Chapter 8

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol

CCB3013 - Chemical Process Dynamics, Instrumentation and Control 1 7/24/2013

Chapter 8

FEEDBACK CONTROLLERS

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Chapter Objectives

End of this chapter, you should be able to:

1. Explain the concept of feedback control

2. Explain P, I and D controllers

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Historical Perspective:

1930s – Commercial Three-mode controllers with

proportional, integral and derivative

(PID) feedback control action

1940s – Widespread acceptance of pneumatic

PID controllers

1950s – Electronic counterparts in the market

1960s – Computer applications

1980s – Use of digital hardware

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Introduction

Consider the following blending process (Fig. 8.1).

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Control objective:

To keep the tank exit composition x at the desired setpoint by adjusting w2.

Measurement : Composition Analyzer-

Transmitter (AT)

Feedback controller (AC): Automatic Controller

Final control element: Pneumatic control valve

Current-to-pneumatic (I/P) transducer

Control system

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Block Diagram

Figs. 8.2 & 8.3: Flow control system/loop (top) and its

block diagram (bottom).

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Block Diagram

Fig. 11.8: Standard block diagram for a feedback

control system.

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol

THREE MODE CONTROLLER

Proportional, Integral and Derivative

CCB3013 - Chemical Process Dynamics, Instrumentation

and Control 8 7/24/2013

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Feedback Control Error

In feedback control, the objective is to reduce the

error signal to zero.

Define an error signal, e, by

)()()( tytyte mSP (6.1)

where spy = set point

my = measured value of the controlled variable

(or equivalent signal from transmitter)

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


• For proportional control, the controller output is

proportional to the error signal

e(t) Kpp(t) c (6.2)

where p(t) = controller output

p = bias value (adjustable, manual reset)

Kc = controller gain (dimensionless, adjustable,

tuning)

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Function of proportional term

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Proportional Band, PB:

Definition : cK

PB%100

• Applies when Kc is dimensionless

• Small (narrow) PB corresponds to large Kc

• Large (wide) PB corresponds to small Kc

(6.3)

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Proportional controller

In order to derive the transfer function for an ideal

proportional controller, define a deviation variable

as

ptptp )()( (6.4)

Then (6.2) can be written as

)()( teKtp c (6.5)

Taking Laplace transform of (6.5) and rearranging

we get

cKsE

sP

)(

)((6.6)

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Remarks

• An inherent limitation of proportional controller is

that a steady-state error (offset) occurs after a set-

point change or a sustained disturbance.

• Offset can be eliminated by manually resetting

either the set-point or bias after an offset occurs –

impractical.

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Ideal vs. actual

Ideal controller does not

include physical limits

A controller saturates

when its output reaches a

physical limit, either pmax

or pmin.

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Reverse or Direct Acting Controller Kc can be made positive or negative

• Direct-Acting (Kc < 0)

• controller output (p)

increases as input (ym)

increases

• Reverse-Acting (Kc > 0)

• controller output (p)

increases as input (ym)

decreases

)( mSPcc yyKpeKpp

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol



Integral Control

(reset control, floating control)

For integral action, the controller output depends

on the integral of the error signal over time,

t)dte(pp(t)t

I

0

1

(6.7)

where is an adjustable parameter and referred to

as the integral time constant or reset time, has units

of time.

I

The transfer

function: sI

1

E(s)

(s)P

(6.8)

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Integral Control

• An important practical advantage: Eliminates offset.

• For the process being controlled to be at steady state, the

controller output p must be constant so that the manipulated

variable is also constant.

• Eq. (6.7) implies that p changes with time unless e(t) = 0.

• This desirable situation occurs unless the controller output

or the final control element saturates.

• The control action by the integral controller is very little

until the error signal has persisted for sometime.

• On the other hand, proportional controller takes immediate

corrective action as soon as an error is detected.

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Function of integral term

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


PI Controller

Integral control is used in conjunction with

proportional control as the proportional-integral (PI)

controller:

t

I

c tdteteKptp0

)(1

)()(

(6.9)

The corresponding transfer function is:

sK

I

c

11

E(s)

(s)P(6.10)

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


PI Controller

The response of the PI controller to a unit step

change in e(t) is shown in Fig.

- repeats per minute, aka, reset rate. I/1

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


PI Controller

Disadvantages:

• Produces oscillatory response

• Reset windup

When a sustained error occurs, the integral term becomes

quite large and the controller output eventually saturates –

reset windup or integral windup.

Antireset windup: Temporarily halting the integral action

whenever the control output saturates.

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Reset windup

Valve

movement

SP=setpoint

PV=process

variable to

control

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Anti wind-up

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Derivative control (Rate action, pre-act, anticipatory control )

• Anticipate the future error by considering its rate of

change.

• For ideal derivative action,

dt

tdeptp D

)()( (6.11)

where is the derivative time, and has units of time. D

As long as the error is constant de/dt = 0, the controller

output is equal to . p

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol



Derivative control

• Derivative action is never used alone.

• Always used in conjunction with P or PI control.

PD controller has the transfer function

sK Dc

1E(s)

(s)P(6.12)

The derivative control action tends to stabilize the

controlled process.

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


PID Controller

PID control algorithm is given by

t

D

I

cdt

detdteteKptp

0

)(1

)()(

(6.13)

Transfer function of an ideal controller (parallel form)

s

sK D

I

c

11

E(s)

(s)P(6.14)

Transfer function – actual (Series form)

1

11

E(s)

(s)P

s

s

s

sK

D

D

I

Ic

Derivative filter

(6.15)

= [0.05, 0.2]

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


ON-OFF controllers

(“two-position” or “bang-bang” controllers)

Ideal controller More practical controller

(Dead band)

• Special case of proportional controller with very high gain.

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Typical responses of Feedback

control systems

Consider response of a controlled system after a sustained

disturbance occurs (e.g. step change in load variable)

No control

New steady state is reached

P control

Offset reduced

PI control

Offset eliminated

Oscillatory response

PID control

Oscillations reduced

Response time reduced

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol



Too small a value of Kc

Sluggish response

Larger deviation

Too large a value of Kc

Exhibit oscillatory or unstable behavior

Intermediate values of Kc is desirable

Increasing tends to improve the

response by reducing the maximum

deviation, response time, and degree

of oscillation

If is too large, measurement noise

is amplified and the response may

become oscillatory.

D

D

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


• Increasing I makes the controller more sluggish.

• Offset will be eliminated for all values of I.

• For large values of I, it takes very long time to return to

the set-point.

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Summary

1. Two Position (On-Off):

• Inexpensive

• Extremely simple

2. Proportional:

• Simple

• Inherently stable when properly tuned

• Easy to tune

• Experiences offset at steady state

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol



3. Proportional + Integral:

• No offset

• Better dynamic response than reset alone

• Possibilities exist for instability due to lag introduced

4. Proportional + Derivative:

• Stable

• Less offset than proportional alone (use of higher gain

possible).

• Faster response time.

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol



5. Proportional + Integral + Derivative :

• Most complex

• Rapid response

• No offset

• Difficult to tune

• Best control if properly tuned.

Pro

cess D

yn

am

ics a

nd

Co

ntr

ol


Conclusion

• Concept of feedback control

• P, I, D controller modes

– ON-OFF as a special case of P-controller

• Advantages and disadvantages

• Motivation for additional modes

Chapter 8

Documents

Transcript of Chapter 8