University of California, Los Angeles Winter 2000
Distillation Column Design In Separating Ethanol-Water Mixture
Marie Dang Sandy Lao Hang-Tam Nguyen ChE 108A Project Professor Choi
Table of Contents
Introduction
3
Procedure
4
Calculation
5
Result
9
Discussion
11
Conclusion
13
Contribution
14
Index
15
2
Introduction A conventional azeotropic distillation uses entrainer such as benzene to purify products. Thus a distillation process without entrainer will cost more and one would need to adjust the process variables to minimize the cost. This project focuses on designing a system consisting of two distillation columns to obtain 99.9 wt% ethanol from a feed stream that composed of 40 wt% ethanol, 60 wt% water, at a total flow rate of 100kg/hr. The feed enters the first column at 25 C and 1 atm. For the basic case design, the first column will contain 60 stages, with a feed stage at 58 and the recycle-in stage is 10. The top pressure is 0.10 atm and the bottom pressure is 0.12 atm. We use total condenser with a distillate rate of 410 kg/hr with a reflux ratio of 25. The pump’s output pressure is 1.1 atm. As for the second distillation column, the number of stages is 90 and the feed stage is at tray 10. The top pressure is 1.0 atm and the bottom pressure is 1.1 atm. Again, we use the total condenser with a distillate rate of 370 kg/hr and a reflux ratio of 25. With the above parameters in mind, we utilize PRO/II with the NRTL thermodynamic model to design and simulate the base case design. By adjusting the following variables, we can come up with the best separation process design.
1. 2. 3. 4. 5. 6. 7. 8.
Number of trays of column 1 Number of trays of column 2 Position of the feed tray for column 1 Position of the feed tray for column 2 Reflux ratio of column 1 Reflux ratio of column 2 Position of the recycle-in stage for column 1 Flow rate of the recycle stream
We then came up with three different designs in which we minimize the material cost. The ultimate goal is to obtain a final design, which is economically the best, or at the
3
very least, has significant improvement from the basic design. The first design, which is denoted as Optimal 2, uses 15 and 39 trays for the two columns with a saving in cost of 13.6% in comparison to the base cost. The second design, Hang1, requires 37 and 40 trays for the two columns and saving us 83% less in comparison to the Base Run cost. Our last run, Sandy2, which uses 32 and 26 trays, saves us 28% less to the Base Run case.
Procedure For the Basic run, which involves two distillation columns, ProII is utilized to quickly calculate the features of each tray of the distillation column and generate a report for each run. First the basic run design is schematically drawn with ProII. The conditions are then entered into each column and initial estimates are provided. Note that the initial estimates for each stream coming in and out of the distillation reflect the overall component mass balance around the each unit. Once the basic run is generated, we can adjust the variables to come up with the better designs, which we could evaluate based on the economic analysis. By comparing the different cases we would be able to select which of the potential candidate designs would be the best. We note that this may not be the most optimal design but it is certainly presents improvement from the base design and that its set of costs are within reasonable tolerances. Thus by minimizing the number of trays of each column, the reflux ratio of the condensers, and the recycle flow rate, we can reduce the expense considerably. The choice of a design is based on the total annualized costs which would consists of both capital costs and operating costs, and the balanced minimum of the two would lead to the optimal design.
4
Calculation Sample Calculation for Determination of Column Size (Case Sandy2 Column 1) From the generated report for PROII simulation run, we first locate the vapor and liquid flow rates for tray #2 and the bottom tray. We picked tray #2 as our top tray because this is the actual first tray in the column. We also find the density (Rho) for both vapor and liquid at the top and bottom. Summarizing these results provided by PROII:
Top (Tray #2) Bottom
L' RhoL (kg/hr) V' (kg/hr) (kg/m^3) 10554 10660 783.806 60.09 4187 986.644
RhoV (kg/m^3) 0.17735 0.07937
Then we determine Flv, which is defined by the following relation:
F lv
ρ = • g V ' ρ L L '
0 .5
For the top tray: 0.17735 • F lv (top ) = 10660 783.806 10554
0.5
= 0.014893(unitless)
0.07937 F lv (bottom) = • 4187 986.644 60.09
0.5
= 0.000129(unitless)
Now we pick 24” tray spacing and turn to Figure 4.4 in out textbook Systematic Methods of Chemical Process Design and find Csb in ft/s.
C sb (top ) = 0.39
ft s
and
C sb (bottom) = 0.4
ft s
Updating our table: L' RhoL (kg/hr) V' (kg/hr) (kg/m^3) Top (Tray #2) 10554 10660 783.806
RhoV (kg/m^3) 0.17735
Flv (no unit) Csb (ft/s) 0.014893 0.39
5
Bottom
60.09
4187
986.644
0.07937
0.000129
0.4
We now calculate the flooding velocity Unf given by the expression: 0.5
ρ L − ρ g 20 0.2 U nf = Csb ρ g σ Where
σ is the surface tension in dynes/cm. For the first column, we use the surface
tension of ethanol at the top since it is mostly ethanol, and the surface tension of water at the bottom. For the second column, we use the surface tension of ethanol for both top and bottom. Note here that the ethanol surface tension used, is at 50C, as opposed to about 80C, which is the actual temperature of ethanol at the streams since this is the highest temperature that we can find. And for water, we used the surface tension at 30C while the actual stream is at 29C.
σ ethanol (80C ) ≈ σ ethanol (50C ) = 88.475
σ water (30C ) ≅ σ water (29C ) = 71.2
dynes
cm dynes
cm
According to CRC Handbook of Chemistry, 80th edition, 1999-2000. 0.5
0.2
783.806 − 0.17735 • 20 • 0.3048m = 5.869 m U nf (Top ) = 0.39 • 0.17735 s ft s 88.475 ft
0.5
0.2
986.644 − 0.07937 • 20 • 0.3048m = 10.544 m U nf ( Bottom) = 0.40 • 0.07937 s ft s 71.2 ft
Now we assume that we want to operate the column at 80% flooding, then the diameter of the column is given by the expression:
D =
0.8 • π
4V ' • U nf (ε )( ρ g )
6
Where
ε is the fraction of the area available for vapor flow and since we picked the
cheaper sieve tray, ε is 0.75.
4(10660) D(Top ) =
0.8(3.14)(5.869)
kg hr
hr 3600 s
m s
(0.75)(0.17735)
4(4187) D( Bottom) =
0.8(3.14)(10.544)
kg
= 2.457m
m3
kg hr
hr 3600 s
m s
(0.75)(0.07937)
kg
=1.717m
m3
D(max) = max( D(bottom), D(top )) = 2.457.m
And we take this to be the diameter of our column. Summarizing the results:
L' V' (kg/hr) (kg/hr) Top 10554 10660 Bottom 60.09 4187
RhoL (kg/m^3) 783.806 986.644
RhoV (kg/m^3) Flv (no unit) Csb (ft/s) Unf (m/s) D (m) 0.17735 0.014893 0.39 5.869 2.457 0.07937 0.000129 0.4 10.544 1.717 D(max)= 2.457 meters
We perform the same calculation for column 2 as well, except that for column 2, we use the surface tension of ethanol for both top and bottom trays. To determine the column height, we use a rough approximation of the tray spacing of 0.6 meter. So the total tray stack height would be: H ( tray stack ) = ( n − 1) * 0.6m
where n = number of trays, so for column 1: H (C 1) = (32 − 1) * 0.6m = 18.6m
7
Adding this to the extra feed space (1.5mX2 feed stages for column1), Disengagement space 1.5m), and skirt height (1.5m), gives the total height of the column. H ( column ) = H (tray stack ) + H ( feed space) + H ( Disengagement ) + H ( skirt )
Thus for column 1, H ( column1) = 18.6m + 3.0m + 1.5m + 1.5m = 24.6m
8
Results Data Result
As can see from the generated graph of the basic run, Figure 1 attached on the next page, we notice the excess number of trays in both columns one and two were used to obtain a 100% pure ethanol. Thus this is a significant source of waste posed by the extra trays. The dimensions are also quite large. The diameters are 2.47 and 1.45 meters for column one and two of Basic Run respectively. The lengths are 42.9 and 59.4 meters for column one and two. The number of trays is 60 and 90 for columns one and two. The feed enters at stage 58 for the first column, and at stage 10 for the entering recycle stream. For the second column, the feed enters at stage 10. The final product purity is 100% ethanol. However, we only need 99.9% ethanol, therefore we can reduce the amount of trays and reflux ratio. For optimal 2 case, the number of tray is reduced to 20 for the first column and 39 for the second column. The reflux ratio is also decreased to 25 and 18. With such a drastic cut in the tray number, the length column went down to 18.9 and 28.8 meters respectively. The diameter stays relatively the same, 2.47 and 1.24 meters. Here the feed enters at stage 18 and the recycle stream was introduced in stage 2. For the second column, the feed enters at stage 5. For the Hang1 case, the number of tray is 37 and 40 for the two columns. The reflux ratio is 4 and 2.5. As for the dimensions, the diameters are 1.08 and 0.52, however, the column heights reduced to 15.9 and 30.6 meters. Thus, the first column is only almost half of the second column in diameter. The feed enters at stage 25 and 2 for the entering recycle stream for the first column. For column two, the stream enters at second stage.
9
Since we have looked at the two extremes of number of tray and the lowest reflux ratio, now on the final optimal case, we try to even out the number of tray in both columns with the lowest reflux ratio possible to see if this would lower our overall cost. For the last case, Sandy2 case, the tray numbers are 32 and 26 for the two columns. The diameters are 2.46 and 0.95 meters. The heights are 26.10 and 21.0 meters respectively for the two columns. The reflux ratios are 25 and 10. For the entering streams of the first column, the feed enters at stage 22 and the recycle stream at 2. As of the second column, the stream enters at stage 2. Thus the following Table1 summarizes four different trials to provide a quick comparison between different runs. Runs
Columns Num. Of Diameter Height Reflux Condenser Heat Reboiler Heat Pump Work Trays (m) (m) Ratio Duty (M*KJ/HR) Duty (M*KJ/HR) (KW)
Basic Run Column1 60 Column2 90
2.47 1.45
42.9 59.4
25 25
-10.258 -8.6674
10.2103 8.7227
0.0142
Optimal 2 Column1 20 Column2 39
2.47 1.24
18.9 28.8
25 18
-10.2793 -6.3485
10.2357 6.4039
0.0142
Hang1
Column1 37 Column2 40
1.08 0.52
29.1 29.4
4 2.5
-1.9542 -1.1541
1.9109 1.2093
0.0142
Sandy2
Column1 32 Column2 26
2.46 0.95
26.1 21.0
25 10
-10.0178 -3.5678
9.9747 3.6229
0.0142
Cost Analysis Results
This section focuses on the economic factor in designing a separation process. According to the Basic Run, which would cost roughly 16 million dollars to purify 40% ethanol to 99.9% pure.
Comparing this cost value to the optimal runs, we see a significant
improvement. For Optimal2 case, the NPV(cost) is only 13.8million dollars. Yet for Hang1 case, the cost is now only 2.7 million dollars. Thus we have saved around 82.9%
of the Basic Run. Table 2 below summarizes the different types of cost for each run.
10
Runs
Total Capital Cost ($)
Total NPV (Cost) % Saved NPV Annualize ($) (Cost) Utility Cost ($)
Basic Run Optimal2 Hang1 Sandy2
1254360 615332 322485 621785
4918508 4353813 817055 3597620
15980036 13802432 2733632 11483816
0 13.62702812 82.89345531 28.1364823
We can see that the costs of all optimal runs are significantly less than that of the Basic Run. However, each of the trials has its’ own advantage and disadvantage as will be
discussed in detail in the next section. The following Table 3 presents a rough calculation of the profit we would have obtained if the designs were to implement. Note that this represents a very crude calculation of the profit just so we would have an idea if this is actually profitable investment. We see that all the trials seem to yield reasonable gain. Even the Basic Run, which costs much higher than the other three optimal runs, brings 35 fold profits for a 10 years period. This indicates that either the retail-selling price is too high ($30/L of ethanol) or that the process does bring considerable gain.
Either way, this evaluation confirms that the
Hang1 run is still the best in term of economic factor.
Profit Evaluation Basic Run Fixed Capital
Optimal2
Hang1
Sandy2
1254360
615332
322485
621785
Working Capital (0.20 f.c)
250872
123066.4
64497
124357
Fixed and Working Capital
1505232
738398.4
386982
746142
699031.77
699031.77
699031.77
699031.77
Raw Material ($0.08/lb prod)
55922.54
55922.54
55922.54
55922.54
Utilities ($0.012/lb prod)
8388.381
8388.381
8388.381
8388.381
0.015
10485.48
10485.48
10485.48
75261.6
36919.92
19349.1
37307.1
Product Rate (lb)
Labor ($0.015/lb prod) Maintenance (0.06yr f.c.)
11
Supplies (0.02yr f.c.)
25087.2
12306.64
6449.7
12435.7
Depreciation (straightline over ~10yrs)
125436
61533.2
32248.5
62178.5
Taxes, insurance (0.03/year)
37630.8
18459.96
9674.55
18653.55
Total Manufacturing Cost ($0.131/lb)
91573.16
91573.16
91573.16
91573.16
Gross Sales
11982762
11982762
11982762
11982762
Gross Profit (GS-TM)
11891189
11891189
11891189
11891189
1198276
1198276
1198276
1198276
10692913
10692913
10692913
10692913
Taxes (0.50 net profit)
5346456
5346456
5346456
5346456
Net Profit after Taxes
5346456
5346456
5346456
5346456
Return on Investment (ROI) (net income/ f&w cap)
35519%
72406%
138158%
71655%
0.1391369
0.0686598
0.0360817
0.0693757
SARE Expenses (0.10sales tax) Net Profit Before Taxes (GP-SARE)
Payout Time (total cap./net annual profit)
*Assume ethanol costs $30/Liter –From Sigma
Discussion Data Discussion
As can see from Figure 1, the purity of ethanol actually reaches 100% long before the tray number reaches 60 trays for the first column of the Basic Run. Thus this indicates that there are significant number of excess trays in the first column. The extra number of tray would cost us an additional cost to operate this design. In order to reduce the cost yet at the same time achieving the ultimate goal, of producing 99.9% ethanol, the stage number can be cut down to the minimum amount. However, if we push for the border line amount of tray number, the ethanol purity might not reach 99.9%, thus adding an extra 5% of tray number would serve our purpose adequately. The lowest number of tray would give the lowest design dimensions, thus would lower the construction cost of such a design. For the optimized runs, we not only push for the lowest number of tray number but also minimizing the reflux ratio as well as the dimensions of the design. The same purpose would serve for having the lowest reflux ratio, this would give a lower cost for
12
the feeding steam entering the column. Also the feed tray number plays an important role in maximizing the design. The recycle stream and the stream from the first column to second column need to feed in at the top tray. This makes sense because the ethanol concentration in these two stream are saturated with ethanol, thus having them fed in the top trays would separate the water out more efficiently. With all these considerations in mind, we eventually derive the three optimal runs. The first run, Optimal 2, aims for the lowest possible distillation tray-number for the second column, yet still produces 99.9% ethanol.
Since the number of tray for the second column is too high for the Basic Run ,
minimizing this would considerably lower the cost of building such a tall column. Having 20 trays in the first column and 39 trays in the second column results in a 99.9% ethanol release for Optimal2. However, as ProII iterates through the design, the system converges significantly slower than the Basic Run , this could be due to the high number of cycle the recycle stream has to reverse to the first column in order to obtain the desired purity. However, this trial was not considered to be a good design because the reflux ratio was quite high, causing a large heat duty amount in the reboiler and condenser, thus the cost of the feeding steam will be expensive. For the second run, Hang1, we minimize the reflux ratio with an intention that this would lower the utility cost of feeding steam into the columns. The reflux ratios are 4 and 2.5 for the two columns. This design leads to only –1.9542 and –1.1541 MJ in heat duty of the condenser. Comparing this heat duty with that of the Basic Run, which is –10.258 and –8.6674 MJ for the heat duty of the condenser. We see almost a 10 fold decrease in the heat duty. Thus the annualize utility cost of feeding steam is only $2,733,632, which is 83.39% less than the Basic Run for the NPV(cost ). So far this design seems very
13
attractive in term of operating cost. However, there are drawbacks in having such a low reflux ratio design, the column’s height is much higher in comparison to the previous runs. Now it requires more stage number to separate the mixtures to 99.9% ethanol. This however, is compromisable if our intention is to minimize the operation cost. As for our last design, Sandy2, we aim for the lowest number of trays in both columns, thus this would give us a relatively the same number of tray for both column. For this case the number of stage for column one is 32 while we only need 26 on the second column. This design might considered to be more advantageous over the previous two designs in term of space design, because the numbers of trays for both columns are close to each other. This offers a better design in the sense that construction would be much easier. The heights of both columns are not too tall or not too short in comparison to all the other runs. However, the reflux ratio is still high leading to a high utility cost. Once again, this demonstrates the need of priority when it comes to process design. If the intention is to save space and building columns that would fit in a designated area, this design would be more superior to the other two.
Conclusion As can see from the three trials, low tray number does not necessarily mean that it is the better design, there are several other factors involve that can significantly affect the capacity of a design. The reflux ratio seems to dominate over all the other factors in term of cost. Thus the lower the reflux ratio, the lower the cost would be. However, too low of a reflux ratio would require higher distillation stage number. Thus when designing a separation process, one would need to consider how the space and location of where the columns are to be built and from there to determine the priorities in designing the
14
process. For our purpose, we try to obtain the lowest cost operation yet with a relatively not too high number of trays, thus, Hang1 run seems to serve our purpose. This design saves us $13 million in comparison to the Basic Run case. Thus it is important to have as low reflux ratio as possible yet with reasonable column height in order to maximize profit of a design.
15
Contribution Marie Dang: Runs column, designs an optimal case and analyzes that particular case as
well as contributing in writing the report. Sandy Lao: Runs column, designs an optimal case, analyzes that particular case and
participate in writing the report. Hang-Tam Nguyen: Runs the column, designs an optimal case, analyzes that particular
case and writing the report.
Table of Index 1. First Report: Base Case Run Report 2. Second Report: Optimal 2 Run Report 3. Third Report: Optimal 3 Run Report 4. Fourth Report: Sandy2 Run Report 5. Dimensional Analysis Report For All Runs 6. Cost Analysis Report For All Runs
16
