Heat Mass Transfer (2009) 45:1037–1046

DOI 10.1007/s00231-007-0317-9

SPECIAL ISSUE

Evaporation of pure liquids with increased viscosity

in a falling film evaporator
Felix Weise Æ Stephan Scholl

Received: 2 August 2006 / Accepted: 28 May 2007 / Published online: 1 August 2007

Springer-Verlag 2007

Abstract The present study investigated fluid dynamics h heat transfer coefficient (W m–2 K–1)
and heat transfer of viscous pure liquids in a falling film Ka Kapitza number (–)
evaporator. This is of special benefit as it avoids mass MEG Monoethylene glycol
transfer effects on the evaporation behaviour. Experiments Nu Nusselt number (–)
at a single-tube glass falling film evaporator were conducted. Pr Prandtl number (–)
It allowed a full-length optical film observation with a high- Re Reynolds number (–)
speed camera. Additionally the evaporator was equipped Re* Alternative definition of Re (–)
with a slotted weir distribution device. Test fluids provided PG Propylene glycol
viscosities ranging from l = 0.3 to 41 mPa s. The Reynolds
number was between 0.7 and 1,930. Surface evaporation and
Greek symbols
the transition to nucleate boiling were studied to gain :
C Mass flow rate per unit tube circumference (kg m–1 s–1)
information about the film stability at maximum wall
k Thermal conductivity (W m–1 K–1)
superheat. A reliable database for laminar and laminar-wavy
l Dynamic viscosity (Pa s)
viscous single component films was created. The experi-
m Kinematic viscosity (m2 s–1)
mental results show a significant enhancement in the wave
q density (kg m–3)
development due to the film distribution. A wavy flow with
r surface tension (N m–1)
different wave velocities was superposed to the film in each
liquid load configuration without causing a film breakdown
or dry spots on the evaporator tube. It was found that
Subscripts
nucleate boiling can be allowed without causing film insta-
BT Boiling temperature (C)
bilities over a significant range of wall superheat.
CT Condensation temperature (C)
crit critical
i inner
List of symbols
l laminar
A, B, C Constants
turb turbulent
a, b Constants
w wall
cp specific heat capacity (J kg–1 K–1)
wl wavy-laminar
CHEX Cyclohexanol
ov overall
g gravity (m s–2)

F. Weise
S. Scholl (&) 1 Introduction

Institute for Chemical and Thermal Process Engineering (ICTV),
Technical University of Braunschweig, Langer Kamp 7,
38106 Braunschweig, Germany Thin liquid films flowing on vertical, inclined or horizontal
e-mail: s.scholl@tu-bs.de walls or tubes are frequently found in several industrial

123
1038 Heat Mass Transfer (2009) 45:1037–1046

applications as e.g. the (petro-)chemical, pharmaceutical base on his theoretical approach. The film Reynolds
and food processing. Basically liquid films in the range of number is used to quantify the fluid dynamic state of the
0.1–2 mm thickness are predestined for thermal treatment film. For free falling film systems it is defined as:
of heat sensitive products due to relatively high heat
transfer coefficients and low thermal stress. A falling film C_
Re ¼ ð1Þ
evaporator with a single or multiple tube arrangement is l
one design layout of a thin film heat exchanger. Besides the
or
general advantages of thin films, it additionally offers good
controllability, limited liquid hold-up and low-pressure
drop. Limitations in application and operation of falling 4
C_
Re ¼ ð2Þ
film equipment often originate from unfavourable fluid l
properties. Especially increased viscosities and the exis-
tence of solids in the process stream may cause severe in
:
Anglo-Saxon regions. For both definitions it is:
problems in film distribution and stability. In disadvanta- C ¼ mass flow per unit inner tube circumference
geous cases a film breakdown followed by product damage (kg m–1 s–1)
may be the ultimate consequence. The design of falling l = dynamic viscosity (Pa s)
film evaporators is often based on empirical correlations
and rules of thumb, especially for the maximum heat flux For later comparison all correlations using Re* will be
and minimum liquid load. When designed at the border- normalized to Eq. 1.
lines of established operating regions significant over- According to Brauer [3], who followed Eq. 1, the flow in
design may be found. This can largely be attributed to a non-heated aqueous falling film can be classified as
a limited database. It is common knowledge that flow follows:
regimes exert the most pronounced influence on heat laminar flow: Re < 4
transfer, see e.g. [1]. In general the overall performance of transient flow: 4 £ Re £ 400
a falling film heat exchanger depends on: turbulent flow: Re > 400

• the liquids properties The liquid forms a flat film within the laminar regime,
• the geometries of film distribution device and wall/tube which is called ‘‘smooth laminar’’. Under these fluid
surface dynamic conditions, the fluid develops the parabolic
• the liquid load velocity profile described by Nusselt’s theory, see Fig. 1.
• thermal stress When exceeding the critical Reynolds number for the
• the mode of operation (counter- or co-current), vapour laminar condition (Re > 4) typically first waves appear
phase shear stress, etc. after a small fluid dynamic start-up zone. First sinusoidal
Based on the fluid dynamics it can be specified that an waves can be observed followed by horizontal wave peaks,
increase of heat transfer rates is found for wavy and finally see Fig. 2. This structure is the 2-dimensional wavy area.
turbulent film flow.
Since the fluid properties may not be altered signifi-
cantly in a process, the most important factors to affect film
structure are distribution device and tube geometry. This
study shows the influence of a slotted weir film distribution
device, in combination with grooved tube geometry, on the
initiation of waves and the minimization of the fluid dy-
namic inlet zone. Additionally experimental data are ob-
tained for the maximum wall superheat before film
instabilities occur.

2 Theory

The first theory about thin (non-wavy) films was developed

by Nusselt in 1916 [2] and, although his work focussed on
aqueous films, to this day most of the developed correla-
tions to describe the fluid dynamics, heat and mass transfer Fig. 1 Laminar and turbulent velocity profiles

123
Heat Mass Transfer (2009) 45:1037–1046 1039

The flow condition in this transient range is known as 3–4 times the thickness of the film. In a related study,
‘‘wavy laminar’’. When Re is increased further more Wasden and Dukler [8] explained how these waves caused
v-shaped wave peaks develop with different amplitudes recirculation effects in the film, which gave rise to the
and velocities. This region is 3-dimensional but still with a increase of the heat and mass transfer rates. Further
laminar base pattern although developing rolling waves investigations by Jayanti and Hewitt [9] have focused on
might travel faster than smaller ones and the residual film. proving that it is not this recirculation effect that causes
Nevertheless these waves remain small and regular up to transfer enhancement, but rather the effective thinning of
Reynolds numbers of approximately 400. Beyond Rey- the film. They claimed that the evolution of the wave shape
nolds numbers of this critical value for the turbulent con- is too slow to affect the fluid dynamics. They concluded
dition (Re > 400) larger and irregular waves arise and that the overall heat transfer coefficient is primarily
additionally the fluid elements can develop an irregular determined by conduction through the film.
transverse movement, perpendicular to the falling direc- Within the last 20 years the step from statistical analysis
tion. In general the wave velocity of the rolling waves is to numerical simulation of wave evolution in thin falling
decoupled from the residual film whereas it is almost the films was accompanied by the development of measure-
same in the wavy-laminar state. In these conditions the film ment techniques that can visualize local film thickness and
is ‘‘fully turbulent’’ and leads to a significant enhancement velocity profiles [10–12]. Al-Sibai [13] studied the influ-
of the heat and mass transfer rates. Figure 2 summarizes ence of fluid properties on fluid dynamics and heat transfer.
the different wave evolutions with an annular gap distri- Assuming that the Reynolds number is not sufficient to
bution device as it frequently used in experimental setups. describe the film in his state, they used Particle Image
In order to improve understanding of the relationship Velocimetry and an Infrared Camera to prove that liquid
between the wave phenomena and an increase of the heat films with similar Reynolds number form different wave
transfer, many experimental studies have been performed profiles. The Kapitza number was used to consider differ-
in the last decades. Early attempts to test and improve ent fluid properties, especially the influence of surface
correlations were made by Kapitza [4] who applied sinu- tension on wave evolution and heat transfer. It is defined
soidal wave models. These models predict lower transport as:
rates than those from Nusselt’s theory at high Reynolds
number. Further attempts were made by Seban and Faghri q
r3
Ka ¼ ð3Þ
[5] who observed an enhancement of heat transfer due to g
l4
the waves through experiments but were unable to explain
the nature of the waves. For a closer look on the structure with
of these irregular waves, Telles and Dukler [6] and Chu and g= gravity (m s–2)
Dukler [7] carried out a large series of experiments. Their r= surface tension (N m–1)
major finding was that the film structure could be described q= density (kg m–3).
adequately as large waves travelling on a thin and nearly
smooth sublayer. The amplitude of these waves often was To date available computer capacities allow detailed
numerical simulations in 2-D and 3-D. The understanding,
simulation and prediction of falling liquid film behavior
was the objective of several authors. Alekseenko [14],
Miyara [15] and Heidrich [16] summarize the work of the
last decade and present their own correlations and models
for falling liquid film structures but each still in a selected
range of Reynolds, Prandtl and Kapitza numbers.
Besides the understanding of falling films on smooth
surfaces many authors influenced fluid dynamics and heat
transfer with special tube geometries:
• twisted tubes with profiles, see e.g. [17]
• rough, porous surfaces, e.g. [18]
• Gregorig–profiles (double fluted tubes), e.g. [19]
• corrugated tubes, film promoters, e.g. [20, 21]
In each case the heat transfer was increased—partly with
a significant tendency to fouling, film breakdown and in-
Fig. 2 Wave initiation creased pressure drop.

123
1040 Heat Mass Transfer (2009) 45:1037–1046

While Re quantifies the fluid dynamic state of the film Prdþ1=3

its physical properties are reflected by the Prandtl number: Nuturb ¼
ðA1 Pr 3=4 þA2 Pr 1=2 þA3 Pr 1=4 þC1 Þþ ðBKa1=2 Pr 1=2 Þ
l
cp ð10Þ
Pr ¼ ð4Þ
k
with
with A1 = 9.17
cp = specific heat capacity (J kg–1 K–1) A2 = 0.328 p (130 + d+)/d+
k= thermal conductivity (W m–1 K–1) A3 = 0.0289(152100 + 2340 d+ + 7 d+2)/d+2
B= 2.51 · 106 d+0,333 Ka–0.173/(4
Re)(3.49 Ka^0.0675)
It represents the ratio of momentum and thermal diffusiv- C1 = 8.82 + 0.0012 Re
ity. In fact it compares the film thicknesses of the velocity d+ = 0.2868 Re0.8
boundary layer and the thermal boundary layer. By varying
the temperature level it is possible to influence the viscosity It should be noted that Alhusseini et al. [25] suggested the
of the fluid and thus the Prandtl number. reciprocal definition of the Kapitza number for his
For the design of a falling film heat exchanger a number correlations.
of different correlations for the Nusselt number have been The purpose of this study was to evaluate the possibility
reported in the open literature. Based on results of Nusselt to directly affect the film structure with a special distri-
[2], Wilke [22] started with binary mixtures followed by bution device together with a tube geometry that already
Chun and Seban [23] whose correlations for the heat showed better results for the heat transfer in previous
transfer coefficients have been established in many engi- studies. Additionally the question of whether Re and Pr
neering handbooks as for example the VDI Wärmeatlas [1]. numbers sufficiently characterize fluid dynamics and
Palen et al. [24] continued with an experimental work for physical properties of an evaporating falling film was ad-
mixtures. Alhusseini et al. [25] investigated the evapora- dressed. Another topic was to analyze the effect of in-
tion for single component liquids to be able to neglect creased wall superheat on the transition from surface to
mixture effects. To predict the heat transfer performance nucleate boiling without generating film instabilities for the
for single component fluids different mathematical models tested viscous liquids.
may be found. A commonly used correlation is from Chun
and Seban [23] who proposed for evaporation of a gravity
driven falling film in a wavy-laminar and turbulent regime: 3 Experimental Setup
0:22
Nuwl ¼ 0:6045
Re ð5Þ The experimental part of this study was performed at a
single-tube vitreous falling film evaporator with the
Nuturb ¼ 0:0028
Re0:4 Prl0:65 ð6Þ dimensions 43 · 2.3 · 1255 mm, see Fig. 3. It allowed
 1=3 a full-length optical observation of the film using a
a m2l high-speed digital camera, which provided reference
Nu ¼ : ð7Þ
k g measurements of the film distribution, wave evolution and
The VDI Wärmeatlas [1] uses the above equations as a transitions in the type of evaporation. It was able to record
basis but little data is available to verify applicability sequences and single frames with up to 1,000 frames/s.
beyond a Prandtl number of 7 although a range of up to Using a frame-grabber and variable recording times as well
Pr = 50 is given. Another correlation expressing the film as frames per second it was possible to maximize the
Nusselt number as a superposition of wavy laminar and accuracy of the measured film characteristics. The CCD
turbulent film Nusselt numbers was introduced by cam was movable to obtain information about the film
Alhusseini et al. [25]: structure at different positions. The film distribution device
was a rectangular slotted weir in combination with a single
groove around the inner circumference of the evaporator
Nu ¼ ðNu5wl þ Nu5turb Þ1=5 ð8Þ tube, see Fig. 4. A centrifugal pump at the bottom allowed
with variation of the flow rate, which was monitored by a mass
flow meter. Fluid temperature was controlled by a heat
exchanger setting it close to boiling temperature. The
Nuwl ¼ 2:13
Re0:158
Ka0:0563 ð9Þ
test facility was equipped with thermocouples (PT100,
The turbulent film Nusselt number is given as a function of Ni–CrNi) at the relevant positions to allow for energy
the film Reynolds, Prandtl and the Kapitza number: balances on the heating and product side and to calculate

123
Heat Mass Transfer (2009) 45:1037–1046 1041

necessary product properties. Additionally mass flows of

feed, product and condensate were measured with high-
resolution scales. The product vapor was flowing in
co-current direction to the falling film. The process control
was fully automated. Controller alignment allowed running
a test with a minimum of fluctuations, so the reproduc-
ibility was optimized.
Heating was provided by organic vapor from a separate
thermostat, which allowed a good controllability in a wide
temperature range. The outer tubes and vessels were
insulated to avoid thermal losses. Heating vapor conden-
sate at the outer tube side was collected separately to allow
for an improved energy balancing. The experiments were
performed under vacuum conditions with a minimum of
12 mbar absolute. Together with boiling temperature the
viscosity of the evaporating liquid changed. Test fluids
were water, monoethylene glycol, propylene glycol and
cyclohexanol. The liquid load was in the range of 0.01–
2.0 m3 m–1 h–1 corresponding to Reynolds numbers of
0.9–1930. Experiments were conducted with organic vapor
on the heating side and overall temperature differences
of up to 40 K. The heat flux density was in the range of
2–12 kW m–2. A total of more than 100 experiments for
evaporating and non-evaporating films were carried out to
Fig. 3 Experimental setup build up a reliable database for laminar and wavy-laminar
viscous single component films. For all data points 3–6
individual experiments were performed. With Pr < 130 the
overall temperature difference DT was varied from 10 to
40 K, see Table 1 for details. For 150 < Pr < 663 the
experimental setup did not allow an evaporation of the film
thus only the fluid dynamic influence of the film distribu-
tion device was studied. Table 2 summarizes the performed
experiments in this range. As indicated by the Reynolds
number the full fluid dynamic range from laminar to wavy-
laminar to turbulent flow was covered in the experiments.
Especially very low Reynolds values were of interest to
identify possible wave development for an expected fully
laminar film pattern.

4 Results

4.1 Fluid dynamics

Following Brauers suggestion the distinction between

different fluid dynamic regimes is based on the Reynolds
number [3]. While it is generally accepted that film flow at
Re > 400 is fully turbulent, transitions from (wavy-)lami-
nar to turbulent flow was found to occur at significantly
smaller Re numbers, see e.g. Ishigai et al. [26]. Not
depending on fluid properties a range from 75 £ Re £ 400
was identified. Therefore this study focuses on Re < 400.
Fig. 4 Film distribution Ishigai et al. [26] state that for

123
1042 Heat Mass Transfer (2009) 45:1037–1046

Table 1 Experiments for evaporating films Eqs. 11 and 12. It is apparent that the conducted experi-
Liquid Pr Visc. Temp. Flow Remin Remax
ments cover the full range of the characteristic fluid dy-
(–) (mPas) (C) (l/h) (–) (–) namic regimes.
Following definitions for the critical Reynolds number
Water 1.75 0.28 99.6 35–280 240 1930 without the influence of the Kapitza factor, Höhne et al.
MEG 12 0.99 145 40–280 84 586 [27] suggested:
PG 29 1.8 130 40–300 43 319 Laminar: Recrit = 4.42
44 3 95 40–300 27 203 Wavy-laminar: Recrit = 19.55–86.43
CHEX 29 1.2 120 40–300 58 432 Turbulent: Recrit = 382.15
40 1.8 105 40–300 41 305
78 3.8 80 40–300 19 145 The movable CCD camera recorded frames and movies
108 5.5 70 40–300 13 101 from the full length of the evaporator tube. Of special
130 6.7 65 40–240 11 66 interest is the entry zone where an immediate effect of the
distribution device was visible. Its objective to shorten the
fluid dynamic start-up zone and to establish a stable
Table 2 Experiments for fluid dynamics only
laminar-wavy, wavy or even turbulent flow at lower Re
Liquid Pr Visc. Temp. Flow Remin Remax numbers than for a standard device was achieved.
(–) (mPas) (C) (l/h) (–) (–)
Figures 6 and 7 show single frames at the maximum
PG 48 3.3 90 40–160 24.4 97.7 viscosities combined with the minimum flow rates (Rey-
107 7.8 60 40–160 10.5 42 nolds numbers) at 25–55 mm below the circular inner
396 31.2 30 10–60 0.7 4 groove, see Fig. 4. Although the Reynolds number indi-
CHEX 29 1.22 120 40–300 57.6 432 cates a fully smooth laminar film it is obvious that the film
40 1.8 105 40–300 40.7 305 distribution configuration significantly enhances wave
78 3.8 80 10–180 4.8 87 initiation.
108 5.5 70 10–180 3.4 60.3 First single waves are visible and propagate to wave
158 8.4 60 10–130 2.2 28.9 fronts along the flow direction. Slowly increasing the flow
241 13.4 50 14–130 2 18.3 rate producing a maximum of Re = 4 the wave evolution
388 22.6 40 20–120 1.7 10.1 was clearly visible as Fig. 7 shows. According to Eqs. 11
663 40.5 30 20–80 0.9 3.8
and 12 the flow should be in the laminar state, especially
with a non-slotted weir or annular gap distribution. By
optical observation and several test recordings the wave
Re  0:47
Ka0:1 ð11Þ pattern was found to be stable along the tube axis at all
films are generally smooth laminar. With respect to fluid conditions. A fluid dynamic start-up zone could not be
dynamics the present study aims at finding identified with the available measurement technique.
Numerical simulations using CFD methods may give ac-
(a) the characteristic conditions at which transition from cess to this resolution [14–16].
laminar smooth to laminar-wavy and from laminar- The optical access to the film allowed for an analysis of
wavy to turbulent flow occurs, the critical Reynolds number at which a significant change
(b) whether Re number alone is a sufficient criterion for of the wave pattern was observed. A direct comparison to
these transitions or whether surface tension effects the heat transfer results was possible then. Figure 8 shows
have to be accounted for through Ka number, and results for the wave peak velocity and the identification of a
(c) which effect nucleate boiling has on these transitions. critical Reynolds number for propylene glycol. The aver-
The Eq. 11 is based on experiments with vertical plane age film velocity was calculated with the Nusselt film
water films performed by Brauer [3]. Only few investiga- thickness. Clearly two flow regimes may be identified.
tions with other fluids than water or aqueous mixtures have Below Recrit  50 a laminar-wavy flow is observed while
been reported. With increasing Reynolds number the for Re > Recrit turbulent flow conditions are found. Results
adjacent flow regimes of unstable and stable laminar wavy for cylohexanol show similar characteristics.
flow can be defined as
4.2 Heat transfer
0:1
2:2
Ka  Re  75; ð12Þ
The second part of the study investigated heat transfer,
see Ishigai et al. [26]. Figure 5 shows the results for the especially the beginning of nucleate boiling and its influ-
Kapitza and Reynolds values for cyclohexanol relating to ence on fluid dynamics. For a consistent nomenclature to

123
Heat Mass Transfer (2009) 45:1037–1046 1043

Fig. 5 Flow regimes for 1E+11

Propylene glycol and °∆ Propylene glycol
Cyclohexanol
Cyclohexanol based on Kapitza 1E+10

and Reynolds number

1E+09
Re=0.47 * Ka0.1

1E+08

1E+07

Ka [-]
1E+06 laminar film flow
Re=75
1E+05

stable laminar possible

1E+04 wavy flow turbulent flow

1E+03
unstable laminar Re=2.2 * Ka0.1
wavy flow
1E+02
0 1 10 100 1000
Re [-]

describe transition from convective to nucleate boiling the • Pr < 50: Nucleate boiling at wave fronts and in the
overall temperature difference DTov was defined as: main (laminar) film possible
• Pr > 50 and DTi = 8.5 K: Irregular bubble formation
DTov ¼ TCT  TBT ð13Þ only at the front end of waves
and • Pr > 50 and 8.5 K < D Ti < 12.5K (DTov = 40 K):
Bubbles at the front area of waves and irregular single
bubbles in the main (laminar) film.
DTi ¼ Tw;i  TBT ð14Þ
Figures 9 and 10 show examples of the film material for
as the relevant mean tube side wall superheat. Due to the two typical cases. In Fig. 9 the bubbles formed at wave
low conductivity of Borosilicate glass with k = 1.2 W m–1 fronts and in the main layer of the film of pure propylene
K–1 a variation of DTov = 10, 20, 30, and 40 K was chosen glycol. Figure 10 shows the shape of a bubble which was
to allow a discussion about the maximum overall temper- initiated at the wave front and followed the wave shape
ature difference before film instabilities occur. Tube side with downwards flow. It should be noted that for DTov ‡
wall superheat DTi is the governing thermal parameter 30 K and Pr < 50 irregular film instabilities occurred
to influence boiling behaviour on the film side. In general especially when different rolling waves interfered — a
it was found that for DTi < 6.5 K, resp. DTov < 20 K massive bubble formed but without a film break-up.
no nucleate boiling was visible with the camera. With A dry spot or dewetting was observed only for DTov =
DTi = 6.5 K (DTov = 20 K) beginning nucleate boiling was 40 K with Pr < 50 due to the fact that a significant amount
observed depending on Re with Pr < 50. For Pr > 50 a of the film volume was evaporated after a length of approx.
transition to nucleate boiling was found for DTi ‡ 8.5 K 1 m and in conclusion the film thickness decreased below
(DTov ‡ 30 K) only. its critical value. A detailed analysis of this phenomenon is
Analysis of the film material showed the following part of a current study and will be reported in the future.
classification for DTi ‡ 8.5 K, resp. DTov ‡ 30 K: A more detailed analysis of heat transfer is presented in
Fig. 11 for Pr = 78. It shows the Nusselt number as

Fig. 6 Cyclohexanol, Re = 0.9, Pr = 663, l = 41 mPas Fig. 7 Cyclohexanol, Re = 3.8, Pr = 663, l = 41 mPas

123
1044 Heat Mass Transfer (2009) 45:1037–1046

Fig. 8 Wave peak velocity for 2,25

Pr = 395
propylene glycol
2,00 Pr = 107
Pr = 48
1,75 Pr = 44
Pr = 35
1,50 Pr = 29

velocity [m/s]
1,25

1,00

0,75
vaverage (Pr 29)
0,50

0,25
vaverage (Pr 395)

0,00
0,1 1 10 100 1000
Re [-]

determined from the experimental heat transfer coefficients analysis with the CCD camera showed that additionally to
as a function of the Reynolds number with the overall single bubbles travelling surface waves in a 2-D film pro-
temperature difference as parameter. For Re < 55 the da- file occurred. This could lead to increased forced convec-
tapoints for all three DTov-values fall together at the tion, which is independent of the wall superheat. An
respective Re = const. Exceeding a Reynolds number of increasing heat transfer coefficient can therefore be
Re = 55 Nusselt numbers for DTov = 30 K deviate signifi- attributed to nucleate film boiling.
cantly from the other two showing a more pronounced It should be noted that no film break-up was observed
increase with Re. This is due to the fact that an additional for test configurations with Pr > 50 on the full length of the
mechanism for heat transfer is active compared to DTov evaporator tube. Depending on the boiling fluid properties
= 10 or 20 K, resp. The optical inspection of the film and its thermal stability an increase of wall superheat for
evaporation revealed the presence of nucleate boiling at this Pr range is possible using the presented liquid distri-
DTov = 30 K and Re > 55. This corresponds with the fluid bution device. For a detailed analysis of its influence
dynamic findings in Fig. 8, which indicate the transition numerical simulation tools will be considered in the future.
from laminar-wavy to wavy or even turbulent pattern. The
film itself did not destabilize or even break-up. Addition-
ally a general deviation for Pr = 78 from the models was 5 Conclusions
obvious. It should be noted that a comparison is only al-
lowed for the state of surface evaporation. For further Evaporation from free falling liquid films builds on a
analysis of these phenomena additional experimental and complex interaction of fluid dynamics with heat and mass
theoretical investigations will be presented in a future transfer. Physical properties of the fluid, design and oper-
study. ation of the equipment determine the overall performance.
To complement the optical results and the measured To enhance heat transfer and to optimize film evaporation
heat flux density of 3.8–9 kW m–2 tube side wall temper- the influence of film generation and fluid dynamics was
atures were calculated for Pr = 78, see Fig. 12. For very investigated in detail.
low Reynolds numbers the wall superheat is significantly
higher and decreases with larger values of Re. Optical

Fig. 9 Propylene glycol, Re = 102, Pr = 40, DTov = 30 K Fig. 10 Cyclohexanol, Re = 67, Pr = 58, DTov = 30 K

123
Heat Mass Transfer (2009) 45:1037–1046 1045

Fig. 11 Experiments and 2,50

∆ Tov=10 K (∆ Ti ≈ 2.5 K)
theory for Nu, cyclohexanol,
2,25 ∆ Tov=20 K (∆ Ti ≈ 6.5 K)
Pr = 78
∆ Tov=30 K (∆ Ti ≈ 8.5 K)
2,00
Chun & Seban nucleate boiling
1,75 Alhusseini et al. (ATC)

1,50

Nu [-]
no influence of Pr
1,25
and Re
1,00

0,75

0,50

0,25 surface
evaporation
0,00
0 20 40 60 80 100 120 140 160
Re [-]

25
∆Tov=10 K
2. The film distribution device was identified to support
∆Tov=20 K the wave evolution and film stability and wave for-
∆Tov=30 K
20 mation/film pattern had an influence on the critical
Reynolds number where nucleate boiling started.
15 3. Fluid properties had an influence on the beginning of
∆ Ti [K]

nucleate boiling.
10 4. Depending on Pr and Re for DTov ‡ 30 K the heat
transfer was increased significantly and beginning
5
nucleate boiling was observed.
5. With elevated Prandtl numbers a deviation between the
0
0 20 40 60 80 100 120 140 160
known correlations to predict Nu was found.
Re [-]
Further investigations are currently performed to explain
Fig. 12 Calculated wall superheat with constant heat flux density the influence of the surface tension and to extend appli-
(3.8–9 kW m–2) for cyclohexanol, Pr = 78 cability of correlations for larger Pr numbers.

The present study reports results for fluid dynamics and References
combined heat transfer for single component systems in a
single tube glass falling film evaporator with a slotted weir 1. VDI-Wärmeatlas (2002) Verein Deutscher Ingenieure, 9th edn.
film distribution. It was possible to shift the transition zone Springer-Verlag, Heidelberg
from laminar to laminar-wavy to film Reynolds numbers 2. Nusselt W (1916) Die Kondensation des Wasserdampfes. Zeit-
schrift des Vereins Deutscher Ingenieure 60(27):541–546
smaller than 4. A laminar-wavy film structure was changed 3. Brauer H (1956) Strömung und Wärmeübergang bei Rieselfil-
into turbulent flow at significantly smaller regions than a men, VDI Forschungsheft 457B, 22
critical Reynolds number reported in literature. A direct 4. Kapitza PL (1960) Wave flow in thin layers of a viscous fluid,
comparison of the test liquids showed diversity in wave Collected Papers of P. L. Kapitza. Macmillan, New York, pp 663
5. Seban RA, Faghri A (1976) Evaporation and heating with tur-
initiation, propagation and structure, so that an additional bulent falling liquid films. J Heat Transfer 98:315
introduction of the Kapitza number, respective the surface 6. Telles AS, Dukler AE (1970) Statistical characteristics of thin,
tension, needs further experimental and theoretical studies. vertical, wavy, liquids films. Ind Eng Chem Fund 9:412
Reynolds and Prandtl number may not be sufficient to 7. Chu KJ, Dukler AE (1975) Statistical characteristics of thin,
wavy films: III. AIChE J 21:583
properly describe film fluid dynamics coupled with heat 8. Wasden FK, Dukler AE (1989) Insights into the hydrodynamics
and mass transfer. of free falling wavy films. AIChE J 35(2):187–195
Based on the experiments with water, monoethylene 9. Jayanti S, Hewitt GF (1997) Hydrodynamics and heat transfer of
glycol, propylene glycol and cyclohexanol the following wavy thin film flow. Int J Heat Mass Transfer 40(1):179–190
10. Leuthner S (1999) Messungen und Modellierungen zum Energie-
conclusions can be stated for the heat transfer: und Stofftransport in Fallfilmen. Dissertation TU Berlin, VDI
Fortschritt-Berichte, Düsseldorf
1. For the experimental set-up it was possible to increase
11. Maun AH (2004) Experimentelle Untersuchungen zum
the overall temperature difference up to 40 K without Wärmeübergang an verdampfenden Fallfilmen binärer Gemische.
causing visible film break-ups with Pr > 50. Dissertation TU Berlin, VDI Fortschritt- Berichte, Düsseldorf

123
1046 Heat Mass Transfer (2009) 45:1037–1046

12. Adomeit P (1996) Experimentelle Untersuchung der Strömung 20. Lehnberger A (2002) Wärmeübergang im Fallfilmverdampfer in
laminar-welliger Rieselfilme. Dissertation RWTH Aachen glatten und profilierten Rohren bei kleinen Temperaturgefällen,
13. Al-Sibai F (2005) Experimentelle Untersuchung der Strömungs- VDI Fortschritt-Berichte, Düsseldorf
charakteristik und des Wärmeübergangs bei welligen Rieselfil- 21. Kedzierski MA, Kim MS (1998) Convective boiling and con-
men. Online-Dissertation RWTH Aachen densation heat transfer with a twisted-tape insert for R12, R22,
14. Alekseenko SV, Nakoryakov VE, Pokusaev BG (1994) Wave R152a, R134a, R290, R32/R134a, R32/R152a, R290/R134a,
flow of liquid films. Begell House, New York R134a/R600a. Thermal Sci Eng 6(1):113–122
15. Miyara A (2003) Numerical simulation of heat and mass transfer 22. Wilke W (1962) Wärmeübergang an Rieselfilme, VDI-Forsch.-
of wavy falling films. In: Conference on transport phenomena with Heft 490, Düsseldorf
moving boundaries, Berlin, Germany, 09th–10th October 2003 23. Chun KR, Seban RA (1971) Heat transfer to evaporating liquid
16. Heidrich H (2005) Numerische Berechnung des Wärme- und films. J Heat Transfer 93:391–396
Stoffübergangs an welligen Fallfilmen binärer Gemische, Online- 24. Palen JW, Wang Q, Chen JC (1994) Falling film evaporation of
Dissertation TU Berlin binary mixtures. AIChE J 40(2):207–214
17. Ljubicic B (1999) Testing of twisted-tube exchangers transition 25. Alhusseini A, Tuzla K, Chen JC (1998) Falling film evaporation
flow regime. In: International conference on compact heat of single component liquids. Int J Heat Mass Transfer
exchangers and enhancement technology for the process indus- 41(12):1623–1632
tries, July 18–23, 1999, Banff Centre for Conferences, Banff 26. Ishigai S, Nakanisi S, Koizumi T, Oyabi Z (1972) Hydrody-
18. Nakayama W, Daikoku T, Kuwara H, Nakajima T (1979) namics and heat transfer of vertical falling liquid films. Bull
Dynamic model of enhanced boiling heat transfer on porous JSME 15(83):594–602
surfaces. In: Advances in enhanced heat transfer. ASME, New 27. Höhne J, Fratzscher W, Würfel R, Allan F (1997) Beschreibung
York, pp 31–43 einer Filmströmung durch Entropiebewertung, Chem. Technik
19. Schnabel G (1982) Analysis of local and overall heat transfer 49(3):109–112
enhancement on double-fluted condenser/evaporator tubes. Chem
Eng Fundamentals 1(2):64–78

123