Supporting Information

Process simulation and techno-economic

analysis of the production of sodium
methoxide

José F.O. Granjo∗ and Nuno M.C. Oliveira

GEPSI–PSE Group, CIEPQPF, Department of Chemical Engineering, University of

Coimbra, Rua Sílvio Lima – Pólo II, 3030–790 Coimbra, Portugal

E-mail: josegranjo@eq.uc.pt
Phone: +351 239 798 793. Fax: +351 239 798 703

∗
To whom correspondence should be addressed

S1
A Phase and chemical equilibrium modeling and param-

eter estimation

A.1 Single electrolyte speciation modeling

The individual chemical potential of each electrolyte (µca ) is calculated from its ionic species,
according with the chemical equilibrium. Moreover, the electrolytes activities in these sys-
tems are, for convenience, commonly reported in a molality scale, i.e., a(m) = (m± γ±m )ν .
Eq. (S1) shows the relations between the mean activity coefficient in molal and mole frac-
tion scales (respectively, γ±m and γ± ), molality (m± ), and mole fraction (x± ) of a single
electrolyte system 1 . Ms (g mol−1 ) represents the solvent molecular weight, ν the ionic stoi-
chiometric coefficient, and z the ionic charge number.

CA → νc C+zc + νa A−za (sol.)

µca = νc µ+z
c
c
+ νa µ−z
a
a

µca -µoca = RT ln (γ±m m± )ν

νa 1/ν
νc
, m± = (mνc c mνaa )1/ν


γ±m = γm,c γm,a
(S1)
−3
γ± = γ±m (1 + 10 · ν m Ms )

mc = νc m, ma = νa m, xc = νc x, xa = νa x

x = 1/(1 + 10−3 · ν m Ms ), x± = (xνc c xνaa )1/ν

ν = νc + νa

S2
A.2 Electrolyte Non-Random Two-Liquid model

The generalized eNRTL model expresses g E and the activity coefficient of each species using
a unsymmetrical convention (γi∗ ), as a result of the summation of three contributions 2,3 :

g E,∗ = g P DH,∗ + g Born,∗ + g lc,∗

(S2)
ln γi∗ = ln γiP DH,∗ + ln γiBorn,∗ + ln γilc,∗ , i = m, c, a

Here m, c and a represent molecular, cationic and anionic species, respectively.

The Pitzer-Debye-Hückel (PDH) term accounts for the long-range electrostatic forces
caused by the ionic species while the Born equation corrects the change of the reference
state from the mixed-solvent composition to the aqueous solution of the PDH model:
" #
2 1/2 3/2
 1/2 2

1000 2z z Ix − 2I x
ln γiP DH,∗ = − i
ln 1 + ρIx1/2 + i


Aφ 1/2
Ms ρ 1 + ρIx
 1/2  2 1/2
1 2πNA ds Qe
Aφ =
3 1000 εs kT (S3)
2
  2
Born,∗ Qe 1 1 zi
ln γi = − 10−2 , i = c, a
2kT εs εw ri
1X
Ix = xi zi2
2 i

Here

X 1 X x0 X
Ms = x0S MS , = S
, εs = wS0 εS
S
ds S
dS S
xS MS xS
x0S = P , wS0 = P
S xS S MS xS (S4)
NA = 6.022 × 1023 mol−1 , Qe = 4.803 × 10−10 e.s.u.

k = 1.381 × 10−16 erg K molecule−1 , ρ = 14.9

and Aφ is the Debye-Hückel parameter, NA the Avogadro’s constant, Qe is the electron

charge, εw the dielectric constant of water, k Boltzmann’s constant, Ix ionic strength, ri is

S3
Born’s ion radius (Å), and ρ is the closest approach parameter.
The molecular weight MS , dielectric constant εS , mole fraction xS , solutes’ free mole (x0S )
and mass (wS0 ) fraction of a solvent component S are introduced in mixing rules (Eq. (S4))
to calculate the molecular weight Ms , density ds (g cm−3 ), and εs the dielectric constant of
the mixed solvent system.
The contribution of short-range interaction forces to the activity coefficients for molecular
(γm
lc
), cationic (γclc ) and anionic(γalc ) species are calculated as follows:
P
j Xj Gjm τjm
X Xm0 Gmm0  P 
lc k Xk Gkm0 τkm0
ln γm = P + P τmm0 − P +
k Xk Gkm m0 k Xk Gkm0 k Xk Gkm0
X X Ya Xc Gmc,ac  P 
Xk Gkc,ac τkc,ac
P τmc,ac − P k
(S5)
c a k X k Gkc,ac k Xk Gkc,ac
 P 
X X Yc Xa Gma,ca
k Xk Gka,ca τka,ca
+ P τmc,ca − P , j, k = m, c, a
a c k X k Gka,ca k X k Gka,ca

1 X P Xk Gkc,ac τkc,ac X Xm Gcm  P

Xk Gkm τkm

lc k k
ln γc = Ya P + P τcm − P
zc a k X k Gkc,ac m k X k G km k Xk Gkm
 P  (S6)
X X Yc0 Xa Gca,c0 a Xk Gka,c0 a τka,c0 a
+ P τca,c0 a − kP , k = m, c, a
a c0 k Xk Gka,c0 a k Xk Gka,c0 a

1 X P Xk Gka,ca τka,ca X Xm Gam  P

Xk Gkm τkm

lc k k
ln γa = Yc P + P τam − P
za c k Xk Gka,ca m k Xk Gkm k Xk Gkm
 P  (S7)
X X Ya0 Xc Gac,a0 c Xk Gkc,a0 c τkc,a0 c
+ P τac,a0 c − kP , k = m, c, a
c a0 k X k Gkc,a0c
k X k G kc,a0c

The modified mole fractions Xk are defined as Xk = Ck xk , where Ck = zk for ionic

species and Ck = 1 for molecular components. The activity coefficients are then normalized

S4
for aqueous infinite dilution:

ln γilc,∗ = ln γilc − ln γilc,∞

lc,∞
ln γm = τwm + Gmw τmw
1 X (S8)
ln γclc,∞ = Ya τwc,ac + Gcw τcw
zc a
1 X
ln γalc,∞ = Yc τwa,ca + Gaw τaw
za c

where subscript w represents water. The terms Ya , Yc , Gcm , Gam , τmc,ac , τma,ca , αcm , αam ,
τcm , τam , Gmc,ac , and Gma,ca are given by the following mixing rules:

Xa Xc
Ya = P , Yc = P
a0 Xa0 c0 Xc0
X X
Gcm = Ya Gca,m , Gam = Yc Gca,m
a c
X X
αma = αam = Yc αm,ca , αmc = αcm = Ya αm,ca
c a
αca,m αca,m (S9)
τmc,ac = τcm − (τca,m − τm,ca ) , τma,ca = τam − (τca,m − τm,ca )
αmc,ac αma,ca
ln Gcm ln Gam
τcm = − , τam = −
αcm αam
Gmc,ac = exp (−αmc,ac τmc,ac ) = exp (−αcm τmc,ac )

Gma,ca = exp (−αma,ca τma,ca ) = exp (−αam τma,ca )

This model has three groups of adjustable binary parameters:

• Molecule - molecule:
τmm0 τm0 m , αmm0 = αm0 m

• Ion-pair - molecule:
τca,m , τm,ca , αm,ca = αca,m

S5
 • Ion-pair – ion-pair:

τca,ca0 , τca0 ,ca , τca,c0 a , τc0 a,ca

αca,ca0 = αca0 ,ca , αca,c0 a = αc0 a,ca

Similarly to the NRTL model, the non-randomness parameters αmm0 , αca,m and αca,c0 a are
assumed to be fixed and only the τ ’s are adjusted to the data. The binary interaction
parameters τ can also be expressed as function of temperature 4 :

τ = a + b/T + c [(298.15 − T ) /T + ln T /298.15] (S11)

A.3 eNRTL parameter estimation

The expressions for the activity coefficients γ ∗ and γ ∞ presented in Section A.2 are com-
pactly represented as eNRTL(·). For the binary system NaOCH3 + CH3OH, γ ∗ and γ ∞
are computed for two molecular species (methanol and water), one anion (OCH3–), and one
cation (Na+). The non-randomness parameter αNa+OCH3–,CH3OH was set to 0.2. Because the
reference state of the eNRTL model is an aqueous infinite dilution, water must be present in
the electrolyte system to compute the transformation of the reference state of ions and molec-
ular species 4 . Therefore, a water mole fraction of 1 × 10−8 is assumed in all experimental
data points.
The eNRTL binary interaction parameters (BIPs) were estimated using a least squares
formulation (S12), minimizing the residues between the experimental and model predicted
symmetric activity coefficients of methanol (Table S1), subjected to the eNRTL model equa-
tions, lower (LB) and upper bounds (U B) for the model parameters θ. To handle the
non-convexity of this nonlinear parameter estimation problem, which usually leads to multi-
ple local optimal solutions, this problem was implemented in GAMS 5 due to the robustness
of the numerical solvers available in this system. Scaling, algebraic rearrangements and new

S6
Table S1: Experimental (âCH3OH ) and predicted activities of methanol (aCH3OH ) in a mixture
with NaOCH3 at 25 ◦C.

Mole fractiona Molaritya Exp. activitya Calc. activityb

CH3OH(x) NaOCH3 (M) âCH3OH aCH3OH
0.768 6.14 0.291 0.290
0.789 5.58 0.335 0.336
0.801 5.27 0.363 0.365
0.808 5.06 0.386 0.383
0.824 4.65 0.425 0.427
0.842 4.17 0.480 0.481
0.856 3.79 0.529 0.528
%ARD 0.41
a
Reproduced from Freeguard et al. 6 with kind permission of
John Wiley & Sons, inc.
b
Ionic radius 7 : Na+, 1.02 Å; OCH3–,1.54 Å.

terms introduction on the original model equations were done to obtain a resulting model
with decreased nonlinearity. A subset of fundamental model variables was identified, and
proper initialization and bounds were established for them. These values were then propa-
gated throughout the model to initialize and bound the remaining variables, allowing both
a more consistent global model initialization, and avoiding the non-convergence of the NLP
due to the inability of locating feasible solutions. This step was verified to be an essential
task to guarantee the convergence of the parameter estimation.
The MSNLP solver 8 was used in this study. It is a global solver based on various heuristic
multistart algorithms, designed to rapidly find various local optima of smooth constrained
NLPs. The NLP solver used by MSNLP to drive the problem to local optimality was
CONOPT 9 . Parameter estimation with MSNLP was performed on a current Intel Xeon
Linux workstation, allowing a maximum of 300 s of CPU for the runs. For the current data,
two distinct local solutions were identified, with ϕ = 1.0775 × 10−5 and ϕ = 1.0877 × 10−5 .
The availability of a multistart optimizer helped to largely mitigate the effects of the initial
estimates on the quality of the solutions obtained. The values of the eNRTL model BIPs

S7
are presented in Table S2.

np
X

2
min ϕ= ∗
γ̂CH3OH,i − γCH 3OH,i
∞
+ γCH 3OH,i
(S12a)
z
i=1
∗
s.t. γCH 3OH,i
∞
, γCH 3OH,i
= eNRTL(·|θ)i (S12b)

LB ≤ θ ≤ U B (S12c)

γ ∗ , γ ∞ ∈ (R+
0)
3×np
, θ ∈ R2 , i = {1, 2, . . . , np } (S12d)

Table S2: eNRTL BIP values 4 .

Molecule or ion pair i H2Oa,b H2Oa,c CH3OHa,d

Na+ Na+
Molecule or ion pair j CH3OH
OH– OCH3–
aij 38.925 2.732 2.297
aji -20.598 −0.6930 −5.352
bij 4734.3 −617.3 0
bji -1917.5 173.0 0
cij 15.318 0 0
cji -1.874 0 0
αij 0.03 0.3 0.2
a
Parameter values retrieved from ASPEN Plus® Phys-
ical Property Databanks, Version 7.2. Reproduced
with kind permission of Aspen Technology, inc.
b
τij = aij + bij /T + cij [(298.15 − T ) /T + ln T /298.15].
c
τij = aij + bij /T + cij ln T + dij T .
d
Regressed from data.

A.4 Equilibrium constant data

The following values of Khm,c for temperatures between 0–100 ◦C in Table S3 were determined
from the pKa and pKw values.

S8
 Table S3: Values of Khm,c at various temperatures.

Temperature/◦C 0 15 25 40 50 75 100
Khm,c 7.7 5.5 4.5 3.4 2.8 1.9 1.4
a
Reproduced from Murto 10 , with kind permission of
John Wiley & Sons, inc.

S9
B Detailed simulation results of Process II at base case

conditions

Table S4: Complete table results of Process II at base case conditions.

Stream 501 502 504 H2 METHANOL NA NAOCH3

Temperature (◦C) 98.0 31.7 83.0 83.0 30.0 30.0 83.0
Pressure (bar) 1.01 14.27 14.27 14.27 1.01 1.01 14.27
Vapor fraction 0 0 0.083 1 0 0 0
Mole flow (kmol h ) 12.54 76.10 82.39 6.838
−1
76.10 12.54 75.55
Mass flow (kg h−1 ) 288.2 2438 2727 37.89 2438 288.2 2689
Component Mass Flow (kg h−1 )
CH3OH 0 2438 2038 25.73 2438 0 2012
NaOCH3 0 0 675.0 0 0 0 675.0
Na 288.2 0 0.968 0 0 288.2 0.968
H2 0 0 12.59 12.17 0 0 0.427
Mass fraction
CH3OH 0 1 0.747 0.679 1 0 0.748
NaOCH3 0 0 0.248 0 0 0 0.251
Na 1 0 0 0 0 1 360 ppm
H2 0 0 0.005 0.321 0 0 159 ppm

S10
 507

501 502
METHANOL D-502

H-501 506
P-501

WASTE
503 504
NAOH
H-502
D-501
P-502 505

30NAOCH3

Figure S2: Flowsheet of example 1 from Guth et al. 11 , as simulated in ASPEN Plus® .

C Model validation of Process III

C.1 Example 1 in Guth et al. 11

The example 1 given in Guth et al. 11 characterizes the operating conditions, stream mass
flows, number of theoretical stages of D-501 and D-502 and the mass reflux ratio in D-502 of
a pilot plant with a flowsheet similar to the one in Figure S2. In general, a good agreement is
observed between the simulation results and experimental data with typical error predictions
below 1 % in mass flow rates. The estimated overall reaction conversion of NaOH is 0.981,
1.7 % less than the 0.998 reported. The amounts of water and NaOH in the product are
the exception, being clearly overestimated: 40 mg h−1 compared to 4 mg h−1 , and 253 mg h−1
compared to 25 mg h−1 , respectively. However, these differences are within the margin of
experimental error, since they represent, respectively, 0.2 % and 1.7 % of the total water and
NaOH mass flows around D-501. The pilot plant data and simulation results are presented
in Table S5 and Table S6.

S11
Table S5: Comparison between pilot plant data and simulation results of example 1 in Guth
et al. 11 .

Pilot plant data Simulation results

Input variables
Caustic soda feed mass flow (g h−1 ) 27
Caustic soda feed temperature (◦C) 75
Methanol make-up (g h−1 ) 54
D-501
Pressure (bar) 1.01
NT 20
D-502
Pressure (bar) 1.01
NT 29
NF 25
RRmas 1.3
Results
D-501
TD 75 70
Distillate mass flow (g h−1 ) 498 498
Product mass flow (g h−1 ) 61 61
Methanol feed temperature (◦C) 61 61
Methanol feed mass flow (g h−1 ) 532 532
Water content in distillate (wt%) 3.93 3.90
Water content in product (ppm) 60 650
NaOH in product (ppm) 410 4119
NaOCH3 in product (wt%) 30 29
D-502
Distillate mass flow (g h−1 ) 478 478
Bottoms mass flow (g h−1 ) 20 20
Water content in distillate (ppm) 30 11
Methanol content in bottoms (wt%) 1 0.9

S12
 Table S6: Complete stream table results for example 1 in Guth et al. 11 .

Stream 501 502 503 504 505 506 507 WASTE METHANOL 30NAOCH3 NAOH

Temperature (◦C) 30.0 30.0 30.0 75.0 61.3 70.1 64.7 99.1 30.0 94.2 30.0
Pressure (bar) 1.01 1.01 1.01 1.01 1.01 1.01 1.01 1.01 1.01 1.01 1.01
Vapor fraction 0 0 0 0 0 1.0 0 0 0 0 0
−1
Mole flow (mol h ) 1.69 1.69 1.42 1.42 16.61 16.01 14.92 1.08 1.69 1.69 1.42
Mass flow (g h−1 ) 54.0 54.0 27.0 27.0 532.1 497.7 478.1 19.6 54.0 61.4 27.0
Component mass flow (g h−1 )

S13
H 2O 0 0 13.5 13.5 0 19.4 0 19.4 0 0 13.5
NaOH 0 0 13.5 13.5 0 0 0 0 0 0.3 13.5
NaOCH3 0 0 0 0 0 0 0 0 0 17.9 0
CH3OH 54.0 54.0 0 0 532.1 478.3 478.1 0.2 54.0 43.2 0
Mass fraction
H 2O 0 0 0.5 0.5 10 ppm 0.0390 11 ppm 0.9907 0 650 ppm 0.5
NaOH 0 0 0.5 0.5 0 0 0 0 0 4119 ppm 0.5
NaOCH3 0 0 0 0 0 0 0 0 0 0.292 0
CH3OH 1 1 0 0 0.99999 0.961 0.99999 0.009 1 0.704 0
C.2 Simulation results of the process in Shao and Wang 12

A second case study to evaluate the reliability of the model is retrieved from Shao and
Wang 12 . This process differs from the previous one in that anhydrous NaOH is mixed with
methanol and introduced at the top of the distillation column while the methanol vapor is
introduced at the bottom. Because only data for the reactive distillation column is provided,
only D-501 was simulated (Figure S3).
VAPORS

503 504

NAOH
H-502

D-501
P-502

501 502

METHANOL

H-501
NAOCH3
P-501

Figure S3: Flowsheet diagram in ASPEN Plus® of the process described in Shao and
Wang 12 .

Our results predict an overall conversion of NaOH is 0.935, compared to 0.948 obtained
from reported simulation results for similar operating conditions. Discrepancies are observed
in the column temperature profiles where 79–109 ◦C obtained here diverge from 88–122 ◦C
and 105–130 ◦C reported, respectively, from elsewhere’s simulations and industrial data 12 .
However, the data plant being reported is for a wide range of operating conditions, including
the operating pressure (1.4–2.2 bar) or methanol mass flow 1000–1700 kg h−1 , and therefore
no definitive conclusions can be made. The plant data and simulation results from Shao and
Wang 12 and here are presented in Table S7, while in Table S8 are the complete numerical
results for the base case operating conditions.
Additionally, the sensitivity analyses relative to three operating parameters performed in
Shao and Wang 12 were replicated here. Figure S4a shows that the effect of the bottom/top
feed mass ratio on the conversion of NaOH predicted here follows a similar trend of previ-

S14
Table S7: Comparison between simulation results and plant data in Shao and Wang 12 .

Plant data Simulation Simulation

Shao and Wang 12 (this work)
Inputs
NT 20
Pressure (bar) 1.4–2.2 1.5
Top feed mass flow (kg h−1 ) 350–500 360
CH3OH/NaOH ratio in top feed 3.5–4 3.5
Bottom feed mass flow (kg h−1 ) 1000–1700 1440
Water in bottom feed (wt%) ≤ 0.3 0.3
Bottom/top feed ratio 3–4 4
Results
Column temperature range (◦C) 105–130 88–122 79–109
Water in distillate (wt%) 2 2.3 2.4
NaOCH3 in product (wt%) 27.5–31 29.68 29.67
NaOH in product (wt%) 0.77 0.73 1.52
Product mass flow (kg h−1 ) 353 341

Table S8: Complete stream table results for the base case in Shao and Wang 12 .

Stream 501 502 503 504 METHANOL NAOCH3 NAOH VAPORS

Temperature (◦C) 25.1 75.5 25.1 75.0 25.0 108.5 25.0 79.2
Pressure (bar) 1.50 1.50 1.50 1.50 1.01 1.50 1.01 1.50
Vapor fraction 0 1 0 0 0 0 0 1
Mole flow (mol h ) 45.05
−1
45.05 12.74 12.74 45.05 9.500 12.74 46.41
Mass flow (g h−1 ) 1440 1440 360 360 1440 341 360 1459
Component mass flow (g h−1 )
H 2O 4.32 4.32 0 0 4.32 2.35 0 35.67
NaOH 0 0 80.00 80.00 0 5.19 80.00 0
NaOCH3 0 0 0 0 0 101.0 0 0
CH3OH 1436 1436 280.0 280.0 1436 232.0 280.0 1424
Mass fraction
H 2O 0.003 0.003 0 0 0.003 0.007 0 0.024
NaOH 0 0 0.222 0.222 0 0.015 0.222 0
NaOCH3 0 0 0 0 0 0.297 0 0
CH3OH 0.997 0.997 0.778 0.778 0.997 0.681 0.778 0.976

S15
ously reported values. As the ratio of methanol/water rises, the conversion increases since
more water is stripped out of the column moving the Reaction 5 towards the products side.
Similarly, Figure S4b describes the positive effect of increasing the top feed methanol/NaOH
mass ratio in the conversion of NaOH. The higher concentration of methanol in the liquid
phase inside D-501 and the lowering of the boiling point of the mixtures contribute to the
increase of the amount of NaOCH3 produced. Lastly, Figure S4c displays the negative effect
of the moisture level in the methanol vapor on the conversion of NaOH since the presence
of water pulls back the equilibrium to the reactants side.
Figure S4 shows that in general our predictions of the conversion of NaOH are less than
3.5 % apart from the ones presented elsewhere 12 . However, they are more conservative and
less sensitive to the operating parameters studied, which is explained by differences in Khm,c
values reported in Shao and Wang 12 and those in Murto 10 , the latter employed in this
work. For the temperature range 79–109 ◦C, the predicted value of Khm,c decreases from
1.8 to 1.2, compared to Khm,c values of 7.1 × 10−3 to 1.9 × 10−4 using the correlation in
Shao and Wang 12 . These lower values of Khm,c explain the higher variations observed for
the NaOH conversion with the amounts of water and methanol entering in the distillation
column. Moreover, Khm,c estimated with Eq. (4) varies less with temperature, suggesting
that the methoxylation reaction of sodium hydroxide in aqueous solution is less exothermic
than the heat of reaction suggested from correlation in Shao and Wang 12 (less negative
slope of ln Khm,c ). The variation of Khm,c with temperature in Murto 10 is consistent with the
calorimetric data existing in Chandran et al. 13 , where ∆Hrx
−◦
=−18.54 kJ mol−1 . In Tables S9–
S11 are the complete numerical results of the sensitivity analyses performed in Shao and
Wang 12 and replicated in this work.

S16
 0.98

0.96

0.94

NaOH Conversion
0.92

0.90

0.88

Predicted
0.86
Shao and Wang 15

2 3 4 5 6 7
Bottom / Top feeds mass flow ratio
(a)

0.99 Predicted
Shao and Wang 15

0.98
NaOH Conversion

0.97

0.96

0.95

0.94

3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4

Methanol / NaOH mass flow ratio in top feed
(b)

0.98

0.96

0.94
NaOH Conversion

0.92

0.90

0.88
Predicted

0.86 Shao and Wang 15

0.0 0.1 0.2 0.3 0.4

Water content in bottom feed (%wt)
(c)

Figure S4: Effect of the (a) bottom/top feed mass flow ratio; (b) methanol/NaOH mass flow
ratio on the conversion of NaOH; and the (c) water content in bottom feed.

S17
 Table S9: Effect of bottom-to-top feed ratio on conversion of NaOH.

Bottom/Top Conversiona Predicted Relative error

mass ratio Shao and Wang 12 conversion (%)
2 0.910 0.848 6.81
3 0.940 0.924 1.69
4 0.970 0.951 1.96
5 0.980 0.964 1.64
6 0.985 0.972 1.29
7 0.990 0.978 1.21
ARDb 2.43
a
Values retrieved graphically.
b
Average relative deviation.

Table S10: Effect of methanol-to-NaOH mass ratio in top feed on conversion of NaOH.

CH3OH/NaOH Conversiona Predicted Relative error

mass ratio Shao and Wang 12 conversion (%)
3.00 0.940 0.948 0.89
3.25 0.954 0.949 0.54
3.50 0.967 0.950 1.76
3.80 0.978 0.950 2.82
4.12 0.988 0.952 3.59
4.52 0.991 0.956 3.49
ARDb 2.18
a
Values retrieved graphically.
b
Average relative deviation.

S18
Table S11: Effect of water content in bottom feed on conversion of NaOH.

Water content Conversion 12a Predicted Relative error

in feed (wt%) conversion (%)
0.00 0.990 0.951 3.94
0.05 0.985 0.948 3.77
0.10 0.982 0.945 3.77
0.15 0.979 0.942 3.75
0.20 0.971 0.940 3.21
0.25 0.961 0.937 2.46
0.30 0.948 0.935 1.36
0.35 0.928 0.933 0.53
0.40 0.900 0.931 3.42
0.45 0.855 0.929 8.62
ARDb 3.48
a
Values retrieved graphically.
b
Average relative deviation.

S19
D Sensitivity analyses of Process III

D.1 Operating variables of D-501

Figure S5 represents the conversion of NaOH against the ratio methanol/water in NaOH
when D-501 is simulated separately (no methanol recycle), with 20 theoretical stages and
operating at 1 atm. As can be observed, augmenting this ratio increases the NaOH conver-
sion, when the concentration of the bottom product is maintained constant (30 wt% NaOCH3
in a methanol solution).

0.990
NaOH Conversion

0.988

0.986

0.984

0.982
35 40 45 50
Methanol/water in NaOH mass flow ratio

Figure S5: Conversion of NaOH against the mass flow ratio methanol/water in caustic soda.

Figure S6 and Table S12 present the results of the sensitivity analysis in D-501, by varying
the total number of equilibrium stages (NT). The optimal value is located in the region of
NT = 20−22, where a lower number of stages would originate a product out of specifications
for methanol feed streams with 100 ppm of water, without significant gains in terms of the
reboiler duty by further decreasing NT.

S20
 “Near”-optimal NaOH prod. (0 ppm)
number of stages
NaOH prod. (100 ppm ) 0.42
7.060

7.055
0.40

%wt NaOH in bottoms

Qreboiler (GJ/h)

7.050
0.38
7.045

0.36
7.040

7.035 Q reb. (0 ppm ) 0.34

Q reb. (100 ppm)
7.030
18 20 22 24 26
Number of stages

Figure S6: Selection of the total number of stages for D-501. Values in parenthesis represent
the water content in methanol vapor stream.

Table S12: Numerical results of the sensitivity analysis of D-501 as a function of NT.

Mass frac. H2O 100 ppm 0 ppm

NT Reb. duty (GJ h−1 ) wt% NaOH Conv. Reb. duty (GJ h−1 ) wt% NaOH Conv.
17 7.031 0.4285 0.9813 7.041 0.3975 0.9826
18 7.036 0.4182 0.9817 7.046 0.3855 0.9831
19 7.040 0.4095 0.9821 7.049 0.3751 0.9836
20 7.040 0.3993 0.9825 7.051 0.3661 0.9840
21 7.042 0.3957 0.9827 7.053 0.3583 0.9843
22 7.044 0.3903 0.9829 7.055 0.3516 0.9846
23 7.046 0.3857 0.9831 7.057 0.3457 0.9848
24 7.048 0.3818 0.9833 7.059 0.3405 0.9851
25 7.049 0.3784 0.9834 7.060 0.3359 0.9853
26 7.049 0.3754 0.9836 7.061 0.3319 0.9854
27 7.050 0.3729 0.9837 7.063 0.3283 0.9856

S21
D.2 Operating variables of D-502

For the distillation column D-502, the base case values of NT and RRmas were calculated
by minimization of the equivalent annual operating costs (EAOC) of the entire Process III,
using the specifications for column D-501 obtained in the previous Section. The RRmas was
adjusted to maintain the distillate and bottom streams of D-502 within the specifications
described in the manuscript. In the EAOC method, the grassroots capital costs CGR is
spread over equal annuities over the operating life (n = 10 years), the discount rate i is 10 %
and the total manufacturing costs excluding depreciation (COMd ) is expressed as

EAOC = CGR,d + COMd = CGR F(i, n) + 0.18CGR + 2.76COL + 1.23 (CUT + CWT + CRM )

where the discount factor F(i, n) converts, in this case, the grassroots capital costs present
value to annuities (CGR,d ):
i(1 + i)n
F(i, n) = = 0.16
(1 + i)n − 1

Figure S7 shows that the normalized equivalent annual operating costs (EAOCn ) decreases
sharply with the number of stages up to about 29, since the reduction of the D-502 mass
reflux ratio leads to great energy cost savings counterbalanced by a minimal rise of capital
costs, as the extra size of D-502 is amortized by the lower areas for heat exchange in the
reboiler and condenser. EAOCn stays close to the minimum until 37 equilibrium stages are
used, increasing again for higher number of theoretical stages, due to the rise of capital costs
(increased size of D-502) and a diminished variation of the utilities costs term. Figure S8
describes the variation of the total number of stages and optimal feed location as a function
of the reflux ratio in the column required to achieve the product specifications. These results
are summarized together in Table S13.

S22
 1.6

1.5
CGRd,n
1.4 COMd,n
EAOCn
1.3
Normalized cost

1.2

1.1

1.0

0.9
20 25 30 35 40 45 50
Number of stages

Figure S7: Normalized equivalent annual operating costs depending on the number of stages
of D-502.

50

20
40
Number of stages

15
Qreboiler (GJ/h)

30

10
20

10 Feed stage (NF) 5

Total number of stages (NT)
Reb. duty

0 0
1.0 1.2 1.4 1.6
Mass reflux ratio

Figure S8: Selection of the total and feed stages, and mass reflux ratio for D-502.

S23
 Table S13: Numerical results of the sensitivity analysis of D-502.

NT RRmas Q502 /(GJ h−1 ) CGR,d /($M) COMd /($M) EAOC/($M) EAOCn COMdn CGR,dn
20 1.670 23.16 0.948 18.391 19.339 1.000 1.000 1.000
21 1.540 21.23 0.986 18.173 19.159 0.991 0.988 1.040
22 1.440 19.73 0.918 17.893 18.811 0.973 0.973 0.968
23 1.352 18.42 0.957 17.764 18.720 0.968 0.966 1.009
24 1.290 17.50 0.974 17.660 18.634 0.964 0.960 1.028
25 1.235 16.68 0.987 17.564 18.551 0.959 0.955 1.041
26 1.190 16.01 0.976 17.461 18.436 0.953 0.949 1.029
27 1.151 15.43 0.988 17.396 18.384 0.951 0.946 1.041
28 1.120 14.97 1.017 17.326 18.343 0.948 0.942 1.072
29 1.090 14.52 1.041 17.333 18.375 0.950 0.943 1.098
30 1.070 14.22 1.053 17.306 18.359 0.949 0.941 1.110
31 1.045 13.85 1.080 17.287 18.367 0.950 0.940 1.139
32 1.025 13.55 1.092 17.260 18.352 0.949 0.938 1.152
33 1.010 13.32 1.137 17.279 18.416 0.952 0.940 1.199
34 0.990 13.03 1.149 17.252 18.401 0.952 0.938 1.212
35 0.975 12.80 1.172 17.248 18.421 0.953 0.938 1.236
36 0.965 12.65 1.194 17.253 18.447 0.954 0.938 1.260
37 0.950 12.43 1.206 17.236 18.442 0.954 0.937 1.272
38 0.940 12.28 1.246 17.260 18.506 0.957 0.939 1.314
40 0.920 11.98 1.289 17.268 18.557 0.960 0.939 1.360
42 0.905 11.76 1.278 17.226 18.504 0.957 0.937 1.348
44 0.890 11.53 1.334 17.257 18.591 0.961 0.938 1.407
46 0.875 11.31 1.370 17.267 18.637 0.964 0.939 1.445
48 0.865 11.16 1.448 17.333 18.782 0.971 0.943 1.527
50 0.855 11.01 1.501 17.371 18.872 0.976 0.945 1.582

S24
D.3 Effect of the operating pressures in D-501 and D-502

A sensitivity analysis was also considered to elucidate the effect of the D-501 and D-502 op-
erating pressures (respectively, PD501 and PD502 ) on the economic performance of Process III.
Higher pressures reduce the size of distillation units because the mixture densities increase.
In contrast, lowering the pressure facilitates the component separation due to higher relative
volatilities, reducing the reflux ratio and the reboiler duties in distillation columns. This
compromise was studied for Process III, by recalculating the economic factors for the set of
pressures 0.40 bar, 0.55 bar, 0.70 bar, 2 bar and 4 bar in D-501 and D-502, considered indi-
vidually in each column. In the cases where PD502 is equal or higher than PD501 , compressor
C-501 is installed between D-501 and D-502. Moreover, when PD502 exceeds 1 atm a second
pump is installed before H-501 to maintain the pressure head of the methanol vapor recycle
stream.
The results obtained are detailed in Table S14, and can be visualized in Figure S9. An
increase of the break even price index of NaOCH3 can be observed when PD501 and PD502
differ from 1 atm. When both distillation columns operate at vacuum, cost savings of 21 %
with steam consumption can be achieved (for PD501 = PD502 = 0.40 bar), since the relative
volatility of methanol and water rises, lowering the RRmas of D-502 from 1.025 to 0.808, the
heat duty of REB-501 from 6.93 to 5.29 GJ h−1 , and the REB-502 from 13.55 to 10.82 GJ h−1 .
However the capital costs of the distillation columns D-501 and D-502 increase, respectively,
from $650 K to $850 K and from $1.55 M to $2.45 M, due to the larger volumetry of D-501
and D-502 and to the additional cost of installing strengthening rings to avoid the walls
of distillation columns from buckling. Furthermore, when PD502 = 0.40 bar, the cooling
utility costs are about 9 times higher than those at higher pressures, since refrigerated water
($4.43 GJ−1 , 14 ) is required to condensate the vapors below 50 ◦C.
In the opposite direction, when PD501 and PD502 are raised simultaneously up to 4 bar,
an increase of COMd by 11 % to $19.19 M is observed, as a direct result of CUT growing
60 %, despite a decrease of the annual electric costs associated with C-501 by 61 % from

S25
 Table S14: Numerical results of Process III economic performance depending on the operating pressure.

PD501 PD502 CUT a CGR b COMd c NaOCH3 price NaOCH3 PD501 PD502 CUT a CGR b COMd c NaOCH3 price NaOCH3
(bar) (bar) ($/M) ($/M) ($/M) ($/tDM) price index (bar) (bar) ($/M) ($/M) ($/M) ($/tDM) price index
0.40 0.40 3.50 10.00 18.72 3936 1.121 1.01 0.40 2.96 10.20 18.09 3828 1.090
0.40 0.55 2.68 8.82 17.50 3661 1.042 1.01 0.55 2.65 8.38 17.38 3621 1.031
0.40 0.70 2.78 8.46 17.56 3657 1.041 1.01 0.70 2.74 7.63 17.36 3586 1.021
0.40 1.01 2.97 8.01 17.71 3666 1.044 1.01 1.01 2.80 6.46 17.22 3512 1.000
0.40 2.00 3.48 8.43 18.41 3814 1.086 1.01 2.00 3.44 6.88 18.08 3689 1.050
0.40 4.00 4.53 9.24 19.85 4114 1.171 1.01 4.00 4.49 7.69 19.52 3989 1.136
0.55 0.40 2.94 9.96 18.02 3805 1.084 2.00 0.40 3.01 10.60 18.22 3869 1.102
0.55 0.55 2.62 8.08 17.29 3592 1.023 2.00 0.55 2.70 8.81 17.52 3665 1.044

S26
0.55 0.70 2.76 7.78 17.41 3601 1.026 2.00 0.70 2.79 8.08 17.50 3631 1.034
0.55 1.01 2.97 7.38 17.60 3619 1.031 2.00 1.01 2.95 6.90 17.48 3579 1.019
0.55 2.00 3.47 7.79 18.28 3764 1.072 2.00 2.00 3.38 5.73 17.80 3589 1.022
0.55 4.00 4.51 8.62 19.71 4063 1.157 2.00 4.00 4.46 6.51 19.27 3894 1.109
0.70 0.40 2.95 10.00 18.04 3811 1.085 4.00 0.40 3.11 11.30 18.47 3944 1.123
0.70 0.55 2.64 8.19 17.34 3605 1.026 4.00 0.55 2.80 9.51 17.77 3740 1.065
0.70 0.70 2.73 7.44 17.31 3569 1.016 4.00 0.70 2.90 8.80 17.77 3710 1.056
0.70 1.01 2.94 6.98 17.49 3583 1.020 4.00 1.01 3.05 7.68 17.75 3660 1.042
0.70 2.00 3.43 7.42 18.17 3727 1.061 4.00 2.00 3.53 6.57 18.14 3686 1.050
0.70 4.00 4.49 8.27 19.63 4032 1.148 4.00 4.00 4.49 5.83 19.19 3850 1.096
a b c
Annual utilities cost. Grassroots capital cost. Annual manufacturing costs excluding depreciation.
 4.0
1.14 1.13
1.1
3.5 1.11
1.12
3.0 1.07
1.08
1.09

PD502 /bar 2.5

1.06

2.0 1.03
1.05

1.5
1.02

1.0 1.04
1.01
1.02
0.5 1.07 1.05 1.06
1.08
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
PD501 /bar

Figure S9: Approximate contour levels of the effect of PD501 and PD502 on the break even
price index of NaOCH3 in Process III. Reference unit value: $3512 /t DM.

$307 K to $120 K. This trend is due to the fact that with a higher operating pressure, the
volumetric vapor flowrate and the relative volatility of methanol and water both decrease,
raising the composition of water in the liquid at the bottom of D-501. Therefore, a higher
heat duty in REB-501 is necessary (8.73 GJ h−1 ) to keep the product moisture within the
limits as well as a high conversion of NaOH. Moreover, in order to keep the composition of
the distillate and bottoms within the specified limits, the reboiler duty and the RRmas of
D-502 also increased to 26.75 GJ h−1 and 2.10, respectively. As a result, CGR reaches $5.73 M
when PD501 = PD502 = 2 bar, and $5.83 M at 4 bar, and the break even price of NaOCH3
increases to $3589 /tDM and $3850 /tDM, respectively. The increase of CGR when PD501
and PD502 goes from 2 bar to 4 bar is primarily associated with the larger volume of D-502
(higher RRmas ) and with the extra material cost of D-501 and D-502, requiring thicker walls
to safely sustain the higher operating pressures.
Operating D-501 and D-502 with different pressure levels, the annual electric and capital

S27
costs of compressor C-501 will dominate the economic performance of Process III. At PD501 =
4 bar and PD502 = 0.40 bar, CGR of C-501 increases to $5.28 M and the annual electric costs
to $816 K, contrasting with $1.77 M and $120 K, respectively, for PD501 = PD502 = 4 bar.
As the gap between PD501 and PD502 increases, more mechanical work on recompressing the
methanol vapor stream is required, compromising the economy of the process and rising the
break even price of the NaOCH3.
A variation of the current design where D-501 and D-502 operate at atmospheric con-
ditions, the condenser of D-502 is total instead of partial-vapor, and the compressor C-501
is substituted by a second pump was also considered. The estimated break even price of
NaOCH3 is $3700 /tDM, 5.4 % higher than in the base case conditions, primarily justified
with the steam consumption that rises to 39.67 GJ h−1 , compared to 24.03 GJ h−1 at base
case conditions. In this case, the cost of evaporating all the methanol recycle stream does
not compensate the capital and electric costs of installing and operating a compressor C-501.

S28
E Process III simulation at base case conditions

Table S15 and Table S15 describe the operating variables of Process III in the base case
conditions considered.

Table S15: Comparison between the operating conditions for D-501 and D-502, estimated
and those described in literature.

Operating conditions Simulation results

Guth et al. 11 ASPEN Plus®
D-501
Pressure (bar) 1–15 1.01
NT 15–30 20
CH OH H2O
F5043 /FNAOH,502 †
10–50 30.6
Water in product (ppm) < 1000 547
NaOH in product (wt%) 0.39
D-502
Pressure (bar) 1–4 1.01
NT 32
NF 26
RRmas 0.80–1.40 1.025
Water in distillate (ppm) 20–100 95
CH3OH in bottoms (wt%) < 0.10 0.10
†
Ratio of methanol mass flow rate entering in D-501 and the water mass flow rate
in caustic soda.

S29
 Table S16: Complete table results of Process III at base case conditions.

Stream 30NAOCH3 501 502 503 504 505 506 507 METHANOL NAOH WASTE
Temperature (◦C) 95.5 30.2 75.0 65.0 158.0 70.7 65.0 65.0 30.0 30.0 100.0
Pressure (bar) 1.01 3.04 3.04 1.01 3.04 1.01 1.01 1.01 1.01 1.01 1.01
Vapor fraction 0 0 0 1 1 1 1 1 0 0 0
Mole flow (kmol h−1 ) 62.38 53.65 53.65 61.89 484.7 463.5 422.9 484.7 61.89 53.65 40.67
Mass flow (kg h−1 ) 2267 1017 1017 1983 15531 14281 13548 15531 1983 1017 733.0
Component Mass Flow (kg h−1 )

S30
H 2O 1.241 508.5 508.5 0 1.287 733.6 1.287 1.287 0 508.5 732.3
NaOH 8.803 508.5 508.5 0 0 0 0 0 0 508.5 0
NaOCH3 674.9 0.0 0.0 0 0 0 0 0 0 0 0
CH3OH 1582 0 0 1983 15530 13547 13547 15530 1983 0 0.718
Mass fraction
H 2O 547 ppm 0.5 0.5 0 83 ppm 0.051 95 ppm 83 ppm 0 0.5 0.999
NaOH 0.004 0.5 0.5 0 0 0 0 0 0 0.5 0
NaOCH3 0.298 0 0 0 0 0 0 0 0 0 0
CH3OH 0.698 0 0 1 1 0.949 1 1 1 0 979 ppm
F Cash flows of processes II and III

Table S17: Non- and discounted after-tax cash flows of Process II during its project-life
period.
k
Year Investment ATCFa NDCFb NDCCFc DCFd DCCFe
P
dk FCIL − dk Rvn COMd
i=0

0 0.60 0.69 (0.60) (0.60) (0.60) (0.60)

1 0.41 0.69 (0.41) (0.38) (0.98) (1.01)
2 1.66 0.69 (1.66) (1.37) (2.34) (2.67)
3 0.14 0.55 17.79 17.27 0.36 0.36 0.27 (2.07) (2.31)
4 0.22 0.33 17.79 17.27 0.39 0.39 0.27 (1.81) (1.92)
5 0.13 0.20 17.79 17.27 0.36 0.36 0.22 (1.58) (1.56)
6 0.08 0.12 17.79 17.27 0.33 0.33 0.19 (1.40) (1.23)
7 0.08 0.04 17.79 17.27 0.33 0.33 0.17 (1.22) (0.89)
8 0.04 17.79 17.27 0.32 0.32 0.15 (1.08) (0.57)
9 17.79 17.27 0.30 0.30 0.13 (0.95) (0.27)
10 17.79 17.27 0.30 0.30 0.12 (0.83) 0.03
11 17.79 17.27 0.30 0.30 0.11 (0.73) 0.33
12 17.79 17.27 0.30 2.28 0.73 0.00 2.61
a
After-tax cash flow: (Rvn − COMd − dk )(1 − t) + dk . b Non-discounted cash flow.
c
Non-discounted cumulative cash flow. d Discounted cash flow. e Discounted cumulative cash flow.

Table S18: Non- and discounted after-tax cash flows of Process III during its project-life
period.
k
Year Investment ATCFa NDCFb NDCCFc DCFd DCCFe
P
dk FCIL − dk Rvn COMd
i=0

0 0.60 6.46 (0.60) (0.60) (0.60) (0.60)

1 3.88 6.46 (3.88) (3.52) (4.12) (4.48)
2 4.14 6.46 (4.14) (3.42) (7.55) 0.00
3 1.29 5.17 18.96 17.22 1.55 1.55 1.17 (6.38) 1.55
4 2.07 3.10 18.96 17.22 1.88 1.88 1.28 (5.10) 3.43
5 1.24 1.86 18.96 17.22 1.53 1.53 0.95 (4.15) 4.96
6 0.74 1.12 18.96 17.22 1.32 1.32 0.75 (3.40) 6.28
7 0.74 0.37 18.96 17.22 1.32 1.32 0.68 (2.73) 7.60
8 0.37 18.96 17.22 1.17 1.17 0.54 (2.18) 8.77
9 18.96 17.22 1.01 1.01 0.43 (1.75) 9.78
10 18.96 17.22 1.01 1.01 0.39 (1.36) 10.79
11 18.96 17.22 1.01 1.01 0.35 (1.01) 11.80
12 18.96 17.22 1.01 3.17 1.01 (0.00) 14.97
a
After-tax cash flow: (Rvn − COMd − dk )(1 − t) + dk . b Non-discounted cash flow.
c
Non-discounted cumulative cash flow. d Discounted cash flow. e Discounted cumulative cash flow.

S31
 G Monte-Carlo simulation numerical results

Table S19: Numerical results of Monte-Carlo simulations for Process II.

Bins 0 1 2 3 4 5 6 7 8 9 10
All Upper value -9.04 -7.4 -5.76 -4.13 -2.49 -0.86 0.78 2.42 4.05 5.69 7.33
Cumulative 0 230 812 1593 3977 5998 7185 8773 9479 9740 10000
FCIL Upper value -0.23 -0.19 -0.15 -0.12 -0.08 -0.04 0 0.04 0.08 0.12 0.15
Cumulative 0 198 696 1553 2695 4108 5933 7646 8928 9720 10000
Product Upper value -5.05 -4.03 -3.01 -1.99 -0.97 0.05 1.07 2.09 3.11 4.13 5.15

S32
Cumulative 0 170 1198 4394 7357 8988 9286 9541 9694 9899 10000
FCIworking Upper value -0.07 -0.03 0.01 0.06 0.1 0.14 0.18 0.22 0.26 0.3 0.35
Cumulative 0 633 2335 4121 5651 6964 8038 8883 9495 9867 10000
ITR Upper value -0.18 -0.15 -0.11 -0.07 -0.04 0 0.04 0.07 0.11 0.14 0.18
Cumulative 0 196 807 1778 3216 4955 6736 8134 9115 9768 10000
AIR Upper value -0.25 -0.21 -0.17 -0.13 -0.09 -0.05 -0.01 0.03 0.07 0.11 0.15
Cumulative 0 235 831 1721 2921 4359 6068 7777 9015 9745 10000
CRM Upper value -6.66 -5.55 -4.43 -3.32 -2.2 -1.08 0.03 1.15 2.27 3.38 4.5
Cumulative 0 134 725 1765 2519 3444 6383 7629 9323 9878 10000
CUT Upper value/(10−3 ) -3.85 -3.08 -2.31 -1.54 -0.77 0.00 0.772 1.542 2.312 3.081 3.851
Cumulative 0 240 831 1821 3228 4974 6775 8173 9151 9766 10000
 Table S20: Numerical results of Monte-Carlo simulations for Process III.

Bins 0 1 2 3 4 5 6 7 8 9 10
All Upper value -7.9 -6.4 -5 -3.5 -2.1 -0.6 0.8 2.3 3.7 5.2 6.7
Cumulative 0 260 786 1847 3391 5074 7000 8511 9540 9882 10000
FCIL Upper value -2.2 -1.8 -1.5 -1.1 -0.7 -0.4 0 0.4 0.7 1.1 1.4
Cumulative 0 183 695 1526 2689 4170 5960 7682 8943 9722 10000
Product Upper value -5.5 -4.4 -3.3 -2.2 -1.1 0 1.1 2.2 3.3 4.4 5.6
Cumulative 0 207 757 1705 3243 5023 6438 7941 9120 9788 10000
FCIworking Upper value -0.1 0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.3 0.4

S33
Cumulative 0 614 2323 4100 5630 6942 8023 8866 9476 9862 10000
ITR Upper value -0.4 -0.3 -0.2 -0.2 -0.1 0 0.1 0.2 0.2 0.3 0.4
Cumulative 0 212 819 1811 3197 4987 6764 8154 9166 9788 10000
AIR Upper value -0.7 -0.6 -0.5 -0.4 -0.2 -0.1 0 0.1 0.2 0.3 0.4
Cumulative 0 191 753 1630 2797 4243 5957 7694 8981 9748 10000
CRM Upper value -4.6 -3.8 -3.1 -2.3 -1.5 -0.8 0 0.7 1.5 2.3 3
Cumulative 0 137 542 1337 2581 4185 5779 7625 9006 9738 10000
CUT Upper value -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Cumulative 0 227 846 1864 3252 5052 6830 8207 9199 9805 10000
Table S21: Break even prices of NaOCH3 depending on its annual capacity of production.

Capacity of production Process II Process III

($t DM/y) ($/t DM) ($/t DM)
5400 3295 3512
6000 3257 3459
10000 3125 3268
14000 3068 3176
18000 3033 3128

S34
References

1. Robinson, R. A. Electrolyte solutions, 2nd ed.; Dover Publications: Mineola, NY, 2002.

2. Chen, C.-C.; Britt, H. I.; Boston, J. F.; Evans, L. B. AIChE J. 1982, 28, 588–596.

3. Chen, C.-C.; Song, Y. AIChE J. 2004, 50, 1928–1941.

4. Aspen Technology, Aspen Physical Property System: Physical Property Models; Version
Number: 7.2: Burlington, MA, July, 2010.

5. Rosenthal, R. GAMS - A User’s Guide; GAMS Development Corporation: Washington,

DC, March, 2014.

6. Freeguard, G.; Moodie, R. B.; Smith, D. J. G. J. Appl. Chem. USSR 1965, 15, 179–182.

7. Leal, J. P.; de Matos, A. P.; Simões, J. A. M. J. Organomet. Chem. 1991, 403, 1–10.

8. Ugray, Z.; Lasdon, L.; Plummer, J.; Glover, F.; Kelly, J.; Martí, R. A Multistart
Scatter Search Heuristic for Smooth NLP and MINLP Problems. In Operations Re-
search/Computer Science Interfaces Series; Rego, C., Alidaee, B., Eds.; Operations Re-
search/Computer Science Interfaces Series; Kluwer Academic Publishers: Norwell, MA,
2005; Vol. 30; pp 25–57.

9. Drud, A. S. ORSA Journal on Computing 1994, 6, 207–216.

10. Murto, J. Hydroxide-alkoxide ion equilibria and their influence on chemical reac-
tions. In The Hydroxyl Group (1971) [Online]; Patai, S., Ed.; PATAI’S Chemistry
of Functional Groups; John Wiley & Sons, Posted Jan 16, 2010; pp 1087–1115.
http://onlinelibrary.wiley.com/doi/10.1002/9780470771259.ch20/summary (ac-
cessed March 10, 2015).

S35
11. Guth, J.; Friedrich, H.; Sterzel, H.-J.; Kaibel, G.; Burkart, K.; Hoffmann, E. Method for
producing alkali methylates. U.S. Pat. 6759560 B2, July 6, 2004.

12. Shao, D. D.; Wang, C. X. Adv. Mat. Res. 2012, 557-559, 2350–2354.

13. Chandran, K.; Srinivasan, T.; Gopalan, A.; Ganesan, V. J. Chem. Thermodyn. 2007,
39, 449–454.

14. Turton, R.; Bailie, R. C.; Whiting, W. B.; Shaeiwitz, J. A.; Bhattacharyya, D. Analysis,
Synthesis and Design of Chemical Processes, 4th ed.; Prentice Hall: Upper Saddle River,
NJ, 2012.

S36