2011 3rd International Conference on Chemical, Biological and Environmental Engineering IPCBEE vol.

20 (2011) (2011) IACSIT Press, Singapore

Temperature Control of Industrial Gas Phase Propylene Polymerization in Fluidized Bed Reactors Using Model Predictive Control
Ahmad Shamiri 1, 2+, M. A. Hussain 1, Farouq Sabri Mjalli 3 and Arash Arami-Niya
2

Department of Chemical Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia Training Center, Razi Petrochemical Company, P.O. Box 161, Bandar Imam, Iran 3 Petroleum & Chemical Engineering Department, Sultan Qaboos University, Muscat 123, Oman

Abstract. A two-phase dynamic model describing gas phase propylene polymerization in a

fluidized bed reactor which considers the polymerization reaction in both phases was used to explore the control of the reactor temperature by manipulating the heat exchanger cooling water flow rate. A model predictive control (MPC) technique is implemented to control of the nonlinear process and compared its performance with conventional PI controllers tuned using the Internal Model Control (IMC) method as well as the standard Ziegler-Nichols (Z-N) method. The closedloop simulations revealed that the Z-N PI controller produced oscillatory responses and the MPC and the IMC-Based PI controllers were able to track the changes in the set point. However the quality of the MPC set point tracking was superior to that of the IMC-Based PI controller.

Keywords: Temperature Control, Propylene Polymerization, Model Predictive Control.

1. Introduction
Modeling and control of the polymerization process in fluidized bed reactors such as polypropylene production are challenging issues in process and control engineering. This is mainly because of the high nonlinearity of the process dynamics due to complicated reaction mechanisms, complex flow characteristics of gas and solids, various heat and mass transfer mechanisms and the interaction between the control loops. Many studies were reported for the modeling and control of olefin polymerization processes using various types of algorithms [1-2]. A process schematic of an industrial gas-phase fluidized bed polypropylene reactor is shown in Figure 1. To maintain acceptable polymer production rate which is an important goal for industry it is necessary to keep the reactor bed temperature above the dew point of the reactants to avoid gas condensation and below the melting point of the polymer to prevent particle melting and agglomeration and consequently reactor shut down. Choi and Ray [1] used a dynamic model considering bubble and emulsion phases in the bed in the first attempt to describe the dynamics of polypropylene production. They showed that a PI feedback control scheme can be used to control the process transients, but it is limited by the recycle gas cooling capacity. Dadebo et al. [2] showed that for the temperature control of industrial gas phase polyethylene reactors the nonlinear error trajectory controller (ETC) exhibits significantly superior responses in terms of speed, damping and robustness compared with an optimally-tuned PID controller over a wide range of operating conditions. Due to the high nonlinearities and difficulties involved in the dynamics and control of the gas phase propylene polymerization fluidized bed reactor, an efficient process control scheme need to be implemented. However, it is beyond the capability of the conventional controller with fixed controller settings to achieve excellent control of the reactor variables. In order to achieve good control of the reactor variables, a more intelligent and efficient process control scheme is needed where the controller is able to automatically redesign itself in real time according to the changing process dynamics.
19

In the present study, a two-phase model with comprehensive kinetics for propylene homo-polymerization in a fluidized bed reactor that considers the presence of particles in the bubbles and the excess gas in the emulsion phase with polymerization reaction in both phases is considered [3-4]. A Model Predictive Control (MPC) algorithm is used for controlling the reactor temperature by manipulating the coolant flow rate.

2. Mathematical Modeling of Propylene Polymerization

In the present study, the kinetic model of propylene homo-polymerization over a Ziegler-Natta catalyst based on the kinetic model developed by Shamiri et al. [3-5] and the dynamic two-phase flow structure were combined and implemented to provide a more realistic understanding of the phenomena encountered in the bed hydrodynamics. For details of the kinetic scheme, definitions of pseudo-kinetic rate constants and the correlations required for estimating the bubble volume fraction in the bed, the voidage of the emulsion phase and bubble phases, the emulsion phase and bubble phases gas velocities and mass and heat transfer coefficients for twophase model, the reader is referred to a paper by Shamiri et al. [3]. The following dynamic material balances were written for all of the components in the bed. For bubbles:
U A [Mi ]bUbA V [Mi ] b Rv b [Mi ]b Kbe ([Mi ]b [Mi ]e ) b (1 b ) b,(in) b b = d dt A b VPFR Ri dz b

(Vbb[Mi ]b )

(1)

For emulsion:

(1 e )Ri [M i ] U A [M i ]e U e Ae Rv e [M i ]e + Kbe ([M i ]b [M i ]e ) Ve e,(in ) e e e 1

= d dt

(2)

(Ve e [M i ]e )

The direction of mass transfer was assumed to be from bubble to emulsion phase. Furthermore, the energy balances can be expressed as: For bubbles:
m m Ub Ab (T C pi Ub Ab (Tb Tref ) [M i ]bC pi Tref ) [M i ] b,(in ) b,(in ) i =1 i =1 A H R m Rv (Tb Tref ) bC pi [M i ]b + (1 b )polC p, pol + (1 b ) b Rpbdz i =1 VPFR m d Vb Vb b (Tb Tref ) C pi ([M i ]b ) +Hbe (Te Tb ) i =1 dt m d (T Tref ) = Vb (b C pi [M i ]b + (1 b ) polC p, pol ) i =1 dt b

(3)

For emulsion:
m m Tref ) [M i ] Ue Ae (T C pi Ue Ae (Te Tref ) [M i ]eC pi e,(in ) e,(in ) i =1 i =1 m Rv (Te Tref ) eC pi [M i ]e + (1 e )polC p, pol (1 e )Rpe H R i =1 HbeVe

m d (Te Tb ) Ve e (Te Tref ) C pi ([M i ]e ) = i =1 dt 1

m d (Te Tref ) Ve (e C pi ([M i ]e ) + (1 e )polC p, pol ) i =1 dt

(4)

3. Results and Discussion

The process was simulated at the operating conditions shown in Table 1.
20

The presented mathematical model for the gas phase propylene polymerization fluidized bed reactor consists of cooling water flow rate (Fcw) as a manipulated variable and emulsion phase temperature (Te) as a controlled variable. A model-predictive controller was designed using the MPC toolbox GUI and implemented in simulink using the MPC simulink block. The closed loop performance of the MPC scheme in tracking series of setpoint changes for the emulsion phase temperature in the polymerization reactor is evaluated. For this purpose, series of setpoint changes in opposite directions were introduced. The magnitude of the setpoints introduced for this loop was typical of the respective nominal operating range. For comparison purposes, conventional PI controllers tuned using the IMC [6] method and the standard Z-N [7] method, were included in this simulation. Both the IMC and ZN PI controller tuning parameters were calculated based on analyses of the open loop process reaction curve for a particular operational region of the process. To ensure good performance of the MPC controller, the tuning parameters must be appropriately tuned. Although Shridhar and Cooper [8] suggestion were good starting values for tuning the controller parameter but the exact values used in this work were the result of further fine tuning based on actual control performance. In addition to the selection of controller tuning parameters, the values were chosen for the constraints and imposed on loop are based on practical considerations acquired through real operational experience of the authors. The resulting controller had a sampling time of 10 sec, a prediction horizon of 17, a control horizon of 1, an output variables weight of 0.09, and a manipulated variables weight of 0.008. Figs. 2 and 3 show the performance of the MPC as compared to the IMC and Z-N PI controllers in tracking series of setpoint changes for the temperature loop as well as the corresponding controller moves. From Fig. 2, the failure of the Z-N PI controller to track the changes in the setpoints for the temperature loop was obvious. The Z-N PI controller produced oscillatory responses. Furthermore, as shown in Fig. 3, the controller moves observed were vigorous. However the shortcomings exhibited by the Z-N PI controller were not observed for the cases of MPC and the IMC-Based PI controllers. In general, both the MPC and the IMC-Based PI controllers were able to track the changes in the set point. However, the quality of the set point tracking as demonstrated by the MPC was superior to the IMC-Based PI controller in terms of their ability to attain minimal overshoot. Moreover, the MPC was able to not only produce controller moves which were well within the specified input constraints, but also the controller moves produced were nonaggressive and smooth for practical implementations. These were attributed to the ability of the MPC to handle constraints in the inputs, of which the conventional PI controller was unable to achieve. To summarize, Table 2 shows the Integral Absolute Error (IAE) for the three controllers in tracking the series of set point changes. The values of the IAE calculated were consistent with the previous discussions, where the performance of the MPC was superior to the conventional PI controllers.

4. Conclusions
The closed-loop simulations revealed that the Z-N PI controller produced an oscillatory response while the MPC and the IMC-Based PI controller were able to track the changes in the setpoint. However, the quality of the setpoint tracking as demonstrated by the MPC was superior to that of the IMC-Based PI controller in terms of the ability to attain minimal overshoot. Moreover, the MPC was able to not only produce controller moves which were well within the specified input constraints, but also the controller moves produced were non-aggressive and smooth for practical implementations. The values of the IAE calculated for the temperature loop indicate that the performance of the MPC was superior to the conventional PI controllers. Tab. 1: Operating conditions considered for modeling fluidized bed polypropylene reactors. Operating conditions V (m3)=50 Tref (K)=353.15
21

Physical parameters (Pa.s)=1.1410-4 g (kg/m3)=23.45

Tin (K)=317.15 P (bar)=25 Propylene concentration (kmol/m3)=0.9 Hydrogen concentration ( kmol/m3)= 0.015 Catalyst feed rate (kg/s)= 0.0003

s (kg/m3)=910 dp (m) =50010-6 mf =0.45

Tab. 2: Integral absolute error (IAE) for the MPC, IMC-Based PI controller, and the Z-N PI controller in tracking series of setpoint changes for the emulsion phase temperature.

Controller MPC IMC-Based PI Controller Z-N PI Controller

MPC

IAE 2527 2712 4655

Recycle Stream Compressor

TT

Reactor

Coolant in, Fcw

Product PP Catalyst in, Fcat

Fresh Feed

Fig. 1: Simplified schematic of the MPC design on the gas phase propylene polymerization fluidized bed reactor.

344 MPC Controller Set point 342

344 IMC-Based PI Controller Set point 342

Emulsion phase temperature (K)

Emulsion phase temperature (K)

340

340

338

338

336 8000

10000

12000

14000

16000

18000

20000

22000

336 8000

10000

12000

14000

16000

18000

20000

22000

Time (s)

Time (s)

344 Z-N PI Controller Set point 342

Em ulsion phase tem perature (K)

340

338

336 8000

10000

12000

14000

16000

18000

20000

22000

Time (s)

22

Fig. 2: Comparison of the performance between the MPC, IMC-Based PI controller (Kc = -18.5345, I = 858.0787, D = 0), and the Z-N PI controller (Kc = -33.3620, I = 266.4696, D = 0) in tracking series of setpoint changes in the emulsion phase temperature (Te) (U0=0.45m/s, Fcat=0.0003kg/s).
60 IMC-Based PI Controller
80 100 MPC Controller

50

Controller moves (%)

40

Controller moves (%)

5000 10000 15000 20000

60

30

40

20

10

20

0 5000 10000 15000 20000

Time (s)

Time (s)

100 Z-N PI Controller 80

Controller moves (%)

60

40

20

0 5000 10000 15000 20000

Time (s)

Fig. 3: Comparison of the corresponding controller moves between the MPC, IMC-Based PI controller, and the Z-N PI controller for the emulsion phase temperature.

Nomenclature
Hbe Kbe [Mi] Rp Ri Rv U0 Ub Ue Vb Ve VPFR b e mf

bubble to emulsion heat transfer coefficient, (W/m3.K) bubble to emulsion mass transfer coefficient, (s-1) concentration of component i in the reactor (kmol/m3) production rate (kg/s) instantaneous rate of reaction for monomer i (kmol/s) volumetric polymer phase outflow rate from the reactor (m3/s) superficial gas velocity (m/s) bubble velocity (m/s) emulsion gas velocity (m/s) volume of the bubble phase volume of the emulsion phase volume of PFR volume fraction of bubbles in the bed void fraction of bubble for Geldart B particles void fraction of emulsion for Geldart B particles void fraction of the bed at minimum fluidization

5. References
[1] K. Y. Choi, and W. H. Ray. The dynamic behavior of fluidized bed reactors for solid catalyzed gas phase olefin polymerization. Chemical Engineering Science, 1985, 40: 2261-2279. [2] S. A. Dadebo, M. L. Bell, P. J. Mclellan, and K. B. Mcauley. Temperature control of industrial gas phase polyethylene reactors. Journal of Process Control, 1997, 7: 83-95. [3] A. Shamiri, M. Azlan Hussain, F. Sabri Mjalli, N. Mostoufi, and M. Saleh Shafeeyan. Dynamic modeling of gas phase propylene homopolymerization in fluidized bed reactors. Chemical Engineering Science, 2011, 66: 11891199. [4] A. Shamiri, M.A. Hussain, F.S. Mjalli. Two phase dynamic model for gas phase propylene copolymerization in fluidized bed reactor, in: Defect and Diffusion Forum, 2011, 312-315:1079-1084.
23

[5] A. Shamiri, M. A. Hussain, F. S. Mjalli, and N. Mostoufi. Kinetic modeling of propylene homopolymerization in a gas-phase fluidized-bed reactor. Chemical Engineering Journal, 2010, 161: 240-249. [6] I. L. Chien, and P. S. Fruehauf. Consider IMC tuning to improve controller performance. Chemical Engineering Progress, 1990, 86: 33-41. [7] J. G. Ziegler, and N. B. Nichols. Optimum settings for automatic controllers. Transactions of the ASME, 64,1942: 759-768. [8] R. Shridhar, and D. J. Cooper. A Tuning Strategy for Unconstrained SISO Model Predictive Control. Industrial and Engineering Chemistry Research, 1997, 36: 729-746.

24