Desalination 169 (2004) 151–160

Condensation of steam with and without the presence of
non-condensable gases in a vertical tube
S.B. Al-Shammaria*, D.R. Webbb, P. Heggsb
Water Technologies Department, Kuwait Institute for Scientific Research, PO Box 24885, Safat 13109, Kuwait
Tel. +965 487-8122-4; Fax +965 487-9238; email:
Department of Chemical Engineering, University of Manchester Institute of Science and Technology (UMIST),
PO Box 88, Manchester, UK M60 1QD

Received 19 March 2003; accepted 20 November 2003

Condensation of steam and steam from non-condensable gases were investigated experimentally inside a vertical
tube. The consistency and reliability of the measurements were checked by detailed energy balance. The experimental
results show the effect of non-condensable gases on heat and mass transfer resistances in the gas phase of the
condensation process. In the steam runs, the gas side heat transfer resistance was very small and the main resistance
was due to condensate film. This resistance increased significantly for steam–air runs, particularly toward the top part
of the tube. This occurred as a result of accumulation of the non-condensable gases near the surface as the condensation
process proceeds, which act as an extra resistance to heat transfer.

Keywords: Condensation; Steam; Non-condensing gases; Vertical tube

1. Introduction
film thickness when the flow is laminar, but
Condensation is important and forms an inte- lower resistance is found for turbulent flow. The
gral part of many engineering systems such as condensation of binary mixtures produces an
steam power plants and seawater desalination extra heat and mass transfer resistance in the gas
units. In condensation of pure saturated vapors, it phase. This extra resistance may become very
is generally assumed that the only rate controlling large if one of the components is a non-
process is that of heat transfer across the conden- condensing gas. The reason for this is that the
sate layer. The resistance to heat transfer of the condensation process effectively sweeps the gas
condensate film is directly proportional to the to the surface where it accumulates as a gas-rich
layer. The transfer of vapor to the surface may
*Corresponding author. then become controlled by the process of

0011-9164/04/$– See front matter © 2004 Elsevier B.V. All rights reserved

25 mm ID while the annulus is made of depending on concentration. which leaves the main test . The inlet section of the condensable gases on heat transfer coefficients. measured using a U-tube mercury manometer. This parameter controlled by means of a thyristor. This heater is transfer analogy approach. Apparatus For experimental work with air.152 S. / Desalination 169 (2004) 151–160 diffusion. and outlet from. currently in the annulus. ratus was designed so that the vapor generated in Sparrow and Marschall [2] improved the theo. The a UMIST heat transfer rig in a university pilot vapor generated in the boiling vessel (V-9) passes plant. The test section was heated by an sation thermal conductivity using a heat and mass electric heater that supplies 36 kW. Six thermocouples are located at equal theory and considered the interactions as due to intervals within the outer tube (annulus) to diffusion in the gaseous mixture.1. In addition. tappings used for thermocouple installation and side and inside vertical tubes was analyzed by pressure measurement. the boiler enters and exits vertically through the retical study by considering the influence of inner pipe while the cooling water flows counter- temperature difference on the effect of non. a relatively slow process compared circulation system and power supply. istics inside vertical tube. The sub-systems: the test section. This paper focuses calibrated differential pressure (DP) cell. annulus was designed to allow the cooling water Experiments on condensation of steam–air to distribute uniformly around the annulus before mixtures outside tubes in a horizontal heat entering the cooling section. The test appa- in steam can reduce the heat transfer by 50%. intervals. the design to measure the temperature in the test stances. [5]. Experimental vacuum. was assumed to contain a single non-condensable a total of 12 thermocouples are incorporated in component and any number of condensing sub. the test tube. The inner tube of the annulus is a copper tube authors. 2. The proposed model was based on film section. They tested different flow patterns and The condenser is equipped with a large number of different operating conditions. This will ensure exchanger were performed by Rashtchian and efficient turbulent mixing of the cooling water. The on investigating experimentally the influence of system pressure at the top of the test section was non-condensing gases on heat transfer character. The vapor phase inlet to. which is obtained by a liquid ring vacuum pump system. The heat transfer rig is well insulated and operated under 2. Condensation out.B. tical concentric pipes forming a cylindrical annu- densation has been extensively studied by many lus. They derived effective conden. air is fed into The experimental work was undertaken using the condenser via a calibrated nozzle box. Webb [3]. A schematic diagram of this apparatus is with a known amount of air to the test section shown in Fig. 1. investigated experimentally and analytically by which are welded into the test tube wall at equal Peterson et al. The absolute allowed the prediction of combined sensible and pressure and differential pressure over the section condensation heat transfer using standard forms were measured using either a manometer or for heat transfer correlations. The measurement of the absolute pressure required the atmospheric pressure scale. Al-Shammari et al. measure the coolant temperature profile. Minkowyez and Sparrow [1] found that with a 28. These are located at every Mazzarotta and Sebastiani [4]. Forced con. Wall vection condensation inside a vertical tube was temperatures are recorded by six thermocouples. the cooling medium uncondensed vapor.7 mm ID. The main with convection. test section (HE-1) is basically 3 m double ver- Non-condensable gases’ effect on steam con. a half percent of air stainless steel with 62. The system is composed of three (HE-1) where it is partially condensed.

Al-Shammari et al. 1.B. . / Desalination 169 (2004) 151–160 153 Fig. S. Schematic diagram of the test section.

All measurements of The energy balance calculations are based on the thermocouples were collected and viewed the amount of condensate collected and the with the SCAN1000 data acquisition system. is calculated by measuring the condensation flow The experimental work was carried out rate and using the latent heat of vaporization of according to the procedures described earlier..6–63. is shown in Fig.22 0. 4. experimental measurements. The energy balance Non-condensing gases are vented from the sys. 2. they are collected in a condensate buffer vessel (V-5) and recirculated to the boiling vessel (V-9). which shows a schematic tem through a liquid ring pump. as shown in Eq. calculating the sensible heat changes of the vapor Some of these runs were immediately rejected mixture and the condensate.77–57. kW 3.54 1. water.43–5. bar 0.2 Coolrant flow rate..33 Feed flow rate.12 43. which is measured by a taken by setting the SCAN1000 to “record flow meter.61 Heat load.075 2. Al-Shammari et al. (1). °C 56. and the inlet and outlet coolant tempera- the system could be switched back to normal by tures. l/min 10–20 10–20 section. Description of energy balance The overall energy balance of the test section was carried out to confirm the reliability of the .19–0. passes to a total condenser (HE-2). positions in the flow loop. and V-4 or they are combined. After completion of the measurements.B. cooling water flow rate.22–8.2. and the condensate mixture.89 Vapor inlet temp.24–11.16–0. °C 55. kg/h 5.73–65 Vapor outlet temp.154 S. (1) 2. / Desalination 169 (2004) 151–160 Table 1 Summary of experimental conditions System Steam Steam–air Pressure. which is directly cooled by the refrigeration loop. kg/h 0. This is corrected by all runs were conducted under reduced pressure. Schematic diagram of the test section. because they were too close to the flooding point.18 52. In either case. 2. Measurements of diagram of the test section illustrating the inlet temperatures and pressures were taken at various and outlet streams. The condensate produced in HE-1 and HE-2 is either passed separately to measurement vessels V-3 Fig. The main conditions under which data were measured are summarized in Table 1.34. The heat gained by the When steady-state conditions were reached in the cooling water can be calculated directly from the test section. the specific heat capacity of the mode”. then the measurements could be flow rate of the coolant.3 Air flow rate.68–62. The energy transferred from the vapor side setting the SCAN1000 to “monitor” mode.49 5.94–5.

For pure steam experiments. for steam and steam–air experiments. the resistance to heat transfer is very high. Energy balance results coefficient is very low.i!Tcool. then the condensing side heat transfer 2.i!Tcool. S. The effect of non-condensing gas (air) on the (3) condensation process is examined by introducing air into the system through the nozzle box.. The second important observation is that the reduction in condensation temperature on the vapor side becomes much larger than in the case of the steam only runs. This reveals that an extra resis- Fig. does not arise from the drop in pressure Assuming that there are no heat losses to the or increase in the concentration of air with ambient.6. if (Tw.6 kg/h was used in the steam–air mixture runs. Now. vapor inlet: if the temperature difference between the wall and the annulus (Tw. of coolant and vapor sides in each interval.. once the test section was estimated using the bottom wall heat flux and wall temperatures are known.6 and (Tcool)i=1. hence. and therefore this side provides the dominant resistance to heat transfer. 3 shows that Figs. Fig. tance has been added to the condensing side due . / Desalination 169 (2004) 151–160 155 The condensate temperature at the bottom of the It is clear from this equation that.. (Tw)i=1.2. with the fact respectively. 4–6 show typical temperature profiles the energy balance are within ±15% for all runs. 3. respect- There are six measurements of the wall and ively. The change in the vapor temperature is much greater than that of the steam only runs.i is very (2) low and. it follows that the composition of the heat flux to be evaluated in each interval and the steam must remain nearly constant. An air flowrate of 2.i) is very low. The measurement of the change of that the vapor phase is never too far from satu- enthalpy of the coolant in each interval allows the ration.. There is a measurements of wall temperature can be used to small drop in measured vapor temperature that calculate the individual heat transfer coefficients does not occur in the saturation value and. hence.i) is very high. therefore. the following expression gives the heat condensation. This is expected behavior because of the accumulation of the non-condensing gases (air) as the conden- sation process proceeds. then the coolant side heat transfer coefficient "cool. the change in annulus temperatures. Temperature profiles quality of the experimental data.34 kg/h and 4. but may show a change in the flux for any interval: degree of superheating of the vapor. An energy balance for each experiment over the test tube was carried out on the condensing vapor and coolant pump-around to assess the 2.. the temperature and the saturation temperature at the heat transfer coefficients may be evaluated.3..2. Al-Shammari et al. Energy balance across the test tube.4. decreases the steam condensing temperature...B. the vapor temperature is small and. On the other hand. which reduces the steam partial pressure in the gas phase and.

Fig. Temperature profile for pure steam runs. / Desalination 169 (2004) 151–160 Fig. . Temperature profile for steam–air runs. Temperature profile for steam–air runs. Fig. 6. Al-Shammari et al.156 S.B. 4. 5.

the intermediate measurements of the coolant temperatures in the analysis are not used. and it is replaced determined by the energy balance [Eq. together with the wall and reliable heat loads in each interval. Eq. opposing demands on the coolant flow. in the analysis by the integral following expression: approach. This flux is inversely proportional to the magnitude of the overall heat transfer resis- tance. Thus. Gas stant along the length of the test tube. The condensing side resistance consists of In the approach followed. did not give tures are known.1. S. the coolant side heat two parts: the resistance due to the gas film and transfer coefficient must be assumed to be con- the resistance due to the condensate film. since the measured intermedi- interval in which inlet and outlet coolant tempera. Therefore.i. (5)]. Results (4) 3. To ensure condensing gas. (7) Since the temperatures of the wall and annulus of the test section are measured and recorded for Combining Eqs. load on the coolant side: ation of the heat transfer resistances present on both the condensing and coolant sides of the test (5) tube wall. In the integral approach. i. more seriously. and six different intervals during experimental runs. together with mined and used to obtain the known total heat the wall temperature variations. understood by considering the unit to be a single Unfortunately. fully developed flow and the absence of condensation of a pure vapor. or. (7) defines the heat load and Eq. The gas side resistance vanishes this constancy. there are six heat fluxes to be deter- heat flux along the test tube. to provide a uniform and large Two approaches were considered: a differ. Al-Shammari et al. (4) defines the (6) coolant heat transfer coefficient. They are best pinch. which. (4) and (7) gives an expression . the annulus temperature is not con- sidered as measured with sufficient accuracy to For each interval.B. ments with condensation of steam alone are used there is no chance then to achieve a measurable to characterize the heat transfer resistance of the temperature change for energy balance purposes condensate film. temperature driving force without a temperature ential and an integral approach. This is a side heat transfer resistance is clearly found in weakness of the present work because there are condensation of a pure vapor from a non. allows an evalu. this resistance is much larger than that due to the the heat load can be estimated using the following condensate. Heat flux variation along the test tube It is necessary to determine the variation of the Therefore. using with the average coolant temperature in the coolant inlet and outlet temperatures by the interval. the latter defining the on the mean of the coolant inlet and outlet temperature driving force. they are based annulus temperatures. In the differential temperatures for the interval: approach (the coolant energy balance). On the other hand. developing boundary layers. Tcool. equation applied in each interval: 3. the heat load can also be provide consistent behavior. it is desirable to have a high and can be considered to be negligible in flowrate. experi. ate coolant temperatures. / Desalination 169 (2004) 151–160 157 to the presence of the non-condensing gases.

B. which is the inlet temperature of the next interval i+1: (8) where Fig. can be calculated for the steam-only experiments The rather high coolant flow rates result in rather from the measurement of local heat flux ( ) and small coolant temperature changes. 7 shows a typical variation of the heat (10) load along the test tube for steam-only and steam–air experiments.158 S. 7. (8) and (9) above are solved using an in the concentration of the non-condensing gas iterative procedure where repeated estimates are (air). For steam experiments where only. The latter assumption is top of the condenser resulting from the increase reasonable because the exchanger is divided into . Tw.2. 3. Al-Shammari et al. This reveals that the condensing side heat negligible gas-side resistance and linear vapor transfer coefficient decreases sharply toward the temperature profile. The trend is reversed from the case of con- made of the coolant heat transfer coefficient until densation of a pure species where the heat trans- the measured and calculated heat loads agree: fer coefficient is smaller at the bottom of the tube. Typical variation of the heat load along the test (9) tube. Eqs. Fig. This is because of an increase in the condensate film thickness. / Desalination 169 (2004) 151–160 for the outlet coolant temperature of interval i. The presence of non-condensing gas on the local heat flux along the test tube shows a reduction in the heat flux toward the top of the This approach is based on the assumption of tube. and neither wall temperature. which increases the heat transfer resistance on the condensing side. the general trend of all runs shows that heat load decreases towards the bottom of the test tube. Condensate heat transfer coefficients The local condensate film heat transfer coeffi- Although differential behavior has been aban- cient ("cond. this approach is considered reasonable.i) in each interval of the test section doned. because there is no gas-side the averaging process nor the assumption of resistance to heat transfer: constant heat transfer coefficient should then be too serious.

The controlling. This occurs because the condensate film stant toward the top of the tube. overall resistance. the gas side resistance never became fully profile in each stream is essentially linear. contributing at most about 50% of the latter assumption was used to obtain a relation. Reliable measurements are reported for the wall temperatures.500 to 10. for steam-only runs. Variation of condensate heat transfer coefficients from top of tube for steam-only runs. which characterize the resis- tances to heat and mass transfer on the con- densing side. 8. condensation process proceeds. the calculated was removed at the tube outlet. These are the condensation of pure steam and steam–air mixtures under reduced pressure. near the vapor out- ship between gas temperature and tube length. particularly toward the top part of the tube. the resistance to heat and mass transfer on the condensing side in- creased. which acted as an extra resistance to heat transfer and decreased the rate of condensation of steam. This resulted tance to heat transfer. However. temperature. most of the heat load was the bottom of the tube. 8 and 9 show the variation of the con. showing almost full gas side control. Tgi. from the increase in heat and mass transfer . In addition. the gas side heat transfer resistance was negligibly small and this explains why the wall temperature measurements approach the vapor temperature at the top of the test tube.B. tube for steam-only runs. the heat and heat transfer coefficient towards the bottom of the mass transfer resistances remained nearly con- tube. This was due to values of ranged from 3. On the other originates at the top and becomes thicker toward hand. the wall temperature measurements decreased which can be used to determine the local gas sharply and sometimes approached the coolant temperature. showed that. 9.000 W/m2 K. exchanged at the bottom of the tube. let. Conclusions Two sets of data were obtained in the experimental work. Analysis of reflux condenser data showed agreement with energy balances of ±15% for steam and steam–air experiments. which increases the resis. The heat load distribution along the test tube densate heat transfer coefficient with tube length. Figs. the large temperature driving force between the The general trend shows a decrease in the local gas phase and coolant. This was due to accumulation of the non- condensing gas (air) near the surface as the Fig. Variation of heat transfer coefficients from top of error in temperature measurement. for steam–air. In the steam runs. The most significant contribution to the errors is the energy balance on the coolant side where the temperature rise is too small to give good precision due to Fig. In the steam–air relatively small increments and the temperature runs. / Desalination 169 (2004) 151–160 159 4. most of heat As can be seen in the figures. When air was introduced. Al-Shammari et al. S.

. kW/m2 steam from mixtures with air in a shell and tube q0 s — Heat flux towards interface.M. 65 (1987) 157. gravity- n — Number of components flow film condensation. Webb. Eng. Rashtchian and D. kW/m2 [3] D.J. 2.B. T0 — Temperature at the wall. Reflux condensation in vertical tube. Can. a wealth of reliable data has been efficient. 89038. °C [5] P. Design.K vapor — Molar heat capacity. Process design of sf. 73 T — Temperature. J. Binary. which reduced "cool — Local coolant heat transfer co- the heat and mass transfer rates in the top section efficient. Condensation of q0 o — Heat flux to coolant. " 6 cool — Average coolant heat transfer co- Thus. J. ASME. 1989. New York. "c — Condensate heat transfer coeffi. Ing. Chem. kg/s Mass Transfer. J/kmol. J/kmol — Latent heat of vaporization. Item No. Sebastiani. J/kg References i — Component number L — Length. )T — Temperature difference. ESDU International Plc.160 S. Nusselt. .M. Trans. Pergamon Press. kg/kmol [2] E. Res. Mazzarotta and E. Peterson. q0 — Heat flux. cient. Heat Transfer. 92 (1969) 205–211. kW/m2 Chem. Al-Shammari et al. Int. Engineer. °C (1995) 456. [7] ESDU. exchange at atmospheric and reduced pressures. kg/m s and mass transfer processes. Kapitsa. Collected Papers by P.E. J. 1965..R. condensers for vapor mixtures in the presence of fer films. Symbols Nu — Nusselt number Re — Reynolds number A — Heat transfer area. difference. Sparrow and E. — Vapor fraction of component i Ver.L. 115 (1993) 98–103. m non-condensable gases. m2 Reg — Reynold number of the inlet cp — Specific heat capacity.L. — Molar mass. Sparrow. Heat — Mass flow rate. W/m2k of the tube. Deut. V. m [1] W. Kapitsa. W/m2k Vol. Schrock and T. Marschall. Kageyama. st — Thickness of heat and mass trans. kW [4] B. Z.K — Molar latent heat of vaporiza- tion. / Desalination 169 (2004) 151–160 resistances toward the tube outlet.F. W/m2k measured that can be used to assess the methods ' — Mass flowrate per unit peri- that are available for the prediction of the heat meter. K [6] W. — Total heat load. Dimensionless groups 5. Engi- Greek neering Sciences Data Unit. J/kg. Surface condensation of water vapor. 9 (1966) 1125. 60 (1916) 541–546. [8] P. Minkowycz and E. K Diffusion layer theory for turbulent vapor conden- )TLMTD — Logarithmic mean temperature sation with non-condensable gases.