Calorimetric Approach and Simulation for Scale-Up of a Friedel-Crafts Reaction

Article abstract of DOI:10.1021/op030043a

The present study, on the optimisation of the scale-up of a Friedel-Crafts acylation, joins together experimental calorimetric techniques and simulation techniques. The adopted methodology involved the execution of isothermal tests at different temperatures, using the reaction calorimeter RC1, to obtain the heat flow versus time data. These data were used to perform a kinetic study of the reaction by using the simulation software BatchCAD. The reliability of the adopted kinetic model was confirmed by comparing the experimental data obtained by operating the desired reaction in adiabatic mode (in a PhiTec II calorimeter) with the predictions of the kinetic model in the same conditions. Using this kinetic equation it was possible to simulate the process under pilot-plant conditions. The good agreement between the model predictions and the experimental data confirmed the accuracy of the kinetic equations. Finally several large-scale production-plant simulations were carried out varying both dosing time and reaction temperature to maximise the space/time yield and minimise the thermal risks. This led to a high quality and safe process, saving a large part of the time (and money) usually spent in traditional scale-up procedures.

Full text of DOI:10.1021/op030043a

Organic Process Research & Development 2003, 7, 1079−1082
Calorimetric Approach and Simulation for Scale-Up of a Friedel-Crafts
Reaction
Lucia Gigante, Angelo Lunghi, Simone Martinelli, and Paolo Cardillo*
Stazione sperimentale per i Combustibili, V.le A. De Gasperi, 3-20097 San Donato Milanese, Italy
Luca Picello, Roberto Bortolaso, and Marco Galvagni
Fabbrica Italiana Sintetici, Via Milano, 26-36041 Alte di Montecchio Maggiore, Vicenza, Italy
Renato Rota
Politecnico di Milano, Dip. di Chimica, Materiali e Ingegneria Chimica G. Natta, Via Mancinelli 7-20131 Milano, Italy
Abstract:
and environment. Both safety and economical aspects are
critical points, particularly for SMEs.
The present study, on the optimisation of the scale-up of a
Friedel-Crafts acylation, joins together experimental calori-
metric techniques and simulation techniques. The adopted
methodology involved the execution of isothermal tests at
different temperatures, using the reaction calorimeter RC1, to
obtain the heat flow versus time data. These data were used to
perform a kinetic study of the reaction by using the simulation
software BatchCAD. The reliability of the adopted kinetic model
was confirmed by comparing the experimental data obtained
by operating the desired reaction in adiabatic mode (in a Phi-
Tec II calorimeter) with the predictions of the kinetic model in
the same conditions. Using this kinetic equation it was possible
to simulate the process under pilot-plant conditions. The good
agreement between the model predictions and the experimental
data confirmed the accuracy of the kinetic equations. Finally
several large-scale production-plant simulations were carried
out varying both dosing time and reaction temperature to
maximise the space/time yield and minimise the thermal risks.
This led to a high quality and safe process, saving a large part
of the time (and money) usually spent in traditional scale-up
procedures.
The main aim of the present study was to develop a kinetic
model for a Friedel-Crafts reaction using calorimetric data,
obtained from the Mettler RC1 and other calorimetric
1
techniques. This kinetic model was used with a commercial
2
,3
computer program (BatchCAD ) to simulate and optimise
the actual process in terms of productivity and safety.
Methodology
Study of the Desired Reaction. In this work we have
4
studied the Friedel-Crafts acylation of toluene with succinic
3
anhydride and AlCl as catalyst at low temperature.
The reaction scheme is shown as the following:
Preliminary DSC runs on the final reaction mass, reagents,
and products showed that all samples do not decompose or
oxidise exothermically within the temperature range 30-
Introduction
2
80 °C during a standard dynamic test (scanning rate 5 °C/
The Fine Chemical Industries are increasingly under huge
pressure because of increasing R&D costs and growing
competition and pricing pressures; therefore, they need to
speed-up the time-to-market to achieve market leadership.
In Italy, most of the Chemical Industry is constituted by
small- and medium-sized enterprises (SMEs) with a limited
number of workers in each firm, and with the strong tendency
to resort to outsourcing. This situation often prevents the
complete identification of all the kinetic parameters related
to the chemical production of new compounds that are
essential to ensure the company’s productivity and to comply
with stricter and stricter laws on health and safety for people
min, static air atmosphere, sealed steel crucible, medium
pressure). Therefore, we can assume that the final reaction
mass is stable up to 280 °C.
The kinetic behaviour of the reaction system was studied
by carrying out a series of isothermal reaction calorimeter
(RC) experiments. The tests were run in batch mode, adding
all succinic anhydride to the organic substrate (toluene) at
the beginning of the experiment, using toluene as solvent
(1) http://www.rxeforum.com/start_8_16-0.htm. Accessed 2003.
(2) BatchCAD, version 7.2; http://www.hyprotech.com/batchcad/default.asp.
Accessed 2003.
(
3) Lunghi, A.; Al o´ s, M. A.; Gigante , L.; Feixas, J.; Sironi, E.; Feliu, J. A.;
Cardillo, P. Org. Process Res. DeV. 2002, 6, 926.
*
Author for correspondence. Telephone +39 02 51604209. Fax +39 02
(4) Groggings, P. H. Unit Process in Organic Synthesis, 3rd ed.; McGraw-
5
14286. E-mail: cardillo@ssc.it.
Hill: New York, 1947.
1
0.1021/op030043a CCC: $25.00 © 2003 American Chemical Society
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
1079
Published on Web 11/06/2003

Figure 1. Experimental heat flow from RC1 tests.
Table 1. RC1 experiments at 10, 20, 30 °C
Figure 2. BatchCAD-fitted plot for the RC1 isothermal
experiment at 20 °C.
Table 2. Estimate of reaction order
∆
H
reac referred
specific heat
of reaction
mass
process
∆Hreac
to succinic
anhydride
[kJ/mol]
reaction
temperature obtained
∆Tad
temperature
∆Hr1
∆Hr2
[
°C]
[kJ]
[J/(kgmix × K)] [°C]
[°C]
k1 [1/s]
k2 [1/s]
m
n
[kJ/mol] [kJ/mol]
-
3
3
3
-3
1
2
3
0
0
0
-60.52
-64.09
-61.11
-86.46
-91.55
-87.29
1623
1739
1716
50.1
49.5
47.9
10
20
30
2.8 × 10
4.3 × 10
6.5 × 10
8.3 × 10
2.6 × 10
7.5 × 10
1.5 0.6
1.0 0.8
1.7 0.9
10.0
10.0
10.0
-96.5
-101.6
-97.3
-
-
-2
-2
Table 3. Estimate of kinetic constants
reaction temperature [°C] k1 [1/s]
and aluminium chloride as catalyst to yield 3-(4-methyl
benzoyl) propionic acid.
k2 [1/s]
The RC1 tests showed that the reaction is exothermic:
the shape of the RC1 curves obtained is typical of a reaction
that accumulates a certain percentage of reactant (Figure 1).
Accumulation increases when decreasing temperature. The
reaction is usually performed close to 0 °C. Table 1
summarises the calorimetric data at three different temper-
atures deduced from the experiments summarised in Figure
-
-
-
3
3
3
-2
-2
-2
10
20
30
2.4 × 10
4.3 × 10
7.2 × 10
3.9 × 10
4.2 × 10
6.3 × 10
Table 4. Estimate of Arrhenius parameters for the acylation
A [m kmol s-1
3
-1
]
E1 [kJ/mol]
m
1
1
.
nucleophile attack
product formation
37406
38.960
1
n
As the measured ∆Tad is about 50 °C, the final temperature
3
-1
A
2
[m kmol s-1
]
E
2
[kJ/mol]
for the reaction mass cannot exceed 280 °C. Since DSC tests
showed that the reaction mass is stable up to 280 °C, from
these preliminary RC1 tests we can deduce that, in these
conditions, the process does not show any runaway hazard.
The equations proposed to describe the system were:
38.420
16.332
1
give the best fit of the heat flow versus time curves were
determined using the “kinetic fitting environment” of the
software BatchCAD from each isothermal experiment, as
reported in Table 2.
1
C H O + C H + AlCl
9
8 activated complex + AlCl3
4
4
3
7
8
3
2
activated complex 98C H O
11 12 3
After the reaction orders m and n were determined from
all isothermal experiments, they were rounded and fixed as
n ) 1 and m ) 1, and the fitting procedure was repeated to
determinate the isothermal rate constants at three different
temperatures, as summarised in Table 3. For the sake of
example, Figure 2 shows a comparison between RC1
isothermal data and model simulation at 20 °C.
From the data summarised in Table 3, activation energy
and preexponential factor values for the two reactions were
determined according to the Arrhenius equation. The calcu-
lated values are summarised in Table 4 and Figure 3.
These kinetic data were assumed to model the thermal
behaviour of the studied reaction. Kinetic equation was
verified by comparison of simulated adiabatic Phi-Tec II runs
The reaction rate equations proposed for these reactions are
the following:
m
r ) k [C H O ]
1
1
4
4
3
n
r ) k [activated complex]
2
2
This simplified reaction scheme, according to literature,4
involves the formation of a not-isolating substance, the so-
called “activated complex”, and then the formation of the
product. The influence of the toluene concentration on the
reaction rates was not investigated since in all the experi-
ments toluene was in large excess with respect to the
stoichiometry. The partial reaction orders with respect to the
succinic anhydride, m, and to the activated complex, n, which
1080
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development

Figure 5. Pilot plant sketch. Fluid conditions are steam at 127
C, and cooling water-glycol ether at -20 °C.
°
Figure 3. Arrhenius plot of the experimental data.
Figure 6. Comparison between pilot-plant experiments and
model predictions.
Figure 4. Comparison between predicted and experimental
data under adiabatic conditions.
to -4 °C, 75 kg of succinic anhydride is charged in semi-
batch mode, during 7.5 h. At the end of the dosing, the
temperature is raised to 25 °C. Under these conditions, no
accumulation of succinic anhydride occurs in the reactor,
and temperature profile could be classified as QFS (quick
onset, fair conversion and smooth profile).
The process has been then simulated under normal
conditions, comparing the real and the simulated data. Figure
Table 5. Phi-Tec II experimental conditions
toluene
35 g
10.5 g
4 g
adiabatic
200 rpm
1.35
aluminum chloride
succinic anhydride
conditions
stirrer
Φ factor
7
6
shows reactor temperature and addition as a function of
time in the two cases.
(
using BatchCAD) and actual Phi-TecII5,6 test cell data. The
Phi-Tec II experimental conditions are summarised in Table
, while the observed and calculated adiabatic temperature
We can see that experimental and predicted profiles are
quite similar, thus confirming the reliability of the developed
kinetic model.
5
plots are shown in Figure 4.
The good agreement between experimental and predicted
data confirms the reliability of the developed kinetic model.
Optimisation of the Pilot-Plant Process by Simulation
According to the DSC tests and to RC1 experimentation,
the process does not involve any thermal hazard. The final
reaction mass is thermally stable in the 30-280 °C range of
Pilot-Plant Process and Simulation
3
Pilot-plant process was carried out using a 1.5 m steel
temperature, and the measured ∆T of the desired reaction
ad
jacketed reactor. The heat-transfer system is composed by
is 50 °C which leads to a maximum final temperature of
3
two service fluids, which are fed into a 0.21 m jacket on a
about 50 °C in the pilot-plant conditions. However, such
operating conditions require large times for cooling, dosing,
and heating. Consequently, several large-scale production-
plant simulations using the BatchCAD simulation environ-
ment were run to optimise the process in terms of produc-
tivity without introducing any thermal hazard.
At first, dosing time was strongly reduced, and the final
heating was eliminated. In this case the optimal configuration
found, in terms of both safety and productivity, was a dosing
drain/fill mode. A manual controller allows heating and
cooling the reactor to the required operation temperature.
Figure 5 illustrates the reactor configuration.
During normal operation conditions, the reactor is charged
with 649.5 kg of toluene, 199.9 kg of aluminium chloride,
and after 2.5 h, during which reactor temperature is brought
(
5) Singh, J. Enhanced vent sizing calorimeter: application and comparison
with existing devices. International Symposium on Runaway Reactions,
CCPS, Boston, MA, 1989; p 313.
(6) http://www.helgroup.co.uk. Accessed 2003.
(7) Steensma, M., Westerterp, K. R. Ind. Eng. Chem. Res. 1990, 29, 1259.
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
1081

Figure 7. Simulated temperature profile for a semibatch
process with dosing time equal to 0.5 h.
Figure 10. Temperature profile in batch reactor with initial
temperature equal to 20°C.
temperature profiles in a batch reactor operating at the same
temperature and charge conditions as the semi-batch reactor.
We can see that also in the batch mode the reactor
temperature is lower than 20 °C during the whole process.
This means that, as previously discussed, the process can
be performed in safe conditions also in a batch reactor.
Furthermore comparing the conversion of the succinic
anhydride in semi-batch and batch mode (Figure 9) we can
see that, as expected, the batch mode leads to a complete
conversion in a shorter time.
Even more severe conditions can be utilised without
compromising the process safety, as shown in Figure 10,
where the reactor temperature profile in batch mode and an
initial temperature of 20 °C is reported.
Comparing the results of this last figure with those of
the normally adopted operating conditions (reported in Figure
Figure 8. Temperature profile in batch reactor with initial
temperature equal to 0°C.
6
), we can note the large reduction in the process time
without compromising the process safety.
Conclusions
This study presents an application of coupling calorimetric
experiments with simulation models to obtain a short-cut
methodology able to optimise the scale-up of chemical
processes, both in safety and economic terms. Limited
targeted experiments with Mettler RC1 along with the
capabilities of the software BatchCAD easily produced a
reliable kinetic model that has been verified by comparison
with adiabatic calorimeter and pilot-plant data. The simula-
tion in different industrial conditions has been used to
optimise the whole process in terms both of productivity and
safety.
Good characterisation of the reactant system allows a
quicker scale-up, promptly recognising any mistakes and
inadequacy. In this particular case the applied methodology
permitted drastic reduction of the reaction time and avoidance
of the initial cooling period.
Figure 9. Comparison of conversion for batch and semi-batch
reactors.
time equal to 0.5 h as shown in Figure 7. We can see that
also in this case no safety problems arise, while the reaction
is completed in about 1 h after completing the dosing of
succinin anhydride.
Therefore, the more severe batch conditions were inves-
tigated. Figure 8 shows the simulated reactor and jacket
Received for review September 9, 2003.
OP030043A
1082
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development