Presented at AIChE Spring 98 Meeting, New Orleans

Design Issues in Converting to Super-Condensed Mode Operation for

Polyethylene
Ashuraj Sirohi and Sundaram Ramanathan
Aspen Technology, Inc.
Ten Canal Park
Cambridge, MA 02141
Copyright 1998 Ashuraj Sirohi and Sundaram Ramanathan

Abstract
With the emerging new technologies, there is a great economic benefit and competitive
advantage in moving from traditional catalyst based gas-phase polymerization to improved
operation / design, e.g. super-condensed mode (SCM), and novel catalyst systems, e.g.
metallocene catalysts. There are various issues that need to be addressed when a move from a
conventional fluidized bed polyethylene plant is made to the super-condensed mode operation
with/without changing to a new catalyst system. A detailed first principles polymerization
process model is used to investigate the various process issues involved in such a transition. A
process engineer armed with this type of knowledge/tools can make better decisions about
debottlenecking existing plants or selecting a new technology.
The capital investment and energy requirements to operate gas-phase reactors are relatively low
compared to other processes. A typical gas-phase reactor has a very high recycle gas flow rate
and conversion per pass is usually less than 5%. The rate of polymer production is limited by the
rate at which heat can be removed from the adiabatically operated gas-phase reactor. The heat of
polymerization is removed as sensible heat by circulating the hot recycle gas stream from the
reactor through the compressor and heat exchanger and back into the reactor after mixing with
fresh feed. Previously it was believed that cooling the recycle stream below its dew point could
lead to plugging of recycle lines, heat exchanger and could lead to non-uniform monomer
concentrations inside the fluidized bed [1]. The amount of heat removed by the circulating gas
determines the reactor productivity. The gas flow rate is limited to prevent excessive entrainment
of solids. As a consequence, the productivity of these reactors was severely limited by the size of
these reactors. The problem was exacerbated when a reactor that was sized for a homopolymer
was used to manufacture co-polymers with higher alpha-olefin co-monomers. The presence of
higher mol. wt. component in recycle stream raised the dew point of the mixture leading to
formation of liquid in the recycle stream.
Contrary to these beliefs, several companies have discovered that the recycle stream can
be cooled below the dew point resulting in condensing a portion of the recycle stream. The
partially condensed stream returns to the reactor without plugging or agglomeration problems in

Presented at AIChE Spring 98 Meeting, New Orleans

the reactor. An increase in polymer production is obtained as a result of increased cooling

capacity. The super-condensed mode provides a way to increase the plant capacity without
resizing the reactor. Using this technology, new plants can be built with smaller reactor diameters
than what was possible earlier.
Traditionally new design and operational issues have been solved through lengthy pilot
plant trials. More recently, process simulation using packages like ASPEN PLUS / SPEEDUP
/ POLYMERS PLUS have been used to reduce the process development time. In this work, a
detailed POLYMERS PLUS model that can simulate both gas-phase and SCM operation is used
to investigate the process and catalyst changes on cooling capacity, compressor power, polymer
properties, etc. The predictions for polymer properties (Melt Index, Density), ash content, MWD
etc. will be compared for multi-site Ziegler-Natta, Metallocene and Cromocene based catalysts
for different modes of operation.

Process Description
A steady state ASPEN PLUS flowsheet describing the polyethylene fluidized bed reactor
operation is developed and is shown in Figure 1. The fluidized bed reactor is modeled as a
continuous stirred tank reactor.
It is assumed that the polymer
phase in the reactor is well
mixed. The gas-phase residence
time in the reactor is 20-30
seconds and the polymer-phase
residence time is several hours.
The catalyst is present in
polymer phase where the
reaction takes at the catalyst
surface. The solubility of
monomers and other light
components in polymer phase is
predicted
using
POLYSL
(Polymer
Sanchez-Lacombe)
equation of state (available in
POLYMERS
PLUS).
No
Figure 1 : A Schematic of Gas-phase Polyethylene Reactor
reaction takes place in the vapor
phase. In the real process the
polymer
is
withdrawn
intermittently from the reactor by opening a discharge pump. In ASPEN PLUS flowsheet the
polymer is withdrawn continually from the flash tank (SEP). In a real reactor there is a pressure
drop across the reactor which is proportional to the reactor cross-sectional area and the weight of
the fluidized bed. This pressure drop is taken into account in the SEP block. The vapor from the
flash tank (SEP) is compressed to reactor pressure (COMP) and then cooled in a heat exchanger
(COOLER). The fresh feed is then mixed with the recycle and fed to the reactor. The catalyst and

Presented at AIChE Spring 98 Meeting, New Orleans

co-catalyst are fed directly into the reactor. The typical operating conditions used in SCM
simulation are chosen from Table 2 of pages 15-16 of reference [1] and are shown below.
Melt Index
Resin Density (g/cc)
Reactor Inlet Temperature (oC)
Reactor Temperature (oC)
Liquid in recycle (wt %)
Reactor Pressure (psig)
Production Rate (Tons/hr)

0.92
0.9187
50
85
22.2
314
32.3

Recycle Stream Compositions

Ethylene
Butene-1
Hydrogen
Isopentane
Nitrogen

Mole %
52.6
20.0
9.7
10
7.7

The above conditions are for condensed mode of operation. To simulate the conventional
mode of operation, no Isopentane is used and Nitrogen mole percent in recycle stream is
increased to 17.7 mole %.

Catalyst Kinetics
A detailed Ziegler-Natta kinetics model (available in POLYMERS PLUS) is used to
model the polymerization kinetics. The kinetic parameters used for the above copolymerization
kinetics are obtained from the literature [2]. The reactions used in the process model are :
Site Activation by Cocatalyst

C ps + Am P0k

Chain Initiation

P0k + M i Pki ,i

Chain Propagation

Pnk,i + M j Pnk+i ,i

Chain Transfer to Hydrogen

Pnk,i + H 2 Dnk + P0k

Chain Transfer to Monomer

Pnk,i + M j Dnk= + Pki ,i

Spontaneous Site Deactivation

P0k Cds ; Pnk,i Cds + Dnk

The above kinetic scheme can model either single site metallocene or muti-site ZieglerNatta kinetics. Three types of activity profiles are chosen to represent three different catalysts. A
lab-scale semi-batch reactor is simulated where a fixed amount of catalyst is charged into the
reactor and the feed rate of monomer is varied to maintain constant total pressure. The
instantaneous feed rate profiles for the three catalysts, shown in Figure 2, illustrate the
differences in catalyst deactivation kinetics. The total pressure is maintained at 3 bar and the

Presented at AIChE Spring 98 Meeting, New Orleans

temperature is kept constant at 85

o
C. Figure 3 shows the fraction of
catalyst deactivated as a function of
time. All three catalysts have fast
activation.
This
simulation
characterizes
the
activation,
propagation
and
deactivation
kinetics for the three different
catalysts. Figure 4 shows the
accumulated mass of polymer in the
reactor as a function of time. There
is no new polymer formed by
Catalyst C after 2 hr. In practice, the
model can be matched to the
experimental data profiles using the
Figure 2 : Instantaneous Feed Rates (kg/hr)
Data-fit / Optimization feature in
ASPEN PLUS process simulator.
Catalyst A is a slow deactivating catalyst. After a batch time of 5 hrs, 50% of the catalyst is still
active. Catalyst B is a medium deactivating catalyst. It is almost 80% deactivated in 2 hr. Catalyst
C is a rapidly deactivating catalyst. It is totally inactive after 2 hr.
These figures also illustrate the
differences in the activities of the
three catalysts. The deactivation
kinetics observed in a continuous
back-mixed reactor would be
different from that observed in a
batch reactor due to the residence
time distribution effects. For
example, a continuous reactor with
a residence time of 2 hr would have
approximately 60 % of catalyst B
deactivated (in the outlet stream)
instead of 80 % as seen in the semibatch experiment simulation. These
three catalyst systems are now used
to study the effect of switching to
the condensing mode of operation
Figure 3 : Fraction of Catalyst Deactivated
for the fluidized bed reactor.

Simulation Results
The kinetics of the three catalysts were characterized using the semi-batch simulation as
shown in the above section. These same catalyst kinetics and operating conditions shown before

Presented at AIChE Spring 98 Meeting, New Orleans

are now used in fluidized bed polymerization reactor flowsheet (Figure 1) to simulate both the
conventional and super-condensed mode of operation.
In these simulations, the total mass
flow rate of the recycle stream is
1E6 kg/hr. The conversion per pass
is 2.5 % for non-condensed mode
and 6.5 % for condensed mode
operation. The reactor is operated
adiabatically.
The
heat
of
polymerization is removed as
sensible heat in non-condensing
mode and as both sensible and
latent heat in condensed mode. The
three different catalysts were used
in both non-condensing and
condensing mode and various
process outputs are compared for
the two modes of operation in
Figure 4: Mass of PE in Batch (kg)
Table 1. The polymer properties
(Resin density and Melt Index) are
a function of catalyst kinetics (and ratios of concentrations) and do not change significantly for
some catalyst systems when used in both conventional and condensed mode operation.

Table 1 : Results for Non-condensed and Condensed mode Operation

Non-Cond.
Cat A
Cat B
Cat C

Polymer

Catalyst

Res. time

Catalyst

Liq. at inlet

Compressor

(Tons/hr)

(kg/hr)

Cooler Duty Ash Content Cat. Activity

(GJ/hr)

(x 1E-4)

(x 1E4)

(hr)

Deact.

mole %

( MWatt)

11.18
11
10.8

0.9
2.45
1.06

44.2
44.1
44

0.805
2.227
0.981

1.242
0.449
1.019

4.95
5
5

41%
80%
95%

0
0
0

1.4
1.4
1.4

29.6
28
28

4.7
7.9
2.68

107.8
102.3
102.3

1.588
2.821
0.957

0.630
0.354
1.045

1.63
1.72
1.72

19%
56%
87%

10.6
9.8
9.8

1.26
1.26
1.26

164.8
154.5
159.3

422.2
222.4
152.8

143.9
132.0
132.5

97.2
26.7
-2.5

-49.3
-21.1
2.5

-67.1
-65.6
-65.6

SCM
Cat A
Cat B
Cat C

Change
Cat A (%)
Cat B (%)
Cat C (%)

-10.0
-10.0
-10.0

In all the above simulation cases, the cooler outlet temperature (almost same as reactor
inlet temperature) is assumed to be at 50oC. Using 10 mole % Isopentane (equivalent to 20 wt %)
in the recycle stream, the amount of liquid predicted in the recycle stream at the cooler outlet is
close to 22 wt %. The exact amount of each component in liquid phase is predicted by POLYSL

Presented at AIChE Spring 98 Meeting, New Orleans

(Polymer Sanchez-Lacombe) equation of state. The polymer produced and cooler duty are a
function of how much heat can be removed from the reactor. For all three catalysts in any mode
of operation, the polymer produced, cooler duty and the amount of condensation in recycle
stream are the same. These process parameters are independent of catalyst kinetics. There is
about 160 % increase in productivity if 10 mole % of liquid is present in the recycle stream. The
increase in catalyst flow rate to achieve the increased productivity in condensed mode depends
on the activity profile of a catalyst. The catalyst with shortest half life (Catalyst C) from semibatch simulation is a very active (g-polymer/g-cat) catalyst in continuous operation. For a
residence time of 5 hr, Catalyst C is almost 95 % deactivated and its activity is still similar to
only 41 % deactivated Catalyst A. The ash content is defined as gram of catalyst per unit gram of
polymer. Catalyst activity reported in Table 1 is inverse of ash content. The ash content increases
in condensed mode for an active catalyst (most of the sites still active). For a catalyst which has a
very short half life (Catalyst C), switching to super-condensed mode does not change its ash
content and its activity (as defined above). In our simulations, the Catalyst C was 95%
deactivated in non-condensed mode. Switching to condensed mode decreases the residence from
5 hr to 1.7 hr and now Catalyst C is only 87 % deactivated. The power of the compressor is a
function of gas composition and a 10% decrease in power requirement is observed when supercondensed mode is used.
We have shown some of the process variables in Table 1 for the two modes of operations.
It was illustrated through simulations that catalyst with a very short half life will result in almost
no change in ash content and catalyst activity when a switch is made to condensed mode
operation. This type of analysis can easily be performed using process simulator packages like
POLYMERS PLUS / ASPEN PLUS / SPEEDUP. Ease of simulating a real process using
these tools can greatly reduce the research and development time and can aid a process engineer
in making better informed decisions about new technology. More results and analysis (e.g., slow
activating Cromocene based catalyst system, multi-site and single-site catalyst systems) will be
presented at the conference.
The authors would like to thank Aspen Technology, Inc. for the support of this work.

References
[1]
[2]

US Patent 5,352,749
Sirohi, A. and K.Y. Choi, Ind. Eng. Chem. Res. Vol. 35, No. 4, 1332 (1996).