Ind. Eng. Chem. Res.

2010, 49, 719–734 719

Design and Control of the Cumene Process

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

The chemistry of the cumene process features the desired reaction of benzene with propylene to form cumene
and the undesirable reaction of cumene with propylene to form p-diisopropylbenzene. Both reactions are
irreversible. Since the second has a higher activation energy than the first, low reactor temperatures improve
selectivity of cumene. However, low reactor temperatures result in low conversion of propylene for a given
reactor size or require a large reactor for a given conversion. In addition, selectivity can be improved by
using an excess of benzene to keep cumene and propylene concentrations low, but this increases separation
costs. Therefore, the process provides an interesting example of plantwide economic design optimization in
which there are many classical engineering trade-offs: reactor size versus temperature, selectivity versus recycle
Downloaded from pubs.acs.org by UNIV OF LOUISIANA AT LAFAYETTE on 01/19/19. For personal use only.

flow rate, and reactor size versus recycle flow rate. Design optimization variables affect both energy costs
and capital investment. They also affect the amount of reactants required to produce a specified amount of
cumene product. The economic effect of reactant consumption is very large, an order of magnitude greater
than the impact of energy or capital. The process is presented in the design book by Turton et al. (Analysis,
Synthesis and Design of Chemical Processes, 2nd ed.; Prentice Hall: Saddle River, NJ, 2003) and consists of
a cooled tubular reactor and two distillation columns. The liquid fresh feeds and the benzene recycle stream
are vaporized, preheated, and fed into the vapor-phase reactor, which is cooled by generating steam. Reactor
effluent is cooled and fed to the first column that produces a distillate stream of benzene that is recycled back
to the reactor. The second column separates the desired cumene product from the undesired p-diisopropyl-
benzene. 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
Ind. Eng. Chem. Res. 2010.49:719-734.

handling large disturbances in production rate.

1. Introduction the amount of undesirable byproduct that is produced, but

increasing either of these variables raises capital and/or energy
The synthesis of a chemical process involves science, art, costs.
innovation, intuition, inspiration, experience, and common sense.
Of more importance is the effect of these design optimization
Success also requires a lot of hard work and at times a little
variables on conversion and selectivity, which translates into
luck. The old saying that “There are many ways to skin a cat”
changes in the amounts of the reactants required to produce a
certainly applies to the development of a process flowsheet. We
fixed amount of cumene. For example, if a bigger reactor is
start with raw materials (fresh feed streams of reactants) and
used, reactor temperature can be lower, which improves
want to produce the desired product(s) while, at the same time,
selectivity while achieving the same conversion and also reduces
achieving a low capital investment, low operating costs, and
the amount of raw materials. If the reactor is inexpensive (cheap
safe nonpolluting operation. The design engineer needs to
catalyst), the optimum design is a large reactor. However, if
consider a multitude of issues and objectives that are often
the reactor is expensive, the optimum design is a small reactor.
conflicting and frequently qualitative in nature.
As Douglas2 pointed out (Douglas Doctrine) two decades ago,
Engineering trade-offs are important parts of chemical process the costs of raw materials and products are usually much larger
synthesis. The classical example is the trade-off between costs than the costs of energy or capital in a typical chemical process.
in the reaction section of the plant versus costs in the separation Therefore the process must be designed (investing capital and
section of the plant. Bigger reactors are more expensive (vessel paying for energy) so as to not waste feed stocks or lose products
and catalyst capital investment) but can produce higher conver- (particularly in the form of undesirable products). Process
sion that requires a less expensive separation section (smaller economics dictate that the conversion of reactant must be quite
recycle flow rates that require smaller diameter distillation high. These principles are clearly illustrated in the cumene
columns with smaller reboilers and condensers). Discussions
process in several ways.
of these issues and examples of their application to typical
chemical processes are presented in most chemical engineering (1) The fresh propylene feed steam contains some propane
design textbooks. Turton et al.1 provide a number of useful impurity, which is inert in the reactor and must have a place to
examples, one of which is the cumene process studied in this get out of the process. Since the separation of propylene and
paper. The authors do not consider optimization or control of propane is difficult, the economics strongly favor designing the
this process. reactor for a very high conversion of propylene. The propane
and any unreacted propylene are flashed off and burned, so they
The purpose of this paper is to use the cumene process to
only have fuel value. High propylene conversion can be
illustrate some interesting design optimization features and
achieved by either running at high temperatures or increasing
principles. The two dominant design optimization variables are
the size of the reactor. The former increases the production of
reactor size and benzene recycle. Increasing either one reduces
undesirable byproduct. The latter increases capital cost.
* E-mail: [email protected]. Tel.: 610-758-4256. Fax: 610-758- (2) The undesirable byproduct is also burned so it only has
5057. value as fuel. Since it takes reactants to produce this product
10.1021/ie9011535  2010 American Chemical Society
Published on Web 11/23/2009
720 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010

Table 1. Reaction Kinetics kinetics of the reactions are given. This latter information is
R1 R2 quite useful since it is often difficult to find kinetic information
in the literature. The authors provide a good description of the
K 2.8 × 107 2.32 × 109
E (kJ/kmol) 104174 146742
basic units in the process with equipment sizes and operating
concentration terms (kmol/m3) CPCB CCCP conditions. Their intent is not to determine the optimum design
but to illustrate a typical multiunit chemical process that features
instead of producing cumene, there is a strong incentive to keep a reaction section coupled with a separation section and
its production rate small. connected by a recycle stream.
The economic analysis presented in the paper will show that 2.1. Reaction Kinetics. The production of cumene (isopro-
raw material fresh feed costs are an order of magnitude larger pylbenzene) involves the reaction of benzene with propylene
that energy or annual capital costs. in a high-temperature, high-pressure gas-phase reactor.

2. Process Studied C6H6 + C3H6 f C9H12 (1)

The cumene process is described in Turton et al.1 in a fair There is also a sequential reaction of cumeme and propylene
amount of detail. The basic flowsheet is presented, and the to form p-diisopropylbenzene (PDIB).

Figure 1. Txy diagram: (a) for benzene/cumene; (b) for cumene/PDIB.

 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010 721
C9H12 + C3H6 f C12H18 (2) were converted and none were lost, the fresh feed flow rate of
1 the pure benzene would have to satisfy the stoichiometry of
Table 1 gives the reaction kinetics provided by Turton et al.
All reaction rates have units of kmol s-1 m-3. Concentration the reaction: (110)(0.95) ) 94.5 kmol/h. But some of the
units are molarity. The reactions occur in the vapor phase in propylene is consumed in the second reaction, and some is lost.
the presence of a solid catalyst (assumed to have 0.5 void The result is a benzene fresh feed flow rate of 104.2 kmol/h,
fraction and a 2000 kg/m3 solid density). The reactor is run at which shows that over 10 kmol/h of benzene is being undesir-
high pressure (25 bar) since the moles of reactants are more ably consumed or lost. The production rate of cumene is 92.86
than the moles of product (LeChatier’s principle). kmol/h in the Turton et al. design.
Notice that the activation energy of the undesirable reaction The liquid fresh feeds are combined with a liquid recycle
is larger than that of the desirable reaction. Therefore low reactor stream (benzene recycle D1, the distillate from the first column
temperatures improve selectivity. In addition selectivity is C1) and fed into a vaporizer. The total benzene fed to the reactor
improved by keeping the concentration of cumene and propylene (sum of fresh plus recycle) is 207 kmol/h. The saturated gas
low in the reactor. This can be achieved by using a large excess leaves the vaporizer at 209 °C and 25 bar. It is preheated in
of benzene, but the excess must be recovered and recycled. two heat exchangers. The first recovers heat from the hot reactor
2.2. Phase Equilibrium. Assuming that a separation between effluent at 427 °C. The second adds additional heat to bring
propylene and propane is uneconomical, the only separations the reactor inlet temperature up to 360 °C.
required are fairly easily achieved by distillation. The normal 2.3.1. Reaction Section. The reactor is a cooled tubular
boiling points of benzene, cumene, and PDIB are 80.1, 152.4, reactor that generates high-pressure steam from the exothermic
and 209.8 °C, respectively. The NRTL physical property reactions. There are 342 tubes, 0.0763 m in diameter, and 6 m
package is used in the Aspen simulations used in this paper. in length. They are filled with a solid catalyst with a void fraction
Figure 1 gives the Txy diagrams for the benzene/cumene system of 0.5 and a solid density of 2000 kg/m3. The temperature on
and for the cumene/PDIB system at atmospheric pressure. The the steam side of the reactor is 360 °C. An overall heat-transfer
curves are fairly fat, so the required number of trays and the coefficient of 0.065 kW m-2 K-1 is used. We assume that the
reflux ratios in the distillation columns are fairly low. reactor inlet temperature is set to the same value as the coolant
2.3. Flowsheet. Figure 2 shows the flowsheet of the process (steam) temperature in the reactor.
with the equipment sizes and conditions used by Turton et al.1
The fresh feed streams of benzene and mixed C3 (propylene Reactor effluent leaves at 427 °C, is cooled to 322 °C in the
and propane) enter the process as liquids. The fresh feed flow feed-effluent heat exchanger (FEHE), and sent to a condenser
rate of the C3 feed is set at 110 kmol/h. The composition of in which it is cooled to 90 °C using cooling water. The two-
this feed is 95 mol % propylene and 5 mol % propane. Since phase stream from the heat exchanger is fed to a flash tank.
propane does not react, the (110)(0.05) ) 5.5 kmol/h of propane The gas from the tank is used as fuel. The liquid is fed into the
must leave the process somewhere. It is vented in a gas stream first distillation column C1.
from the flash drum. It is important to note that this gas stream in the Turton design
The fresh feed of benzene is 104.2 kmol/h. Note that if there is very large (21.77 kmol/h), and it contains a large concentration
were only the cumene reaction (eq 1) and if all the propylene of unreacted propylene (31.9 mol % propylene). Clearly this

Figure 2. Turton design flowsheet.

722 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010

Figure 3. Conversion 99%.

operation would be quite uneconomical because of the exces- reflux ratio is low (RR ) 0.63). Energy consumption is 1.2 Gcal/
sively large losses of the raw material propylene. h, and the column diameter is 1.26 m.
As we will show in the optimization studies in the next The design specification is to attain high-purity cumene in
section, the propylene conversion must be drastically increased. the distillate and minimize the loss of cumene in the bottoms.
There are 8.9 kmol/h of propylene in the reactor effluent in the The bottoms composition is set at 0.1 mol % cumene. The
Turton design, which has a reactor inlet temperature and a steam distillate composition is 99.9 mol % cumene using the 0.63
temperature of 360 °C. It is a simple matter to increase the reflux ratio.
propylene conversion by increasing these two temperatures. Notice that the bottoms B2 flow rate is 1.952 kmol/h in this
However, the higher temperature will increase the production flowsheet. This represents a loss of reactants, and the stream
of the undesirable product. So it may be more economical to only has fuel value. The production of the undesired product
increase the size of the reactor (use more tubes), which would can be reduced by lowering reactor temperatures (which would
give a higher propylene conversion but at a lower temperature. require a larger reactor) or increasing benzene recycle (which
This is one of the important design trade-offs in the cumene would require a larger distillation column C1 and increase
process. energy consumption in the vaporizer and in the C1 reboiler).
2.3.2. Benzene Recycle Column C1. The column has 15
stages and is fed on stage 6, which is the optimum feed stage 3. Economic Optimization
to minimize reboiler heat input. The operating pressure is 1.75 The design discussed in the previous section has not been
bar, which gives a reflux-drum temperature of 60 °C, so cooling optimized from the standpoint of economics. The low conversion
water can be used in the condenser. The reflux ratio is small of propylene in the reactor and the subsequent loss of propylene
(RR ) 0.44). The distillate is mostly benzene and is recycled in the gas stream is an obvious issue that must be considered.
back to the reactor. Its composition is 95.6 mol % benzene with
Of the 104.5 kmol/h of propylene in the fresh feed, the
small amounts of propylene and propane that are in the liquid
amount lost in the gas stream is 6.94 kmol/h. So reactor
from the flash drum. propylene conversion is only 93%. In addition, the production
The design specification is to keep benzene from dropping of 1.952 kmol/h of the undesirable product consumes 3.904
out of the bottom and affecting the purity of the cumene product kmol/h of propylene. So the conversion of propylene to the
leaving in the distillate of the downstream column. Since the desired product cumene is only 90%.
specified cumene purity is 99.9 mol %, a very small benzene 3.1. Increasing Propylene Conversion. As previously men-
composition in the bottoms (0.05 mol %) is required. The tioned, propylene conversion in the reactor can be easily
column diameter is 1.36 m, and the reboiler heat input is 1.64 increased by increasing reactor inlet temperature, Tin., and steam
Gcal/h. Both of these increase if more benzene recycle is used. temperature, TS. However, selectivity will be adversely affected.
2.3.3. Cumene Product Column C2. The column has 20 Figure 3 gives a flowsheet in which the reactor inlet
stages and is fed on stage 12. The operating pressure is 1 bar temperature and coolant temperature have been increased from
psia, which gives a reflux-drum temperature of 152 °C. The 360 to 374 °C. Notice that the production rate of cumene is
 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010 723

Figure 4. Effect of benzene recycle.

kept the same as the original case (D2 ) 92.86 kmol/h). The of gas increases. Notice that the reactor temperatures must be
gas stream is greatly reduced from 21.77 to 10.10 kmol/h. The increased as more material flows through the reactor to maintain
concentration of propylene in the gas stream drops from 31.9 the 99% propylene conversion because of the smaller residence
to 5.4 mol %. Therefore much less fresh propylene is fed into time. Despite the higher temperatures, selectivity improves
the system; fresh benzene feed drops from 104.2 to 99.81 kmol/ because of the higher benzene concentrations in the reactor.
h. This represents a very large savings in raw material costs, as Figure 5 shows how changes in reactor size impact the
we will quantify in the next section. performance of the system. As the number of reactor tubes is
Notice, however, that the higher reactor temperatures also increased, more cumene is produced and less undesirable
result in a higher production of the undesirable product. The product must be burned. There is also a small decrease in gas
flow rate of the bottoms B2 from column C2 increases from flow rate. The important effect of making the reactor larger is
1.952 to 2.588 kmol/h, so the loss of both propylene and benzene that lower reactor temperatures can be used to achieve the same
that this stream represents increases. 99% propylene conversion, as shown in the bottom right graph
How can we keep propylene conversion high but reduce the of Figure 5. Comparing the results in Figure 4 with those in
production of the undesired product? There are two ways to Figure 5 shows that the lower reactor temperatures have the
achieve these objectives. The temperatures in the reactor could beneficial effects of making more cumene, less undesirable
be lowered if a bigger reactor were used. Or the recycle of product, and less gas.
benzene could be increased. Both of these involve higher capital These preliminary scoping studies indicate that reactor size
and/or energy costs. In the following section we quantify the has a stronger impact on performance that benzene recycle. Of
economic effects of these two design optimization variables course, increasing either of these variables means an increase
while maintaining a high propylene conversion. in capital and/or energy costs. To quantify the economics of
3.2. Effects of Design Optimization Variables. To get some these design optimization variables, we must quantitatively
feel for the effects of reactor size and benzene recycle, a series estimate the capital cost of equipment (heat exchangers, reactor
of runs were made in which one of the two design optimization vessel, catalyst, and distillation columns), the operating cost of
variables was changed over a range of values while holding energy (vaporizer, preheater, and reboilers), and the cost of
the other variable constant. In these runs the fresh feed of the purchasing the two raw materials (propylene and benzene).
propylene/propane stream was held constant at the original value 3.3. Economic Basis. The sizing and cost estimation for the
of 110 kmol/h in the Turton et al. design. Note that the equipment is taken from Luyben3 and is presented in Table 2.
production rate of cumene D2 is not held constant in this series The reactor capital cost is based on its heat exchange area and
of runs. the amount of catalyst it contains. The very important effect of
Figure 4 shows how changes in the benzene recycle affect a the price of catalyst is explored later in the paper. As we would
number of important variables. The total benzene feed to the expect, expensive catalyst favors a small reactor design while
reactor is the sum of the fresh benzene feed and the recycle inexpensive catalyst leads to a large reactor with many tubes.
benzene, which is the distillate D1 of the first distillation column. High-pressure steam (254 °C with a cost of $9.83/GJ) is used
The top left graph shows that the production of the undesirable in the vaporizer and in both reboilers because of their high
product (PDIB in B2) decreases as more benzene is recycled. temperatures. The very high reactor inlet temperatures (360 °C)
The middle left graph shows that the production of the desired require a high-temperature heating medium or electricity, so a
product (cumene in D2) increases. These very beneficial effects cost of $16.8/GJ is used. Steam generated in the reactor is
are offset by the required increase in the energy consumption assumed to be worth $6.67/GJ.
in the column C1 reboiler (QR1) and in the vaporizer (Qvap), The cost of raw materials is based on data from the Internet
as shown in the bottom two graphs. In addition, the production Web site icis.com that was provided by Miles Julian. Propylene
724 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010

Table 2. Basis of Economics and Equipment Sizing

column diameter: Aspen tray sizing reactor (cooled)
column length: NT trays with 2 ft spacing plus 20% extra length heat-transfer coefficient ) 0.0 65 kW/(K · m2)
column vessel (diameter and length in meters) differential temperature ) TR(z) - TS
capital cost ) 17640(D)1.066(L)0.802 capital cost ) 7296(area)0.65
condensers (area in m2)
heat-transfer coefficient ) 0.852 kW/(K · m2) energy cost
differential temperature ) 13.9 K HP steam ) $9.83/GJ
capital cost ) 7296(area)0.65
reboilers, FEHE and preheater (area in m2): capital cost
TAC ) + energy cost
heat-transfer coefficient ) 0.568 kW/(K · m2) payback period
differential temperature ) 34.8 K
capital cost ) 7296(area)0.65 payback period ) 3 years

Table 3. Economic Effects of Design Optimization Variables (Cumene Production, D2 ) 92.85 kmol/h)
Turton 99% increase increase economic
design conversion benzene recycle reactor size optimum
reactor tubes 342 342 342 1000 1500
total benzene recycle kmol/h 207 207 250 207 207
Tin ) TS °C 360 374 382 363 358
Tout °C 427 419 428 364 358.5
gas kmol/h 21.77 10.10 11.12 10.01 9.979
B2 kmol/h 1.952 2.588 1.898 1.85 1.550
FFC3 kmol/h 110.0 104.1 102.6 102.55 101.93
FFB kmol/h 104.2 99.81 100.24 99.07 98.78
C1 capital 106 $ 0.407 0.414 0.455 0.389 0.407
C1 energy 106 $/year 0.591 0.609 0.767 0.609 0.607
C2 capital 106 $ 0.338 0.338 0.339 0.338 0.338
C2 energy 106 $/year 0.429 0.429 0.429 0.429 0.429
HX1 capital 106 $ 0.0201 0.0261 0.0322 0.0212 0.0192
HX1 energy 106 $/year 0.0500 0.0744 0.103 0.0542 0.0464
HX2 capital 106 $ 0.0700 0.0690 0.0691 0.0609 0.0601
vaporizer capital 106 $ 0.222 0.219 0.241 0.219 0.219
vaporizer energy 106 $/year 1.18 1.15 1.33 1.16 1.15
FEHE capital 106 $ 0.232 0.242 0.245 0.369 0.354
reactor vessel capital 106 $ 0.410 0.410 0.410 0.823 1.07
catalyst at $10/kg 106 $ 0.374 0.374 0.374 1.09 1.64
reactor steam credit 106 $/year 0.359 0.350 0.401 0.545 0.401
fuel credit B2 106 $/year 0.624 0.828 0.607 0.592 0.713
fuel credit Gas 106 $/year 3.41 1.61 1.87 1.60 1.59
feed C3 cost 106 $/year 33.051 31.274 30.828 30.813 30.627
feed benzene cost 106 $/year 62.612 59.979 60.230 59.535 59.36
total capital 106 $ 2.07 2.09 2.16 3.31 4.11
total energy 106 $/year 2.24 2.26 2.63 2.24 2.23
TAC 106 $/year 94.204 91.427 91.539 90.961 90.883

is priced at $0.37/lb (34.3 $/kmol), and benzene is priced at There are two waste streams that are burned and carry a fuel
$2.90/gal (68.6 $/kmol). credit in the economic analysis. The bottoms stream from

Figure 5. Effect of reactor size.

 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010 725

Figure 6. Effect of reactor size for fixed production rate.

column C2 is essentially pure p-diisopropylbenzene. Its heat of sion has been increased by raising reactor temperatures. Pro-
combustion is estimated to be 1624 kcal/mol. The gas stream pylene conversion in the reactor is raised by specifying a much
contains propylene, propane, benzene, and cumene. The heats smaller molar flow rate of 0.9 kmol/h in the reactor effluent,
of combustion of these components are 460, 485, 747, and 1183 which is achieved by adjusting the reactor inlet and steam
kcal/mol, respectively. temperature (raised from 360 to 374 °C).
The values of these streams were estimated by calculating There is little change in total capital investment or energy,
the amount of air needed to completely combust each component and fuel credits are reduced. Gas flow rate is cut in half, but
to water and carbon dioxide, with 5% excess oxygen. The molar the production of the undesirable product increases somewhat
flow rates and heat capacities of each component were used to (B2 changes from 1.952 to 2.588 kmol/h because of the higher
find the sensible heat needed to increase the combustion gas reactor temperatures).
temperature from 25 to 285 °C (the assumed stack gas However, the flow rates of the raw materials are both reduced,
temperature). This sensible heat was subtracted from the heat which produce a very large reduction in the cost of raw
of combustion to find the net heating value of each component. materials. The fresh C3 feed is reduced from 110 to 104.1
Using $6/GJ value of energy, the calculated heating values of kmol/h and fresh benzene feed drops from 104.2 to only 99.81
propylene, propane, benzene, cumene, and p-diisopropylbenzene kmol/h (see also Figure 3). The raw material cost is cut from
used in the economic analysis are $10.4, $10.9, $27.9, $26.6, $95,660,000 per year to $91,250,000 per year! Remember that
and $36.5 per kmol, respectively. the production rate of cumene is the same in both of these
3.4. Economic Optimization Results. Table 3 gives detailed flowsheets.
information for a number of alternative designs. All of these These results illustrate very dramatically the Douglas Doctrine
cases produce the same amount of cumene (D2 ) 92.86 kmol/ that process designs must minimize the losses of raw material
h). But the equipment sizes and fresh feed flow rates are different and products.
in each case. 3.4.3. Higher Benzene Recycle. The third column in Table
Capital costs of equipment, energy costs, fuel credits, and 3 gives economic results for the case where benzene recycle
raw material costs are provided. The results in this table use a is increased from 207 to 250 kmol/h of total benzene fed to
catalyst price of $10/kg. The effect of catalyst price is explored the reactor. High propylene conversion is maintained by
later in this section. raising reactor temperatures to 382 °C. Higher temperatures
3.4.1. Turton Design. The first column in Table 3 gives are required because of the smaller residence time in the
economic results for the original Turton et al. design. Total reactor.
capital investment is $2,070,000. The net total energy cost Total capital and energy costs both increase due to a larger
is $2,240,000/year (energy consumption in columns, heat vaporizer and a larger recycle column C1. Production of the
exchangers, and vaporizer reduced by the steam generated undesirable product is reduced, despite the higher reactor
in the reactor and reduced by the fuel credits from burning temperatures, because of the higher benzene concentrations in
B2 and gas). the reactor. Gas flow rate is slightly higher.
The dominating factor is the cost of raw materials. The The fresh feed flow rates change in different directions. Less
fresh C3 feed at 110 kmol/h and fresh benzene feed at 104.2 propylene fresh feed is required, but more benzene fresh feed
kmol/h cost a total of $95,660,000/year! There are fairly large is needed. The net result is a slight reduction of the raw material
fuel credits from burning gas and B2, but these are swamped costs of about $200,000/year. The TAC that combines all these
by the cost of the purchased reactants. The low propylene factors increases from $91,427,000 per year in the previous case
conversion gives a molar flow rate of 9 kmol/h of propylene to $91,539,000 in this higher benzene recycle case.
in the reactor effluent, and this is lost in the gas steam. So increasing benzene recycle does not improve the econom-
3.4.2. High Conversion Design. The second column in Table ics of the process because more capital and more energy are
3 gives economic results for the case where propylene conver- required.
726 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010

Figure 7. Effect of reactor size and catalyst cost on TAC for fixed production rate.

Figure 8. Optimum design for $10/(kg of catalyst) (1500 tubes and 207 kmol/h total benzene).

3.4.4. Larger Reactor. The fourth column in Table 3 gives Table 3 shows that reactor temperatures can be reduced to
economic results for the case in which reactor size is 363 °C with this larger reactor, compared to the second column
increased. The number of reactor tubes is increased from 342 where 374 °C is required with the base-case number of tubes.
to 1000. Total benzene recycle is set at the base-case 207 Gas production is about the same, but the production of
kmol/h. This larger reactor will be more expensive in terms undesirable product is reduced to 1.85 kmol/h. Energy cost is
of vessel cost (more heat-transfer area) and catalyst cost. about the same, but capital cost increases due to the larger
However, a larger reactor should permit lower reactor reactor.
temperatures, which should improve selectivity. Propylene However, both fresh feed flow rates are reduced, which cuts
conversion is kept at 99%. raw material costs to $90,348,000/year. The economic results
 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010 727

Figure 9. Reactor profiles: (a) temperature; (b) composition.

show a TAC of $90,961,000/year, which is better than the conversion is 99%, total benzene recycle is 207 kmol/h, and
smaller reactor case. catalyst price is $10 per kg in this figure.
So increasing the number of tubes from 342 to 1000 improves There is a minimum in the TAC curve at 1500 reactor tubes
the economics. What happens for larger reactors? Figure 6 gives where the increasing cost of capital is balanced by the reduction
results for a range of the number of reactor tubes for a constant in raw material cost. Energy costs change very little with reactor
production rate of cumene. These results are different from those size. Energy consumption depends primarily on benzene recycle.
shown in Figure 5 in which the production rate of cumene is 3.4.5. Effect of Catalyst Price. All of the results given in
not fixed, but the flow rate of fresh propylene is fixed. Propylene Table 3 and Figure 6 are based on a catalyst price of $10/kg.
728 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010

Figure 10. Profiles in column C1: (a) temperature; (b) composition.

Since catalyst price has a major impact on capital cost, it will 4. Plantwide Control
affect the optimum economic design.
The steady-state Aspen Plus model shown in Figure 4 does
Figure 7 shows the effect of catalyst price on TAC over a
not include pumps and control valves. The first step in studying
range of reactor sizes. As catalyst price increases, the optimum dynamics using a pressure-driven dynamic simulation is to add
number of reactor tubes decreases. But even with very expensive these items and to determine the volumes of all vessels. The
catalyst, the optimum number of reactor tubes is substantially units with dynamics include the vaporizer, heat exchangers, flash
larger than the original Turton design. drum, and distillation columns.
Figure 8 gives flowsheet conditions for the optimum 1500 Reflux drums, column bases, the vaporizer, and the flash drum
tube reactor case assuming $10/(kg of catalyst) cost. The last are sized to provide 5 min of holdup when at 50% level. The
column in Table 3 gives details of the economics and conditions tube side and shell side volumes of the FEHE are calculated
for this case. using the heat-transfer area of 460 m2 and assuming 0.0762 m
Figure 9 gives the temperature and composition profiles in diameter tubes, 6 m in length. The tube volume is 8.76 m3 in
the reactor for the optimum case. Figures 10 and 11 give 320 tubes. Shell volume is assumed equal to tube volume. The
temperature and composition profiles for columns C1 and C2. volume of HX2 is 0.5 m3 on the basis of its heat-transfer area
 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010 729

Figure 11. Profiles in column C2: (a) temperature; (b) composition.

of 25.6 m3. The volume of the vaporizer is 4.59 m3, and the (4) Reactor inlet temperature is controlled by heat input to
volume of the flash drum is 4.15 m3. HX1.
Figure 12 shows the plantwide control structure developed (5) Reactor exit temperature is controlled by manipulating
for this process using process control methodology.3,4 Conven- the pressure (temperature) of the steam being generated on the
tional PI controllers are used in all loops. The various loops shell side of the tubular reactor.
are listed below with their controlled and manipulated variables. (6) The back-pressure on the reactor system is controlled by
(1) Fresh C3 feed is flow controlled. This is the throughput manipulating the control valve located after the FEHE. This
handle. valve drops the pressure from 23 bar in the reactor to 2 bar in
(2) Total benzene (fresh feed plus benzene recycle D1 from the flash tank.
column C1) is ratioed to the C3 flow rate. The fresh feed of (7) Temperature leaving the condenser HX2 is controlled by
benzene is manipulated to achieve the desired flow rate of total heat removal.
benzene. (8) The level in the flash tank is controlled by manipulating
(3) Vaporizer level is controlled by vaporizer heat input. There the flow rate of liquid leaving the tank and going to column
is no liquid outlet stream. C1.
730 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010

Figure 12. Plantwide control structure.

Table 4. Temperature Controller Parameters (11) Base levels in all columns are controlled by bottoms
reactor exit, recycle column product column flow rates.
TCR C1 stage 11, TC1 C2 stage 16, TC2 (12) Reflux drum levels in all columns are controlled by
distillate flow rates.
SP 358.5 °C 162.2 °C 200.6 °C
transmitter range 300-400 °C 100-200 °C 100-200 °F
(13) Reflux flow rates are ratioed to the column feed.
OP 358 °C 3.37 Gcal/h 3.37 Gcal/h (14) The temperature on stage 11 in column C1 is controlled
OP range 300-400 °C 0-1.68 Gcal/h 0-1.18 Gcal/h by manipulating the reboiler heat input. This location is selected
dead time 1 min 1 min 1 min by looking at the temperature profile shown in Figure 10a.
KC 2.4 0.66 0.51 (15) The temperature on stage 16 in column C2 is controlled
τI 11 min 11 min 23 min by manipulating the reboiler heat input. This location is selected
by looking at the temperature profile shown in Figure 11a.
(9) The pressure in the flash tank is controlled by manipulat- Item 2 above is one of the key plantwide control features of
ing the gas stream. this structure. The total flow rate of benzene is controlled.
(10) Pressures in all columns are controlled by condenser Note that the flow rates of the fresh benzene and the recycle
heat removal. benzene are about the same in magnitude (98.78 and 108.2

Figure 13. Controller face plates.

 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010 731

Figure 14. 20% changes in C3 feed flow rate.

kmol/h) in this flowsheet. The flow rate of the fresh feed is being used to control the reflux-drum level in this column. In
used to control the total with the distillate from column C1 another process in which the flow rate of the fresh feed is much
732 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010

Figure 15. Propylene feed composition disturbances.

smaller than the recycle, this structure would not be effective. the flow rate of the recycle, and the liquid level in the reflux
In that situation, the total should be controlled by manipulating drum should be controlled by manipulating the fresh feed stream.
 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010 733

Figure 16. Reactor exit temperature disturbances.

It is important to realize that changing the fresh feed flow rate total flow, so the total flow controller adjusts the distillate flow
has an immediate effect on the reflux-drum level even though to hold the total flow constant. Therefore the reflux-drum level
it is not introduced into the drum. Changing fresh feed affects is immediately affected.
734 Ind. Eng. Chem. Res., Vol. 49, No. 2, 2010

All liquid level controllers are proportional only using Kc ) 358 to 350 °C (dashed lines). As expected, increasing reactor
2. Reactor exit temperature loop and column temperature loops temperature produces more undesirable product B2. There is
contain 1 min dead times and are tuned using relay-feedback little change in the cumene production D2 for an increase in
tests and Tyreus-Luyben tuning rules. Table 4 gives controller reactor temperature, but there is a significant decrease for a
tuning parameters for various cases. Figure 13 shows the decrease in reactor temperature (see Figure 16c). This occurs
controller face plates from the Aspen Dynamics simulation. Note because the lower conversion of propylene in the reactor
that the flow controller on the total benzene is on cascade, produces a large increase in the gas (see Figure 16a).
receiving its set point from the multiplier that gets its input from
the flow transmitter on the fresh C3 feed. 5. Conclusion
The effectiveness of this structure is demonstrated in Figures
14, 15, and 16 for several large disturbances. In Figure 14 the The cumene process exhibits some interesting design and
set point of the C3 fresh feed flow controller is changed at time control features. It provides a classical example of the engineer-
) 0.2 h. The solid lines are responses for a 20% increase, and ing trade-off between reactor size and recycle flow rate. Design
the dashed lines are for a 20% decrease. Stable regulatory control optimization variables affect capital and energy costs. But they
is achieved with the product quality of the cumene product also affect conversion and selectivity, which have a very large
xD2(cucmene) being maintained above the desired 99.9 mol % impact on raw material costs. The economics of the cumene
specification (upper right graph in Figure 14c). An increase in process illustrate that reducing reactant losses and the production
fresh feed results in the production of more gas and more of undesirable products can produce savings that are much larger
undesirable product B2. The temperature of the steam TS in than capital or energy costs.
the reactor must be reduced to hold the same exit temperature. Effective dynamic control of the multiunit process is achieved
Column temperatures are well-controlled. There is a fairly large by the use of conventional controllers and a control structure
dynamic transient in the composition of the undesirable product that fixes the total benzene feed to the reactor (fresh feed plus
xB2(PDIB). This could be greatly reduced by using a steam-to- recycle). Large disturbances in throughput and feed composi-
feed ratio in column C2 with the ratio reset by the column tions are handled with products held close to their specified
temperature controller. values.
Figure 15 gives responses for disturbances in the composition
Literature Cited
of the fresh C3 feed. The solid lines are when more propane is
fed. The feed composition is changed from 95 to 93 mol % (1) Turton, R., Bailie, R. C., Whiting, W. B., Shaelwitz, J. A. Analysis,
propylene with a corresponding increase in propane from 5 to Synthesis and Design of Chemical Processes, 2nd ed.; Prentice Hall: Upper
7 mol % propane. The dashed lines are when less propane is Saddle River, NJ, 2003.
(2) Douglas, J. M. Conceptual Design of Chemical Processes; McGraw-
fed (5 mol % changed to 3 mol % propane). The control Hill: New York, 1988.
structure handles this disturbance well. When more propane (3) Luyben, W. L. Disillation Design and Control Using Aspen
enters the system, the flow rate of fresh benzene is gradually Simulation; Wiley: New York, 2006.
reduced. More gas is removed, and less cumene is produced. (4) Luyben, W. L., Tyreus, B. D., Luyben, M. L. Plantwide Process
Control; McGraw-Hill: New York, 1999.
Benzene recycle increases, which results in a small decrease in
the production of the undesirable product. ReceiVed for reView July 18, 2009
The final disturbance is a change in the set point of the reactor ReVised manuscript receiVed October 3, 2009
Accepted November 2, 2009
exit temperature controller. Figure 16 gives responses for a
change from 358 to 365 °C (solid lines) and for a change from IE9011535