Ind. Eng. Chem. Res.

2007, 46, 6953-6962 6953

CO2 Separation from Purge Gas and Flue Gas in the Methanol Process, Using
NLP Model Optimization
Anita Kovač Kralj* and Peter Glavič
Faculty of Chemistry and Chemical Engineering, UniVersity of Maribor, SmetanoVa 17, Maribor, SloVenia

The concentration of CO2 in the atmosphere has to be stabilized, requiring a reduction in current emission
rates in existing plants. This will be done by reducing the environmental burden imposed in such areas as
materials input and CO2 emission reduction and using cleaner production, resources, and energy recycling.
Any opportunities for emission reduction and CO2 reuse largely depend on existing plant and energy systems.
CO2 can be separated from the outlet stream (purge gas) and from flue gas by a membrane or absorption
system (absorber and regenerator) or adsorption system and reused as a reactant in a reactor system. Therefore,
product yield can be increased and CO2 emissions reduced, simultaneously. CO2 emissions can be reduced at
the source. The authors of this paper studied CO2 reuse in a methanol process, in which electricity can be
generated using an open gas turbine, followed by a separator. Simultaneous optimization of a process structure
and its parameters using simplified nonlinear programming (NLP) ensures an additional annual profit, influenced
by reusing the flow rate of CO2. The additional electricity cogeneration and additional flow rates of the raw
material could generate an additional profit of 2.79 MEUR/a.

1. Introduction contributor to the problem due to its abundance, thus becoming

The relationship between energy efficiency and flue gas
emission is clear. The higher the energy efficiency, the less fuel
is burned and, hence, the smaller the flue gas emission. There
are two major optimization approaches: pinch analysis11 and
mathematical programming methods13 can be used to improve
energy efficiency through better heat integration and utility
savings and, hence, reduced flue gas emissions. Pinch analysis
does not guarantee an optimal global solution because it cannot
be used simultaneously as amount and enthalpy balances, but
it quickly proposes beneficial, integrated structures in nontrivial
complex processes. Mathematical optimization methods can be
classified as simultaneous approaches.3 These simultaneous
approaches can account for capital and energy trade-off ac-
curately and can thus yield a better solution, but convergence
is difficult for complex and energy-intensive processes because
the number of variables increases with the number of combina-
tions.
In this paper we have analyzed the possibility of reusing the
CO2 flow rate in a cogeneration system and decreasing the CO2
emissions. First the techniques used for CO2 separation are
presented, the pressure swing adsorption (PSA) being selected
because the existing inactive one already exists in the process.
The only simulated absorption model of CO2 separation, using
the ASPEN PLUS simulator, is then described, because the real
adsorption and membrane models cannot be simulated using
the ASPEN PLUS.
2. CO2 Separation
Currently, the desire to alleviate the problem of global
warming has resulted in environmental concerns over a reduction
in greenhouse emissions from industrial sources. The major
greenhouse gases are methane (CH4), carbon dioxide (CO2),
nitrous oxide (N2O), halogens such as chlorofluorocarbons, and
hydrofluorocarbons. From among them, CO2 is the primary
* To whom correspondence should be addressed. Tel.: +386 02
2294454. Fax: +386 02 2527 774. E-mail: anita.kovac@uni-mb.si.
10.1021/ie0702173 CCC: $37.00 © 2007 American Chemical Society
Published on Web 09/19/2007
a major target for reduction. Concern over global climate change
has led to a need for reducing CO2 emissions from chemical
plants.5 The technologies for removing CO2 from flue gas are
the use of polymer or ceramic membranes, adsorption or
absorption systems.
2.1. Membrane. The word membrane comes from the Latin
word, ’membran’ which means ‘a skin’.16 Today’s use of
‘membrane’ has been extended to describe a thin flexible sheet
or film, acting as a selective boundary between two phases
because of its semipermeable properties. Physically a membrane
could be solid or liquid. It functions as a separation agent that
is very selectively based on the differences in diffusivity
coefficient, electric current, or solubility.17,10 There are several
ways to classify membranes. Based on their materials, mem-
branes are classified as polymeric membrane, liquid membrane,
solid, (ceramics) membrane, and ion exchange membrane. Based
on what they do and how they perform, membranes are
classified as fine filtration (microfiltration/MF, ultrafiltration/
UF, nanofiltration/NF, and reverse osmosis/RO), dialysis, elec-
trodialysis (ED), gas separation (GS), carried-mediated transport,
control release, membrane electrode, and pervaporation (PV).
For the separation of gases, especially oxygen/nitrogen and
methane/carbon dioxide, new glassy polymers and inorganic
materials such as zeolites are used to produce membranes with
better selectivity and higher fluxes.
2.2. Absorption. Absorption is a process by which one or
more components of a gas mixture are transferred to a liquid
where it is soluble. The operation of absorption can be
categorized, on the basis that the interaction is natural between
absorbent and absorbate, into the following two types: (1)
physical absorption, the component being absorbed is more
soluble in the liquid absorbent than the other gas components
with which it is mixed but does not react chemically with the
absorbent and (2) chemical absorption, characterized by the
occurrence of a chemical reaction between the gas component
being absorbed and a component in the liquid to form a
compound.
To date, absorption is still a powerful tool for gas separation
and purification. However, current processes for the enhance-
6954 Ind. Eng. Chem. Res., Vol. 46, No. 21, 2007

Table 1. Data of the Existing Flue Gas

component mass flow rate, q (kg/h)
CO 1000
CO2 3300
H2O 6000
O2 2000
N2 13000

reverse reaction is endothermic. Solvents with higher reactivity,

Figure 1. Basic flow sheet for the absorption system (absorber and
regenerator).
ment of gas-liquid absorption rate are still limited to those
chemical reactions, which occur in the liquid phase between
gas component and liquid solution. Commercial CO2 removal
units utilize a physical solvent of alkanolamine, such as
monoethanolamine (MEA), diethanolamine (DEA), and meth-
yldiethanolamine (MDEA). These technologies require large
units and high regeneration energy requirement, while exhibiting
solvent and equipment degradation. However, the cost of CO2
capture using conventional solvents such as aqueous solutions
of MEA is prohibitively high. Nearly half of the total cost is
due to operational costs, of which about 70-80% can be
attributed to the cost of solvent regeneration. In addition, the
conventional solvents are rather corrosive and can be degraded
by many of the contaminants present in the flue gas streams. A
major reason for the high cost is that the conventional solvents
have a variety of practical problems including the following:
(1) a high-energy consumption for solvent regeneration, (2) a
high rate of process equipment corrosion, (3) a fast evaporation
rate causing high solvent losses, and (4) a high rate of
degradation in the presence of oxygen. All of these problems
translate into high capital and operating costs.
The partial pressure of CO2 in the feed gas, as a first
approximation, determines the solvent’s circulation rate. In the
countercurrent column, the feed gas contacts the fully loaded
rich solvent at the bottom of the column (Figure 1).
At a minimum, there must, therefore, be sufficient solvent
to absorb all the CO2 in the feed and provide a driving force
for absorption at the bottom of the column. The partial pressure
of CO2 in the product determines the degree of regeneration
required. The equilibrium partial pressure of CO2 in the
regenerated solvent must be lower than the partial pressure of
the product gas. Otherwise, there will not be a positive driving
force for absorption at the top of the absorber. Therefore, the
two most important cost factorsscirculation rate and regenera-
tion energy dutysare established, to a significant extent, by the
CO2 partial pressure in the feed and the product. In addition, a
significant portion of the investment of a chemically reactive
system is dependent on the regeneration energy duty. CO2
absorption with a chemically reacting solvent involves mass-
transfer using chemical reaction. The role of the chemical
reaction is to speed up the mass-transfer rate as well as to
provide greater solvent loading capacity under lower CO2 partial
pressure conditions. The reaction between dissolvent CO2 and
amine is reversible and proceeds through a series of reactions.
In general, the chemical reaction proceeds forward in the
absorption direction under low temperature and high partial
pressure conditions and the reverse in the desorption (regenera-
tion) direction under high temperature and low partial pressure
conditions. Ideally, higher reaction rate constants in both forward
and reverse reactions are favorable. This allows the absorption
and regeneration to be accomplished within as short a time frame
as possible. The forward reaction is exothermic, while the
as indicated by higher forward reaction rate constants, also tend
to have higher heat reaction, and, therefore, they require more
energy for regeneration and are more difficult to regenerate.
As such they tend to have higher residual solvent loadings (lean
loadings) than the less reactive ones. It should be noted that
the higher the rate constant, the greater the mass-transfer
coefficient, thus a lower capital cost for the absorbers. CO2
partial pressure under absorber conditions is more than 10 bar.
CO2 partial pressure under regenerator conditions is 1 bar.
Other feasible technologies for large-scale CO2 recovery are
hot K2CO3 absorption and low-temperature absorption by
methanol.
The overall reactions between absorbents and CO2 can be
written as follows:10
CO2 removal:
2NaOH + CO2 f Na2CO3 + H2O
Na2CO3 + H2O + CO2 f 2NaHCO3
CO2 removal:
CO2 + 2HOCH2CH2NH2 T
HOCH2CH2NHCOO- + HOCH2CH2NH2+
CO2 + 2(HOCH2CH2)2NH T
(HOCH2CH2)2NCOO- + (HOCH2CH2)2NH+
Regenerable CO2 capture process chemistry with aqua
ammonia (or urea15):
CO2 removal:
2NH3 + CO2 f NH2COONH4
NH2COONH4 + CO2 + 2H2O f 2NH4HCO3
NH2COONH4 + H2O f NH4HCO3 + NH3
NH3 + CO2 + H2O f NH4HCO3
2NH3 + CO2 + H2O f (NH4)2CO3
(NH4)2CO3 + CO2 + H2O f 2NH4HCO3
CO2 regeneration:
2NH4HCO3(aq) f (NH4)2CO3(aq) + CO2(g) + H2O
NH4HCO3(aq) f NH3(aq) + CO2(g) + H2O
(NH4)2CO3(aq) f 2NH3(aq) + CO2(g) + H2O
2.2.1. Simulated Absorption Model of CO2 Separation
Using ASPEN PLUS. Simulation models of CO2 separation
using an absorbent system with different absorbents are
presented, using existing flue gas data (Tables 1 and 2) and
using the ASPEN PLUS simulator.18 The aqua absorbent bases

Table 2. Absorption Data
mass flow rate data in
absorbent of absorbent, q/(kg/h) absorber
K2CO3 40 p ) 20 bar
water 100 T ) 30 °C
MEA 40 p ) 20 bar
water 100 T ) 30 °C
DEA 40 p ) 20 bar
qater 100 T ) 30 °C
NH3 40 p ) 20 bar
water 100 T ) 30 °C
use electrolyte chemistry. In ASPEN PLUS, an electrolyte
system is defined as one in which some of the molecular species
dissociate partially or completely into ions in a liquid solvent.
The presence of ions during the liquid phase causes highly
nonideal thermodynamic behavior. ASPEN PLUS provides a
specialized thermodynamic model (the ELECNRTL) for the
electrolytes.
The absorption model involves the formation of ionic species.
In electrolytic systems molecular species dissociate. The
chemistry of electrolytes must be included in the model. In order
to specify the electrolytic chemistry, the following must be
defined: (1) complete set of components present, (2) concentra-
tion of reactants, (3) temperature of reaction, and (4) liquid phase
inlet flow rate.
The best CO2 separation is in aqua ammonia absorbent using
the same data and conditions. If there were no nitrogen, CO2
capture from flue gas would be greatly simplified.
2.3. Adsorption. Adsorption separation of CO2 from simu-
lated flue gas containing CO2, O2, and N2, with and without
moisture, was investigated using a novel nanoporous adsorbent
based on polyethylenimine (PEI)-modified mesoporous molec-
ular sieve MCM-41 in a flow system. The CO2 adsorption
capacity and CO2 separation selectivity of MCM-41 were greatly
improved by loading PEI into its nanosized pore channels, which
caused the resulting adsorbent to operate like a ‘molecular
basket’ for CO2. The CO2 adsorption capacity of the MCM-
41-PEI adsorbent for the simulated moist flue gas was higher
than that for the simulated dry flue gas.
Pressure swing adsorption (PSA) is a commonly used
technology for purifying gases. PSA technology was introduced
commercially in the 1960s, and today PSA is used extensively
in the production and purification of CO2, oxygen, nitrogen,
and hydrogen for industrial uses. PSA processes can be used to
separate gases in a mixture because different gases tend to be
Figure 2. Adsorption (PSA) column.
Ind. Eng. Chem. Res., Vol. 46, No. 21, 2007 6955
CO2 absorb in data in CO2 regeneration
absorbent, q (kg/h) regenerator from absorbent, q (kg/h)
1668 p ) 1 bar 1610
T ) 70 °C
1665 p ) 1 bar 1607
T ) 70 °C
1669 p ) 1 bar 1611
T ) 70 °C
1671 p ) 1 bar 1613
T ) 70 °C
attracted to different solid surfaces more or less strongly. It
operates at near ambient temperatures. PSA is the technology
used to separate certain species from a gas under pressure,
according to these species molecular characteristics and affinity
for an adsorbent material.20 Special adsorptive materials (e.g.,
zeolites) are used as a molecular sieve, preferentially adsorbing
the undesired gases at high pressure. PSA can separate the
carbon dioxide (CO2) from outlet gas. Application of the PSA
method was examined at 26 bar and 35 °C using zeolite 13 X
as adsorbent for CO2 separation from the flue gas (Figure 2).
The main process comprised PSA and a super cold separator.
Carbon dioxide, present at about 10% concentration in the flue
gas, was initially concentrated by PSA, and then the concentrated
CO2 was liquefied by a super cold separator (condenser) up to
a concentration of more than 99%.
We studied the development of a new material for cost-
effective separation with high CO2 adsorption capacity using
activated carbon particulates (ACP) and K2CO3.9 K2CO3 is
loaded onto the support by pseudoincipient wetness at various
loadings by varying the solution concentration. The nanodis-
persed nature of K2CO3, combined with the use of small support
particulates, promotes high K2CO3 utilization, high contact
efficiency, and a high accessibility of K2CO3. At equivalent bed
volumes, the K2CO3/ACP entrapped materials provide longer
breakthrough times for CO2 removal as compared to packed
beds of K2CO3 pellets. At 87% relative humidity, maximum
capacity at room-temperature based on the total weight of CO2
adsorption is found to be around 0.037 g with loading of w(K2-
CO3) ) 22% for packed beds corresponding to 33% utilization
of impregnated K2CO3. With equivalent K2CO3 loading, mi-
crofibrous entrapped CO2 sorbents show an improvement in
breakthrough capacity and pressure drop. The use of entrapped
microfibrous sorbents provides reduced external mass transfer
resistance, enhanced adsorption rates, and less heat required for
regeneration. This novel adsorbent allows mass reduction in the
overall system and volume reduction for continuous removal
of CO2 traces from gas streams with high levels of relative
humidity, namely a fuel processor stream.
Regeneratable CO2 capture process chemistry with potassium
carbonate (low temperature application):
CO2 removal:
K2CO3 + CO2 + H2O f 2KHCO3
CO2 regeneration:
2KHCO3 + heat f K2CO3 + CO2 + H2O
Regeneratable CO2 capture process chemistry with calcium
oxide (high temperature application):
CO2 removal:
CaO + CO2 f CaCO3
6956 Ind. Eng. Chem. Res., Vol. 46, No. 21, 2007

CO2 regeneration: The methanol process is composed of three subsystems: (1)

production of synthesis gas, (2) production of crude methanol,
CaCO3 + heat f CaO + CO2 and (3) purification of methanol (F301, D301-D304).
We have studied the possible reuse of the CO2 flow rate from
Lithium zirconate (Li2ZrO3) is one of the most promising purge gas and flue gas, the reduction of CO2 emissions,
materials for CO2 separation from flue gas at high temperatures.7 retrofitting the Lurgi process for low-pressure crude methanol
This material is known for its ability to absorb a large amount production, and extending a simplified flow sheet by using a
of CO2 at around 400-700 °C. However, the mechanism of gas turbine (Figure 4).
the CO2 sorption/desorption process on Li2ZrO3 is unknown, Natural gas is first desulfurized (D101) and then heated up
as yet. Li2ZrO3 powders were prepared from lithium carbonate in a steam reformer (REA-1) where synthesis gas is produced
(Li2CO3) and zirconium oxide (ZrO2) by the solid-state method, from natural gas and steam:
and the CO2 sorption/desorption property was examined by
thermogravimetric analysis (TGA). It was shown that pure Li2- 3C2H6 + 6.5H2O f 2CO + 12H2 + 1.75CH4 + 2.25CO2
ZrO3 absorbs a large volume of CO2 at high temperature with ∆rH298 ) 196.17 kJ/mol (R1)
a slow sorption rate. The addition of potassium carbonate (K2-
CO3) and Li2CO3 into the Li2ZrO3 remarkably improves the 3C3H8 + 10H2O f 3.5CO2 + 17H2 + 3CO + 2.5CH4
CO2 sorption rate of the Li2ZrO3 materials. The reaction between ∆rH298 ) 277.88 kJ/mol (R2)
Li2ZrO3 and CO2 is reversible.
The adsorbents for adsorption can include synthetic (Z10- 3C4H10 + 13.5H2O f 4.75CO2 + 22H2 + 4CO + 3.25CH4
08, Z10-10, ...) and/or natural (Na Alumino Silicate 1, Na ∆rH298 ) 361.48 kJ/mol (R3)
Alumino Silicate 2, Ca K Alumino Silicate, ...) zeolites.
In this paper, CO2 separation from purge gas and flue gas CH4 + H2O / CO + 3H2 ∆rH298 ) 206.08 kJ/mol
using the PSA system was used as a case study because the (R4)
existing inactive PSA already exists in the plant.
CO + H2O / CO2 + H2 ∆rH298 ) -41.17 kJ/mol
3. Case Study (R5)

This case study is based on a Lurgi methanol process (Figure The hot stream of the synthesis gas is cooled in an E107
3). The retrofitted process can be operated within existing boiler, in E109, E110, and E111 heat exchangers in an EA101
parameters. air cooler, and in an E112 water cooler. The condensate is

Figure 3. Process flow diagram of a low-pressure Lurgi methanol plant.


Figure 4. Simplified flow sheet of the methanol plant using a gas turbine (additional units are shaded).
expanded in flashes: F1, F2, F107, and F108. The synthesis
gas is compressed in G201I and G201II two-stage compressors.
In the second subsystem, methanol is produced by the
catalytic hydrogenation of carbon monoxide and/or carbon
dioxide in a REA-2 reactor with three main reactions (r ) R6,
R7, R8):
CO + 2H2 / CH3OH ∆rH298 ) -90.77 kJ/mol
CO2 + 3H2 / CH3OH + H2O ∆rH298 ) -49.58 kJ/mol
(R7)
CO2 + H2 / CO + H2O ∆rH298 ) 41.19 kJ/mol (R8)
The high-pressure reactor REA-2 is operated within the
existing parameters, and unconverted gas is recycled. The high
recycle ratio and operating pressure of the reactor are exploited
to produce electricity, using a gas turbine (TUR) placed
downstream of the reactor. The turbine uses REA-2 outlet gas
as a working fluid. The inlet stream of the reactor is heated by
a process stream (HEPR) or by high-pressure steam (HEST) or
a combination of both. The stream leaving the turbine is cooled
using air (HEA) and water (HEW) coolers before entering the
Ind. Eng. Chem. Res., Vol. 46, No. 21, 2007 6957
flash (SEP). The liquid stream of the separation is the product,
and the recycled gas stream is compressed to 51 bar in a new,
two-stage compressor (COMP1, 2) with intermediate water
cooling (HEW1). The purge gas is separated from the crude
methanol in the flash F301.
The high-pressure purge gases from separators SEP and F301
can be mixed, and then CO2 can be separated, before the gas is
(R6) burned in the furnace, by using membrane or absorption or
adsorption system. CO2 can be separated from the flue gas, too,
but at a higher cost. CO2 separation from high-pressure purge
gas is much cheaper. The producer has to make a choice. The
producer can use the existing, inactive PSA column for H2
separation for additional H2 flow rate or use another adsorbent
such as mesoporous molecular sieve MCM-41 or zeolite 13X
for CO2 separation.
In the retrofitted methanol process, the recycled CO2 can be
reused as the reactant, and CO2 emission can be reduced. The
recycled CO2 reduces the CO2 flow rate in the purge gas before
the gas is burned, and, therefore, a lower CO2 concentration
remains in the flue gas. The CO2 in the flue gas can also be
separated. The methanol process parameters are optimized using
a nonlinear programming (NLP) model.4 The parameters in the
6958 Ind. Eng. Chem. Res., Vol. 46, No. 21, 2007

retrofitted model of the heat exchanger networks, flashes, Additional equations for the mass balance of the uth separa-
compressors, mixer, splitters, reactors,6 and turbine8 were tion (u ) F1, F2, F107, F108, SEP, and F301; u ) 1, ..., U) for
simultaneously optimized using the GAMS/MINOS.4 This NLP all the components (s ) 1, ..., S) are as follows:
can be solved with a large-scale reduced gradient method (e.g.
u ) Fu + Fοut,l (for u ) 1, ..., U)
οut,v
MINOS). The model is nonconvex; it does not guarantee a Fin u (3.9)
global optimization solution but quickly gives good results for
u ‚xu,s ) Fu ‚xu,s + Fοut,l
u ‚xu,s
οut,v οut,v οut,l
nontrivial, complex processes. The NLP model contains vari- Fin in

ables of all the process parameters: molar heat capacities,

(for u ) 1, ..., U; s ) 1, ..., S) (3.10)
material flow rates, heat flow rates, and temperatures, which
are limited by real constraints. The NLP model has variable S
heat capacity flow rate for all the streams, and the structure
can also be varied by using them. The NLP model contains
∑s xu,sοut,v ) 1 (for u ) 1 ..., U) (3.11)
equations for structural and parametric optimization. The NLP
model is formulated using a simplified process superstructure S

without binary parameters between heat and power integration ∑s xu,sοut,l ) 1 (for u ) 1 ..., U) (3.12)
and additional CO2 separation from purge gas and flue gas.
The structure of process units models in Figure 4 presents
the synthesis gas reactor REA-1 and the crude methanol reactor Ku,s ) du,s + cu,sTi,n
οut
+ bu,s(Ti,n
οut 2
)
REA-2. We have considered all the reactions r (where r ) 1, ..., (for u ) 1 ..., U; s ) 1, ..., S) (3.13)
R) of the synthesis gas and the crude methanol production,21,22
including those equations for the equilibrium constants of οut,v
xu,s ) Ku,s‚ xu,s
οut,l
(for u ) 1, ..., U; s ) 1, ..., S) (3.14)
reactions r (eqs 3.1 and 3.2) and the equilibrium conversions
of components s (CO2, CO, H2, CH4, C2H6, C3H8, C4H10, H2O, The inlet amount flow rate for the uth separation is the sum
CH3OH, CH3OCH3, CH3CH2CH2OH; where s ) 1, ..., S) (eq of the outlet amount flow rates of the vapor and liquid phases
3.3).21 The equilibrium constant Kr depends on the temperature (eq 9). Equation 10 includes the amount of flow fractions. The
(eqs 3.2 and 3.4): equilibrium constant of the sth component in the uth separation
is a function of temperature.
∑ (ys)νs ) pν‚Kr (for r ) 1, ..., R) (3.1) The temperatures at the outlets of the compressors CPc
out
(T CPc ; c ) G201I, G201II, COMP1, and COMP2) depend on
in
Kr ) e(-∆Gr/RT) (3.2) the inlet temperatures (TCPc ), and they can be calculated by the
equation
dns
) dωr (for s ) 1, ..., S; r ) 1, ..., R) (3.3) out
TCPc ) aCPc + bCPcTCPc
in
ys
(for c ) G201I, G201II, COMP1, and COMP2) (3.15)

[
∆Gr ) Jr - RT ∆ar ln T + ∆br
T
()
2
+ ∆cr
T2
6 () ∆dr
+ 2 + Ir
2T ] The mass balance of the PSA column is simplified:

Fin,p ) Fοut,p,CO2 + Fοut,p

(for r ) 1, ..., R) (3.4) (3.16)

This model was extended by additional equations regarding Fοut,p,CO2 ) FCO2‚ηPSA (3.17)
heat balances for the retrofit of heat exchangers:
The inlet stream (Fin,p) is the sum of the outlet streams of
in
(Ti,n - οut
Ti,n )‚CFi,n ) Φn (3.5) the pure CO2 component (Fout,p,CO2), and the outlet stream of
the remaining purge gas (Fout,p). FCO2 is the flow rate of CO2 in
for the purge gas. PSA uses an adsorber packed with 50% efficiency
(ηPSA ) 0.5). Fout,p,CO2 can be varied from 0 to 640 kg/h for
n ) E107, E109, E110, E111, EA101, E112, E201, purge gas and 0-950 kg/h for flue gas.
HEPR, HEA, and HEW∈ N The designed medium pressure of the turbine can be varied.
Its power (Ptur) is a function of the outlet (Ttur,out) temperature,
in
(Ti,n - Tj,n
οut
) - (Ti,n
οut
- Tj,n
in
) molar heat capacity (Cm), and the amount of the flow rate (F;
∆lnTn ) (3.6) eq 3.18). The inlet temperature (Ttur,in) is constant:
in
ln[(Ti,n - Tj,n
οut οut
)/(Ti,n - Tj,n
in
)]
Ptur ) Cm(Ttur,in - Ttur,out)Fηturηgen (3.18)
Φn
An ) n∈N (3.7)
∆lnTn‚Un The thermodynamical efficiency of the medium-pressure turbine
(ηtur) and the mechanical efficiency of the generator (ηgen) are
n gAn -
Aadd ∑ An,ex
n∈N
n∈N (3.8) supposed to be 85% each.
A mathematical model is applied, including integration of
heat flows, generation of electricity, CO2 reuse, increased
Equation 3.5 denotes the heat flow rate of the nth heat production, and decreased CO2 emissions. The flow rate of CO2
exchanger (the same is true for the equations of new heat can be varied, and it can affect the optimal process structure
exchanger m; m ) HEST and HEW1). Equation 3.6 calculates and production rate.
the log-mean temperature difference. Equation 3.7 represents Simultaneous optimization can generate additional annual
the heat exchange area of the nth heat exchanger. Equation 3.8 profit, which is strongly influenced by the flow rate of CO2 in
selects any additional area of the nth heat exchanger. the process. The CO2 flow rate depends on the following: (1)
 Ind. Eng. Chem. Res., Vol. 46, No. 21, 2007 6959

Table 3. Cost Items for the Example Processa

cost item expression or valuea ref
installed cost of heat exchanger (euros) (8600 + 670A0.83)× 3.5 × 2 b Tjoe and Linnhoff12
depreciation of compressor, Ccom (euros) 2605P0.82 c Biegler et al.3
depreciation of gas turbine, Ctur (euros/a) (22 946 + 13.5Ptur) × 4 c Biegler et al.3
price of methanol, CMd 115.0 euros/t
cost and taxes of CO2 emissions, Ctax 22 euros/t Axelsson et al.2
cost of electricity, Cel 435.4 euros/(kW a) Swaney14
cost of CO2 purification, CCO2p 78 euros/t CO2
cost of CO2 purification, CCO2f 210 euros/t CO2
cost of H2 purification in existing PSA column and inlet injection in recycle CH2 0.1 euros/kg
price of 37 bar steam (C37) () 106.3 euros/(kW a) Swaney14
a The published cost equations for the equipment are adjusted to the real, higher industrial costs, by a multiplier of 2 or 4. b A ) area, given in units of

m2. c P ) power, given in units of kW. d 10-yr average.

Figure 5. Simplified flow sheet of the methanol plant with additional CO2 flow rate from purge and flue gases using a gas turbine (additional units are
shaded).

the additional flow rate of H2 before the two-stage compressor
(G201I-II), which is separated from the purge gas by an existing
(nonactive) PSA column,8 which varies from 0 to 488 kg/h, (2)
the additional flow rate of CO2 recycled from the purge gas
(by high pressure) and flue gas (by a new compressor with
power PCOM,f) by using a new adsorption PSA system and the
existing one with a new adsorbent (at 26 bar and 35 °C), (3)
the decreased raw material (the high-pressure steam) mass flow
rate, from 33.1 t/h to 32 t/h in the reactor REA-1, (4) separation
(SEP) and reaction systems (REA-1, REA-2), (5) increased
methanol production, (6) electricity cogeneration, (7) heat
integration of enthalpy flows, (8) energy savings in HEPR, and
(9) the fraction of the exhaust flow rate.
The CO2 flow rate depends very much on the degrees of
conversion in the reactors REA-1 and REA-2 and catalyst
selectivity in REA-1. The NLP model, used to increase
6960 Ind. Eng. Chem. Res., Vol. 46, No. 21, 2007

cogeneration and product production as well as to optimize the Table 4. Comparsions between Case and Optimized Streams
flow rate of CO2 and reduce CO2 emissions,6 includes the parameter case stream optimized stream
following: (1) amount and enthalpy balances of process units, Raw Material-Natural Gas
(2) equations for degrees of conversion, (3) process constraints mass flow rate, q (kg/h)
(amount flow rates, temperatures, and pressures), (4) retrofit CH4 10 214 10 214
equations for reusing existing process units (heat exchangers), C2H6 86 86
(6) changing the raw material inlet of high-pressure steam in C3H8 38 38
C4H10 16 16
the reformer REA-1, (7) additional flow rate of H2, which is CO 0 0
separated from purge gas by using an existing PSA column,8 CO2 25 25
(8) injecting the additional flow rate of CO2, which is separated H2 0 0
from the purge gases (from the separators SEP and F301), and temperature °C) 0 0
pressure (bar) 24.5 24.5
flue gas by using adsorption systems, (9) decreasing CO2
emissions, and (10) objective function. After Reactor REA-1
mass flow rate, q (kg/h)
The model contains equations for structural and parametric CO 10 846 12 517
optimization, together with process operating constrains. The CO2 6986 8500
objective function (eq 3.19) of the NLP model is to maximize H2 3606 4100
the annual profit and includes the cost of CO2 reuse (CCO2p) CH4 1616 1555
temperature °C) 825 825
from purge gas and flue gas (CCO2f), and taxes for CO2 emissions
pressure (bar) 15 15
(Ctax; Table 3):
After Compressor G201II
Maximal additional annual profit for retrofit: mass flow rate, q (kg/h)
CO 10 846 12 216
Vmax ) CelPturηturηgen + CM∆FM + C37∆Fsteam - CO2 6986 8189
CtaxFCM,CO2 - C37ΦHEST - (22 946 + 13.5Ptur)4 - H2 3606 4089
CH4 1616 1726
[2605PCOMP10.82- 2605PCOMP20.82 - ∑
m
(8600 + H2O
temperature °C)
136
112
90
112

∑n 670∆AHE,n0.83 × 3.5 × 2]r -

pressure (bar) 51 51
670AHE,m0.83) × 3.5 × 2 -
After Reactor REA-2
mass flow rate, q (kg/h)
CCO2pFCO2p - CCO2fFCO2f - 2605PCOM,f 0.82
r - CH2FH2 CO 8169 11 800
CO2 13 447 15 700
(3.19) H2 22 874 34 600
CH4 30 526 41 400
for CH3OH 18 330 23 000
temperature °C) 250 250
n ) E107, E109, E110, E111, EA101, E112, E201, pressure (bar) 49 49
HEPR, HEA, and HEW Crude Methanol
mass flow rate, q (kg/h)
and CO 20 12
CO2 350 100
m ) HEST and HEW1 H2 4 3
CH4 140 100
The inlet stream of the reactor REA-2 can be heated by a process CH3OH 16 031 18 350
stream in the heat exchanger (HEPR) and/or by high-pressure temperature °C) 40 37
steam (HEST) using varying heat flow rates. pressure (bar) 40 37
In the model, the existing areas can be used (AHE,ex), enlarging Stream SEP-Gas
them by additional areas (∆AHE,add) if necessary. Additional mass flow rate, q (kg/h)
CO 7672 11 217
annual depreciation of the enlarged and new areas (AHE,new) of CO2 12 149 12 060
the heat exchangers and compressors (Table 3) is multiplied H2 20 828 34 211
by the payback multiplier (r ) 0.216)1 and corrected for CH4 23 567 40 819
inflation. CH3OH 2188 4048
temperature °C) 40 37
A retrofitted methanol process (Figure 5) with a maximum pressure (bar) 40 37
additional CO2 flow rate of 640 kg/h from the purge gas to be
Purge Gas after Splitter S1
reused, a maximum additional CO2 flow rate of 950 kg/h from mass flow rate, q (kg/h)
flue gas, electricity generation using gas turbine pressure drop CO 481 673
from 49.7 to 37 bar, and an outlet temperature, Ttur,out ) 110 CO2 761 723
°C, was selected as a starting flowsheet. H2 1306 2052
CH4 1477 2449
The flow rate of CO2 is separated from the purge gas and
CH3OH 137 242
injected into the G201I compressor. The existing PSA column temperature °C) 40 37
for the purification of H2 is not used, and additional H2 is pressure (bar) 40 37
unnecessary because of the new catalyst selectivity of the reactor
REA-1. A PSA with a new adsorbent can be used for CO2 This structure enables 20 MW of electricity power to be
separation from purge gas and CO2 separation from flue gas generated. The steam exchanger (HEST) needs 25 MW of heat
by the new adsorption system. CO2 separation from purge gas flow rate. The integrated process stream exchanges 6 MW of
(Table 4) and flue gas (Table 1) reduced the local CO2 heat flow rate in HEPR. The powers of the first and the second
emissions. The total additional annual methanol production compressor stage are 3.4 MW and 4.7 MW, respectively. The
(including all the effects of the additional flow rates of CO2 HEW1 is supposed to exchange 3.6 MW, and the coolers HEW
and reduced flow rate of steam) is estimated to be 20 mol/s. and HEA are supposed to exchange 6 MW and 8.2 MW of

heat flow rate, respectively. The purge gas outlet flow rate
fraction is decreased from 5.9% to 5.4%. The existing coolers
of the synthesis gas (E107, E109, E110, E111, EA101, E112,
and E201) need not be enlarged. The additional annual
depreciation of the gas turbine, new heat exchangers (HEST,
HEW1, having 942 m2 and 324 m2 of area, respectively) and
the new two-stage compressor, is estimated to be 3.04 MEUR/
a. The cost of the high-pressure steam used in HEST will be
2.64 MEUR/a, and the CO2 purification cost from purge gas in
the existing PSA column and the inlet injection cost in the
recycle are estimated to be 0.4 MEUR/a. The CO2 purification
cost from flue gas will be 1.6 MEUR/a and 0.5 MEUR/a for
compressor depreciation. In the depreciation account for retrofit
we have included additional costs to the new units only: 32
kEUR/a for the instrumentation cost (which is estimated to be
15% of the additional plant direct cost19), 12 kEUR/a for the
contingency (estimated at 5% of the additional plant direct
cost19), 5 kEUR/a for the maintenance cost (estimated as 2%
of the additional plant direct cost), and 21 kEUR/a for the turbine
down time (estimated as 5% of the additional plant direct cost).
The annual income from the additional production of electric-
ity will be 8.7 MEUR/a, and that of the methanol 2 MEUR/a.
The steam flow rate can be reduced by 9192 t/a with additional
annual savings of 0.06 MEUR/a. The emission tax is reduced
by 0.28 MEUR/a. The additional profit of the process optimiza-
tion including cogeneration and additional methanol production
is estimated to be 2.79 MEUR/a, with a payback time of about
tPB ) 2 a. In Table 4 the stream data for the case and the
optimized process are compared.
4. Conclusions
This paper presents an efficient use of NLP model formula-
tions for simultaneous cogeneration of electricity using gas
turbine and process heat integration, with CO2 reuse as reactant.
The NLP model is formed using a simplified process super-
structure, and it is a very effective tool for the solution of
complex process retrofit problems. This simultaneous math-
ematical optimization method is giving a near optimum solution
but not a global one. CO2 emissions can be reduced at the
source. The simultaneously optimized NLP model of the low-
pressure Lurgi methanol plant includes the effects of raw
material flow rate, degrees of conversion in the reactors,
temperature efficiency of the separator, the fraction of exhaust
flow rate, energy savings, and CO2 reuse to decrease the CO2
emissions in the purge and flue gases. Higher equilibrium
conversion in both reactors can increase the conversion of
methanol by 14.4%, producing 18 000 t/a of additional methanol.
We have carried out simultaneous heat, power, product, and
emissions optimization with a potential additional profit of 2.79
MEUR/a. Improved process integration and cogeneration can
decrease environmental emission by 12 800 t/a.
Nomenclature
NLP ) nonlinear programming
PSA ) pressure swing adsorption
Indices
c ) compressor
i ) hot process or utility stream
j ) cold process or utility stream
l ) liquid phase
m ) new heat exchanger
n ) existing heat exchanger
Ind. Eng. Chem. Res., Vol. 46, No. 21, 2007 6961
r ) reaction
s ) component
u ) separator
v ) vapor phase
Sets
N ) {n| n is an existing heat exchanger}
R ) {r| r is a reaction}
S ) {s| s is a component}
Parameters
aCP ) temperature constant for polytropic compression
An,ex ) area of an existing heat exchanger, m2
bCP ) temperature constant for polytropic compression
bu ) equilibrium constant in separation u
cr ) heat capacity constant of reaction r
cu ) equilibrium constant in separation u
dr ) heat capacity constant of reaction r
du ) equilibrium constant in separation u
Ir ) integration constant of reaction r
Jr ) integration constant of reaction r
Ku,s ) equilibrium constant of component s in separation u
r ) pay back multiplier
U ) overall heat transfer coefficient, W/(m2 K)
Variables
n ) additional area of heat exchanger n, m
Aadd 2
An ) total area of existing heat exchanger n, m2
CFi,n ) heat capacity flow rate of hot stream i in heat exchanger
n, W/K
Cm ) molar heat capacity, J/(mol K)
G ) Gibbs energy, J
F ) amount flow rate, mol/s
Kr ) equilibrium constant of reaction r
p ) pressure, bar
T ) temperature, K
x ) amount fraction
ν ) stoichiometric coefficient
ωr ) equilibrium conversion, mol/s
∆lnTn ) log-mean temperature difference of heat exchanger n,
K
Φ ) heat flow rate, W
Literature Cited
(1) Ahmad, S. Heat exchanger networks: Cost tradeoffs in energy and
capital, Ph.D. Thesis, University of Manchester, Manchester, U.K., 1985;
pp 113-306.
(2) Axelsson, H.; Harvey, S.; Asblad A.; Berntsson, T. Potential for
greenhouse gas reduction in industry through increased heat recovery and/
or integration of combined heat and power. Appl. Therm. Eng. 2003, 23,
65-87.
(3) Biegler, L. T.; Grossmann, I. E.; Westerberg, A. W. Systematic
methods of chemical process design; Prentice Hall: Upper Saddle River,
NJ, 1997; pp 1-408.
(4) Brooke, A.; Kendrick, D.; Meeraus, A. GAMS: A User’s Guide;
Scientific Press: Palo Alto, CA, 1992.
(5) Davis, M. L.; Cornwell, D. A. Introduction to enVironmental
engineering, 2nd ed.; McGraw-Hill, Inc.: New York, 1991.
(6) Kralj, A. Kovač; Glavič, P.; Kravanja, Z. Retrofit of complex and
energy intensive processes II: stepwise simultaneous superstructural
approach. Comput. Chem. Eng. 2000, 24 (1), 125-138.
(7) Ida, J.; Lin, Y. S. Mechanism of high-temperature CO2 sorption on
lithium zirconate. EnViron. Sci. Technol. 2003, 37 (9), 1999-2004.
(8) Kralj, A. Kovač; Glavič, P. Additional flow rate of H2 in methanol
process optimization using NLP model. Proceedings of ECOS’2005,
Trondheim, Norway, 2005; pp 547-552.
6962 Ind. Eng. Chem. Res., Vol. 46, No. 21, 2007
(9) Noppadon, S.; Hongyun, Y.; Donald, R. C.; Yong, L.; Bruce, J. T.
Adsorptive filtration of carbon dioxide from wet gases utilizing microfibrous
filter media entrapped K2CO3. http://aiche.confex.com (accessed month,
2005).
(10) Zhikang, X. Separation and fixation of carbon dioxide using
polymeric membrane contactor; Institute of Polymer Science, Zhejiang
University: China, 2006.
(11) Linnhoff, B.; Townsed, D. W.; Boland, P.; Hewitt, G. F.; Thomas,
B. E. A.; Guy, A. R.; Marsland, R. H. A User Guide on Process Integration
for the Efficient Use of Energy. Chem. Eng. Sci. 1982, 39 (3), 619-620.
(12) Tjoe, T. N.; Linnhoff, B. Using pinch tehnology for process retrofit.
Chem. Eng. 1986, 28, 47-60.
(13) Shivakumar, K.; Narasimhan, S. A robust and efficient NLP
formulation using graph theoretic principles for synthesis of heat exchanger
networks. Comput. Chem. Eng. 2002, 26, 1517-1532.
(14) Swaney, R. Thermal integration of processes with heat engines and
heat pumps. AIChE J. 1989, 35 (6), 1010.
(15) McNulty, B. WE energies to host innovative carbon separation
project. www.we-energies.com (accessed month 2006).
(16) Jones, A. J. Membrane and separation technology: The Australian
perspectiVe; Australian Government Publishing Service: Canberra, Aus-
tralia, 1987.
(17) Wenten, I. G. Recent development in membrane science and its
industrial aplications. Membr. Sci. Technol. 2002, 24 (suppl).
(18) Aspen Technology. ASPEN PLUS User Manual Release 11.1;
Aspen Technology, Inc.: Cambridge, MA, 2002.
(19) Valle-Riestra, J. F. Project eValuation in the chemical process
industries; McGraw-Hill, Inc.: New York, 1983; p 262.
(20) http://en.wikipedia.org/ (accessed month year).
(21) McKetta, J. J.; Cunningham, W. A. Encyclopedia of chemical
processing and design; Marcel Dekker: New York, 1985; Vol. 29, pp 418-
474.
(22) Smith, J. M.; Van Ness, H. C. Introduction to chemical engineering
thermodynamics; McGraw-Hill, Inc.: New York, 1987; pp 496-518.
ReceiVed for reView February 7, 2007
ReVised manuscript receiVed June 29, 2007
Accepted July 10, 2007
IE0702173