ChE 182 TAB HXY

de Torres. Magueflor. Maningding

Control Design Project for a Jacketed Reactor

Phosgene-free Polyurethane Production (MPC Synthesis)

Executive Summary
A jacketed reactor was controlled through feedback and feedforward control strategies
with the controlled variable being the temperature of the tank, the manipulated variable being
the flow rate of the steam used as a heating fluid and the disturbance variables being the inlet
temperature of the tank and the initial concentrations of the reactants, namely aniline and
DMC. These were made to maintain safety within the area of the reactor as a higher
temperature than the operating condition may result to explosion risks and diffusion of harmful
vapours. It would also affect the quality of the product produced.
Series PID with filter controller was used because the process gave a second order
transfer function.
The response plots obtained from direct synthesis, internal model control, ZiegerNichols and Tyreus-Luyben were superimposed for comparison. The Zieger-Nichols control
settings gave the best response as it approached stability in the least time possible with the
least overshoot.
The Bode plots gave same responses for the two controller settings, being stable at -180
degrees so direct substitution was used to obtain values for Kcu and omega. These same values
were used for the robustness analysis giving a stable system within a range of 2.6271.7226.
It can be concluded the control strategies used were effective enough to control the
equipment chosen.

Process Description
ProcessDesign
The production of phosgene-free polyurethane using aniline and DMC as the raw
materials consists of four main steps. The first step is the synthesis of MPC which will be
followed by MDC synthesis then MDI synthesis. MDI will undergo polymerization which will
produce polyurethane. For this study, the output temperature of the reactor in MPC synthesis
is to be controlled for safety purposes. As shown in Figure 1, the endothermic reaction of
aniline with DMC will produce MPC proceeds at 150 degrees Celsius. Any temperature higher
than the operating condition may result to risks of explosion and diffusion of harmful vapours.

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

Figure 1. Chemical reaction of Aniline and DMC to produce MPC

Figure 1. Process flow diagram of MPC Synthesis

As shown in the process flow diagram in Figure 2, two feeds streams (aniline and DMC)
are pressurized, mixed, heated in the fed reactor. The outlet stream of the reactor contains
MPC (product), excess reactants and bi-products. After this step separation and purification of
MPC follows by a series of distillation.
Control Strategy
The reactor is being heated by a jacket with steam as the heating fluid. The variable to
be controlled is the output temperature of the tank (T) by manipulating the mass flow rate (Q)
of the steam entering the jacket. The disturbance variables are the input concentrations of the
reactants (CAi and CBi) and the input temperature of the tank (Ti) .

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

For this control system, the control strategy chosen are the following:
1. Feedback Control

2. Feedback-Feedforward Control

Derivations

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

A. PROCESS TRANSFER FUNCTION (from Mass and Energy Balances), Gp

Calculated data needed for the derivation of Gp:
q = 13,384.85 kg/hr
Vtank = 102 m3
CAi = 0.129
CBi = 0.87
k = 1.03
= 1241 kg/m3
Cp = 0.2033 kJ/mol-K
U = 680 W/m2K
A = 93. 75 m2
Vj = 20.462 m2
3
steam = 5.6657 kg/m
Cpsteam = 1.8294 kJ/mol-K
Aj = 18.75 m2
Tji = 184 C
Tj = 175 C
T = 150 C
Component Balance A

Linearization

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

+ (
*

) *
+

+ (

) [ ]

[ ]

To get

= 0.128

()

()

()

Transfer functions

OMB Tank

Energy Balance Tank

(
Linearization

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

( )

()
Transfer Functions

()
()
()
()
OMB Jacket

)(

)(

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

Energy Balance (jacket)

(

Linearization

To get
(

()
Transfer function

()

()

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

Over-all Process Transfer Function

(

)(

Instrumentation
The range for all the electronic signals is 4 to 20 mA.
500-AH Transducer
Current to Pressure Transducer: It acts as a device with negligible
dynamics. The output signal changes from 0.2 10 bars when the input signal
changes from 4 20 mA.

Steam Control Valve with I/P positioner integrated into a compact pneumatic
actuator
Control Valve: Control valve with a negligible dynamics of Kv = 9.7.
Input 15.9 60 bars and output of 0-16 kg/h of steam.
Resistance Temperature Detector
Transmitter: It acts as a device with negligible dynamics. The output
signal changes from 30-800 degrees Celsius when the input signal changes from
4 20 mA.

Calculations for Tuning and Response Analysis

CONTROLLER TRANSFER FUNCTION, Gc
A. FEED-BACK CONTROL
A.1. Direct Synthesis (DS) Method

)(

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

)(

)(

)(

)(

From the expression of Gc (2nd order) it can be seen that PID controller should be used to
control the process.
)

PID controller settings (

3446

A.2. Internal Model Control (IMC) Method

)(

)(

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

)(

)(

)(

The extracted PID controller settings from the Gc expression are:

3346
Values are consisted with the controller settings obtained using Direct Synthesis method.
A.3. ZIEGER-NICHOLS AND TYLUS , KCU and Pu
(

)(

)(

)(

)(

)(

Kc (

)(

)(

)(

)
)(
)(
)

)(
)(

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

B.

FEEDFORWARD-FEEDBACK CONTROL

Applying a filter
(

Using an

value of 0.5

Response Plots
Feedback - Disturbances

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

Plots for the disturbances (Ti, CAi, CBi) show the same trend having the best
response for the Zieger-Nichols control settings with the lowest overshoot and
the least time to reach stability.
Feedback - Setpoint

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

For the plot shown above for setpoint tracking, still the Zieger-Nichols
control settings gave the best response with the least overshoot and time to
reach stability.
Feedforward (on-line tuning)

Further tuning on feedforward control is not necessary for the set Kc has
given a stable response.
Bode and Nyquist Plots

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

Zieger-Nichols

The system stabilizes at -180 degrees so the values of Kcu and Pu were
obtained from solving through the direct substitution method for the plot would
give an almost infinite number omega for the -180 degrees. The Kcu and omega
was used for the robustness analysis.

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

Tyreus Luyben

Same is true with the response of T-L analysis. The system gave a stable
response at -180 degrees so the location of omega from the plot would result to a
number of values. Again, direct substitution was used and the obtained Kcu and
omega were used for the robustness analysis.

ChE 182 TAB HXY

de Torres. Magueflor. Maningding

Analysis of Robustness
From the value of omega (47.44) obtained from the direct substitution method, the gain
margin was calculated with a result of 2.627 a value greater than 1 implying a stable system.
Considering the process gain, the gain shift allowed for the system to be stable is equal to
1.7226.
The phase margin was not used because of a return value of zero because of the
absence of time delay in the process.

References

http://www.controlair.com/index.php/ip-ep-pi-transducers/t500-electropneumatic-transducerip-ep
http://www.tlv.com/global_pdf/tuk/u-cvcos-hp.pdf
A Safer Ruote to MDI
Process Dynamics and Control, Seborg et. Al.