Ind. Eng. Chem. Res.

2007, 46, 9223-9231 9223

Reducing Diameters of Distillation Columns with Largest Calculated Diameter at

the Bottom
Phillip C. Wankat*
School of Chemical Engineering, Purdue UniVersity, 480 Stadium Mall DriVe,
West Lafayette, Indiana 47907-2100

Methods are developed to reduce column diameter for distillation when the calculated diameter is largest at
the bottom of the column. This diameter can often be reduced by generating vapor above the reboiler with
an intermediate reboiler or with two-enthalpy feed. For a saturated liquid feed that is 50% n-butane and 50%
n-pentane the two-enthalpy feed system reduced column volume by 8.2% with no change in energy use. For
distillation of dilute acetic acid (5%)-water (95%) mixtures the best system (an intermediate reboiler) resulted
in decreases in column volume as large as 53.9% with no increase in energy use. For a 50% acetic acid
mixture the two-enthalpy feed system resulted in up to 32.5% reduction in column volume, while vaporizing
the entire feed resulted in 31.5% volume reduction. Vapor bypass was effective in concentrated absorbers.

In an earlier paper1 we showed how to reduce the diameter use sieve trays, and diameters were calculated with the procedure
of distillation columns with vapor feeds when the largest developed by Fair6 and incorporated in the commercial Aspen-
calculated diameter is in the enriching section. The driving Plus 2004 process simulator. The optimum feed trays that gave
philosophy behind this research was to consider a large the maximum product purities are reported.
difference in the calculated diameters as potential to either Liquid Feed of 50-50 n-Butane and n-Pentane Base Case.
reduce the column volume or process more material in a A distillation column is separating 1000 kmol/h of a saturated
retrofitted column. Essentially, by condensing some of the vapor liquid feed that is 50 mol % n-butane and 50 mol % n-pentane.
below the column condenser, the vapor flow rate at the location The column has 28 stages plus a partial reboiler and a total
with the largest calculated diameter was reduced. The methods condenser (N ) 30 in AspenPlus notation). Calculations were
explored were to use an intermediate condenser,1,2 condition done with the AspenPlus simulator using the Peng-Robinson
the feed by cooling the entire feed stream,1,3,4 and use a two- vapor-liquid equilibrium (VLE) correlation. For an external
enthalpy feed that cools only part of the feed stream.1,5 reflux ratio ) 0.879, the distillate was 99.90 mol % n-butane
In this paper we explore reducing the diameter of distillation and the bottoms was 99.90 mol % n-pentane. Table 1 shows
columns when the largest calculated diameter is at the bottom the results and conditions for this base case, and the calculated
of the column. Most of the cases involve liquid feeds. The diameters are listed in Table 2. Although the vapor flow rate is
strategy employed is to reduce the vapor flow rate at the bottom highest at the top, the calculated diameter is highest at the
of the column by generating some of the vapor required for the bottom.
separation at a higher location in the column. Four approaches Acetic Acid (5 mol %), Water (95 mol %) Liquid Feed
for doing this will be examined: (1) Use an intermediate Base Case. The acetic acid-water system is economically quite
reboiler,2 which consists of withdrawing a liquid stream from important and represents a rather difficult separation with high-
the column, vaporizing it, and then returning it to the column energy loads. In fact, the separation is difficult enough that an
as a vapor. (2) Condition the feed by heating or vaporizing the entrainer is often used.7 Table 1 lists the details of this base
entire feed stream.3,4 (3) Heating or vaporizing a part of the case, and Table 2 shows the diameters calculated for this
feed while leaving the other part vapor with a two-enthalpy feed distillation column. The column has 40 stages plus a kettle
system.5 In two-enthalpy feed operation the original liquid feed reboiler, a total condenser (N ) 42), and L/D ) 2.3. Although
is split into a liquid part FL and a part FV that is heated or the NRTL-HOC VLE correlation is recommended,7 we found
vaporized in a heat exchanger. (4) Use a vapor bypass to input a better fit to the VLE data in Perry’s Handbook8 with the
vapor from the reboiler or from vaporized bottoms product at Wilson HOC VLE correlation regressed to this data. The
a location above the bottom of the column. With these systems percentage error in measured and predicted values of (y - x)
there will either be an energy penalty resulting in an increase of the more volatile component (MVC) at equilibrium was
in overall heating and cooling loads or more stages will be calculated.
needed to produce the desired purity. However, since the column
diameter will be decreased, there will often be capital cost % error in (yMVC - xMVC) )
savings that can be significant. These methods for balancing 100{[(y - x)predicted - (y - x)data]/(y - x)data} (1)
the calculated diameters are applicable to standard and high
capacity trays, and to standard and high capacity packings. This equation measures the error in the distance between the
equilibrium curve and the total reflux operating line. The percent
Base Cases for Standard Distillation Systems error in (yW - xW) ) 0.3% at xW ) 0.3063, 2.2% at xW )
To illustrate the differences in calculated diameters that are 0.6750, and 4.6% at xW ) 0.9578. Thus, the fit is good at low
possible, we will first consider three base cases. The columns and moderate mole fractions of water, but slightly overstates
the difference between vapor and liquid mole fractions at high
* Tel.: (765) 494-0814. Fax: (765) 494-0805. E-mail: wankat@ mole fractions of water. The results for the diameter calculations
ecn.purdue.edu. for this separation should be accurate, but the number of
10.1021/ie0709887 CCC: $37.00 © 2007 American Chemical Society
Published on Web 11/23/2007
9224 Ind. Eng. Chem. Res., Vol. 46, No. 26, 2007

Table 1. Simulation Conditions and Results for Base Casesa

NFeed N dia A vol tray QR Qc L/D D
Liquid Feed of 50-50 n-Butane and n-Pentane Base Case
19 30 3.10 7.54 133.3 29 6 150 -5 860 0.879 500
Acetic Acid (5 mol %), Water (95 mol %) Liquid Feed Base Case
25 42 6.14 29.56 554.1 41 35 560 -35 550 2.3 950
Methanol (5 mol %), Water (95 mol %) Liquid Feed Base Case
9 20 1.20 1.13 9.8 2 2 410 -2 270 3.6 50
a F ) 1000 kmol/h, all feeds are saturated liquids, D is distillate flow rate in kmol/h, and N ) number of trays + condenser + reboiler. Tray spacing is
24 in. ) 0.6096 m (butane-pentane) or tray spacing ) 18 in. ) 0.4572 m (acetic acid-water and methanol-water). Operation is at 80% of flooding. dia
) maximum calculated diameter in m, tray ) tray at which maximum diameter occurred, A ) maximum calculated column area in m2, vol ) column
volume in m3 ) A(N - 2 + 1)(tray spacing), where N - 2 is the number of trays and the +1 is for disengagement space for liquid and vapor, QR and Qc
are the reboiler and condenser duties in kW, p ) column pressure in atm ) 1.0, pressure drop in psi/tray ) 0, Nfeed is the optimum feed stage, and condenser
is labeled as stage 1 in all cases.

Table 2. Diameters Calculated for Standard Distillation Base Casesa

n-butane, acetic acid- methanol-
n-pentane, liquid water, liquid water, liquid
tray V dia A tray dia A tray dia A
2 940 2.64 5.47 2 3.64 10.42 2 1.20 1.13
18 887 2.60 5.29 25 3.86 11.68 8 0.98 0.76
19 885 2.94 6.77 41 6.14 29.56 9 1.18 1.09
29 855 3.10 7.54 19 1.15 1.04
a Conditions and results are listed in Table 1. Vapor flow rate V is in kmol/h.

equilibrium stages required in the enriching section may be vaporize stream FV. What is of critical importance is to separate
slightly underpredicted. The distillate is 99.79 mol % water, the feed into two parts before one part is heated or vaporized.
and the bottoms is 4.00 mol % water. Table 2 shows there is The method is also useful if the original feed will flash and
little variation in the calculated diameter in the enriching section, become a two-phase feed in the column. This approach is similar
and the diameter is largest at the bottom of the column. A single to the use of two-enthalpy feed to condense part of a vapor
column with a diameter of ∼6.1 m would be built. feed.1 Conditioning a liquid feed by heating or vaporizing the
Methanol (5 mol %), Water (95 mol %) Liquid Feed Base entire feed stream3,4 is well-known and is similar to condensing
Case. Table 1 also shows the results and conditions for this a vapor feed.1
base case, and the calculated diameters are listed in Table 2. We also discovered that vapor bypass can be useful for
The column has 18 stages plus a kettle reboiler and a total diameter control in certain distillation applications and for
condenser (N ) 20). The distillate was 95.45 mol % methanol, concentrated absorbers may be the method of choice. One
and the bottoms was 99.76 mol % water. Calculations were done version of vapor bypass for distillation is shown in Figure 3a,
with the AspenPlus simulator using the NRTL VLE correlation. where part of the vapor from the reboiler bypasses the bottom
The largest calculated error in (yM - xM) was 3.6%; thus, the stages in the column and is input higher up the column. This
correlation fits the VLE data in Perry’s Handbook8 quite well. bypass vapor is in equilibrium with the bottoms product. An
In this case the variation in the calculated column diameter is alternative if the reboiler does not have sufficient capacity is
modest, and one would probably not try to improve performance shown in Figure 3b, where part of the bottoms stream is
with the methods employed in this paper. vaporized and input into the column. This bypass vapor is the
The first example in Table 2 shows a modest variation in same mole fraction as the bottoms product. In general, the
column diameter, while the second example represents a fairly bypass can withdraw vapor from anywhere in the column and
extreme, but not unusual case of diameter variation. Concen- feed it at a different location. In absorption columns part of the
trated absorbers also show a larger diameter at the bottom of feed gas bypasses the bottom one or more stages in the column
the column. There are also many systems similar to the (Figure 4).
separation of a liquid feed of methanol and water that show
very modest variations in calculated column diameter. If there Application to Distillation of Liquid Feeds of Light
is no change in calculated diameter, then diameter control Hydrocarbons
methods are not needed.
In the distillation of homologous series the calculated diameter
is usually largest at the bottom of the column when the feed is
Methods To Adjust Column Diameters
a liquid. The differences in calculated diameter depend upon
The use of an intermediate reboiler2 to control the diameter the physical properties of the compounds. Generally speaking,
of a column is well-known and is obviously related to our as the ratio of molecular weights becomes larger, the difference
previous study with intermediate condensers for vapor feeds.1 in calculated diameters will also become larger. We will study
This method is shown in Figure 1. Use of two-enthalpy feed to applications of vapor bypass, two-enthalpy feed, and intermedi-
control column diameter for liquid feeds has not been reported ate reboilers for the separation of n-butane from n-pentane and
in the literature. The two-enthalpy feed system is shown in for the ternary separation of propane, n-butane, and n-pentane.
Figure 2 when the original feed is a liquid.5 In normal operation Vapor Bypass for Separation of 50-50 n-Butane from
the entire feed is often heated (set FL ) 0 in Figure 2) and will n-Pentane. The vapor bypass systems shown in Figure 3 were
be partially or totally vaporized. In two-enthalpy feed operation studied, and the results are given in Table 3. In general, as the
the feed is split into a liquid part FL and a part FV that is bypass rate increased at constant QR,total the calculated diameter
vaporized in the heat exchanger. It is not necessary to totally at the bottom of the column (tray 29) decreased, the diameter
 Ind. Eng. Chem. Res., Vol. 46, No. 26, 2007 9225

Table 3. Results for Vapor Bypass, Two-Enthalpy Feed, and Intermediate Reboiler for a Liquid Feed That Is 50% n-Butane and 50%
n-Pentanea
FV NF,L NF,V N dia A vol tray QR,total Qc
0 (base) 19 ) 30 3.10 7.54 133.3 29 6150 -5860
Vapor Bypass (N ) 30)
FBy NF,V NBy,V Figure dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
78.9 18 23 3b 2.99 7.03 124.3 23/29 6190 -5900 6.6 0.6
78.9 18 23 3a 2.99 7.03 124.3 23/29 6190 -5900 6.6 0.6
95.0 18 22 3a 2.97 6.94 122.7 22/29 6210 -5920 7.9 0.9
Two-Enthalpy Feeds
FV NF,L NF,V N dia A vol tray QR,total Qvaporize Qc decr vol (%) incr QR,total (%)
107 19 22 30 2.93 6.74 119.2 22/29 6200 757.8 -5920 10.5 0.9
115 19 23 31 2.92 6.69 122.3 23/19 6150 814.5 -5860 8.2 0
Two-Phase Feed (Qheat_feed ) 814.5 kW)
NF N dia A vol tray QR,total Qheat_feed Qc incr vol (%) incr QR,total (%) xD,B
19 31 3.26 8.34 152.5 19 6150 814.5 -5860 +14.4 0 0.9936
Intermediate Reboilers (N ) 30):
Fwithdr NF,liq NL,with NV,ret dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
89.8 19 22 23 2.95 6.82 120.6 22/23/29 6170 -5880 9.4 0.3
107 19 21 22 2.94 6.80 120.2 21/19 6190 -5890 9.8 0.5
a Unless specified otherwise, mole fractions of n-butane in the distillate and n-pentane in the bottoms are both 0.9990. Pressure is 1.0 atm, and tray

spacing is 24 in. ) 0.6096 m. Total condenser is labeled stage 1. FV is the kmol/h in the vapor portion of the feed for two-enthalpy feed, FBy is the kmol/h
vapor in the vapor bypass, and Fwithdr is the kmol/h that is withdrawn and then vaporized in the intermediate reboiler. NF,L is the optimum feed stage for the
liquid feed, NF,V is the optimum feed stage for the vapor portion of the feed in two-enthalpy feed systems, NBy,V is the feed location for the vapor bypass,
NL,with is the optimum withdrawal plate for the liquid that is sent to the intermediate reboiler, NV,ret is the return location for the vapor for an intermediate
reboiler. QR,total is the sum of the energy required in the reboiler QR plus the energy Qvaporize required to vaporize the vapor bypass, the portion of the feed,
or in the intermediate reboiler, or QR,total ) QR + Qheat_feed, where Qheat_feed is the energy used to heat the entire feed to produce a two-phase feed.

Figure 1. Intermediate reboiler system. Figure 2. Two-enthalpy feed for a liquid feed.

at the return stage increased, and the purity decreased. As the were used for Figure 3a. Since the results are essentially
vapor was returned to a higher stage (lower number) in the identical, vapor bypass results for other systems will be done
column, the diameter at the return stage decreased, but the purity only for Figure 3b, which did not have the convergence
also decreased. Purity could be increased by increasing QR,total, problems that occurred for Figure 3a with high withdrawal rates.
which also increased both diameters. The rate of the vapor The improvements (6.6-7.9% decrease in column volume)
bypass, the stage to return the bypass, and the value of QR,total shown in Table 3 for vapor bypass are modest.
were varied until the calculated column diameters were balanced Two-Enthalpy Feed or Intermediate Reboiler for Separa-
at stage 29 and the return stage. For example, in the Figure 3b tion of 50% n-Butane, 50% n-Pentane Liquid Feed. We can
result in Table 3, the calculated diameters were the same for also use the two-enthalpy feed system for this separation by
return tray 23 and for tray 29. The notation 23/29 for “tray” vaporizing a portion of the feed. The results in Table 3 show
indicates the largest calculated diameter occurred simultaneously larger but still modest decreases in column volume. The result
at both trays 23 and 29. Note that there are also conditions where for FV ) 107 was obtained with the identical column as the
the diameters will match at return tray 24 and tray 29, at return base case by increasing QR until the purities matched the base
tray 22 and tray 29, and so forth. By returning the bypass vapor case. The result for FV ) 115 used the same QR,total as the base
higher up the column, increasing the bypass rate, and increasing case, but added an additional stage to match the base case purity.
QR,total, a slightly lower column diameter resulted. The calcula- Because of the added height of the additional stage, the column
tion was first done for Figure 3b, and then the same conditions volume decreases less than for the previous system. Since the
9226 Ind. Eng. Chem. Res., Vol. 46, No. 26, 2007

have been for binary systems, these methods also apply to

multicomponent separations. For the ternary base case a
distillation column is separating 1000 kmol/h of a saturated
liquid feed that is 30 mol % propane, 25 mol % n-butane, and
45 mol % n-pentane. The column has 19 stages plus a kettle
reboiler and a total condenser (N ) 21). Calculations were done
with the AspenPlus simulator using the Peng-Robinson VLE
correlation. Propane was selected as the light key and the
distillate flow rate was 300 kmol/h. For an external reflux ratio
) 0.45, the distillate was 95.82 mol % propane and 4.18 mol
% n-butane, and the bottoms was 1.79 mol % propane, 33.92
mol % n-butane, and the remainder n-pentane. The calculated
diameters and vapor flow rates are stage 2 ) 1.55 m, V ) 435.0
kmol/h; stage 10 ) 1.51 m, V ) 407.3 kmol/h; stage 11 )
1.92 m, V ) 398.4 kmol/h; and stage 20 ) 2.12 m, V ) 400.2
kmol/h. Although the vapor flow rate is highest at the top, the
calculated diameter is highest at the bottom. Other information
Figure 3. Vapor bypass systems for distillation. (a) Bypass part of vapor
from reboiler. (b) Vaporize and bypass part of bottoms stream. for the base case is in Table 5.
The results for vapor bypass, two-enthalpy feed, and inter-
mediate reboiler systems are shown in Table 5. The optimum
values of the flow rate of material vaporized FV and the optimum
input and (for intermediate reboiler) withdrawal stages were
found by matching the column diameters at stage 20 and one
of the input stages and then increasing the absolute value of Qc
until the propane mole fraction in the distillate matched the base
case (95.82 mol %). For this ternary hydrocarbon system the
two-enthalpy feed system had the largest decrease in column
area. Comparing the results in Table 5 to those in Tables 3 and
4 for the separation of n-butane from n-pentane, we see that
the reductions in column area are larger for this ternary system
for all three techniques. This occurred because the ratio of the
smallest calculated diameter divided by the largest calculated
Figure 4. Vapor bypass for absorption.
diameter is smaller for the ternary base case (0.71) than for the
50-50 binary separation (0.84). Thus, there is more room for
column diameters were identical at two trays, increasing FV improvement with this ternary system. This ratio is smaller for
further had no effect on the column diameter and eventually the ternary base case to a large extent because the molecular
will cause the purity to decrease below that of the base case. weights of the distillate and bottoms differ by a larger amount.
The results obtained for the intermediate reboiler (Table 3) are We also simulated a ternary distillation with the same system
very similar to those obtained with the two-enthalpy feed system. but with n-butane as the light key (D ) 550 kmol/h) (not shown
Table 3 also shows the results for a two-phase feed using in Table 5). The column had a total condenser, a kettle reboiler,
the same amount of energy to heat the entire feed as the two- 19 equilibrium trays (N ) 21), and liquid feed on stage 12.
enthalpy feed system used to vaporize FV (FV ) 115). Both liquid and vapor flow rates decreased from stage 15 to
Unfortunately, this system increases the maximum calculated stage 20, but the largest diameter was on the bottom tray (no.
diameter of the column and does not meet the product 20). Although there is less variation in calculated diameter in
specification. The distillate purity drops because a two-phase this case, the qualitative conclusions are essentially the same
feed increases the minimum reflux ratio;1 thus, operation at the as for the base case in Table 5. Variations in physical properties
same L/D as the two-enthalpy feed system is at a lower can cause significant increases in the calculated diameter even
multiplier of the minimum reflux ratio. when vapor or liquid flow rates decrease.
Separation of 25% n-Butane, 75% n-Pentane and 75%
n-Butane, 25% n-Pentane Liquid Feeds. To explore the effect Application of Processes for Distillation of Acetic Acid
of feed concentration, the results for the separation of 25-75 and Water
and 75-25 mixtures of n-butane and n-pentane are given in
Table 4. For all feed concentrations the two-enthalpy feed has The results in Table 1 show that, with a liquid feed that is
the largest decrease in column volume, but at the price of a dilute in acetic acid, the calculated diameter is largest at the
larger increase in QR. The intermediate reboiler system has bottom of the distillation column. We will see later in this
slightly less decrease in column volume, but the smallest section that this statement is also true for more concentrated
increase in QR. The vapor bypass system appears to work better (50-50) feeds. For a liquid feed of water-acetic acid the
at low feed mole fractions of the more volatile component than calculated diameters will be balanced by vaporization of part
it does at 50 or 75% n-butane in the feed. The intermediate of the feed, with vapor bypass, or with an intermediate reboiler.
reboiler and two-enthalpy feed systems appear to have a higher Distillation of a 95 mol % Water, 5 mol % Acetic Acid
volume reduction at 50 mol % n-butane and lower volume Liquid Feed. The base case was presented in Table 1 and is
reduction at 75 mol % n-butane in the feed. The increases in repeated in Table 6, and the calculated diameters for the base
energy use are very modest. case were presented in Table 2. The column had 40 equilibrium
Ternary Distillation of Propane, n-Butane, and n-Pentane. trays, a total condenser, and a kettle reboiler (N ) 42). The
Although all the previous examples in this paper and in ref 1 two-enthalpy feed system was employed by vaporizing a portion
 Ind. Eng. Chem. Res., Vol. 46, No. 26, 2007 9227

Table 4. Results for Base Case, Vapor Bypass, Two-Enthalpy Feed, and Intermediate Reboiler for Liquid Feeds That Are 25 mol % n-Butane,
75% n-Pentane and 75% n-Butane, 25% n-Pentanea
25% n-Butane and 75% n-Pentane
FV NF,L NF,V dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
0 (base) 13 2.70 5.72 101.1 29 4330 -4045
Two-Enthalpy Feed
87 13 19 2.57 5.17 91.4 19 4360 -4070 9.6 0.7
Vapor Bypass
FBy NF,V NBy,V Figure dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
58 13 21 3b 2.60 5.32 94.0 21/29 4340 -4050 7.0 0.2
Intermediate Reboiler
Fwithdr NF,L NL,with NV,ret dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
55 13 23 22 2.59 5.28 93.3 22/29 4330 -4050 7.8 0.07
75% n-Butane and 25% n-Pentane
FV NF,L NF,V dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
0 (base) 17 3.27 8.40 148.5 29 7410 -7260
Two-Enthalpy Feed
160 21 26 3.15 7.81 138.1 21 7410 -7260 7.1 0.07
Vapor Bypass
FBy NF,V NBy,V Figure dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
60 21 24 3b 3.20 8.03 142.0 29 7420 -7270 4.5 0.15
Intermediate Reboiler
Fwithdr NF,L NL,with NV,ret dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
49 21 27 26 3.20 8.02 141.8 21/29 7410 -7260 4.6 0.02
a The distillation system had 28 stages, a total condenser, and a kettle-type reboiler (N ) 30). F ) 1000 kmol/h, p ) 1.0 atm, and tray spacing is 24 in.
) 0.6096 m. For the 25% n-butane feed with L/D ) 1.582, the base purity was xD,C4 ) 0.9802 and xB,C5 ) 0.9934. For the 75% n-butane feed with L/D
) 0.550, the base purity was xD,C4 ) 0.9941 and xB,C5 ) 0.9822. Notation is the same as in Table 3.

Table 5. Results for Ternary Distillationsa

FV NF,L NF,V dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
0 (base) 11 2.12 3.52 42.9 20 2955 -2307
Two-Enthalpy Feed
80 10 16 1.93 2.93 35.7 16/20 2957 -2308 16.9 0.06
Vapor Bypass
FBy NF,V NBy,V dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
75 9 16 Figure 3b 1.97 3.04 37.1 16/20 2957 -2308 13.8 0.05
Intermediate Reboiler
Fwithdr NF,L NL,with NV,ret dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
71 9 17 16 1.95 2.99 36.5 16/20 2957 -2308 15.0 0.05
a Pressure is 1.0 atm, N ) 21, and tray spacing is 24 in. ) 0.6096 m. Distillate mole fraction xD,C3 ) 0.9582. Notation is the same as in Table 3.

of the feed. With FV ) 600 kmol/h and the same N and same energy loads as the base case, the maximum volume reduction
energy inputs (second entry in Table 6) the column volume occurs close to FV ) 800 kmol/h.
decreased significantly, but the purity also dropped. The purity Table 6 also shows the results for a two-phase feed using
could be made to match the base case purity by increasing QR,total the same amount of energy to heat the entire feed (9006.4 kW)
and Qc with the same N (third entry), or by increasing N with as the two-enthalpy feed system with FV ) 800. To meet the
the same QR,total and Qc (fourth entry). In all cases the largest product specification, the number of stages had to be increased.
diameter is at the bottom of the column and diameters are not The two-phase feed system has less volume reduction than either
balanced. For the fourth entry the diameters were 5.53 m at the two-enthalpy feed with FV ) 800 or the fully vaporized
stage 45, 3.84 m at stage 32 (vapor feed stage), and 3.64 m at feed.
stage 2. Increasing the amount vaporized to FV ) 800 further Vapor bypass results using the configuration in Figure 3b
decreases the column diameter, but more stages need to be added are also shown in Table 6. First, at a bypass rate of 500 kmol/h
to obtain the same purity as the base case (entry 5). The energy there is an 8% decrease in diameter, but the product purities
used to vaporize the feed with FV ) 800 is 9006.4 kW. Even are less than those of the base case. By adding one stage and
with all of the feed vaporized (FV ) 1000 kmol/h, which is adjusting the tray location for return of the bypass vapor, the
identical to the normal system of conditioning all of the feed) base purities are met. In general, with a lower location in the
the diameter drops significantly, but so does the purity (not column for the return of the bypass (larger NBy,V), the purities
shown). The base case purity could again be obtained either by are slightly better but the diameter is slightly larger. The bypass
increasing QR or increasing the number of stages (entries 6 and runs were repeated with FBy values of 750 and 1000. Note in
7, respectively), but the required increase in number of stages Table 6 that with FBy ) 1000 the largest diameter was calculated
becomes quite large. If one wants to operate with the same at the bypass return, not the bottom of the column. The optimum
9228 Ind. Eng. Chem. Res., Vol. 46, No. 26, 2007

Table 6. Results for Distillation of 95% Water, 5% Acetic Acid, Liquid Feeda
Base
FV NF,L NF,V N dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
0 25 42 6.14 29.6 554.1 41 35 560 -35 550
Two-Enthalpy Feed
600 22 28 42 5.48 23.6 442.7 41 35 560 -35 550 20.1
Distillate (xD,W ) 0.9974) and bottoms (xB,W ) 0.0490) do not match base purities.
600 23 28 42 5.60 24.7 463.0 41 36 430 -36 400 16.4 2.4
600 25 32 46 5.53 24.0 493.8 45 35 560 -35 550 10.9
800 26 34 49 5.31 22.2 487.2 48 35 560 -35 550 12.1
1000 27 42 5.30 22.0 412.4 41 37 730 -37 720 25.6 6.1
1000 38 55 5.08 20.24 498.7 54 35 560 -35 550 10.0
Two-Phase Feed (Qheat_feed ) 9006.4 kW)
NF N dia A vol tray QR,total Qheat_feed Qc decr vol (%) incr QR,total (%)
34 51 5.30 22.1 505.2 50 35 560 9006.4 -35 550 8.8
Vapor Bypass (Figure 3b)
FBy NF,L NBy,V N dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
500 24 38 43 5.89 27.3 524.2 42 35 560 -35 550 5.4
750 25 38 43 5.76 26.1 501.2 42 35 560 -35 550 9.6
1000 25 40 43 5.94 27.7 531.9 40 35 560 -35 550 4.0
Two-enthalpy feed (FV ) 600) plus vapor bypass (FBy ) 750):
25 31 42 47 5.10 20.5 431.1 46 35 560 -35 500 22.2
Intermediate Reboiler
Fwithdr NF,L NL,with NV,ret N dia A vol tray QR,total Qc decr vol (%) incr QR,total (%)
200 25 33 34 42 5.94 27.8 521.1 41 35 560 -35 550 6.0
1800 25 33 34 42 4.17 13.6 254.9 33/41 35 560 -35 550 54.0
Distillate (xD,W ) 0.9978) and bottoms (xB,W ) 0.0402) do not match base purities.
1800 25 33 34 43 4.11 13.3 255.4 42/33 35 560 -35 550 53.9
a F ) 1000 kmol/h, D ) 50 kmol/h, tray spacing is 18 in. ) 0.4572 meters, operation is at 80% of flooding, N ) 30, and p ) 1.0 atm. Same notation
as in Table 3. Base results are in Tables 1 and 2. Base purities are xD,W ) 0.9979 and xB,W ) 0.0400. Unless specified otherwise, distillate and bottoms match
or slightly exceed base purities.

Table 7. Results for Distillation of 95% Water, 5% Acetic Acid, Saturated Vapor Feeda
Fwithdr NF NL,with NV,ret N dia A vol tray QR,total Qc decr vol (%)
Vapor Feed Base Case
0 27 42 5.30 22.0 412.4 41 26 470 -37 720
Intermediate Reboiler
50 27 33 34 42 5.24 21.6 404.9 41 26 470 -37 720 1.8
1100 27 33 34 43 3.92 12.1 232.3 42/27 26 470 -37 720 43.7
a F ) 1000 kmol/h, D ) 50 kmol/h, tray spacing is 18 in. ) 0.4572 m, operation is at 80% of flooding, p ) 1.0 atm, x
D,W ) 0.9979, and xB,W ) 0.0400.
Same notation as in Table 3. All simulations have the same value of QR,total. The decrease in column volume is compared to the vapor feed base case. All
results match base case distillate and bottoms purities.

operation of this system is close to a bypass rate of 750 kmol/ to vaporize the liquid withdrawal, 19 690 kW, is greater than
h. We also tried combining the two-enthalpy feed system with the amount required in the reboiler, 15 870 kW, but QR,total is
vapor bypass (Table 6). The addition of two-enthalpy feed plus unchanged.
vapor bypass is an improvement compared to each individually. Distillation of a 95 mol % Water, 5 mol % Acetic Acid
The best results for this separation were obtained with an Vapor Feed. If the feed is input as a vapor, we would normally
intermediate reboiler (Table 6). With low liquid withdrawal rates expect the diameter to be largest at the top of the column.1
(Fwithdr ) 200 in Table 6) the intermediate reboiler meets the However, with this system the parameter effect on the diameter
purity specifications with no further changes. However, the is very large and the diameter is still largest at the bottom of
decrease in volume is modest and the column diameters are the column. The new base case with a vapor feed can be
not balanced since the diameter at stage 41 (5.94 m) remains obtained from the two-enthalpy feed results with total vaporiza-
much greater than the diameter at the vapor return stage (4.21 tion of the feed in Table 6, but with QR,total reduced by the energy
m). With much larger withdrawal rates (Fwithdr ) 1800), large required to vaporize the liquid feed (Qvaporize ) 11 086 kW).
reductions in diameter are possible and the column diameters This result is shown as the base case in Table 7. This system is
essentially balance (4.13 m at stage 41 and 4.17 m at stage 33). unusual in that not only is less energy needed with a vapor feed,
By increasing the number of stages by one, we can match the but also the column diameter is significantly less than with a
base purities and balance the column diameters (4.11 m at stage liquid feed because much of the energy required is input with
42 and 4.10 m at stage 33). For this water-acetic acid separation the vapor feed away from the stage that requires the largest
the use of an intermediate reboiler has the advantages that liquid diameter. Base case column diameters are tray 2, 3.76 m; tray
can be withdrawn away from the pinch point that occurs at high 27, 3.91 m; tray 28, 3.38 m; and tray 41, 5.30 m. The size of
water concentrations and the withdrawal rate can be greater than a column with vapor feed can be reduced with vapor bypass or
the feed rate. With Fwithdr ) 1800 the amount of energy needed with an intermediate reboiler, but since the intermediate reboiler
 Ind. Eng. Chem. Res., Vol. 46, No. 26, 2007 9229

Table 8. Results for Distillation of a Liquid Feed of 50% Water, 50% Acetic Acida
FV NF,L NF,V N dia A vol tray QR,total Qc decr vol (%)
0 (base) 30 42 5.15 20.8 389.9 41 22 704 -22 679
Intermediate Reboiler
Fwithdr NF,L NL,with NV,ret N dia A vol tray QR,total Qc decr vol (%)
625 31 34 35 43 4.57 16.4 314.9 35/42 22 704 -22 679 19.2
900 31 34 35 43 4.56 16.3 313.0 34 22 704 -22 679 19.7
1000 31 34 35 43 4.58 16.5 316.8 34 22 704 -22 679 18.7
Two-Enthalpy Feed
FV NF,L NF,V N dia A vol tray QR,total Qc decr vol (%)
430 30 32 43 4.69 17.3 324.3 42 22 704 -22 679 16.8
500 31 33 44 4.62 16.7 328.3 43 22 705 -22 678 15.8
980 31 33 46 4.03 12.8 263.3 45 22 704 -22 679 32.5
1000 34 47 4.02 12.7 267.1 34 22 705 -22 678 31.5
a Feed rate is 1000 kmol/h and operation is at 1.0 atm. L/D ) 3.0, D ) 500 kmol/h, tray spacing is 18 in. ) 0.4572 m, p ) 1.0 atm, x
D,W ) 0.994 33,
and xB,W ) 0.005 67. All simulations have essentially the same value of QR,total. Decrease in volume is compared to the base case in this table. Notation is
the same as in Table 3. All systems meet purity requirements.

was much more effective only this result is shown (Table 7). Table 9. Relative Heat Exchanger Areas, A ) Q/(U∆T), for
Distillation of a Liquid Feed of 5% Acetic Acid and 95% Water
The intermediate reboiler system with Fwithdr ) 50 kmol/h
Comparing the Intermediate Reboiler System (Fwithdr ) 1800
matches the base case product purities, but the reduction in kmol/h) with the Base Case (see Table 6)a
column volume is only 1.8%. If the withdrawal rate is increased
steam T (°C) ∆TR ∆Tint_reb Atotal/Abase
to Fwithdr ) 75 kmol/h (not shown in the table), the bottoms
purity requirement is no longer met. If we increase N to 43 and 122 5.6 20.9 0.60
132 15.6 30.9 0.73
keep increasing the withdrawal rate until the diameters at the 142 25.6 40.9 0.79
feed stage and the bottom tray balance (Fwithdr ) 1100 kmol/ 152 35.6 50.9 0.83
h), we obtain the same purities as the base case, but with a a The column reboilers are at 389.55 K, and the liquid withdrawn from
column volume that is 43.7% smaller (Table 7). stage 33 in the intermediate reboiler system is at 374.25 K. QR,total ) 35 560
Distillation of a 50 mol % Water, 50 mol % Acetic Acid kW for both systems; for the intermediate reboiler system QR ) 15 870
Liquid Feed. The base case column had 40 equilibrium trays, kW, and Qint_reb ) 19 690 kW. Heat transfer coefficient U is assumed to
a kettle reboiler, and a total condenser (N ) 42). External reflux be the same for all systems. Atotal/Abase is the ratio of the total heat exchanger
area of the intermediate reboiler system divided by the heat exchanger area
ratio was 3.0. The optimum feed location was NF ) 30. Product of the base case.
purities are xD,W ) 0.994 33 and xB,W ) 0.005 67. Results are
shown in Table 8. The calculated diameters are 2.94 (stage 2),
Heat Exchangers
4.06 (feed stage ) 30), and 5.15 m (stage 41). This more
concentrated feed is still a difficult separation, but the diameters The effects of intermediate reboilers and two-enthalpy feed
and energy loads are reduced despite a considerably higher on the cost of heat exchangers and the cost of heating depends
purity bottoms product. on the situation in each plant. Since the intermediate reboiler
The intermediate reboiler results (Table 8) show a decrease and the heater for the feed operate at lower temperatures than
in calculated column volume as Fwithdr increases up to Fwithdr ) the column reboiler, they can use a lower temperature heating
625 kmol/h, where the calculated column diameters are identical medium. If this lower temperature heating medium is less
at the bottom of the column and stage 35 (vapor return stage). expensive, then operating costs can be reduced.2,5
Above this withdrawal rate (e.g., to Fwithdr ) 900 kmol/h) stage Since the cost per area of heat exchangers decreases as the
34 becomes the stage with the largest calculated diameter and area increases,9 splitting the single reboiler of the base case into
there is very little advantage to increasing the withdrawal rate. two heat exchangers for the intermediate reboiler and two-
Increasing the withdrawal rate to Fwithdr ) 1000 kmol/h causes enthalpy feed systems would be expected to increase the costs
the diameter to increase, and the purities just match the base of the heat exchangers. However, if the same heating medium
case. Further increases of the withdrawal rate will cause the is used, the driving force ∆T for heat transfer is larger for the
purities to not meet the product specifications. intermediate reboiler and the feed heater than for the column
The two-enthalpy feed method was also simulated for this reboiler. This will result in a smaller total area for systems with
separation (Table 8). The run with FV ) 430 is better than the an intermediate reboiler or feed heater when QR,total is constant.
run with FV ) 500 because the increase in the amount of feed This reduction in area can be large if the temperature of the
vaporized requires an extra tray to reach the desired purity. The heating medium is close to the boiling temperature in the
best results were obtained with FV ) 980, although the fully reboiler. For example, consider the comparison of the intermedi-
vaporized feed system (FV ) 1000) is almost as good and is ate reboiler system to the base case for the acetic acid-water
simpler. With this feed concentration the two-enthalpy feed distillation delineated in Table 9. For the lower steam temper-
system works well because the feed stage is no longer in the atures the difference between the two driving forces is quite
region of very low volatilities. large, and the total heat exchanger area for the intermediate
Subtracting the energy required to vaporize the feed (9171.5 reboiler system is considerably less than for the base case.
kW) from the FV ) 1000 results in Table 8 gives the results Despite the cost advantage per area of the larger heat exchanger,
for a saturated vapor feed of the 50-50 mixture. Since the the cost for the heat exchangers for the intermediate reboiler
diameter at stage 34 (4.02 m) is almost identical to the diameter system will be very competitive. Since the smaller distillation
at stage 46 (4.01 m), no further reductions in column volume column of the intermediate reboiler system is less expensive
are possible for a saturated vapor feed. than the base case distillation column,9 the overall system costs
9230 Ind. Eng. Chem. Res., Vol. 46, No. 26, 2007

Table 10. Results for Absorption of n-Pentane from Methanea

FBy NBy,V dia A vol tray L0 LN V1 decr vol (%) incr L0 (%)
0 (base) 2.01 3.17 15.5 7 466 666 800
140 4 1.90 2.85 13.9 7/4 473 673 800 10.1 1.4
a Methane purity in outlet gas matches the base case purity (99.03%). Both gas and liquid are at 20 °C and 3.0 atm. Column pressure is 3.0 atm, the
column is adiabatic, and the plate spacing is 24 in. ) 0.6096 m. The feed is 99.41% vapor and 0.59% liquid. FBy is the kmol/h vapor in the vapor bypass,
and NBy,V is the feed location for the vapor bypass. Flow rates of entering solvent L0, exiting liquid LN, and treated vapor V1 are in kmol/h.

will be less. With higher steam temperatures the heat exchanger applied to different types of trays and different random and
areas of the two systems start to approach each other and the structured packings. Retrofitting, which is illustrated elsewhere,1
heat exchangers for the intermediate reboiler system will become is probably easier with tray systems than with packed systems.
more expensive; however, the overall costs may still be less Once the final design has been determined, a tray rating or
for the intermediate reboiler system. For accurate economic packing rating analysis should be done to ensure proper
estimates, the costs need to be determined on a case-by-case hydraulic behavior.1 The ternary example showed that this
basis. method of diameter balancing can be applied to multicomponent
systems. This also applies to azeotropic and extractive distillation
Vapor Bypass for Concentrated Absorbers if there is a major change in column diameter in the column.
The processing method that maximized volume reduction with
Since gas enters at the bottom of the absorber and much of constant energy use depended upon the separation being
the solute is transferred to the liquid, the vapor flow rate performed and the feed concentration. Other factors such as ease
decreases as one goes up the absorber. This effect is particularly of operation or low-cost retrofitting may also be important. For
marked for concentrated feed streams. Thus, the largest calcu- example, if a distillation system with excess reboiler and
lated diameter for concentrated absorbers is at the bottom of condenser capacities is limited by the vapor flow rate at the
the column. Although in the particular chemical system studied bottom of the column, modest debottlenecking can easily be
here condensation of a portion of the gas stream is reasonably achieved with vapor bypass (Figure 3a). Vapor bypass would
convenient, in many cases this is not true. When the gas stream probably not be the method of choice for a new column.
is difficult to partially condense, neither two-feed nor intermedi- When there are very large changes in calculated column
ate reboiler systems are convenient. Thus, we will explore only diameter, distillation systems are often constructed with two
the use of vapor bypass (Figure 4) for concentrated absorbers. columns of different diameters. Use of the methods in this paper
Absorption of n-Pentane from Methane Gas with Heavy will reduce the diameter of the larger column (in this paper the
Solvent. Table 10 shows the results for the base case absorption stripping section), and may allow use of a single, constant
column with seven equilibrium stages processing 1000 kmol/h diameter column.
of a feed stream that is 80 mol % methane and 20 mol % Obviously, the economically optimum system for any given
pentane. The inlet solvent is pure and is assumed to have the separation problem depends upon the detailed economic costs
same properties as n-nonane. Calculations were done with the and needs to be calculated for each specific case. Keller10 found
AspenPlus simulator using the Peng-Robinson VLE correlation. for distillation that the ratio of capital cost to utility costs ranged
For the base case the inlet liquid flow rate was 466 kmol/h; the from about 0.55 to 3.5; thus, the capital cost is always
outlet gas stream is 99.03 mol % methane, 0.73% n-pentane, significant. The large reductions in column volume for the acetic
and 0.25% solvent; and the outlet liquid stream is 1.13% acid systems will clearly result in significant capital savings.
methane, 29.16% n-pentane, and 69.71% solvent. The calculated The significance of the savings for the other systems needs to
diameters and vapor flow rates for the base case are stage 1 ) be determined from more detailed economic calculations.
1.67 m, V1 ) 800 kmol/h; and stage 7 ) 2.01 m, V7 ) 952
kmol/h. Nomenclature
For vapor bypass part of the concentrated feed gas bypasses
the bottom of the column where the calculated diameter is A ) column cross-sectional area, m2
largest. The best results with the same outlet methane mole D ) distillate flow rate, kmol/h
fraction (99.03%) as the base case are shown in Table 10. dia ) calculated column diameter, m
Putting the bypass lower in the column improved the purity F ) feed rate, kmol/h
and required less increase in solvent flow rate to match the purity FBy ) rate of vapor bypass, kmol/h
of the base case, but there was less reduction in vapor flow FL ) feed rate of liquid in two-enthalpy feed column, kmol/h
rate and hence a larger final diameter. Putting the bypass higher FV ) feed rate of vapor in two-enthalpy feed column, kmol/h
in the column reduced the purity more but reduced the calculated Fwithdr ) withdrawal rate of liquid stream sent to intermediate
diameter more. Repeated simulations showed that stage 4 was reboiler, kmol/h
the best compromise to input the bypass vapor feed. The Lj ) liquid flow rate leaving stage j, kmol/h
calculated diameters and vapor flow rates for the vapor bypass L/D ) external reflux ratio
run with FBy ) 140 are stage 1 ) 1.67 m, V1 ) 800 kmol/h; N ) number of trays + condenser + reboiler
stage 4 ) 1.91 m, V4 ) 887 kmol/h; and stage 7 ) 1.91 m, V7 NBy,V ) feed location for vapor bypass
) 816 kmol/h. Nfeed ) optimum feed stage with condenser labeled as stage 1
NF,L, NF,V ) optimum feed locations for the liquid and vapor
Discussion portions of feed
NV,ret ) optimum location to return vapor from intermediate
Since the diameter balancing method is a processing technique reboiler
for distillation and absorption, it is generally applicable to any NV,with ) optimum withdrawal plate for the vapor Fwithdr
type of equipment used for these unit operations. Thus, the p ) column pressure, atm
methods developed in this and the previous paper1 can be Qc ) condenser duty, kW
 Ind. Eng. Chem. Res., Vol. 46, No. 26, 2007 9231

Qheat_feed ) energy used to heat the entire feed to produce a (3) Liebert, T. Distillation Feed PreheatsIs it Energy Efficient?
two-phase feed, kW Hydrocarbon Process. 1993, 72 (10), 37.
QR ) reboiler duty, kW (4) Ognisty, T. P. Distillation/Energy Management. In Encyclopedia of
Separation Science; Wilson, I. D., Ed. in Chief, Cooke, M., Poole, C. F.,
QR,total ) total heating duties including intermediate reboiler and Eds.; Academic Press: New York, 1999; Vol. 3, pp 1005-1012.
vaporizing portion of feed, kW (5) Wankat, P. C.; Kessler, D. P. Two-Feed Distillation: Same
Qvaporize ) energy required to vaporize the vapor bypass, the Composition Feeds with Different Enthalpies. Ind. Eng. Chem. Res. 1993,
portion of the feed, or in the intermediate reboiler, kW 32 (12), 3061-3067.
Vj ) vapor flow rate leaving stage j, kmol/h (6) Fair, J. R. Gas Absorption and Gas-Liquid System Design. In Perry’s
vol ) column volume, m3 Chemical Engineers’ Handbook, 7th ed.; Perry, R. H., Green, D. W., Eds.;
McGraw-Hill: New York, 1997; pp 13-23 to 13-38.
xB,i ) mole fraction component i in bottoms product
(7) Huang, H.-P.; Lee, H.-Y.; Gau, T. K.; Chien, I. L. Design and Control
xD,i ) mole fraction component i in distillate of Acetic Acid Dehydration Column with p-Xylene or m-Xylene Feed
Subscripts Impurity. I. Importance of Feed Tray Location on the Process Design. Ind.
Eng. Chem. Res. 2007, 46, 505-517.
AA ) acetic acid (8) Seader, J. D. Distillation. In Perry’s Chemical Engineers’ Handbook,
C3 ) propane 7th ed.; Perry, R. H., Green, D. W., Eds.; McGraw-Hill: New York, 1997;
C4 ) n-butane Section 13.
C5 ) n-pentane (9) Turton, R.; Baillie, R. C.; Whiting, W. B.; Shaeiwitz, J. A. Analysis,
Synthesis and Design of Chemical Processes, 2nd ed.; Prentice Hall PTR:
feed_st ) feed stage Upper Saddle River, NJ, 2003; Appendix A.
M ) methanol (10) Keller, G. E., II. Separations: New Directions for an Old Field;
W ) water AIChE Monograph Series; American Institute of Chemical Engineers: New
York, 1987; No. 17.
Literature Cited
ReceiVed for reView July 20, 2007
(1) Wankat, P. C. Balancing Diameters of Distillation Column with ReVised manuscript receiVed September 19, 2007
Vapor Feeds. Ind. Eng. Chem. Res. 2007, 46, DOI: 10.1021/ie0705554. Accepted October 2, 2007
(2) King, C. J. Separation Processes, 2nd ed.; McGraw-Hill: New York,
1981; pp 704-707. IE0709887