You are on page 1of 21

Process Modelling, Simulation and Control for Chemical Engineers.

Solved
problems. Chapter 11: Laplace domain analysis of advanced control sys-
tems.
This document contains my own solutions to the problems proposed at the end of each chapter of
the book ”Process Modelling, Simulation and Control for Chemical Engineers”, Second Edition,
by William L. Luyben. As such, I can’t guarantee that the proposed solutions are free from errors.
Think about them as a starting point for developing or as a means of checking your own solutions.
Any comments or corrections will be appreciated. Contact me at francisco.angel.rod@gmail.com
The computer programs developed for this chapter (Python) are available at:
https://www.dropbox.com/sh/t9fzrq4cky1npaj/AABSBhQKgpV01YwNki9pIEGaa?dl=0

Problem 1
The load and manipulated variable transfer functions of a process are:
X 1
= GM =
M (s + 1)(5s + 1)
X 2
= GL =
L (s + 1)(5s + 1)(0.5s + 1)
Derive the feedforward controller transfer function that will keep the process output X constant
with load changes L.

Solution
The feed forward transfer function is:

F = −GL /GM
2
=−
0.5s + 1

Problem 2
Repeat problem 1 with:
0.5s + 1
GM =
(s + 1)(5s + 1)

Solution
The feed forward transfer function is:

F = −GL /GM
2
=−
(0.5s + 1)2

Problem 3
The transfer functions of a binary distillation column between distillate composition xD and feed
rate F , reflux rate R, and feed composition z are:

xD KF e−DF s
=
F (τF s + 1)2
xD Kz e−Dz s
=
z (τz s + 1)2
xD KR e−DR s
=
R τR s + 1

1
Find the feedforward controller transfer functions that will keep xD constant, by manipulating
R, despite changes in z and F . For what values of parameters are these feedforward controllers
physically realizable?

Solution

Kz (τR s + 1)
Fz,R = −e−(Dz −DR )s
KR (τz s + 1)2
KF (τR s + 1)
FF,R = −e−(DF −DR )s
KR (τF s + 1)2
The controllers are realizable if:

Dz >= DR
DF >= DR

Problem 4
Greg Shinskey has suggested that the steadystate distillate and bottoms compositions in a binary
distillation column can be approximately related by:
xD /(1 − xD )
=S
xB /(1 − xB )

where S is a separation factor. At total reflux it is equal to αNT +1 where α is the relative volatility
and NT is the number of theoretical trays. Assuming S is a constant, derive the nonlinear steady
state relationship showing how distillate drawoff rate D must be manipulated, as feed rate F
and feed composition z vary, in order to hold distillate composition xD constant. Sketch this
relationship for several values of S and xD .

Solution
The mass balances around the distillation tower are:

F z = DxD + BxB (1)


F =D+B (2)

Rearranging the Shinskey expression we get:


xD
xB = (3)
S(1 − xD ) + xD

Replacing equation (2) and (3) into equation (1) we get:


xD
F z = DxD + (F − D)
S(1 − xD ) + xD
 
zS(1 − xD ) − xD (1 − z)
D=F
xD S(1 − xD ) − xD (1 − xD )

Curves showing the ratio D/F as a function of z, for different values of xD and S = 2.0 are shown
in Figure 1. The values of z are in the range xB <= z <= xD , where the values of xB , calculated
from equation (3) are: xB = [0.14, 0.33, 0.60] for xD = [0.25, 0.50, 0.75], respectively. The ratio
D/F is a straight line with slope equal to 1/(xD − xB ). The difference xD − xB for S = 2.0 is
shown in Figure 2 as a function of xD , a maximum value, and consequently, a minimum slope for
the D/F curve is located around xD = 0.6.
Curves showing the ratio D/F as a function of z, for different values of S and xD = 0.5 are shown
in Figure 3. The values of xB are [0.49, 0.33, 0.20] for S = [1.05, 2.0, 4.0], respectively. The slope
of the curve D/F decreases as the value of S increases.

2
Figure 1: D/F ratio for different values of xD and S = 2.0.

Figure 2: xD − xB as a function of xD .

Figure 3: D/F ratio for different values of S and xD = 0.5.

3
Figure 4: Root locus plot for first order open loop stable process.

Problem 5
Make root locus plots of first and second order openloop unstable processes with PI feedback
controllers.

Solution
The first order open loop transfer function is:
Kp
GM =
τp s − 1
The characteristic equation is:
Kp K(τ s + 1)
1+ =0
(τp s − 1) τs
τp τ s2 + τ (Kp K − 1)s + Kp K

A root locus plot is shown in Figure 4, for Kp K from 0 to 5, the roots start at [0, 0.3125] and end
at [-0.625+0.4841j, -0.625-0.4841j], when Kp K = 1 the roots are [0.3536j, -0.3536j].
The second order open loop transfer function is:
Kp
GM =
(τp1 s + 1)(τp2 s − 1)
The characteristic equation is:
Kp K(τ s + 1)
1+ =0
(τp1 s + 1)(τp2 s − 1) τs
τp1 τp2 τ s3 + τ (τp2 − τp1 )s2 + τ (Kp K − 1)s + Kp K = 0

Replacing a root of the form r = iw, we get:

−τp1 τp2 τ w3 + τ (Kp K − 1)w = 0 (1)


2
−τ (τp2 − τp1 )w + Kp K = 0 (2)

Replacing equation (1) into equation (2) we get:


τ (τp2 − τp1 )
Kp K =
τ (τp2 − τp1 ) − τp1 τp2

4
Figure 5: Root locus plot for second order open loop stable process.

For the system to be stable a necessary condition is that the previous value has a positive sign so:

τp2 − τp1 > 0


τ (τp2 − τp1 ) > τp1 τp2

A root locus plot is shown in Figure 5, for Kp K from 0 to 2, the roots start at [0, 0.2809, -
3.125] and end at [0.2889j, -0.2889j, -2.884], when Kp K = 1.095 the roots are [-0.1380+0.3870j,
-0.1380+0.3870j, -2.568].

Problem 6
Find the closed loop stability requirements for a third order open loop unstable process with a
proportional controller:
Kp Kc
BGM =
(τ1 s + 1)(τ2 s + 1)(τ3 s − 1)

Solution
The characteristic polynomial is:

τ1 τ2 τ3 s3 + (τ1 τ3 + τ2 τ3 − τ1 τ2 )s2 + (τ3 − τ1 − τ2 )s + Kp Kc − 1 = 0

The Routh matrix is:


τ3 − τ1 − τ2
 
τ1 τ2 τ3 0
 τ1 τ3 + τ2 τ3 − τ1 τ2 Kp Kc − 1 0
τ3 − τ1 − τ2 − τ1 τ2 τ3 (Kp Kc −1)
 
τ1 τ3 +τ2 τ3 −τ1 τ2 0 0
Kp Kc − 1 0 0

So the conditions for stability are:


τ1 τ2
τ3 >
τ1 + τ2
(τ3 − τ1 − τ2 )(τ1 τ3 + τ2 τ3 − τ1 τ2 )
Kp Kc < 1 +
τ1 τ2 τ3
Kp Kc > 1

5
Problem 7
Find the value of the feedback controller gain Kc that gives a closed loop system with a damping
coefficient of 0.707 for a second order open loop unstable process with τ2 > τ1 :
Kp Kc
BGM =
(τ1 s + 1)(τ2 s − 1)

Solution
The characteristic equation is:

τ1 τ2 s2 + (τ2 − τ1 )s + Kp Kc − 1 = 0

Comparing with the standard form of the characteristic equation we get:



τ2 − τ1 2 0.5
=
τ1 τ2 τp
Kp Kc − 1 1
= 2
τ1 τ2 τp

The solution is:


√τ1 τ2
τp = 2
τ2 − τ1
(τ2 − τ1 )2
Kp Kc = 1 +
2τ1 τ2

Problem 8
What is the ultimate gain and period of the system with a positive zero:
−3s + 1
GM =
(s + 1)(5s + 1)

(a) With a proportional controller?


(b) With a PI controller for τI = 2?

Solution
(a) The characteristic equation is:

K(−3s + 1)
1+
(s + 1)(5s + 1)
5s2 + (6 − 3K)s + 1 + K = 0

For a root of the form r = iw, the associated equations are:

−5w2 + 1 + K = 0
(6 − 3K)w = 0

The solution is K = 2 and w2 = 3/5.


(b) The characteristic equation is:

(−3s + 1) K(1 + 2s)


1+
(s + 1)(5s + 1) 2s
10s3 + (12 − 6K)s2 + (2 − K)s + K = 0

6
For a root of the form r = iw, the associated equations are:
−10w3 + (2 − K)w = 0 (1)
2
−(12 − 6K)w + K = 0 (2)

Replacing equation (1) into equation (2) gives a second order equation for K:
6K 2 − 34K + 24 = 0

The valid solution is K = 0.827 which corresponds to w2 = 0.117.

Problem 9
(a) Sketch the root locus plot of a system with an openloop transfer function:
K
BGM =
(s + 1)(s + 5)(s − 0.5)

(b) For what values of the gain K is the system closed loop stable?
(c) Make a root locus plot of this system in the log(s) plane.

Solution
(a) A root locus plot is shown in Figure 6, for K from 0 to 13.5, the roots start at [-5, -1, 0.5] and
end at [-5.5, 1.414j, -1.414j].
(b) The characteristic equation is:
s3 + 5.5s2 + 2s + K − 2.5 = 0

For a root of the form r = iw the associated equations are:


−w3 + 2w = 0
−5.5w2 − 2.5 + K = 0

The first solution is w = 0, K = 2.5 which corresponds to the roots [-5.108, -0.3915, 0], the
second solution is w2 = 2, K = 13.5 which corresponds to the roots [-5.5, 1.414j, -1.414j], so
the system is closed loop stable for 2.5 < K < 13.5.
(c) A root locus plot is shown in Figure 7.

Problem 10
A two heated tank in series is shown in Figure 8 and the corresponding process block diagram is
shown in Figure 9. Design a feedforward controller, the load disturbance is inlet feed temperature
T0 .
500000 BTU/min
GV =
16 mA
16 mA
GT =
100 ◦ F

Solution
The feedforward control transfer function is:
F = −GL /GM
= −2160 BT U/min ◦ F

7
Figure 6: Root locus plot.

Figure 7: Root locus plot in the log(s) plane.

Figure 8: Two heated tank process diagram.

8
Figure 9: Two heated tank block diagram.

Problem 11
Modify your feedforward controller design of Problem 10 so that it can handle both feed temper-
ature and feed flow rate changes and uses a feedback temperature controller to trim up the steam
flow.

Solution
In problem 10, we have for the first tank:
dT1 Q
τ1 = T0 − T1 +
dt Cp ρF̄
 
1 Q
T1 = T0 + (1)
τ1 s + 1 Cp ρF̄

Where:

τ1 = V1 /F̄ = 1
1
= 1/2160
Cp ρF̄

For the second tank we have:


dT2
τ2 = T1 − T2
dt
T1
T2 = (2)
τ2 s + 1
Where:

τ2 = V2 /F̄ = 5

Replacing equation (1) into equation (2) we get the relation between T2 , T0 and Q:
 
1 Q
T2 = T0 +
(τ1 s + 1)(τ2 s + 1) Cp ρF̄

9
To allow for flow disturbances, the equation for the first tank mus be modified, for simplicity we
assume the outflow is equal to the inflow at all times:

T̄0 F T̄1 F Q
τ1 sT1 = T0 + − T1 − +
F̄ F̄ Cp ρF̄
 
1 T̄0 − T̄1 Q
T1 = T0 + F+ (3)
τ1 s + 1 F̄ Cp ρF̄

For the second tank we have:


T̄1 T̄2
τ2 sT2 = T1 + F − T2 − F
F̄  F̄ 
1 T̄1 − T̄2
T2 = T1 + F (4)
τ2 s + 1 F̄

Replacing equation (3) into equation (4) we get:


 
1 Q T̄0 + τ1 sT̄1 − (τ1 s + 1)T̄2
T2 = T0 + + F
(τ1 s + 1)(τ2 s + 1) Cp ρF̄ F̄

A feedforward controller to eliminate disturbances due to F , with Q as the manipulated variable


is:

F = −GL /GM
= −Cp ρ(T̄0 + τ1 sT̄1 − (τ1 s + 1)T̄2 )

Problem 12
A ”valve position controller” is used to minimize operating pressure in a distillation column (Figure
10). Assume that the openloop process transfer function between column pressure and cooling
water flow GM = P/Fw is known.
(a) Sketch a block diagram of the closed loop system.
(b) What is the closed loop characteristic equation of the system?

Figure 10: Valve position controller.

10
Figure 11: Block diagram for valve position controller.

Solution
(a) A block diagram is shown in Figure 11. Where: GV : valve gain, GP T : pressure transmit-
ter gain, BP : pressure controller gain, BV P C : valve position controller gain, SPP : pressure
controller set point.
(b) From the pressure controller we get:

P = GM FW
P = GM GV CO
P = GM GV BP (SPP − P GP T )
P (1 + GM GV BP GP T ) = GM GV BP SPP

From the SPP loop we get:

SPP = BV P C (SP − CO)


SPP = BV P C (SP − BP (SPP − P GP T ))
SPP (1 + BV P C BP ) = BV P C (SP + BP P GP T )
BV P C (SP + BP P GP T )
SPP =
1 + BV P C BP

Combining both expressions we get:

P (1 + GM GV BP GP T )(1 + BV P C BP ) = GM GV BP BV P C (SP + BP GP T P )
GM GV BP BV P C
P = SP
1 + GM GV BP GP T + BV P C BP

Problem 13
A proportional only controller is used to control the liquid level in a tank by manipulating the
outflow. It has been proposed that the steady state offset of the proportional only controller could
be eliminated by using the combined feedforward feedback system sketched in Figure 12.
The flow rate into the tank is measured. The flow signal is sent through a first order lag with time
constant τF . The output of the lag is added to the output of the level controller. The sum of these
two signals sets the outflow rate. Assume that the flow rate follows the setpoint signal to the flow
controller exactly.

(a) Derive the closed loop transfer function between liquid level h and inflow rate F0 .
(b) Show that there is no steady state offset of level from the setpoint for step changes in inflow
rate.

11
Figure 12: Level control in a tank.

Solution
(a) The level inside the tank and the inflow are related by (G: level transmitter gain, K: level
controller gain, A: cross sectional area of the tank):
F0
sAh = F0 − hGK −
τF s + 1
τF s
h= F0
(sA + GK)(τF s + 1)

(b) The steady state value of h for a step change in F0 is:


τF s
∆h = ∆F0 lim =0
s→0 (sA + GK)(τF s + 1)

Problem 14
A process has a positive zero:
−3s + 1
GM =
(s + 1)(5s + 1)
When a proportional only feedback controller is used, the ultimate gain is 2. Outline your procedure
for finding the optimum value of τD if a proportional derivative controller is used. The optimum
τD will give the maximum value for the ultimate gain.

Solution
The characteristic equation is:
(1 − 3s)
1+ K(1 + τD s)
(s + 1)(5s + 1)
(5 − 3KτD )s2 + (6 + K(τD − 3))s + 1 + K = 0

Replacing a solution of the form r = iw gives:

−(5 − 3KτD )w2 + 1 + K = 0


(6 + K(τD − 3))w = 0

The solution is:


6
K= (1)
3 − τD
1+K
w2 = (2)
5 − 3KτD

12
We require w2 to be a non negative value so:
1+K
>= 0
5 − 3KτD
5 − 3KτD > 0
5
K<
3τD
Replacing equation (1) we get:
5
K<
3τD
6 5
<
3 − τD 3τD
τD < 15/23
From equation (1), we observe that K grows as τD → 3 so the value that maximize K is τD =
15/23 − .

Problem 15
Draw a block diagram of a process that has two manipulated variable inputs (M1 and M2 ) that
each affect the output (X). A feedback controller B1 is used to control X by manipulating M1 since
the transfer function between M1 and X (GM 1 ) has a small time constant and small deadtime.
However, since M1 is more expensive than M2 , we wish to minimize the long term steady state
use of M1 . Therefore, a ”valve position controller” B2 is used to control M1 at M1set . What is the
closed loop characteristics equation of the system?

Solution
The block diagram for the valve position controller is shown in Figure 13.

Figure 13: Block diagram for valve position controller.

From the block diagram we have:


X = GM 1 M1 + GM 2 M2 (1)
set
M1 = B1 (X − X) (2)
M2 = B2 (M1set − M1 ) (3)
Replacing equation (2) and (3) into equation (1) we get:
X = GM 1 B1 (X set − X) + GM 2 B2 (M1set − B1 (X set − X))
GM 1 B1 − GM 2 B2 B1 GM 2 B2
X= X set + M set
1 + GM 1 B1 − GM 2 B2 B1 1 + GM 1 B1 − GM 2 B2 B1 1

13
Problem 16
Make root locus plots for the two processes given below and calculate the ultimate gains and the
gains which give closed loop damping coefficients of 0.707 for both processes.
(a)

K
BGM =
(10s + 1)(50s + 1)

(b)

K(−τ s + 1)
BGM =
(10s + 1)(50s + 1)

With τ = 6.
For part b derive your expression for the ultimate gain as a general function of τ .

Solution
(a) The root locus plot is shown in Figure 14, for K from 0 to 10, the roots start at [-0.1, -0.02]
and end at [-0.06+0.136j, -0.06-0.136j].
The characteristic equation is:

500s2 + 60s + 1 + K = 0

The ultimate gain can be found after replacing a root of the form r = iw, which results in the
following equations:

−500w2 + 1 + K = 0
60w = 0

The solution is w = 0, K = −1, so the system is stable for every value of K. When ζ = 0.707
we have:
1+K 1
= 2
500 τp
60 2 ∗ 0.707
=
500 τp

The solution is K = 2.6, τp = 11.79 and corresponds to the roots [-0.06+0.06j, -0.06-0.06j].
(b) The root locus plot is shown in Figure 15, for K from 0 to 10, the roots start at [-0.1, -0.02]
and end at [0.1483j, -0.1483j].
The characteristic equation is:

500s2 + (60 − Kτ )s + 1 + K = 0

The ultimate gain can be found after replacing a root of the form r = iw, which results in the
following equations:

−500w2 + 1 + K = 0
(60 − Kτ )w = 0

14
Figure 14: Root locus plot for K/(10s + 1)(50s + 1).

The solution is w2 = (τ + 60)/(500τ ), K = 60/τ , for τ = 6, K = 10. When ζ = 0.707 we have:


1+K 1
= 2 (1)
500 τp
60 − Kτ 2 ∗ 0.707
= (2)
500 τp

Replacing equation (2) into equation (1) gives a second order equation for K:
τ 2 K 2 − (120τ + 1000)K + 2600 = 0

The solution is:


p
120τ + 1000 ± (120τ + 1000)2 − 4τ 2 ∗ 2600
K=
2τ 2
For τ = 6, the solution is K = [1.563, 46.22], the smallest being the stability limit and corre-
sponds to the roots [0.0506+0.0506j, 0.0506-0.0506j].

Problem 17
An openloop unstable process has a transfer function containing a positive pole at +1/τ and a
negative pole at −1/aτ . Its steady state gain is unity. If a proportional only controller is used,
what is the value of a that gives a closed loop damping coefficient of 0.5 when the controller gain
is ten times the minimum gain?

Solution
The characteristic equation is:
K
1+  1 1
 =0
s− s+
τ aτ
aτ 2 s2 + τ (1 − a)s + aτ 2 K − 1 = 0
We are looking for the value of K that moves the original positive root to the imaginary axis, so
we replace r = iw in the characteristic equation, which gives:
−aτ 2 w2 + aτ 2 K − 1 = 0
τ (1 − a)w = 0

15
Figure 15: Root locus plot for K(1 − τ s)/(10s + 1)(50s + 1).

The solution is Kmin = 1/(aτ 2 ). Replacing K = 10Kmin and ζ = 0.5, and comparing with the
standard form of the characteristic equation we get:
2 ∗ 0.5 1−a
= (1)
τp aτ
1 9
= 2 (2)
τp2 aτ

Replacing equation (2) into equation (1) we get:

a2 − 11a + 1 = 0

There are two solutions, a = [0.09167, 10.91], we select a = 0.09167, because it makes the system
stable for the selected value of K. A root locus plot is shown in Figure 16, for K from 0 to 10Kmin ,
the dashed line corresponds to ζ = 0.5.

Figure 16: Root locus plot for Problem 17.

16
Problem 18
The ”Smith Predictor” (Figure 17) for deadtime compensation is a feedback controller that has
been modified by feeding back the controller ouput into the controller input through a transfer
function F = GM W D − GM . The transfer function GM W D is the portion of the process open
loop transfer function GM which does not contain the deadtime D. What is the closed loop
characteristic equation of the system?

Figure 17: Smith predictor block diagram.

Solution
From the inner loop we get:

M = B(xset − x − (GM W D − GM )M )
B(xset − x)
M=
1 + B(GM W D − GM )
From the outer loop we get:

x = LGL + M GM
B(xset − x)
x = LGL + GM
1 + B(GM W D − GM )
GL (1 + B(GM W D − GM )) BGM
x= L+ xset
1 + BGM W D 1 + BGM W D
So the characteristic equation contains no deadtime:

1 + BGM W D = 0

For a transfer function that contains no dead time, GM W D = GM , and the equation for x reduces
to:
GL BGM
x= L+ xset
1 + BGM 1 + BGM

Problem 19
We want to analyze the floating pressure ”valve position control” system proposed by Shinskey.
Let the transfer function between cooling water flow and column pressure be:
0.1
GM = psi/gpm
s(s + 1)

17
Figure 18: Block diagram for valve position control.

The span of the pressure transmitter is 50 psi. The control valve has linear installed characteristics
and passes 600 gpm when wide open. A proportional only pressure controller is used.
B1 = K1
An integral only valve position controller is used.
(a) Draw a block diagram of this VPC closed loop system.
(b) What is the closed loop characteristic equation of this system?
(c) Considering only the pressure control loop, determine the value of the pressure controller gain
K1 that gives a closed loop damping coefficient of 0.707 for the pressure control loop. Sketch
a root locus plot for the pressure control loop.
(d) Using the value of K1 found above and with the pressure controller on automatic, determine
the VPC tuning constant (K2 ) that makes the entire VPC/pressure controller system critically
damped, i.e. closed loop damping coefficient of unity. Sketch a root locus plot for the VPC
controller with the pressure controller on automatic.

Solution
(a) The block diagram is shown in Figure 18.
(b) From the pressure loop we get:
P = Fw GM
P = 600 ∗ CO ∗ GM
 
set P
P = 600K1 P − GM
50
P (50 + 600K1 GM ) = 50 ∗ 600K1 GM P set (1)

From the P set loop we get:


xset − CO
P set =
τs  
P
P set τ s = xset − K1 P set −
50
50xset + K1 P
P set = (2)
50τ s + 50K1
Replacing equation (2) into equation (1) we get:
50xset + K1 P
P (50 + 600K1 GM ) = 50 ∗ 600K1 GM
50τ s + 50K1
30000K1 GM
P = xset
50τ s + 50K1 + 600K1 GM τ s

18
Figure 19: Root locus plot for pressure loop.

(c) The pressure characteristic equation is:

50 + 600K1 GM = 0
0.1
50 + 600K1 =0
s(s + 1)
50s2 + 50s + 60K1 = 0

Comparing with the standard form of the characteristic equation we get:



2 0.5/τp = 1
6K1 /5 = 1/τp2

The solution is τp = 2, K1 = 5/12. A root locus plot is shown in Figure 19, for K between 0
and 1, the roots start at [0, -1] and end at [-0.50+0.97j, -0.50-0.97j]. For K1 = 5/12 the roots
are [-0.5+0.5j, -0.5-0.5j].
(d) Replacing K1 = 5/12 into the VPC loop characteristic equation we get:
5 5 0.1τ s
50τ s + 50 ∗ + 600 ∗ ∗ =0
12 12 s(s + 1)
300τ s2 + (300τ + 125)s + 125 + 150τ = 0

Comparing with the standard form of the characteristic equation we get:


2 3τ + 1.25
= (3)
τp 3τ
1 1.25 + 1.5τ
= (4)
τp2 3τ

Replacing equation (3) into equation (4) we get a second order equation for τ :

9τ 2 + 7.5τ − 1.252 = 0

The positive solution is τ = K2−1 = 0.1726, the corresponding root is -1.707. A root locus
plot is shown in Figure 20, for τ from 0.1 to 0.5, the roots start at [-4, -1.167] and end at
[-0.9167+0.7022j, -0.9167-0.7022j].

19
Figure 20: Root locus plot for VPC loop.

Problem 20
Two open loop transfer functions GM 1 and GM 2 are connected in parallel. They both have the
same input M but each has its own output x1 and x2 , respectively. In the closed loop system, a
proportional controller K1 is installed to control X1 by changing M .
However, a cascade system is used where another proportional controller K2 is used to control x2
by changing the setpoint of the K1 controller. Thus we have a parallel cascade system.
(a) Draw a block diagram of the system.
(b) Derive the closed loop transfer function between x2 and xset
2 .

(c) What is the closed loop characteristic equation? Suppose the openloop process transfer func-
tions are:
1
GM 1 =
(s + 1)2
1
GM 2 =
5s + 1

(d) Determine the value of K1 that gives a closed loop damping coefficient of 0.5 in the secondary
loop.
(e) Determine the ultimate gain in the primary K2 loop when the secondary loop gain is K1 = 3.

Solution
(a) The block diagram is shown in Figure 21.
(b) For the secondary loop we have:

x1 = GM 1 K1 (K2 (xset
2 − x2 ) − x1 )
GM 1 K1 K2 (xset
2 − x2 )
= (1)
1 + GM 1 K1

For the primary loop we have:

x2 = GM 2 K1 (K2 (xset
2 − x2 ) − x1 ) (2)

20
Figure 21: Block diagram for parallel cascade process.

Replacing equation (1) into equation (2) we get:

K1 K2 GM 2 xset
2
x2 =
1 + K1 GM 1 + K1 K2 GM 2

(c) The characteristic equation is:

1 + K1 GM 1 + K1 K2 GM 2 = 0
K1 K1 K2
1+ + =0
(s + 1)2 5s + 1
5s3 + (11 + K1 K2 )s2 + (7 + 5K1 + 2K1 K2 )s + 1 + K1 + K1 K2 = 0

(d) The characteristic equation for the secondary loop is:

1 + K1 GM 1 = 0
2
s + 2s + 1 + K1 = 0

Comparing with the standard form of the characteristic equation we get:

2 ∗ 0.5/τp = 2
1/τp2 = 1 + K1

The solution is τp = 0.5 and K1 = 3.


(e) Replacing K1 = 3 in the characteristic equation for the secondary loop we get:

5s3 + (11 + 3K2 )s2 + (22 + 6K2 )s + 4 + 3K2 = 0

Replacing r = iw into the previous equation we get:

−5w3 + (22 + 6K2 )w = 0 (3)


2
−(11 + 3K2 )w + 4 + 3K2 = 0 (4)

The solution is K = −3.25 ± 1.33j, so the secondary loop is stable for every value of K2 .

21

You might also like