You are on page 1of 22

Article No : o08_o01

Distillation, 2. Equipment
JOHANN STICHLMAIR, Technische Universit€at M€unchen, Garching, Germany

1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . 456 3.1. Design Principles . . . . . . . . . . . . . . . . . . . . . 466


2. Tray Columns . . . . . . . . . . . . . . . . . . . . . . . 456 3.2. Operation Region of Packed Columns . . . . . 467
2.1. Design Principles . . . . . . . . . . . . . . . . . . . . . 456 3.3. Two-Phase Flow in Packed Columns . . . . . . 470
2.2. Operation Region of Tray Columns. . . . . . . 457 3.4. Mass Transfer in Packed Columns . . . . . . . 472
2.3. Two-Phase Flow on Trays . . . . . . . . . . . . . . 460 References . . . . . . . . . . . . . . . . . . . . . . . . . . 475
2.4. Mass Transfer in the Two-Phase Layer . . . . 464
3. Packed Columns . . . . . . . . . . . . . . . . . . . . . 466

Symbols Greek symbols


a: volumetric surface, m2 =m3 a: orifice discharge coefficient
A: area, m2 b: mass transfer coefficient, m/s
CG : gas capacity factor, m/s e: relative gas content, porosity
d: diameter, m p: circular constant
D: diameter, m; diffusion coefficient, m2/s r: density, kg/m3
DE : dispersion coefficient, m2/s Dr: density difference, kg/m3
pffiffiffiffiffiffi
F: F-factor, Pa1=2 , F  uG  rG s: surface tension, kg/s2
g: acceleration of gravity, m/s2 t: time, residence time, s
_
G: molar gas flow, kmol/s j: relative free area
h: height, m h: viscosity, kg m1 s1
hL : liquid hold-up x; z: orifice coefficient, friction factor
H: height, plate spacing, m
lw : weir length, m Subscripts
L_: molar liquid flow, kmol/s
m: slope of equilibrium line ac: active
^
M: molar mass, kg/kmol c: column
N: transfer units C: condenser (cooler)
N_ : molar flow, kmol/s cl: clearance under downcomer (skirt
p: pressure, bar clearance)
Dp: pressure drop, bar d: downcomer
T: temperature,  C f: froth, froth layer
u: superficial velocity, m/s F: feed
V_ : volume flow, m3/s h: hole, hydraulic
x: molar fraction of liquid G: gas, vapor
y: molar fraction of gas irr: irrigated

 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


DOI: 10.1002/14356007.o08_o01
456 Distillation, 2. Equipment Vol. 11

L: liquid OG: overall gas


max: maximum OL: overall liquid
min: minimum p: particle
n: number of stage V: valve
N: nominal w: weir
o: single, initial

1. Introduction ment. In this step all geometrical structures and


dimensions of the equipment (type and dimen-
Distillation isbyfarthemostimportant technology sions of columns and internals) have to be deter-
for fractionating fluid mixtures. This article pre- mined. This includes the calculation of the
sents the present state of the art on design and diameter and the height of the column (i.e., the
dimensioning of mass transfer equipment. In the number of actual trays or the height of the pack-
focus are tray columns and packed columns. The ing) as well as the dimensioning of reboiler and
design principles and the feasible operating re- condenser.
gions of both types of columns are presented. Two-phase flow and mass transfer in distilla-
Separate sections concern the two-phase flow tion columns are rather complex mechanisms
withinthecolumnandtheinterfacialmasstransfer. that cannot be modeled from first principles
Distillation units typically consist of a separa- alone. At present, models for column dimension-
tion column, in which the interfacial mass transfer ing are – to a large extent – based on empirical
takes place, a reboiler at the bottom, and a con- know how developed over decades of industrial
denser at the top of the column for providing a application. Most know how is well described in
countercurrent flow of gas and liquid within the the literature [1–6]. Additionally, there is a huge
column. Normally, heat is supplied by steam via a number of research reports from universities.
thermosyphon reboiler and removed by cooling Important contributions to the state of the art
water via a tube heat exchanger. The requirements originate from industry funded research institutes
on effective mass transfer equipments can be un- as Fractionation Research Inc. (FRI) in Still-
derstood from the basic equation of mass transfer: water, OK, and Separation Research Program
(SRP) of the University of Texas at Austin. In
N_ ¼ Akðy yÞ ð1Þ
recent years some computer software for column
To achieve a high mass transfer rate N_each factor dimensioning is also available [7–9].
in Equation (1) should be large. The driving force
ðy yÞ is high at a countercurrent flow of gas and
liquid. A large interfacial area A is provided by 2. Tray Columns
column internals like trays or packings. The mass
transfer coefficient k has high values if the inter- A perspective view of a tray column is shown in
facial area is steadily renewed. All these require- Figure 1. The gas flows upwards and the liquid
ments are well met by tray columns and by flows downwards in the column. On the trays itself
packed columns. there exists a cross flow of the two phases. The
At the present state of the art dimensioning of horizontal trays have openings (holes, caps,
distillation columns is performed in two steps. The valves) for allowing the gas flow through the plates.
first step is the thermodynamic simulation of the
processes taking place in the column (! Distil-
lation, 1. Fundamentals). The results of this rigor- 2.1. Design Principles
ous thermodynamic simulation are the flow rates
of gas and liquid within the column, the energy Three principles for the design of the openings in
demand, the number of equilibrium stages, the the plates can be distinguished (Fig. 2):
internal profiles of concentrations and tempera-
ture, and the product qualities. These data are the . Sieve trays have a large number of round holes
basis for the second step, the sizing of the equip- in the plate (6–13 mm in diameter) with an
Vol. 11 Distillation, 2. Equipment 457

the tray on its way to the outlet weir (3–8 cm


high) and the downcomer. Through a side open-
ing at the lower end of the downcomer the liquid
is fed to the tray beneath. The downcomers are
alternately arranged at the left-hand and right-
hand side of the trays. Typically, tray spacing is
in the range of 0.3–0.6 m. The active area on the
trays is about 80 % of column cross-section. In
large diameter columns special trays are often
used with several downcomers evenly distribut-
ed over the active area (multiple downcomer
trays). In systems with high fouling risk trays
without downcomers (dual-flow trays) are re-
commended. Some characteristic design para-
meters of trays are listed in Table 1.
The gas load, expressed as F-factor, is typi-
cally in the range of 1–2 Pa1/2. The liquid load V_L ,
referred to weir length lw , can be varied from ca.
2–100 m3m1h1. The pressure drop of tray
columns is rather high, typically 7 mbar per
equilibrium stage. Thus, tray columns are not
very well suited for vacuum services.

2.2. Operation Region of Tray


Columns

Tray columns can only be operated within certain


limits of gas and liquid flow as is shown in Figure
Figure 1. Perspective view of a tray column a) Downcomer;
b) Tray support; c) Sieve trays; d) Man way; e) Outlet weir;
3. The upper limits (bold lines) are strong limits
f) Inlet weir; g) Side wall of downcomer; h) Liquid seal which can never be crossed without causing
mechanical damages. The lower limits (dashed
lines) are soft limits which can be crossed if some
losses of mass transfer efficiency are accepted.
open area of about 8–15 % of the active area. The operation point should allow for a sufficient
The weeping of liquid is prevented by the safety margin to the operation limits.
upward flow of gas through the holes
. In bubble cap trays weeping of liquid is pre- Maximum Gas Load. The gas load in the
vented by design. A chimney with a cap covers column is a very important quantity since it
the openings in the plate. The bubble-caps have governs the diameter of the column. By conven-
a diameter of 50–80 mm. The relative free area tion, the gas load in columns is expressed by the
is about 10–20 %. The pressure drop of bub- F-factor according to the definition:
ble-cap trays is rather high
. Valve trays have movable elements in the tray pffiffiffiffiffiffi
F  uG  rG ð2Þ
openings for adjusting the free area to the
actual gas load. This widens the operating
range and reduces the pressure drop at high The term uG denotes the superficial gas velocity
gas loads referred to the active area Aac of the tray. i.e., the
cross-sectional area of the column minus the area
The two-phase layer on the trays is approxi- of two downcomers (Fig. 2). The gas load has to
mately 10–30 cm high and has a large volumetric be kept between a maximum and a minimum
interface of up to 1000 m2/m3. The liquid crosses value.
458 Distillation, 2. Equipment Vol. 11

Figure 2. Schematic of a tray column showing the most important design parameters A) Vertical section
dcap ¼ cap diameter; dV ¼ valve diameter; dh ¼ hole diameter; hcl ¼ height of skirt clearance of downcomer; hw ¼ weir
height; H ¼ tray spacing; V_L ¼ volumetric flow rate of liquid
B) Cross section a) Small holes with narrow spacing; b) Bubble caps with wide spacing
Ad ¼ cross-sectional area of downcomer; Aac ¼ active cross-sectional area; Dc ¼ column diameter; lL ¼ length of liquid
path; lw ¼ weir length

Table 1. Characteristic dimensions of industrial tray designs

Vacuum Atmospheric High


pressure pressure

Tray spacing, m 0.4–0.6 0.4–0.6 0.3–0.4


Weir length, m (0.5–0.6)  Dc (0.6–0.7)  Dc (0.6–0.8)  Dc
Weir height hw , 0.02–0.03 0.03–0.07 0.04–0.1
m
Skirt clearance 0:7hw 0:8hw 0:9hw
Bubble-cap 0.08–0.15 0.08–0.15 0.08–0.15
diameter, m
Bubble-cap 1:25dcap ð1:25--0:14Þdcap 1:5dcap
spacing, m
Valve diameter 0.04–0.05 0.04–0.05 0.04–0.05
dV , m
Valve spacing, m 1:5dV 1:5dV 1:5dV
Hole diameter 0.004–0.013 0.004–0.013 0.004–0.013
Figure 3. Operation region of tray columns
dh , m
Operating boundaries due to: a) Entrainment; b) Froth height;
Hole spacing (2.5–3)  dh (3–4)  dh (3.5–4.5)  dh
c) Downcomer capacity; d) Weeping; e) Minimum crest over
(triang.), m
weir
Aac ¼ active tray area; V_G ¼ volumetric flow rate of vapor;
Relative free 15–10 10–6 7.5–4.5
area j, %
V_L ¼ volumetric flow rate of liquid
Vol. 11 Distillation, 2. Equipment 459

The maximum gas load is often determined


from the equation of Souders and Brown (1934):
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
F ¼ CG  rL rG ð3Þ

The gas load factor CG , defined by Equation (3),


has to be determined from experimental data or
from empirical correlations of such data [1, 3, 5].
Well known and often used is the correlation
of Fair [10] originally developed for sieve trays
(Fig 4).
A theoretically better founded approach has Figure 4. Correlation of the maximum gas load according to
been developed by Stichlmair [4, 11]. A balance Fair [10]
Plate spacing H: a) 0.91 m; b) 0.61 m; c) 0.46 m; d) 0.30 m;
of weight, buoyancy, and drag forces on a single e) 0.23 m; f) 0.15 m
0
droplet whose size is determined by a critical Values for CG are shown for weir height hw  0.15 H, hole
Weber number yields: diameter dh  6.35 mm, and liquid surface tension ¼ 0.02
0
N/m. The capacity factor is calculated from the chart value CG
0  
 1=4 s 0:2
with the formula: CG ¼ CG 0:02 .
Fmax ¼ 2:5 j2 sðrL rG Þg or ð4Þ Corrections must be made for the relative free hole area j as
 1=4 follows:
0
g j  0.10, CG ¼ CG
CG max ¼ 2:5 j2 s j ¼ 0.08, CG ¼ 0:9CG
0
rL rG 0
j ¼ 0.06, CG ¼ 0:8CG

According to these equations the maximum fea-


sible gas load depends on system properties minimum gas load in the holes, Fh min ¼ Fmin =j,
(density r of gas and liquid, surface tension s) decreases with increasing hole diameter dh . At
as well as on tray design (relative free area j). large hole diameters, however, the minimum gas
Equation (4) describes the entrainment flooding load Fmin =j increases with increasing hole
of trays with very large tray spacing. For smaller diameter since the condition of no-weeping is
tray spacing the froth height on the tray sets a decisive in this case.
lower limitation (see Section 2.3).
Maximum Liquid Load. The liquid flow
Minimum Gas Load. A limitation of the downwards the downcomers is enforced by grav-
minimum gas load is set either by weeping of
liquid through the openings in the tray or by non-
uniform gas flow through those openings. The
condition for uniform gas flow through the holes
of a sieve tray is according to MERSMANN [12]:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Fmin ¼ j 2s=dh ð5Þ

According to Ruff [13], weeping of liquid


through the holes of a sieve tray is prevented if
the gas load is higher than:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1=4
Fmin ¼ j ð0:37dh gðrL rG Þ5=4 Þ=rG ð6Þ

Equations (5) and (6) are determined for sieve Figure 5. Minimum gas load of sieve trays for three different
trays operated with three different systems as systems a) Air–water; b) Benzene–toluene; c) Isobutane–n-
shown in Figure 5. At low hole diameters the butane
460 Distillation, 2. Equipment Vol. 11

ity forces which results in a limitation of the distribution of liquid. Accordingly, it is recom-
maximum liquid load. In practice, four empirical mended to ensure a minimum liquid-flow
rules of thumb are applied to the determination of rate over the exit weir larger than
maximum liquid load: V_L =lw  2 m3 m1 h1 . In small diameter col-
umns, however, the liquid load can be consider-
. The relative weir load V_L =lw should be lower ably lower.
than 60 m3m1h1.
. The liquid velocity in the downcomer should
not exceed the value 0.1 to 0.2 m/s. 2.3. Two-Phase Flow on Trays
. The volume of the downcomer should permit a
residence time of the liquid of more than 5 s. Decisive for the performance of tray columns is
. The height of the clear liquid in the downcomer the behavior of the two-phase layer on the trays.
should not exceed half of the tray spacing H. Three different structures of the two-phase layer
can be distinguished:
These rules have been developed by long term
practical experience. However, they do not per- . Bubble regime: The liquid constitutes the con-
mit any insight into the limiting physical phe- tinuous phase. The gas rises in the liquid in
nomena. In essence, liquid flow through a down- form of discrete bubbles
comer is comparable with the flow of liquid out of . Drop regime: The gas forms the continuous
a vessel, which is described by Torricelli’s equa- phase and the liquid is dispersed into fine
tion: droplets
. Froth regime: This regime represents the in-
pffiffiffiffiffiffiffiffiffiffiffiffi
uLd ¼ a 2gH ð7Þ termediate state between bubble and drop
regime. The two-phase layer is intensively
agitated and no univocally dispersed phase
This basic equation has to be adapted to the exists
conditions present at the outlet area of a down-
comer. According to Stichlmair [4, 11] the fol- The froth regime is the dominant regime at
lowing relationship holds for a standard tray normal column operation. Only in vacuum ser-
design: vices drop regime may sometimes exist on trays.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
V_L max r r hDp þhL Relative Liquid Hold-Up. Relative liquid
¼ aeLd hcl  2gH L G  1 ð8Þ
lw rL eLd H hold-up eL is a very important parameter of the
two-phase layer on a tray. It is defined as the ratio
of clear liquid height hL and froth height hf :
Where eLd denotes the relative liquid hold-up in
the downcomer, hcl the height of the downcomer hL
eL ¼ ð9Þ
clearance, hDp the pressure loss of a tray ex- hf
pressed in clear liquid height, and hL the clear
liquid height on the tray (see Section 2.3). The There exists a large data base of relative liquid
orifice discharge coefficient a has, in most cases, hold-up data and many empirical correlations in
a constant value of 0.61. For non-foaming sys- literature. Most published data are well summa-
tems, a value of 0.4 is recommended for the rized in the following correlation [4, 11]:
relative liquid hold-up eLd in the downcomer.
eL ¼ 1ðF=Fmax Þ0:28 ð10Þ

Minimum Liquid Load. In principle, a col-


umn tray can be operated even with very small The term Fmax (see Eq. (4)) accounts for gas load,
liquid loads because the necessary height of the several properties of the system and tray design.
two-phase layer (froth) on the tray is provided by Liquid hold-up decreases significantly with in-
the exit weir. At extremely low liquid loads, creasing gas load. Its dependency on liquid load
however, the liquid will flow in an uneven pattern is very small. In most cases the relative liquid
across the tray resulting in some degree of mal- hold-up eL on a tray is as low as 5–10 %.
Vol. 11 Distillation, 2. Equipment 461

Froth height. Knowledge of the height hf of


the two-phase layer on a tray is very important
because it has to be significantly lower than tray
spacing to avoid excessive liquid entrainment.
The decisive mechanism for froth height is the
liquid flow over the weir at the end of the tray.
Thus, froth height cannot fall below weir height
hw . The flow of the two-phase mixture ðV_L =eL Þ
over the outlet weir length (lw ) additionally
contributes to froth height as well as the lifting
pffiffiffiffiffiffi
effect ðF0:2 rG Þ of the gas moving upwards
through the two-phase layer. From a large data
base the following correlation has been devel-
oped by Stichlmair [4, 11]:
 2=3  pffiffiffiffiffiffi
1:45 V_L =eL 125 F0:2 rG 2
hf ¼ hw þ  þ 
g1=3 lw ðrL rG Þg 1eL

pffiffiffiffiffiffi
for F  0:2 rG ð11Þ

Figure 6. Correlation of liquid entrainment versus gas load.


An increasing liquid load increases the froth The parameter lines stand for tray spacing. Curves are plotted
at indicated values of the parameter (H  hf)  Dr  g/F2max
height hf according to the second term in Equa- for a hole diameter of less than 10 mm
tion (11) (i.e., Francis formula applied to a two- H ¼ plate spacing; hf ¼ height of two-phase froth layer;
phase system). An increasing gas load F results Dr ¼ density difference; g ¼ acceleration due to gravity
in an increase of froth height according to both a
decreasing relative liquid hold-up eL (in the umn is not disturbed severely but mass transfer
second term) and an increasing third term of efficiency will be reduced to some degree. At
Equation (11). very high entrainment rates, however, the pres-
At normal operation the froth height has sure drop rises significantly and, eventually, the
values of ca. 0.1–0.3 m. This requires a tray countercurrent flow in the column breaks down
spacing of 0.3–0.5 m for safe operation. (entrainment flooding).

Entrainment of Liquid. Gas flowing Liquid Mixing. The two-phase layer on a


through the two-phase layer on a tray always tray is heavily moved and agitated by the gas
entrains some liquid in form of small droplets. emerging from the tray openings. This intensive
This liquid entrainment is unfavorable because it motion effects a partial mixing of the liquid on
affects the countercurrent flow of gas and liquid the tray resulting in leveling out concentration
within the column and, in turn, the separation gradients. This mechanism affects mass transfer
efficiency of the column. Figure 6 presents an efficiency of the trays (see Section 2.4).
empirical correlation that summarizes nearly all The degree of liquid mixing on a tray is
data published in the literature so far [4, 11]. expressed by the dimensionless Peclet number
The volume of the entrained liquid strongly whose definition is:
increases with increasing gas load ðV_E / F 5 Þ.
Thus, the entrainment data spread over 5 dec- l2L
Pe  ð12Þ
DE tL
ades. The sharp increase of the entrainment at a
relative gas load of F=Fmax  0:65 is obviously
caused by the transition from froth regime to drop The term tL denotes the residence time of the
regime (phase inversion). liquid in the two-phase layer. A value Pe ¼ 0
In practice, the entrainment of ca. 10 % of the stands for complete liquid mixing on a tray, i.e.,
liquid fed to a tray is tolerated. Even at higher no concentration gradients exist. A value Pe!¥
entrainment rates two-phase flow within the col- stands for no back-mixing and, in turn, for plug
462 Distillation, 2. Equipment Vol. 11

Figure 7. Liquid isotherms in the two-phase layer of a bubble-cap tray operated with hot water and air. Data were determined on a
bubble-cap plate with a diameter of 2.3 m, at a relative liquid flow rate V_ L =lw of 10 m3 m1 h1 A) Gas load F ¼ 0.47 Pa1/2;
B) Gas load F ¼ 1.61 Pa1/2;

flow of the liquid across the tray. Here, significant in the froth. The shape of the isotherms in Figure
concentration gradients exist in the liquid on the 7 proves increased liquid velocities near the walls
tray due to cross flow of gas and liquid. and in the center of the tray.
Predicting the degree of liquid mixing, Liquid maldistribution affects mass transfer
expressed in Peclet Pe-number, requires knowl- efficiency of columns. However, in tray columns
edge of the dispersion coefficient DE . Only few maldistribution is restricted to a single tray since
experimental data have been published in litera- the liquid is well mixed in the subsequent down-
ture. The following correlation is recommended: comer and evenly redistributed to the next tray.
 1=2 Therefore, tray columns are better suited for
eL large-diameter columns than packed columns
DE ¼ 1:06hf F ð13Þ
rL rG where the detrimental effects of maldistribution
accumulate over the packing height (see Section
3.4).
Maldistribution of Liquid. Undisturbed
plug flow of liquid across the tray provides the Interfacial Area. One of the most important
highest mass transfer rate attainable in a tray quantities of the two-phase layer on a tray is the
column. In small diameter columns the plug flow size of the interfacial area contributing to mass
is disturbed by back-mixing of some liquid. In transfer. Present knowledge on interfacial
large diameter columns, however, back-mixing area on trays is rather poor and contradicting.
of liquid is of minor importance. Here maldistri- Figure 8 shows the result of a theoretical estima-
bution, i.e., non-uniform liquid flow across the tion [4, 11] of the relative interfacial area a
tray, is the prevailing mechanism. Some charac- (interfacial area referred to the volume of the
teristic flow patterns are shown in Figure 7 two-phase layer) for three systems with different
[4, 11]. The lines plotted in the active area of surface tensions. This rough estimation shows
the trays are liquid isotherms experimentally that in aqueous systems the relative interfacial
determined in the system air–water (hot). In such area is in the range of 500–600 m2/m3. In organic
a system the mass transfer from water into air systems, however, the relative interfacial area
causes a temperature decrease of water (like in a typically reaches values up to 1000 m2/m3 or
cooling tower). In a first approximation, the even larger.
isotherms are lines of constant residence time of
the liquid in the froth. In case of plug flow the Pressure Drop. The pressure drop of a gas
liquid isotherms are supposed to be parallel flowing through a tray is a very important quan-
straight lines. Any deviations from straight lines tity. Generally, the pressure drop (better pressure
are caused by local variations of liquid velocities loss) should be as low as possible, in particular in
Vol. 11 Distillation, 2. Equipment 463

hold-up hL on the tray:


rG 2
Dp ¼ x u þhf eL rL g ð14Þ
2 Gh

The first term in Equation (14) refers to the dry


pressure loss Dpdry of the plate. It depends on the
gas velocity uGh in the tray openings. The second
term accounts for the liquid head on the tray.

Dry Pressure Loss of Sieve Trays. In sieve


trays, the orifice coefficient x of the holes can be
taken from Figure 10 [4, 11]. The reading yields
the orifice coefficient xo of a plate with a single
hole, i.e., for a relative open area j!0. This
Figure 8. Rough estimation of the relative interfacial area in value is transformed into the orifice coefficient of
the froth of sieve trays a) Isobutane–n-butane; b) Acetone– a tray with j > 0 by:
benzene; c) Air–water
– – – Bubble regime; ——— Drop regime; ——— Pre- pffiffiffiffiffi
x ¼ xo þj2 2 xo for s=dh !0 or
sumed real values 2 ð15Þ
x ¼ xo þj 2j for s=dh > 0

vacuum services. Pressure loss of the gas signifi- Dry Pressure Loss of Bubble-cap and Valve
cantly depends on both gas and liquid load as Trays. Bubble-cap and valve trays have very
shown in Figure 9. During operation the pressure complex geometrical structures that make the
drop exerts an upward force on the mechanical rigorous prediction of the orifice coefficients
structure of the tray. rather difficult. As such elements are built in
The pressure loss is caused by the gas flow great numbers, the orifice coefficients are exper-
through the openings in the tray and by the liquid imentally determined by the vendors.

Figure 9. Pressure drop of a sieve tray versus gas load. The parameter lines denote constant liquid flow rates across the outlet
weir.
Liquid load V_L =lw , m3/(m  h): a) 0 (dry pressure drop); b) 4; c) 10; d) 20; e) 40; f) 60
464 Distillation, 2. Equipment Vol. 11

Figure 10. Orifice coefficient jo for a very small relative free area j and for sharp edged holes of a sieve tray
Dashed lines indicate theoretical limits.

For the calculation of the dry pressure loss of 2.4. Mass Transfer in the Two-Phase
bubble-cap and valve trays a slightly different Layer
approach is recommended. The orifice coeffi-
cient is not related to the velocity in the smallest At the present state of the art the mass transfer on
geometrical open area but to the whole active a tray is expressed by tray efficiencies. Two
area of the tray since the quality of tray design definitions of tray efficiencies have to be distin-
depends not only on the pressure loss of a single guished, i.e., point efficiency EOG and tray effi-
element but also on the number of elements ciency EOGM .
arranged on the tray. The following definition
implies both quantities:

0 r
Dpdry  x  u2 ð16Þ
2

A compilation of tray orifice coefficients x0 is


presented in Figure 11 [4, 11]. The symbols are
plotted for the recommended number of elements
per tray. The parameter lines show the orifice
coefficients when less elements are installed.
Such a modification is often necessary for geo-
metrical reasons. The abscissa in Figure 11 is the
smallest relative free area j0 which can be, for
instance, either the slots or the chimney of a
bubble-cap. For comparisons sake, the orifice
coefficient x0 of a sieve tray with xo ¼ 1:7 is
also plotted in Figure 11. The quality of a tray Figure 11. Correlation of the orifice coefficients j0 of bubble-
design is determined by the distance of the rele- cap and valve trays on the relative free hole area j
vant parameter line from the line of a sieve tray Bubble-cap plates: a) Bayer; b) Montz KSG; c) Linde;
d) ACV; e) Thormann 1963; f) Kirschbaum 1951
with xo ¼ 1:0, which is the theoretical minimum Valve trays: g) Koch TK8; h) Stahl VV 12/2; i) Koch A, T;
of a sieve tray with rounded edges. j) Glitsch V-4; k) Glitsch V-1; l) Nutter type 1972
Vol. 11 Distillation, 2. Equipment 465

The point efficiency is discribed by the con-


centrations along an individual stream line of the
gas moving vertically through the froth:

yyn1
EOG  ð17Þ
y ðxÞyn1

The subscript OG indicates that the overall mass


transfer resistance is formally placed into the gas
phase. The point efficiency can reach a maximal
value of 1 assumed the liquid is thoroughly
mixed over the height of the froth. This assump-
tion is well met. Hence, point efficiency is an Figure 12. Relationship between tray and point efficiency
important theoretical quantity. under consideration of liquid back-mixing represented by the
In practice, tray efficiencies (Murphree effi- Peclet number Pe
ciency) are used which are defined by:
yn yn1
EOGM  ð18Þ mixing of the liquid. At total back-mixing, tray
y ðxn Þyn1
efficiency EOGM is equal to point efficiency EOG .
where the total gas flow through the tray is At very low back-mixing Equation (19) holds.
considered. The actual concentrations change For large-diameter trays very large values of
ðyn yn1 Þ is related to
 the maximal  feasible tray efficiency are found (Fig. 12) since large
concentration change y ðxn Þyn1 . The gas trays have a very small degree of liquid back-
concentration y ðxn Þ is in state of equilibrium mixing. However, large trays normally have
with the liquid leaving the tray. This definition is some degree of maldistribution which also exerts
arbitrary because the liquid concentration varies a detrimental effect on tray efficiency. In practice
on the tray due to the cross flow of the phases. tray efficiencies are only slightly higher than
Hence, values of tray efficiencies higher than 1.0 point efficiencies.
are feasible.
Equations (17) and (18) are the definitions of Point Efficiency. EOG is calculated from
gas-side efficiencies. Sometimes liquid-side ef- fundamental laws of mass transfer. Adjusting
ficiencies are used to describe the interfacial these laws to tray columns yields:
mass transfer. However, the gas-side definitions EOG ¼ 1expðNOG Þ with ð20Þ
are preferred since gas-phase resistance is the
dominant mechanism in distillation processes. abG hf =uG
NOG ¼
1þmðbG =bL ÞðM^L =M^G ÞðrG =rL Þ
Relationship between Point and Tray Effi-
ciency. In case of plug flow of the liquid the The most crucial quantities in this equation are
following relationship between point and tray the interfacial area a (see Section 2.3), the
efficiency holds: gas-side mass transfer coefficient bG , and the
liquid-side mass transfer coefficient bL .
   
EOGM L=_ G_ 1 m The mass transfer coefficients can be estimat-
¼   exp EOG 1 ð19Þ ed by the assumption of an unsteady mass trans-
EOG m EOG _ G_
L=
fer [4, 11]. Decisive is the residence time tG of
the gas in the froth since it is the gas present in the
Plug flow prevails on very large trays only. In liquid that generates the interfacial area and, in
smaller trays there is always some degree of turn, enables the interfacial mass transfer:
back-mixing of liquid, which is quantitatively rffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffi
4DG 4DL
expressed by the Peclet number (see Section 2.3). bG ¼ and bL ¼ with tG ¼ hf eL =uG
ptG ptG
Figure 12 presents the relationship between tray ð21Þ
and point efficiency under consideration of back-
466 Distillation, 2. Equipment Vol. 11

Figure 13. Typical experimental values of gas-side point


efficiency a) NG ¼ 2, NL ¼ 15; b) NG ¼ 2, NL ¼ 2; c) NG
1, NL ¼ 1; d) NG ¼ 0.5, NL ¼ 0.5
Experimental data: e) Distillation; f) Vacuum distillation;
g) Absorption, hL ¼ 10–20 cP; h) Absorption of carbon
dioxide in water; i) Desorption of oxygen from water

In organic systems with low values of the inter-


facial tension, Equation (21) sometimes gives
too low values due to a steady renewal of the
interfacial area.
The weak point of mass transfer prediction,
however, is poor knowledge of the interfacial
area a that contributes to mass transfer (see Figure 14. Perspective view of a packed column (Sulzer)
a) Liquid distributor; b) Liquid collector; c) Structured pack-
Section 2.3). Therefore, empirical values of tray ing; d) Support grid; e) Man way; f) Liquid re-distributor
efficiencies are normally used in practice. In
distillation systems values of tray efficiencies of pressure drop of the gas and a liquid flow which
70 % are often recommended. However, if a is uniform over the cross section of the column.
significant liquid-side mass transfer resistance
exists, tray efficiencies can be much lower. Some
empirical values are presented in Figure 13. 3.1. Design Principles

Two different design principles of packings can


3. Packed Columns be distinguished:

Packed columns are as important as tray columns . Random packings consist of a dumped bed of
in the process industry. Due to novel develop- particles made of ceramic, metal, or plastic
ments of packing elements the industrial use of (Fig. 15). The shape of the particles should
packed columns is steadily increasing. In packed ensure a homogeneous structure of the dumped
columns there exists a genuine countercurrent bed and a low pressure drop of gas flow.
flow of gas and liquid as is shown in Figure 14. Standard particles (Raschig rings, Berl sad-
An intimate contact between gas and liquid dles) have a closed surface. Modern particles
phases is established by packings that represent are characterized by net structures (Pall rings)
a solid structure with large porosity and large or grid structures (Super rings). The size of the
internal surface. The liquid proceeds downwards particles must not exceed 5–10 % of the col-
on the packing in form of thin films or rivulets. umn diameter. The porosity of a bed is as high
Decisive for a good performance are a low as 70 % (ceramic particles) and more than
Vol. 11 Distillation, 2. Equipment 467

Figure 15. Examples of random packing elements


Ceramic packings are shown in the top row, metal packings in the middle row, and plastic packings in the bottom row

90 % (metal particles). The volumetric surface sure drop per equilibrium stage is typically as
of a bed depends on particle size and on column low as 0.5 mbar.
diameter. Typical values are 50–300 m2/m3
(Table 2). The separation efficiency of random Very important for all types of packings is a
packings strongly depends on column diame- uniform liquid distribution at the top of the bed
ter. Typical values of industrial columns are in and a bed-height limitation of 6–8 m. Beneath
the range of 1–2 equilibrium stages per meter each bed, the liquid has to be collected, mixed,
packing height. and redistributed. These measures intend to sup-
. Structured packings have been increasingly press the maldistribution of liquid because it
used since the mid 1960s because they better strongly affects mass transfer rates. The design
meet the requirements for homogeneous bed of liquid distributors, liquid collectors, support
structures. The standard design consists of grids etc. should provide a high open area in order
parallel corrugated metal sheets, ca. 20 cm in to assure the countercurrent flow of gas and liquid.
height, with alternating orientation of subse-
quent layers (Fig. 16). The porosity of struc-
tured beds is significantly higher than 95 %. 3.2. Operation Region of Packed
The specific surface area is typically Columns
250 m2 =m3 or higher (Table 3). Structured
packings have a lower pressure drop and a Packed columns can be operated within certain
better mass transfer efficiency than random limits of gas and liquid loads only. A typical
packings. Mass transfer efficiency is as high operation region of packed columns is shown in
as 2–4 equilibrium stages per meter. The pres- Figure 17. The mechanisms that set limitations to
468 Distillation, 2. Equipment Vol. 11

Table 2. Characteristic data of some random packings

Type dN, mm e a, m2/m3

Raschig rings (ceramic) 10 0.65 440


25 0.726 192
38 0.76 140
50 0.77 98
Raschig rings (metal) 10 0.744 520

Raschig super rings (metal) 20 0.970 250


25 0.98 180
30 0.98 150
38 0.98 120
50 0.98 100
Pall rings (metal) 25 0.94 210
35 0.945 141
50 0.96 102
90 0.97 65
Figure 16. Structured packing (Sulzer)
Pall rings (ceramic) 25 0.73 220
35 0.76 165
50 0.78 120
an operating pressure loss of ca. 3 mbar=m is
Raflux rings (metal) 25 0.94 215
35 0.95 145 often recommended for column dimensioning.
50 0.97 120
90 0.97 65 Maximum Gas and Liquid Load. The
Hiflow (metal) 25 0.97 200 upper operation limit of packed columns (flood-
28 0.97 200 ing) is caused by a too high pressure drop that
50 0.98 99 blocks off the countercurrent flow of gas and
58 0.987 92
liquid. There exist numerous empirical correla-
Hiflow (ceramic) 20 0.72 280 tions for predicting flooding of packed columns
35 0.76 128
50 0.78 102
[3]. The correlation of MERSMANN [14, 15] is
75 0.76 70 illustrated in Figure 18. In principle, the gas load
IMTP (metal) 15 0.964 282
is plotted versus the liquid load in this diagram.
25 0.972 225 However, the gas load is substituted by the dry
40 0.981 150 pressure drop of the gas flowing through the
50 0.979 100 packing. The abscissa is a dimensionless liquid
Torus saddles (ceramic) 25 0.724 185 load. The parameter lines denote constant values
38 0.76 128 of the dimensionless pressure drop of the irrigat-
50 0.76 142
75 0.78 92
90 0.78 68 Table 3. Characteristic data of some structured packings

Type e a, m2/m3

Flexipac (metal) 0.91 558


0.94 223
gas and liquid flow rates are flooding and poor 0.96 135
surface wetting. Flooding is a strong limitation 0.98 70
that cannot be surpassed. Insufficient surface Gempak (metal) 0.9475 525
wetting is a soft limitation that can be crossed 0.96 394
at the expense of poor mass transfer efficiency. It 0.961 262
0.974 131
is important to note that the operation range of the
liquid is further narrowed down by the liquid Mellapak (metal) 0.975 250
0.975 500
distributor. The gas load can be very low in a
packed column. Montzpak (metal) 0.972 300
The operation point has to assure a sufficient 0.98 200
0.99 100
safety margin to the operation limit. In practice,
Vol. 11 Distillation, 2. Equipment 469

Figure 17. Operation region of packed columns a) Minimum liquid load; b) Operating region of liquid distributor; c) Flooding;
d) Operating region of packing

ed packing. Flooding point and loading points are the following equation [16]:
also marked in this diagram.
 2=9  
A more rigorous approach to flooding is based r s3 g 1=2
uL min ¼ 7:7106  L4  ð22Þ
on pressure drop models of irrigated packed beds hL g a
(see Section 3.3).

Minimum Liquid Load. The minimum liq- For industrial practice the following values are
uid load of random packings can be estimated by recommended:

Figure 18. Correlation of flooding of packed columns according to Mersmann [14, 15]
Calculated cuves are shown for different values of the pressure drop in an irrigated packing, Dpirr / (H  rL  g): a) 0.2; b) 0.1;
c) 0.05; d) 0.04; e) 0.03; f) 0.02; g) 0.01; h) 0.005
470 Distillation, 2. Equipment Vol. 11

. For aqueous systems: uLmin  10 m3m2h1 dispersed particles is [4, 11]:


. For organic systems: uLmin  5 m3m2h1
Dpdry 3 1e u2
¼ xo  4:65 rG  G ð25Þ
H 4 e dp
Structured wire mesh packings can be operat-
ed with organic systems down to 0.2 m3m2h1.
This value is in the order of magnitude of a heavy Equation (25) differs from the well known Ergun
tropical rainfall. equation [3] by the exponent 4.65 on the porosity
An additional limitation to liquid load is e. The improved porosity term allows the predic-
set by the liquid distributor as shown in Figure tion of the pressure loss of a bed of dispersed
17. In most cases the operation of the distrib- particles from knowledge of the friction factor xo
utor is less flexible than that of the packing of a single particle. Hence, Equation (25) holds
itself. for the whole range of bed porosity. Figure 19
presents a graphical correlation of the friction
factors xo for several dumped and structured
3.3. Two-Phase Flow in Packed packings [4, 11].
Columns In a limited range of Reynolds numbers, the
values presented in Figure 19 are correlated by:
The geometrical structures of packed beds are too
uG dp rG
complex for mathematical description. There- xo ¼ bReð2þcÞ
o with Reo  ð26Þ
hG
fore, the real bed structures are replaced by model
structures, e.g., by a system of parallel channels
or a system of dispersed particles. The model Some values of the factor b and the exponent c are
structures should have the same interfacial area a listed in Table 4.
and the same porosity e as the packed bed. From
both conditions follows the hydraulic diameter Pressure Drop of Irrigated Beds. Typical
dh of the channel model structure: values of the pressure drop of dry and irrigated
beds are presented in Figure 20. The pressure
dh ¼ 4e=a ð23Þ loss of irrigated beds is always significantly
higher than that of dry beds since the character-
The same conditions lead to the particle diameter istic quantities of the bed are changed by the
dp of the particle model structure: liquid accumulated within the bed (liquid hold-
up hL ).
1e
dp ¼ 6 ð24Þ
a

It must be noted that dh characterizes the open


structure and dp the solid structure of the packed
bed, respectively. The round channels intersect at
large porosities of the bed. The particles
(spheres) of the particle structure do not touch
each other like in a fluidized bed. Thus, the
particle model is more realistic and preferred in
this chapter.

Pressure drop. The pressure drop of a gas


flowing through a packed bed is one of the key
quantities in the design of packed columns. It is
dominated by the gas and liquid load as well as by
the structure of the bed.
Figure 19. Correlation of the friction factor jo according to
the particle structure model a) Raschig rings, metal;
Pressure Drop of a Dry Bed. The pressure b) Raschig rings, ceramic; c) Pall rings; d) Saddles;
drop of a gas flowing through a dry bed of e) Cylinders; f) Spheres; g) Structured packings
Vol. 11 Distillation, 2. Equipment 471

Table 4. Values of the factor b and the exponent c in Equation (26)

Packing material Factor b Exponent c Reynolds number

Raschig rings (ceramic) 7.4 0.17 6101 to 3103


Raschig rings (metal) 16 0.13 1101 to 2102
Pall rings (metal) 4.8 0.15 1102 to 3103
Saddles 5.5 0.20 7101 to 6103
Spheres 1.7 0.20 2102 to 5103
Cylinders 3.0 0.23 6101 to 2103
Structured packings 4.8 0.38 4101 to 3103

The liquid hold-up hL changes the bed char-


acteristics in different ways. Of great importance
are the reduction of the bed porosity e, the Figure 21. Liquid hold-up hL as a function of pressure drop
enlargement of the particle diameter dp and the in the irrigated packing Dpirr. Packed column with 25 mm
Bialecki rings (metal) [17]
reduction of the friction coefficient xo . All three Liquid velocity uL, m/h: a) 80; b) 60; c) 40; d) 20; e) 10
effects can be modeled giving [4, 11]:

  
Dpirr 1eð1hL =eÞ ð2þcÞ=3 hL 4:65
¼  1 ð27Þ
Dpdry 1e e
packing volume, are presented in Figure 21 [17].
Here, the experimental values of the hold-up hL
Evaluation of this equation demands knowledge are plotted versus the pressure drop in the packed
of the liquid hold-up hL in the bed. bed Dpirr =ðHrL gÞ. Two regions can be distin-
guished in this diagram:
Liquid Hold-up in Packed Beds. Liquid
hold-up is a decisive parameter of two-phase . A region of low pressure drop (low gas load)
flow in packed beds. Typical values of the where the hold-up depends on liquid load only
hold-up hL , defined as ratio of liquid volume to (horizontal parameter lines)
. A region of high pressure drop (high gas load)
where the hold-up is further increased by
pressure drop

The transition between these two regions is


called loading point. The parameter lines in
Figure 21 end at the flooding point.
In the region below the loading point the
liquid hold-up hLo depends on liquid load, liquid
properties, and packing structure only. The
following correlation holds for random packings
[18]:
 2 0:33  2 3 0:125  2 0:1
u a h a sa
hLo ¼ 3:6 L  L2  ð28Þ
g rL g rL g

At column operation above the loading point,


the liquid hold-up is further influenced by the
Figure 20. Pressure drop Dp of a packed column composed pressure drop of the column. This influence is
of 25 mm metal Bialecki rings (porosity 0.94). The parameter described by:
lines denote constant liquid load [17]
Liquid velocity uL, m/h: a) 0; b) 10; c) 20; d) 40; e) 60; f) 80 "   #
Measurement were made using air–water at 20 C and 1 bar in Dpirr 2
hL ¼ hLo  1þ20 ð29Þ
a column with packing height of 1.5 m and a diameter of HrL g
0.15 m.
472 Distillation, 2. Equipment Vol. 11

Flooding of Packed Columns. As shown in


Figure 20, the condition for flooding is an infinite
increase of pressure drop with increasing gas
load. Taking the dry pressure loss as measure
for the gas load gives the following flooding
condition:
qDpirr qDpdry
¼ ¥ or ¼0 ð30Þ
qDpdry qDpirr

Combining Equations (27) and (28) to (29) and


performing the derivation (see Equation (30))
gives [4, 11]:
1 40ð2þcÞ=3hLo 186hLo
¼ þ with ð31Þ
D2 1eþhLo ð1þ20D2 Þ ehLo ð1þ20D2 Þ

Dpirr
D
HrL g Figure 22. Pressure drop and flooding of a packed column in
an irrigated packing composed of 25-mm Bialecki rings
Equation (31) can be solved by iteration only. As (metal). The values of the abscissa are determined from
Equation (28)
flooding of random packings typically occurs at Calculated curves are shown for different values of the
values of the dimensionless pressure drop pressure drop in an irrigated packing, Dpirr/(H  rL  g):
Dpirr =ðHrL gÞ  0:1 to 0:3, a value of D ¼ 0:1 a) 0.3; b) 0.2; c) 0.1; d) 0.06; e) 0.04; f) 0.02; g) 0.01;
is a good first estimation. h) 0.006; i) 0.004; j) 0.002
Figure 22 presents the result of the evaluation
of Equation (27) together with Equations (28)
and (29) for a random packing of Bialecki rings. stage (HETP) are also determined and applied to
This plot is almost identical with the empirical column dimensioning. A more rigorous approach
flooding correlation of Mersmann (Fig. 18). Just is based on the following equation valid for
the dimensionless liquid load is replaced by the undisturbed plug flow of gas and liquid within
liquid hold-up hLo below loading. the packing:
!
uG b M ^ r
HOG ¼  1þm G  L  G ð33Þ
3.4. Mass Transfer in Packed Columns bG aeff bL M^G rL

By convention, mass transfer in packed columns


is described by the model of transfer units since This rather simple equation contains three cru-
a countercurrent flow of gas and liquid prevails cial quantities that depend on design and opera-
in packed beds. According to this model the tion of the packed column. These quantities are
required height H of the packing within the the effective interfacial area aeff , the gas-side
column is: mass transfer coefficient bG , and the liquid-side
mass transfer coefficient bL . According to the
H ¼ HOG NOG ð32Þ
present state of the art, no models based on first
principles are available for the prediction of
The term NOG denotes transfer units which is a these rather crucial quantities. Therefore, sev-
measure of the difficulty of the separation ! eral empirical correlations have been published
Distillation, 1. Fundamentals. The mass transfer so far [3].
in a packed bed is expressed by the height of a
transfer unit HOG . In practice empirical values of Empirical Correlations for Mass Transfer
the height of a transfer unit are often used for Prediction. Often recommended is the empiri-
column dimensioning. Sometimes, empirical va- cal model developed by Onda [19]. His correla-
lues of the height equivalent of an equilibrium tion of the effective interfacial area is:
Vol. 11 Distillation, 2. Equipment 473

Table 5. Values of critical surface tension in Equation (34) [19] Table 6. Selected values of the factors CV and CL in Equations (39)
and (40) according to [21]
Packing material scrit , N/m
Packing CV CL
Polyethylene 0.033
Poly(vinyl chloride) 0.040 Raschig rings (ceramic) 50 0.210 1.416
Ceramic 0.061 25 0.412 1.361
Glass 0.073
Raschig super rings (metal) 1 0.440 1.290
Stainless steel 0.071
2 0.400 1.323
Steel 0.075
3 0.300 0.850
Bialecki rings (metal) 50 0.302 1.721
35 0.390 1.412
25 0.331 1.461
aeff
¼ 1exp
a Pall rings (metal) 50 0.410 1.192
"      0:05  2 0:2 # 35 0.341 1.012
scrit 0:75 uL rL 0:1 u2L a u r
1:45    L L 25 0.336 1.440
s ahL g sa
Pall rings (ceramic) 50 0.333 1.278
ð34Þ
Intalox (ceramic) 0.488 1.677
Some values of the critical surface tension scrit Ralupak (metal) YC-250 0.385 1.334
are listed in Table 5. The corresponding equa-
Impuls packing (metal) 250 0.270 0.983
tions for the mass transfer coefficients are: (ceramic) 100 0.327 1.317
   1=3
rG uG 0:7 hG Montz packing (metal) B1-200 0.390 0.971
bG ¼ C  ðadN Þ2 ðaDG Þ ð35Þ (metal) B2-300 0.422 1.165
ahG rG DG

     
ghL 1=3 rL uL 2=3 rL DL 1=2
bL ¼ 0:0051   ðadN Þ2=5
rL aeff hL hL  
uL DL 1=2
ð36Þ bL ¼ CL 121=6  ð40Þ
hL dh
The factor C is 2 for small ðdN < 15 mmÞ and
5.24 for larger packing elements. Values of the factors CV and CL are listed in
BRAVO and FAIR [20] published a model that Table 6.
combines the correlations of Onda for the mass
transfer coefficients with a new correlation for Problems of Mass Transfer Prediction. At
the effective interfacial area: the present state of the art all models for predic-
  tion of mass transfer in packed columns are
aeff h uL uG rG 0:392
¼ 19:76s0:5 H 0:4  L  ð37Þ insufficient to some degree. One problem is a
a s ahG
poor knowledge of the interfacial area effective
for mass transfer. Published studies differ signif-
In this correlation, the term H 0:4 accounts for the icantly and their results are often contradicting.
influence of maldistribution on mass transfer. Furthermore, the existing models for predicting
Equation (37) is valid for the units m and kg/s2 mass transfer coefficients are not sufficiently
for the height H and the surface tension s, reliable.
respectively. The main reason for the deficiencies of the
BILLET and SCHULTES [21] developed the present knowledge is, however, the maldistri-
following correlations for all three unknowns: bution of gas and liquid in packed beds. Many
   0:75  2 0:45 studies reveal that there exist large deviations
aeff r uL dh 0:2 rL u2L dh uL
¼ 1:5ðadh Þ0:5  L   from plug flow of gas and in particular of liquid
a hL s gdh
within the bed. The degree of maldistribution as
ð38Þ
well as its effect on mass transfer is unknown
 1=2    1=3 and hence not accounted for in existing mass
a r uG 3=4 hG
bG ¼ CV ðehL Þ1=2  DG  G  transfer models. Thus, the published data for
dh ahG rG DG
interfacial area and mass transfer coefficients
ð39Þ
comprise the maldistibution in an undefined
474 Distillation, 2. Equipment Vol. 11

Figure 23. Maldistribution in a packed column indicated by liquid isotherms. Column diameter 0.63 m, packing height 6.8 m.
A) Uniform liquid feed; B) Single-point liquid feed
Bottom row: Orientation of vertical cross sections.

manner. The data are not true values but pseudo- causes a decrease of liquid temperature which
values which are not predictable within the plug can be easily measured. The lines drawn within
flow model. the packed bed are liquid isotherms that should
Figure 23 presents some experimental results be horizontal and straight in case of plug flow.
of a packed column operated with hot water and Any deviations from straight lines are caused by a
air (like a cooling tower) [6]. In this system the non-uniform gas and, in particular, liquid flow
mass transfer of water from liquid into gas phase through the packed bed.
Vol. 11 Distillation, 2. Equipment 475

In the experiments performed with uniform 8 SulCol of Sulzer, Switzerland, www.sulzerchemtech.


liquid distribution at the top (Fig. 23 A), the com.
degree of maldistribution increases on the way 9 TrayHeart of Welchem, Germany, www.welchem.com.
10 J.R. Fair,: ‘‘How to predict sieve tray entrainment and
of the liquid downwards through the bed. At non-
flooding’’, Petro/Chem. Eng. 33 (1961) no. 10, 45 – 52.
uniform initial liquid distribution (Fig. 23 B), 11 J. Stichlmair: Grundlagen der Dimensionierung des Gas/
however, the degree of maldistribution decreases. Fl€ussigkeit-Kontaktapparates Bodenkolonne, Verlag Che-
Thus, the degree of maldistibution depends on mie, Weinheim 1978.
both the liquid distributor and the packed bed 12 A. Mersmann: ‘‘Wann werden alle L€ocher einer Siebbo-
itself. At present it is not possible to predict either den-Lochplatte durchstr€omt?’’, Chem.-Ing. Tech. 35
the degree of maldistribution or its effect on mass (1963) no. 2, 103 – 107.
transfer. As the effects of maldistribution accu- 13 K. Ruff, Th. Pilhofer, A. Mersmann: ‘‘Vollst€andige
Durchstr€omung von Lochb€oden bei der Fluid-Dispergier-
mulate over packing height, it is important to
ung’’, Chem.-Ing.-Tech. 48 (1976) no. 9, 759 – 764.
suppress the maldistribution by dividing the pack- 14 A. Mersmann: ‘‘Zur Berechnung des Flutpunktes in
ing into several short sections. This measure for F€ullk€orpersch€uttungen’’, Chem. Ing. Tech. 37 (1965)
suppressing the maldistribution requires the in- 218 – 226.
stallation of multiple support grids, liquid collec- 15 A. Mersmann, K. Bornh€utter: ‘‘Druckverlust und
tors, liquid mixers, and liquid distributors. Flutpunkt in berieselten Packungen’’, VDI-W€ armeatlas,
VDI-Verlag, D€usseldorf 2002.
16 R. Schmidt: ‘‘The lower capacity limit of packed
columns’’, Inst. Chem. Eng., Symp. Ser. 56 (1979) , no.
References 2, 3.1 – 1-14.
17 R. Billet, J. Mackowiak: ‘‘How to use the absorption data
1 R.H. Perry, D.W Green: Perry’s Chemical Engineers’ for design and scale-up of packed columns’’, Fette Seifen
Handbook, McGraw-Hill, New York 2008. Anstrichmittel 86 (1984) no. 9, 349 – 358.
2 H.Z. Kister: Distillation Operation, McGraw-Hill, New 18 V. Engel: ‘‘Fluiddynamik in Packungskolonnen f€ur
York 1989. Gas–Fl€ussig-Systeme’’, Fortschrittsberichte VDI, Reihe
3 H.Z. Kister: Distillation Design, McGraw-Hill, New 3, no. 605, VDI Verlag D€usseldorf 1999.
York 1992. 19 K. Onda, H. Takeuchi, Y. Okumoto: ‘‘Mass transfer
4 J.G. Stichlmair, J.R. Fair: Distillation - Principles and coefficients between gas and liquid phases in packed
Practice, Wiley-VCH, Weinheim 1998. columns’’, Journal of Chemical Engineering of Japan 1
5 M. J. Lockett: Distillation tray fundamentals, Cambridge (1968) 56 – 62.
University Press, New York 1986. 20 J.L. Bravo, J.R. Fair: ‘‘Generalized correlation for mass
6 A. Mersmann, M. Kind, J. Stichlmair: Thermische transfer in packed distillation columns’’, Ind. Eng. Chem.
Verfahrenstechnik, Springer-Verlag, Berlin, Heidelberg, Process Des. Dev. 21 (1982) Vol. 1, 162 – 170.
New York 2005. 21 R. Billet, M. Schultes: ‘‘Prediction of mass transfer
7 KG-Tower of Koch-Glitsch, USA, www.kochglitsch. columns with dumped and arranged packings’’, Trans
com. IChemE 77 (1999) Part A, 498 – 504.

You might also like