Reactions and Separations

22 www.aiche.org/CEP September 2006 CEP

M
ost strategic product- and process-development
decisions are made very early in the life of a
project detailed data on reaction kinetics,
phase behavior, chemical properties, etc., are incomplete
or unknown, and investment decisions must be made
based on mostly qualitative and some quantitative infor-
mation obtained during the discovery phase. In addition,
there is increasing pressure to reduce the development
time for new products and processes.
Process development is usually performed in an evolu-
tionary way, in which the process flowsheet is developed
based on experience and proven best practice rather than on
a systematic framework for generating process alternatives.
Although this approach yields useful designs quickly and
reliably, it does not promote the use of novel technologies,
nor does it allow for the comparison of alternative designs.
The main danger in the traditional approach is jumping to a
solution before the problem is really understood.
This article outlines how possible process options are
identified using mostly qualitative information, some key
quantitative information, and teamwork by discovery
chemists and conceptual design engineers. Identification of
the options can aid investment decisions, and can also help
steer experimental and process development programs.
Process development can be viewed as consisting of
three phases:
1. discovery chemistry and preliminary process devel-
opment (the first 46 weeks)
2. process chemistry and conceptual design (the next
36 months)
3. detailed mechanical and equipment design (the next
612 months).
Phase 1: Discovery chemistry
and preliminary process development
Phase 1 is the most critical phase in the process devel-
opment for value-added chemicals. Many chemical routes,
which may require different raw materials, are pursued.
Exploratory experiments are performed to verify whether
the reactions in a proposed route produce the desired inter-
mediates and products. Preliminary experimental data
include qualitative information on the reaction rates and
quantitative data on conversion, yield and selectivity.
Decisions must be made to pick a chemical route so
that further work can be concentrated on that route.
Conventionally, the chemical route is chosen based on raw
materials costs. However, other important factors must be
taken into account before making the selection.
A preliminary flowsheet sketch can indicate the com-
plexity of the manufacturing process. The economic
advantage of cheaper raw materials may be diminished by
much more expensive manufacturing. After taking into
account factors such as the availability and cost of raw
materials (which depend on reaction selectivities), manu-
facturing process feasibility, and complexity, the most
promising chemical route is selected. If the raw materials
costs for a particular route are higher than the product
price, the development of that route is terminated.
The preliminary process scheme also identifies key
areas where further experimental efforts must be focused.
Potential waste treatment problems must be identified as
early as possible; therefore, the economics of waste treat-
ment must be a part of Phase 1 activities.
A few methodologies are available for the early stages
of process development. A method published by Sharratt
Use this structured approach, which
combines preliminary experimental data
with predictive methods and heuristics,
to quickly generate and screen
process alternatives at the early stages
of a new venture.
Sagar B. Gadewar
GRT, Inc.
Gerhard Schembecker
Univ. of Dortmund
Michael F. Doherty
Univ. of California, Santa Barbara
Rapid Process Design
CEP September 2006 www.aiche.org/CEP 23
et al. (1) uses qualitative knowledge of the reaction system
to determine alternatives in the early stages of process
development. It involves the identification of rate process-
es (such as reaction, mass transfer, etc.) and phases pres-
ent, and the use of driving forces to manipulate these rate
processes and phases to generate a process recipe.
This article extends the methodology developed by
Schembecker et al. (2) for generating and screening alterna-
tives based on qualitative and semi-quantitative data during
Phase 1 of process development. Outcomes of Phase 1 are
the chemical route(s), estimated raw materials costs, a pre-
liminary sketch of the manufacturing process scheme(s),
and the scope of a research program to determine the key
data and models needed for process design in Phase 2.
An important aspect of Phase 1 is the generation of
intellectual property. A critical ingredient for success in
Phase 1 is a team of discovery chemists and conceptual
design engineers working shoulder-to-shoulder under the
same leader in the same lab.
Phase 2: Process chemistry
and conceptual design
Phase 2 focuses on getting a better estimate of the
flowsheet and unit operations for the chosen chemical
route. Experiments are conducted to determine the effect
of parameters such as reactant molar ratios, catalyst com-
position, temperature and pressure on selectivities and
conversions. Experiments are also performed to determine
the physical properties (identified in Phase 1) required for
conceptual design. Phase 2 aims to determine equipment
sizes, utilities usage, raw materials consumption, and
investment and operating costs.
The information generated in this phase must be suffi-
cient for a reliable economic analysis that gives the desired
internal rate of return (IRR) for the investment. Potential
flowsheet alternatives are analyzed by varying a few key
variables (e.g., reaction conversion, reactant molar feed ratio,
etc.) to study their effect on IRR. The experiments in this
phase continue until there is no big improvement in the IRR.
If the IRR for the process does not
make the venture profitable, the
development work is terminated.
Potentially profitable flowsheet
alternatives are screened to deter-
mine the most economically attrac-
tive operable process scheme.
Simulations are performed to verify
the predictions of the conceptual
design exercise. Optimization is per-
formed for the key process variables.
The outcome of Phase 2 is a
process flowsheet with estimated equipment sizes, process
operating conditions, and flowrates for all the streams.
New intellectual property is generated to protect the final
manufacturing scheme and all economically viable alter-
natives that have been identified, as well as new catalysts,
compositions of matter, operating conditions, etc.
As in Phase 1, a critical ingredient for success is the close
collaboration between chemists and conceptual design engi-
neers. The engineers should deliver the stopping criterion
(i.e., no further significant change in the IRR) for ending a
line of research, as well as define the incentive for opening
new lines of research or continuing existing ones.
Phase 3: Detailed mechanical
and equipment design
Once a project is found to be profitable in Phase 2, all
the flows in the flowsheet are fixed. Materials of construc-
tion for important process equipment are identified.
Physical and chemical stability of the catalytic materials
are then determined.
Detailed mechanical and equipment design is per-
formed to arrive at reasonably accurate equipment sizes,
and more simulations are performed at this stage. A reli-
able estimate of the capital and operating costs is deter-
mined based on vendor quotes.
The outcome of Phase 3 is a technology package suit-
able for handing off to a construction company for pilot
plant implementation.
Rapid design in Phase 1
Figure 1 shows some of the tools used during each phase
of process development. Unlike Phase 1, systematic meth-
ods for Phases 2 and 3 are more mature (3). The remainder
of this article introduces a new methodology (incorporating
existing techniques), and demonstrates its application for
rapid design in Phase 1 of process development.
This methodology generates reactor and separation alter-
natives using systematic methods that rely on qualitative
and semi-quantitative information. The alternatives are then
Phase 3
Materials of Construction
Catalyst Stability
ASPEN PLUS
ICARUS
PIPEPHASE
VISUAL FLOW
Lab-Scale Reactors
Catalyst Screening
High-Throughput
Experimentation
FTIR, SEM, TEM
GC, HPLC
MATERIAL VISUALIZER
PROSYN
Mole Balances
Phase 1
Experimental Tools
ASPEN PLUS
DISTIL
HEXTRAN
ICARUS
MAYFLOWER
PRO II
PROSYN
Phase 1
Phase 2
I Figure 1. Many tools are available for process development.
screened to select economically
attractive chemical routes and
process schemes. The genera-
tion of alternatives also identi-
fies areas where more experi-
mentation is required for
selecting and designing unit
operations (e.g., determining
phase equilibrium to design a
liquid-liquid extraction unit).
The reactor type, its charac-
teristics and the reaction condi-
tions are important issues that
must be considered at an early
stage of process synthesis, since
they affect not only the per-
formance of the reactor but also
the separation section that fol-
lows. Furthermore, only basic
qualitative data from initial
experiments are available at the
early stages of process develop-
ment. Influences such as tem-
perature, catalyst, reactant con-
centration, operating mode,
conversion and type of reactor
can only be accomplished with
a large experimental effort and
more-detailed calculations that
are justified only after a project
has reached Phase 2.
Acomputer-aided method-
ology based on a tool called
PROSYN (PROcess
SYNthesis) employs databases,
numeric routines and heuristics
to aid flowsheet generation
using limited data (4). The procedure consists of first identi-
fying alternative reactor types and process conditions that
can achieve the desired goals for the reaction system.
Methods that support the tools for identifying reaction
and separation alternatives are very important. For example,
it is useful to generate correct mole balances to be able to
estimate the raw materials consumption of a process. This
task can be tricky for complex chemistries with many reac-
tions, and it is helpful to use algorithmic methods that can
reduce the effort involved in such calculations (5, 6).
Physical properties are critical in process synthesis.
Reference 7 is an excellent review of the state-of-the-art of
fluid-phase system thermodynamics. The desired thermo-
dynamic data are often not known. There is extensive lit-
erature on efforts to estimate missing properties from
known molecular structures, usually by dividing the mole-
cule into its constituent groups and estimating the contri-
bution of each group to a particular thermodynamic prop-
erty. The UNIFAC method is commonly used to estimate
vapor-liquid equilibrium (VLE) and liquid-liquid equilibri-
um (LLE) for multi-component mixtures of nonelec-
trolytes (8). Group contribution methods can also be used
to predict pure-component properties, including the boil-
ing point, critical pressure, enthalpy of formation, etc. (9).
Conceptual design tools normally associated with
Phase 2 can be used to evaluate the feasibility of prelimi-
nary separation schematics in Phase 1. The feasibility
tools in the MAYFLOWER and FORTUNE programs are
Table 1. Known information on the reaction system for each route.
Reaction T, C P, atm Phase Conversion Selectivity Attribute
24 www.aiche.org/CEP September 2006 CEP
Reactions and Separations
Measured Assumed or Estimated
Acetone Route
1 25 1 Liquid >95% >95% Reversible
2 80100 1 Liquid 70% 100% Reversible
3a 80100 1 Liquid 30% 100% Reversible
3b 120160 1 Liquid 100% 100% Reversible
4 80150 8 Liquid 90% 77% Reversible
Table 2. Binary interaction parameter matrix for the acetone route.
Acetone Ammonium
Acetone HCN Cyanohydrin H
2
SO
4
Methanol Water MMA Bisulfate
Acetone I L A D D D A
HCN I L A L D I A
Acetone
Cyanohydrin L L A I I I A
H
2
SO
4
A A A A L A A
Methanol D L I A D D A
Water D D I L D D A
MMA D I I A D D A
Ammonium
Bisulfate A A A A A A A
MMA = Methyl methacrylate
A = Approximated; D = Retrieved from a database; I = Treated as ideal; L = From the literature
t-Butanol Route
1 420 13 Gas 94% 94% Irreversible
2 300 23 Gas 89% 96% Irreversible
3 4060 1 Liquid 70% 99% Reversible
Ethylene Route
1 130170 2351 Gas-Liquid 90% 100% Reversible
2 130170 2351 Gas-Liquid 7585% 99% Reversible
3a 300 2 Gas 515% 99% Irreversible
3b 300 2 Gas 436% 99% Irreversible
4 4060 1 Liquid 70% 99% Reversible
CEP September 2006 www.aiche.org/CEP 25
useful for generating process schemes (10). Commercial
equivalents include the DISTIL and SPLIT programs
from Aspen Technology.
To use these tools, the physical properties of the system
must be described by thermodynamic models, such as the
NRTL equation, UNIFAC model, etc. During the discovery
stage, many properties, including binary interaction param-
eters for the VLE and LLE, are unknown. This rapid
design approach utilizes predictive methods and approxi-
mates the properties of components for which the predic-
tive methods cannot be used. The thermodynamic package
established in this way is not accurate, but in many cases it
is sufficient to generate sketches of the separation system.
Once a collection of preliminary process flowsheets
has been rapidly identified, they must be ranked to weed
out the inferior alternatives. It would be best to rank
them on the basis of IRR, but this is not possible in
Phase 1, since neither equipment sizes nor utility usages
are known. However, it is possible to rapidly estimate all
the material flowrates throughout the flowsheet (5, 6),
and this information provides a basis for ranking the
flowsheet alternatives (discussed later).
Example: Phase 1 MMA process development
Methyl methacrylate (MMA) is used to produce poly-
methyl methacrylate (e.g., Plexiglas), a hard transparent
plastic used in construction, furniture, business signs and
automobiles. MMA can be manufactured by various chem-
ical routes (11, 12), three of which are considered here.
The first task in generating and screening alternatives is
to compile all the reaction data generated by the chemists
and relevant data from the literature. These include con-
version, yield and selectivity data, together with the tem-
perature and pressure during reaction.
In this example, data for the reaction conditions and their
effects on the rates for each of the routes are obtained from
the literature. Kinetic models are not available for any of
the reactions. Qualitative information on the effect of reac-
tion parameters on the rates, and reaction conditions (e.g.,
the temperature range, pressure range, etc.), are known,
however, and are used to determine the operating conditions
for the reactions. The reaction data are given in Table 1.
Next, a property package to
predict the thermodynamic phase
behavior of the mixtures for each
route is prepared. The pure-com-
ponent properties are readily
available, but the binary interac-
tion parameters for phase equilib-
rium models are not available for
many component pairs. Some of
the properties can be obtained from commercially available
databases, while the remaining properties are either deter-
mined using predictive methods or approximated. Table 2 is a
matrix of the sources of binary interaction parameters for the
acetone route; similar tables are compiled for the other routes.
Reactor selection is based on the objective of highest
selectivity. Reactor analysis is performed using heuristics
and mathematical analyses together with the experimental
data, as well as rules described by Schembecker, et al. (2).
The analysis determines reaction parameters, including the
mixing patterns, temperature and pressure profile, degree
of conversion, etc., and considers the interdependency of
the reaction parameters as well.
Acetone route
The reaction of acetone with hydrogen cyanide (HCN)
to produce acetone cyanohydrin takes place in the liquid
phase at 25C and 1 atm; it is reversible with an equilibri-
um constant close to 14. There are no known side reac-
tions, and the rate is second-order with respect to HCN. A
plug-flow reactor (PFR) is selected for this reaction step
because it has higher selectivity than other reactors. The
separation of unreacted acetone and HCN can be carried
out using distillation.
Acetone cyanohydrin is reacted with sulfuric acid to pro-
duce the intermediate methacrylamide sulfate. Due to the
presence of water, the byproduct -hydroxyisobutyramide
is formed, which at higher temperatures can be converted to
methacrylamide sulfate. Reactor analysis suggests using a
PFR for producing methacrylamide sulfate from acetone
cyanohydrin at 90C, and also using a PFR for converting
the unwanted -hydroxyisobutyramide into methacrylamide
sulfate at 160C. Other byproducts of the reaction include
acetone, HCN, acetone sulfonate and formamide sulfate.
Sulfuric acid and methacrylamide sulfate are the heavies,
and are removed in the bottoms of a distillation column; the
distillate is fed to another column, which recovers acetone
cyanohydrin for recycle back to the reactor.
Methacrylamide sulfate is then reacted with methanol
at a pressure of 810 atm to form MMA and ammonium
bisulfate. Dimethyl ether (DME) is normally formed as a
byproduct; a series of continuous stirred-tank reactors
H
3
CCCH
3
=
O
H
3
CCCH
3
OH
CN

H
2
C=CCNH
2
H
2
SO
4
=
O

CH
3
H
2
C=CCOCH
3
=
O

CH
3
HCN
(1)
H
2
SO
4
(3a)
CH
3
OH
(4)
H
3
CCCNH
2
H
2
SO
4
=
O

CH
3
OH

H
2
SO
4
+ H
2
O
(2)
(3b) H
2
O
Acetone Route

26 www.aiche.org/CEP September 2006 CEP

(CSTRs) can be used to avoid or reduce its formation.
The products pass through a gravity settler, which
removes the heavy components. The process stream is
then fed to a distillation system, where an indirect split
removes sulfuric acid in the bottoms from the first col-
umn. A mixture of methanol, water and MMA is
removed as a side stream at the top of the column, and
light ends are removed as the distillate.
Figure 2 is a residue curve map for the methanol-water-
MMAmixture, where the LLE region is indicated by the
hatched area. (Most commercial process simulators can be
used to plot residue curve maps; this one was generated using
the tools in MAYFLOWER.) It shows that the MeOH-MMA
azeotrope is the lowest boiler, and can be removed as a distil-
late. The bottoms composition lies on the water-MMAedge,
and a liquid-liquid split yields pure water as the aqueous
phase. The organic phase, consisting of water and MMA, is
sent to a distillation column, where the light-boiling water-
MMAazeotrope is a distillate and
pure MMAis the bottoms. The
water-MMAazeotrope is recycled
back to the decanter.
Figure 3 is a schematic of the
process. This scheme is not very
complicated and has relatively
easy separation tasks. No other
promising acetone-route alterna-
tives were found.
To design and cost this flow-
sheet in Phase 2, further experi-
ments are needed to determine the
kinetics for the reaction of ace-
tone cyanohydrin with sulfuric
acid, and to estimate the forma-
tion of DME during esterification.
The presence of sulfate salts
makes the physical property pre-
dictions inaccurate. Experiments
must therefore be performed to
determine the physical properties
required in designing some of the
unit operations shown in Figure 3.
t-Butanol route
MMAis made commercially
from t-Butanol (t-BuOH) by
companies in Japan.
t-Butanol is oxidized in the
vapor phase over a molybdenum-
based catalyst at 300-420C to
form methacrolein. Use of a PFR at
Reactions and Separations
I Figure 3. MMA production via the acetone route.
Acetone
HCN
25C
PFR
Acetone
HCN
Acetone
Cyanohydrin
H
2
SO
4
Water
PFR PFR
90C 140C
Experiments
Required
Experiments
Required
Physical Property
Data Required
Physical Property
Data Required
Acetone Cyanohydrin
Water
(Acetone, HCN)
H
2
SO
4
Methacrylamide Sulfate
Heavy Impurities
MeOH-MMA Azeotrope
MMA-Water
Azeotrope
Methyl
Methacrylate
(MMA)
Water
LL
Split
H
2
SO
4

H
2
SO
4
MeOH
Water
MMA
DME
Light Impurities
Settler 80C
CSTRs
MeOH
Ammonium Bisulfate
Heavy Impurities
*
*
*
*
*
*
*
*
*
*
*
*
*
**
*****
*********
****
******
Water
MMA MeOH
(100.0C)
(81.6C)
(100.5C) (64.4C) (64.5C)
VLE Model: NRTL
Pressure: 1.0 atm
Residue Curve
I Figure 2. Residue curve map for the MeOH-water-MMA system.
H
3
CCCH
3
OH

CH
3
O
2
(1)
H
2
C=CCH
=
O

CH
3
O
2
(2)
H
2
C=CCOH
=
O

CH
3
CH
3
OH
(3)
H
2
C=CCOCH
3
=
O

CH
3
t-Butanol Route
CEP September 2006 www.aiche.org/CEP 27
the highest temperature possible yields the highest conversion,
assuming there are no side reactions. Methacrolein is further
oxidized over a phosphomolybdic-acid-based catalyst at
270350C to produce methacrylic acid (MAA). The oxida-
tion of methacrolein is not affected by the presence of unreact-
ed t-BuOH, so no separation is required after the t-BuOH oxi-
dation step. Methacrolein oxidation is performed in a PFR at
the highest possible temperature to achieve high conversion.
CO and CO
2
are formed as byproducts in the reaction.
Analysis of the methacrolein-t-BuOH-water-MAA mix-
ture indicates that the methacrolein-water azeotrope is the
lowest boiler, and can be removed as a distillate. All of the
unreacted methacrolein is recycled back to the methacro-
lein oxidation stage. The t-BuOH-water-MAA mixture is
sent to a distillation column, where the t-BuOH-water
azeotrope is removed as a distillate and a mixture of MAA
and water is obtained as the bottoms stream.
Esterification of MAA with methanol produces MMA.
Since the reaction is carried out at atmospheric pressure
and relatively low temperature (80C), no significant for-
mation of DME is expected. A PFR is preferred over a
CSTR because it requires a lower reactor volume.
The lowest boiler among the reaction products is the
MeOH-MMAazeotrope (consisting mostly of MeOH),
which is removed as a distillate and recycled to the esterifi-
cation reactor. The mixture of water, MAAand MMAis sent
to a decanter, where water is removed as the aqueous phase.
The organic phase, consisting of
MAA, MMAand some water, is
sent to a distillation column.
MAAis obtained as the bottoms
stream and is recycled to the
esterification reactor. The distil-
late consists of MMAand water;
it is decanted to get an aqueous
phase of pure water and an organ-
ic phase consisting of MMAand
some water. The organic phase is
fed to a distillation column that
produces pure MMAas the bot-
toms stream. The distillate is the
MMA-water azeotrope, which is
recycled to the decanter. Figure 4
is a flowsheet for this process.
Experiments needed to
design and cost this flowsheet in
Phase 2 include the measure-
ment of the reaction kinetics of
the oxidation reactions, and the
determination of the VLE and
LLE data for a mixture contain-
ing methanol, water, MAA and MMA.
Several alternatives to this scheme can be generated. For
example, if excess MeOH is used in the esterification reac-
tor to ensure the complete conversion of MAA, the process
would be as shown in Figure 5a. The feed rates to the sepa-
I Figure 4. MMA production via the t-butanol route.
420C
Experiments
Required
Physical Property
Data Required
Water
LL
Split
350C
Air
80C
Methacrolein
t-BuOH
Water
MAA
CO
2
, CO
Methacrolein-Water Azeotrope
t-BuOH
Air
t-BuOH-Water Azeotrope
t-BuOH
Water
MAA
MAA
Water
MeOH
MeOH-MMA
Azeotrope
LL
Split
Water
MMA
MMA
Water
MMA-Water
Azeotrope
MMA
Water
Water
MAA
MMA
Water
MAA
MMA
MeOH
Water
MAA
MMA
MAA
I Figure 5a. Alternative scheme for the t-butanol route using
excess MeOH in the esterification reactor for complete
conversion of MAA.
Water
80C
t-BuOH
Water
MAA
MAA
Water
Excess
MeOH
MeOH-MMA
Azeotrope
LL
Split
Water
MMA
MeOH
Water
MAA
MMA
Water
MMA
MMA-Water Azeotrope
t-BuOH-Water
Azeotrope
Same steps
as in Figure 4
28 www.aiche.org/CEP September 2006 CEP
rations that follow the esterification will be larger, but the
separation system is much simpler. MAApolymerizes at
high temperatures, especially when the MAAconcentration
is high. Therefore, the distillation columns for mixtures con-
taining MAAmust be operated at a lower pressure.
Alternatively, liquid-liquid extraction can be used
for separating MAA from other components, as shown
in Figure 5b.
The alternatives in Figures 4 and 5 were not found in
the patent or published literature, thereby demonstrating
the ability of the rapid design methodology to generate
new intellectual property.
Ethylene route
MMAcan be manufactured from ethylene via
methacrolein, propionaldehyde or (as considered
here) propanoic acid.
The first step is the carbonylation of ethylene
to form propanoic acid (PROP) by reacting the
ethylene with CO and water at a pressure of
2351 atm and a temperature of 130170C (13).
Propanoic anhydride (PAA) is formed as a
byproduct. The unconverted ethylene and CO are
recovered in a flash drum. Water formed during
the reaction could be removed before the next
reaction step, but since water is a product in the
next step, it is not necessary to do so.
Propanoic acid undergoes condensation reaction
with formaldehyde to form MAA. Propanoic anhy-
dride reacts with formaldehyde to form MAAand
propanoic acid (14). Unreacted formaldehyde is
removed as a distillate with water, and the bottoms
stream is fed to a distillation column, where MAA
is removed as the bottoms steam. Amixture of
propanoic acid, PAAand water is fed to another dis-
tillation column, where the low-boiling PAA-water
azeotrope, the distillate, forms two liquid phases;
after phase-split, pure water is produced as the
aqueous stream. The organic phase with PAAand
water is recycled back to the condensation reaction.
MAA is fed to the esterification reactor with
methanol. The separation scheme for the outlet of
the esterification reactor is identical to the scheme used in
the t-butanol route (Figure 4).
The flowsheet for this reactor and separation system is
shown in Figure 6.
To design and cost this flowsheet in Phase 2, experi-
ments must determine the reaction kinetics for the con-
densation reaction, as well as the VLE and LLE data for
a mixture of PROP, PAA, MAA and water.
The flowsheet can be simplified by using an excess of
methanol in the esterification reaction, as shown in Figure
7a. Alternatively, because MAA polymerizes at higher
temperatures, after the condensation step, MAA can be
recovered using liquid-liquid extrac-
tion instead of low-pressure distilla-
tion, as shown in Figure 7b.
Several other process alternatives can be
generated for this chemical route, but they
are all variations of the themes already
identified in Figures 6 and 7. They can be
evaluated further in Phase 2 if the ethylene
route survives the Phase 1 evaluation crite-
ria and makes it to Phase 2.
Reactions and Separations
Water
LL
Split
80C
t-BuOH
Water
MAA
MAA
Water
MeOH
LL
Split
Water
MMA
MMA
Water
MMA-Water
Azeotrope
MMA
Water
Water
MAA
MMA
Water
EXTR EXTR
Solvent
Water
MAA
MAA
MAA
Water
MAA
MMA
MeOH
Water
MAA
MMA
MAA
MeOH-MMA Azeotrope
t-BuOH-Water
Azeotrope
Same steps
as in Figure 4
without recycle of
Methacrolein-Water
Azeotrope
I Figure 5b. Alternative scheme for the t-butanol route in which MAA is
separated by liquid-liquid extraction.
H
2
C=CH
2
CH
3
CH
2
COH
=
O
CO +
H
2
O
(1)
HCHO
(3a)
H
2
C=CCOH
=
O

CH
3
CH
3
OH
(4)
H
2
C=CCOCH
3
=
O

CH
3
H
3
CH
2
CC
=
O
H
3
CH
2
CC
=
O
O

(2) CO +
H
2
C=CH
2
(3b)
HCHO
Ethylene Route
CEP September 2006 www.aiche.org/CEP 29
The alternatives in Figures
6 and 7 were not found in
existing patent or published
literature, again demonstrat-
ing the ability of the rapid
design methodology to gener-
ate new intellectual property.
Early cost estimates
The outcome of Phase 1
(discovery/development) is
a set of potentially attractive
process alternatives for each
chemical route under con-
sideration. The complexity
of carrying out the reactions
and separations for each
route is important in getting
a qualitative estimate of the
manufacturing costs.
The process alternatives
for the t-butanol and the ethyl-
ene routes are more complex
than the acetone route. The
acetone route, however, generates much more waste,
and the waste-treatment costs must be considered.
The availability of the raw materials for each of
I Figure 6. MMA production using the ethylene route, via propanoic acid as an intermediate.
Water
LL
Split
150C
40 atm
80C
MeOH
LL
Split
Water
MMA
MMA
Water
MMA-Water
Azeotrope
MMA
Water
Water
MAA
MMA
MAA
MAA
Water
MAA
MMA
MeOH
Water
MAA
MMA
MeOH-MMA Azeotrope
Flash
Ethylene
CO
Water
Ethylene
CO
300C
PROP
PAA
Water
HCHO
Water
HCHO
PROP
PAA
MAA
Water
PROP
PAA
Water
PAA-Water
Azeotrope
LL
Split
Water
PAA
Water
PAA-PROP
MAA
Experiments
Required
Physical
Property
Data
Required
MAA
I Figure 7a (left). Alternative scheme for the ethylene route uses excess MeOH in the esterification reactor for complete conversion of MAA.
I Figure 7b (right). MAA is separated by extraction.
PAA*
Water
Solvent
PROP
Solvent
PROP*
Water
LL
Split
80C
MeOH
LL
Split
Water
MMA
MMA
Water
MMA-Water
Azeotrope
MMA
Water
Water
MAA
MMA
MAA
Water
MAA
MMA
MeOH
Water
MAA
MMA
MAA
MeOH-MMA Azeotrope
MAA
Water
PROP
PAA
MAA
Water
Same steps
as in Figure 6
except streams
marked *
EXTR
EXTR
Water
80C
Excess
MeOH
LL
Split
MMA
MMA-Water
Azeotrope
MMA
Water
Water
MMA
MeOH
Water
MMA
MeOH-MMA Azeotrope
PROP
PAA
MAA
Water
PROP
PAA
Water
PAA-Water
Azeotrope
LL
Split
Water
PAA
Water
PAA
PROP
MAA
MAA
Same steps
as in Figure 6
30 www.aiche.org/CEP September 2006 CEP
the routes will depend on the company implementing the
project. For example, if acetone is produced as a byprod-
uct from some other chemical process that the firm oper-
ates and is available as a cheap raw material, the acetone
route might be desirable. Transportation of hydrogen
cyanide might be problematic, so a site that does not pro-
duce HCN might not prefer the acetone route.
Raw materials costs for the three routes are compared in
Table 3. (Prices were taken from the Chemical Market Re-
porter, Nov. 2003.) Known selectivity data (Table 1) were
taken into account to determine raw materials consumptions.
The difference between the product price and the raw
material cost in Table 3 represents the margin available for
manufacturing costs and the return on investment. Assuming
that the raw materials are not produced on-site for any of the
routes, the ethylene route has the lowest raw materials costs.
The total product manufacturing cost of MMA is the
sum of raw materials costs, capital costs and operating
costs. During the first phase of rapid design, equipment
sizes are not known. Capital and operating costs can be
estimated based on the observation that the size of equip-
ment is strongly dependent on the inlet flowrates.
Arough estimate of the total capital and operating costs
can, therefore, be obtained by summing the inlet flows to all
the units in a flowsheet. The size of the equipment depends
on the flowrates, and the number of units in the flowsheet is
accounted for by the number of flowrates that are summed.
The flowsheets combined mass flowrate is assumed to be
directly proportional to its capital and operating costs.
The mass index is defined as the ratio of the sum of all
inlet mass flowrates to the production rate of the desired
product. The higher the mass index, the higher the expect-
ed capital and operating costs. The mass index allows a
qualitative comparison of the flowsheets, which is suitable
for decision-making at the Phase 1 stage of a new venture.
The mass indexes for the three MMA routes appear in
the first column of Table 4.
To compare the routes, the sum of raw material costs
together with the capital and operating costs must be esti-
mated. A fair comparison can be accomplished by normal-
izing the raw materials cost and the mass index (which
represents the capital and operating costs).
Since the t-butanol route has the lowest mass index, it is
assigned a normalized mass index value of 1.0, and the
other routes mass indexes are normalized relative to that.
Similarly, the ethylene route has the lowest raw materials
cost, so its normalized raw materials cost is 1.0, and the
other routes costs are normalized relative to that (Table 4).
The normalized total cost for each route is the weighted
sum of the normalized raw materials cost and the normal-
ized mass index. For instance, if the raw materials cost is
50% of the total product cost, the normalized total cost is
the sum of 0.5 times the normalized raw materials cost and
0.5 times the normalized mass index. The normalized total
cost for each route as a function of the weight assigned to
the raw materials cost is illustrated in Figure 8.
Figure 8 shows that the t-butanol route is more expensive
than the acetone route at the lower end of the typical raw
materials cost curve. As raw materials account for more of the
total costs, the t-butanol route is much more expensive than
the other two routes. Analysis of Figure 8, therefore, suggests
that the t-butanol route should not be pursued further.
For commodity chemicals, raw materials typically
account for 37% to 87% of the total product manufactur-
ing costs (15). Considering this range, the acetone and eth-
ylene routes look economically attractive. The acetone
route is much less sensitive to the weight assigned to raw
materials cost than the ethylene route.
For oxidation products such as MMA, raw materials usu-
ally account for 75% of the total product costs (15), where
the normalized total cost for both the acetone route and the
Reactions and Separations
I Figure 8. Normalized total cost vs. raw materials cost as a
fraction of total cost.
Typical Range
for
Commodity
Chemicals
Raw Materials Cost as a
Fraction of Total Product Cost
N
o
r
m
a
l
i
z
e
d

T
o
t
a
l

C
o
s
t
2.5
2
1.5
1
0.5
0
0 0.2
0.4 0.6 0.8 1
Ethylene
Acetone
t-Butanol
Table 3. Economic analysis based only on
raw materials costs to produce 220.4 million lb/yr of MMA.
Product Price Minus
Costs of Raw Materials Raw Materials Costs
Route $MM/yr $/lb MMA $MM/yr $/lb MMA
Acetone 80.34 0.365 79.45 0.360
t-Butanol 128.59 0.583 31.20 0.142
Ethylene 63.95 0.290 95.84 0.435
Table 4. Mass index and normalized costs
to produce 220.4 million lb/yr of MMA.
Normalized
Normalized Raw
Mass Mass Materials
Route Index Index Cost
Acetone 22.39 1.35 1.26
t-Butanol 16.56 1.00 2.01
Ethylene 25.85 1.56 1.00
CEP September 2006 www.aiche.org/CEP 31
ethylene route are very close. These routes must be analyzed
further in Phase 2, where a more-detailed estimate of the cap-
ital and operating costs together with the raw materials costs
will identify the more economically attractive route. These
more-detailed costs require better estimation of the thermody-
namic properties so that the conceptual design of equipment
can be performed together with the economic analysis.
Phase 2: What should the chemists
and engineers work on?
Acetone route. The reaction of acetone cyanohydrin
with sulfuric acid must be studied in more detail, since
reaction performance has a large effect on the overall
process. The reaction conditions for the conversion of the
unwanted intermediate (-hydroxyisobutyramide) into the
required intermediate (methacrylamide sulfate) must be
studied to allow for reactor design. Process chemists and
engineers must determine the effects of temperature on the
reaction rates, conversion of acetone cyanohydrin to both
intermediates, and the transformation of -hydroxyisobu-
tyramide into methacrylamide sulfate.
The esterification reaction (Step 3) produces dimethyl
ether as a byproduct. Experimental measurement of the selec-
tivity to MMAwill determine the methanol consumption.
A better estimate of the VLE between water and ace-
tone cyanohydrin is required to verify that pure water can
be removed from its mixture with acetone cyanohydrin
after Step 2. The schematic in Figure 3 shows that a liq-
uid-liquid split gives pure water as the aqueous phase.
Experimentally measured LLE data are needed for reliable
conceptual design of the separation system.
Ethylene route. The reaction kinetics of the condensa-
tion reaction must be determined. The conversion plays an
important role in deciding the size of the separation sys-
tem that recovers unreacted propanoic acid and PAA from
MAA and water. The condensation of propanoic acid and
PAA has a very low conversion. This results in larger
recycle flows and, therefore, larger equipment.
Experiments must be conducted to design new catalysts
and study the effects of reaction parameters to increase the
conversion and reduce the total cost for this route.
The liquid-liquid equilibrium behavior of the mixture
of water, MAA and MMA must be measured experimen-
tally to verify the separation scheme outlined in Figure 6.
Because MAA polymerizes at high temperatures and the
distillation columns must be operated at a lower pressure,
the effect of temperature on MAA polymerization in the
presence of water and other components must be known
before deciding on a threshold temperature (and hence
pressure) for the columns. Operating low-pressure distilla-
tion columns is usually expensive, so the use of extraction
columns might become economically attractive.
Depending on the outcome of the experiments for thermal
stability of MAA, the extraction column alternative may
be explored, which will require experimental measurement
of LLE together with extraction-solvent selection.
Closing thoughts
The use of systematic methods coupled with a single
team of discovery chemists and conceptual design engi-
neers leads to the quick generation of process options
using limited experimental data. The rapid design method-
ology presented here identifies the key areas for further
experimentation and other process development activity.
Its focus is to guide the process development efforts away
from dead-ends and toward routes leading to profitable
world-class manufacturing processes.
Important decisions that must be made during the dis-
covery stage must rely not only on the business aspects but
also on the chemical and engineering aspects of a chemical
route. Therefore, the availability and price of the raw mate-
rials together with the complexity of the process must be
taken into account before picking a chemistry route.
The goal of Phase 2 is to select the chemistry route and
the manufacturing process using better cost estimates. The
rapid design methodology can be effectively used to guide
the development program and at the same time generate
new intellectual property.
SAGAR B. GADEWAR is a senior research engineer at GRT, Inc. (861 Ward Dr.,
Santa Barbara, CA 93111; Phone: (805) 696-6660 x104; Fax: (805) 696-
6761; E-mail: [email protected]). His research interests include
reaction combined with distillation, conceptual design for complex
chemistries, and conversion of natural gas to liquid hydrocarbons. He
received his BTech in chemical engineering from the Univ. of Mumbai
(UDCT), and his PhD in chemical engineering from the Univ. of
Massachusetts at Amherst, and he is a member of AIChE.
GERHARD SCHEMBECKER was appointed professor for plant and process
design at Dortmund Univ. (Dept. of Biochemical and Chemical
Engineering, Univ. of Dortmund, D-44221 Dortmund, Germany; Phone:
+49-231-755-2524; Fax : +49-231-755-2341; Email: [email protected]
dortmund.de) in September 2005. He is co-founder of the consultancy
firm Process Design Center and has worked on more than 100 industrial
process synthesis projects. His research interests include the conceptual
design of (bio)chemical processes with special interest in
chromatographic and crystallization processes. He received his diploma
(1988) and his PhD (1992) in chemical engineering from Dortmund Univ.
MICHAEL F. DOHERTY is a professor of chemical engineering at the Univ. of
California, Santa Barbara (Dept. of Chemical Engineering, Univ. of
California, Santa Barbara, CA 93106-5080; Phone: (805) 893-5309; Fax:
(805) 893-4731; E-mail: [email protected]). His research and
teaching interests include chemical process design and synthesis, and
separation systems design with special interest in simultaneous reaction
and separation, and crystallization of organic materials. He received his
BSc in chemical engineering from Imperial College, London, and his PhD
in chemical engineering from Trinity College, Cambridge, and he is a
member of AIChE.
Article continues with Literature Cited on next page.
32 www.aiche.org/CEP September 2006 CEP
Reactions and Separations
Literature Cited
1. Sharratt, P. N., et al., Generating Innovative Process
Designs using Limited Data, J. Chem. Technol. Biotechnol.,
78 (23), pp. 156160 (Feb.Mar. 2003).
2. Schembecker, G., et al., Synthesis of Chemical Process
Flowsheets by Means of Cooperating Knowledge Integrating
Systems, Chem. E. Symp. Ser., 133, pp. 333341 (1994).
3. Barnicki, S. D., and J. J. Siirola, Process Synthesis
Prospective, Comp. Chem. Eng., 28 (4), pp. 441446
(April 2004).
4. Schembecker, G., et al., READPERT Development,
Selection and Design of Chemical Reactors, Chem. Eng.
Proc., 34 (3), pp. 317322 (June 1995).
5. Gadewar, S. B., et al., A
Systematic Method for Reaction
Invariants and Mole Balances for
Complex Chemistries, Comp. Chem.
Eng., 25 (910), pp. 11991217
(Sept. 2001).
6. Gadewar, S. B., et al., Reaction
Invariants and Mole Balances for
Plant Complexes, Ind. Eng. Chem.
Res., 41 (16), pp. 37713783 (Sept.
2002).
7. Prausnitz, J. M., and F. W.
Tavares, Thermodynamics of Fluid-
Phase Equilibria for Standard
Chemical Engineering Operations,
AIChE J., 50 (4), pp. 739761 (Apr.
2004).
8. Fredenslund, A., et al., Group-
Contribution Estimation of Activity
Coefficients in Nonideal Liquid
Mixtures, AIChE J., 21 (6), pp.
10861099 (Nov. 1975).
9. Marrero, J., and R. Gani, Group-
Contribution Based Estimation of
Pure Component Properties, Fluid
Phase Equilibria, 183-184, pp.
183208 (July 2001).
10. Huss, R. S., et al., Computer-Aided
Tools for the Design of Reactive
Distillation Systems, Comp. Chem.
Eng., 23 (special issue), pp.
S955S962 (1999).
11. Watkins, K. J., Expanding
Capacity Unconventionally, Chem.
Eng. News, 79 (12), pp. 16 17 (Mar.
19, 2001).
12. Kirk and Othmer Encyclopedia of
Chemical Technology, John Wiley
& Sons, Hoboken, NJ, Vol. 16, pp.
487497 (1995).
13. Spivey, J. J., et al., Novel Catalysts
for the Environmentally Friendly
Synthesis of Methyl Methacrylate,
Ind. Eng. Chem. Res., 36 (11), pp.
46004608 (Nov. 1997).
14. Gogate, M. R., et al., Annual
Technical Report for DE-AC22-
94PC94065, U.S. Dept. of Energy,
Pittsburgh Energy Technology
Center, Pittsburgh, PA (Dec. 1996.)
15. Grumer, E. L., Chem. Eng., 74 (9),
pp. 190193 (Apr. 1967).
CEP
www.cepmagazine.org or Circle No.121