Académique Documents
Professionnel Documents
Culture Documents
H I G H L I G H T S
art ic l e i nf o
a b s t r a c t
Article history:
Received 6 January 2014
Received in revised form
8 April 2014
Accepted 12 April 2014
Available online 21 April 2014
Falling-lm evaporators, which are used for concentrating dairy liquids and juice, often include multipass bodies (calandrias). The ow of vapours through multi-pass bodies was modelled and simulated.
Equations were developed to describe the pressure drop of vapour as the ow rate changes with
evaporation. It was shown that, even though all the vapour leaves at the base of the evaporator, some of
the vapour ow upwards in some tubes and downwards in other tubes. The vapour velocities across the
tube sheet of a ve-pass evaporator concentrating milk from 16% solids to 39% solids were simulated to
be up to 25 m s 1. The vapour ow that resulted was likely to cause signicant deection of incoming
liquid jets with the potential increase fouling. Alternative designs were evaluated.
& 2014 Elsevier Ltd. All rights reserved.
Keywords:
Falling-lm evaporator
Distributor
Vapour ow
Pressure drop
1. Introduction
Falling lm evaporators are commonly used in the food
industry for the concentration of liquids such as milk, whey, sugar
and juice (Morison and Hartel, 2007). In nearly all such evaporators, heated liquid enters at the top of the evaporator body
(calandria) and ows down the inside of many vertical tubes
which are externally heated by steam (water vapour). Water is
evaporated from the liquid as it passes down, thus concentrating
the solids content of the liquid. The vapour and concentrated
liquid are separated at the bottom of the tubes. Signicant energy
savings can be made by having several stages in series whereby
the vapour produced by one stage is used to heat a subsequent
stage which operates at a lower temperature and pressure. Each of
these stages with a different pressure is termed an effect, and
n
Corresponding author.
E-mail addresses: ken.morison@canterbury.ac.nz (K.R. Morison),
sbroome@transeldworley.co.nz (S.R. Broome).
1
Present address: WorleyParsons, 25 Gill Street, New Plymouth, New Zealand.
http://dx.doi.org/10.1016/j.ces.2014.04.015
0009-2509/& 2014 Elsevier Ltd. All rights reserved.
2. Model
It is useful to start with an outline of the cause for upward
vapour ows. In some multi-pass evaporators the top part of all
the passes is at the same pressure, and all the passes have the
same pressure at the base, so the pressure drop down the tubes is
_ evap ) produced in
the same for all passes. The ow rate of vapour (m
each tube is directly related to the heat transfer coefcient, U, (Eq.
Vapour
Feed
UA T
hv
Fan
Vapour
Liquid
_ f eed
wf eed m
_s
m
Vapour
Liquid
Fig. 2. Cross-section of two possible distributor designs: a) without vapour tubes and b) with vapour tubes.
of the wall, k:
1
A
t 1
i
U he Ae k hi
l l v g
2l
"
he
0:00402kl Re0:4
l
Pr 0:65
l
#1=3
;
l l v g
2l
Rel 4 30
#1=3
;
4
l
_l
m
D
10
It was found that the Reynolds number remained low enough that
Eq. (8) was never used.
The stainless steel tubes typically have an outside diameter of
50.8 mm (2 in.), a wall thickness of 1 mm and a thermal conductivity of 17 W m 2 K 1. The local internal lm heat transfer
coefcient was chosen based on indicative values for overall heat
transfer coefcients given in Morison and Hartel (2007) to be
hi 6200 11400w
11
When combined with the other terms, the overall heat transfer coefcient ranged from 370 W m 2 K 1 at 50% solids to
2750 W m 2 K 1 for water.
Liquid typically enters each effect at a temperature higher than
that of the effect. Some amount of liquid ashes to vapour
according to Eq. (9):
_ f lash
m
12
The net ow of vapour down each tube must equal the rate of
ashed vapour. This can be expressed as
i 1;npass
_ v;i m
_ f lash
ni m
13
14
dx
dx sh
dx a
dx f
dP
dx
v g
sh
15
dx f
D
where f is the friction factor and vv is the specic volume of the
vapour. The absolute value of mass ux is used to allow for
reverse ows.
Hajiloo et al. (2001) reviewed friction factors for such ows but
their gas and liquid ow rates were mostly higher than those
expected in an evaporator. Their data were inspected and extrapolated to the conditions in the tubes, but in all cases the
experimental friction factors were less than or very close to the
friction factor for a smooth tube. Wallis (1969) collected a range of
annular lm ow results for horizontal and vertical lms and
obtained an equation for friction factor:
f 0:005 1 300
17
D
From this work they concluded that the equivalent pipe roughness
was four times the lm thickness (for water/air). However for
more viscous liquids the lm is thicker and the friction predicted
by Eq. (17) becomes unreasonably high. Hence Haalands equation
for friction factors (Haaland, 1983) was applied using a uniform
roughness () of 1.1 mm corresponding to the waves formed in a
water/air system. Subsequent sensitivity analysis showed that the
vapour velocities were not at all sensitive to the choice of roughness:
2
1:11 6:9
f 1= 3:6log 10
3:7D
Re
18
dx
dP dx
21
The vapour mass ux, G, changes with evaporation down the tube:
_v
dG 1 dm
dx A dx
22
dx D2 hv
D hv
Hence
4UT
dP
dvv dP
signG 2Gvv
G2
dx a
Dhv
dP dx
Combining all the terms
!
4UT
dP
g
dvv dP 2f G2 vv
signG 2Gvv
G2
Dhv
dx
vv
dP dx
D
23
24
25
dP
dvv
g
1 signGG2
2signG
dx
vv
dP
!
4UT f G2 v
v
Gvv
Dhv
D
Table 1
Parameters used in the simulation.
26
dP g=vv 2signGjGjvv 4UT=Dhv f G2 vv =D
dx
1 signGG2 dvv =dP
27
28
1=
2
milk
967 1:3T 1635 2:6T 0:02T
water T
Here wf at ; wNFS and wwater are the mass fraction of fat, non-fatsolids (NFS) and water respectively, and T is temperature in 1C.
The viscosity was calculated using the equations given by
Morison et al. (2013) as follows:
milk waterT exp
i
ai wi
wwater
29
where the sum is over components protein, lactose and fat with
alactose 3:35 0:0238T 1:25 10 4 T 2
30
31
RT 2wb ln xwater
hv
32
Parameter
Effect 1
Effect 2
9.0
2
0.18
650
457
16.1
5
0.20
327
247
190
153
127
66
3.5
70
70
3.5
75
the resulting number of tubes in each pass was not always the best
for physical construction. Other parameters are given in Table 1.
2.4. Distributor
When the distributor shown in Fig. 2a is used, the vapour ows
to the circumference of the distributor plate. The arc length (at the
circumference) available for ow in each pass was assumed to be
in direct proportion to the number of tubes, but it is acknowledged
that this does not need to be so. The distributor was assumed to be
40 mm above the tube plate. The design in Fig. 2b allows vapour to
pass directly from evaporator tubes into corresponding vapour
tubes (or vice versa) so vapour does not need to ow horizontally.
Fig. 3a shows the top view of a small tube sheet and Fig. 3b
shows a photo of a similar system. For the tube diameters and
pitch given above, the minimum distance between the centre of a
distributor hole and the inner edge of an evaporator tube is
10.2 mm. Typically distributor holes and jets are 68 mm in
diameter so the distance between the outer edge of the jet and
in tube is 67 mm.
With the distributor design shown in Fig. 2a there is possibility
of interaction between the falling liquid jet and the vapour as
shown for an individual jet in Fig. 4. If the deection of the falling
jet is high enough, the distribution of liquid into the tubes and the
subsequent wetting of the tubes will be signicantly affected. The
horizontal vapour velocity is assumed to be the greatest after the
outer tubes near the circumference of the distributor.
By considering drag over a cylinder it can be shown that the
horizontal liquid velocity ul;y of a piece of falling uid is given by
dul;y 2C d v u2v
dt
l D
34
p
2gh gt
35
Eqs. (34) and (35) can be used to nd the trajectory of an ideal jet
of falling liquid.
Fig. 3. a) A scale top view of 7 evaporator tubes (large circles) and position of vertical jets from the distributor (small circles). b) Vertical jets of sucrose solution falling from a
distributor onto a model tube sheet.
-2.5
Liquid
Vapour
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
Fig. 4. Side view of a distorted jet of liquid falling from a distributor plate onto a
tube sheet with a cross ow of vapour.
3
Pass number
Fig. 6. Vapour velocity at the top and base of a single tube in each of ve passes in
Effect 2. The grey area indicates the amount of evaporation in each pass. The black
areas represent ow from passes 1 and 2.
ow rate into the top of each pass was guessed. The pressure was
then calculated from the base of the tubes to the top also using
Modied Euler's method (in the style of Holland and Bragg, 1995).
Excel Solver was then used to set all the pressures at the top of the
passes equal while requiring the total of the vapour ows into the
passes to equal the amount of vapour ashed.
-0.25
0.00
0.25
0.50
0.75
1.00
1.25
3. Results
1.50
1.75
2.00
2.25
Pass number
Fig. 5. Simulated total vapour ow rates at the top and base of each of ve passes
in Effect 2. The y-axis is inverted to show the physical direction of ow. The grey
area indicates the amount of evaporation in each pass. The black areas represent
ow from passes 1 and 2.
14.0
12.0
10.0
8.0
6.0
4.0
2.0
0.0
-2.0
-4.0
0
0.05
0.1
0.15
0.2
0.25
30.0
25.0
20.0
15.0
10.0
5.0
0
5
10
Vertical distance, mm
15
20
25
30
35
40
-10
-5
0
5
10
15
Horizontal distance, mm
Fig. 9. The calculated deection of a jet of milk falling from a distributor hole onto
a tube sheet with a vapour velocity of 25.6 m s 1. The width of the curve indicates
approximate width of the jet.
0.0
-5.0
-10.0
-15.0
-20.0
0
0.1
0.2
0.3
For the rst effect, with parameters as given in Table 1 but with
a feed temperature ranging from 70 1C to 80 1C (010 1C ash), the
worst case for pass 1 was an upward vapour ow rate from the
pass of 0.48 m3/s when there was no ash, and for pass 2 the
worst case occurred at 10 1C ash with a downward vapour ow
rate of 1.1 m3/s. To obtain the required pressure drop was found
that in. outside diameter tubing (1.5 mm wall thickness) could
be used for most of the passes; hence this diameter was chosen for
nearly all the passes. In pass 1 of Effect 1 with one vapour tube for
Table 2
Velocity and pressure drops for vapour tubes.
Effect Pass Tube
ID
(mm)
Vapour tube
Evap:
Total vapour
vapour tube ow rateb (m3/ velocity
(m s 1)
ratioa
s)
1
1
2
2
2
2
2
3
3
1
3
3
1
11
1
2
1
2
3
4
5
16.1
22.4
16.1
16.1
16.1
16.1
22.4
0.49c
1.1d
1.0
0.27
0.24
00.59
0.83
11.7c
18.4d
15.5
16.4
19.0
19.2
16.6
Pressure
drop (Pa)
21
55
34
38
51
51
38
4. Discussion
The simulation conrmed and quantied the direction of
vapour ows in the evaporator bodies. The results are not
expected to be particularly sensitive to any particular parameters.
While the pressure drop calculation was effective, the extent of
vapour ows is dependent on the relative pressure drops of all the
passes, and not on the absolute pressure drops, so this calculation
is not critical. Linear scaling of the pressure drop has no effect on
the vapour ows. The ow of vapour from one pass to another was
greater in the ve-pass calandria because this one had a large
range of solids content and hence a larger range of evaporation in
different passes. The velocities predicted are high enough to cause
concern that they might disturb liquid ows.
Experimental validation of the model results is desirable, but it
would be very difcult to measure vapour velocities in an industrial
evaporator while maintaining vacuum and without signicant unacceptable modications to the structure. Thus validation relies on the
authors observation of upward ows noted above.
It was difcult to nd minor variations in the design of the
distributor in Fig. 2a that could remove the vapour velocities. The
gap between the distributor plate and the tube sheet was varied
but had very little impact on the jet deection; a greater gap
reduced the vapour velocities but increased the interaction
between the vapour and liquid. Upward vapour ow would be
substantially reduced if the ve-pass effect was split into two
calandrias, but this would add to the capital costs of the evaporator. Alternatively the incorporation of vapour tubes as shown in
Fig. 2b enabled the elimination of high vapour velocities.
This analysis has assumed that all passes have a common pressure
at the top of the calandria. If each pass was sealed at the top, vapour
could not ow from one pass to another and this problem would be
avoided. Such a design is complicated by the need to be able to open
and inspect the top of the calandria, while ensuring that all parts can
be cleaned by normal clean-in-place procedures.
5. Conclusions
The analysis presented here used a model to simulate local heat
transfer, vapour ows and pressure drops in falling-lm evaporators
with multiple passes. When the solids content of the liquid gets
higher, reduced heat transfer causes less evaporation and vapour
ows. The result of this is signicant upward ows of vapour in the
rst pass and increased downward ows in the last pass. The extent of
this phenomenon was found to be greater in a ve-pass effect because
of the wide range in solids concentrations and hence heat transfer
coefcients. For a distributor plate without vapour tubes, the vapour
velocities are sufcient to deect the liquid jets falling from the
distributor plate which is likely to cause deterioration in heat transfer.
When vapour tubes are installed in the distributor plate, there is little
interaction between vapour and liquid ows but the tubes must be
large enough to handle the required vapour ow rates with minimal
pressure drop.
While the designs of industrial evaporators continue to
develop, there are very many installed evaporators with distributors that might cause unnecessary interactions between vapour
and liquid ows. A retrot could be considered for these if there
are performance problems of the type discussed here.
Nomenclature
a
A
Cd
Ch
Cp
D
f
G
h
h
hv
k
K
l
_
m
n
P
Pr
Re
t
t
T
u
U
v
w
x
x
viscosity coefcient
heat transfer area (m2)
drag coefcient
hole discharge coefcient
specic heat capacity (J kg 1K 1)
tube inside diameter (m)
Fanning friction factor
mass ux (kg s 1 m 2)
lm heat transfer coefcient (W m 2K 1)
liquid height (m)
heat of vaporisation (J kg 1)
thermal conductivity (W m 1 K 1)
friction loss coefcient
length (m)
mass ow rate (kg s1)
number of tubes
pressure (Pa)
Prandtl number
Reynolds number
wall thickness (m)
time (s)
temperature (1C or K)
velocity (m s 1)
overall heat transfer coefcient (W m 2 K 1)
specic volume (m3 kg 1)
mass fraction
distance down a tube (m)
mole fraction
Greek letters
lm thickness (m)
roughness (m)
wetting rate (kg m 1 s 1)
viscosity (Pa s)
density (kg m 3)
Subscripts
a
b
e
f
acceleration
boiling point elevation
external
friction
i
l
P
s
sh
v
x, y
internal
liquid
pressure
solids
static head
vapour
direction
References
Adib, T.F., Vasseur, J., 2008. Bibliographic analysis of predicting heat transfer
coefcients in boiling for applications in designing liquid food evaporators.
J. Food Eng. 87, 149161.
APV, 1999. Evaporator Handbook4th ed. APV Nordic, Anhydro, Denmark.
Billet, R, 1989. Evaporation Technology: Principles, Applications and EconomicsVCH
publishers, Weinheim, Germany.
Bouman, S., Brinkman, D.W., de Jong, P., Waalewijn, R., 1988. Multistage evaporation in the dairy industry: energy saving, product losses and cleaning. In: Bruin,
S. (Ed.), Preconcentration and Drying of Food MaterialsElsevier Science,
Amsterdam.
Chen, P, Chen, X.D., Free, K.W., 1996. Measurement and data interpretation of the
freezing point depression of milks. J. Food Eng. 30, 239253.
GEA, 2003. Evaporation Technology Using Mechanical Vapour RecompressionGEA
Process Engineering Inc., Columbia, MD, USA.
GEA, 2000. Falling Film Evaporators for the Dairy IndustryNiro France S.A.S,
Montigny le Bretonneux, France.
Haaland, S.E., 1983. Simple and explicit formulas for the friction factorJ. Hydraulics
Div.: ASCE 10289
Hajiloo, M., Chang, B.H., Mills, A.F., 2001. Interfacial shear in downward two-phase
annular co-current ow. Int. J. Multiph. Flow 27, 10951108.
Hewitt, G.F., Hall-Taylor, N.S., 1970. Annular Two-Phase Flow. Pergamon Press,
Oxford.
Holland, F.A, Bragg, R., 1995. Fluid Flow for Chemical Engineers2nd ed. Edward
Arnold, London.
Lapple, C.E., Shepherd, C.B., 1940. Calculation of particle trajectories. Ind. Eng.
Chem. 32 (605621), 1940.
McNaught, J.M., Butterworth, D., 1998. Film condensation of pure vapour, Section
2.6.2. In: Hewitt, G.F. (Ed.), Heat Exchanger Design Handbook. Begell House Inc.,
Redding, CT, USA, p. 1998.
Morison, K.R., Hartel, R.W., 2007. Evaporation and freeze drying. In: Heldman, D.R.,
Lund, D.B. (Eds.), Handbook of Food Engineering, 2nd ed CRC Press, Boca Raton,
pp. 495552.
Morison, K.R., Phelan, J.P., Bloore, C.G., 2013. Viscosity and non-Newtonian behaviour of concentrated milk and cream. Int. J. Food Prop. 16, 882894.
Morison, K.R., Worth, Q.A.G., ODea, N.P., 2006. Minimum wetting and distribution
rates in falling lm evaporators. Food Bioprod. Process. 84, 302310.
Paramalingam, S., Winchester, J., Marsh, C., 2000. On the fouling of falling lm
evaporators due to lm break-up. Trans. Inst. Chem. Eng. 78C, 7984.
Pseck, J., 1997. Handbook of Milk Powder Manufacture. Niro A/S, Copenhagen,
Denmark.
Sharma, S., Rangaiah, G.P., Cheah, K.S., 2012. Multi-objective optimization using MS
Excel with an application to design of a falling-lm evaporator system. Food
Bioprod. Process. 90, 123134.
Wallis, G.B, 1969. One-dimensional Two-phase Flow. McGraw-Hill Book Co., New
York.
Westergaard, V., 2004. Milk Powder Technology: Evaporation and Spray Drying.
Niro A/S, Copenhagen, Denmark.