65 "c) €i: + + €zs + $ g &20G Lr o io i e ;XX: ?ls E

:ro6 C) F- Fr. 5

0. 1 0.2 0.3 0.4 0.5 0.6 0.7

Frrw-ideal

Figure 7 ldeal and experimental heat transfer rates versus vapour Froude number corresponding to ideal heat transfer rate

R & D lournal, 2000, l6( I ) 7 then Journal of HeaI Mass Transfer, 1992,35, pp. 1)sa : -ideal Qia.dl Ut g Po A"t) (e) 2203-22t8. from which the vapour Froude number can be calculated. 3. Obinelo IF, Round GF k Chang JS. Condensation In Figure 7 the ideal heat transfer rate calculated assum- Enhancement by Steam Pulsation in a Reflux Con- ing that flooding does not occur is plotted versus the denser . International J ournal of H eat and Ftuid corresponding ideal vapour Froude number. To demon- FIow, 1994, 15, pp .20-29. strate the effect that flooding has on the thermal effectiveness of a reflux condenser, the measured heat transfer 4 Russell CMB. Condensation of Steam in a Long rate is also plotted versus the ideal Froude number. Reflux Tube. Heat Transfer and Fluid Flow Ser- Figure 7 clearly shows the effect that flooding has on uice, 1980, AER-E Harwell and National Engineer- the heat transferred by an air-cooled reflux condenser ing Laboratory. tube. When flooding occurs in the elliptical tube, the thermal effectiveness thereof decreases, resulting in the 5. Reuter H k Kroger DG. Pressure Change and actual heat transfer rate being lower than the heat trans- Flooding in Vertical and Inclined Tubes during fer rate under ideal conditions. Reflux Condensation of Steam. gth International Symposium on Transport Phenomena in Thermal- Conclusion Fluids Engineering, Singapore. ISTP-9, 1996, 2, pp.727 -732. The pressure drop in a reflux condenser tube can be predicted using the twephase flow pressure drop model ap- 6 Bellstedt M0. Condensation of Low-Pressure Steam plicable to reflux condensers that was proposed by Zapke in Inclined Air-Cooled Tubes. PhD thesis, 1991, k Kroger.e As is the case for adiabatic flow, the vapour University of Stellenbosch, South Africa. Froude number is the governing dimensionless group at 7 Groenewald W. Heat Transfer and Pressure flooding in reflux condensers. The flooding correlation Change in an Inclined Air-Cooled Flattened Tube during proposed by Zapkel0 for flooding during adiabatic coun- Condensation of Steam. MEng thesis, 1993, tercurrent flow can be simplified and applied with confi- uni- versity of Stellenbosch, South Africa. dence to determine the flooding velocity at the inlet to reflux condenser tubes or ducts wherein steam between 8. zapke A & Kroger DG. The Influence of Fluid Prop- 45oC and 65oC is condensed. It has also been shown that erties and Inlet Geometry on Flooding in Vertical a decrease in the thermal effectiveness of a dephlegmator and Inclined Tubes. International Journal of Mul- occurs at flooding. tiphase Flow, 1996 ,22, pp .46I-472. References I Zapke A k Kroger DG. Vapor-Condensa,te Interac- tions during Counterflow in Inclined Reflux Con- 1. Banerjee S, Chang JS, Girard R & Krishnan VS. Re- densers, HTD-Vo1.342. I,{ ational H eat Transf er flux Condensation and Transition to Natural Circu- Conference, L997, 4, ASMB. lation in a Vertical U-Tube . Journal of Heat Trans- 10 zapke A. The Characteristics of frr, 1983, 105, pp .717 -727 . Gas-Liquid Coun- terflow in Inclined Ducts with particular reference 2 to Reflux Condensers. PhD thesis, 1997, University ffi:l J*urf,i?l'":i;.1iH1 3;" ff i:':;,::;- of Stellenbosch, South Afrrca.

R & D lournal, 2000, I6( l ) 8