Revised: 1-3-04

Chapter 14

14.1

Let GOL(jc) = R + jI

where c is the critical frequency. Then, according to the Bode stability

criterion

| GOL(jc)| = 1 = R 2 + I 2
GOL(jc) = - = tan 1 (I/R)

Solving for R and I: R = -1 and I = 0

Substituting s = jc into the characteristic equation gives,

1 + GOL(jc) = 0

I + R + jI = 0 or R = -1 , I = 0

Hence, the two approaches are equivalent.

14.2

Because sustained oscillations occur at the critical frequency

2
c = = 0.628 min 1
10 min
(a) Using Eq. 14-7,

1 = (Kc)(0.5)(1)(1.0) or Kc = 2

(b) Using Eq. 14-8,

= 0 + 0 +(-c) + 0

or = = 5 min
c

Solution Manual for Process Dynamics and Control, 2nd edition,

Copyright 2004 by Dale E. Seborg, Thomas F. Edgar and Duncan A. Mellichamp

14-1
14.3

(a) From inspection of the Bode diagrams in Tables 13.4 and 13.5, the
transfer function is selected to be of the following form

K ( a s + 1)
G(s) =
s (1 s + 1)( 2 s + 1)

where a, 1, 2 correspond to frequencies of = 0.1, 2, 20 rad/min,

respectively.

Therefore, a = 1/0.1 = 10 min

1=1/2 = 0.5 min

2= 1/20 = 0.05 min

For low frequencies, AR |K/s| = K/

At = 0.01, AR = 3.2, so that K = ()(AR) = 0.032

Therefore,

0.032(10s + 1)
G(s) =
s(0.5s + 1)(0.05s + 1)

(b) Because the phase angle does not cross -180, the concept of GM is
meaningless.

14.4

The following process transfer can be derived in analogy with Eq. 6-71:

H1 ( s ) R1
=
Q1 ( s ) ( A1R1 A2 R2 ) s 2 + ( A1R1 + A2 R1 + A2 R2 ) s + 1

For R1=0.5, R2 = 2, A1 = 10, A2 = 0.8:

14-2
 0.5
Gp(s) = (1)
8s + 7 s + 1
2

0.5
For R2 = 0.5: Gp(s) = (2)
2s + 5.8s + 1
2

(a) For R2 = 2

7c 0.5
Gp= tan-1 2
, |Gp| =
1 8c (1 8c ) + (7c ) 2
2 2

For Gv = Kv = 2.5, v=0, |Gv| = 2.5

1.5 1 .5
For Gm = , m= -tan-1(0.5) , |Gm| =
0.5s + 1 (0.5c ) 2 + 1

Kcu and c are obtained using Eqs. 14-7 and 14-8:

7c
-180 = 0 + 0 + tan-1 2
tan-1(0.5c)
1 8c

Solving, c = 1.369 rad/min.

0.5 1.5
1 = ( K cu )(2.5)
2 2 (0.5 ) 2 + 1
(1 8c ) + (7c )
2
c

Substituting c = 1.369 rad/min, Kcu = 10.96, cKcu = 15.0

For R2=0.5

5.8c 0.5
Gp = tan-1 2
, |Gp| =
1 2c 2 2
(1 2c ) + (5.8c )
2

5.8c
-180 = 0 + 0 + tan-1 2
tan-1(0.5c)
1 2c

Solving, c = 2.51 rad/min.

Substituting c = 2.51 rad/min, Kcu = 15.93, cKcu = 40.0

14-3
(a) From part (a), for R2=2,

c = 1.369 rad/min, Kcu = 10.96

2
Pu = = 4.59 min
c

Using Table 12.6, the Ziegler-Nichols PI settings are

Kc = 0.45 Kcu = 4.932 , I= Pu/1.2 = 3.825 min

Using Eqs. 13-63 and 13-62 ,

c= -tan-1(-1/3.825)

2
1
|Gc| = 4.932 +1
3.825

Then, from Eq. 14-7

1 7c
-180 = tan-1 -1
+ 0 + tan 2
tan-1(0.5c)
3.825c 1 8c

Solving, c = 1.086 rad/min.

Using Eq. 14-8,

Ac = AROL|=c =

2
1
+ 1 (2.5)
0 .5
= 4.932


3.825 c

2 2
(1 8 c ) + (7c )
2

1.5

(0.5 ) 2 + 1
c

= 0.7362

Therefore, gain margin GM =1/Ac = 1.358.

Solving Eq.(14-16) for g

AROL|=c = 1 at g = 0.925

14-4
 Substituting into Eq. 14-7 gives g=|=g = 172.7.

Therefore, phase margin PM = 180+ g = 7.3.

14.5

(a) K=2 , = 1 , = 0.2 , c=0.3

Using Eq. 12-11, the PI settings are

1
Kc = =1 , I = = 1 min,
K + c

Using Eq. 14-8 ,

1
-180 = tan-1 0.2c tan-1(c) = -90 0.2c
c

/2
or c = = 7.85 rad/min
0.2

Using Eq. 14-7,

1 2 2
Ac = AR OL = +1 = = 0.255
= c
c
2 2
c + 1 c

From Eq. 14-11, GM = 1/Ac = 3.93.

(b) Using Eq. 14-12,

g = PM 180 = 140 = tan-1(-1/0.5g) 0.2g tan-1(g)

Solving by trial and error, g = 3.04 rad/min

1
2

+1 2
AR OL = 1 = Kc
= g 0.5 g + 1
2
g

Substituting for g gives Kc = 1.34. Then from Eq. 14-8

14-5
 1
180 = tan-1 0.2c tan-1(c)
0.5c

Solving by trial and error, c =7.19 rad/min.

From Eq. 14-7,

1
2 2
Ac = AR OL = 1.34 + 1 = 0.383

0.5c
= c
2
c + 1

From Eq. 14-11, GM = 1/Ac = 2.61

(c) By using Simulink-MATLAB, these two control systems are compared for
a unit step change in the set point.

1.4

1.2

0.8
y
part (a)
0.6
part (b)

0.4

0.2

0
0 1 2 3 4 5
time

Fig S14.5. Closed-loop response for a unit step change in set point.

The controller designed in part a) (Direct Synthesis) provides better

performance giving a first-order response. Part b) controller yields a large
overshoot.

14-6
14.6

(a) Using Eqs. 14-7 and 14-8,

Ym 4 2 + 1 2 0.4
AR OL = = Kc (1.0)
Ysp 0.01 2
+ 1 0.25 2 + 1 25 2 + 1

= tan-1(2) tan-1(0.1) tan-1(0.5) (/2) tan-1(5)

2
Bode Diagram
10

0
10
AR/Kc

-2
10

-4
10
-90

-135
Phase (deg)

-180

-225

-270
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/sec)

Figure S14.6a. Bode plot

(b) Using Eq.14-12

g = PM 180 = 30 180 = 150

From the plot of vs. : g = -150 at g = 1.72 rad/min

14-7
 AR OL AR OL
From the plot of vs : = 0.144
Kc Kc = g

1
Because AR OL = g
=1 , Kc = = 6.94
0.144

(c) From the phase angle plot:

= -180 at c = 4.05 rad/min

AR OL AR OL
From the plot of vs , = 0.0326
Kc Kc = c

Ac = AR OL = c
= 0.326

From Eq. 14-11, GM = 1/Ac = 3.07.

2
Bode Diagram
10

0
10
AR/Kc

-2
10

-4
10
-90

-135
Phase (deg)

-180

-225

-270
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/sec)

Figure S14.6b. Solution for part (b) using Bode plot.

14-8
 2
Bode Diagram
10

0
10
AR/Kc

-2
10

-4
10
-90

-135
Phase (deg)

-180

-225

-270
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/sec)

Figure S14.6c. Solution for part (c) using Bode plot.

14.7

(a) For a PI controller, the |Gc| and Gc from Eqs. 13.62 and 13.63 need to
be included in the AR and given for GvGpGm to obtain AROL and OL.
The results are tabulated below

AR |Gc|/Kc AROL/Kc Gc OL
0.01 2.40 250 600 -3 -89.8 -92.8
0.10 1.25 25.020 31.270 -12 -87.7 -99.7
0.20 0.90 12.540 11.290 -22 -85.4 -107.4
0.50 0.50 5.100 2.550 -41 -78.7 -119.7
1.00 0.29 2.690 0.781 -60 -68.2 -128.2
2.00 0.15 1.601 0.240 -82 -51.3 -133.3
5.00 0.05 1.118 0.055 -122 -26.6 -148.6
10.00 0.02 1.031 0.018 -173 -14.0 -187.0
15.00 0.01 1.014 0.008 -230 -9.5 -239.5

From Eq. 14-12, g = PM 180 = 45 180 = -135.

Interpolating the above table, OL= -135 at g = 2.5 rad/min and

14-9
 AR OL
= 0.165
Kc = g

1
Because AR OL = g
=1 , Kc = = 6.06
0.165

(b) From the table above,

AR OL
OL= -180 at c = 9.0 rad/min and = 0.021
Kc = c

Ac = AR OL = c
= 0.021 Kc = 0.127

From Eq. 14-11,

GM = 1/Ac = 1/0.127 = 7.86

(c) From the table in part (a),

OL= -180 at c = 10.5 rad/min and AR = = 0.016.

c

2 1
Therefore, Pu = = 0.598 min and Kcu = = 62.5.
c AR =
c

Using Table 12.6, the Ziegler-Nichols PI settings are

Kc = 0.45 Kcu = 28.1, I = Pu/1.2 = 0.50 min

Tabulating AROL and OL as in part (a) and the corresponding values of M

using Eq. 14-18 gives:

|Gc| Gc AROL OL M
0.01 5620 -89.7 13488 -92.7 1.00
0.10 563.0 -87.1 703 -99.1 1.00
0.20 282.0 -84.3 254 -106.3 1.00
0.50 116.0 -76.0 57.9 -117.0 1.01
1.00 62.8 -63.4 18.2 -123.4 1.03
2.00 39.7 -45.0 5.96 -127.0 1.10
5.00 30.3 -21.8 1.51 -143.8 1.64
10.00 28.7 -11.3 0.487 -184.3 0.94
15.00 28.3 -7.6 0.227 -237.6 0.25

Therefore, the estimated value is Mp =1.64.

14-10
14.8

Kcu and c are obtained using Eqs. 14-7 and 14-8. Including the filter GF
into these equations gives

-180 = 0 + [-0.2c tan-1(c)]+[-tan-1(Fc)]

Solving,

c = 8.443 for F = 0
c = 5.985 for F = 0.1

Then from Eq. 14-8,

2 1
1 = (K cu )
2 2 2
c + 1 F c + 1

Solving for Kcu gives,

Kcu = 4.251 for F = 0

Kcu = 3.536 for F = 0.1

Therefore,

cKcu = 35.9 for F = 0

c Kcu= 21.2 for F = 0.1

Because cKcu is lower for F = 0.1, filtering the measurement results in

worse control performance.

14.9

(a) Using Eqs. 14-7 and 14-8,

1 5 1
AR OL = K c + 1 (1.0)
25 2
100 + 1 + 1
2 2

= tan-1(-1/5) + 0 + (-2 tan-1(10)) + (- tan-1())

14-11
 2
Bode Diagram
10

1
10

0
AR/Kc
10

-1
10

-2
10
-100

-150
Phase (deg)

-200

-250

-300

-350
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/sec)

Figure S14.9a. Bode plot

(b) Set = 180 and solve for to obtain c = 0.4695.

Then AR OL = c
= 1 = Kcu(1.025)

Therefore, Kcu = 1/1.025 = 0.976 and the closed-loop system is stable for
Kc 0.976.

(c) For Kc = 0.2, set AROL = 1 and solve for to obtain g = 0.1404.

Then g = = = -133.6
g

From Eq. 14-12, PM = 180 + g = 46.4

(d) From Eq. 14-11

14-12
 1 1
GM = 1.7 = =
Ac AR OL =
c

From part (b), AR OL = c

= 1.025 Kc

Therefore, 1.025 Kc = 1/1.7 or Kc = 0.574

2
Bode Diagram
10

1
10

0
AR/Kc

10

-1
10

-2
10
-150

-180
Phase (deg)

-200

-250

-300

-350
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/sec)

Figure S14.9b. Solution for part b) using Bode plot.

2
Bode Diagram
10

1
10

0
AR/Kc

10

-1
10

-2
10
-150

-180
Phase (deg)

-200

-250

-300

-350
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/sec)

Figure S14.9c. Solution for part c) using Bode plot.

14-13
14.10

0.047 5.264
(a) Gv ( s ) = 112 =
0.083s + 1 0.083s + 1

2
G p ( s) =
(0.432s + 1)(0.017 s + 1)

0.12
Gm ( s ) =
0.024 s + 1

Using Eq. 14-8

-180 = 0 tan-1(0.083c) tan-1(0.432c) tan-1(0.017c)

tan-1(0.024c)

Solving by trial and error, c = 18.19 rad/min.

Using Eq. 14-7,

5.624 2
1 = ( K cu )
(0.083 ) 2 + 1 (0.432 ) 2 + 1 (0.017 ) 2 + 1
c c c

0.12

(0.024 ) 2 + 1
c
Substituting c=18.19 rad/min, Kcu = 12.97.

Pu = 2/c = 0.345 min

Using Table 12.6, the Ziegler-Nichols PI settings are

Kc = 0.45 Kcu = 5.84 , I=Pu/1.2 = 0.288 min

(b) Using Eqs.13-62 and 13-63

c = Gc = tan-1(-1/0.288)= -(/2) + tan-1(0.288)

2
1
|Gc| = 5.84 +1
0.288

Then, from Eq. 14-8,

14-14
 - = (/2) + tan-1(0.288c) tan-1(0.083c) tan-1(0.432c)

tan-1(0.017c) tan-1(0.024c)

Solving by trial and error, c = 15.11 rad/min.

Using Eq. 14-7,

2
1 5.264
Ac = AR OL =c = 5.84
+ 1
0.288c
(0.083 ) 2 + 1
c

2 0.12

(0.432 ) 2 + 1 (0.017 ) 2 + 1 (0.024 ) 2 + 1
c c c

= 0.651

Using Eq. 14-11, GM = 1/Ac = 1.54.

Solving Eq. 14-7 for g gives

AR OL = g
=1 at g = 11.78 rad/min
Substituting into Eq. 14-8 gives

g = = = (/2) + tan-1(0.288g) tan-1(0.083g)

g

tan-1(0.432g) tan-1(0.017g) tan-1(0.024g) = -166.8

Using Eq. 14-12,

PM = 180 + g = 13.2

14.11

(a)
10 1.5
| G |= (1)
2 2
+ 1 100 + 1

= tan-1() tan-1(10) 0.5

14-15
 Bode Diagram
2
10

1
10

AR 0
10

-1
10

-90
Phase (deg)

-180

-270
-2 -1 0 1
10 10 10 10
Frequency (rad/sec)

Figure S14.11a. Bode plot for the transfer function G=GvGpGm.

(b) From the plots in part (a)

G = -180 at c = 1.4 and |G|=c = 0.62

AR OL = c
= 1= (- Kcu) |G|=c

Therefore, Kcu = -1/0.62 = -1.61 and

Pu = 2/c = 4.49

Using Table 12.6, the Ziegler-Nichols PI-controller settings are:

Kc = 0.45Kcu = -0.72 , I = Pu/1.2 = 3.74

Including the |Gc| and Gc from Eqs. 13-62 and 13-63 into the results of
part (a) gives

1
2
15
AR OL = 0.72 + 1

3.74 + 1 100 + 1
2 2

14-16
 2.89 14.0 2 + 1
=
2 + 1 100 2 + 1
= tan-1(-1/3.74) tan-1() tan-1(10) 0.5

Bode Diagram
4
10

2
10
AR

0
10

-2
10

90

-90
Phase (deg)

-180

-270

-360
-2 -1 0 1
10 10 10 10
Frequency (rad/sec)

Figure S14.11b. Bode plot for the open-loop transfer function GOL=GcG.

(c) From the graphs in part (b),

= -180 at c=1.15

AR OL = c
= 0.63 < 1

Hence, the closed-loop system is stable.

14-17
 Bode Diagram
4
10

2
10

AR
0
10

-2
10

-90

-180
Phase (deg)

-270

-2 -1 0 1
10 10 10 10
Frequency (rad/sec)

`
Figure S14.11c. Solution for part (c) using Bode plot.

(d) From the graph in part b),

amplitude of ym (t )
AR OL = 0.5
= 2.14 =
amplitude of ysp (t )
Therefore, the amplitude of ym(t) = 2.14 1.5 = 3.21.

(e) From the graphs in part (b), AR OL = 0.5

= 2.14 and =0.5 =-147.7.

Substituting into Eq. 14-18 gives M = 1.528. Therefore, the amplitude of

y(t) = 1.528 1.5 = 2.29 which is the same as the amplitude of ym(t)
because Gm is a time delay.

(f) The closed-loop system produces a slightly smaller amplitude for = 0.5.
As approaches zero, the amplitude approaches one due to the integral
control action.

14-18
14.12

(a) Schematic diagram:

TC
Hot fluid

TT

Cold fluid

Mixing Point Sensor

Block diagram:

(b) GvGpGm = Km = 6 mA/mA

GTL = e-8s

GOL = GvGpGmGTL = 6e-8s

If GOL = 6e-8s,

| GOL(j) | = 6

GOL (j) = -8 rad

Find c: Crossover frequency generates 180 phase angle = radians

-8c = - or c = /8 rad/s

14-19
 2 2
Find Pu: Pu = = = 16s
c / 8
1 1
Find Kcu: Kcu = = = 0.167
| G p ( j c ) | 6

Ziegler-Nichols decay ratio settings:

PI controller:

Kc = 0.45 Kcu = (0.45)(0.167) = 0.075

I = Pu/1.2 = 16/1.2 = 13.33 sec

PID controller:

Kc = 0.6 Kcu = (0.6)(0.167) = 0.100

I = Pu/2 = 16/2 = 8 s
D = Pu/8 = 16/8 = 2 s

(c)

1.4

1.2

0.8
y
PID control
0.6 PI control

0.4

0.2

0
0 30 60 90 120 150
t

Fig. S14.12. Set-point responses for PI and PID control.

14-20
 (d) Derivative control action reduces the settling time but results in a more
oscillatory response.

14.13

(a) From Exercise 14.10,

5.264
Gv ( s ) =
0.083s + 1
2
G p ( s) =
(0.432s + 1)(0.017 s + 1)
0.12
Gm ( s) =
(0.024s + 1)
1
The PI controller is Gc ( s ) = 51 +
0.3s
Hence the open-loop transfer function is

GOL = Gc Gv G p Gm

Rearranging,

6.317 s + 21.06
GOL = 5
1.46 10 s + 0.00168s 4 + 0.05738s 3 + 0.556s 2 + s
5

14-21
 By using MATLAB, the Nyquist diagram for this open-loop system is

Nyquist Diagram
1

0.5

-0.5

-1
Imaginary Axis

-1.5

-2

-2.5

-3

-3.5

-4
-3 -2.5 -2 -1.5 -1 -0.5 0
Real Axis

Figure S14.13a. The Nyquist diagram for the open-loop system.

1
(b) Gain margin = GM =
AR c

where ARc is the value of the open-loop amplitude ratio at the critical
frequency c. By using the Nyquist plot,

Nyquist Diagram
1

0.5

-0.5

-1
Imaginary Axis

-1.5

-2

-2.5

-3

-3.5

-4
-3 -2.5 -2 -1.5 -1 -0.5 0
Real Axis

Figure S14.13b. Graphical solution for part (b).

14-22
 = -180 ARc = | G(jc)| = 0.5

Therefore the gain margin is GM = 1/0.5 = 2.

14.14

1
To determine max | em | < , we must calculate Mp based on the CLTF
Mp
with IMC controller design. In order to determine a reference Mp, we
~
assume a perfect process model (i.e. G G = 0 ) for the IMC controller
design.

C *
= Gc G
R

Factoring,
~ ~ ~
G = G+ G
~ ~ 10
G+ = e s , G =
2s + 1
* 2s + 1
Gc = f
10

Filter Design: Because = 2 s, let c = /3 = 2/3 s.

1
f =
2 3s +1

* 2s + 1 1 2s + 1
Gc = =
10 2 3 s + 1 20 3 s + 10

C 2 s + 1 10e s 10e s

*
= Gc G =
=
R 20 3 s + 10 2 s + 1 20 3s + 10

M p =1

The relative model error with K as the actual process gain is:

14-23
 Ke s 10e s
~
G G 2s + 1 2s + 1 K 10
em = ~ = =
G 10e s 10
2s + 1
K 10
Since Mp = 1, max | em | = <1
10

K 10
<1 K < 20
10

K 10
> 1 K > 0
10

0 < K < 20 for guaranteed closed-loop stability.

14.15

Denote the process model as,

~ 2e 0.2 s
G=
s +1

and the actual process as:

2e 0.2 s
G=
s + 1

The relative model error is:

~
G ( s ) G ( s ) (1 ) s
( s) = ~ =
G ( s) s + 1

Let s = j. Then,

(1 ) j | (1 ) |
= = (1)
j + 1 | j + 1 |

14-24
or

| (1 ) |
=
2 2 + 1

Because | | in (1) increases monotonically with ,

|1 |
max | | = lim | | = (2)

Substituting (2) and Mp = 1.25 into Eq. 14-34 gives:

|1 |
< 0.8

This inequality implies that

1
< 0.8 1 < 1.8 > 0.556

and

1
< 0.8 0.2 < 1 <5

Thus, closed-loop stability is guaranteed if

0.556 < < 5

14-25