Indian Journal of Chemical TechnologyVol. 4, March 1997, pp. 66-72

Heat transfer in boiling of mixtures in a thermo siphon reboiler

Mohd KamilInstitute of Petroleum Studies and Chemical Engineering, AJigarh Muslim University, AJigarh 202 002, India

Received 14 February 1996; accepted 2 September 1996

Boiling heat tr~sfer da.ta for water, methanol and their seven binary mixtures are presented. Theheat transfer section consisted of an electrically heated stainless steel tube of 25.56 mm i.d. and1900 mm len~ '. The heat flux values ranged from 4.1 to 4.30 kW/m2• The liquid submergence le-vels were ma.mt~I.ned at 100,. 75: 50 and 30 per cent. All the data were generated at atmosphericpressure. A significant reduction In the heat transfer coefficient is observed for binary mixtures attri-butable to the effect of mass diffusion on heat transfer. Possible reasons for the low heat transferrates in mixtures are suggested in relation to bubble growth and nucleation.

Vertical thermosiphon reboilers with boiling ofmixtures on the tube side are common processequipment in the petroleum and chemical indus-tries. The process fluid entering the tubes of heatexchanger gets heated and moves upwards due todensity difference between the liquid and boilingtwo phase fluid. The boiling of liquids in verticaltubes of closed loop circulation systems as above,finds its application also extending to refrigera-tion, power plants and nuclear reactors.

The nucleate boiling heat transfer coefficient inbinary or multicomponent liquid mixfures isstrongly influenced by the composition of themixture':". The heat transfer coefficient werefound to be lower than that from the pure com-ponent liquids. Similar observations were alsomade in flow boiling of liquid mixtures"!", Withnatural circulation boiling of liquid mixtures, asencountered in thermosiphon reboiler, therefore,the composition becomes an additional controll-ing parameter.

The available data on boiling of liquid mixturesgive insufficient and inacurate information aboutthe local heat transfer coefficient and circulationrate under the influence of governing parameters.In spite of the industrial importance of the designof reboilers for distillation colullUls, the studies onthermo siphon boiling of liquid mixtures in verticaltubes have been reported only by a few workers!"!".

Therefore, the present experimental study, wasundertaken to obtain such data for natural circu-lation boiling of binary mixtures and to study theeffect of heat flux and submergence on heat trans-fer to boiling binary mixtures in a single verticaltube thermosiphon reboiler.

Experimental ProcedureThe main unit was a U'-shaped circulation loop

made of two long vertical tubes and one shorthorizontal tube. The experimental reboiler andcooling system used in this study was described indetail elsewhere 11,14. In order to monitor the heattransfer surface temperatures along the tubelength, twenty one copper-constantan thermocou-ples were spot welded on the outer surface of thetube at intervals of 50 mm upto a length of 200mm from the bottom end and 100 mm over theremaining length. A copper-constantan thermo-couple probe was placed in the viewport to mea-sure the inlet liquid temperature. The temperatureof the boiling liquid before entry to the vapour-li-quid separator was measured by another travers-ing thermocouple probe. The entire set up wasthoroughly lagged with asbestos rope and glasswool and finally covered with a thin aluminiumsheet to reduce the heat losses, which were lessthan ± 2.5 per cent.

For conducting a series of runs, the test liquid.was boiled off for about 8-10 h to remove the lasttraces of dissolved air which was indicated by thedisappearance of the air bubbles in the bubbler.Then a desired heat flux was impressed and cool-ing water flow rate to the condenser was regulat-ed to give a maximum temperature rise consistentwith no loss of vapour due to inadequate conden-sation. The liquid level in the down-flow pipe wasadjusted by adding/draining the requisite amountof test liquid. When steady state conditions wereestablished, the readings of the ammeter, voltme-ter, thermocouples and rotameters were recorded.The liquid level in the downflow pipe was ob-

KAMIL: HEAT TRANSFER IN BOIUNG OF MIXTURES

Table I-Range of experimental parametersSystem q x 10- 4, S, % !1 T,ub' °C Concentration

W/m? of more volatilecomponent, wt, %

Distilled 0.57-4.3 30,50, 0.5-4.6water 75, 100Methanol 0.41-2.1 30,50 1.0-3.7

75,100Methanol- 0.57-2.9 30,50 0.5-4.6 5,10,18,26,30,Water 75,100 38,58

served. The maximum liquid head used in thepresent study corresponded to the liquid levelequal to the top end of the .reboiler tube. Thiscondition has been termed as 100% submergence.The experimental data were generated for fourdifferent levels of liquid submergence and variousheat fluxes. The liquid samples from the inlet ofthe reboiler tube, the .vapour-liquid separator andthe condensed .vapour were withdrawn with thehelp of a specially designed double walled sampl-ing device made of glass through drain cocks C

"C3 and C4, respectively. The temperature betweenthe two walls of sampling device was maintainedat OcC by placing melting ice in it. This arrange-ment proved quite effective for instant and com-plete cooling of the sample and thus avoiding thelosses of vapour. After cooling the samples, theinner tube was taken out and kept in a refrigera-tor. The refractive index of the samples were de-termined and the composition read from the re-fractive index composition chart of the system un-der study. -A similar procedure was followed forvarious compositions of binary liquid mixture.

Distilled water and methanol were used to pre-pare binary mixtures of methanol-water of re-quired compositions. The composition of binarymixtures were judiciously selected based on thebehaviour of (y- x) versus x plot of the system.The operating parameters with each binary liquidmixture were heat flux, submergence and inlet li-quid subcooling. The mixture concentrations andrange of other experimental parameters studiedare tabulated in Table 1.

Based on thermal equilibrium mode" the circu-lation rates and liquid bulk temperature distribu-tion in the thermo siphon reboiler have been de-termined by making a heat balance on the testsection. In order to determine the liquid circula-tion rate, it was required to know the length of ef-fective non-boiling or sensible heating region overwhich the liquid temperature varied linearly. Theeffective boiling and non-boiling zones over theentire heated length Were determined from theamount of net vapour generation. This could be

5 - 100-Q 1M•. - r--

XI Xoe: 122811 II 10·17 IO·I~1.1. 16'99 ,. 0·17 1o~629 21305 ~7 0168 O·I6C'V 25241 1'2 o~a 0.159

94 Jl)---------,--.--.--

86~--~----~---L----~--~--~Q..4 ~ \.Z 1.6 W

Z.ft!

67

Fig. I-Variation of wall temperature along the tube lengthwith heat flux as parameter for methanol-water system

116

11;>

loa

104

u 100° ,~•..

96

92

OB,$- --

8a/

/

/

84

-- --$------Q - 16599-

fs <'>T,." , Xi x,~ 100 25 0·300 0.2964 '75 1·8 0-289 0·269It 150 1·2 0·28 0228

90~ __ _L ~ __ ~ ~ __ ~ -Lo 04 0.8 1.2 1.6

Z,m2.0

Fig. 2-Variation of wall temperature along the tube lengthwith submergence as parameter for methanol-water system

68 INDIAN 1. CHEM. TECHNOL., MARCH 1997

obtained by the vapour condensed in the conden-ser. The details have been discussed earlier!"!".

The vapour fraction of two-phase mixture atthe exit of test section is:

... (1)

The local heat transfer coefficients in boiling aswell as non-boiling sections were calculated by di-viding wall heat flux with local values of tempera-ture difference between the wall and liquid as ex-pressed by equation:

... (2)

where the temperature drop, ~ T; between thethermocouple bead and the inside surface was es-timated using the equation of conductive heattransfer with internal heat generation for cylindri-cal wall.

The length-mean values of heat transfer coeffi-cient in the boiling region were calculated as giv-en below:

... (3)

Results and DiscussionFigs 1 and 2 shows the typical variation of wall

temperature with heat flux and liquid submerg-ence as parameters for methanol-water system.The wall temperature, T; rises at a fast rate withZ from its inlet (lower) end up to a point beyondwhich a steep fall sets in followed by a gradualdecrease, over the remaining portion of the heat-ed tube. However, the curves at higher heat fluxvalues get shifted to higher wall temperatures.The typical behaviour, as observed above, re-maining the same at other concentrations of liquidmixtures, the values of wall temperatures, 'locationof peak values and the lengths of various zonesare different. The location of wall temperaturepeaks get shifted towards the tube inlet and thecurves move to lower values of Twas the liquidsubmergence is reduced from 100 to 30 per cent.At the submergence of 100 and 75 per cent, allthe regimes of heat transfer are present while atlower submergences of 50 and 30 per cent the in-itial portions of curves before peak are almost ab-sent. The liquid temperature increases linearlyalong the tube length as observed for pure liquids,upto a point where it attains the saturation value(bubble point). The liquid temperature remainssaturated with its temperature increasing at a verylow rate. This may be attributed to the vapouriza-tion of liquid accompanied with the decrease inthe concentration of low boiling component as theliquid moves upwards. The bubble point thus in-

120

112

84

76~0----O~!4----0~8----~I~----~16----J;-0---~

l,m

Fig. 3-Variation of wall temperature along the tube lengthwith concentration as parameter for methanol-water system

creases along the tube length inspite of the reduc-tion in hydrostatic head.

The typical variation of wall and liquid temper-atures as observed indicates that there exist diffe-rent regimes of heat transfer in a reboiler tube.The linear rise in the temperature of liquid as itmoves upwards through the tube results from sen-sible heating under uniform heat flow. Since thedegree of subcooling in the present study is small,the wall temperatures attain values above satura-tion right in the close viscinity of tube inlet. Whenthe minimum wall superheat required is attained,the bubbles start nucleating at the surface but col-lapse there due to the presence of subcooled li-quid core. The onset of subcooled boiling thuscreates additional turbulence at the surface. Thisexplains why the linearly increasing wall tempera-ture corresponding. to convective heat transfer,starts varying at decreasing rate eventually be-coming zero at peak values. Once the bulk liquidtemperature attains saturation value, the bubblesgenerated at the surface grow to their maximumsize and get detached resulting in the existence ofvapour phase in the tube. All the heat supplied

3600

3200

2800

2400

2000.u..•

E 1600

~,s:

1200

800

KAMIL: HEAT TRANSFER IN BOILING OF MIXTURES 69

400

o ~ __ ~ ~ ~ ~ ~ __ ~o 0.8 ,l.O1.2

Z,mFig. 4- Variation of heat transfer coefficient along the tubelength with heat flux as parameter for methanol-water system

gets absorbed as latent heat of vaporization con-verting the liquid to vapour. The two phases flowupwards through the tube with increasing quantityof vapour and hence changing flow patterns. Thiscorresponds to saturated boiling regime as exhi-bited by the slowly decreasing wall and liquidtemperature profiles. As the value of heat flux israised, the wall temperature also increases in or-der to provide adequate temperature differencefor transferring the additional heat. In convectivemode, this should be almost in the same ratio asthat of heat flux change. But in nucleate boiling itis not so because the increased heat flux enableslarger number of nuclei for bubble generation be-coming active and thus enhancing the heat trans-fer coefficient and requiring small temperaturedifference.

The effect of liquid concentration on wall tem-perature distribution has been shown in Fig. 3. Ata given heat flux the nature of variation is similarfor all the concentrations of a system. However,the curves get shifted to lower wall temperatureswith increase in concentration for saturated boil-ing as well as initial non-boiling/subcooled boilingregions. There does not seem to be definite trendnear the peak values.

woo ~--------------------~.-----.

5400

N

E

I

"00 f1

4200 ~

I

3600 ~I

3000

2400

1800

1200Q = 29516

S ~Tsub Xi X

0.4 0·9 2012

Z, m

Fig. 5-Variation of heat transfer coefficient along the tubelength with submergence as parameter for methanol-water

system

The typical variation of heat transfer coefficientalong the test section length with heat flux as par-ameter has been shown in Fig. 4. At approximate-lythe same value of inlet liquid subcooling andsubmergence, the characteristic behaviour ofcurves for higher values of heat flux get shifted tohigher heat transfer coefficients than those forlower heat flux values. As the heat flux is raised,the initial portions of the curves (decreasing andslowly rising) get shortened shifting the point ofsteep rise towards tube inlet. Fig. 5 shows an ef-fect of submergence on the variation of h with Zat a constant heat flux. The curves, at the samevalue of heat flux, generally shift to higher valuesof heat transfer coefficient as the submergence islowered from] 00 to 30 per cent.

The characteristic behaviour of h versus Zcurves are indicative of heat transfer mechanismgoverned by various parameters. The initial por-tions of curves with h decreasing may be assignedto the entrance effect. This is observed to be pro-nounced at low heat flux values and ! or highsubmergence where the heat transfer is dominant-ly convective. The increase in h with Z at a sJO\\'rate, after the entrance effect may also be due tothe convective heat transfer followed by progress-

70 INDIAN J. CHEM. TECHNOL., MARCH 1997

6400

5600

4100

4000V...e

3200-~.s: 2400

1600

.00

5000

sooe

7000

6000

" • Z!l516S.50 4000

Ill,"b Xi X.2-3 0.00 0.001.1 .258 0.2411.5 0.287 0-2190·9 0.576 0.521

• 1.3 1.0 1.0

o '---...L----'--_.l..-._--'-_--'-_--l1OO0o I~ l6

Z,m

. Fig. 6-Variation of heat transfer coefficient along the tubelength with concentration as parameter for methanol-water

system

ively increasing subcooled boiling. The steep risein the values of heat transfer coefficient corre-sponds to the onset of saturated boiling. The en-hancement of transfer coefficient at varying ratesalong the tube length may be accounted for bythe additional turbulence due to nucleation,growth and detachment of vapour bubbles result-ing in two phase with different flow patterns. Withincrease in the value of heat flux larger number ofnuclei for bubble generation become active en-hancing the heat transfer coefficient. The changesin the characteristic behaviour of transfer coeffi-cient profiles with submergence seems mainly dueto the interaction between circulation rate andbubble dynamics in the tube. The circulation rateincreases with S, resulting in the dominant con-vective heat transfer with appearance of entranceeffect and slowly rising h regions. At low submer-gences, the circulation rate falls and the nucleateboiling dominates showing its effect as observedand discussed. The point, where saturated boilingsets in, gets shifted to a lower level in the tube asthe submergence is reduced from 100 per centprobably due to the change in circulation rates.The decrease in the value of liquid submergencereduces the driving force for liquid circulationand hence its rate through reboiler tube. At alower rate of liquid circulation the rate of changein its temperature along the tube length becomeshigher and the saturation temperature is attained

8000

,,

5=100q Ll.Tsub

e 12289 3·3 - 4.6A 16618 3J - 4.1e 21305 as - 4-4

4200

3800

3400·

3000

:.'2600

N

E

~- 1200

'".""1800

1400

\000

5000

,,,

3000

2000

0.4 06 1.0Cone eont fa t ion I X

Fig. 7-Variation of heat transfer coefficient with concentra-tion for methanol-water system

at much smaller length from the inlet. When thesubmergence is low, the net flow of liquid is smalland local circulation between the tube wall and li-quid bulk dominates resulting in a condition simi-lar to pool boiling.

Fig. 6 shows the effect of mixture concentrationon variation of h with Z, at same heat flux and li-quid submergence. While the main feature makingthe shape of the curves are same at all concentr-ations as observed earlier, the relative magnitudes,positions of boiling incipience and nature of var-iations change with composition. The change inthe behaviour of heat transfer coefficient profilesfor binary mixtures may be attributed mainly tothe physical properties of mixtures being differentfrom those of their constituents and presumablythe effect of mixture composition on the mechan-ism of the boiling process. Further, with the prog-ress of vaporization, the larger quantities of morevolatile component go into the vapour phase re-ducing its concentration into the liquid phase, thesaturation temperature increases and the aboveeffect is intensified. But the higher concentrationof vapour phase than that of the coexisting liquidphase in the tube presents the necessary concen-tration gradient for mass diffusion which might af-fect adversely the bubble growth rate. Thus the

KAMIL: HEAT TRANSFER IN BOIUNG OF MIXTURES 71

4200

3800

3400

3000

~NE 2600

~W 2200.c

1800

1400

1000

6000

q = 16599I 5 I l>Tsub I

.1100 31 41OI?5.18!d..i~5OT1:f-461!,,130126451

0·2 0.4 0.6 0.\1 1.0

Concentra1ionl XFig. 8-Variation of heat transfer coefficient with concentra-

tion for methanol-water system

variations of h with Z for binary systems are go-verned by all the. parameters involved for purecomponents and the additional effect of mass dif-fusion on bubble growth rate in a very complexmanner.

The average values of heat transfer coefficientfor saturated boiling of binary mixtures have beenplotted against their concentrations in Figs 7 and8, respectively. These figures clearly demonstratethat hB decreases with increase in concentrationof more volatile component in the mixture upto acertain value beyond which it increases showing aminima. The general nature of variation for a sys-tem remains essentially the same at all heat fluxesand liquid submergences. The shifting of curveswith q and S follows the same trend as observedearlier for pure components and mixtures. Themixture concentration at which the heat transfercoefficient becomes minimum is nearly the sameat all heat fluxes and submergences for a system.The experimentally obtained values of hB for mix-tures are invariably lower than their weighted av-erage at any given concentration. The differencebetween the two values is maximum at the con-centration corresponding to the minimum value ofh« of a system.

The reduction in the values of heat transfer co-efficient for mixtures as compared to the weightedaverage of the constituent components may be at-tributed to the change in properties on mixingand the effect of mass diffusion on boiling pro-cess. When a vapour bubble forms and grows in aliquid mixture, the concentration of more volatilecomponent in the vapour bubble, which is in eq-uilibrium with the surrounding liquid, becomeshigher than that in the surrounding liquid. Thedifference between the vapour and liquid phasecompositions (y-x) provides the necessary concen-tration gradient for the diffusion of more volatilecomponent from the bubble to the liquid. Thisprocess being just opposite of the bubble growthand hence the rate of bubble growth gets re-tarded. The micro-convective contribution to heattransfer, thus gets reduced resulting in a lowerheat transfer coefficient than the expected valueof weighted average. The rate of diffusion andhence the retardation of bubble growth is maxi-mum at a concentration at which (y- x) is maxi-mum. This is the reason why the values of fiB areminimum at the concentrations which are almostsame as those corresponding to the maximum va-lue of (y- x).Conclusion

The values of heat transfer coefficient duringboiling of binary mixtures are generally lowerthan the weighted average of the pure componentvalues. This reduction may be due to the expect-ed retardation of bubble growth rate by the diffu-sion of more volatile component from the vapourbubble to the saturated liquid.Nomenclature

= heat transfer coefficient, W/m'oC= average heat transfer coefficient, W/m2 °C= average boiling heat transfer coefficient, W/m' °C= circulation rate, kg/s= liquid flow rate from condenser, kg/s= heat flux, W/m= submergence, %= temperature,°C(K)= temperature difference, ( T; - TL), °C= degree of subcooling, ( Ts - TL), °C= vapour fraction.= mass fraction of component in liquid phase= mass fraction of more volatile component in vapour

phaseZ = distance along the test section, mSubscriptsB = boiling

= inlet= liquid= logarithmic average= outlet= onset of boiling= wall

hhitBmM,qST6.T6. 7;uhXO

xy

L1moOBw

72 INDIAN J. CHEM. TECHNOL., MARCH 1997

References1 Alam S S, Kamil M & Jamal A, Indian J Chern Technol;

2(1995)287.2 Sardesai R G, Palen J W & Thome J R, Heat transfer in

nucleate boiling of binary mixtures, AIChE winter AnnualMeeting, paper 127a, 1986, Miami Beach, Fla.

3 Schlunder E J, Int Chern Eng, 23 (1983) 589.4 Thome J R & Shock RAW, In Advances in Heat Trans-

fer, Vol 16, edited by Hartnett J P & Irvine T F Jr (Acad-emic Press, New York) 1981,59.

5 lung D S, Mclinden M, Ri Radermacher & Didion D, IntJ Heat Mass Transfer, 32, (1989) 1751.

6 Kusumowardhoyo M A & Hardianto T, Int J Heat FluidFlow, 6, (1985) 265.

7 Ponter A B & Peir W, Int J Heat Mass Transfer, 21 (1978)1025.

8 Ross H, Radermacher R & di Marzo M, Int J Heat Mass

Transfer, 30 (1987) 979.9 ThomeJ R, Heat Transfer Eng; 10, (1989) 29.

10 Toral H, Kenning D B R & Shock RAW, Flow boilingof ethanol/cyclohexane mixtures, Proc of the 7th Int HeatTransfer Conf, 4, 255-260,1982, paper FBI4.

11 Kamil M, Studies on a vertical tube thermosiphon reboiler;Ph D Thesis, Aligarh Muslim University, Aligarh (1992).

12 Kamil M, Alam S S & Ali H, Chern Eng World,XXVII(2), (1992) 67.

13 Kamil M, Alam S S & Ali H, Heat transfer to boiling bi-nary mixtures in a reboiler tube, Proc of Nat Seminar onApplication of Chemical Engineering in the Utilization ofNatural Resources (CHEMlNAR-93), F. 8.1-F.8.5., Feb.,1993,24.

14 Kamil M & Alam S S, Int J Heat Mass Transfer, 39,(1996) 1489.

15 Saha P & Zuber N, Int J Heat Mass Transfer, 21 (1978)415.