Izmir Institute of Technology

Dept. of ChE.

Process Dynamics
and Control-ChE 402

Chapter10 –
PID controller design
and tuning

Asst. Prof. Dr. Can Kızılkaya

Learning objectives for Chapter10

• List the performance characteristics of FBC systems

• Explain how to design controllers using Direct Synthesis

(DS) and Internal Model Control (IMC) techniques

• Discuss the concepts of desired closed-loop transfer

function (Y/Ysp)d, the desired closed-loop time constant
(tc ) and their effect on closed loop performance

• Outline the tuning of controllers using Integral-of-Time

weighted-Absolute-Error (ITAE) technique

• Compare robustness vs performance of controllers based

on different design and tuning methods
Quote of the week

• It's not what happens to you, but how you react to it

that matters.

• Epictetus (50-135)
• Stoic philosopher
 Performance criteria of FBC systems

• Stable
• Good disturbance rejection (Fast, effects of disturbances are
minimized)
• Good set-point tracking (Fast, smooth responses to set-point
changes)
• Eliminates offset
• Avoids excessive controller action
• Robust (Insensitive to changes in process conditions and to
inaccuracies in the process model, i.e. suitable over a wide
range of operating conditions )

• Impossible to satisfy all, so controller should be optimized

to achieve desired criteria  Controller Design and Tuning
FBC Design and Tuning

• FBC Design and Tuning is the selection of the

optimum controller (PI or PID) and the adjustment of
the controller parameters (Kc, tI and tD)

• FBC Design and Tuning requires a tradeoff between

system performance and robustness.

• There is also a tradeoff between set-point tracking

and disturbance rejection (Controllers that give the
best disturbance rejection give the worst
performance for set-point tracking)
FBC Design and Tuning

• FBC Design and Tuning requires a tradeoff between

system performance and robustness.
• Performance: Rapid and smooth responses to
disturbance changes, with an acceptable degree of
oscillation
• Large values of Kc, small values of tI , Aggressive controller

• Robustness: Satisfactory performance for a wide range

of process conditions and for a reasonable degree of
model inaccuracy
• Small values of Kc, large values of tI, Conservative controller
Output response of FBC systems for a step change in
disturbance variable, for various controller settings
Output response of FBC systems for a step change in
disturbance variable, for various controller settings

• As Kc increases or tI decreases, the response to the step

disturbance becomes more aggresive (faster response but
more oscillations).
• Controller 1 produces an unstable response (growing
oscillations)
• Controller 5 provides the best performance (low degree of
oscillations and fast response)
Controller Design and Tuning Techniques

• Direct Synthesis (DC) Method

• Internal Model Control (IMC)

• Controller Tuning Relations (ITAE)

Direct Synthesis (DS) Method

• This method is based on:

• Process model

• Desired closed loop transfer function for set-point

change

• Desired time constant for the closed loop transfer

function

• DS method produces PI or PID controllers based on

process model
Standard block diagram of FBC
Direct Synthesis (DS) Method

Y GC G
 ( G includes Gp, Gm, Gv)
Ysp 1  GC G

1. Specify closed-loop response (transfer function)

Y 
 
Y 
 sp  d
 (= GPGMGV)
2. Need process model, G

3. Solve for Gc,

 Y  
   
1   Ysp  d 
GC    
G  
1  Y  
  Ysp  
 d 
 Direct Synthesis (DS) Method - Specify
desired closed loop transfer function
Y  e s
   (12  6)
 Ysp d t c s  1

(first – order response)

t c  speed of response, θ = process time delay in G 

tc = Desired time constant of the closed loop transfer function

 Y  
   
1   Ysp  d 
GC    
G  
1  Y  
  Ysp  
 d  e-s - 1- s.
Taylor’s series approx.
Direct Synthesis (DS) Method - Derivations of
PI controllers for FOPTD process

Consider the standard first-order-plus-time-delay model,

remembering G includes Gp, Gm, Gv
θs
Ke
G  s   (12-10)
τs  1
Specify closed-loop response as FOPTD (12-6),
but approximate e-s - 1- s.

Substituting and rearranging gives a PI controller,

Gc  Kc 1  1/ τ I s  , with the following controller settings:
1 τ
Kc  , τI  τ (12-11)
K θ  τc
Properties of PI controller designed by DS
method
• tI = t Slow processes have large values of t, tI should
also be large for satisfactory control)
• As tc decreases, Kc increases. (A faster output
response [because of decreasing tc ] requires more
control action, larger values of Kc, i.e. aggressive
controller)
• Kc is inversely proportional to K (gain of the process
model)

Gc  Kc 1  1/ τ I s  ,
1 τ
Kc  , τI  τ (12-11)
K θ  τc
Direct Synthesis (DS) Method - Derivations of
PID controllers for SOPTD process
Consider a second-order-plus-time-delay model,
 θs
Ke
G  s   (12-12)
 τ1s  1 τ2 s  1
Use of FOPTD closed-loop response (12-6) and time delay
approximation gives a PID controller in parallel form,

 1 
Gc  K c  1   τDs  (12-13)
 τI s 
where
1 τ1  τ 2 τ1τ 2
Kc  , τ I  τ1  τ 2 , τD  (12-14)
K τc   τ1  τ 2
Example: Using DS method to design PID controllers

Use DS method to calculate PID controller settings for the process

(Combined TF for process, sensor and valve):
2e  s
G
10s  1 5s  1
Consider three values of the desired closed-loop time constant:
τc = 1, 3, and 10. Evaluate the controllers for unit step changes in
both the set point and the disturbance, assuming that Gd = G.
Perform the evaluation for two cases:

a. The process model is perfect ( G = G).

b. The model gain is K = 0.9, instead of the actual value, K = 2.
This model error could cause a robustness problem in the
controller for K = 2. s
0.9 e
G 
10s  1 5s  1
Solutions to the example (Correct model gain, K = 2)

Decreasing tc
(Increasing Kc)
Decreasing tc
(Increasing Kc)

Regulator problem
Servo problem Unit step change
Unit step change in D
in ysp (D’=1)
(y’sp=1)
Solutions to the example (Incorrect model gain, K=0.9)
Conclusions from the example

 1 
Gc  Kc 1   τDs  (12-13)
 τI s 
1 τ1  τ 2 τ1τ 2
Kc  , τ I  τ1  τ 2 , τD  (12-14)
K τc   τ1  τ 2

• The values of Kc increase as tc decreases but the

values of tI and tD do not change.

• As tc decreases (i.e. Kc increases) the

disturbance rejection gets better (Response
becomes faster, and max. deviation becomes
smaller ), but the set-point tracking gets worse.

• As K is changed from 2 to 0.9 Kc increases. (For

very high values of Kc response becomes too
oscillatory or unstable)

• The controller which gives the best disturbance

rejection (lowest value of tc and highest value of Kc)
gives the worst set-point response
Internal Model Control (IMC)

• IMC is based on a comparison of actual and model process

transfer functions.

• While DS method produces controllers for FOPTD and SOPTD

process models only, IMC method produces various controllers
for different process transfer functions.

• In the IMC method, time delay for the process model must be
approximated by Taylor’s Series or 1st order Pade
approximations. Based on different time delay approximations,
IMC method produces different controllers.

• If Taylor’s series approximation is used for the time delay term

in FOPTD or SOPTD models, IMC and DS methods produce
identical controllers. (REMEMBER: For DS method, only Taylor
series approximation was used for approximating time delay!)
Internal Model Control (IMC)
IMC based PI/PID controller settings for
various process models
Example: Using IMC method to design PI or PID
controllers for FOPTD model
Example: Using IMC method to design PI or
PID controllers for FOPTD model
• Using Taylor’s series approximation (Row G in IMC
table) for the time delay in IMC method produces a PI
controller for an FOPTD process model, which has
identical settings with a controller designed by DS
method.

• Using Pade approximation (Row H in IMC table) for

the time delay in IMC method produces a PID
controller for an FOPTD process model.

• The IMC model can produce either PI or PID

controllers based on the approximation of time delay
for an FOPTD process model.
IMC based PI/PID controller settings for
various process models

FOPTD model, Taylor’s approx. = PI controller

FOPTD model, Pade approx. = PID controller

IPTD, Taylor’s approx. = PI controller

IPTD, Pade approx. = PID controller

Example: Using IMC method to design PI or
PID controllers for Integrating Process Model
Example: Using IMC method to design PI or
PID controllers for Integrating Process Model
• For PI controller, Use row M in IMC table.

• For PID controller, Use row N in IMC table.

 Example: Using IMC method to design PI or
PID controllers for Integrating Process Model

PI Increasing tc

Increasing tc

• For increase of tc , tc = 15 compared to

tc = 8, the response is less oscillatory and
slower (more sluggish) (Because of the
decrease of Kc and increase of tI)

• For increase of tc , tc = 15 compared to tc =

8, overshoot is smaller for set-point
change (better set-point tracking) and
maximum deviation is larger after
disturbance change (worse disturbance
rejection) (Because of decrease of Kc,
similar to the previous example) Increasing tc Increasing tc

• PID controller provides a better

disturbance response than PI controller
PID
with a smaller max. deviation.

• In addition, PID controller has a very

short settling time for tc = 8 , which gives
it the best performance of controllers
considered.
How to select tc ?

Model-based design methods such as DS and IMC produce PI or

PID controllers for certain classes of process models, with one
tuning parameter tc (see Table 12.1)

How to Select tc?

• Several IMC guidelines for t c have been published for the
model in Eq. 12-10:

1. τc / θ > 0.8 and τc  0.1τ (Rivera et al., 1986)

2. τ  τc  θ (Chien and Fruehauf, 1990)
3. τc  θ (Skogestad, 2003)
Controller tuning by ITAE (Integral of the Time-
weighted Absolute error) method
1. Integral of square error (ISE)

ISE   e( t ) dt
2

2. Integral of0absolute value of error (IAE)


IAE   e( t ) dt
3. Time-weighted
0
IAE

ITAE   t e( t ) dt
0
Pick controller parameters to minimize integral.

IAE allows larger deviation than ISE (smaller overshoots)

ISE longer settling time
ITAE weights errors occurring later more heavily

Approximate optimum tuning parameters are correlated

with K, , t (Table 12.3)
ITAE method produces different controllers optimized for disturbance
and set point changes!
Controller tuning by ITAE method
Controller tuning by ITAE method
Example: Controller tuning by ITAE method
Comparison of Controller Design-Tuning
Relationships

1. The controller gain Kc should be inversely proportional to the product of

the other gains in the feedback loop (KPKVKM) .

2. Kc should decrease as /t, the ratio of the time delay to the dominant
time constant increases. In general, the quality of control decreases as /t
increases due to longer settling times and larger maximum deviations from
the set point

3. Both tI and tD should increase as /t increases. For many controller

tuning relations, the ratio, tD/ tI, is between 0.1 and 0.3. As a rule of
thumb, use , tD/ tI = 0.25 as a first guess.

4. When integral control action is added to a proportional-only controller,

Kc should be reduced. The further addition of derivative action allows Kc
to be increased to a value greater than that for proportional-only control.

5. To reduce oscillation, decrease KC and increase tI .

Learning objectives for Chapter10

• List the performance characteristics of FBC systems

• Explain how to design controllers using Direct Synthesis

(DS) and Internal Model Control (IMC) techniques

• Discuss the concepts of desired closed-loop transfer

function (Y/Ysp)d, the desired closed-loop time constant
(tc ) and their effect on closed loop performance

• Outline the tuning of controllers using Integral-of-Time

weighted-Absolute-Error (ITAE) technique

• Compare robustness vs performance of controllers based

on different design and tuning methods
 THE END

Next week, there will be SIMULINK problem solving

sessions (tutorials) during lecture hours.