See discussions, stats, and author profiles for this publication at: [Link]

net/publication/257566743

Pilot-scale experimental validation of unsaturated polyesters synthesis by

reactive distillation

Article in Chemical Engineering Journal · December 2012

DOI: 10.1016/[Link].2012.10.003

CITATIONS READS

14 671

4 authors, including:

Anton A. Kiss Edwin Zondervan

Delft University of Technology University of Twente
351 PUBLICATIONS 10,261 CITATIONS 212 PUBLICATIONS 2,943 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Anton A. Kiss on 23 January 2019.

The user has requested enhancement of the downloaded file.

 1 Pilot-scale experimental validation of unsaturated
2 polyesters synthesis by reactive distillation
3

4 Mayank Shah,1,2 Anton A. Kiss,1* Edwin Zondervan,2 André B. de Haan2

1
5 AkzoNobel Research, Development & Innovation, Process Technology ECG, Zutphenseweg
6 10, 7418 AJ Deventer, The Netherlands. Tel: +31 26 366 9420,
7 E-mail addresses: [Link]@[Link], [Link]@[Link]
2
8 Process Systems Engineering, Department of Chemical Engineering and Chemistry,
9 Eindhoven University of Technology, 5600 MB, The Netherlands
10
11 Keywords: reactive distillation, polyesters, experimental validation, rate based modeling
12
13 Abstract
14 Reactive distillation (RD) is an integrated operation combining both reaction and separation,
15 thus improving the productivity and selectivity, reducing the energy usage, eliminating the
16 need for solvents and leading to intensified highly-efficient systems. In previous work, we
17 performed the conceptual design and modeling of a reactive distillation process for synthesis
18 of unsaturated polyesters. The simulation results of our previous work clearly demonstrated
19 that reactive distillation is a promising alternative for the polyesters synthesis.
20 This paper presents our pilot scale experimental validation work on the proposed RD process
21 for polyester synthesis. Two configurations are investigated: 1. a RD column alone and 2. a
22 RD column coupled with a pre-reactor. For the first configuration, we demonstrated that
23 mostly monoesters are formed in the RD column, due to the short residence time at pilot
24 scale. Since the combination of a RD column with pre-reactor allows sufficient residence
25 time also at pilot scale, we demonstrated that polyesters are formed in the RD column. The
26 extended rate-based model developed previously was used to simulate the pilot plant RD
27 column. The model adequately predicts and describes the experimental data obtained from
28 the pilot plant. Moreover, the product specification of the polyester produced in the
29 continuous RD column was compared with the polyester product manufactured at industrial
30 scale. Remarkably, it was found that the product specifications of the polyester produced in
31 the RD column is comparable to that of the polyester produced in industry.
*
32 Corresponding author’s e-mail address: [Link]@[Link], TonyKiss@[Link]

-1-
 1 1. Introduction
2 Most chemical processes involve reaction and separation operations that are typically carried
3 out in different sections of the plant and use different equipment types (e.g. reactors and
4 distillation columns) operated under a wide variety of conditions. Recycle streams between
5 these units are often used to improve conversion and selectivity, to minimize the production
6 of undesired by-products, to reduce the energy requirements, and to improve the process
7 controllability (Luyben and Yu, 2008). Recent economic and environmental considerations
8 have encouraged industry to focus on technologies based on process intensification. Reactive
9 distillation (RD) is such an integrated operation that combines reaction and separation into a
10 single unit allowing the simultaneous production and removal of products, thus improving
11 selectivity and productivity, reducing the energy use and the need for solvents while also
12 leading to high-efficiency systems with green engineering attributes (Malone and Doherty,
13 2000; Taylor and Krishna, 2000; Sundmacher and Kienle, 2003; Shah et al., 2011a, 2012a).
14 Some of these benefits are obtained by using reaction to improve separation (e.g. overcoming
15 azeotropes, avoiding or eliminating difficult separations) while others are achieved by using
16 separation to improve reactions (e.g. overcoming reaction equilibrium limitations, improving
17 selectivity) – the maximum effect being achieved when both aspects are important (Malone
18 and Doherty, 2000; Harmsen, 2007). However, the application of RD is somewhat limited by
19 constraints, such as: common operating range for distillation and reaction, proper boiling
20 point sequence, providing proper residence time characteristics and more difficult control.
21 Remarkably, RD has found many industrial applications during the past decades (Tuchlenski
22 et al., 2001; Schoenmakers and Bessling, 2003; Stankiewicz, 2003; Hiwale et al., 2004;
23 Harmsen, 2007, 2010). In particular, (trans-)esterification and etherification are some of the
24 most used equilibrium reactions (Malone and Doherty, 2000; Sundmacher and Kienle, 2003;
25 Bisowarno et al., 2004; Shah et al., 2011a, 2012b; Keller et al., 2012). Nowadays, the
26 application with the largest number of installations is methyl tertiary butyl ether (MTBE) that
27 is used in gasoline blending (Luyben and Yu, 2008). Other esters such as ethyl tertiary butyl
28 ether (ETBE), tert-amyl methyl ether (TAME) or fatty acid methyl esters (FAME), are also
29 produced now by RD (Harmsen, 2007, 2010; Kiss and Bildea, 2012).
30 In previous work we made a review of the available modeling approaches for RD, and
31 applied the findings to the polyester synthesis – the main focus being on extending the
32 dynamic rate-based model (Baur et al., 2000) to take into account the liquid back mixing
33 (Shah et al., 2012b, 2012c). The simulation results of previous work demonstrated that
34 reactive distillation is a promising alternative for polyesters synthesis (Shah et al., 2011a,

-2-
 1 2012a). The production time was reduced by 85% and the required volume to produce
2 polyester at industrial scale (typically 100 kton/yr) was reduced by 74%. Moreover, the RD
3 model predicted the polymer attribute, isomerization and saturation composition of the
4 polymer to be in the range of industrial production data for the polyester process. Although
5 the ideal reaction temperature of the polyester process is between 210°C and 220°C – to
6 avoid the degradation of reactants and products (Parker, 1966) – the reaction temperature of
7 the reactive distillation was varied between 185°C and 270°C. It was assumed that there is
8 significantly less risk of destructing the unsaturated acid in the RD process due to the short
9 residence time requirement and the free acid (or anhydride) only being present in the top
10 three stages of the RD column, where the reaction temperature remains between 185°C and
11 210°C as shown previously (Shah et al., 2011a).
12 Following the modeling work, it is very important to prove experimentally that polyesters can
13 be indeed produced in a RD column, and that the assumption given for operating the RD
14 column at temperatures over 220°C is reasonable. This paper discusses the experimental
15 work carried out at pilot scale, in a reactive distillation column. The polyester produced in the
16 pilot plant is compared with the polyester produced in industry. Moreover, the experimental
17 results are compared with the simulation results to check the validity of the previously
18 developed model, and the product specifications of the polyester produced in the RD column
19 are compared to that of polyester produced in the traditional industrial batch-reactor setup. To
20 our knowledge, the synthesis of unsaturated polyesters by reactive distillation was never
21 attempted before at pilot-scale – let alone at large scale – so our work is the first experimental
22 study that reports the successful manufacturing of polyesters in a RD column at pilot-scale,
23 while also validating the previous modeling work.
24
25 2. Materials and methods
26 2.1 Process chemistry
27 The reaction scheme is rather complex and accounts four types of reactions: ring opening,
28 poly-esterification, isomerisation and saturation. These types of reactions form a network of
29 nine reactions in total. The order of reaction is changing in time from 1 to 2 with respect to
30 the carboxylic acid, due to the autocatalytic effect of the carboxylic acid. The main four
31 chemical reactions are described hereafter:

-3-
 1
O CH3 O O CH3
O O
+ HO CH2 CH OH HO C CH CH C O CH2 CH OH
(1)

MAD PG COOH (MA) + E + OH (PG)

2
O O CH3 O O CH3
n HO C CH CH C O CH2 CH OH HO C CH CH C O CH2 CH O H + (n-1) H2O (2)
n

COOH + OH POLY Water

3
OH
OH O
O
O OH O (3)
OH

Cis (MA) Trans (FA)

4
O O
O O CH3 HO C CH CH2 C OH
+
HO CH2 CH OH O CH2 CH OH
(4)
HO C CH CH C OH
CH3

Dbb (MA) OH (PG) SatDbb (Sat acid)

5
6 Maleic anhydride (MAD) reacts with propylene glycol (PG) and produces an acid (COOH)
7 and alcohol (OH) end group with an ester (E) bridge (eq. 1). The carboxylic acid end groups
8 react with glycol end groups and form a polyester (POLY) and water (eq. 2). Unsaturated
9 dicarboxylic acids are used in the industrial production of polyesters. The presence of
10 unsaturated carboxylic acids essentially leads to side reactions such as isomerization (eq. 3)
11 and double bond saturation (eq. 4). The isomerized acid (FA) and saturated acid (SACID)
12 also esterify and produce polyester (POLY2D) and saturated polyester (POLYs), respectively.
13 Polyester (POLY1D) formed from maleic anhydride also isomerizes and saturates to produce
14 isomerized polyester (POLY2D) and saturated polyester (POLYs), respectively. The
15 isomerized acid (FA) and isomerized polyester (POLY2D) also saturates and produces
16 saturated acid (SACID) and saturated polyester (POLYs), respectively. The three reactions,
17 esterification, isomerization and saturation form a network of nine reactions, which are
18 summarized in the appendix along with their rate equations and kinetic parameters.

-4-
 1
2 2.2 RD configurations
3 The reaction between maleic anhydride and propylene glycol is taken as a model reaction
4 system for the synthesis of unsaturated polyesters. In previous work (Shah et al., 2011a,
5 2012a), we performed the conceptual design and modeling of the RD process for polyesters
6 synthesis and evaluated different configurations. In this work, two main configurations are
7 investigated:
8
9 Configuration 1: This configuration represents the polyester synthesis in a RD column by
10 feeding maleic anhydride as a liquid at the top of the column and propylene glycol as a vapor
11 at the bottom of the column. A schematic view of this configuration is shown in Figure 1 (a).
12
13 Configuration 2: This configuration represents the polyester synthesis in a RD column by
14 feeding monoesters as a liquid at the top of the column and propylene glycol as a vapor at the
15 bottom of the column. In this case, propylene glycol acts also as a stripping agent. The mono-
16 esters are prepared in a pre-reactor by mixing maleic anhydride and propylene glycol in the
17 ratio of 1:1.15. A schematic view of this configuration is shown in Figure 1 (b).
18 Note that using pre-reactors or side-reactors in combination with RD was previously applied
19 successfully to increase the residence time and hence the conversion of equilibrium limited
20 reactions (Bisowarno et al., 2004; Harmsen, 2007; Smejkal, 2009).
21
22 2.3 Experimental setup
23 A schematic view of the pilot plant setup is given in Figure 2, while Figure 3 shows a picture
24 of the pilot plant setup. The column has an inner diameter of 50 mm and a height of 1.7 m, as
25 well as an outer jacket for circulating hot oil. The material of construction is stainless steel.
26 Note that the wall effects rise sharply at diameters below 50 mm, while at diameters larger
27 than 100-150 mm the wall effects are almost independent of the diameter. Consequently, the
28 RD column of this experimental setup can be acceptable for scale-up purposes. The column is
29 equipped with 7 units of Sulzer BX packing, with a total packed section height of 1.19 m. The
30 space between top of the column and the top packing unit is 0.25 m and the space between
31 bottom of the column and the last packing unit is 0.25 m. Along the RD column, 8
32 temperature sensors are mounted to measure the temperature profile inside a column and 8
33 samples points are available to collect samples. The liquid holdup inside the column is
34 monitored by differential pressure measurement over the column. Two differential pressure

-5-
 1 transmitters (GE Sensing LX8381) are used to measure the pressure difference over the
2 packed section, and to measure the pressure difference in the bottom section of the column,
3 respectively. Four tanks of 50 L are used for storage – these tanks containing a level
4 controller and a temperature sensor. In addition, the tanks for anhydride and the bottom
5 product storage are equipped with an agitator and an outer jacket for heating or cooling. Two
6 heating thermostats (supplied by LAUDA, model E200) are used to heat the storage tanks.
7 The heating thermostats (supplied by LAUDA, model Proline P5) are used for heating up the
8 column to the required temperature.
9 A condenser containing a helical tube of 6 m is connected to the column to condense the
10 vapor leaving the column. The temperature of the condenser is controlled by a cooling
11 thermostat (supplied by Julabo, model F25). The reactants are fed to the column by a
12 magnetic gear pump (supplied by Bronkhorst, model MZR-7255). The mass flow rates of
13 reactants are controlled by a Coriolis mass flow controller (supplied by Bronkhorst, model
14 M53-RAD-220-B). The evaporator – model aSTEAM DV1C supplied by aDROP
15 Feuchtemeztechnik GmbH – is used to vaporize the glycol. The flow of vapor leaving the
16 column is controlled with a mass flow controller which is regulated by pressure
17 measurement. The pilot plant set up is monitored and controlled by a custom made program
18 with a graphical interface, coded in National Instruments LabVIEW.
19
20 2.4 Experimental procedure
21 The RD column is heated to the required isothermal temperature in both configurations. The
22 temperature difference between the top and bottom section of the column is maximum ± 3°C.
23 The column is operated at 1 bar in both configurations, the pressure being controlled by
24 manipulating a valve at the top of the RD column, which regulates the vapor flowrate. After
25 achieving the isothermal temperature and setting the required pressure in the column,
26 propylene glycol is fed to the column via the evaporator where propylene glycol is vaporized.
27 The evaporator is equipped with an additional heater to keep the vapor temperature above the
28 dew point. The evaporator requires at least 30 minutes to achieve full operational conditions,
29 and then it continuously evaporates the propylene glycol. The propylene glycol lines from the
30 evaporator to the column are also traced up to 300°C to avoid condensation of the propylene
31 glycol vapor. After achieving a constant vapor flow and an isothermal temperature in the
32 column, the liquid is fed at the top of the column. The liquid feed line is passing through the
33 column jacket to heat up the liquid feed to the column temperature. In case of the first

-6-
 1 configuration, maleic anhydride is preheated up to 90°C in the storage tank in order to liquefy
2 it. In case of the second configuration, monoesters are prepared in the storage tank by mixing
3 maleic anhydride and propylene glycol in the molar ratio of 1:1.15 at 90°C. Since this tank is
4 equipped with an agitator and an outer jacket for heating and cooling, this storage tank acts as
5 a pre-reactor. The distillate samples are collected at the outlet of the condenser. The samples
6 are collected at various points along the length of the column in case of first configuration
7 (RD column alone). In case of the RD column with pre-reactor, due to the high viscosity of
8 the liquid flowing inside the column, the samples could not be collected from the 1.5 mm
9 sample lines mounted on the column. Thus, the samples are only collected from the liquid
10 stream leaving the bottom of column.
11
12 2.5 Analytical methods

13 The acid number (AV) is defined as the amount of potassium hydroxide (KOH) expressed in
14 mg, required for neutralizing the free or unreacted carboxyl groups in 1 g of a sample. The
15 AV is determined by titration (with an accuracy of ±0.5%), using the following formula:
.
V NMWKOH
16 AV = (5)
W
17
18 The isomerization and saturation concentration of a sample is measured by 1H-NMR, with an
19 accuracy of ±5%. The 1H-NMR spectrum is recorded on a Varian VXR 400 spectrometer. A
20 sample is dissolved in dimethyl sulfoxide (DMSO) in 5 mm NMR tubes and tetra methyl
21 silane (TMS) is used as internal standard for quoting chemical shifts in all spectra. The
22 maleate-formed acid and ester compounds appear at peaks from 6.3 ppm to 6.5 ppm and the
23 fumarate-formed acid and ester compounds appear at the peaks from 6.6 ppm to 6.9 ppm
24 (Curtis et al., 1964). The peaks for maleate and fumarate compounds are well separated from
25 each other, and their area can be easily determined. The area of an NMR peak is directly
26 proportional to the number of protons causing the peak (Curtis et al., 1964); thus the relative
27 concentrations of maleate and fumarate present in the sample can be easily determined. The
28 molecular weight distribution is measured using gel permeation chromatography (GPC), with
29 an accuracy of ±5%. The calibration of GPC is performed by using polystyrene standards,
30 while tetrahydrofuran (THF) is used as solvent to dissolve the sample. The quantity of water
31 in the distillate is measured by Karl Fischer titration using a 652 KF Coulometer.
32

-7-
 1 2.6 Pilot plant modeling
2 The RD column is modeled as a packed column containing the Sulzer BX packing. The
3 extended rate-based model developed previously is used to simulate the column (Shah et al.,
4 2012b, 2012c). Aspen Custom Modeler (ACM) is used as a multipurpose CAPE tool to
5 extend the traditional rate-based model such that the complete model accounts for the effect
6 of the axial dispersion on the RD process. The extended rate-based model is described by the
7 following equations (see also the list of notation provided separately in this paper):

dniL, j Ci, j−1 − 2Ci, j +Ci, j+1

8 = Lj−1Ci, j−1 − Lj Ci, j + DaxM j + M j Ri, j + NiL, j + LF Ci,F + LRCi,R (6)
dt h2
dn Lj . . i =k i =k . .
9 = L j −1 − L j + S j +1 + M j ∑ Ri , j + ∑ N iL, j + L F + L R (7)
dt i =1 i =1

dniV, j yi , j
10 = V j +1 yi , j +1 − V j yi , j − N iV, j + VF yi , F (8)
dt
dnVj i =k
11 = V j +1 − V j − ∑ N iV, j + VF (9)
dt i =1

dE Lj . .  L * i =n . L
L  . .
12 = L j −1 H L
j −1 − L j H + S j +1 H
L
j
L
j +1 + a h j (T j − T j ) + ∑ N i , j H i , j  + L F H FL + L R H RL (10)
 L

dt  i =1 
dE Vj  i=n . V
V 
13 = V j +1 H Vj+1 − V j H Vj − a hVj (T jV − T j* ) + ∑ N i , j H i , j  + VF H FV (11)
dt  i =1 
i=n . L i =n . V
h (T − T ) + ∑ N i , j H = h (T − T ) + ∑ N i , j H i , j
L * L L V V * V
14 j j j i, j j j j (12)
i =1 i =1

15 yi , j Pj = xi , j γ i , j Pi ,satj (13)

16 N iV, j = N iL, j (14)

17 N iL, j = k l a ( xi*, j − xi , j ) (15)

18 N iV, j = k g a ( yi , j − yi*, j ) (16)

19
20 Note that the term LFCi,F in eq. (6) is applicable only to the feed stage and the term LRCi,R is
21 applicable only to the reflux stage. Since the poly-esterification process involves autocatalytic
22 reactions, the reactions take place throughout the whole column. As the system is operated at
23 low pressure (1 bar), the activity coefficient approach is sufficient to represent the vapor-
24 liquid equilibrium. In this work the polymer NRTL non-ideal thermodynamic model is used
25 to describe vapor-liquid equilibrium. The interaction parameters for the polymer NRTL

-8-
 1 model are reported in our earlier work (Shah et al., 2011b). The kinetics and thermodynamic
2 models, as well as the corresponding parameters from previous work are used here (Shah et
3 al., 2011b). For the reader’s convenience, these parameters are also given in appendix. The
4 Dirichlet and Neumann boundary conditions are applied to solve the model for the top stage
5 (j =1) and bottom stage (j =j), respectively (Shah et al., 2012c). The feed flow rates and
6 compositions, liquid holdup and reflux rate are used as initial conditions to solve the model.
7 Since we concluded in our previous work (Shah et al., 2012a, 2012b) that the polyester
8 synthesis by reactive distillation is kinetically controlled (Damkohler number lower than 1)
9 and the reaction takes place only in the liquid bulk (Hatta number less than 1), the reaction in
10 the liquid film is neglected. The liquid holdup, pressure drop and mass transfer are calculated
11 using the correlation for Sulzer BX packing as proposed by Bravo et al. (1992). Particularly,
12 the mass transfer correlations have a significant impact on correctly predicting the RD
13 column behaviour (Svandova et al., 2008). The axial dispersion coefficient is calculated using
14 the correlation proposed by Kushalkar et al. (1990).
15
16 3. Results and discussion
17 3.1 Configuration 1: RD column
18 In this configuration, the reactions between maleic anhydride and propylene glycol in the RD
19 column are investigated. The column is operated counter-currently with maleic anhydride fed
20 at the top and propylene glycol fed at the bottom of the column, as previously described in
21 the experimental procedure. Since the liquid anhydride feed tube passes through the column
22 jacket, the preheated maleic anhydride is further heated up to the column temperature. The
23 boiling points of maleic anhydride and propylene glycol are 202°C and 188°C, respectively.
24 Therefore, the column should be operated around 195°C to prevent significant vaporization
25 of maleic anhydride in the liquid feed line and excessive condensation of propylene glycol at
26 the bottom of the column. The operating parameters are listed in Table 1. After operating the
27 column for 2 h at steady column temperature, liquid and vapor flows, samples are taken from
28 the liquid feed stream tube, the liquid stream leaving the bottom of column and the sample
29 points at 1.2 m and 1.45 m of the column. The acid value of the samples is determined by
30 titration. The conversion of the carboxylic acid is determined from the acid value as follows:
AVMAD − AV product
31 X MAD = (17)
AVMAD
32 where, AVMAD and AVproduct are the acid value of maleic anhydride (reactant) and the product

-9-
 1 produced in the reactive distillation column, respectively.
2 The experimental acid value (AV) and the conversion (X) are compared with the simulation
3 results in Figure 4 (a) and Figure 4 (b), respectively. This comparison shows that the model
4 proposed previously (Shah et al., 2012b) adequately predicts the performance of the pilot-
5 scale RD column. Moreover, the acid value of maleic anhydride (572 mg/g) has decreased to
6 360 mg/g in the product leaving at the bottom of the column, thus indicating that 37%
7 conversion is achieved. The mean relative error between the experimental and simulated acid
8 value and conversion is 5% and 9%, respectively. The model also predicts 14% liquid holdup
9 in the column. Moreover, the residence time of the liquid is calculated from the liquid holdup
10 and the liquid flow rate, which is approximately 0.32 hours (19 minutes). Due to the short
11 liquid residence time of approximately 0.32 hours and the relatively low column temperature
12 of 195°C, a maximum conversion of 37% is achieved. Due to the lower conversion achieved
13 in the pilot reactive distillation column, mostly monoesters have been formed – as confirmed
14 also by the corresponding acid value for the monoesters.
15
16 3.2 Configuration 2: RD column with pre-reactor
17 The RD column is operated counter-currently with the monoester fed as a liquid at the top of
18 the column and propylene glycol fed as vapor at the bottom of the column. The monoesters
19 are prepared in the storage tank at 90°C by mixing maleic anhydride and propylene glycol in
20 the ratio of 1.0:1.15. Due to the formation of monoester in the storage tank, the acid value of
21 the mixture is reduced from 572 mg/g to 264 mg/g after 1 hour of mixing. The acid value of
22 the monoester reduces further with the reaction time, as shown in Figure 5. The monoesters
23 of variable acid value are fed to the column and the measured acid value of the monoester is
24 used to calculate the conversion achieved in the reactive distillation column as,
AVmonoester − AV product
25 X monoester = (18)
AVmonoester
26 where, AVmonoester and AVproduct are the acid value of monoester and the product produced in
27 the reactive distillation column, respectively.
28 A total of 8 experiments with varying operating conditions were performed in order to
29 validate the RD model and to quantify the influence of the temperature and the glycol vapor
30 feed flow rate on the polyesters synthesis. The experimental operating conditions are listed in
31 Table 2. After operating the column at steady state (constant temperature, liquid and vapor
32 flow rates), samples were taken at 15 minutes intervals from the liquid feed stream, the liquid

- 10 -
 1 product stream leaving at the bottom of column and the top distillate product. Note that the
2 steady state operation is reached in about 3-4 hours from the start-up of the system, while
3 changing from one steady state to another takes nearly 1-1.5 hrs.
4 The acid values of the bottom product at different operating conditions of the reactive
5 distillation column are shown in Figure 6. The acid values for different samples of the bottom
6 product at the particular operating condition are constant, which confirms that the samples
7 were taken when the column was operating under steady-state conditions. It can be seen from
8 Figure 6 (a) that the acid value of the bottom product reduces at higher temperature in the
9 column, which clearly indicates that a higher conversion is achieved by operating the column
10 at higher temperature. Figure 6 (b) and (c) show that the acid value of the bottom product
11 reduces with an increase in the glycol vapor feed flow rate, which indicates that a higher
12 conversion is achieved with an increase of the glycol vapor feed flow rate.
13 Figure 7 illustrates the conversion of carboxylic acid in the reactive distillation column for
14 different operating conditions. Figure 7 (a) shows the influence of column temperature on the
15 conversion for experiments E1, E2, E3 and E4. It can be seen that the conversion increases
16 significantly with an increase of the column temperature. A conversion of 21% is achieved at
17 a column temperature of 190°C which increases up to 73% at a column temperature of
18 250°C. The conversion versus the glycol vapor feed flow rate is shown in Figure 7 (b) for the
19 experiments performed at the monoester feed flow of 1.5 kg/hr and, at 230°C (exp. E3, E5
20 and E6) and 250°C (exp. E4, E7 and E8). At both temperatures, the conversion increases with
21 an increase in the glycol vapor feed flow rate. From Figure 7 it can be clearly observed that a
22 conversion of >70% is achieved at a column temperature of 250°C and a conversion of >45%
23 is achieved at a column temperature of 230°C. Figure 7 also indicates that the model
24 adequately predicts the experimental data. The mean relative error between experimental and
25 predicted conversion from the model is 9%.
26 Note that the conversion shown in Figure 7 is calculated from the acid values of the mono-
27 ester and the bottom product. Thus, this conversion refers to the conversion achieved in the
28 reactive distillation column. The overall conversion should be calculated from the acid value
29 of the reactant (maleic anhydride) and the bottom product according to eq. (17). Figure 8 (a)
30 and (b) show that a minimum conversion of 90% is achieved at a column temperature of
31 250°C while a minimum conversion of 77% is achieved at a column temperature of 230°C.
32 The model predicts 16% liquid holdup in the column. Moreover, the residence time of the
33 liquid is calculated from the liquid holdup and the liquid flow rate, which is approximately

- 11 -
 1 0.55 hours (32 minutes). This indicates that 90% conversion is achieved within 0.55 hours at
2 a temperature of 250°C in the RD column coupled with the pre-reactor; which confirms the
3 polyester formation in the reactive distillation column.
4 Furthermore, the isomerization and saturation fraction of the liquid feed and the bottom
5 product has been determined by 1H-NMR. For experiment E4, Figure 9 shows an illustrative
6 example for determination of the isomerized fraction in the liquid feed and the bottom
7 product from 1H-NMR spectra, while the isomerization percentage at various temperatures in
8 the bottom product is shown in Figure 10 (a). It can be seen that the bottom product contains
9 83% of isomerized compounds when the column is operated at 250°C and 79% of isomerized
10 compounds when the column is operated at 230°C. The mean relative error between the
11 experimental and predicted isomerization percentage from the model is 12%.
12 The concentrations of the (un-)saturated compounds are also determined from the 1H-NMR
13 peak (Umare and Chandure, 2008). Furthermore, the saturation fraction is calculated from the
14 concentration of (un-)saturated compounds present in the sample. Then the change in the
15 saturation fraction of the bottom product is calculated with respect to the saturation fraction
16 of the liquid feed. The bottom product obtained at 230°C and 250°C contains 12% and 17%
17 saturated compounds, respectively as shown in Figure 10 (b). The mean relative error
18 between the experimental and predicted saturation percentages from the model is 22%. Note
19 that the mean relative error for the saturation percentage is higher, since there are two sources
20 of error encountered from calculating the unsaturation concentration, and from calculating the
21 saturation fraction with respect to the liquid feed. For experiment E4; the number-average
22 molecular weight (MWN) and the degree of polymerization (DP) are experimentally
23 determined by GPC, and compared with the simulation results. Table 3 shows that the model
24 predicted the polymer attributes adequately.
25 Table 4 makes a comparison of the acid value, conversion, isomerization and saturation
26 percentages and polymer attributes of the bottom product – produced at a column temperature
27 of 250°C – against a polyester product typically produced in an industrial setup (Feuer et al.,
28 1954; Fradet et al., 1982; Korbar et al., 1993; Jedlovčnik et al., 1995). The conversion
29 achieved in the reactive distillation column at 250°C is close to the conversion achieved in
30 the industrial polyester process. Moreover, the isomerization and saturation fraction of the
31 bottom product is in the range of the polyester produced in industry. The number-average
32 molecular weight and the degree of polymerization of the bottom product produced in the
33 reactive distillation column are lower as compared to the polyester produced in industry, but

- 12 -
 1 this is due to lower conversion that was achieved in the pilot reactive distillation column. The
2 polymer attributes of the polyester produced in the RD column are expected to be comparable
3 to the polyester produced in industry, if a conversion of 95% is achieved in the RD column.
4 The examination of the liquid feed, bottom and top product samples revealed that color of the
5 liquid feed and the bottom product are similar. However, the color of bottom product is a
6 little darker yellow compared to the pale yellow color of monoester feed. Since the bottom
7 product sample is taken in open air, an oxidation reaction may have occurred that influenced
8 the color of the bottom product. Moreover, the distillate is colorless which indicates that a
9 good separation took place in the column.
10 The experimentally determined water fraction of the distillate is compared with the predicted
11 water fraction of the distillate from the model as shown in Figure 11. It can be seen that the
12 model predicted the water fraction of distillate within the range of experimental data. The
13 mean relative error between experimental and predicted water fraction from model is 14%.
14 Figure 11 also shows that the water concentration in the distillate reduces with the increase in
15 temperature. Note that in fact, the absolute amount of water may remain constant or even
16 increase with the increase of temperature. However, this effect is hidden since an increase in
17 the temperature also increases the glycol evaporation rate, and thereby the glycol content in
18 the distillate. Moreover, the excess un-reacted glycol will strip out from column since the RD
19 column is operated isothermally and well above the boiling point of glycol. For this reason,
20 the overall water concentration is reduced due to this dilution effect of the evaporated glycol.
21 Note that this effect diminishes when full conversion is achieved in the column.
22
23 4. Conclusions
24 The polyester synthesis was successfully carried out in a pilot scale RD column equipped
25 with Sulzer BX gauze packing, using two configurations: 1. a RD column alone and 2. a RD
26 column coupled with a pre-reactor. In case of the RD column alone, a maximum conversion
27 of 37% was achieved – due to the rather short residence time (0.32 hours) and an operating
28 temperature of 190°C – which indicates monoester formation in the RD column. In the case
29 of the RD column coupled with a pre-reactor, a conversion of 90% was achieved within 0.55
30 hours, at a temperature of 250°C – which confirms the polyester formation in the RD column.
31 The extended rate-based model developed in previous work was effectively used to simulate
32 the pilot-scale RD column. The model predicted adequately the experimental data – in terms
33 of acid value, conversion, isomerization and saturation fraction, number-average molecular

- 13 -
 1 weight, degree of polymerization and water fraction in the distillate – within an error range of
2 5-22%. Remarkably, the product specifications of the polyester produced at 250°C in the
3 pilot-scale RD column is in the same range as the polyesters produced in the conventional
4 industrial setup. Furthermore, discoloration of the polyester was hardly noticed, even though
5 the column was operated at a relatively high temperature of 250°C. From the experimental
6 validation of the RD process for polyester synthesis, we conclude that unsaturated polyesters
7 can be produced in a RD column with a short residence time, leading to comparable product
8 specifications as the polyesters produced in industry.
9
10 Acknowledgment
11 The financial support of the Institute for Sustainable Process Technology (ISPT project SC-
12 00-005) and industrial partners (DSM, AkzoNobel) is greatly acknowledged. The technical
13 and experimental support from Wouter Gerritsen is also acknowledged and appreciated.
14
15 Appendix
16 This appendix provides a summary of the model parameters related to reaction rates, kinetics,
17 and binary interaction parameters used in the property model. More detailed information can
18 be found in our earlier work (Shah et al., 2011a, 2011b, 2012a, 2012b).
19 Table 5 presents the chemical reactions, as well as the reaction rate equations corresponding
20 to each reaction, while Table 6 lists the kinetic parameters (pre-exponential factor and
21 activation energy) for the poly-esterification reaction between maleic anhydride and pro-
22 pylene glycol. In addition, Table 7 provides the binary interaction parameters (BIP) for the
23 Poly-NRTL property model used in the Aspen Custom Modeler simulations of the RD setup.
24

- 14 -
 1 Notation

2 a interfacial area (m2)

3 AV acid value (mg/g)
4 C concentration of component (mol/kg)
5 Dax axial dispersion coefficient (m2/s)
6 EL, EV liquid and vapor energy holdup (J)
L
7 H liquid molar enthalpy (J/mol)
V
8 H vapor molar enthalpy (J/mol)
L
9 H partial molar enthalpy of liquid (J/mol)
V
10 H partial molar enthalpy of vapor (J/mol)
11 hL heat transfer coefficients of liquid (w/m2-k)
12 hV heat transfer coefficients of vapor (w/m2-k)
13 h height of a stage (m)
14 kl mass transfer coefficients of liquid (mol/m2-s)
15 kg mass transfer coefficients of vapor (mol/m2-s)
16 L liquid mass flow rate (kg/s)
.
17 L liquid molar flow rate (mol/s)
18 M liquid holdup (kg)

19 m [
order of reaction (-) ( m = 1 − (1 − 21−q ) X ]
1 /(1− q )
)
20 MW molecular weight (kg/kmol)
21 N normality of titer (mol/l)
22 NL liquid side mass transfer rate (mol/s)
23 NV vapor side mass transfer rate (mol/s)
.
24 Ni mass transfer flux (mol/m2/s)
25 n number of moles of component (mol)
26 P pressure (bar)
27 q adjustable exponent (-) (q = 7)
28 R rate of reaction (mol/kg-s)
29 S side stream molar flow rate (mol/s)
30 t time (s)
31 Tj temperature on stage j (K)
32 V vapor flow rate (mol/s)
.
33 V volume of titer (l)
34 v liquid velocity (m/s)

- 15 -
 1 W weight of sample (kg)
2 x liquid mole fraction (mol/mol)
3 X Conversion (-)
4 y vapor mole fraction of (mol/mol)
5 Greek letters
6 εj reaction volume on stage j (kmol)
7 γI,j activity coefficient of component i on stage j (-)
8 νI,m stoichiometry coefficient of component i in reaction m (-)
9 τ residence time of the liquid in the column (hr)
10 Subscripts
11 F feed stage
12 i component indices
13 j stage indices
14 R reflux
15 Superscripts
16 * equilibrium
17 L liquid phase
18 V vapor phase
19 sat saturated
20 Abbreviations
21 COOH acid end group
22 FA fumaric acid
23 MA maleic acid
24 MAD maleic anhydride
25 OH hydroxyl group
26 POLY polyester
27 POLY1D maleate formed polyester
28 POLY2D fumarate formed polyester
29 POLYs saturated polyester
30 PG propylene glycol
31 SACID saturated acid
32

- 16 -
 1 References
2 1. Baur R., Higler A. P., Taylor R., Krishna R., Comparison of equilibrium stage and
3 nonequilibrium stage models for reactive distillation, Chemical Engineering Journal, 76
4 (2000), 33-47.
5 2. Bisowarno B. H., Tian Y. C., Tade M. O., Application of side reactors on ETBE reactive
6 distillation, Chemical Engineering Journal, 99 (2004), 35-43.
7 3. Bravo J. L., Rocha J. A., Fair J. R., A comprehensive model for the performance of
8 columns containing structured packings, IChemE Symposium Series, 128 (1992), A439.
9 4. Curtis L. G., Edwards D. L., Simons R. M., Trent P. J., Von Bramer P. T., Investigation
10 of maleate-fumarate isomerization in unsaturated polyesters by nuclear magnetic
11 resonance, Industrial & Engineering Chemistry Product Research and Development, 3
12 (1964), 218-221.
13 5. Feuer S. S., Bockstahler T. E., Brown C. A., Rosenthal I., Maleic-fumaric isomerization
14 in unsaturated polyesters, Industrial & Engineering Chemistry, 46 (1954), 1643-1645.
15 6. Fradet A., Marechal E., Study on models of double bond saturation during the synthesis
16 of unsaturated polyesters, Die Makromolekulare Chemie, 183 (1982), 319-329.
17 7. Harmsen G. J., Reactive distillation: The front-runner of industrial process intensification:
18 A full review of commercial applications, research, scale-up, design and operation,
19 Chemical Engineering and Processing, 46 (2007), 774-780.
20 8. Harmsen G. J., Process intensification in the petrochemicals industry: Drivers and hurdles
21 for commercial implementation, Chemical Engineering and Processing, 49 (2010), 70-73.
22 9. Hiwale R. S., Bhate N. V., Mahajan Y. S., Mahajani S. M., Industrial applications of
23 reactive distillation: Recent trends, International Journal of Chemical Reactor
24 Engineering, 2 (2004), R1.
25 10. Jedlovčnik R., Šebenik A., Golob J., Korbar J., Step-growth polymerization of maleic
26 anhydride and 1,2-propylene glycol, Polymer Engineering & Science, 35 (1995), 1413-
27 1417.
28 11. Keller T., Holtbruegge J., Gorak A., Transesterification of dimethyl carbonate with
29 ethanol in a pilot-scale reactive distillation column, Chemical Engineering Journal, 180
30 (2012), 309-322.
31 12. Kiss A. A., Bildea C. S., A review on biodiesel production by integrated reactive
32 separation technologies, Journal of Chemical Technology and Biotechnology, 87 (2012),
33 861-879.

- 17 -
1 13. Korbar J., Golob J., Šebenik A., Process of unsaturated polyester resin synthesis on a
2 laboratory and industrial scale, Polymer Engineering & Science, 33 (1993), 1212-1216.
3 14. Kushalkar K. B., Pangarkar V. G., Liquid holdup and dispersion in packed columns,
4 Chemical Engineering Science, 45 (1990), 759-763.
5 15. Luyben W. L., Yu C. C., Reactive distillation design and control, Wiley-AIChE, 2008.
6 16. Malone M. F., Huss R. S., Doherty M. F., Green Chemical Engineering Aspects of
7 Reactive Distillation, Environ. Sci. Technol., 37 (2003), 5325-5329.
8 17. Parker E. E., Unsaturated polyesters, Industrial & Engineering Chemistry, 58 (1966), 53-
9 58.
10 18. Schoenmakers H. G., Bessling B., Reactive and catalytic distillation from an industrial
11 perspective, Chemical Engineering and Processing, 42 (2003), 145-155.
12 19. Shah M., Zondervan E., Oudshoorn M. L., de Haan A. B., A novel process for the
13 synthesis of unsaturated polyester, Chemical Engineering and Processing, 50 (2011a),
14 747-756.
15 20. Shah M., Zondervan E., de Haan A. B., Process modeling for the synthesis of unsaturated
16 polyester, Polymer Engineering & Science, 51 (2011b), 2495-2504.
17 21. Shah M., Kiss A. A., Zondervan E., de Haan A. B., A systematic framework for the
18 economical and technical evaluation of reactive distillation processes, Chemical
19 Engineering and Processing, 60 (2012a), 55-64.
20 22. Shah M., Kiss A. A., Zondervan E., de Haan A. B., Influence of liquid back mixing on a
21 kinetically controlled reactive distillation process, Chemical Engineering Science, 68
22 (2012b), 184-191.
23 23. Shah M., Kiss A. A., Zondervan E., de Haan A. B., Extended rate-based model validation
24 for polyester synthesis by reactive distillation, Computer Aided Chemical Engineering, 30
25 (2012c), 1182-1186.
26 24. Smejkal Q., Kolena J., Hanika J., Ethyl acetate synthesis by coupling of fixed-bed reactor
27 and reactive distillation column – Process integration aspects, Chemical Engineering
28 Journal, 154 (2009), 236-240.
29 25. Stankiewicz A., Reactive separations for process intensification: an industrial perspective,
30 Chemical Engineering and Processing, 42 (2003), 137-144.
31 26. Svandova Z., Markos J., Jelemensky L., Impact of mass transfer coefficient correlations
32 on prediction of reactive distillation column behaviour, Chemical Engineering Journal,
33 140 (2008), 381-390.

- 18 -
 1 27. Sundmacher K., Kienle A. (Eds.), Reactive distillation: Status and future directions,
2 Wiley-VCH, Weinheim, 2003.
3 28. Umare S. S., Chandure A. S., Synthesis, characterization and biodegradation studies of
4 poly(ester urethane)s, Chemical Engineering Journal, 142 (2008), 65-77.
5 29. Taylor R., Krishna R., Modelling reactive distillation, Chemical Engineering Science, 55
6 (2000), 5183-5229.
7 30. Tuchlenski A., Beckmann A., Reusch D., Dussel R., Weidlich U., Janowsky R., Reactive
8 distillation – Industrial applications, process design & scale-up, Chemical Engineering
9 Science, 56 (2001), 387-394.
10

- 19 -
 1 Tables
2
3 Table 1. Operating parameters of experiments
Temperature Pressure MAD flow PG flow
(oC) (bar) (kg/hr) (kg/hr)
195 1 2.12 1.98
4
5
6
7
8 Table 2. Operating parameters of experiments
Exp. no. Temperature Pressure Monoester flow PG flow Acid value of
(oC) (bar) (kg/hr) (kg/hr) monoester (mg/g)
E1 195 1 1.5 0.25 250
E2 200 1 1.5 0.25 237
E3 230 1 1.5 0.25 235
E4 250 1 1.5 0.25 217
E5 230 1 1.5 0.5 224
E6 230 1 1.5 0.75 224
E7 250 1 1.5 0.5 224
E8 250 1 1.5 0.75 216
9
10
11
12
13
14
15 Table 3. Polymer attributes of a bottom product for experiment E4
Product specifications Experimental Simulation
Average-molecular weight number [-] 1404 1497
Degree of polymerization [-] 8.89 9.47
16
17
18

- 20 -
1 Table 4. Comparison of product specifications
Product specifications Bottom product Polyester product
produced at 250°C produced in industry*
Acid value [mg/g] 58 25
Conversion [-] 90 95
Isomerization [%] 83 90-95
Saturation [%] 17 15-20
Average-molecular weight number [-] 1404 3100
Degree of polymerization [-] 8.89 25
2
*
3 Note: As previously reported in literature by Korbar et al. (1993), Jedlovčnik et al. (1995),
4 Feuer et al. (1954), Fradet and Marechal (1982).
5
6 Table 5. Chemical reactions and rate equations
Reactions Rate equations

Esterification reactions:

MA + PG POLY1D + H2O r1 = Cm-1MA (k1CMACPG– k’1CPOLY1DCH2O) (19)

FA + PG POLY2D + H2O r2 = Cm-1MA (k2CFACPG– k’2CPOLY2DCH2O) (20)

SACID+ 0.5PG POLYS + H2O r3 = Cm-1MA (k9CSACIDCPG– k’9CPOLYSCH2O) (21)

Isomerization reactions:

MA FA r4 = Cm-1MA (k3CMA– k’3CFA) (22)

POLY1D POLY2D r5 = Cm-1MA (k4CPOLY1D– k’4CPOLY2D) (23)

Saturation reactions:

MA + 0.5 PG SACID r6 = Cm-1MA (k5CMACPG– k’5CSACID) (24)

FA + 0.5PG SACID r7 = Cm-1MA (k6CFACPG – k’6CSACID) (25)

POLY1D + 0.5PG POLYS r8 = Cm-1MA (k7CPOLY1DCPG – k’7CPOLYS) (26)

POLY2D + 0.5PG POLYS r9 = Cm-1MA (k8CPOLY2DCPG– k’8CPOLYS) (27)

- 21 -
1 Table 6. Kinetic parameters for the poly-esterification reaction: MA + PG
Reaction Forward reaction Backward reaction
ko Ea ko’ Ea
-1 -1 -1 -1
(kg mol hr ) (J/mol) (kg mol hr ) (J/mol)
Esterification 72000000 75000 37200 59000
Isomerization 7620000 56000 10680 41700
Saturation 16380 47000 10560 49600
2
3 Table 7. Binary interaction parameters for the Poly-NRTL property model

Component i Component j bij bji

PG H2O -248.25 790.27
PG FA 303.67 -229.11
PG SACID 398.41 -217.16
PG POLYS 431.80 -221.43
PG POLY1D 540.32 -129.89
PG POLY2D 201.60 93.53
H2O FA 761.89 -354.33
H2O SACID 1421.77 -567.46
H2O POLYS 2109.51 -509.44
H2O POLY1D 1374.88 -266.50
H2O POLY2D 1565.88 272.87
FA SACID -597.18 965.66
FA POLYS -1113.58 3084.73
FA POLY1D -631.65 1126.21
FA POLY2D 520.63 -229.98
SACID POLYS -108.42 109.36
SACID POLY1D -56.25 290.92
SACID POLY2D -56.25 290.92
POLYS POLY1D 581.21 -416.24
POLYS POLY2D 581.21 -416.24
POLY1D POLY2D 128.83 -114.78
MA POLY1D -631.65 1126.21
MA POLY2D 520.63 -229.98
MA FA 128.83 -114.78
MA SACID -597.18 965.66
PG MA 303.67 -229.11
H2O MA 761.89 -354.33
MA POLYS -1113.58 3084.73
4

- 22 -
 1 Figure captions
2
3 Figure 1. Schematic view of configuration 1 and 2: reactive distillation column (a) and RD
4 column coupled with a pre-reactor (b)
5
6 Figure 2. Schematics of the reactive distillation pilot plant setup
7
8 Figure 3. Photo of the pilot plant setup
9
10 Figure 4. Acid value (a) and conversion (b) profiles over the column height
11
12 Figure 5. Acid value of the monoester in the liquid feed stream
13
14 Figure 6. Acid value of the bottom product leaving the column for experiments E1, E2, E3,
15 E4 (a), for experiments E3, E5, E6 (b) and for experiments E4, E7, E8 (c)
16
17 Figure 7. Conversion as function of temperature for experiments E1 to E4 (a) and conversion
18 as function of glycol vapor feed rate for experiments E3 to E8 (b)
19
20 Figure 8. Overall conversion as a function of temperature for experiments E1 to E4 (a) and
21 overall conversion as function of glycol vapor feed rate for experiments E3 to E8 (b)
22
23 Figure 9. 1H-NMR spectra of maleate-fumarate isomerization in the liquid feed (a) and the
24 bottom product (b)
25
26 Figure 10. Isomerization percentage (a) and saturation percentage (b) in the bottom product
27 for experiments E1 to E4
28
29 Figure 11. Water fraction in the distillate as a function of column temperature for
30 experiments E1 to E4 (a) and glycol vapor feed rate for experiments for E3 to E8 (b)
31

- 23 -
 maleic propylene
maleic anhydride glycol
anhydride

water,
propylene glycol, water,
acid propylene glycol,
monoester acid

(a) (b)

propylene polyester
glycol propylene polyester
glycol
1
2 Figure 1. Schematic view of configuration 1 and 2: reactive distillation column (a) and RD
3 column coupled with a pre-reactor (b)
4

A – Glycol feed tank

B – Anhydride feed tank
C – Bottom product tank
D – Top product tank
E – Pump
F – Flow meter
G – Evaporator
H – Oil bath
I – Condenser
J – Mass flow control valve
T – Temperature indicator
s – Sample point
dpi – Differential pressure indicator

5
6 Figure 2. Schematics of the reactive distillation pilot plant setup

- 24 -
1
2 Figure 3. Photo of the pilot plant setup
3

600 0.4
Acid value / [mg/g]

(a) (b)
550
Conversion / [-]

0.3
500
450 0.2
400
simulation 0.1 simulation
350 experimental experimental
300 0
0 0.3 0.6 0.9 1.2 1.5 1.8 0 0.3 0.6 0.9 1.2 1.5 1.8
Column height / [m] Column height / [m]
4
5 Figure 4. Acid value (a) and conversion (b) profiles over the column height
6

300
Acid value / [mg/g]

280
260
240
220
200
1 2 3 4 5 6 7 8
Time / [hr]
7
8 Figure 5. Acid value of the monoester in the liquid feed stream

- 25 -
 1
250
(a)

Acid value / [mg/g]

200
190 C
150
200 C
100 230 C
250 C
50
L = 1.5 kg/hr, V = 0.25 kg/hr
0
1 2 3 4 5 6 7 8
Samples
140 65
(b) (c)

Acid value / [mg/g]

Acid value / [mg/g]

L = 1.5 kg/hr, T = 230 C L = 1.5 kg/hr, T = 250 C

130 60
120
V = 0.25 kg/hr 55 V = 0.25 kg/hr
110 V = 0.5 kg/hr V = 0.5 kg/hr
50
100 V = 0.75 kg/hr V = 0.75 kg/hr
90 45

80 40
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Samples Samples
2
3 Figure 6. Acid value of the bottom product leaving the column for experiments E1, E2, E3,
4 E4 (a), for experiments E3, E5, E6 (b) and for experiments E4, E7, E8 (c)
5
6
7
8
9
0.8 1.0
experimental exp. T = 230 C sim. T = 230 C
Conversion / [-]
Conversion / [-]

0.7 0.9
simulation exp. T = 250 C sim. T = 250 C
0.6 0.8
0.5
0.7
0.4
0.6
0.3
0.2 0.5
(a) (b)
0.1 0.4
190 210 230 250 0.1 0.3 0.5 0.7 0.9
o
Temperature / [ C] Vapor flowrate / [kg/hr]
10
11 Figure 7. Conversion as function of temperature for experiments E1 to E4 (a) and conversion
12 as function of glycol vapor feed rate for experiments E3 to E8 (b)

- 26 -
 1
1 1
(a) (b)
Conversion / [-]

Conversion / [-]
0.9 0.9

0.8 0.8

0.7 0.7 T = 250 C

T = 230 C
0.6 0.6
190 210 230 250 0.1 0.3 0.5 0.7 0.9
o Vapor flowrate / [kg/hr]
Temperature / [ C]
2
3 Figure 8. Overall conversion as a function of temperature for experiments E1 to E4 (a) and
4 overall conversion as function of glycol vapor feed rate for experiments E3 to E8 (b)
5
6
7
8

maleate fumarate

(a) (b)

maleate
fumarate

9
10 Figure 9. 1H-NMR spectra of maleate-fumarate isomerization in the liquid feed (a) and the
11 bottom product (b)
12

- 27 -
 1
100 20
Isomerization / [%]

(a) (b)

Saturation / [%]
90
80 15
70
10
60
50 experimental 5 experimental
40 simulation simulation
30 0
190 210 230 250 190 210 230 250
Temperature / [oC] Temperature / [oC]
2
3 Figure 10. Isomerization percentage (a) and saturation percentage (b) in the bottom product
4 for experiments E1 to E4
5
6
7
8
9
10
11
12
13
14
15
16
17
0.6 0.6
Water fraction / [-]
Water fraction / [-]

simulation sim. T = 250 C

0.55 0.5
experimental exp. T = 250 C
0.5 0.4 sim. T = 230 C
0.45 exp. T = 230 C
0.3
0.4 0.2
0.35 0.1
(a) (b)
0.3 0
190 210 230 250 0.1 0.3 0.5 0.7 0.9
Temperature / [°C] Vapor flow / [kg/hr]
18
19 Figure 11. Water fraction in the distillate as a function of column temperature for
20 experiments E1 to E4 (a) and glycol vapor feed rate for experiments for E3 to E8 (b)
21

- 28 -

View publication stats