Prediction of the Effect of Weeping on

Distillation Tray Efficiency

The reduction in distillation tray efficiency caused by uniform weeping of liquid M. J. LOCKETT,
from the tray has been calculated. Numerical results are presented for all three
Lewis cases over the range of variables of interest to tray designers. Analytical M. A. RAHMAN and
equations are given for Lewis’ cases 2 and 3 for plug flow of liquid on the tray. H. A. DHULESIA
Department of Chemical Engineering
UMlST
Manchester, England

SCOPE
One consequence of the present economic climate is that throughputs, it is important to predict the reduction in tray ef-
many distillation columns are operating at less than their design ficiency which results from weeping. Previous work, by
capacity. At reduced vapor loads, weeping becomes a problem Kageyama, deals only with the case where the vapor is com-
in trayed columns because it reduces tray efficiency and so pletely mixed between trays (Lewis Case 1).This is a reasonable
adversely affects the separation achieved by the column. To assumption only for small pilot-plant-size columns. The objec-
maintain the required separation, it is possible to overreflux. tive of the present study was to extend the analysis so as to be
This reduces the number of theoretical trays required and also, applicable to industrial columns by making the assumption that
by maintaining a high internal vapor load, reduces the tendency the vapor is unmixed between trays. In addition, both alternate
to weep. However, overrefluxing increases energy consump- direction (Lewis Case 3) and parallel liquid flow on successive
tion. trays (Lewis Case 2) have been considered.
To determine the best strategy of operation at reduced

CONCLUSIONS AND SIGNIFICANCE

In the presence of weeping, distillation tray efficiency de- the liquid Peclet number is high and conditions approximate
pends on the fraction of the liquid which weeps, liquid Peclet to plug flow. Analytical solutions are given for this case suitable
number, stripping factor, point efficiency and the particular for inclusion in tray design computer programs. The significance
Lewis case. Tabulated results are presented from which the of the work is that, coupled with prediction methods for weep-
apparent Murphree tray efficiency can be determined as a ing fraction and point efficiency under weeping conditions, it
function of each of these five variables. The range of variables should now be possible to predict tray efficiency when weeping
covered is of interest to tray designers. For large diameter trays, occurs.

INTRODUCTION (parallel flow), liquid on the tray is partially mixed, vapor and
weeping liquid are unmixed between trays.
Kageyama (1966, 1969) presented the results of a model from Case 3: liquid flows in alternate directions on successive trays,
which the effect of weeping on Murphree tray efficiency was liquid on the tray is partially mixed, vapor and weeping liquid are
calculated. The conditions approximated to those for Lewis’ Case unmixed between trays.
1 (1936). That is, the vapor was completely mixed and the weeping Case 1 corresponds to small-diameter columns as indicated
liquid was unmixed between trays. This model is only relevant to above. Case 2 corresponds to high-efficiency parallel flow trays
small pilot-plant-scale columns in which complete transverse vapor as used,for example, by Union Carbide (Smith and Delnicki, 1975)
mixing between trays can reasonably be assumed. The only other and recently by other tray designers (Haselden et al., 1982;Jenkins,
related study is that of O’Brien (1966), but in his study liquid 1981).Case 3, which gives the lowest tray efficiency, corresponds
mixing on the tray was not considered. The present work addresses to the majority of trays installed.
the problem of the effect of weeping on tray efficiency for all three For Case 1, simultaneousassumptions are made of no mixing of
cases proposed by Lewis. Both analytical and numerical solutions the weeping liquid but complete vapor mixing between trays. This
to the problem have been obtained. seems reasonable because of the large size of the weeping drops.
The three cases considered are: In a previous study dealing with entrainment (Lockett et al., 1983,
0 Case 1: liquid flows in alternate directions on successive trays, we have considered Case 1 for entrainment to involve complete
liquid on the tray is partially mixed, vapor is completely mixed mixing of both entrained drops and vapor between trays. In view
between trays, weeping liquid is unmixed between trays. of the generally smaller drop size obtained during entrainment than
Case 2: liquid flows in the same direction on successive trays during weeping, this seems a justifiable assumption. An even more
rigorous analysis than that presented below would allow for partial
mixing both of weeping liquid and vapor. The benefits to be gained
Correspondenceconcerning this paper should be addressed to M. 1. h k e t t . from such a study are questionable, however, because the extent
Present addresses: M. J. k k e t t , Union Carbide Corp., Linde Div., Tonawanda, NY; M. A. of mixing is not known and the results would differ only marginally
+
Rahman, Ahu Dhabi National Oil CO.,T P Directorate, Abu Dhahi, and H. A. Dhulesia, Direction
Research Operationelle, Compgnie Francaise des Petroles, Paris, France. from those given in this paper.

AlChE Journal (Vol. 30, No. 3) May, 1984 Page 423

 v v

I b -
Figure lb. Flow rates wlth weeping.

(EQUIMOLAR OVERFLOW) The four terms in Eq. 1 represent transport of material from or to
Figure l a . Flow rates wlthout weeping (equlmolar overflow). a point on the tray by the following mechanisms respectively: liquid
mixing, flow of liquid, and wecping and mass transfer with the
vapor.
NUMERICAL SOLUTIONS As shown in Figure 1, the liquid flow rate on each tray, I,, in the
presence of weeping, is less than the externally supplied reflux flow
rate Lo. It follows that L = Lo - L,. It is convenient to redefine
Main Assumptions the stripping factor and the weeping fraction in terms of the ex-
1. Constant liquid, vapor and weeping flow rates from tray to ternal flow rates, i.e.,
tray. mV
A, = - and Po = -
LlU
2. Weeping uniformly distributed over the tray. L O LO
3. Constant point efficiency.
4. No mass transfer between vapor and weeping liquid-or if So that Eq. 1 becomes:
any occiirs, it is included in the point efficiency.
5. Rectangular tray.
6. Linear equilibrium relationship y f = mxn b +
Kageyama (1966) has given the basic differential equation for
the liquid concentration on a general tray n as a function of dis-
tance from inlet weir, z'.
The usual boundary conditions are used for Eq. 2, ix.,
a) Danckwerts boundary condition at the inlet weir 2 =o
(Danckwerts, 1953),

Page 424 May, 1984 AlChE Journal (Vol. 30, No. 3)

 b) At the exit weir, z = 1,
dx,=() (4)
dz w =O w = l
n-1
The calculation procedure involved the following steps:
(i) Values of A,, E, Pe and 0,were set.
(ii) Arbitrary values of xiR and (xN)out were set such that x i R
< ( X N ) ~ ~where
~, xiR is the liquid composition in equilibrium with
the vapor entering the bottom tray from the reboiler (assumed
completely mixed) and (xN)outis the liquid composition leaving
the bottom tray. The calculated efficiencies are independent of
these assumed values. n
(iii) Equation 2 was integrated numerically to obtain the liquid
concentration profile on the bottom tray. For the first iteration, 0,
was set equal to zero. From the definition of point efficiency and
the equilibrium relationship

n+l
and this equation was used to obtain the xin profile leaving the
bottom tray.
(iv) The liquid concentration leaving the second tray was de-
termined from Eq. 3.

+
(v) The vapor concentration profile entering the next tray was
determined from the vapor concentration profile leaving the froth
of the tray below according to the Lewis case assumed.
(vi) The calculations were repeated tray by tray up the column
for a total of ten trays.
(vii) The liquid concentration profiles so obtained were used
in the next overall column iteration to estimate the concentration
profile of the weeping liquid leaving each tray. The whole proce- x n + dxn
dure was repeated with a finite value of p,, until convergence was
achieved on the liquid concentration profiles on each tray.
Xn Yn+ I
Calculation of Tray Efficiency
Figure 2. Nomenclaturefor Lewis Case 2 with weeping.
In the presence of weeping, two alternative definitions of tray
efficiency can be used. These are the apparent efficiency, ELv,
analogous to that proposed by Colburn (1936)in the presence of of view, the tray designer in practice need not concern himself with
entrainment, and the reduced efficiency, ELv, proposed by how they are defined, but simply use them as he would a normal
Standart and Kastanek (1966). These have been proved to be nu- Murphree tray efficiency. In his work Kageyama (1966)used ELv.
merically identical in the case of entrainment (Lockett et al., 1983; However, since ELV and EGv are numerically identical, we prefer
Rahman and Lockett, 1981),and an exactly analogous proof can to refer to the efficiency as the apparent efficiency. This then
be made to show that they are also numerically identical in the makes the terminology consistent with what we have used for en-
presence of weeping. This was also confirmed using the calculated trainment and as originally used by Colburn (1936).
concentrationsfrom the computer program. The definitions of the The apparent tray efficiency was determined from the calcu-
two efficiencies are: lated concentrations around each tray. As the bottom tray receives
completely mixed vapor, this has a different efficiency than the
(5) others for Lewis Cases 2 and 3. However, by the fourth tray from
the bottom of the column, the tray efficiency becomes constant and
where the reported efficiencies are those for the sixth tray.

and ANALYTICAL SOLUTIONS FOR PLUG FLOW OF LIQUID

Lewis Case 2

and also From a component material balance around a differential slice

of the dispersion on tray G,Figure 2, we have

[(~n - Yn+ 1) -f(xn-1- xn)lVdw = Ldxn (7)

where where f = L,/V and dw = dvf V
y; = m(XL)out +b Using the equilibrium relationship, y* = mx
= mV/L, Eq. 7 becomes,
+ b, and with X
and
Ltu
(%)out = izn)out - - (6%)out - X,1
L O
The point efficiency E is
The reduced concentration (XG),ut is, in fact, the average concen-
tration of the liquid leaving tray n. Either E S v or ELv can be used (9)
in place of the usual Murphree vapor-phase tray efficiency to give
the effective efficiency in the presence of weeping. From that point Substituting and simplifying,

AlChE Journal (Vol. 30, No. 3) May, 1984 Page 425

 O n substituting into Eq. 18,

n- I Equation 19 is the final working equation for Lewis Case 2. It may

be used to determine a for particular values of A,, E and Po. Before
dealing with how this is then used to determine efficiency, we will
derive equivalent equations for Lewis Case 3.

n Lewis Case 3
For this case, we define the following similarity relations fol-
lowing Rahman and Lockett (1981). The nomenclature is shown
in Figure 3.
n t l z n = Yn - (Ynh (20)
Z n + 1 = Yn+ 1 - (Yn+ 111
Figure 3. Nomenclature for Lewls Case 3 with weeping.
kn = ( Y n h - (Yn+ 111 (21)
L+1= ( Y n + 1 ) 1 - (Y~+z)o
@n)w - kn
(22)
a=(Zn+~)l-w kn+l
The basic differential-difference equation, Eq. 12, also holds for
Differentiating Eq. 9 with respect t o w , -
this case and substituting Eqs. 20-22 gives:

d ~ i - 1 d ~ n( l i E ) d ~ n + i
a(Em - Ef + 2f) ( z n ) w - (Em-Ef + f + a 2 (zn11-w
dw E d w dw m j
'
)
Substituting Eq. 11 into Eq. 10 ,

Equation 23 can be written as

This equation corresponds to Lewis' Eq. 3 when f = 0 (Lewis,

1936).
Following Lewis we define the usual similarity relations,
Zn = Yn - ( Y n h (13)
Zn+ 1 = Yn+ 1 -(Yn+ 110 Where

kn =(~n)o-(Yn+I)o (14) A=
a(Em - Ef + 2f)
m
k n + l = (Yn+I)o-(Yn+z)o Em - Ef + f + a2f
B= (25)
Because of the similarity of concentration profiles on adjacent trays m
we also have
a = z n / Z n + 1 = kn/kn+ 1 (15)
On substituting Eqs. 13-15 into Eq. 12 and simplifying we find, Following Lewis, Eq. 24 may be solved by shifting the origin from
w = 0 to w = 0.5. Thus setting s = w - 0.5 and

Integrating Eq. 16 from w = 0, Zn = 0 t o w = w , Z, = Zn gives

after rearranging
Equation 24 becomes
+ kn-1 - ( a - l)X(Em - Ef + f - a
In [(a- 1)Zn
kn-1 1- m ( a - 1 + E)
f ) ~

(17)
We now assume a solution to Eq. 27 of the form,
Lewis (1936)showed that for parallel liquid flow k, = (Zn+ 1)1. The
same relation has been shown to hold in the presence of entrain- u = a (sin(X8s) - y cos(h8s)) (28)
ment (Lockett e_tal., 1983). An exactly similar proof can be used
to show that it also holds in the presence of weeping. where a is an arbitrary coefficient.
Consequently, we have k,-1 = (Zn)l so that Eq. 17 becomes Equation 28 is substituted into Eq. 27 and setting the coefficients
m(a - 1 + E )
= [(a- 1)(Em - f ( a
]
- 1 + E ) ) lna (18)
of the trigonometric functions to zero, so that the equation is sat-
isfied for all s, enables 8 and y to be determined. The remainder
of the mathematical development exactly parallels that given
It is generally more convenient to work in terms of the external previously for entrainment (Rahman and Lockett, 1981).The final
liquid flow rate, so that with Po = L,/L, we have X = XJ(1 - Po) working equation, after converting to the external liquid flow rate
and f = mP,/Xo as a basis, is

Page 426 May, 1984 AlChE Journal (Vol. 30,, No. 3)

 TABLE 1. APPARENT TRAYEFFICIENCY
E&,FOR LEWISCASE 1

Po =0 Po = 0.1
PdAo 0.5 1.0 2.0 3.0 PdAO 0.5 1.0 2.0 3.0
E = 0.4 E = 0.4
0 0.40 0.40 0.40 0.40 0 0.40 0.40 0.40 0.40
2 0.42 0.44 0.48 0.52 2 0.42 0.43 0.47 0.50
10 0.43 0.47 0.56 0.66 10 0.43 0.46 0.53 0.58
20 0.44 0.48 0.58 0.71 20 0.44 0.47 0.54 0.61
1,000 0.44 0.49 0.61 0.77 1,000 0.44 0.48 0.56 0.64
E = 0.6 E = 0.6
0 0.60 0.60 0.60 0.60 0 0.60 0.60 0.60 0.60
2 0.64 0.68 0.77 0.87 2 0.63 0.67 0.74 0.81
10 0.68 0.77 0.99 1.28 10 0.67 0.74 0.88 1.00
50 0.70 0.81 1.12 1.57 50 0.68 0.77 0.95 1.10
1,000 0.70 0.82 1.16 1.68 1,000 0.68 0.78 0.97 1.12
E = 0.8 E = 0.8
0 0.80 0.80 0.80 0.80 0 0.80 0.80 0.80 0.80
2 0.87 0.95 1.12 1.31 2 0.86 0.93 1.05 1.16
10 0.94 1.12 1.57 2.20 10 0.92 1.06 1.30 1.48
50 0.97 1.20 1.86 2.98 50 0.94 1.12 1.42 1.61
Loo0 0.98 1.22 1.97 3.32 1,000 0.95 1.14 1.46 1.64
E = 1.0 E = 1.0
0 1.00 1.00 1.00 1.00 0 1.00 1.00 1.00 1.00
2 1.11 1.24 1.51 1.84 2 1.09 1.20 1.39 1.55
10 1.23 1.52 2.32 3.55 10 1.18 1.41 1.77 1.98
50 1.28 1.67 2.94 5.38 50 1.22 1.51 1.95 2.11
1.30 1.72 3.18 6.30 1,000 1.23 1.55 2.00 2.14

Po = 0.3 Po = 0.5
Pe/A, 0.5 1.0 2.0 3.0 Pe/Ao 0.5 1.0 1.5 2.0
E = 0.4 E = 0.4
0 0.40 0.40 0.40 0.40 0 0.40 0.40 0.40 0.40
2 0.41 0.42 0.44 0.46 2 0.41 0.41 0.42 0.42
10 0.42 0.44 0.47 0.49 10 0.41 0.42 0.43 0.43
20 0.42 0.44 0.48 0.50 20 0.41 0.42 0.43 0.43
1,000 0.43 0.45 0.48 0.50 1,m 0.41 0.42 0.43 0.43
E = 0.6 E = 0.6
0 0.60 0.60 0.60 0.60 0 0.60 0.60 0.60 0.60
2 0.63 0.65 0.69 0.71 2 0.62 0.63 0.64 0.64
10 0.65 0.69 0.74 0.77 10 0.63 0.64 0.65 0.66
50 0.66 0.70 0.76 0.77 50 0.63 0.65 0.66 0.66
1,000 0.66 0.71 0.77 0.78 1,OO0 0.63 0.65 0.66 0.65
E = 0.8 E = 0.8
0 0.80 0.80 0.80 0.80 0 0.80 0.80 0.80 0.80
2 0.84 0.88 0.94 0.98 2 0.83 0.85 0.86 0.87
10 0.89 0.95 1.03 1.05 10 0.85 0.87 0.88 0.88
50 0.90 0.98 1.05 1.04 50 0.86 0.88 0.88 0.87
1,000 0.91 0.99 1.05 1.04 1,000 0.86 0.88 0.88 0.79
E = 1.0 E = 1.0
0 1.00 1.00 1.00 1.00 0 1.00 1.00 1.00 1.00
2 1.07 1.13 1.21 1.26 2 1.04 1.07 1.09 1.10
10 1.13 1.23 1.32 1.31 10 1.08 1.11 1.11 1.11
50 1.16 1.28 1.34 1.27 50 1.09 1.12 1.11 1.08
1,000 1.17 1.29 1.34 1.25 1,000 1.09 1.12 1.10 1.03

Eq. 27 may be obtained by assuming a solution of the form

u = c (Cinh (A&) - y cosh (Ah)). (30)

(29)

Equation 29 can be used to determine a for specified values of E,

Po and A,. It is used when a (and A,) < 1.
To avoid imaginary numbers when a > 1, another solution to An exactly parallel development leads to Eq. 31

AlChE Journal (Vol. 30, No. 3) May, 1984 Page 427

 TABLE2. APPARENT TRAYEFFICIENCY
E'&yFOR LEWISCASE 2

Po =o Po = 0.1
wx, 0.5 1.0
E = 0.4
2.0 3.0 pe/ X, 0.5 1.0
E = 0.4
2.0 3.0

0 0.40 0.40 0.40 0.40 0 0.40 0.40 0.40 0.40

2 0.42 0.44 0.48 0.52 2 0.42 0.44 0.47 0.51
10 0.44 0.48 0.57 0.68 10 0.43 0.47 0.55 0.63
20 0.44 0.49 0.60 0.74 20 0.44 0.49 0.58 0.66
1,m 0.45 0.50 0.63 0.81 1,o@J 0.44 0.49 0.59 0.71
E = 0.6 E =0.6
0 0.60 0.60 0.60 0.60 0 0.60 0.60 0.60 0.60
2 0.64 0.69 0.79 0.89 2 0.64 0.68 0.76 0.85
10 0.69 0.80 1.05 1.37 10 0.68 0.77 0.96 1.16
50 0.71 0.84 1.20 1.72 50 0.70 0.81 1.06 1.34
1,o@J 0.71 0.86 1.25 1.85 1,OOo 0.70 0.82 1.10 1.40
E = 0.8 E = 0.8
0 0.80 0.80 0.80 0.80 0 0.80 0.80 0.80 0.80
2 0.88 0.97 1.15 1.36 2 0.87 0.95 1.10 1.25
10 0.98 1.19 1.73 2.46 10 0.96 1.14 1.50 1.88
50 1.02 1.30 2.11 3.45 50 1.00 1.23 1.73 2.26
1.03 1.33 2.25 3.89 1,m 1.00 1.25 1.80 2.38
E = 1.0 E = 1.0
0 1.00 1.00 1.00 1.00 0 1.00 1.00 1.00 1.00
2 1.13 1.28 1.59 1.95 2 1.12 1.24 1.49 1.73
10 1.32 1.70 2.68 4.14 10 1.29 1.58 2.19 2.79
50 1.41 1.93 3.55 6.58 50 1.37 1.76 2.61 3.44
1,m 1.43 2.00 3.90 7.82 1,000 1.39 1.82 2.75 3.65

Po = 0.3 Po = 0.5
Pe/ A, 0.5 1.0 2.0 3.0 WX, 0.5 1.0 1.5 2.0
E = 0.4 E =0.4
0 0.40 0.40 0.40 0.40 0 0.40 0.40 0.40 0.40
2 0.41 0.43 0.45 0.47 2 0.41 0.42 0.43 0.43
10 0.43 0.45 0.50 0.55 10 0.42 0.44 0.45 0.47
20 0.43 0.46 0.52 0.56 20 0.42 0.44 0.46 0.47
l,OOo 0.43 0.47 0.53 0.59 1,m 0.42 0.44 0.46 0.48
E = 0.6 E = 0.6
0 0.60 0.60 0.60 0.60 0 0.60 0.60 0.60 0.60
2 0.63 0.66 0.72 0.77 2 0.62 0.64 0.66 0.68
10 0.66 0.73 0.83 0.93 10 0.64 0.68 0.72 0.75
50 0.68 0.75 0.89 1.00 50 0.65 0.70 0.74 0.77
1,m 0.68 0.76 0.90 1.02 1,o@J 0.66 0.71 0.75 0.78
E = 0.8 E =0.8
0 0.80 0.80 0.80 0.80 0 0.80 0.80 0.80 0.80
2 0.86 0.91 1.01 1.09 2 0.84 0.88 0.91 0.93
10 0.93 1.04 1.23 1.37 10 0.89 0.96 1.01 1.05
50 0.95 1.10 1.32 1.50 50 0.91 0.99 1.05 1.10
1,000 0.96 1.11 1.35 1.53 1,m 0.91 1.00 1.06 1.11
E = 1.0 E = 1.0
0 1.00 1.00 1.00 1.00 0 1.00 1.00 1.00 1.00
2 1.09 1.17 1.33 1.45 2 1.07 1.12 1.17 1.21
10 1.23 1.40 1.68 1.87 10 1.14 1.26 1.33 1.39
50 1.29 1.51 1.83 2.06 50 1.18 1.32 1.40 1.46
1,00o 1.30 1.54 1.88 2.11 1,000 1.19 1.33 1.42 1.48

r
x, =

tanh

Equation 31 is used to determine a when a (and A,) 3 1.

Calculation of Apparent Tray Efficiency CY-1

E&" = - (321
When cy has been determined from Eq. 19, Eq. 29 or 31 as ap- x, - 1
propriate, the apparent Murphree tray efficiency EMv may be Equation 32 wasoriginally derived by Lewis (1936) in the absence
determined from the following equation of weeping or entrainment. It has been proved to hold in the

Page 428 May, 1984 AlChE Journal (Vol. 30,No. 3)

 TABLE3. APPARENT TRAYEFFICIENCY FOR LEWISCASE 3

Po =0 pa = 0.1
Pe/b 0.5 1.0 2.0 3.0 PdX, 0.5 1.0 2.0 3.0
E = 0.4 E =0.4
0 0.40 0.40 0.40 0.40 0 0.40 0.40 0.40 0.40
2 0.42 0.44 0.47 0.51 2 0.42 0.43 0.46 0.49
10 0.43 0.47 0.55 0.65 10 0.43 0.46 0.52 0.57
20 0.44 0.48 0.57 0.69 20 0.43 0.47 0.53 0.60
1,m 0.44 0.49 0.60 0.75 1,OOO 0.44 0.48 0.55 0.62
E =0.6 E = 0.6
0 0.60 0.60 0.60 0.60 0 0.60 0.60 0.60 0.60
2 0.64 0.68 0.76 0.85 2 0.63 0.67 0.73 0.79
10 0.68 0.76 0.95 1.20 10 0.67 0.73 0.85 0.96
50 0.69 0.80 1.06 1.44 50 0.68 0.76 0.91 1.03
1,oOO 0.70 0.81 1.10 1.53 0.68 0.77 0.93 1.05
E = 0.8 E = 0.8
0 0.80 0.80 0.80 0.80 0 0.80 0.80 0.80 0.80
2 0.87 0.94 1.09 1.26 2 0.86 0.92 1.02 1.12
10 0.93 1.07 1.44 1.94 10 0.91 1.02 1.21 1.36
50 0.96 1.14 1.65 2.49 50 0.93 1.07 1.29 1.43
1,000 0.97 1.16 1.73 2.72 1,m 0.94 1.08 1.32 1.44
E = 1.0 E = 1.0
0 1.00 1.00 1.00 1.00 0 1.00 1.00 1.00 1.00
2 1.10 1.20 1.44 1.72 2 1.08 1.17 1.33 1.48
10 1.17 1.39 1.98 2.89 10 1.14 1.30 1.57 1.74
50 1.20 1.47 2.33 3.95 50 1.17 1.35 1.65 1.75
1,000 1.21 1.50 2.46 4.43 1,m 1.17 1.36 1.64 1.72

Do = 0.3 Po = 0.5
Pe/X, 0.5 1.0 2.0 3.0 Pe/ x, 0.5 1.0 1.5 2.0
E = 0.4 E = 0.4
0 0.40 0.40 0.40 0.40 0 0.40 0.40 0.40 0.40
2 0.41 0.42 0.44 0.45 2 0.41 0.41 0.42 0.42
10 0.42 0.44 0.46 0.48 10 0.41 0.42 0.42 0.43
20 0.42 0.44 0.47 0.49 20 0.41 0.42 0.43 0.43
1,OOO 0.43 0.45 0.48 0.49 0.41 0.42 0.43 0.43
-
E =-
0.6 E = 0.6
0 0.60 0.60 0.60 0.60 0 0.60 0.60 0.60 0.60
2 0.62 0.65 0.68 0.71 2 0.62 0.63 0.63 0.64
10 0.64 0.68 0.72 0.74 10 0.62 0.64 0.64 0.64
50 0.65 0.69 0.73 0.74 50 0.63 0.64 0.64 0.63
1,OOO 0.66 0.70 0.74 0.73 1,000 0.63 0.64 0.63 0.62
E =0.8 E =0.8
0 0.80 0.80 0.80 0.80 0 0.80 0.80 0.80 0.80
2 0.84 0.88 0.93 0.96 2 0.83 0.84 0.85 0.86
10 0.87 0.92 0.97 0.98 10 0.84 0.85 0.85 0.84
50 0.89 0.94 0.97 0.93 50 0.84 0.84 0.82 0.79
1,OOO 0.89 0.94 0.96 0.91 1,m 0.84 0.84 0.82 0.76
E = 1.0 E = 1.0
0 1.00 1.00 1.00 1.00 0 1.00 1.00 1.00 1.00
2 1.06 1.11 1.18 1.22 2 1.03 1.05 1.07 1.08
10 1.09 1.15 1.20 1.18 10 1.04 1.05 1.04 1.02
50 1.10 1.16 1.15 1.06 50 1.03 1.02 0.98 0.92
1,m 1.10 1.15 1.14 1.00 1.000 1.03 1.00 0.95 0.86

presence of entrainment (Rahman and Lockett, 1981; Lockett et Figure 4 shows the effect of weeping fraction for typical values
al., 1983), and an exactly similar proof shows it to hold also for of E , A, and Pe found in distillation. It is assumed here that the
weeping. It was in fact used previously by Kageyama (1966)for point efficiency does not change as weeping increases. It is clear
Lewis Case 1 weeping. that the efficiency advantage of Lewis Case 2 is maintained over
the other two cases even in the presence of substantial weeping.
Figure 5 shows the effect of Peclet number, and indirectly of
TABULATED RESULTS tray size, when weeping occurs. For a small tray, or a short flow
path length, giving a low Peclet number, weeping causes only a
The numerical and analytical solutions were used to construct small reduction in tray efficiency. This can be understood by
Tables 1-3. These show the apparent tray efficiency E& as a considering the limiting case of a perfectly mixed tray (Pe = 0).
function of the Lewis case, point efficiency E , stripping factor A,, It is then immaterial how the liquid on the tray progresses to the
liquid Peclet number and fractional weeping Po. tray below. Both weeping liquid and liquid passing down the
downcomer have the same concentration; so in this case weeping
causes no reduction in efficiency. This is the main difference be-
DISCUSSION tween the effect of entrainment and weeping on tray efficiency.
Entrainment causes a reduction in tray efficiency even for a per-
The effect of each of the five variables considered on the ap- fectly mixed tray which can be calculated from Colburn’s (1936)
parent tray efficiency is shown graphically in Figures 4-7. equation.

AlChE Journal (Vol. 30, No. 3) May, 1984 Page 429

 1.3.

1.2 -
WEEPING FRACTION = 0

/
1
>
ox
W 1.1 -
CASE 2
1.0 *
>
0s
W
0.9- CASE I

CASE 3

0.74
0 0.1 0.2 0.3 0.4 0.5

Figure 4. Apparent tray efficiency vs. weeping fraction for E = 0.8, A,, =
1.0 and Pe = 50.

1.4-

-
/
LEWIS CASE 2 0 I I I I I I
1.3 - 0 I 2 3 4 5

LO
Flgure 6. Apparent tray efficiency vs. stripping factor for Lewis Case 3, E
= 0.4, Pe = 00.

///A LEWIS CASE I 8. - 0.1

-
A// / LEWIS CASE 3 8. - 0.3 1.6

1.4 - 8,- 0.1

PI
1.2 - 8, = 0.3
Figure 5. Apparent tray efficiency vs. Peclet Number for A, = 1.0, E =
0.8. >
1.0 - so= 0.5
The total reduction in tray efficiency caused by weeping will
generally be larger than that shown in Figures 4 and 5, because they 0.8 -
are plotted for a constant value of point efficiency E . Since weeping
reduces the flow rate of liquid flowing over the weir, the froth
height is reduced and so also is the point efficiency. The extent of 0.6 -
the latter depends on the relative magnitudes of the liquid crest
height over the weir and the weir height, and so will vary from case
to case. To take this into account the calculated results given in 0.4-
Tables 1-3 must be combined with a model to predice point effi-
ciency under weeping conditions.
0.2 I I
Figure 6 shows the effect of large values of the stripping factor 1 I
0 0.2 0.4 0.6 0.8 1.0
such as may be encountered in absorption or stripping. It shows
that, as the stripping factor increases, the penalty paid in efficiency E
from only a small amount of weeping becomes severe. As weeping Figure 7. Apparent tray efflciency vs. point efficiency for Lewis Case 3, A,
increases, the tray efficiency is approximately equal to the point = 1.0. Pe = 50.
efficiency.
Figure 7 shows that, under distillation conditions, weeping has
only a small effect when the point efficiency is low. Trays designed Even so,a tray designed to exploit enhancement of tray efficiency
for high point efficiency are more severely affected by weeping. over point efficiency will still give a higher efficiency than a point
In fact all these results show that trays designed to give a high tray efficiency tray even under weeping conditions.
efficiency, by exploiting crossflow enhancement of tray efficiency To use the above models it is necessary to predict the weeping
over point efficiency, are particularly vulnerable to a failure to fraction. Existing correlations in the literature leave something to
achieve that enhancement as a result of weeping. A tray which is be desired and we are presently concluding a comprehensive ex-
only essentially a point efficiency tray can tolerate much more perimental study on this aspect of the problem. Most designers in
weeping than one designed to achieve maximum tray efficiency. practice have access to proprietary correlations, however, such as

Page 430 May, 1984 AlChE Journal (Vol. 30, No. 3 )

that of FRI, so this should not prove to be a barrier to incorporation a = similarity ratio, Eq. 15 or 22
of this present work into tray design methods. P = Lw/L
P O = Lw/Lo
r,8 = constants in Eqs. 28 and 30
NOTATION x = stripping factor mV/L
A0 = stripping factor mV/Lo
= variable defined by Eq. 25 PF = holdup fraction of liquid in fro1 (m3 liquid/m3
= coefficient in Eqs. 28 and 30 froth)
= variable defined by Eq. 25 p;. = clear liquid density ( k m ~ l - m - ~ )
= constant in equilibrium line equation
= variable defined by Eq. 25
Superscripts
= eddy diffusion coefficient of mixing (rn2-s-l)
= Murphree vapor-phase point efficiency - = mean value
= apparent Murphree vapor-phase tray efficiency
= reduced Murphree vapor-phase tray efficiency
Subscripts
= weeping fraction based on vapor flow Lw/V
= froth height (m) n = on or leaving tray n
= defined by Eq. 14 or 21 N = on or leaving bottom tray
= liquid flow rate on tray in presence of weeping = Lo R = leaving reboiler
- L, (kmo1.s-1)
=liquid flow rate on tray in absence of weeping
(kmol-s-I) LITERATURE CITED
= weeping liquid flow rate (kmo1.s-I)
= slope of equilibrium line Colburn, A. P., “Effect of Entrainment on Plate Efficiency in Distillation,”
= liquid Peclet number Ind. Eng. Chem., 28, p. 526 (1936).
-
= w 0.5 Danckwerts, P. V., “Continuous Flow Systems. Distribution of Residence
= function defined by Eq. 26 Times,” Chem. Eng. Sci., 2, p. 1 (1953).
= vapor flow rate (kmol-s-l) Haselden, G. G., K. S. Teo, and M. J. Markwell, “A Cartridge Tray Assembly
= tray width normal to liquid flow (m) for Upgrading the Performance of Small Industrial Columns,” I. Chem.
= position variable on tray defined in Figures 2 and 3 E . Symp. Ser., No. 73, p. D13 (1982).
Jenkins, A. E. O., “The Design of Distillation Columns ‘for the Times,’ ”
= liquid concentration (mole fraction)
I. Chem. E . Symp. Ser., No. 61, p. 73 (1981).
= mean concentration of weeping liquid tray n Kageyama, O., “Tray Efficiency of Distillation Tower with Weeping,”
= mean concentration of liquid leaving tray n via the Kuguku Kogaku, 30, p. 629 (1966).
downcomer Kageyama, O., “Plate Efficiency in Distillation Towers with Weeping and
= reduced liquid concentration leaving tray n Entrainment,” I. Chem. E . Symp. Ser., No. 32,2:72 (1969).
= liquid concentration in equilibrium with vapor of Lewis, W. K., “Rectification of Binary Mixtures: Plate Efficiency of Bubble
concentration yn Cap Columns,” Ind. Eng. Chem., 28, p. 399 (1936).
= liquid concentration at inlet of tray n Lockett, M. J., M. A. Rahman, and H. A. Dhulesia, “The Effect of En-
= liquid concentration at inlet weir of tray n trainment on Distillation Tray Efficiency,” 38, p. 661 (1983).
O’Brien, N. G., “Calculation of Effect of Liquid Leakage on Tray Effi-
= apparent vapor concentration ciency and Tray Requirements,” Ind. Eng. Chem. Proc. Des. & Dev.,
= vapor concentration (mole fraction) 5, No. 4, p. 383 (1966).
= vapor concentration in equilibrium with liquid of Rahman, M. A., and M. J. Lockett, “Efficiency Reduction of Large Dis-
concentration x tillation Trays by Entrainment,” I. Chem. E . Symp. Ser., No. 61, p. 111
= vapor concentration in equilibrium with liquid con- (1981).
centration (Xi)out Smith, V. C., and W. V. Delnicki, “Optimum Sieve Tray Design,” Chem.
= vapor concentrationleaving points 0 and 1 respectively Eng. Prog., 71, No. 8, p. 68 (1975).
on tray n Standart, G., and F. Kastanek, “The Effect of Carryover on Plate Effi-
= defined by Eqs. 13 and 20 ciency,” Separation Sci;, 1, (l),p. 27 (1966).
= liquid flow path length (m)
= dimensionless distance d / Z n Manuscript received February 9,1983; revision received May 31, and accepted June
2’ = distance from inlet weir (mj 10, 1983.

AlChE Journal (Vol. 30, No. 3) May, 1984 Page 431