1206 Ind. Eng. Chem. Res.

2011, 50, 1206–1218

Design and Control of the Acetone Process via Dehydrogenation of 2-Propanol

William L. Luyben*
Department of Chemical Engineering, Lehigh UniVersity, Bethlehem, PennsylVania 18015

Acetone is produced via several alternative processes, one of which is the dehydrogenation of 2-propanol
(IPA). The endothermic gas-phase reaction converts IPA to acetone and hydrogen. The process consists
of a vaporizer, heated tubular reactor, flash tank, absorber, and two distillation columns. The liquid
fresh feed is a mixture of IPA and water. It is combined with a small IPA/water recycle stream, vaporized,
and fed into the vapor-phase reactor, which is heated by high-temperature molten salt. Reactor effluent
is cooled and fed to a flash tank. The gas from the tank contains most of the hydrogen but also some
acetone. This gas is fed to an absorber in which a water stream is used to recover acetone. The liquids
streams from the base of the absorber and the flash tank are fed to the first distillation column, which
produces high-purity acetone out the top. There is also a vapor vent stream leaving the reflux drum of
this column to remove the small amount of hydrogen dissolved in the feed. The second distillation
column produces a high-purity water bottoms and a distillate with a composition near the IPA/water
azeotrope, which is recycled back to the vaporizer. There are a number of interacting design optimization
variables in this process, which illustrate some interesting design trade-offs. Removing the hydrogen
without losing too much product acetone is the main challenge. Losses can occur in both the absorber
off-gas and the column vent. Raising absorber pressure decreases off-gas losses but increases vent losses.
Raising absorber pressure has several other effects. It raises the vaporizer pressure, which raises the
required temperature and cost of the vaporizer heat source. It adversely affects kinetics because the
reaction is nonequimolar and conversion decreases with increasing pressure. A higher reactor temperature
is required to achieve the desired conversion. The purpose of this paper is to develop the economically
optimum design considering capital costs, energy costs, and raw material costs and then to develop a
plantwide control structure capable of effectively handling large disturbances in production rate.

1. Introduction quantitative illustration of the dominating economic effects of

reactant costs (order of magnitude larger than energy or capital
The development of an economical, safe, environmentally costs). This effect is completely generic (“Douglas Doctrine”)
friendly chemical process involves many engineering trade-offs. and of vital importance in process design.
The classical trade-off is between reactor size and recycle. But
The purpose of the current paper is to use the acetone process
there are many other trade-offs that come into play during the
to illustrate some additional interesting design optimization
design stages when topology, equipment, and operating condi-
features and principles. The principle feature of the acetone
tions are being specified.
process is the efficient removal of the light hydrogen product
A number of design trade-offs have been illustrated in several
so that losses of reactant 2-propanol (IPA) and product acetone
typically large processes that have been described in the
are minimized.
literature over the years. The pioneering work by Douglas1
presented the first and most widely studied example, the hydro-
dealkylation of toluene to form benzene (HDA). A series of 2. Process Studied
recent papers2–5 have illustrated a number of design trade-offs, Figure 1 shows the acetone process described in Turton et
in addition to demonstrating many fundamental principles of al.6 The authors provide a good description of the basic units
design and control. These papers deal with specific chemical in the process with equipment sizes and operating conditions.
processes, but each of these processes contains generic features Their intent is not to determine the optimum design but to
that are encountered in a wide variety of chemical processes. illustrate a typical multiunit chemical process that features a
The alkylation process2 illustrates the general problem of how reaction section coupled with a separation section and connected
to purge out inert components entering in the fresh feeds. It by a recycle stream.
also illustrates that the location in the flowsheet where the fresh The fresh feed contains 34.81 kmol/h of IPA. Notice that
feed is introduced can have a dramatic effect on both design
the absorber is operated at low pressure (1.5 atm), which results
and control. In addition, it deals with a case in which low reactor
in a loss of acetone in the off-gas of 2.52 kmol/h (7.2% loss).
temperatures require expensive refrigeration, which impacts the
The acetone lost in the vent stream from the first distillation
economic optimization.
column reflux drum is small (0.0335 kmol/h) because there is
The MIPA process3 illustrates the general problem encoun-
very little hydrogen in the column feed at the low operating
tered when there are two recycle streams that are affected in
pressure in the flash drum and absorber.
opposite directions by changing reactor conditions. The ethyl
benzene process4 illustrates a process in which one product is We will consider in a later section a revised flowsheet in
“recycled to extinction.” The cumene process5 provides a which the absorber pressure is raised to reduce off-gas acetone
losses. However, higher pressure increases dissolved hydrogen
* To whom correspondence should be addressed. E-mail: WLL0@ in the column feed and results in larger losses of acetone in the
Lehigh.edu. Tel.: 610-758-4256. Fax: 610-758-5057. vent stream.
10.1021/ie901923a  2011 American Chemical Society
Published on Web 03/01/2010
 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011 1207

Figure 1. Turton design.

Table 1. Reaction Kinetics find the effect of pressure on conversion. At 623 K (the
modified reversible temperature in the Turton flowsheet), the RGIBBS equilibrium
Turton
irreversible forward reverse
conversion is 97.1% at 2 atm. This should be compared with
the 90% conversion in the Turton design. As pressures are raised
k 3.51 × 10 5
22 × 10 6
1000 to 5, 10, and 15 atm, the equilibrium conversions decrease to
E (kJ/kmol) 72 380 72 380 9480 93.3, 87.8, and 83.2%. Clearly the effect of pressure must be
concentration terms CIPA CIPA CacetoneCH2
(kmol/m3) considered.
The kinetics were modified to assume a reversible reaction:
2.1. Reaction Kinetics. The production of acetone involves (CH3)2CHOH S (CH3)2CO + H2 (2)
the dehydrogenation of IPA in a high-temperature gas-phase
reactor. The forward and reverse overall reaction rates are given in eq
3, with units of kmol s-1 m-3:
(CH3)2CHOH f (CH3)2CO + H2 (1)
RF ) CIPAkFe-72380/
(RT)

6 (3)
Turton et al assumed the reaction to be irreversible and specify RR ) CacetoneCH2kRe-9480/
(RT)

that the per-pass conversion of IPA is 90% to prevent side

reactions. The reaction occurs in the vapor phase in the presence The activation energy of the forward reaction used is the value
of a solid catalyst (assumed to have 0.5 void fraction and a given in Turton et al (72 380 kJ/kmol). The activation energy
2000 kg/m3 solid density). of the reverse reaction was calculated using the heat of reaction
Using the reactor size and kinetics given in Turton et al as (+62 900 kJ/kmol).
shown in Table 1 gave the 90% conversion specified by the
authors if the presence of the solid catalyst was neglected. Of λ ) E F - ER (4)
course, the catalyst is present, so a bigger reactor would have
to be used to give the same conversion. An empirical trial-and-error procedure was employed to find
The kinetics used by Turton et al are also unrealistic in that the unknown values of the pre-exponential factors kF and kR
the reaction cannot be irreversible. If it were irreversible, we that satisfied two conditions. First, they should give ap-
could operate at high pressure, which would raise the concentra- proximately the same pressure dependence of conversion in a
tion of IPA and drive the reaction toward the products. But large tubular reactor as predicted by RGIBBS. Second, they
LeChatier’s principle tells us that raising pressure should drive should give 90% conversion when used in the Turton tubular
the reaction toward the reactant since there are two moles of reactor (including the presence of catalyst). Table 1 shows the
product generated from one mole of reactant. parameters used in the rest of this study. To be consistent with
In an attempt to modify the kinetics so as to capture these the assumption made by Turton, the conversion in the reactor
realistic effects, an RGIBBS reactor was run in Aspen Plus to is held at 90% as other design parameters are changed.
1208 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011

Figure 3. (A) Txy diagram for acetone/water; (B) xy diagram for acetone/
water.
Figure 2. (A) Txy diagram for IPA/water; (B) xy diagram for IPA/water.
The separation is fairly easy, so the column will not require
2.2. Phase Equilibrium. Two phase equilibrium features many trays and can run with a low reflux ratio.
dominate the separations required in this process. The first is The normal boiling point of acetone is 329.4 K, so it is the
the high volatility of hydrogen compared to the other compo- low-boiler in the system if hydrogen is disregarded. Figure 3
nents. After cooling the mixture leaving the reactor, it is flashed gives the Txy and xy diagrams for acetone and water at
in the separator. Most of the hydrogen goes into the gas stream, atmospheric pressure. There is no azeotrope, but there is a pinch
but a small amount is dissolved in the liquid. This hydrogen at the high acetone end of the diagrams. This means that
ends up in the condenser of the first distillation column and producing high-purity acetone will require a column with many
would require a very low temperature to condense. Therefore a trays and a fairly high reflux ratio.
Figure 4 shows the ternary diagram for the acetone/IPA/water
small vent stream is removed from the top of the reflux drum.
system at 1 atm. There are two distillation regions. The feed to
Any acetone in this vent stream is lost and only has fuel value.
the first distillation column lies in the lower region. The residue
The second important phase equilibrium feature is the curves indicate that a high-purity acetone can be produced in
existence of an azeotrope in the IPA/water system. The normal the distillate, and the bottoms will be a mixture of mostly IPA
boiling points of IPA and water are 355.4 and 373 K, and water. The location of the feed, distillate, and bottoms
respectively. The azeotrope contains 67.32 mol % IPA at 1 atm stream compositions for the first distillation are shown and lie
and 353.4 K. Figure 2 gives the Txy and xy diagrams for IPA/ on a straight line.
water at atmospheric pressure. UNIQUAC physical properties Figure 4 also gives results from the Aspen Conceptual Design
are used in the Aspen simulations. The existence of the program. The feed composition is specified at 44.3 mol %
azeotrope means that the IPA that is unconverted in the reactor acetone, 5.3 mol % IPA, and 50.4 mol % water. The three
and is recycled from the top of the second distillation column compositions specified are xD(W) ) 0.001, xD(IPA) ) 0.000001
will have a composition close to the azeotropic composition. and xB(A) ) 0.0001. Setting the reflux ratio at 3.9 produces a
 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011 1209

Figure 4. Ternary diagram for acetone/IPA/water.

column with 50 stages feeding on stage 45. These results are

similar to those of the Turton design.

3. Turton Flowsheet
Figure 1 shows the flowsheet of the process with the
equipment sizes and conditions presented by Turton et al.6 The
fresh liquid feed is 51.96 kmol/h of a mixture that is 67 mol %
IPA and 33 mol % water.
3.1. Vaporizer. The fresh feed is fed into a vaporizer along
with 5.88 kmol/h of a recycle stream that is 65 mol % IPA and
35 mol % water. The vaporizer operates at 2.6 atm and 389 K
in this design because the absorber downstream is operated at
low pressure (1.5 atm). Low-pressure steam at 433 K can be
used as the heat source in the vaporizer with a cost of $7.78
per GJ.
If the pressure in the absorber is raised to improve acetone
recovery, the pressure and temperature in the vaporizer will
increase. At higher pressures, the vaporizer will have to use
more expensive medium- pressure steam (457 K at $8.22 per Figure 5. Turton reactor temperature profile.
GJ) or high-pressure steam (527 K at $9.83 per GJ). These
alternative operating conditions are considered in the next The heat-transfer coefficient in the reactor is 60 W m-2 K-1.
section. Temperatures in the reactor vary with length. Figure 5 shows
3.2. Reactor. The vapor stream is fed into a tubular reactor the reactor temperature profile. The reactor exit temperature is
containing 450 tubes, 0.0504 m in diameter and 6.096 m in 623 K. Since the reaction is endothermic, the higher the
length. The endothermic heat of reaction is provided by a hot temperature is, the larger is the chemical equilibrium constant.
molten salt heating medium at 624 K. The heat duty is 0.960 As we will see in the next section, if reactor pressure is
MW, which is assumed to be provided by a fired furnace at a increased, which drives the reaction toward the reactant, the
price of $9.83 per GJ. The catalyst in the reactor tube has a reactor temperature must be increased to maintain the same
void fraction of 0.5 and a solid density of 2000 kg/m3. conversion.
1210 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011

Figure 6. The 17-atm design.

Table 2. Effect of Absorber Pressurea

absorber pressure off-gas acetone vent acetone total acetone vaporizer reactor heating
(atm) loss (kmol/h) loss (kmol/h) loss (kmol/h) temp (K) medium temp (K)
5 2.336 0.1033 2.469 409 676
7 1.270 0.1953 1.465 420 695
9 0.7679 0.2152 0.9831 429 710
11 0.4858 0.2711 0.7569 437 723
13 0.3110 0.3269 0.6487 444 734
15 0.1990 0.3826 0.5816 450 743
17 0.1241 0.4383 0.5624 456 752
19 0.0745 0.4938 0.5683 461 760
21 0.0422 0.5493 0.5915 466 767
a
Water flow rate ) 20 kmol/h; 300 reactor tubes.

Table 3. Effect of Absorber Watera

absorber water off-gas acetone vent acetone total acetone reboiler reboiler
(kmol/h) loss (kmol/h) loss (kmol/h) loss (kmol/h) duty C1 (WM) duty C2 (WM)
20 0.1241 0.4383 0.5624 1.066 0.1158
25 0.0522 0.4581 0.5103 1.073 0.1158
28 0.0225 0.4698 0.4923 1.078 0.1158
30 0.0103 0.4776 0.4879 1.080 0.1159
32 0.0040 0.4854 0.4894 1.083 0.1159
35 0.0008 0.4970 0.4978 1.088 0.1160
a
Absorber pressure ) 17 atm; 300 reactor tubes.

3.3. Heat Exchangers, Flash Tank, and Absorber. The hot stage absorber. Water is fed on the top tray at 20 kmol/h and
reactor effluent is cooled with cooling water to 318 K and then 320 K. The off-gas from the absorber is mostly hydrogen, but
cooled to 293 K using refrigeration. A low temperature helps 2.52 kmol/h of acetone are lost in this stream (39.76 kmol/h
to reduce the loss of acetone in the gas stream leaving the flash with 6.34 mol % acetone). The diameter of the absorber is 0.217
tank and subsequently in the absorber off-gas. But refrigeration m.
is much more expensive than cooling water. If the absorber and The liquid from the flash tank and the liquid from the base
flash tank were run at higher pressure, refrigeration would not of the absorber are fed to the first distillation column. They
be required. The refrigeration load is 0.22 MW at a cost of $4.43 both are mixtures of acetone, IPA, and water with a small
per GJ at this temperature level. amount of dissolved hydrogen (0.02 mol % H2 in the flash tank
The two-phase mixture from the partial condenser is sent to liquid and 0.01 mol % H2 in the absorber bottoms). This
a flash drum. The vapor stream is fed into the bottom of a 10- hydrogen goes overhead in the first distillation column and will
 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011 1211
ratio required to achieve the design specification of 99.9 mol
% acetone is 2.78. The specification for the bottoms is 0.01
mol % acetone impurity. The column diameter is 0.8915 m and
the reboiler heat input is 1.217 MW.
3.5. Water Column C2. The column has 20 stages and is
fed on stage 16. The operating pressure is 1 atm, which gives
a reflux-drum temperature of 353 K. The reflux ratio is low
(RR ) 0.849). Energy consumption is 0.120 MW, and the
column diameter is 0.251 m. The design specifications are 0.1
mol % IPA in the bottoms (the water product) and 65 mol %
IPA in the distillate (reasonably close to the azeotropic
composition).

4. Revised Flowsheet
The design discussed in the previous section has not been
optimized from the standpoint of economics. The loss of acetone
is quite high. Refrigeration is required in one heat exchanger.
An obvious parameter to vary is the absorber pressure. Figure
6 gives the revised flowsheet for a design in which the absorber
pressure is raised to 17 atm.
4.1. Effect of Absorber Pressure. As previously mentioned,
raising absorber pressure should reduce acetone losses in the
absorber off-gas. But it will raise the pressure in the reactor,
which adversely affects conversion. To keep conversion at 90%,
reactor temperatures will have to be increased. There may be a
maximum temperature limitation due to loss of catalyst activity.
Howard7 gives typical reactor operating temperatures of 673-773
K, so we will assume that the temperature of the hot medium
heating the reactor cannot be higher than 760 K. Notice that
the temperature of the heating medium in the Turton design is
624 K, which is far below this limitation.
Raising absorber pressure will also raise the pressure and
temperature in the vaporizer, which means a higher-temperature,
more expensive energy source must be used. On the other hand,
a higher pressure in the absorber should mean some steam can
be generated from the hot reactor effluent. In addition, cooling
water instead of refrigeration can be used in the condenser
Figure 7. (A) Reactor temperature profile, 17-atm design; (B) reactor because a low temperature is not required at the higher pressure.
composition profiles, 17-atm design. Table 2 shows the effects of absorber pressure on a number
of parameters. As absorber pressure increases, the acetone losses
Table 4. Optimum Distillation Column Design
in the absorber off-gas decrease, but the acetone losses in the
C1 distillation column reflux-drum vent increase. The minimum
NT 67 77 87 97 total loss occurs at an absorber pressure of 17 atm.
ID (m2) 0.932 0.905 0.8847 0.868 At this pressure, the vaporizer temperature has increased to
QR1 (MW) 1.188 1.128 1.085 1.050 456 K from the Turton design of 389 K. Low-pressure steam
QC1 (MW) 1.135 1.072 1.027 0.9893 at 433 K can be used in the Turton design, in which the heat
TAC (106 $/y) 0.5075 0.5014 0.4996 0.5001
duty in the vaporizer is 0.7577 MW. High-pressure steam at
C2 527 K must be used in the 17-atm design, in which the heat
NT 15 20 25 duty is 0.866 MW. Low-pressure steam costs $7.78 per GJ.
ID (m2) 0.262 0.2505 0.2465 High-pressure steam costs $9.83 per GJ. Therefore, vaporizer
QR1 (MW) 0.1255 0.1159 0.1133
QC1 (MW) 0.1298 0.1189 0.1150 energy cost is higher in the 17-atm design.
TAC (106 $/y) 0.05546 0.05461 0.05606 However, low-pressure steam can be generated in the first
heat exchanger by the hot gases leaving the reactor instead of
blanket the condenser unless vented off. With the low-pressure wasting the heat by using cooling water. The heat duty in this
operation of the flash tank and absorber, the acetone lost in the steam generator is 0.4944 MW, which constitutes an energy
vent stream is quite low (0.0335 kmol/h of acetone). Raising credit. In addition, refrigeration is not needed in the second heat
the pressure in the flash tank and absorber will decrease the exchanger. The Turton design requires 0.2203 MW of refrigera-
acetone losses in the absorber off-gas but increase the acetone tion at a cost of $4.2 per GJ.
losses in the column vent. Increasing absorber pressure also requires an increase in
3.4. Acetone Column C1. The column in the Turton design reactor temperature to maintain the desired 90% per pass
has 67 stages and is fed on stage 54, which is the optimum conversion of IPA. Reactor energy consumption increases from
feed stage to minimize reboiler heat input. The operating 0.960 MW in the Turton design to 1.098 MW in the 17-atm
pressure is 1 atm, which gives a reflux-drum temperature of design. An energy cost of $9.83 per GJ is assumed to be required
320 K, so cooling water can be used in the condenser. The reflux in all designs.
1212 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011

Table 5. Basis of Economics and Equipment Sizing

column
diameter Aspen tray sizing
length NT trays with 2 ft spacing plus
20% extra length
vessel (diameter and capital cost )
length in meters) 17640(D)1.066(L)0.802
condensers (area in m2) heat-transfer coefficient )
0.852 kW/K-m2
differential temperature )
reflux-drum temperature -
310 K
capital cost ) 7296(area)0.65
reboilers (area in m2): heat-transfer coefficient )
0.568 kW/K-m2
differential temperature ) 403
K - base temperature
capital cost ) 7296(area)0.65
reactor (heated) heat-transfer coefficient ) 60
W/K-m2
differential temperature ) TH
- TR(z)
capital cost ) 7296(area)0.65
vaporizer (area in m2) heat-transfer coefficient )
0.568 kW/K-m2
differential temperature )
Tsteam - Tvaporizer
capital cost ) 7296(area)0.65
flash tank (diameter and capital cost )
length in meters) 17640(D)1.066(L)0.802
absorber (diameter and
length in meters)
diameter aspen tray sizing
length NT trays with 2 ft spacing plus
20% extra length
vessel (diameter and capital cost ) 17,640
length in meters) (D)1.066(L)0.802
energy costs fired furnace ) $9.83 per GJ
HP steam ) $9.83 per GJ
MP steam ) $8.22 per GJ
LP steam ) $7.78 per GJ
refrigeration ) $4.42 per GJ
TAC ) (capital payback period ) 3 years
Figure 8. (A) Column C1 temperature profile; (B) column C2 temperature cost)/(payback
profile. period) + energy cost
The higher reactor temperatures require an increase in the
temperature of the heating medium in the reactor. If there is a Since conversion in the reactor is fixed at 90% and reactor
maximum catalyst temperature constraint, this may impose a pressure is fixed at 17.5 atm by the absorber pressure of 17
constraint on the maximum absorber pressure. In the Turton atm, the required reactor temperature does not change as the
design, the temperature of the heating medium is 624 K. In number of tubes is changed. The required heat duty also does
17-atm design, the temperature of the heating medium is 761 not change. So the only thing that increases as the number of
K. Literature7 values of operating temperature indicate a range tubes is decreased is the temperature of the heating medium.
from 673 to 773 K, so the 17-atm design falls under this limit. Since high-temperature energy is assumed to be used in
4.2. Effect of Water Solvent and Absorber Stages. The providing this heat (fired furnace), the temperature level does
two design optimization parameters in the absorber are the flow not affect energy cost.
rate of the water and the number of stages. Very little However, it does affect the wall temperatures in the reactor,
improvement in acetone recovery was observed by increasing which could affect catalyst activity if a high temperature
the number of stages above 10. limitation exists. The literature suggests a limit of 773 K.
Increasing the water flow rate should recover more acetone, Using 300, 250, 200, 150, and 100 tubes in the reactor gives
but it may increase vent losses and increase heat duties in the heating medium temperatures of 752, 754, 761, 778, and 826
distillation column. More water consumption will increase water K, respectively. Therefore a reactor with 200 tubes is selected
procurement costs, but process water is very inexpensive ($0.067 to be well under the maximum catalyst temperature limitation.
per 1000 kg). A flow rate of 30 kmol/h corresponds to a water The capital cost of the reactor in the Turton design with 450
bill of only $317 per year. tubes is $378,000. The capital cost of the 200 tube reactor is
Table 3 shows the effect of water flow rate on acetone losses $223,000.
and reboiler heat duties. Acetone loss is minimized using 30 Reactor temperature and composition profiles are given in
kmol/h of water in the absorber. There is a very small 2% Figure 7.
increase in reboiler duty in the first distillation column in going 4.4. Optimum Distillation Design. The number of trays in
from 20 to 30 kmol/h. Reboiler duty in the second column is the distillation columns in the 17-atm design that produced the
essentially constant. minimum total annual cost were determined. Capital and energy
4.3. Effect of Reactor Size. The final design optimization costs were calculated for a range of number of stages. Table 4
variable in the reaction section of the process is reactor size. presents these results.
 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011 1213
Table 6. Economic Comparison
units Turton design 17-atm design
absorber pressure atm 1.5 17.0
no. reactor tubes 450 200
reactor area m2 434 193
reactor capital investment 106 $ 0.3782 0.2232
reactor energy (fired furnace) MW 0.960 1.098
reactor energy cost 106 $/yr 0.2976 0.3398
HX1 energy MW 0.8993 0.4944
HX1 area m2 11.3 20.6
HX1 capital investment 106 $ 0.0353 0.0521
HX1 energy credit (LP steam) 106 $/yr 0.1213
vaporizer energy MW 0.7577 0.866
vaporizer area m2 73.0 34.8
vaporizer capital investment 106 $ 0.1185 (LP steam) 0.07335 (HP steam)
vaporizer energy cost 106 $/yr 0.1859 0.2685
condenser energy MW 0.2203 0.7814
condenser area m2 12.66 14.72
condenser capital investment 106 $ 0.0380 0.0419
condenser energy cost 106 $/yr 0.0307 (refrigeration)
separator capital investment 106 $ 0.0146 0.0158
absorber capital investment 106 $ 0.0171 0.00888
C1 area reboiler m2 64.9 65.6
C1 area condenser m2 123 120
C1 HX capital investment 106 $ 0.276 0.274
C1 shell capital investment 106 $ 0.428 0.424
C1 energy cost 106 $/yr 0.2086 0.2650
C2 area reboiler m2 8.31 8.00
C2 area condenser m2 4.18 4.17
C2 HX capital investment 106 $ 0.0474 0.0466
C2 shell capital investment 106 $ 0.0319 0.0319
C2 energy cost 106 $/yr 0.0205 0.0284
total capital 106 $ 1.764 1.415
total energy 106 $/yr 0.8420 0.7801
TAC 106 $/yr 1.430 1.252

The optimum number of stages in the first column C1 was 5. Economic Comparison
found to be 87, which is 20 more than in the Turton design.
The optimum number of stages in the second column C2 was The basis for the economic and sizing calculations below are
found to be 20, which is the same as in the Turton design. taken from Douglas,1 and Turton et al.,6 Table 5 summarizes
Temperature profiles for both columns are given in Figure 8. this information.

Figure 9. Plantwide control structure.

1214 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011

Table 7. Column Control Structure Selection simulated using the Aspen Radfrac model. Heat exchangers are
feed reflux-to-feed reflux simulated using the Aspen Heater model. The vaporizer and
composition ratio ratio flash tank are simulated using the Aspen Flash2 model.
0.37602 acetone 0.9222 2.436
6.1. Control Structure. Figure 9 shows the plantwide control
0.06401 IPA structure developed for this process using plantwide process
C1 design 0.39602 acetone 0.9696 2.433 control methodology.8,9 Conventional PI controllers are used
0.04401 IPA in all loops. The various loops are listed below with their
0.41602 acetone 1.018 2.432 controlled and manipulated variables. Items 12, 13, and 16
0.02401 IPA
require some discussion.
0.05317 IPA 0.07986 0.9935 1. Fresh feed of the mixture of IPA and water is flow
0.94675 water controlled. This is the throughput handle.
0.06317 IPA 0.08293 0.8658 2. Vaporizer level is controlled by vaporizer heat input.
0.93675 water
C2 design 0.07317 IPA 0.09407 0.8459
There is no liquid outlet stream.
0.92675 water 3. Reactor outlet temperature is controlled by the temper-
0.08317 IPA 0.1061 0.8383 ature of the molten salt.
0.91675 water 4. The temperature of the process stream leaving the steam
0.09317 IPA 0.1185 0.8345 generator is controlled by manipulating the generation
0.90675 water
rate of low-pressure steam.
5. The temperature of the process stream leaving the
Table 6 gives detailed information of equipment sizes, energy
condenser is controlled by manipulating the flow rate of
consumptions, capital investments, and energy consumptions
cooling water.
for the Turton design and for the 17-atm design. Figure 1 gives
6. The liquid level in the flash tank is controlled by
the Turton flowsheet. Figure 6 gives the 17-atm flowsheet.
manipulating the liquid leaving the bottom of the vessel.
The first column in Table 6 gives economic results for the
7. The pressure in the absorber is controlled by manipulating
original Turton design. Total capital investment is $1,764,000.
the flow rate of the off-gas. This holds the pressure in
The total energy cost is $842,000 per year (energy consumption
the system all the way back to the vaporizer.
in columns, reactor, refrigerated condenser, and vaporizer).
8. The liquid level in the base of the absorber is controlled
The second column in Table 6 gives economic results for
by manipulating the liquid leaving the bottom of the
the case where the pressure in the absorber is raised to 17 atm.
absorber.
Total capital investment is $1,415,000. The total energy cost is
$780,000 per year (energy consumption in columns, reactor and 9. The flow rate of water to the top of the absorber is ratioed
vaporizer with an energy credit for LP steam generated). to the gas feed coming into the absorber from the flash
tank.
In addition to these favorable energy and capital investment
10. Pressures in the two distillation columns are controlled
numbers, the 17-atm design produces 34.31 kmol/h of acetone
by condenser heat removals.
product, compared to the 32.25 kmol/h produced in the Turton
11. Base levels in the distillation columns are controlled by
flowsheet. Using an acetone price of $0.50 per lb, the increase
in annual product sales is $1,150,000. bottoms flow rates.
12. Reflux drum level in the first column is controlled by
Raising the absorber pressure provides a much more eco-
manipulating reflux flow rate because of the fairly high
nomically attractive design. Capital investment and energy
reflux ratio.
consumption are lower, and product yield is increased.
13. Distillate flow rate in the first column is ratioed to reflux
flow rate with the ratio being adjusted by a distillate
6. Plantwide Control
composition controller.
The steady-state Aspen Plus model shown in Figure 6 does 14. The temperature on stage 75 in the first column is
not include pumps and control valves. The first step in studying controlled by manipulating reboiler heat input.
dynamics using a pressure-driven dynamic simulation is to add 15. The reflux drum level in the second column is controlled
these items and to determine the volumes of all vessels. The by manipulating the distillate flow rate.
units with dynamics include the vaporizer, the reactor, the flash 16. The reflux flow rate in the second column is ratioed to
drum, the absorber, and the two distillation columns. Of course, the column feed.
the size of the reactor is known from the steady-state design. 17. The temperature on stage 17 in the second column is
The other units are sized to provide 5 min of liquid holdup when controlled by manipulating reboiler heat input.
at 50% level. The units with liquid surge capacity are the reflux 6.2. Column Control Structure Selection. Most distillation
drums and column bases in the distillation columns, the columns use some type of single-end temperature control
vaporizer, the absorber base, and the flash tank. The reactor is because of its simplicity and low maintenance cost. However,
simulated as a tubular reactor with a constant-temperature this simple structure may not provide effective control for some
heating medium. The distillation columns and the absorber are columns. Even if a single-end control structure is possible, we

Table 8. Controller Parameters

TCR TC1 TC2 CC
SP 750 K 365.6 K 363.8 K 0.001 mf water
transmitter range 700-800 K 300-400 K 300-400 K 0-0.001858 mf water
OP 764.6 K 258.2 kcal/sec 28.23 kcal/sec 0.40375 D2/R2 ratio
OP range 700-800 K 0-516 kcal/sec 0-55 kcal/sec 0-1 D2/R2 ratio
deadtime 3 min 1 min 1 min 3 min
KC 0.359 2.63 2.41 1.51
τI 14.6 min 14.5 min 10.6 min 76.6 min
 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011 1215

Figure 10. Feed flow rate disturbances with single-ended temperature control.

have to decide how to select the other control degree of freedom. In the second distillation column C2, the feed compositions
Normally there is a choice between holding a constant reflux of the light-key component IPA and the heavy-key component
ratio or holding a constant reflux-to-feed ratio. water are varied around the design point (7.32 mol % IPA and
A. Selecting Reflux Ratio or Reflux-to-Feed Ratio. To 92.68 mol % water). The required changes in both the reflux-
explore this question, a series of steady-state runs were made to-feed ratio and the reflux ratio in this column are fairly large.
in which the effects of changes in feed composition on the Therefore a single-end structure may not be effective. However,
required changes in reflux-to-feed ratio and reflux ratio were the performance of this column is not as critical as the first
determined while holding both products at their specified column since the distillate is recycled and the loss of IPA in
composition. the bottoms is quite small. A reflux-to-feed structure is used in
Table 7 gives results of these calculations for the two this column, and the results given in the next section indicate
distillation columns. In the first distillation column C1, the feed that this structure provides adequate control.
compositions of the light-key component acetone and the heavy- B. Selecting Temperature Control Tray Location. Another
key component IPA are varied around the design point (39.6 important issue in control structure selection is the location of
mol % acetone and 4.40 mol % IPA). As more acetone and the tray used for temperature control. A simple approach that
less IPA are fed to the column, the reflux-to-feed ratio increases usually works quite well is to select a location where the
about 5%. On the other hand, the reflux ratio hardly changes at temperatures are changing significantly from tray to tray.
all. This indicates that a single-end control structure with a fixed Looking at the temperature profile for the first column C1 given
reflux ratio may provide effective control. in Figure 8A, we can see that there is a significant slope in the
1216 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011

Figure 11. Feed flow rate disturbances with dual-end control.

temperature profile near the base of the column. Stage 75 is in the temperature profile near the base of the column. Stage
selected for single-end temperature control with a fixed reflux 17 is selected for single-end temperature control with a fixed
ratio. Reboiler heat input is manipulated to control this tray reflux ratio. Reboiler heat input is manipulated to control this
temperature. tray temperature. Since the important objective in this column
It is important to note that a high-purity acetone distillate is to prevent loss of IPA out the bottoms low, a tray location
product is required from this column. The purity of the bottoms near the bottom should be effective.
is less important because any acetone that drops out the bottom 6.3. Dynamic Performance Results. All liquid level con-
will go overhead in the next column and be recycled back to trollers are proportional only using Kc ) 2. There is one
the reactor in the recycle stream. Conventional distillation exception to this level control tuning. A gain of 4 is used in the
control wisdom suggests that to achieve effective inferential reflux drum level controller in the second column. The level
composition control of the distillate by using a tray temperature,
sets the distillate, which is a recycle stream. Using a higher
a tray near the top of the column should be controlled. However,
gain helps to speed up the overall dynamics of the recycle
the break in the temperature in this column is near the bottom.
system. The required changes in the flow rate of the recycle
For this reason, a single-end temperature control may not be
stream occur more quickly with the higher gain, which helps
effective in this application. We will demonstrate this problem
in the next section and show that a composition controller is to drive the overall process to its new steady state in less time.
required to maintain high-purity acetone. Deadtimes of 3 min are used in the reactor exit temperature
Looking at the temperature profile for the second column C2 loop and in the composition loop. Deadtimes of 1 min are used
given in Figure 8B, we can see that there is a significant slope in the column temperature loops. Temperature and composition
 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011 1217

Figure 12. Disturbances in reactor temperature, absorber pressure, or feed composition.

loops are tuned using relay-feedback tests (as comprehensively specified value. Notice that the distillate-to-reflux ratio is
described by Yu10) and Tyreus-Luyben tuning rules. Table 8 increased as feed flow rate increases. This corresponds to a
gives controller tuning parameters for the three major temper- lower reflux ratio. Keep in mind that the stage 75 temperature
ature loops and for the composition loop. controller is maintaining temperature, so this column has
The initial control structure tested did not use the composition dual-end control.
controller. The first column had a single temperature controller The compositions of the bottoms and distillate products in
and a fixed reflux ratio. Figure 10 gives results for 20% increases
the second column depart somewhat from their specifications,
(solid lines) and 20% decreases (dashed lines) in the fresh feed
but as previously discussed, tight control of these streams is
flow rate. Stable regulator control is achieved for these large
not vital.
disturbances.
However, as the third graph on the left shows, the impurity An alternative to using dual-end control with a composition
of water in the acetone distillate product from the first column analyzer is to over purify the acetone product so that it remains
(xD1(w)) deviates significantly from the specified 0.1 mol % (on above the purity specifications despite disturbances. This would
a relative basis). increase energy consumption in the first distillation column. For
Figure 11 gives results for the same disturbances with the example, changing the water impurity level to 0.05 mol % would
composition controller in service. The composition controller require an increase in the reboiler duty in the first column of
adjusts the reflux ratio to drive distillate impurity to the 5.7%.
1218 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011

Figure 12 gives results for other disturbances using the dual- Effective dynamic control of the multiunit process is achieved
end structure. The solid lines are for a change in reactor exit by the use of conventional controllers. To ensure close control
temperature from 750 to 740 K. Reactor conversion is lower, of the high-purity acetone product, a dual-end control structure
so there is an increase in the recycle stream (D2). More energy must be used. Large disturbances in throughput, feed composi-
is required in the reboiler of the second column. tion, reactor temperature, and absorber pressure are well handled.
The dashed lines in Figure 12 are for a change in the absorber
pressure from 17 to 15 atm. The second graph on the right shows Literature Cited
that the composition of acetone in the absorber off-gas yG
increases from 0.66 mol % up to almost 3 mol %. The vent (1) Douglas, J. M. Conceptual Design of Chemical Processes; McGraw-
stream decreases from 0.66 kmol/h to about 0.60 kmol/h. Hill: New York, 1988.
(2) Luyben, W. L. Design and control of an auto-refrigerated alkylation
The dotted lines in Figure 12 are for a change in the fresh process. Ind. Eng. Chem. Res. 2009, 48, 11081–11093.
feed composition from 67 to 60 mol % IPA with a corresponding (3) Luyben, W. L. Design and control of the monoisopropylamine
increase in water concentration. With less IPA fed, the produc- process. Ind. Eng. Chem. Res. 2009, 48, 10551–10563.
tion rate of acetone decreases (D1), less energy is needed in (4) Luyben, W. L. Design and control of the ethyl benzene process.
AIChE J., submitted for publication.
the reboiler of the first column (QR1), and a higher reflux ratio (5) Luyben, W. L. Design and control of the cumene process. Ind. Eng.
is required in the first column (lower D1/R1 ratio). Chem. Res. 2010, 49, 719–734.
The proposed dual-end control structure handles all these large (6) Turton, R.; Bailie, R. C.; Whiting, W. B.; Shaelwitz, J. A. Analysis,
disturbance quite well with the acetone product held close to Synthesis and Design of Chemical Processes, 2nd ed.; Prentice Hall: Upper
its specified purity. Saddle River, NJ, 2003.
(7) Howard, W. L. Acetone. In Kirk-Othmer Encyclopedia of Chemical
Technology; Wiley: New York, 2001; Vol. 1, p 160.
7. Conclusion (8) Luyben, W. L. Disillation Design and Control Using Aspen
Simulation; Wiley: New York, 2006.
The acetone process illustrates a number of interesting design (9) Luyben, W. L.; Tyreus, B. D.; Luyben, M. L. Plantwide Process
trade-offs. Effective removal of the very light hydrogen product Control; McGraw-Hill: New York, 1999.
while minimizing loss of the acetone product is the main feature (10) Yu, C. C. Autotuning of PID Controllers, 2nd ed.; Springer: New
of this process. There is an optimum absorber pressure that York, 2006.
balances the opposing effects of pressure on losses in the
absorber off-gas stream and losses in the distillation column ReceiVed for reView December 5, 2009
vent stream. High pressure also adversely affects reaction ReVised manuscript receiVed January 23, 2010
Accepted February 15, 2010
kinetics, so reactor temperature must be increased as pressure
is increased. IE901923A