ch32b Midterm2 2004
ch32b Midterm2 2004
FC
FI
Process Inlet
Flow, F(t)
Temperature, Ti
TC
Process Outlet
Flow, F(t)
Temperature, T(t)
Heat
Exchanger
Trap
The following Laplace domain expression relates changes in the steam and process fluid
flowrates to the exchanger outlet temperature:
T ( s ) G ( s ) FS ( s ) GF ( s ) F ( s )
0.242 e 6 s
0.01e 8 s
FS ( s )
F ( s)
7.5 s 1
15 s 1
(1)
Page 1 of 7
The time units of the transfer function parameters are minutes. The overbar in Eqn. (1)
indicates (the Laplace transform of) the deviation of the system variables from the
steady-state values listed below.
Tss 90 C ,
FS , ss 82.7 kg/min ,
The thermocouple is calibrated for the range 0-100 C and exhibits a random error of
mean zero and variance 0.01 (C)2. The maximum steam flow setpoint (corresponding to
a temperature controller output of 100%) is 150 kg/min.
(a) The steam flow control valve is air-to-open, i.e. fail-shut. Should the direction of the
steam flow controller be reverse (increase/decrease) or direct (increase/increase)? Should
the temperature controller be configured as reverse-acting or direct-acting?
[10 marks]
Page 2 of 7
(b) Build a Simulink model named heater.mdl to simulate 2 hours of PID control of the
heat exchanger outlet temperature, introducing a step change of +500 kg/min in the
process fluid flowrate at time t = 20 minutes. The setpoint of the temperature controller
is to remain fixed at its initial value ( SP ( s ) 0); the dynamics of the steam flow
control loop may be neglected (i.e. assume perfect steam flow control). Tune the
controller using the trial-and-error procedure covered in Section 2.3.2 of the lecture
notes. Write a MATLAB program (call it afterheater.m) which plots the setpoint and
measured temperature in the upper subplot and the controller output in the lower subplot.
Record the integral of the squared error as a measure of the performance of your
temperature controller.
[55 marks]
Kc
ISE =
(c) Repeat part (b) using a Proportional-Integral controller designed according to the
Ziegler-Nichols tuning method.
Table 1 - Ziegler-Nichols tuning rules.
D
Controller Type
KC
PI
0.9
K D
3.33 D
__
PID
1.2
K D
2.0 D
0. 5 D
[6 marks]
Kc
ISE =
Page 3 of 7
ISE =
Which of the three tuning methods (trial-and-error, ZN PI, ZN PID) would you
recommend for this application? Explain your answer.
[9 marks]
Page 4 of 7
(e) In order to improve the performance of the Ziegler-Nichols PID controller, the
measured temperature will be filtered by the first-order transfer function
GF ( s)
1
F s 1
(2)
with time constant F 0.3 min. Prove that the frequency response of this PV filter is
given by
| G F ( j ) |
(3)
1 F2 2
G F ( j ) tan 1 ( F )
(4)
Page 5 of 7
(f) Repeat part (d) assuming that the PV signal is passed through G F (s ) prior to feeding
it back to the controller. Save your model as heaterfilt.mdl and comment upon your
results.
[10 marks]
ISE =
Page 6 of 7
Page 7 of 7