Professional Documents
Culture Documents
Abstract- The PID control system design has been proposed based Chien et al. [9] method for the FOPDT system with an
on the IMC principle for integrating, first order and second order integrator and with/without a zero.
systems with an integrator, and with/without zero. The proposed Due to the simplicity and better performance of internal model
method of PID controller design is based on the disturbance control (IMC) based tuning rule, the analytically derived IMC-
rejection, which causes the overshoot in the setpoint response. The
PID tuning [12-14] methods attracted the attention of industrial
two-degree-of-freedom (2DOF) control structure has been
suggested to eliminate the overshoot in setpoint response. The users recently. The IMC-PID tuning rule has only one user-
simulation results show the superiority of the proposed tuning defined tuning parameter, which is directly related to the
rule over other existing methods, when the controller is tuned to closed-loop time constant. The IMC-PID controller provides
have the same robustness level by evaluating the peak of the good set-point tracking but sluggish disturbance response
maximum sensitivity ( M s ) . The multiplicative uncertainty bound especially for the process with a small time-delay/time-
has been proposed based on norm-bound uncertainty principle constant ratio. However, for many process control applications,
disturbance rejection is much more important than set-point
( )
for the selection of closed-loop time constant λ .
tracking. Therefore, controller design that emphasizes
I. INTRODUCTION disturbance rejection rather than set-point tracking is an
important design problem that has received renewed interest
In the process control, more than 95% of the control loops are recently.
of the proportional-integral-derivative (PID) type. The main However, the methods of PID controllers designing for the
reason is its relatively simple structure, which can be easily FOPDT system with an integrator and with/without a zero are
understood and implemented in practice. Finding design not discussed extensively. Form the literature it is clear that
methods that lead to the optimal operation of PID controllers is there is a need of unified framework for the IMC-PID tuning
therefore of significant interest. The integrating processes, first rule based on disturbance rejection. Therefore, the present
order and second order systems with an integrator and work is directed to design the PID controllers for such systems
with/without zero are frequently encountered in the process by extending the IMC method for the integrating and first and
industries. For the first order systems with an integrator and second order system with an integrator, and with/ without zero
with/without zero, if the zero is a positive, the system exhibits for the disturbance rejection. The concept of 2DOF control
inverse response and if the zero is negative, then the system structure is used to cope with setpoint performance. The
shows large overshoot in the response. performance of the proposed tuning rule has been compared
A recent trends show that the tuning of the controllers for a with other tuning methods, when the controller is tuned to the
time-delay integrating process with/without zero has been an same robustness level by evaluating the peak of the maximum
active area of research in the literature [1-11]. Integrating sensitivity ( M s ) . The multiplicative uncertainty bound has
processes are frequently encountered in the process industries
and is very convenient for process identification because it been proposed based on norm-bound uncertainty principle for
contains also two parameters. The design of the integrating the selection of λ .
process is suggested in [1-11]. II. CONTROLLER DESIGN ALGORITHM
Zhang et al. [6] have proposed a PID tuning method for first
order plus dead time (FOPDT) system with an integrator. An Figures (1-a&b) show the block diagram of IMC control and
optimization method shows the performance of their method is equivalent classical feedback control structures, where GP is
better than that of Poulin and Pomerleau [5]. Wang and Cai [7] the process, G P is the process model, and q is the IMC
have proposed simple tuning formulae for PID controller for
integrating and unstable processes with time delay. controller. The controlled variables are related as
Chien et al. [9] have proposed a design method for the PID G q 1 − G q (1)
C= P
R+ P
GD d
controller for the second order plus dead time (SOPDT) (
1 + q GP − G P ) ( )
1 + q GP − G P
systems with a positive zero/negative zero respectively. Their For the nominal case (i.e., GP = G P ), the set-point and
method is based on direct synthesis for getting the tuning
parameters. Recently, Anil and Padma Sree [10] extended the disturbance responses are simplified as
where PM is the portion of the model inverted by the controller; K (θ − β + 2λξ ) (θ − β + 2λξ )
PA is the portion of the model not inverted by the controller (it θ 3
6 −
βθ 2
2
2
+ βθ )
τI (θ − β + 2λ ξ )
and/or right half plane zeros); PA ( 0 ) = 1 .
The value of β is calculated by solving
Step 2: The idealized IMC controller is the inverse of the ( )
1−( βs +1) e−θs λ2s2 +2λξs +1 s=−1τ =0
invertible portion of the process model. to cancel the slow pole of the process and β value is given as
q = PM −1 (5) β = τ 1− ( λ 2 − 2λξτ +τ 2 ) e−θ τ τ 2
.
To make the IMC controller proper, it is mandatory to add the
filter. Thus, the IMC controller is designed as B. Delay Integrating Process (DIP)
(6) The commonly used delay integrating process model for
= PM-1 f
q = qf
chemical industries is given below
The ideal feedback controller equivalent to the IMC controller
Ke−θ s (14)
can be expressed in terms of the internal model, G P , and the G p = GD =
s
IMC controller, q : The DIP process can be modeled as the FOPDT by
q (7) approximating as:
Gc =
1 − G P q G = G =
K e −θ s
=
K e −θ s
=
ψ K e −θ s (15)
p D
Since the resulting controller has not a standard PID controller s s + 1 /ψ ψ s +1
form, the remaining issue is to design the PID controller that where ψ is an arbitrary constant with a sufficiently large value
resemble the equivalent feedback controller most closely. Lee i.e., ψ >> 1 . The proposed filter is f = ( β s + 1) ( λ 2 s 2 + 2λξ s + 1)
et al. [13] proposed an efficient method for converting the ideal
for the DIP model.
feedback controller Gc to a standard PID controller. Since Gc Therefore, the resulting IMC controller
has an integral term, it can be expressed become, q = (ψ s +1)( βs +1) Kψ ( λ2s2 + 2λξs +1) and the ideal
f (s) (8)
G =
c feedback controller equivalent to the IMC controller is
s
Expanding Gc in Maclaurin series in s gives GC =
(ψ s +1)( β s +1) (16)
( )
Kψ λ2 s2 + 2λξ s +1 − e−θ s ( β s +1)
2257
Analogous to the FOPDT model, from the Eq. (11a~c) the PID ( K + ∆K ) e −(θ +∆θ )s (22)
GP ( s ) =
parameters can be obtained as: (τ s + 1)( ∆τ s + 1)
kC =
τI
τ I = (ψ + β ) −
(λ 2
2
−θ
+ βθ
2
) ( K + ∆K ) e−(θ +∆θ )s − Ke−θ s ( K + ∆K ) e −∆θ s − K (23)
ψ K (θ − β + 2λξ ) (θ − β + 2λξ ) ∆m ( s) =
(τ s + 1)( ∆τ s + 1) τ s + 1 = ( ∆τ s + 1)
θ 3 βθ 2 Ke−θ s K
6 − 2 τ s +1
τD =
(ψ β ) −
(θ − β + 2 λξ ) −
(
λ 2 −θ 2 + βθ
2
) (
λ − 2λξψ +ψ e
β =ψ 1−
2 2
) −θ ψ
Therefore,
τI (θ − β + 2 λξ ) ψ2 ∆K
− ∆θ s
1 + e (24)
Based on the IMC design principle, PID tuning rules for ∆ ( s ) = K −1
( ∆ τ s + 1) m
Putting s = jω in above equation we obtained as Example 2: Stable FOPDT System with an Integrator
(21) Consider the integrating process transfer function used by [5-
2
{ (
τ 1 − λ − 2λξτ + τ e
2 2 −θ τ
τ2) }ω
2
2
+ 1 8,10]
1
<
(1 − λ ω ) + ( 2λξω )
2 2 2 ∆m ( s ) 1e −0.2 s (27)
GP =
2
∞
∞ s ( s + 1)
Suppose we have uncertainty in all the three process
parameters, i.e., θ , τ , K The control performance of the proposed method is compared
2258
TABLE 1 PID controller tuning rules
G = GD =
Ke − θ s kC =
τI
τI = (ψ + τ + β ) −
( −θ 2
2
+ θβ1 − β 2 + 2λ 2 + 4λ 2ξ 2 )
s (τ s + 1) ψ K ( 4λξ + θ − β1 ) 1
( 4λξ + θ − β1 )
θ 3 β1θ 2 + θβ + 4λ 3ξ
6− 2 2
τD =
( β2 + (τ +ψ ) β1 +ψτ ) − ( 4λξ + θ − β1 ) −
( −θ 2
2
+ θβ1 − β 2 + 2λ 2 + 4λ 2ξ 2 )
τI ( 4λξ + θ − β1 )
2 2
λ2 2λξ λ 2 2λξ −θ ψ
τ2 − + 1 e−θ τ −ψ 2 2 − + 1 e + ψ 2 −τ 2 ( ) λ 2 2λξ
2
τ τ ψ ψ
2
β1 = β 2 = τ 2 − + 1 e−θ τ −1 + β1τ
(ψ − τ ) τ
2
τ
G = GD =
Ke −θ s
kC =
τI
τ I = (τ 1 + τ 2 + β1 ) −
( −θ
2
2
+ θβ1 − β 2 + 2 λ 2 + 4λ 2ξ 2 )
(τ 1 s + 1)(τ 2 s + 1) K ( 4λξ + θ − β1 ) ( 4λξ + θ − β1 )
θ 3 β1θ 2
6− + θβ 2 + 4λ 3ξ
2
τD =
( β2 + (τ 1 + τ 2 ) β1 + τ 1τ 2 ) − ( 4λξ + θ − β1 ) −
( −θ 2
2 + θβ1 − β 2 + 2λ + 4λ ξ
2 2 2
)
τI ( 4λξ + θ − β1 )
2 2
λ2 2λξ λ 2 2λξ −θ τ 2
τ 12 − + 1 e −θ τ1 − τ 2 2 2 − + 1 e + τ 22 − τ12 ( ) λ 2 2
τ τ1 τ2 τ2
2
2λξ
β1 = β 2 = τ 12 + 1 e −θ τ 1 − 1 + β1τ1
1
−
(τ 2 −τ1 ) τ 1
2
τ1
G = GD =
K (τ a s + 1) e −θ s kC =
τI
τ I = (ψ + τ + β1 ) −
( − θ 2 + θβ1 − β 2 + 2λ 2 + 4 λ 2ξ 2
2
)
s (τ s + 1 ) ψ K ( 4λξ + θ − β1 ) ( 4λξ + θ − β1 )
θ 3 β1θ 2 + θβ + 4λ 3ξ
6− 2 2
( β 2 + (τ +ψ ) β1 +ψτ ) − ( 4λξ + θ − β1 ) ( −θ 2
2
+ θβ1 − β2 + 2λ 2 + 4 λ 2ξ 2 ) Lag filter , t F =
1
τD =
τI
−
( 4λξ + θ − β1 ) (τ a s + 1)
2 2
λ2 2λξ λ 2 2λξ −θ ψ
τ2 − + 1 e −θ τ −ψ 2 2 − + 1 e + ψ 2 −τ 2 ( ) λ 2 2λξ
2
τ τ ψ ψ
2
β1 = β 2 = τ 2 − + 1 e−θ τ − 1 + β1τ
(ψ − τ ) τ
2
τ
2259
the problem of controlling processes that exhibit both inverse TABLE 3 PID controllers setting for Example 2
response and dead time.
Example 4: Biochemical Reactor Tuning Kc τI τD set- disturbance
methods point
A biochemical reactor with several components has the
IAE IAE
following linearized model Bequette [15], which relates the
biomass concentration to the dilution rate: Proposed 5.45 1.60 0.46 1.12 0.29
1 . 7 5 ( − 3 s + 1 ) ( − 5 s + 1 ) e − 1 .2 5 s (29) λ = 0.353
G =
(1 0 s + 1 )( 4 s + 1 )
p 2
Anil & P. Sree 5.48 1.88 0.47 1.3 0.35
For high-order processes, the IMC design methods do not yield Wang & Cai 3.09 1.18 0.86 2.23 0.64
PID controllers directly. Thus the model order must be reduced,
or the resulting controller must be approximated by a PID
controller. Skogestad [14] has proposed a simple method of
approximating the high-order models with low-order models. It TABLE 4 PID controllers setting for Example 3
is clear that an ‘‘inverse response time constant’’ T0inv
(negative numerator time constant) may be approximated as a Tuning Kc τI τD tF set- disturbance
methods point
time delay ( −T0inv +1 ) ≈ e−T s .This is reasonable since an inverse
0
inv
IAE IAE
response has a deteriorating effect on the control, similar to
that of a time delay. For the model reduction of the above Proposed 33.8 62.10 14.84 300 27.91 14.3
λ = 7.47
biochemical reactor, we have used “Half rule”. The half rule is
discussed in detail Skogestad [14]. The above model can be Anil & 39.2 42.12 10.50 300 31.29 15.34
reduced in the SOPDT form using Half rule as P. Sree
Wang & 1.6 26.0 2.0 0.2 14.87 16.63
G p = 1.75 e −11.25 s (1 2 s + 1 )( 4 s + 1 ) . Cluett
The PID controller has been designed by proposed method for
Ms = 2.0 . Figure 5 shows the closed-loop output response for TABLE 5 PID controllers setting for Example 4
biochemical reactor model with a unit-step setpoint change
occurring at t = 0 , and a unit-step disturbance occurring at t = 100 . Tuning Kc τI τD set- disturbance
The closed loop response of controller shows the smooth and methods point
fast setpoint and disturbance rejection. b = 0.6 has been used in IAE IAE
the 2DOF controller for the setpoint response. The robustness
of the controller is evaluated by perturbing the 100% Proposed 0.477 17.094 3.536 26.67 38.69
λ = 6.214
uncertainty in dead time (θ = 2.5 ) . The perturbed response is
also shown in Fig. 5 and performance index is listed in Table 5. Proposed 0.477 17.094 3.536 29.19 41.14
It is clear that the proposed controller design method of the (100% θ
uncertainty)
biochemical reactor is simple and robust.
2260
VI. CONCLUSIONS
The IMC filter has been modified for several representative
processes to improve disturbance rejection performance of the
PID controller. Based on the proposed filter, tuning rules for
the PID controller was derived by using the generalized IMC-
PID method by Lee et al. [13]. Undershoot in the disturbance
rejection can be eliminated by the overdamped IMC filter for
the system with an integrator and a negative zero. The
processes which have inverse response can be treated by
reducing them into FOPDT/SOPDT model. The model
reduction techniques can be utilized to design the PID
controller for the inverse response process maintaining the
performance and robustness level. The simulation results
demonstrated superiority of the proposed method.
ACKNOWLEDGEMENT
The authors wish to thank and express their appreciation to
2006 Energy Resource and Technology Project and second-
phase of BK (Brain Korea) 21 program for providing the
Fig. 3. Simulation results for Example 2 financial support.
REFERENCE
[1] I. L. Chien, and P. S. Fruehauf, “Consider IMC tuning to improve
performance,” Chem. Eng. Prog. Vol. 10, 33, 1990.
[2] W. L. Luyben, “Design of proportional integral and derivative
controllers for integrating dead-time processes,” Ind. Eng. Chem. Res.,
vol. 35, 3480, 1996.
[3] A. Visioli, “Optimal tuning of PID controllers for integral and unstable
processes,” IEE Proc. Control Theory, 148, 180, Appl 2001.
[4] M. Chidambaram, and R. Padma Sree, “A simple method of tuning PID
controller for integrator/dead-time processes, Computer and Chemical
Engineering, 27, 211-215, 2003.
[5] E. Poulin, and A. Pomerlaeu, “A PID tuning for integrating and unstable
processes, IEEE Proceedings – CTA, vol.143, 429-435, 1996.
[6] W. Zhang, X. Xu, and Y. Sun, “Quantitative performance design for
integrating processes with time delay, Automatica, vol. 35, pp. 719-723,
1999.
[7] Y. G. Wang, and W. J. Cai, “Advanced proportional integral derivative
tuning for integrating and unstable processes with gain and phase
margin specifications, Ind. Eng. Chem. Res., vol. 41, pp. 2910-2914,
2002.
[8] W. Tan, J. Liu, P. K. S. Tam, PID tuning based on loop shaping H∞
control. IEEE Proc.-Control Theory Appl. Vol. 145, 485, 1998.
Fig. 4. Simulation results for Example 3
[9] I. L. Chien, Y. C. Chung, B. S. Chen, and C. Y. Chuang, Simple PID
controller tuning method for processes with inverse response plus dead
time or large overshoot response plus dead time, Ind. Eng. Chem. Res.,
2.5
vol. 42, pp. 4461-4477, 2003.
[10] C. Anil, and R. Padma Sree, “Design of PID controllers for FOPTD
2 systems with an integrator and with/without a zero, Indian chem. Engr.,
Section A. vol. 47, 4, pp. 235-242, 2005.
[11] L. Wang, and W. R. Cluett, “Tuning PID controllers for integrating
1.5 processes” IEEE Proceedings- CTA., vol. 144, 385, 1997.
Process R esponse
2261