Geoffrey Widdison
1656 Sunnyside Avenue
Salt Lake City, Utah, 84105
Dr. Terry Ring
Department of Chemical Engineering
University of Utah
Salt Lake City, Utah, 84102
Dear Dr. Ring,
On October 18, 2005, you assigned the members of project group E the task of
operating the distillation column located in the MEB Senior Lab. We were asked to use
the column to separate a water-isopropanol mixture under conditions of total reflux, as
well as under continuous feed with a reflux ratio of twice the minimum. We were asked
to determine the efficiency of each tray, and to make recommendations for the operation
of the column.
We determined that measured efficiency rates within the column varied widely
within the column and from one set of operating conditions to another. The minimum
reflux ratio was found to be 1.11. The average efficiency of the column was found to be
approximately 31.1% under total reflux conditions and 23.84% with a reflux ratio twice
the minimum. The values of individual trays, however, ranged from near 100% to
approximately 1%. While this may be partly due to defects and variation within the
column, it seems important to reexamine the methods used to measure composition at
each tray.
The lack of a vapor sampling port on each tray inhibited the data range we were
able to gather, installation of additional vapor ports could increase the accuracy of models
developed for the separation process. Syringes that form better connections with the
liquid sampling ports could also facilitate the testing process.
Examination of the separation data leads us to conclude that the ninth tray is at or
near the optimal feed location for most operating conditions, though additional testing
may be desired to confirm this. The range of conditions we used are near to those used to
separate a number of mixtures. The information on tray and column efficiency, and the
related information concerning column capacity and flow rates make it possible for a
wide variety of separations to be modeled, the reaction of this specific column to be
predicted.
Sincerely,
Geoffrey Widdison
OPERATION OF A BUBBLE-CAP CONTINUOUS DISTILLATION COLUMN
By
Geoffrey Widdison
Project No. 2M
Projects Laboratory I
Assigned: October 18, 2005
Due: November 15, 2005
Submitted: November 16, 2005
Project Team Members for Group E:
Brandon Grimm
Stanley Holbrook
Geoffrey Widdison
Geoffrey Widdison
INTRODUCTION
According to Seader (1998), distillation processes are a major component of
chemical processing with a great deal of scientific and economic significance. A large
amount of industrial distillation takes place in staged columns. Each stage consists of a
tray designed to contact rising vapor with descending liquid, which increases the vapor
2
composition of the more volatile component. This process allows the separation of a
wide variety of mixtures into separate species.
The relationship between the configuration of a distillation column and the degree
of separation achieved can be predicted using a combination of theoretical and
empirically derived models. Since complete equilibrium is rarely achieved in any stage
in distillation processes, experimental data must be gathered in order to determine the
efficiency of the column.
By collecting and analyzing samples from different stages, the efficiency of the
distillation operation can be calculated. These calculations can be used to confirm the
mathematical model for the distillation operation, and modify it for the specific
equipment being used.
THEORY
The mathematical basis for staged distillation columns is well established and
presented in a number of texts. The following treatment is adapted from Seaders (1998).
The separation dynamics are most heavily dependent on the vapor pressures of the
species in the column. The more volatile component is also known as the light key, the
less volatile component is also known as the heavy key. For a two-component
separation:
α=KL/KH.
(1)
KL= PL/P.
(2)
KH= PH/P.
(3)
3
All symbols are defined in the nomenclature section at the end of the text. By Raoult’s
law, this means that relative volatility can be described in terms of mole fractions in the
liquid and vapor phases.

yL / xL
yL (1  xL )

.
yH / x H x L (1  yL )
(4)
( x L )
.
1  x L (  1)
(5)
This can be rearranged to give
yL 
The traditional approach to modeling distillation processes is known as the McCabeThiele approach. This model requires making the following assumptions:

That the relative volatility is constant over the temperature range in the column.

That the components have equal and constant molar enthalpy.

That enthalpy changes and heat of mixing are negligible.

That the pressure is uniform within the column.
These assumptions allow equations to be derived relating flow rates to molar
compositions in each stage of the column. The column is divided into a stripping section,
in which the more volatile component is selectively driven into the vapor phase, and the
rectification section, in which the less volatile component is selectively condensed. In
each section, the liquid and vapor compositions can be related to flow rates.
Rectification section:
Vn+1yn+1=Lnxn+DxD.
(6)
yn+1=(Ln/ Vn+1)+(D/ Vn+1) xD.
(7)
Where the stages are numbered consecutively, beginning with the top stage.
L/V=R/(R+1).
(8)
4
 R 
 1 
yn  
 xn  
x
 R  1
 R  1 D .
(9)
Stripping section:
yn=(Ln/Vn)xn-(Bn/Vn)xB.
(10)
VB=V/B
(11)
L V  B VB  1


.
V
V
VB
(12)
 VB  1
 1
yn  
 xn    x B .
 VB 
 VB 
(13)
Given equilibrium data for the components of the distillation, these equations can
be used to model the stages graphically, as shown in Figure 1.
5
Figure 1. Sample McCabe-Thiele diagram. Each step indicates a theoretical stage
going to complete equilibrium. The calculations are shown for the rectification
section only. Reprinted from Seader (1998).
Because true vapor-liquid equilibrium is unlikely to be achieved on each stage,
this method must be adapted in order to properly model the distillation system. A widely
used method for this is known as Murphree Efficiency.
 yn  yn 1 
E MV  
,
 yn *  yn1 
(14)
where y* used in this equation is derived from experimentally measured liquidphase composition values. The Murphree Efficiency can be used to alter the equilibrium
line in the Mcabe-Thiele graph. This adapted graph can be used to more accurately
predict the separation under different conditions of feed composition and reflux ratios.
6
APPARATUS AND PROCEDURE
The primary piece of equipment used was the distillation column located in the
Merrill Engineering Building on the University of Utah campus. The top of the column,
control system and condenser were located in room 3290 on the third floor. The column
extended directly downward to the second floor, where the reboiler and bottoms storage
tank were located. The feed tank and pump, as well as the distillate storage tank were
located on the first floor, directly below the column.
The distillation column contained 12 bubble cap trays spaced at 11” intervals. The
trays were numbered in descending order, with the top tray designated tray 1. Each tray
had a diameter of approximately 14” and contained 18 bubble-caps. The bubble-caps
were 3” in diameter with square openings. The total column height was 145”. A full
schematic can be found in Appendix C.
Figure 2. Illustration of a bubble cap. Reprinted from Seader (1998)
7
The column design allowed for variation in the feed point, but for the purposes of
our experiments it remained stationary above tray 9. T-type thermocouples were
positioned inside each tray with digital readouts to the control computer. Each tray was
equipped with a sample port for the liquid phase. Sample ports for the vapor phase were
located on even-number trays only. Additional sample ports were available to sample
liquid from the feed, distillate and bottoms. Orifice plate meters equipped with digital
pressure gauges were used to measure flow rate of the feed, reflux, distillate and bottoms.
A computer equipped with Opto 22 control software was used to alter valve
positions to control feed, distillate and reflux flow rates, as well as the flow of steam to
the reboiler, and the flow of cooling water to the condenser. The software also received
and reported data from the thermocouples and flow meters.
Composition analysis was done with two pieces of equipment. A Mettler/Par
DMA46 densitometer (SN: 451038) was used to calculate sample densities. A ReichertJung ABBE Mark II digital refractometer (SN: 10952-9) was used to measure refractive
indices of samples. Assuming binary mixtures at every point, these measurements allow
the composition to be calculated. The correlations between densitometer and
refractometer data and composition were developed by creating and testing standards
with known volumetric compositions. These compositions were then converted to mole
fractions. For details, see Appendix A.
The column was first operated at total reflux. The pump was used to mix the
feed, then a sample of the feed was taken prior to distillation. Approximately 17.5
gallons of feedstock was introduced into the column, and then the valve was closed and
the pump shut down. The steam flow valve was opened until the steam flow rate was
0.456 kg/min, which required manual adjustment of the valves. The flow rate of the
8
cooling water was set at 50% of capacity using the computer control system, which
corresponded to a flow-meter reading of 136.8 kg/min. The temperature profile within
the column was monitored until the temperatures stabilized. When the temperatures no
longer changed significantly, the column was assumed to be at steady state.
A syringe with a plastic connection hose was used to extract a liquid sample from
each the top six trays. A glass syringe with a stainless steel needle was used to sample
the vapor from trays 2, 4 and 6. Samples of the distillate and bottoms were also taken.
All of these samples were saved and tested for refractive index. The samples with
sufficient volume to allow a densitometer reading were tested for density as well. Once
the samples were taken, the steam was shut down, and valves were opened to drain the
bottoms, the distillate, and any remaining fluid in the column into the feed tank. Once
the column had cooled below 70°C, the cooling fluid was shut off, and the valves were
closed.
Due to time constraints, only the top six trays were tested in this fashion on the
first distillation run. The column was subsequently started and brought to steady state
using the same procedure previously described, with a new sample of the feed taken
before running the column. Once the column was at steady state, liquid samples were
taken from trays 6 through 12, as well as the distillate and bottoms. Vapor samples were
taken from trays 8, 10 and 12. As before, all of these samples were tested for refractive
index, and as many as had sufficient volume were tested for density. The column was
shut down as previously described.
The feedstock analysis allowed the calculation of inlet composition. This
composition was used to graphically determine the minimum reflux ratio for a distillate
product of 60 mol% isopropanol. Given this data, it was calculated that 69% of the
9
distillate needed to be recycled for the column to run at twice the minimum reflux (see
Appendix D). The column was once again brought to steady state using the previously
described process. Once steady state was reached, the feed pump was turned on and the
valve was opened until the flow-meter reported a flow rate of 4.4 kg/min. The bottoms
valve on the column was immediately opened to release fluid at a similar flow rate, and
the reflux settings were changed to recycle 69% of the distillate.
Rotameters which reported the flow rates of the feed and the bottoms were
monitored and valves were adjusted to maintain a similar flow rate from each.
Additionally, the fluid level in the reboiler was observed, and any significant changes
were compensated for by adjustment of the flow rates of the feed and the bottoms.
As in previous experiments, the temperature profile of the column was watched.
When the profile no longer changed significantly, the column was assumed to be at
steady state. Liquid samples were taken from each tray, and vapor samples were taken
from the even numbered trays, using the methods previously mentioned. The samples
were tested for refractive index and density, and the compositions of the vapor and liquid
at each tested point was calculated.
Using known vapor-liquid equilibrium data for water-isopropanol systems and the
experimental data, the efficiency of each tray was estimated. These efficiencies, which
reflect the specific characteristics of the column, could then be used to predict the
behavior of other distillation systems run on the same equipment.
10
RESULTS AND DISCUSSION
The calculated mole fraction in each tray was as follows:
Table 1: Liquid and vapor compositions from the first run of the distillation column
under conditions of total reflux. Given in mole fractions of isopropanol in water.
Calculated from density and refractive index data.
Tray
Liquid Composition
Vapor Composition
1
0.55±0.015
2
0.42±0.015
0.45±0.015
3
0.33±0.015
4
Error
0.41±0.015
5
0.33±0.015
6
0.04±0.015
0.39±0.015
7
0.00±0.015
Feed
0.027±0.015
Table 2: Liquid and vapor compositions from the second run of the distillation column
under conditions of total reflux. Given in mole fractions of isopropanol in water.
Calculated from density and refractive index data.
Tray
Liquid Composition
Vapor Composition
6
0.496±0.015
0.612±0.015
7
0.383±0.015
8
0.374±0.015
0.524±0.015
9
0.091±0.015
10
0.027±0.015
0.505±0.015
11
0.016±0.015
12
0.020±0.015
0.103±0.015
Bottoms
0.014±0.015
Feed
0.027±0.015
11
Table 3: Liquid and vapor compositions from the distillation column with a reflux ratio
of 2.22. Given in mole fractions of isopropanol in water. Calculated from density and
refractive index data.
Tray
Liquid
Vapor
Composition
Composition
1
Error
2
0.603±0.015
0.225±0.015
3
0.573±0.015
4
0.549±0.015
0.232±0.015
5
0.504±0.015
6
0.450±0.015
0.160±0.015
7
0.028±0.015
8
0.026±0.015
0.121±0.015
9
0.025±0.015
10
0.022±0.015
0.092±0.015
11
0.020±0.015
12
0.018±0.015
0.070±0.015
Feed
0.023±0.015
Bottoms
0.017±0.015
The error points indicate samples that became tainted or otherwise inaccurate due
to mistakes in the sampling or handling process. Further experimentation is necessary to
determine the steady-state compositions at these points. Because the distillate (the liquid
from tray 1) was improperly sampled, and because the composition at tray 2 is near the
isopropanol-water azeotrope, it is assumed that the distillate for the partial reflux
operation is near the composition of the tray 2 liquid.
Because vapor samples could not be taken from every tray, the Murphree vapor
efficiency for each tray could not be calculated. Therefore, the liquid composition at
each point was used, with published vapor-liquid equilibrium data (Perry’s, 1999) to
create a McCabe-Thiele graph for the process. From this graph, efficiency for each tray
was be estimated.
12
Figure 3: Representation of liquid composition on trays 1 through 6 of the
distillation column at conditions of total reflux, plotted against graph of
vapor-liquid equilibria for the isopropanol-water system.
Figure 4: Representation of liquid composition on trays 7 through 10 of
the distillation column at conditions of total reflux, plotted against graph
of vapor-liquid equilibria for the isopropanol-water system.
13
Figure 5: Representation of liquid composition on trays 9 through 12 of the
distillation column at conditions of total reflux, plotted against a partial graph of
vapor-liquid equilibria for the isopropanol-water system.
Figure 6: Representation of liquid composition on trays 1 through 7 of the
distillation column with a reflux ratio of 2.22, plotted against a partial graph of
vapor-liquid equilibria for the isopropanol-water system.
14
Figure 7: Representation of liquid composition on trays 7 through 10 of the
distillation column with a reflux ratio of 2.22, plotted against a partial graph of
vapor-liquid equilibria for the isopropanol-water system.
The efficiency of each stage under each set of conditions can be approximated
graphically using an inverse lever-arm technique. A perpendicular line connecting the
operating line to the equilibrium line is drawn for each step. The length of the line below
the step divided by the length of the line overall is taken to be the stage efficiency. Since
these are estimates, it is difficult to quantify the error margins.
Table 4: Efficiency of each distillation column tray under conditions of total and partial
reflux. Values were estimated graphically.
Tray
1
2
3
4
5
6
7
8
9
10
11
12
Total Reflux
20.00%
30.40%
5.76%
78.60%
53.80%
8.69%
61.53%
45.70%
18.20%
5.10%
14.30%
Partial Reflux
50.00%
40.00%
38.90%
27.30%
100.00%
1.00%
1.00%
1.00%
1.00%
1.00%
1.00%
15
There is a clear trend toward a heightened efficiency in trays 5 and 6 as compared
to the other trays. Under the partial reflux regime, most of the separation takes place in
tray 6. Under partial reflux, it is also evident that very little separation occurs on trays 7
through 12. Further testing is necessary to determine whether these phenomena occur
consistently in this column.
There is a clear difference in plate efficiencies between conditions of partial and
total reflux. While the precise reasons for this are unknown, the liquid and vapor flow
rates in each plate were notably lower than for the partial reflux conditions. It appears
that the differences in flow effect the performance of certain trays. This is a reasonable
conclusion, particularly if the flow rates are significantly below those assumed in the
column design.
16
CONCLUSIONS AND RECOMMENDATIONS
The data gathered suggests that most of the separation processes are occurring on
a few trays, with the rest displaying little performance. An additional testing regime to
gather data over a wider variety of operating conditions may be of use.
Sample extraction proved to be a very challenging aspect of the experiment.
However, the process was much simpler under continuous feed conditions. Thus it is
likely that the difficulties experiences earlier where the result of too little fluid in the
column. Increasing the amount of feed used for total reflux from 17.5 gallons to 25
gallons might alleviate this problem considerably. It should be noted that, when
gathering samples from taps on the feed tank, tops line or bottoms line, the taps should be
purged before taking a sample. Failure to do so is a possible cause of errors in our data.
As was mentioned, additional testing is necessary to confirm the results gathering
in this experimental regime. Future tests could be facilitated by the installation of vapor
sampling ports on the odd-numbered trays. This would allow the Murphree vapor
efficiency of each tray to be calculated directly.
Preliminary data suggests that some of the trays, particularly trays 3 and 4 may be
operating at significantly lower than expected efficiency. If further experiments confirm
this conclusion, it may become necessary to inspect these trays for defects, damage or
signs of wear.
The end compositions show that a reasonable separation of the species in the feed
stock was effected. However, data concerning the steps within the process are not
consistent with what one would expect from the theory. Experimental data and defects
within the equipment are both potential causes of this inconsistency, and both should be
addressed in order to ensure accuracy in modeling this process.
17
REFERENCES
Seader, J.D., Henley; Ernest J., Separation Process Principles, John Wiley & Sons,
Danvers, MA, p 359-387. (1998)
Perry, Robert H.; Green, Don W.Perry’s Chemical Engineer’s Handbook, McGraw-Hill
Professional, New York, NY. (1997)
18
NOMENCLATURE
Symbol
Definition
Units
α
Relative volatility
B
Bottoms Flow Rate
l/min
D
Distillate Flow Rate
l/min
EMV
Murphree Vapor Efficiency
KH
Fractional Pressure of Heavy Key
KL
Fractional Pressure of Light Key
L
Total Liquid Flow Rate in Column
mol/min
Ln
Liquid Flow Rate in Stage n
mol/min
P
Total Pressure in Column
psi
PH
Partial Pressure of Heavy Key
psi
PL
Partial Pressure of Light Key
psi
R
Recycle Ratio
RI
Refractive Index
T
Period of Vibration as Read by a Densitometer
V
Total Vapor Flow Rate in Column
VB
Boilup Ratio
Vn
Vapor Flow Rate in Stage n
xB
Mole Fraction of Light Key in Bottoms Product
xD
Mole Fraction of Light Key in Distillate
xH
Mole Fraction of Heavy Key in Liquid Phase
xL
Mole Fraction of Light Key in Liquid Phase
yH
Mole Fraction of Heavy Key in Vapor Phase
yL
Mole Fraction of Light Key in Vapor Phase
y*n
Equilibrium Mole Fraction of Light Key in Vapor
mol/min
mol/min
at Stage n, Based on Measured Liquid Composition
19
APPENDIX A
Full Data
The standardization curves were developed from the following data.
Refractive Index
Composition (%)
Run1
Run2
Run3
1.3330
1.3408
1.3472
1.3543
1.3605
1.3641
1.3684
1.3721
1.3747
1.3767
1.3767
1.3327
1.3401
1.3477
1.3541
1.3608
1.3649
1.3691
1.3723
1.3747
1.3763
1.3764
1.3326
1.3409
1.3480
1.3548
1.3608
1.3648
1.3687
1.3721
1.3751
1.3763
1.3764
Density
Composition (vol%) T1
0 5.7023
10 5.6839
20 5.6712
30 5.6533
40 5.6287
50 5.6018
60 5.5723
70 5.5405
80 5.5054
90 5.4702
100 5.4270
T2
5.7024
5.6840
5.6712
5.6533
5.6286
5.6016
5.5722
5.5403
5.5047
5.4702
5.4271
T3
Ave
5.7023
5.7023
5.6859
5.6846
5.6716
5.6713
5.6535
5.6534
5.6258
5.6277
5.6020
5.6018
5.5711
5.5719
5.5359
5.5389
5.5058
5.5053
5.4698
5.4701
5.4270
5.4270
0
10
20
30
40
50
60
70
80
90
100
AVG
(vol%)
1.3328
1.3406
1.3476
1.3544
1.3607
1.3646
1.3687
1.3722
1.3748
1.3764
1.3765
STDEV Mole % (IPA)
0.0002
0.0004
0.0004
0.0004
0.0002
0.0004
0.0004
0.0001
0.0002
0.0002
0.0002
0.0000
0.0255
0.0557
0.0918
0.1359
0.1909
0.2614
0.3550
0.4855
0.6798
1.0000
St Dev Density (IPA) Mole % (IPA)
0.0001
0.9980
0.0000
0.0011
0.9837
0.0255
0.0002
0.9731
0.0557
0.0001
0.9587
0.0918
0.0016
0.9383
0.1359
0.0002
0.9178
0.1909
0.0007
0.8942
0.2614
0.0026
0.8683
0.3550
0.0006
0.8421
0.4855
0.0002
0.8148
0.6798
0.0001
0.7817
1.0000
From these data sets, empirical correlations were found to relate instrument
measurements to composition
Refractive Index:
xL =(RI110.47)*(10-15.617)
Densitometer:
xL = -101.92*T3+291.13*T2-279.46*T+90.244
20
Refractive Index vs Mol% IPA
1.3800
1.0500
1.3750
1.0000
Refractive Index
1.3700
1.3650
0.9500
1.3600
1.3550
0.9000
RI
Densitometer
1.3500
0.8500
1.3450
1.3400
0.8000
1.3350
1.3300
0.0000
0.2000
0.4000
0.6000
0.8000
0.7500
1.0000
Mol% IPA
Figure A1. Refractive indices and densities of standard mixtures of isopropanol and
water, plotted against mole fraction of isopropanol.
These correlations were then used to determine the composition of samples taken
from the distillation column.
21
Continuous Feed Conditions
Table A1. Densitometer periods and refractive indices for liquid and vapor samples
taken from the distillation column operating with a reflux ratio of 2.22.
Tray
1
2
3
4
5
6
7
8
9
10
11
12
Bottoms
Feed
RI of Liquid RI of Vapor Period of Liquid Period of Vapor
5.5643
5.5643
5.5639
5.4860
5.4798
5.4860
5.4798
5.4859
5.4796
5.4923
5.4923
5.4922
5.4974
5.4893
5.4972
5.4893
5.4971
5.4892
5.5072
5.5072
5.5072
5.5198
5.5559
5.5197
5.5558
5.5197
5.5558
1.3412
1.3409
1.3408
1.3404
1.3417
1.3405
1.3415
1.3403
1.3416
1.3403
1.3401
1.3400
1.3393
1.3407
1.3395
1.3406
1.3392
1.3404
1.3388
1.3390
1.3391
1.3384
1.3401
1.3384
1.3400
1.3383
1.3400
1.3374
1.3377
1.3375
1.3397
1.3396
1.3395
22
Total Reflux Conditions
Table A2. Refractive indices for liquid and vapor samples taken from the distillation
column operating operating at total reflux.
Tray
1
2
3
4
5
6
7
RI of Liquid
1.3752
1.3751
1.3756
1.3740
1.3734
1.3739
1.3716
1.3712
1.3715
1.3640
1.3640
1.3641
1.3723
1.3704
1.3712
1.3433
1.3445
1.3424
1.3332
1.3339
1.3339
RI of Vapor
1.3744
1.3748
1.3750
1.3735
1.3736
1.3736
1.3727
1.3728
1.3732
Table A3. Refractive indices for liquid and vapor samples taken from the distillation
column operating operating at total reflux.
Tray
6
7
8
9
10
11
12
RI of Liquid
1.3760
1.3760
1.3761
1.3729
1.3729
1.3727
1.3725
1.3724
1.3727
1.3553
1.3549
1.3549
1.3402
1.3401
1.3400
1.3340
1.3341
1.3343
1.3364
1.3372
1.3368
RI of Vapor
1.3759
1.3760
1.3758
1.3750
1.3751
1.3752
1.3751
1.3747
1.3748
1.3564
1.3564
1.3553
23
APPENDIX B
Error calculations
The standard deviation for the refractometer was 0.0004. Since the highest
refractive index used in this testing regime is 1.375, we can calculate the margin of error
as
ΔxL =(1.3750110.47)*(10-15.617)-( 1.3754110.47)*(10-15.617)=0.015.
This is the margin of error for all compositions determined by refractometry.
APPENDIX C
24
Equipment Schematics
Figure A1. Schematic of the distillation column used in this experiment.
APPENDIX D
25
Calculations
The calculations for minimum reflux are based upon the graphical “pinch point”
analysis, as developed in Seader (1998). The feed stock was calculated to be 2.7 mol%
isopropanol, and this was assumed to be a saturated liquid, therefore, a vertical line
connected this point to the equilibrium line. The distillation point was assumed to be 60
mol% isopropanol, this point on the equilibrium line was identified. The line drawn
between these two points indicates a rectification section with an infinite number of steps.
Thus:
Rmin
.
Rmin  1
(A1)
Slope
.
1  Slope
(A2)
Slope 
Rmin 
The slope was measured to be 0.525.
Rmin 
0.525
 110
. .
0.475
(A3)
Therefore, twice the minimum reflux ratio is 2*1.10=2.20.
2.20 
L
L
L

.
D 1 L
2.20
 0.69.
1  2.20
(A4)
(A5)
Thus, the reflux valves must be set to recycle 69% of the distillate in order to produce a
reflux ratio of twice the minimum.
26