Extended Abstract

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Operational Safety of the Polymerization Reactors

S. Silva, J. Bordado, R. Luís, P. Correia

Instituto Superior Técnico, Av. Rovisco Pais 1, 1049-001 Lisboa, Portugal

SGL Composites, S.A., Lavradio, 2835-908 Barreiro, Portugal

Abstract
At SGL Composites, S.A., the formation and constant safety awareness aims to make known
the right procedures, as well as to develop and adjust the employees’ behavior. However, it is
also necessary to conduct a detailed risk analysis to implement the right specific control
instrumentation, in order to correct all the possible perturbations so the stationary state can be
achieved as soon as possible. In the scope of FM Global, it was found necessary to evaluate the
polymerization reactors’ zone to comprehend if there is any possibility/probability, even slightly,
of explosion or runaway, despite the existing control instrumentation. Therefore, after studying
the polymerization area and the chemical reactivity of the polymerization system – reaction
kinetics and an adiabatic system simulation –, it was determined and analyzed the worst-case
scenarios that are possible to occur in the polymerization process, using the software Aspen Plus
and the Pilot Unit of the firm. Hereupon, a critical review of the control instrumentation was
proceeded and it was proposed to be considered the installation of some devices, such as a level
sensor, a bubble separator, a filter, among others. Lastly, it was possible to verify that SGL
Composites, S.A. respects all the Process Safety Management umbrella relevant aspects, always
considering the safety, the quality and the environment.

Keywords: acrylonitrile, vinyl acetate, polyacrylonitrile, acrylic fibers, acrylonitrile free-radical


copolymerization

Introduction
nitrile rubber received special support due to
ACRYLONITRILE their strategic importance. This way,
Acrylonitrile was first synthetized in 1893 acrylonitrile became recognized as a
by the French Chemist Ch. Moureau, monomer of commercial importance and, in
although it had at the time no significant 1950, the acrylonitrile demand began to
technical or commercial applications until increase due to the acrylic fibers. [2, page 3]
the late 1930s. Just before the second world Nowadays, all acrylonitrile is produced
war, IG Farben industry [1] introduced a by direct catalytic conversion of propene,
synthetic rubber based on a copolymer of oxygen (as air) and ammonia, and its world
butadiene and acrylonitrile and, during the production is approximately 5 million tons
same period, in USA, projects related to per year. In addition, acrylonitrile is, today,

1
an industrial intermediate used the versatility of being able to be chemically
predominantly in the production of polymeric modified and thus to produce a wide variety
materials: acrylic fibers accounts for 60%, of materials: [3]
while the plastics accounts for 25% of world
• Fibers;
consumption. [2, page 3]
• Carbon Fibers;

POLYACRYLONITRILE • Adhesives;

Polyacrylonitrile (PAN) was first • Engineering plastics;

synthetized by Dr. º Hans Fikentscher and • Superabsorbent polymers.

Dr. º Claus Heuck, in 1930, in the


This polymer is essential for textile and
laboratories of IG Farben, in Ludwigshafen,
high technology applications due to its
a German company that patented its
special properties, such as thermal stability,
polymerization method. However, since the
high strength and modulus of elasticity, UV
polymer was insoluble in most common
degradation stability, non-fusible and
solvents, the substance was considered to
chemical resistance. Therefore, the main
be unusable. Even so, the investigations
application is in textile industry, while the
continued and, in 1931, Dr. º Herbert Rein,
second one is related to the production of
also from IG Farben but in Bitterfeld,
carbon fibers. However, there are other
discovered that PAN, obtained from a
minor applications which use is growing,
sample when visiting the Ludwigshafen
such as: [3]
plant, could be dissolved in the ionic liquid 1-
Benzylpyridinium chloride and converted • Fibers for cement reinforcement;

into fibers. Later in 1942, the same chemist • Filtration membranes;


discovered a better solvent for the polymer – • Awning fabrics and outdoor
dimethylformamide, a discovery that allowed applications;
to develop the spinning process to produce • Oxidized PAN fibers for thermal and
fibers and films. Due to the second world acoustic insulation;
war, the investigations were interrupted and • Anti-flame fibers;
the polymerization process was not used in • Felts’ manufacture for hot air
an industrial scale. [3]
filtration.

Despite the interruption, when the war


Acrylonitrile Polymerization
time was over, in 1946, DuPont introduced
the large-scale production of PAN to at SGL Composites, S.A.
produce the Orlon fibers and patented the The acrylonitrile polymerization at SGL
wet-spinning process. Later, the Dormagen Composites, S.A. is based on a free-radical
Bayer company introduced the production of reaction of the acrylonitrile and vinyl acetate.
Dralon fibers and patented the dry-spinning [4] This exothermic reaction takes place in an
process. [3]
aqueous medium and occurs continuously
under quasi steady state conditions with an
Throughout the production of PAN, it
88% conversion. The feeding aqueous
was realized that among its main features is

2
solutions are previously prepared and the redox system consists of ferrous sulfate (catalyst),
ammonium persulfate (main initiator) and sodium bisulfite (chain-transfer agent that regenerates
the oxidized form of the ferrous ion and, hence, acts simultaneously as an initiator). To keep a
constant pH, a sulfur acid filtered solution is used. [5, page 1] The polymerization reactional
mechanism is as follows: [6, page 12]

Initiation

𝑆2 𝑂82− + 𝐹𝑒 2+ → 𝑆𝑂42− + 𝑆𝑂4− ∗ + 𝐹𝑒 3+

𝐻𝑆𝑂3− + 𝐹𝑒 3+ → 𝑆𝑂3− ∗ + 𝐹𝑒 2+ + 𝐻 +

Propagation 1

Concerning the termination step, free-radical polymer chains can terminate by three mechanisms:

Chain Transfer (more likely to happen) [6, page 11]


𝑆𝑂4 (𝐶𝐻2 𝐶𝐻𝐶𝑁 ∗ )𝑛 + 𝐻𝑆𝑂3− → 𝑆𝑂4 (𝐶𝐻2 𝐶𝐻𝐶𝑁)𝑛 𝐻 − + 𝑆𝑂3− ∗

Radical Recombination 1


𝑆𝑂4 𝐶𝐻2 𝐶𝐻𝐶𝑁 ∗ − + 𝑆𝑂4 (𝐶𝐻2 𝐶𝐻𝐶𝑁 ∗ )𝑛−1 → 𝑆𝑂4 (𝐶𝐻2 𝐶𝐻𝐶𝑁)𝑛 𝑆𝑂4 2−

Disproportionation 1

𝑆𝑂4 𝐶𝐻2 𝐶𝐻𝐶𝑁 ∗ − + 𝑆𝑂4 𝐶𝐻2 𝐶𝐻𝐶𝑁 ∗ − → 𝑆𝑂4 𝐶𝐻 = 𝐶𝐻𝐶𝑁 − + 𝑆𝑂4 𝐶𝐻2 𝐶𝐻2 𝐶𝑁 −

1 Mechanisms shown and explained in a presentation at SGL Composites, S.A. on 04/2018

3
Moreover, the reactional mixture is evaluated through the reactional mixture
continuously agitated by a propeller-type temperature.
stirrer, describing an axial movement. While
the reaction occurs, the polymer suspension ASPEN PLUS
is expelled out of the respective reactor by After establishing the stationary state,
overflow. In the reactors discharge tube is the worst-case scenarios were simulated as
added the inhibitor solution to prevent the sensibility analysis and it was possible to
reaction to last. [5, pages 1 and 2] The reactional understand how a manipulated variable
temperature is controlled by the cooling impacts the variable to be studied. The
water that circulates in the cooling jackets of reactor here simulated was a 12 m3 one and,
the reactors, in a closed circuit and with a after running the known stationary state, it
high flow. was obtained an overflow temperature of
In addition, after the reaction step, the 62,0°C. The reactional mixture temperature
slurry proceeds to a column so the was measured at the overflow.
unreacted monomers can be separated.
This is followed by filtration and washing PILOT UNIT
steps and, after that, the polymer is While in Aspen Plus a 12 m3 reactor
pelletized, dried, ground and finally stored could be defined, in the Pilot Unit the
for later spinning. [7, page 836] To produce PAN experiments were carried out in a 5,7 L
fibers after obtaining the polymer, there are reactor (consequently, the industrial
those who apply the dry-spinning process, conditions had to be adjusted to this
but SGL Composites, S.A. follows the volume). To do so, the following
Mitsubishi Rayon example and applies the experimental setup was prepared:
wet-spinning process using
dimethylacetamide as solvent. Inlet streams

Overflow
Experimental Section
To diagnose the probability of runaway
or explosion at the polymerization reactor,
the worst-case scenarios were analyzed. To
Cooling water
do so, the software Aspen Plus was used to circulation
understand the behavior of the reactional
mixture towards the flow variation of the Figure 1. Experimental setup

feeding solutions, while in the Pilot Unit it


and a check-list was completed. Besides,
was studied the behavior of the reactional
the materials were yet prepared, bearing in
mixture towards the cooling and agitation
mind the following information:
loss, the lack of inert media, the pH variation
and the unexpected introduction of
unreacted MB. These deviations were

4
Table 1. Necessary quantities to prepare the solutions Regarding Pilot Unit experiments, the
msolid (g) Total (g) worst-case scenarios with more impact on
CD 1592 80000 the reactional mixture temperature were the
CE 404 80000 agitation loss, the cooling loss and the
CB 1800 80000 unexpected introduction of unreacted MB –
HS 200 19000 when each condition was tested, the reactor
11383 (AN) inner’s temperature increased infinitely and
AN+AV 14376
1013 (AV) uncontrollably and the reactional mass even
solidified. In case any of these worst-case
scenarios happens industrially, all the
Besides the preparations above, it was
control loops of the reactors zone actuate
also necessary to weight the following
automatically and have the right indicators
quantities from the previous preparations,
and alarms to warn the operators of what is
during the reactor’s pre-start:
happening at the time. Besides, all the
Table 2. Required quantities of CB and HS solutions operators well know the right procedures in
for the reactor's pre-start
case an intervention is needed.
mliquid (g)
CB (1:1000) 103 The figures Figure 2 and Figure 3,

HS 26 presented below, are an illustration of one of

DIW2 3975 the four simulations completed in Aspen and


one of the 6 experiments done in the Pilot
Unit, respectively.

Results and Discussion


Concerning Aspen simulations, from all
the variations of the feeding flows, the
variation of the MB feeding flow is the one
that impacts the reactional mixture
temperature the most (Figure 2) – the
temperature of the reactional mixture
decreases with the MB feeding flow
increasing and vice-versa. In fact, if there are
more monomers reacting, there are more
monomer radicals forming, which means
that more chains are growing at the same
time. However, the residence time remains,
so each chain does not grow as much and,
thus, the final polymer chains are smaller.
This results in polymeric chains with lower
molecular weight. The energy released is
lower than the usual and, consequently, the
temperature decreases.

5
64,0

63,5
Treactional mixture (°C)

63,0

62,5

62,0

61,5

61,0

60,5

60,0
2900 3100 3300 3500 3700 3900 4100 4300 4500

QMB (L/h)

Figure 2. Reactional mixture temperature dependence on MB feeding flow

Figure 3. Impact of the unexpected introduction of MB on the reactional mixture’s temperature

6
After testing all the worst-case scenarios and CE feeding flows: in MB case the
to understand the reactor’s behavior, the temperature variation is linear, while in CD
most common anomalies and the existing and CE cases the temperature does not
control instrumentation were analyzed so change with any flow variation. This impact
some improvements could be proposed. difference might be related to the fact that
Briefly, and knowing the improvements CD and CE are present in faster reactions
made over time, the implementations (initiation and/or termination), while MB is
proposed were: the principal reagent during propagation (a
reaction that takes up almost all the free-
• a rupture disk in the polymerization radical reaction time). Nonetheless, it cannot
reactors; [8, page 2]
be forgotten that, although in reality the
• a manual self-cleaning filter adjustment of the manipulated variables is
connected to the overflow; [9, page 2]
automatic, the possible controller lag and the
• an obstruction detector in the possibility of an operator intervention are not
overflow; considered, so the differences between the
• a level sensor near to the DSF results returned by the software and the real-
entrance; [10, page 3] life results are to be expected. Even though
• a bubble separator before the it was confirmed that, so far, the existing
flowmeters of the reactors’ feeds; control loops and the used controllers, as
• a second measurement system. well as the room operators’ training and
experience, have proved to be able to
Conclusions prevent a nefarious occurrence, such as a
runaway, the controller response can be
In the scope of FM Global, this
improved to avoid the “human error”.
dissertation was directed to the
Returning to the results, despite changing,
polymerization reactors in order to study the
almost all temperature variations are within
occurrence probability of explosion or
the alert limits – bigger MB feeding flow
runaway. This part of the thesis was based
variations might result in unacceptable
on computational simulations and
temperatures; however, those MB feeding
experimental work.
flow values are unlikely. Summarizing the
computational simulations, there is no
Regarding the computational
reason to be threaten by any variation of
simulations, the software used was Aspen
reagents feeding flow. There is no probability
Plus, Polymers template, and the deviations
of explosion or runaway: no result was
triggered were just related to the variation of
alarming and it was confirmed that SGL
the reagents feeding flow. By these results,
Composites, S.A. has the necessary control
it was witnessed the temperature varies with
instrumentation to a correct automatic
the variation of any reagent feeding flow, but
adjustment and the required equipment to
CB. However, the variation of MB feeding
alert the operators of the anomalies that are
flow has a different impact on the reactional
occurring at the time.
mixture temperature than the variation of CD

7
Concerning the experimental work, it can be made gradually over time is the
could be concluded that the pH increasing, improvement of the controllers’ response,
pH decreasing and the lack of inert media do decreasing its slowness that sometimes may
not affect the reactor safety. But, the pH require the intervention of an operator.
variation scenarios affect the product
specification. Actually, during the pH References
increasing experiment, the NSP was sent to [1] A. Huang, G. Matheus, J. Carneiro, L.
measure: 0,164, while the central value is
Zanuto, L. Moraes, and Y. David,
0,175. Concerning the agitation loss, cooling
“Acrylonitrile.” [Online]. Available:
loss, adiabatic system and the unexpected
introduction of unreacted MB scenarios, http://acrilonitrila.wixsite.com/eq1
these affect the reactor safety and are hard 01/blank. [Accessed: 25-Jun-2018].
to control since the reactional mixture
[2] Commonwealth of Australia,
solidifies in minutes and its temperature
increases infinitely. Even so, accidents are “Acrylonitrile,” 2000.
avoided because, meanwhile, the reactor is
[3] N. P. IGTPAN (Alves Granado,
immediately stopped and cleaned out – no
“Polyacrylonitrile.” [Online].
one lets the reactor to reach its maximum.
Additionally, and regarding the existing
Available:

instrumentation, it was once more confirmed http://www.igtpan.com/Ingles/poli


that SGL Composites, S.A. has the acrilonitrila.asp. [Accessed: 16-Apr-
necessary equipment to alert the operators 2018].
of the anomalies that are occurring at the
time. Summarizing the experimental work, it [4] H. Tamura, “United States Patent,
is necessary to be vigilant of any system ‘Process for Producing Acrylonitrile
alteration but still, nefarious occurrences are Polymers,’” 3,915,942, 1975.
prevented, at least any one triggered by the
scenarios here simulated. [5] R. Conceição, “SGL Composites,
S.A., Manual A, CP Area,” A-CP1-1.1,
The process instrumentation was
2017.
critically reviewed and some proposals were
made to contribute to the overall process [6] H. F. Mark, “Encyclopedia of
safety and to increase the “intrinsically safe” Polymer Science and Technology,”
components.
vol. 9, 2004.

Summing all up, there is no need to be [7] B. G. Frushour and R. S. Knorr,


threaten about a possible explosion or
“Acrylic Fibers,” Handb. Fiber Chem.
runaway, at least considering the operating
Third Ed., vol. Volume 8, pp. 812–
procedures, the operators’ training and
958, 2006.
experience and the existing control
instrumentation. However, a progress that

8
[8] J. Hash, “Rupture Disk,” no. 475000,
p. 5775, 2008.

[9] SPX Corporation, “Self-cleaning


Filters,” pp. 0–3, 2011.

[10] BALLUFF, “Solutions for Level


Detection Optimized for your
application,” pp. 1–4.

You might also like