Sunteți pe pagina 1din 20

INDIAN CHEMICAL ENGINEER © 2014 Indian Institute of Chemical Engineers

Vol. 56 No. 4 December 2014, pp. 336–354


Print ISSN: 0019-4506, Online ISSN: 0975-007X, http://dx.doi.org/10.1080/00194506.2014.927181

Two-degree-of-freedom Controllers for Double


Integrating Systems with Time Delay and Unstable
First Order plus Time Delay Systems with an
Integrator

C.V. Nageswara Raoa and R. Padma Sreeb*


a
Department of Chemical Engineering, Gayatri Vidya Parishad College of Engineering (A),
Visakhapatnam 530048, India
b
Department of Chemical Engineering, AU College of Engineering (A), Visakhapatnam 530 003, India

Abstract: In this paper, design of two-degree-of-freedom Proportional Integral and Derivative (PID)
controller for Double Integrator with Time Delay systems and Unstable First Order plus Time Delay
systems with an Integrator is proposed. The robustness of the proposed controller for uncertainty in
model parameters is evaluated considering one parameter at a time using Kharitonov’s theorem. The
stability region for all model parameters is evaluated for the proposed controller and compared with the
controllers designed by recently reported methods. The proposed controller is applied to various transfer
function models and the performance comparison of the proposed controller is made with the PID
controllers designed by recently reported methods to show the efficiency of the proposed controllers.

Keywords: Two-degree-of-freedom controller, Double-integrating systems with time delay, Unstable


FOPTD system with an integrator, Kharitonov’s theorem, PID controller.

Introduction
Some processes in chemical and industrial practice are unstable because of their integrating
character. For example, once the current balance status of such a process is disturbed, the
process output varies continuously at a certain speed rather than automatically recovering the
previous value or reaching a new stable value. The phenomenon is very disadvantageous and

*Author for Correspondence. Email: padvan@gmail.com


Two-Degree-of-Freedom Controllers for Integrating Systems 337

dangerous on most occasions. Therefore, how to reliably control such kinds of processes is
always a classical research subject in control field.
Generally two typical plant models of second-order integrating processes are Unstable
kp eLs
First Order plus Time Delay systems with an Integrator (UFOPTDI) sðss1Þ and Double
Ls
k e
Integrator with Time Delay system (DITD) p s2 .
Double integrator systems are simple but an important class of second-order systems as
they model single-degree-of-freedom translational and rotational systems. Applications arise
in steering control systems [1], low-friction, free rigid-body motion, such as single-axis
spacecraft rotation [2] and rotary crane motion [3]. Many industrial processes, such as
aerospace control systems, direct current motors and high-speed disc drives, show the
dynamics of such double-integrator type [4, 5]. Oxygen control in a fermenter (filamental
fungal growth) represents double integrator with a zero [6].
Open-loop unstable systems, observed in the process industries (exothermic stirred
reactors with back mixing, batch reactors and combined feed/effluent heat exchanger with
adiabatic exothermic reactor) in operating the process at unstable steady state, can be
modelled as a second-order dead time process with two unstable poles and unstable FOPTD
systems with an integrator. They are difficult to control than open-loop stable processes due
to presence of unstable poles and due to the difficulty in achieving the desired performance.
Very limited literature is available on the methods for designing Proportional Integral
and Derivative (PID) controllers for double-integrating systems with time delay. Rao and
Bernstein [7] have given a review of various methods to control a double-integrator process.
But the effect of time delay is not studied. Later, Skogestad [8] proposed Internal Model
Control (IMC) method for the design of controllers for double-integrating systems with time
delay and Liu et al. [4] proposed two-degree-of-freedom controller based on the H2 optimal
controller.
The literature methods available for the control design of unstable FOPTD system
with an integrator include polynomial approach [9], two-degree-of-freedom control scheme
[10–12], IMC method [13] and direct synthesis method (DS-s) [5].
However, for the above plant types, very few analytical control design methods have
been reported in literature. Therefore, further development on the simple and analytical
design methods of practical PID controller for the DITD systems and UFOPTDI systems is
desirable.
In this paper, it is intended to design two-degree-of-freedom controllers for DITD and
UFOPTDI system. The DITD systems and UFOPTDI systems are stabilised in the inner
loop by using proportional derivative (PD) controller. To the stabilised system, PID
controller is designed by equating the denominator of the stabilised system with the
numerator of the outer-loop PID controller. Using the phase angle criteria for the combined
stabilised system and outer-loop PID controller, the crossover frequency (wc) is obtained. The
ultimate value of controller gain (outer loop) is obtained by using gain margin (GM)
criterion. Using a GM of 1.5–2.5, the design value of controller gain of the outer-loop
controller is obtained.
The performance of the two-degree-of-freedom controller (TDF) for various transfer
function models is compared with the controller designed by recently reported methods like
INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014
338 R AO AND S REE

DS-s [5], TDF-s [11] and IMC method (IMC-PID controller) [14] for both servo and
regulatory problems.

Design of the Controller forLsPure Double-integrator Process with Time Delay


k e
The process transfer function p s2 is stabilised by a PD controller in the inner loop (refer to
Fig. 1).
With the PD controller [k1(1+τds)] in the inner-feedback loop, the inner closed-loop
transfer function Gp1 (s) is written as:

GC1 Gp ðsÞ k1 ð1 þ sd sÞ kp eLs


Gp1 ðsÞ ¼ ¼ 2 ð1Þ
1 þ k1 ð1 þ sd sÞGp ðsÞ s þ k1 kp ð1 þ sd sÞeLs

Using a Taylor series expansion, the time-delay term in the denominator of Equation (1) is
approximated by:

eLs ffi 1  Ls þ 0:5 L2 s2 ð2Þ

Substituting Equation (2) in the denominator of Equation (1), Gp1 (s) is expressed as:

k1 kp ð1 þ sd sÞ eLs
Gp1 ðsÞ ¼
s3 ð0:5 L2 k1 kp sd Þ þ s2 ð1 þ 0:5 L2 k1 kp  Lsd k1 kp Þ þ sðLk1 kp þ sd k1 kp Þ þ k1 kp
ð3Þ

The characteristic equation (denominator) of Equation (3) should have negative (stable) poles
so that the inner loop is stable. The following are the necessary conditions that must be
satisfied from the Routh stability criteria.

sd > L ð4Þ

1
k1 < ð5Þ
kp ðsd L  0:5 L2 Þ

Fig. 1. Two-degree-of-freedom control loop.

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


Two-Degree-of-Freedom Controllers for Integrating Systems 339

The above constraints are considered while tuning the inner-loop PD controller. The
denominator of Equation (3) is written as:

ðsf s þ 1Þ ðr1 s2 þ r2 s þ r3 Þ ð6Þ


On expanding Equation (6), we get:

s3 ðr1 sf Þ þ s2 ðr2 sf þ r1 Þ þ s ðr3 s f þ r2 Þ þ r3 ð6aÞ


3 2
Equating the coefficients of s , s , s and constant term in the denominator of Equation (3) and
Equation (6a), the following equations are obtained:

0:5 k1 kp L2 sd ¼ r1 sf ð7Þ

1 þ 0:5 k1 kp L2  Lsd k1 kp ¼ r2 sf þ r1 ð8Þ

Lk1 kp þ sd k1 kp ¼ r3 sf þ r2 ð9Þ

k1 kp ¼ r 3 ð10Þ
Using Equations (7), (9) and (10) in Equation (8), the following equation for τf is obtained.

s3f ðk1 kp Þ þ s2f ðLk1 kp þ sd k1 kp Þ  sf ð1 þ 0:5 L2 k1 kp  Lsd k1 kp Þ þ 0:5 L2 k1 kp sd ¼ 0


ð11Þ
Using roots of MATrix LABoratory (MATLAB), Equation (11) is solved for filter time
constant τf. Selecting the positive value from the available values for τf, values of r1, r2, r3 are
obtained from Equations (7), (8) and (10), respectively. Now the inner-loop transfer function
Equation (3) is written as:

k1 kp ð1 þ sd sÞ eLs
Gp1 ðsÞ ¼ ð12Þ
ðsf s þ 1Þ ðr1 s2 þ r2 s þ r3 Þ
A PID controller is designed in the outer loop to control the stabilised process in the inner
loop [Equation (12)].
The outer loop PID controller transfer function is written as:
1
GC ¼ kc ð1 þ þ sD sÞ ð13Þ
sI s
The numerator of the outer-loop PID controller is made equal to the part of the denominator
of the inner-loop stabilised system. Then,
r2 r1
sI ¼ and sI sD ¼ ð14Þ
r3 r3

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


340 R AO AND S REE

The values of τI and τD are obtained from Equation (14). Hence, the loop transfer function is
given by:

k1 kp ð1 þ sd sÞ eLs kc
Gp1 ðsÞ GC ðsÞ ¼ ð15Þ
ðsf s þ 1Þ r3 s1 s

where kc is determined based on phase-angle criterion and GM criterion.


The phase-angle criterion for cascaded process Equation (15) is given by:
p
/ðxÞ ¼   Lx  tan1 ðsf xÞ þ tan1 ðsd xÞ ð16Þ
2
At the crossover frequency ω = ωc, ϕ(ωc)=−π.
Hence, Equation (16) is written as:
p
p ¼   Lxc  tan1 ðsf xc Þ þ tan1 ðsd xc Þ ð17Þ
2
This non-linear equation is solved by using fsolve of MATLAB for ωc. The GM criterion for
the loop transfer function Equation (15) is given by:
 
Gp1 ðxc ÞGc ðxc Þ ¼ 1 ð18Þ

From the above equation, ultimate controller gain is given by:


qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r3 sI xc s2f x2c þ 1
kc;max ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð19Þ
k1 kp ðsd xc Þ2 þ 1

A GM of 1.5–2.5 is used to obtain the design value of kc.

Design of Controller for Unstable FOPTD Systems with an Integrator


The process transfer function given by Equation (20) is stabilised by a PD controller in the
inner loop (refer to Fig. 1).

kp e Ls
Gp ¼ ð20Þ
s ðs s  1Þ

With the PD controller in the inner-feedback loop, the inner closed-loop transfer function
Gp1 (s) can be obtained as:

GC1 Gp ðsÞ k1 ð1 þ sd sÞ kp eLs


Gp1 ðsÞ ¼ ¼ ð21Þ
1 þ GC1 Gp ðsÞ s ðs s  1Þ þ k1 kp ð1 þ sd sÞ eLs

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


Two-Degree-of-Freedom Controllers for Integrating Systems 341

Using a Taylor series expansion for the time-delay term [Equation (2)], in the denominator of
Equation (21), the inner-loop transfer function is given by:

k1 kp ð1 þ sd sÞ eLs
Gp1 ðsÞ ¼
s3 ð0:5 L2 k1 kp sd Þ þ s2 ðs þ 0:5 L2 k1 kp  L sd k1 kp Þ þ s ð1  L k1 kp þ sd k1 kp Þ þ k1 kp
ð22Þ
The characteristic equation (denominator) of Equation (22) should have negative (stable)
poles so that the inner loop is stable. The followings are the necessary conditions that must be
satisfied from the Routh stability criteria.
1
k1 > ð23Þ
kp ðsd  LÞ

s þ 0:5 k1 kp L2
sd < ð24Þ
Lk1 kp
The above constraints are considered while tuning the inner-loop PD controller. The
denominator of Equation (22) can be written as:

ðsf s þ 1Þðq1 s2 þ q2 s þ q3 Þ ð25Þ


On expanding Equation (25), the following equation is obtained.

s3 ðq1 sf Þ þ s2 ðq2 sf þ q1 Þ þ sðq3 sf þ q2 Þ þ q3 ð26Þ


3 2
Equating the coefficients of s , s , s and constant term in the denominator of Equation (22)
and Equation (26), the following equations are obtained.

0:5 k1 kp L2 sd ¼ q1 sf ð27Þ

s þ 0:5 k1 kp L2  L sd k1 kp ¼ q2 sf þ q1 ð28Þ

1  Lk1 kp þ sd k1 kp ¼ q3 sf þ q2 ð29Þ

k 1 k p ¼ q3 ð30Þ
Using Equations (27), (28) and (30) in Equation (29), the following equation for τf is
obtained.

s3f ðk1 kp Þ þ s2f ð1  L k1 kp þ sd k1 kp Þ  sf ð s þ 0:5 L2 k1 kp  L sd k1 kp Þ


þ 0:5 L2 k1 kp sd ¼ 0 ð31Þ
Using roots of MATLAB, Equation (31) is solved for filter time constant τf. Selecting the
positive value from the available values for τf, the values of q1, q2, q3 are obtained from
INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014
342 R AO AND S REE

Equations (27), (28) and (30), respectively. Now, Equation (22) can be written as:

k1 kp ð1 þ sd sÞ eLs
Gp1 ðsÞ ¼ ð32Þ
ðsf s þ 1Þ ðq1 s2 þ q2 s þ q3 Þ
As the unstable FOPTD system with an integrator is stabilised with the PD controller in the
inner loop, a PID controller is designed in the outer loop to control the stabilised process in
the inner loop. The PID controller transfer function is written as in Equation (13).
The numerator of the outer loop PID controller is made equal to the part of the
denominator of the inner-loop stabilised system. Then,
q2 q1
sI ¼ and sI sD ¼ ð33Þ
q3 q3
The values of τI and τD are obtained from Equation (33).
Hence, the loop transfer function is given by:

k1 kp ð1 þ sd sÞ eLs kc
Gp1 ðsÞ GC ðsÞ ¼ ð34Þ
ðsf s þ 1Þ q3 s1 s
The phase-angle criterion for Equation (34) is given by:
p
/ðxÞ ¼   Lx  tan1 ðsf xÞ þ tan1 ðsd xÞ ð35Þ
2
At the crossover frequency ω = ωc, φ(ωc) = −π. Hence, Equation (35) is written as:
p
p ¼   Lxc  tan1 ðsf xc Þ þ tan1 ðsd xc Þ ð36Þ
2
This non-linear equation is solved by using fsolve of MATLAB for ωc. The GM criterion for
the open-loop transfer function Equation (34) is given by:
 
Gp1 ðxc ÞGc ðxc Þ ¼ 1 ð37Þ

From the above equation, maximum controller gain is given by:


qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
q3 sI xc s2f x2c þ 1
kc; max ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð38Þ
k1 kp ðsd xc Þ2 þ 1

A GM of 2.0 is used to obtain the design value of kc.

Robustness of the Controller


A control system is said to be robust if the closed-loop system is stable even when the model
parameters of the actual process are different than that used for the controller design. To
compare the robustness of the different controller-design methods, the range of uncertainty in
each of the model parameters for which the controller is stable is to be calculated. The
INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014
Two-Degree-of-Freedom Controllers for Integrating Systems 343

robustness of the closed-loop system for the perturbation in time delay, time constant and
process gain separately is analysed theoretically by Kharitonov’s theorem [15].
The process transfer function is given by:

kp e Ls
Gp ðsÞ ¼ ð39Þ
s ðs s  aÞ
When a = 0 and τ = 1, it becomes a double-integrating system with time delay and when a =
1, it is an unstable FOPTD system with an integrator. With the PD controller in the inner-
feedback loop, the inner closed-loop transfer function Gp1(s) can be obtained as:

GC1 Gp ðsÞ k1 ð1 þ sd sÞ kp eLs


Gp1 ðsÞ ¼ ¼ ð40Þ
1 þ GC1 Gp ðsÞ sðss  aÞ þ k1 kp ð1 þ sd sÞeLs
The controller transfer function is given by:
1
Gc ¼ kc ð1 þ þ sD sÞ ð41Þ
sI s
The loop transfer function is given by:

1 k1 kp ð1 þ sd sÞeLs
Gc Gp1 ¼ kc ð1 þ þ sD sÞ ð42Þ
sI s sðss  aÞ þ k1 kp ð1 þ sd sÞeLs
The characteristic equation of the system with PID controller is given by:
1 þ Gc Gp1 ¼ 0 ð43Þ
Using second-order Pade’s approximation for time delay, the following polynomial in s is
obtained.

FðsÞ ¼ z1 s5 þ z2 s4 þ z3 s3 þ z4 s2 þ z5 s þ z6 ð44Þ
where,
z 1 ¼ y1 þ sd y5 ð45Þ

z 2 ¼ y2 þ y5 þ sd y6 ð46Þ

z 3 ¼ y3 þ y6 þ sd y7 ð47Þ

z 4 ¼ y4 þ y7 þ sd y8 ð48Þ

z 5 ¼ y8 þ sd y9 ð49Þ

z 6 ¼ y9 ð50Þ

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


344 R AO AND S REE

y1 ¼ x2 s sI ð51Þ

y2 ¼ x1 s sI  ax2 sI ð52Þ

y3 ¼ s sI  ax1 sI ð53Þ

y4 ¼ a sI ð54Þ

y5 ¼ x2 k kc sI sD ð55Þ

y6 ¼ x1 k kc sI sD þ x2 k sI þ x2 k kc sI ð56Þ

y7 ¼ k kc sI sD  x1 k s I  x1 k kc sI þ x2 k kc ð57Þ

y8 ¼ k sI þ k kc sI  x1 k kc ð58Þ

y9 ¼ k kc ð59Þ

x1 ¼ 0:5 L ð60Þ

x2 ¼ 0:0833 L2 ð61Þ

k ¼ k1 kp ð62Þ
Kharitonov’s polynomials for z
< zi <
i zþ
(i = 1, 2, 3, 4, 5, 6) are given below where
i z
i and

i are the lower bound and upper bound for zi, respectively.

z  4 þ 3 þ 2  
1 s þ z2 s þ z3 s þ z4 s þ z5 s þ z6 ¼ 0
5
ð63Þ

zþ þ 4  3  2 þ þ
1 s þ z2 s þ z3 s þ z4 s þ z5 s þ z6 ¼ 0
5
ð64Þ

zþ  4  3 þ 2 þ 
1 s þ z2 s þ z3 s þ z4 s þ z5 s þ z6 ¼ 0
5
ð65Þ

z þ 4 þ 3  2  þ
1 s þ z2 s þ z3 s þ z4 s þ z5 s þ z6 ¼ 0
5
ð66Þ
For fixed value of kp and τ a perturbation in time delay L, i.e., when (L – ΔL) ≤ L ≤ (L + ΔL),
is substituted in the above coefficients and Kharitonov’s polynomials are checked for stability
using the Routh–Hurwitz method [16]. Similar procedure is repeated to find stability region
for kp and τ.
INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014
Two-Degree-of-Freedom Controllers for Integrating Systems 345

Table 1. GM and PM.

Transfer Function PM and GM Two-degree-of-freedom IMC-PID DS-s TDF-s


e0:8s
s2 GM 1.49 1.35 1.55 –
PM 38.61 16.46 30.48 –
e5s
s2 GM 2.32 1.35 1.55 –
PM 68.49 16.44 30.49 –
2:2649 e0:3311s
s2 GM 1.49 1.57 1.43 –
PM 39.30 20.87 26.23 –
e0:2s
sðs1Þ GM 1.99 1.43 2.007 1.93
PM 56.87 15.38 36.07 20.41

Simulation Results
In this section, simulation results for various transfer function models are given for both servo
and regulatory problems. The performance comparison of the proposed TDF controller is
made with the PID controllers designed by DS-s [5], TDF-s [11] and IMC method (IMC-PID
controller) [14] in terms of integral performance criteria, Integral Square Error (ISE), Integral
Absolute Error (IAE) and Integral Time weighted Absolute Error (ITAE). The GM and
phase margin (PM) are evaluated for all the case studies and are given in Table 1. The GM
and PM indicate the robustness of the controller. The larger values of GM and PM indicate
that the controller is robust.

Case Study 1
0:8s
The transfer function model Gp ¼ e s2 [5] is considered. The PID controller parameters for
various methods are reported in Table 2.
The closed-loop response of the system for unit-step change in the set point at 0 s and for
unit-step change in load at time, t = 30 s for all the proposed controllers is shown in Fig. 2.
The performance of the controllers is given in terms of ISE, IAE and ITAE in Table 3. The
servo and regulatory performance of the proposed TDF is superior in comparison with all the
other controllers.

Table 2. PID parameters for Case Study 1.

PD inner-loop
parameters

Method kc τI τD θ γ1 γ2 k1 τd

Two-degree-of-freedom (GM = 1.5) 0.7656 1.1126 0.6134 – – – 0.1953 3.6


IMC-PID (λ = 0.9 L) 0.6205 4.0 1.6960 0.4 0.03 0.2085 – –
DS-s (λ = 1.5 L) 0.6807 4.4 1.7091 0.4 – 0.01069 – –

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


346 R AO AND S REE

2.5 IMC-PID
TDF
DS-s
2
Response

1.5

0.5

0
0 10 20 30 40 50 60
Time (s)

Fig. 2. Servo response and regulatory response of exp (−0.8s)/s2 (Case Study 1).

The stability regions of the model parameters for the proposed PID controller are calculated
by using Kharitonov’s theorem considering uncertainty in one parameter at a time and are
tabulated in Table 4. The DS-s controller gives superior region of stability compared to all the
other controllers for uncertainty in the model parameter kp. The proposed TDF controller gives
better region of stability than the IMC-PID controller for uncertainty in the model parameter kp.
The proposed TDF controller gives better region of stability compared to the IMC-PID
controller and the DS-s controller. Ms value is calculated and listed in Table 4. The proposed
controller is robust compared to the controllers designed by IMC method and Ds-s method.

Case Study 2
5s
The transfer function model Gp ¼ es2 is considered. The PID controller parameters for
various methods are reported in Table 5. The closed-loop response of the system for unit-step
change in the set point is shown in Fig. 3 and the response of the system for unit-step change
in load for all the controllers [5, 14] is shown in Fig. 4. The regulatory response of the TDF
controller is separately shown in Fig. 5.
The performance of the controllers is given in terms of ISE, IAE and ITAE in Table 3.
The servo and regulatory performance of the proposed TDF controller is superior to the
other controllers.
The stability regions of the model parameters for the proposed PID controllers are
calculated by using Kharitonov’s theorem considering uncertainty in one parameter at a time
and are tabulated in Table 4. Considering uncertainty in model parameter kp, the DS-s
controller gives a superior region of stability compared to the IMC-PID and the proposed
TDF controller. Considering uncertainty in model parameter L, the proposed TDF controller
gives superior region of stability compared to the IMC-PID controller and the DS-s

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


Two-Degree-of-Freedom Controllers for Integrating Systems 347

Table 3. Performance comparison.

Servo problem Regulatory problem

Transfer
function Method ISE IAE ITAE ISE IAE ITAE
0:8s
e
s2 Two-degree-of- 1.96 3.10 8.82 0.24 1.45 8.96
freedom
IMC-PID 2.46 3.71 10.41 8.14 6.45 27.10
DS-s 1.86 3.28 10.36 6.23 6.46 30.96
e5s
s2 Two-degree-of- 12.91 19.65 346.62 3.64 13.58 503.39
freedom
IMC-PID 15.36 23.17 406.49 7.74 × 104 1.57 × 104 4.13 × 104
DS-s 11.58 20.49 404.59 5.96 × 104 1.58 × 103 4.73 × 104
2:2649 e0:3311 s
s2 Two-degree-of- 0.80 1.27 1.50 0.113 0.58 1.31
freedom
IMC-PID 1.18 1.89 2.87 1.60 2.15 4.70
DS-s 0.74 1.27 1.46 0.25 0.78 1.39
e0:2s
sðs1Þ Two-degree-of- 0.53 0.89 0.84 0.099 0.54 1.12
freedom
IMC-PID 0.74 1.18 1.13 0.044 0.281 0.387
DS-s 0.69 1.33 1.96 0.092 0.55 1.15
TDF-s 0.80 1.60 2.89 0.22 0.94 2.39

Two-degree-of-freedom, proposed method; IMC-PID, Internal Model Control Method (Nageswara Rao and Padma
Sree [14]); DS-s, Direct Synthesis method (Seshagiri Rao, V.S.R. Rao and Chidambaram [5]); TDF-s, Two-degree-of-
freedom controller (Seshagiri Rao and Chidambaram [11]).

Table 4. Comparison of stability regions and Ms values.

Stability region

Transfer function Method kp L τ Ms value


e0:8s
s2 Two-degree-of-freedom ±33.45% ±28.47% – 3.27
IMC-PID ±27.4% ±16.94% – 4.3171
DS-s ±42.0% ±23.5% – 2.8985
e5s
s2 Two-degree-of-freedom ±29.5% ±35% – 1.7621
IMC-PID ±27.4% ±16.8% – 4.3278
DS-s ±42% ±23.67% – 2.8896
2:2649 e0:3311 s
s2 Two-degree-of-freedom ±30% ±28.8% – 3.26
IMC-PID ±31.92% ±27% – 3.1857
DS-s ±41.2% ±19.5% – 3.4297
e0:2s
sðs1Þ Two-degree-of-freedom ±23.95% ±29.5% ±24.7% 2.157
IMC-PID ±25.426% ±23.18% ±15.368% 4.0308
DS-s ±39.35% ±40.9% ±37.13% 2.0471
TDF-s ±39.3% ±47.95% ±41.7% 1.8415

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


348 R AO AND S REE

Table 5. PID parameters for Case Study 2.

PD inner-
loop
parameters

Method kc τI τD θ γ1 γ2 k1 τd

Two-degree-of-freedom (GM=2.5) 0.3322 4.5556 5.2474 – – – 0.0063 20


IMC-PID (λ = 0.9 L) 0.0159 25 10.6 2.5 1.1725 1.3028 – –
DS-s (λ = 1.5 L) 0.0174 27.5 10.6818 2.5 – 0.06683 – –

controller. Ms value is calculated and listed in Table 4. The proposed controller is robust
compared to the controllers designed by IMC method and Ds-s method.

Case Study 3
0:3311s
The transfer function model Gp ¼ 2:2649se2 [17] is considered. The PID controller
parameters for various methods are reported in Table 6. The closed-loop response of the
system for unit-step change in the set point at time, t = 0 s and for unit-step change in load at
time, t = 20 s for all the controllers [5, 14] is shown in Fig. 6.
The performance of the controllers is given in terms of ISE, IAE and ITAE in Table 3.
The servo performance of the proposed TDF controller and the DS-s [5] controller is almost
similar and better than the IMC-PID controller. For the regulatory problem, the proposed
TDF controller gives superior performance when compared to the IMC-PID controller and
the DS-s controller.

1.8

1.6

1.4

1.2
Response

0.8

0.6
IMC-PID
0.4 TDF
DS-s
0.2

0
0 20 40 60 80 100 120 140
Time (s)

Fig. 3. The servo response of exp (−5s)/s2 (Case Study 2).

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


Two-Degree-of-Freedom Controllers for Integrating Systems 349

80

70

60
IMC-PID
50 TDF
DS-s
Response

40

30

20

10

-10
0 20 40 60 80 100 120 140
Time (s)

Fig. 4. The regulatory response of exp (−5s)/s2 (Case Study 2).

The stability regions of the model parameters for the proposed PID controllers are
calculated by using Kharitonov’s theorem considering uncertainty in one parameter at a time
and are tabulated in Table 4. The DS-s controller gives superior region of stability compared
to all the other controllers for uncertainty in model parameter kp. The IMC-PID controller
and the TDF controller give almost similar region of stability for uncertainty in model
parameter kp. The proposed TDF controller gives better region of stability for uncertainty in
model parameter L. GM and PMs in Table 1 also indicate the same stability criteria as

0.45

0.4

0.35 TDF REG


0.3
Response

0.25

0.2

0.15

0.1

0.05

0
0 20 40 60 80 100 120 140
Time (s)

Fig. 5. The regulatory response of exp (−5s)/s2 for proposed TDF controller (Case Study 2).

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


350 R AO AND S REE

Table 6. PID parameters for Case Study 3.

PD parameters

Method kc τI τD θ γ1 γ2 k1 τd

Two-degree-of-freedom 0.6122 0.3549 0.3205 – – – 0.6099 1.3244


(GM = 1.5)
IMC-PID (λ = 1.2 L) 1.0028 2.1522 0.8991 0.1656 0.0104 0.1315 – –
DS-s (λ = 1.3 L) 2.0846 1.6224 0.6399 0.1656 – 0.00384 – –

explained above. Ms value is calculated and listed in Table 4. The proposed controller is
robust compared to the controllers designed by IMC method and Ds-s method.

Case Study 4
e0:2s
The transfer function model for unstable FOPTD system with an integrator Gp ¼ sðs1Þ [5] is
considered. The PID controller parameters for various methods are reported in Table 7.
The closed-loop response of the system for unit-step change in the set point at time, t = 0 s
and for unit-step change in load at time, t = 10 s for all the controllers [5, 11, 14] is shown in
Fig. 7.
The performance of the controllers is given in terms of ISE, IAE and ITAE in Table 3.
The proposed TDF controller gives superior performance compared to the reported DS-s
controller [5], IMC-PID [14] and the TDF-s controller [11] for servo problem.
The IMC-PID controller gives superior performance compared to all the other
controllers, the DS-s controller and the TDF-s controller for regulatory problem. The

2.5

2 IMC-PID
TDF
DS-s
Response

1.5

0.5

0
0 5 10 15 20 25 30 35 40
Time (s)
2:2649e0:3311s
Fig. 6. The servo and regulatory response of s2 (Case Study 3).

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


Two-Degree-of-Freedom Controllers for Integrating Systems 351

Table 7. PID parameters for Case Study 4.

PD inner-loop
parameters

Methods kc τI τD θ γ1 γ2 k1 τd

Two-degree-of-freedom (GM = 2) 0.3180 0.1725 0.2691 – – – 2.5165 1.3524


IMC-PID (λ = 1.25 L) 4.807 1.35 0.771 0.1 0.0035 0.073 – –
DS-s 3.6615 2 1.003 0.1 – 0.00395 – –
TDF-s 2.584 2.44 1.274 0.1 – 0.0044 – –

performance of the proposed TDF controller and the DS-s controller for regulatory problem
is almost similar and better than the TDF-s controller.
The stability regions of the model parameters for the proposed PID controllers are
calculated by using Kharitonov’s theorem considering uncertainty in one parameter at a time
and are tabulated in Table 4. The controllers designed by DS-s and TDF-s methods give
superior region of stability compared to all the other controllers for uncertainty in the model
parameter kp. The IMC-PID controller gives better region of stability compared to the TDF-s
for uncertainty in parameter kp.
Considering uncertainty in the model parameter L, controller designed by TDF-s method
gives superior region of stability when compared to all the other controllers. The reported
DS-s controller gives better region of stability than the IMC-PID controller and the proposed
TDF controller. The proposed TDF controller gives better region of stability than the IMC-
PID controller.

1.6

1.4

1.2

1
Response

0.8

0.6
IMC-PID
0.4 TDF
DS-s
0.2 TDF-s

0
0 5 10 15 20
Time (s)
e0:2s
Fig. 7. The servo and regulatory response of sðs1Þ (Case Study 4).

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


352 R AO AND S REE

Considering uncertainty in the model parameter τ, the TDF-s controller and DS-s
controller gives superior region of stability when compared to all the other controllers. The
proposed TDF controller gives better region of stability when compared to the IMC-PID
controller. GM and PMs in Table 1 also indicate the same stability criteria as explained
above. Ms value is calculated and listed in Table 4. The proposed controller is robust
compared to the controllers designed by IMC method, TDF-s and Ds-s method.

Conclusions
TDF controller for pure double-integrating systems with time delay and unstable FOPTD
systems with an integrator is proposed. The performance of the proposed controller is better
than controllers designed by the recently reported methods. The stability region for various
model parameters considering uncertainty in one parameter at a time is obtained using
Kharitonov’s theorem. The stability region for the proposed controller is compared with that
of the controllers designed by literature reported methods. The proposed controller is robust
for model uncertainties. One can also calculate Ms value to check the robustness of the
controller. For stable systems, Ms value range given by Seborg et al. [18] is 1.4–2 for robust
control. However, for unstable systems, Ms value can go up to 3–3.5 [19]. Simulation results
of various transfer function models of DITD and UFOPTDI systems and a transfer function
model of vertical take-off airplane [17] are given to show the effectiveness of the proposed
controllers.

Nomenclature
a Defined by Equation (39)
Gc Controller transfer function
Gp Process transfer function
Gp1 Inner closed-loop transfer function defined by Equations (1) and (21)
k1 PD controller gain (inner loop) as in Equations (1) and (21)
kc Controller gain
kc,max Ultimate controller gain
kp Process gain
L Time delay of the process
r1, r2, r3 Roots of polynomial in denominator of inner closed-loop transfer function (Equation 6)
q1, q2, q3 Roots of polynomial in denominator of inner closed-loop transfer function(Equation 25)
x1 Defined by Equation (60)
x2 Defined by Equation (61)
y Output of the process
yr Set point
y1−y9 Defined by Equation (51) to Equation (59)
z1−z6 Defined by Equation (45) to Equation (50)

Greek Letters
τ Process time constant
τI Integral time of PID controller

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


Two-Degree-of-Freedom Controllers for Integrating Systems 353

τD Derivative time of PID controller


τf Denominator filter time constant of a PID controller
τd Derivative time constant of the inner-loop PD controller
λ Adjustable tuning parameter
γ1, γ2 Denominator filter time constants of a PID controller
θ Numerator time constant of a PID filter

References
[1] Chen, C. and Tan, H. “Steering Control of High Speed Vehicles: Dynamic Look Ahead and Yaw
Rate Feedback”, Proceedings of the 3rd IEEE Conference on Decision and Control, 1989, Tampa,
FL (1998).
[2] Hughes, P.C. Spacecraft Attitude Dynamics, Wiley, New York, NY (1986).
[3] Gustafsson,T. “On the Design and Implementation of a Rotary Crane Controller”, Eurorpean J.
Control., 2, pp. 166–175 (1996).
[4] Liu, T., He, X., Gu, G.Y. and Zhang, W.D. “Analytical Decoupling Control Design for Dynamic
Plants with Time Delay and Double Integrators”, IEE Proc. Control Theory Appl., 151(6), pp.
745–753 (2004).
[5] Seshagiri Rao, A., Rao, V.S.R. and Chidambaram, M. “Direct Synthesis based Controller Design
for Integrating Processes with Time Delay”, J. Franklin Inst., 346, pp. 38–56 (2009).
[6] Bodizs, L., Titica, M., Faria, N., Srinivasan, B., Dochain, D. and Bonvin, D. “Oxygen Control for
an Industrial Pilot Scale Fed Batch Filamentous Fungal Fermentation”, J. Process Control, 17, pp.
595–606 (2007).
[7] Rao, V.G. and Berstein, D.S. “Naive Control of the Double Integrator”, IEEE Control Syst. Mag.,
21(5), pp. 86–97 (2001).
[8] Skogestad, S. “Simple Analytic Rules for Model Reduction and PID Controller Tuning”, J.
Process Control, 13, pp. 291–309 (2003).
[9] Dostal, P., Bobal, V. and Sysel, M. “Design of Controllers for Integrating and Unstable Time
Delay Systems Using Polynomial Method”, Proc. Am. Conf. (ACC), pp. 2773–2778 (2002).
[10] Liu, T., Zhang, W. and Gu, D. “Analytical Design of Two-degree-of-freedom Control Scheme for
Open-loop Unstable Processes with Time Delay”, J. Process Control, 15, pp. 559–572 (2005).
[11] Seshagiri Rao, A. and Chidambaram, M. “Enhanced Two-degrees-of-freedom Control Strategy
for Second-order Unstable Processes with Time Delay”, Ind. Eng. Chem. Res., 45, pp. 3604–
3614 (2006).
[12] Shamsuzzoha, M. and Lee, M. “Enhanced Performance for Two-degree-of-freedom Control
Scheme for Second Order Unstable Processes with Time Delay, International Conference on
Control”, Autom. Syst., October 17–20, 2007 in COEX, Seoul, Korea (2007).
[13] Panda, R. “Synthesis of PID Controller for Unstable and Integrating Processes”, Chem. Eng. Sci.,
64, pp. 2807–2816 (2009).
[14] Nageswara Rao, C.V. and Padma Sree, R. “IMC Based Controller Design for Integrating Systems
with Time Delay”, Indian Chem. Eng., 52(3), pp. 194–218 (2010).
[15] Kharitonov, V.L. “Asymptotic Stability of an Equilibrium Position of a Family of Systems of
Linear Differential Equations”, Differential ’nye Uraveniya, 14, pp. 1483–1485 (1978).

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


354 R AO AND S REE

[16] Sinha, N.K. Control Systems, New Age International (P) Ltd., Publishers, New Delhi, pp. 142–
144 (2002).
[17] Filatov, N.M., Keuchel, U. and Unbehauen, H. “Dual Control of Unstable Mechanical Plant”,
IEEE Control Syst., 16(4), pp. 31–37 (1996).
[18] Seborg, D.E., Edgar, T.F. and Mellichamp, D.A. Process Dynamics and Control, 2nd edn, John
Wiley & Sons, Singapore (1989).
[19] Shamsuzzoha, M. and Lee, M. “Analytical Design of Enhanced PID Filter Controller for
Integrating and First Order Unstable Processes with Time Delay”, Chem. Eng. Sci., 63, pp. 2717–
2731 (2008).

INDIAN CHEMICAL ENGINEER Vol. 56 No. 4 December 2014


Copyright of Indian Chemical Engineer is the property of Taylor & Francis Ltd and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.

S-ar putea să vă placă și