Está en la página 1de 10

06-202 Unit Operations Spring, 2000

Key to Homework #8

1a.) (Prob. 11.7-4 from Geankoplis) The table below summarizes product distribution from the
partial solution obtained in last week’s assignment:

Component FzFi zFi DxDi xDi BxBi xBi


1 40 0.40 40.00 0.6178 0.00 0.0000
2 25 0.25 23.75 0.3668 1.25 0.0355 light key
3 20 0.20 1.00 0.0154 19.00 0.5390 heavy key
4 15 0.15 0.00 0.0000 15.00 0.4255
totals = 100 1 64.75 1 35.25 1

1b.) The top temperature of the column is the dew point of the distillate product. Choosing yi =
xDi, we invert (11.7-1) to obtain
P
xi = o yi
Pi T bg
The we vary T by trial-and-error until ∑ xi = 1. The total pressure P is given as 405.3
kPa = 4 atm = 3040 mmHg. Using the empirical equations for the vapor pressures Pio T bg
given in the problem statement, we find the dew point to be 65.74°° C. The results for this
final iteration are given in the table below.

i yi = x Di
e
Pio 65.74 o C j xi =
P
yi αi =
Pio
Pio o
PHK
(atm)
1 0.6178 7.262 0.3404 8.041
2 0.3668 2.482 0.5913 2.748
3 0.0154 0.903 0.0682 1.000
4 0.0000 0.354 0.0000 0.392
totals = 1.0000 0.9999

To obtain the temperature at the bottom of the column, we assigned the bottoms
composition to the liquid mole fraction: xi = xBi, then calculated the vapor mole fractions
from

yi = xi
Pio T bg
P

The temperature is varied until ∑ yi = 1. The bubble point turns out to 130.3°° C. The
results for this final iteration are given in the table below, along with α which is average
of the relative volatilities at 65.75 and 130.3°C.
06-202 page 2 on HWK #8 Spring, 2000

i xi = x Bi e
Pio 130.3o C j Po
yi = xi i αi =
o
Pio
α = α Dα B
(atm) P PHK
1 0.0000 25.881 0.0000 5.293 6.524
2 0.0355 11.063 0.0982 2.262 2.494
3 0.5390 4.890 0.6589 1.000 1.000
4 0.4255 2.284 0.2429 0.467 0.428
totals = 1.0000 1.0000

1c.) To calculate the minimum number of stages, we use Fenske’s equation (11.7-12). To use
this equation, we need to average the reflux ratio over the range of temperatures found in
our column. From (11.7-13):

α LK , HK = 2.748 × 2.262 = 2.493

Substituting into Fenske’s equation:

F x Di x Bi I F 0.3668 0.0355I
ln GH x Dj xBj JK lnGH 0.0154 05390
.
JK
N min = = = 6.45
lndαij i lnb2.493g

ANSWER: 6.45 theoretical stages (including reboiler)

In calculating our product distribution in part a), we assumed that all components heavier
than the heavy key were negligible in the distillate stream. In this problem, we assume that
heptane was negligible. Let’s check that assumption by using Fenske’s equation to
calculate the pentane mole fraction in the distillate stream. We continue to use hexane (the
heavy key) as the reference for volatility, so we must also use it as the “j” in Fenske’s
equation. Solving for xDi:

d iN d i6.45 005390
x Dj .0154
x Di = α ij x Bi = 0.392 × 0.467 . × 10 −5
0.4255 = 51
min

x Bj .

ANSWER: xD,heptane = 5.1×10-5

Similarly, we neglecting all components lighter than the light key in the bottoms. In
particular, we neglected butane. So let’s go back and calculate it from Fenske’s equation.
Solving the last equation for xBi:

x Di 0.6178
x Bi = = . × 10 −4
= 12
dα ij i N min
x Dj
x Bj
d 8.041 × 5.293 i 6.45 0.0154
0.5390

ANSWER: xB,butane = 1.2×10-4

Both of these mole fractions are too small to change the temperatures.
06-202 page 3 on HWK #8 Spring, 2000

1d.) Using Underwood’s method, we first determine the root of (11.7-19):

α z
∑ α i −Fiφ = 1 − q
i i

Substituting in known values:

6.524 × 0.40 2.494 × 0.25 1 × 0.20 0.428 × 015


.
+ + + = 1− 1 = 0
6.524 − φ 2.494 − φ 1− φ 0.428 − φ
1444444444 424444444444 3
f φ bg
The root which lies in the interval 1 ≤ φ ≤ 2.494 is φ = 1.222. Once we have the value of
φ, we calculate the reflux ratio from (11.7-20):

α x
Rmin + 1 = ∑ i Di
i
αi − φ

Substituting in known values:

6524
. × 0.6178 2.494 × 0.3668 1 × 0.0154
Rmin + 1 = + + = 1410
.
6524
. − 1.222 2.494 − 1.222 1 − 1.222

where we have omitted components heavier than the heavy key from this sum. In this case,
including component 4 would not have changed the result to the accuracy shown.

ANSWER: Rmin = 0.41

1e.) To estimate the number of trays required at

R = 1.3Rmin = 0.544

To use the Erbar-Maddox correlation, we need to evaluate

R Rmin
= 0.348 and = 0.291
R+1 Rmin + 1

N min
Using Fig. 11.7-3, we read = 0.41
N

In the Erbar-Maddox correlation, N and Nmin refer to the total number of steps (including
the reboiler), so substituting Nmin = 6.45, we obtain, N = 15.7 total steps or

ANSWER: 14.7 stages + reboiler


06-202 page 4 on HWK #8 Spring, 2000

1f.) We first calculate the ratio of the number of enriching trays Ne to the number of stripping
trays Ns using (11.7-21):

LM F I 2 OP 0.206 = LM 0.20 3525


. F 0.0355 I 2 O
0.206

N 0.25 64.75 H 0.0154 K PQ = 1189


z F , HK B x B , LK
Ne
Ns
=
MN GH
z F , LK D x D, HK JK PQ . (1)

We also know the total number of stages:

Ne + Ns = 15.7 (2)

Solving (1) and (2) simultaneously:

ANSWER: Ne = 8.5 and Ns = 7.2

2a.) MS&H Problem 19.1. The table below summarizes product distribution from the partial
solution obtained in last week’s assignment:

Component FzFi zFi DxDi xDi BxBi xBi


1 5 0.05 5 0.1053 0 0
2 42 0.42 41.58 0.8754 0.42 0.0080 light key
3 46 0.46 0.92 0.0193 45.08 0.8587 heavy key
4 7 0.07 0 0 7 0.1333
totals = 100 1 47.50 1 52.50 1

To estimate the minimum number of trays, we can apply Fenske’s equation (11.7-12) to the
two key components (i = 2 and j = 3):

F x Di xBi I F 08754 0080 I


ln GH xDj x Bj JK lnGH 0..0193 00..8587 JK
N min − 1 = = = 16.00
lndα ij i lnb17
. 1g

ANSWER: 15.00 trays plus reboiler

xD,4 can be estimated from Fenske’s equation by using component 4 as i or j. Fenske’s


equation for i = 4 and j = 3 gives:

FG xD,4 01333
. IJ
16.00 =
ln
H 0.0193 08587
. K
lnb0.65 1g

or xD,4 = 3.0×10-6

Similarly we could find xB,1 = 3.3×10-5

Both of these are small compared to unity so neglecting them in the tables above did not
introduce significant error. If they turned out to be comparable to other mole fractions
06-202 page 5 on HWK #8 Spring, 2000

(particularly of the key components), we could substitute these numbers for xD,4 and xB,1 in
the table and iterate the calculation of Nmin.

2b.) Using Underwood’s method, we first determine the root of (11.7-19):

α z
∑ α i −Fiφ = 1 − q
i i

Substituting into known values:

2.1 × 0.05 17
. × 0.42 1 × 0.46 0.65 × 0.07
+ + + = 1−1= 0
2.1 − φ . −φ
17 1− φ 0.65 − φ
1444444442444444443
bg
f φ

This is a 4th order polynomial in φ. The 10


left-hand side f(φ) is plotted at right. This
1 1.7
equation has four singular points at f( φ)

φ = α1, α2, α3 and α4


0 1 q
These are shown as the 4 vertical lines in
the plot. Notice that there are also 3
different φ’s for which

f(φ) = 1-q = 0 10
2 0 2 4
φ
We are interested in the root in the
interval:

α HK < φ < α LK or 1 < φ < 1.7

which turns to be φ = 1.269. Once we have the value of φ, we calculate the reflux ratio
from (11.7-20):
α x
Rmin + 1 = ∑ i Di
i
αi − φ

Substituting in known values:

. × 01053
21 . . × 0.8754 1 × 0.0193
17
Rmin + 1 = + + = 3.65
21. − 1269
. . − 1269
17 . 1 − 1269
.

where we have omitted components heavier than the heavy key from this sum. In this case,
including component 4 would not have changed the result to the accuracy shown.

ANSWER: Rmin = 2.65


06-202 page 6 on HWK #8 Spring, 2000

2c.) To estimate the number of trays required at

R = 1.3Rmin = 3.44

we use Gilliland’s correlation (Fig. 19.5 of MS&H5). The abscissa is first calculated as

R − Rmin
= 0179
.
R +1

Using Fig. 19.5, we read the corresponding ordinate from the graph:

N − N min
= 0.46
N +1

Substituting Nmin = 15.0+1, we obtain N = 30.5

ANSWER: 29.5 stages + reboiler (Gilliland)

For the Erbar-Maddox correlation, we need to evaluate

R Rmin
= 0.775 and = 0.726
R+1 Rmin + 1

N min
Using Fig. 11.7-3, we read = 056
.
N

In the Erbar-Maddox correlation, N and Nmin refer to the total number of steps (including
the reboiler), so substituting Nmin = 15.0+1, we obtain, N = 28.6 total steps

ANSWER: 27.6 stages + reboiler (Erbar-Maddox)

3a.) A summary of the specifications supplied to ChemSep follows:


Column C:\MYDOCU~1\202S98\G11_7_4.SEP

Components:
n-Butane
n-Pentane
n-Hexane
n-Heptane

Operation:
Simple Distillation
Total (Liquid product) Condenser
Partial (Liquid product) Reboiler
100 Stages
Feeds to stages 50

>
1

< >
-- --- --
-- --- --
06-202 page 7 on HWK #8 Spring, 2000
-- --- --
> 50 -- --- --
-- --- --
-- --- --
-- --- --
-- --- --
<

> >
100

Properties:
Raoult's law K model
Antoine Vapour pressure
Ideal Enthalpy

Specifications:

Column pressure
Condenser pressure 4.000 (atm)
Top pressure 4.000 (atm)

Default efficiency = 1.000 ( )

Feed 1
Stage 50
Pressure (atm) 4.000
Vapour fraction ( ) 0.0000
Temperature ( C)
Component flows (mol/s)
n-Butane 40.00
n-Pentane 25.00
n-Hexane 20.00
n-Heptane 15.00

Condenser
n-Pentane recovery = 0.9500 ( )
Initialization guess: *

Reboiler
n-Hexane recovery = 0.9500 ( )
Initialization guess: *

Solve options:
Automatic Initialization
Newton's method
Flow Step limit 0.5000 ( )
Temperature Step limit 10.00 ( C)
Composition Step limit 1.000 ( )
Accuracy = 0.001
Maximum iterations = 30
Feeds type = Stage below

ChemSep converged in 22 iterations with an accuracy set at 0.001. ChemSep does not
report the reflux ratio directly, but it can be calculated from the flowrates which can be
tabulated. Below are the first few lines of the tables produced by selecting "Results-
Tables-T/P/Flow profiles":
Stage Temperature Pressure Flow rates (mol/s)
( C) (atm) Liquid Vapour Feed Product

1 53.40 4.000 28.72 64.75 64.75 L


2 65.37 4.000 26.92 93.47
3 69.98 4.000 26.21 91.67
06-202 page 8 on HWK #8 Spring, 2000
4 71.96 4.000 25.67 90.96
5 73.17 4.000 25.25 90.42

Generally the liquid and vapor flowrates on the line for stage n are the rates of the streams
leaving stage n, except that stage 1 refers to the condenser plus the stream splitter
following it. The liquid leaving "stage 1" is the reflux stream while the "vapor" leaving
stage 1 refers to the distillate product (regardless of whether it is vapor or liquid). Thus
the reflux ratio is

L 28.72
R= = = 0.444
D 64.75

This is reasonably close to the estimate of 0.41 produced by Underwood's method (Prob.
2d of Hwk#8). The same result (R = 0.444) was obtained for 120 trays, placing the feed
on the 60th tray; however Chemsep did not converge when a larger number of trays was
used. This lack of convergence with a large number of trays is probably caused with
round-off errors (near the “pinch” there is a larger number of trays which have virtually the
same composition).

3b.) The composition of the two product streams can be examined using the
Results/Tables/Streams command. The first few lines are:
Stream Feed 1 Top Bottom

Stage 50 1 100
Pressure (atm) 4.000 4.000 4.000
Vapour fraction (Ä) 0.0000 0.0000 0.0000
Temperature (øC) 67.82 53.40 129.5
Enthalpy (kJ/mol) -18.36 -18.10 -8.006

Mole flows (mol/s)


n-Butane 40.00 40.00 2.573E-21
n-Pentane 25.00 23.75 1.250
n-Hexane 20.00 1.0000 19.00
n-Heptane 15.00 7.613E-18 15.00

Total molar flow 100.0 64.75 35.25

Mole fractions (Ä)


n-Butane 0.4000 0.6178 7.300E-23
n-Pentane 0.2500 0.3668 0.03546
n-Hexane 0.2000 0.01544 0.5390
n-Heptane 0.1500 1.176E-19 0.4255

In the table below, we compare these results with those obtained in Prob. 2a, Hwk#8

xDi xBi
Component from Hwk#8 from Chemsep from Hwk#8 from Chemsep
1 0.6178 0.6178 0.0000 7.300E-23
2 0.3668 0.3668 0.0355 0.03546
3 0.0154 0.01544 0.5390 0.5390
4 0.0000 1.176E-19 0.4255 0.4255

The results for the stream compositions are virtually identical. This shouldn’t be too
surprising since the main assumption of Prob. 2a, Hwk#8 is that Component 4 is negligible
06-202 page 9 on HWK #8 Spring, 2000

in the distillate and Component 1 is negligible in the bottoms. These assumptions are met
by the solution which Chemsep has obtained. The rest of the compositions follow from
mass balances. So agreement is expected.

3c.) Below is the Lotus ® WK1 file (blue font) produced by Chemsep after loading into Excel.
We have hidden rows 7-104 and columns B-N and then added the text (in black font) to
calculate the relative volatilities (α’s).

We calculated the relative volatilities by dividing the K-value for any component by the K-
value for hexane (the reference component). Below is a table comparing the α’s calculated
from the Chemsep results with those reported in Prob. 2b, Hwk#8.

α = α Dα B
i Chemsep Hwk #8
1 6.524 6.910
2 2.494 2.552
3 1.000 1.000
4 0.428 0.410

With the exception of the reference component (3), all the other relative volatilities are
closer to unity in Chemsep than the values used in Hwk#8. This means that Chemsep
thinks the separation is more difficult than we did in Hwk#8. This probably explains why
Chemsep came up with a higher minimum reflux ratio in part a).

Regarding the constancy of α, Chemsep had α13 varying between 5.259 and 9.079 whereas
the corresponding range from Hwk#8 was between 5.293 and 8.041. Neither interval is
very narrow, suggesting the α is really not that constant.
06-202 page 10 on HWK #8 Spring, 2000

3d.) At right is the plot of liquid and vapor flowrates


(horizontal axis) as a function of stage number
(vertical axis). Near the feed, distillate or
bottoms, the flowrate within either rectifying or
stripping section change by 10-20%, but this is not
very much. Thus equimolal overflow is probably
a good assumption.

3e.) Below are plotted the temperature and liquid


composition profiles produced by Chemsep. Note
that there a larger number of adjacent trays having
virtually the same temperature or composition.

Indeed there are two such regions — one in the rectifying section and another in the
stripping section — in which there is virtually no change from one stage to the next. These
are “pinch points”. Recall in the McCabe-Thiele method that (at the minimum reflux ratio)
if the operating line almost touched the equilibrium curve, the steps became very small.
This is what is happening in this more complicated 4-component case.

3f.) The resulting T’s are summarized in the following table. The feed location does not seem
to affect the T’s at all.

feed location rectifying T (°C) stripping T (°C)


30 75.97 95.83
50 75.97 95.83
65 75.97 95.83

3g.) Chemsep stopped converging (in 30 steps) at 60 stages. However, if you move the feed up
toward the top of the column, you can get it to converge for fewer total stages. The
minimum number of stages at which I could get it to converge was for 8 total stages, with
the feed at stage 2.

También podría gustarte