Advanced Process dynamics and control

Lecture 1: Introduction
(revision)
 Process Control
Process control is the study and application of automatic control in the
field of chemical engineering processes.

 process control has become increasingly important in the process

industries as a consequence of global stiffer competition in the chemical
marketplace ,processes become more complicated to operate, rapidly
changing economic conditions, and stringent environmental and safety
regulations
 Process Control
 It is also clear that the scope and importance of process control technology
will continue to expand.

 Consequently, chemical engineers need to master this subject in order to

be able to design and operate modern plants.
 Advantages of Automatic control of a process

Enhanced process safety

Satisfying environmental constraints

Meeting ever-stricter product quality specifications

More efficient use of raw materials and energy

Increased profitability
 Advantages of Automatic control of a process

Increased product throughput

Increased yield of higher valued products

Extended life of equipment

Improved Operability

Decreased production labor

 Illustration
(a) (b) (c)

 Acceptable product quality requires that the impurity of the product is

below the limit.

 we do not want to make the product purer than necessary, because it

would increase the production costs
 Process Dynamics
 A process is a dynamical system, whose behaviour changes over time. Control
systems are needed to handle such changes in the process.

 Thus, it is important to understand the process dynamics when a control system is

designed.

 Mathematically, the process dynamics can be described by differential equations.

Unsteady‐state (or transient) process behaviour then corresponds to a situation,
where (at least some) time derivatives of the differential equations are nonzero.
 Process Dynamics
Transient
Transient operation occurs during important situations such as start‐ups
and shutdowns, unusual process disturbances, and planned transitions
from one product grade to another.
 Process Dynamics …
Even at normal operation, a process does not operate at a steady state (with
all time derivatives of the differential equations exactly zero) because
there are always variations in external variables, such as feed composition
or cooling medium temperature. Thus, knowledge of steady‐state (or
static) process properties, taught in many engineering courses, is not
sufficient for control design.
 Process Dynamics …
 A dynamical system can be defined as a combination of components that act
together to perform a certain objective.
U (t ) Y (t )
System

 input variables u (t) which affect the system behaviour in some way and

 output variables y (t) which give information about the system behaviour

 The argument “t” indicates that the values of the variables change over time.
Chapter
Chapter11
 Control Terminology
controlled variables - these are the variables which quantify the
performance or quality of the final product, which are also called
output variables.

manipulated variables - these input variables are adjusted

dynamically to keep the controlled variables at their set-points.
 Control Terminology…
 disturbance variables - these are also called "load" variables and
represent input variables that can cause the controlled variables to
deviate from their respective set points.
 set-point change - implementing a change in the operating
conditions. The set-point signal is changed and the manipulated
variable is adjusted appropriately to achieve the new operating
conditions. Also called servomechanism (or "servo") control.
 Control Terminology…
disturbance change - the process transient behaviour when a
disturbance enters, also called regulatory control or load change.
A control system should be able to return each controlled variable
back to its set-point.
Example: level control

disturbance
desired value
(set-point)

transmitter

controller

controlled
variable
(measurement)

manipulated
variable
Example: vaporizes liquid butane and mixes
the vapor with compressed air.
 Question
a. Identify at least three controlled variables, which must
be measured.
b. Identify at least one manipulated variable for each of
the controlled variables.
Hint: these must be valves.
c. Identify at least three disturbance variables. (These do
not have to be measured.) For each, determine which
controlled variable(s) are influenced, i.e., disturbed.
Illustrative Example: Blending system

Notation:
• w1, w2 and w are mass flow rates
• x1, x2 and x are mass fractions of component A
Assumptions:
1. w1 is constant
2. x2 = constant = 1 (stream 2 is pure A)
3. Perfect mixing in the tank

Control Objective:
Keep x at a desired value (or “set point”) xsp, despite
variations in x1(t). Flow rate w2 can be adjusted for this
purpose.

Terminology:
 Controlled variable (or “output variable”): x
 Manipulated variable (or “input variable”): w2
 Disturbance variable (or “load variable”): x1
Design Question. What value of w2 is required to have
x  xSP ?

Overall balance:
0  w1  w2  w (1-1)
Component A balance:

w1x1  w2 x2  wx  0 (1-2)

(The over bars denote nominal steady-state design values.)

• At the design conditions, x  xSP. Substitute Eq. 1-2, x  xSP and

x2  1 , then solve Eq. 1-2 for w2 :
xSP  x1
w2  w1 (1-3)
1  xSP
 Equation 1-3 is the design equation for the blending
system.
 If our assumptions are correct, then this value of w2 will keep x
at xSP. But what if conditions change?

Control Question. Suppose that the inlet concentration x1

changes with time. How can we ensure that x remains at or near
the set point xSP ?
As a specific example, if x1  x1 and w2  w2, then x > xSP.

Some Possible Control Strategies:

Method 1. Measure x and adjust w2.
 Intuitively, if x is too high, we should reduce w2;
 Proportional feedback control law,

w2  t   w2  K c  xSP  x  t   (1-4)

1. where Kc is called the controller gain.

2. w2(t) and x(t) denote variables that change with time t.
3. The change in the flow rate, w2  t   w2is
, proportional to
the deviation from the set point, xSP – x(t).
Method 2. Measure x1 and adjust w2. (what kind of control
strategy?)
• Thus, if x1 is greater than x1, we would decrease w2 so that
w2  w2 ;

• One approach: Consider Eq. (1-3) and replace x1 and w2 with

x1(t) and w2(t) to get a control law:
xSP  x1  t 
w2  t   w1 (1-5)
1  xSP
Method 3. Measure x1 and x, adjust w2.
• This approach is a combination of Methods 1 and 2.

Method 4. Use a larger tank.

• If a larger tank is used, fluctuations in x1 will tend to be damped
out due to the larger capacitance of the tank contents.
• However, a larger tank means an increased capital cost.
Classification of Control Strategies

Table. 1.1 Control Strategies for the Blending System

Method Measured Manipulated Category
Variable Variable
1 x w2 FB
2 x1 w2 FF
3 x1 and x w2 FF/FB
4 - - Design change
 Feedback Control:
• Distinguishing feature: measure the controlled variable
• It is important to make a distinction between negative feedback and positive feedback.
• Advantages:
 Corrective action is taken regardless of the source of the disturbance.
 Reduces sensitivity of the controlled variable to disturbances and changes in the
process (shown later).
• Disadvantages:
 No corrective action occurs until after the disturbance has upset the process,
that is, until after x differs from xsp.
 Very oscillatory responses, or even instability…
Feedforward Control:
 Distinguishing feature: measure a disturbance
variable
• Advantage:
 Correct for disturbance before it upsets the process.
• Disadvantage:
 Must be able to measure the disturbance.
 No corrective action for unmeasured disturbances.
 Closed-loop Artificial Pancreas
glucose u
setpoint
y
r

controller pump patient sensor

measured glucose
Chapter 1
Chapter 1

Block diagram for temperature feedback control system

Chapter 1

Figure 1.6 Block diagram for composition feedback control system on Fig. 1.4.
 or
pneumatic controller
Chapter 1
 Control strategies
Open-loop control
In some simple applications, open-loop control without measurements
can be used. In this control strategy
 the controller is tuned using some a priori information (a “model”) about
the process
after the tuning has been made, the control action is a function of the
setpoint only (setpoint = desired value of the controlled variable)
Open-loop control
Example
 Control strategies…
Feedforward control
Control is clearly needed to eliminate the effect of disturbances on the
system output. Feedforward control is a type of open-loop control strategy,
which can be used for disturbance elimination, if
disturbances can be measured
it is known how the disturbances affect the output
It is known how the control signal affects the output
Feedforward is an open-loop control strategy because the output, which
we want to control, is not measured.
Feedforward control
Example
 Control strategies…
Feedback control

Generally, successful control requires that an output variable is

measured. In feedback control, this measurement is fed to the
controller. Thus
the controller receives information about the behaviour of the process

usually, the measured variable is the variable we want to control (in

principle, it can also be some other variable)
Feedback control
Example
Example
Control system development
Transfer function and Input output model

Lecture 2: (revision)
 Mathematical Modeling of Chemical
Processes
Chapter 2

Mathematical Model (Eykhoff, 1974)

“a representation of the essential aspects of an existing system (or a

system to be constructed) which represents knowledge of that
system in a usable form”k

Everything should be made as simple as possible, but no simpler.

 General Modeling Principles
 The model equations are at best an approximation to the real process.
 Adage: “All models are wrong, but some are useful.”
 Modeling inherently involves a compromise between model accuracy and
complexity on one hand, and the cost and effort required to develop the
model, on the other hand.
 Process modeling is both an art and a science. Creativity is required to
make simplifying assumptions that result in an appropriate model.
 Dynamic models of chemical processes consist of ordinary differential
equations (ODE) and/or partial differential equations (PDE), plus related
algebraic equations.
 Table 2.1. A Systematic Approach for
Developing Dynamic Models
1. State the modeling objectives and the end use of the model. They
determine the required levels of model detail and model accuracy.
2. Draw a schematic diagram of the process and label all process
variables.
3. List all of the assumptions that are involved in developing the model.
Try for parsimony; the model should be no more complicated than
necessary to meet the modeling objectives.
4. Determine whether spatial variations of process variables are
important. If so, a partial differential equation model will be required.
5. Write appropriate conservation equations (mass, component, energy,
and so forth).
 Table 2.1. (continued)
6. Introduce equilibrium relations and other algebraic equations (from
thermodynamics, transport phenomena, chemical kinetics, equipment
geometry, etc.).
7. Perform a degrees of freedom analysis to ensure that the model
equations can be solved.
8. Simplify the model. It is often possible to arrange the equations so
that the dependent variables (outputs) appear on the left side and
the independent variables (inputs) appear on the right side. This
model form is convenient for computer simulation and subsequent
analysis.
9. Classify inputs as disturbance variables or as manipulated variables.
 Conservation Laws
Theoretical models of chemical processes are based on conservation laws.

Conservation of Mass
 rate of mass  rate of mass  rate of mass 
    (2-6)
 accumulation   in   out 
Conservation of Component i
 rate of component i rate of component i 
   
 accumulation   in 

 rate of component i  rate of component i

   (2-7)
 out   produced 
 Conservation of Energy
The general law of energy conservation is also called the First Law of
Thermodynamics. It can be expressed as:
rate of energy  rate of energy in  rate of energy out 
   
 accumulation   by convection   by convection 

net rate of heat addition   net rate of work 

   
 to the system from 
 performed on the sys tem  (2-8)
   by the surroundings 
 the surroundings   

The total energy of a thermodynamic system, Utot, is the sum of its internal energy,
kinetic energy, and potential energy:

U tot  U int  U KE  U PE (2-9)

 Black box (empirical)
 Large number of unknown parameters
 Can be obtained quickly (e.g., linear regression)
 Model structure is subjective
 Dangerous to extrapolate
 Semi-empirical
 Compromise of first two approaches
 Model structure may be simpler
 Typically 2 to 10 physical parameters estimated
(nonlinear regression)
 Good versatility, can be extrapolated
 Can be run in real-time
 linear regression
y  c0  c1 x  c2 x 2
 nonlinear regression


y  K 1  e t / 
 number of parameters affects accuracy of model, but confidence limits on
the parameters fitted must be evaluated
 objective function for data fitting – minimize sum of squares of errors
between data points and model predictions (use optimization code to fit
parameters)
 nonlinear models such as neural nets are becoming popular (automatic
modeling)
 Number of
births (West
Germany)

Number of sightings of storks

Uses of Mathematical Modeling

 To improve understanding of the process

 To optimize process design/operating conditions

 TO design a control strategy for the process

 TO train operating personnel

 Mathematical Modelling
Process engineering systems
 •Development of Dynamic Models
•Illustrative Example: A Blending Process

An unsteady-state mass balance for the blending system:

rate of accumulation   rate of   rate of 

    (2-1)
 of mass in the tank  mass in  mass out 
 or d Vρ 
 w1  w2  w (2-2)
dt
where w1, w2, and w are mass flow rates.

• The unsteady-state component balance is:

d Vρx 
 w1x1  w2 x2  wx (2-3)
dt
The corresponding steady-state model was derived in Ch. 1 (cf. Eqs. 1-1 and 1-2).

0  w1  w2  w (2-4)
0  w1 x1  w2 x2  wx (2-5)
 The Blending Process Revisited

For constant  , Eqs. 2-2 and 2-3 become:

dV
  w1  w2  w (2-12)
dt
 d Vx 
 w1x1  w2 x2  wx (2-13)
dt
Equation 2-13 can be simplified by expanding the accumulation term using the “chain
rule” for differentiation of a product:

d Vx  dx dV
  V  x (2-14)
dt dt dt
Substitution of (2-14) into (2-13) gives:

dx dV
V   x  w1x1  w2 x2  wx (2-15)
dt dt
Substitution of the mass balance in (2-12) for dV/dt in (2-15) gives:

dx
V  x  w1  w2  w   w1x1  w2 x2  wx (2-16)
dt
After canceling common terms and rearranging (2-12) and (2-16), a more convenient
model form is obtained:
dV 1
  w1  w2  w  (2-17)
dt 
dx w1 w2
  x1  x    x2  x  (2-18)
dt V  V
 Stirred-Tank Heating Process

Figure 2.3 Stirred-tank heating process with constant holdup, V.

 Stirred-Tank Heating Process (cont’d.)

Assumptions:

1. Perfect mixing; thus, the exit temperature T is also the

temperature of the tank contents.
2. The liquid holdup V is constant because the inlet and outlet
flow rates are equal.
3. The density  and heat capacity C of the liquid are
assumed to be constant. Thus, their temperature dependence
is neglected.
4. Heat losses are negligible.
For the processes and examples considered in this book, it
is appropriate to make two assumptions:

1. Changes in potential energy and kinetic energy can be

neglected because they are small in comparison with changes
in internal energy.
2. The net rate of work can be neglected because it is small
compared to the rates of heat transfer and convection.
For these reasonable assumptions, the energy balance in
Eq. 2-8 can be written as
dU int
dt

  wH  Q  (2-10)
  denotes the difference
U int  the internal energy of
between outlet and inlet
the system conditions of the flowing
H  enthalpy per unit mass streams; therefore
w  mass flow rate  
-Δ wH = rate of enthalpy of the inlet
Q  rate of heat transfer to the system stream(s) - the enthalpy
of the outlet stream(s)
 Model Development - I
For a pure liquid at low or moderate pressures, the internal
energy is approximately equal to the enthalpy, Uint H, and H
depends only on temperature. Consequently, in the subsequent
development, we assume that Uint = H and Uˆ int  Hˆ where the
caret (^) means per unit mass. a differential change in temperature,
dT, produces a corresponding change in the internal energy per
unit mass, dUˆ int ,

dUˆ int  dHˆ  CdT (2-29)

where C is the constant pressure heat capacity (assumed to be

constant). The total internal energy of the liquid in the tank is:
U int  VUˆ int (2-30)
 Model Development - II
An expression for the rate of internal energy accumulation can be
derived from Eqs. (2-29) and (2-30):
dU int dT
 VC (2-31)
dt dt
Note that this term appears in the general energy balance of Eq. 2-
10.
Suppose that the liquid in the tank is at a temperature T and has an
enthalpy, Ĥ . Integrating Eq. 2-29 from a reference temperature
Tref to T gives,
ref 
Hˆ  Hˆ  C T  T ref  (2-32)
where Hˆ ref is the value of Ĥ at Tref. Without loss of generality, we
assume that Hˆ ref  0 . Thus, (2-32) can be written as:

Hˆ  C T  Tref  (2-33)
 Model Development - III
For the inlet stream


Hˆ i  C Ti  Tref  (2-34)

Substituting (2-33) and (2-34) into the convection term of (2-10) gives:

   
 wHˆ  w C Ti  Tref   w C T  Tref 
      (2-35)

Finally, substitution of (2-31) and (2-35) into (2-10)

dT
V C  wC Ti  T   Q (2-36)
dt
 steam-heating: Q  ws H v

dT
V C  wC (Ti  T )  ws H v (1)
dt

0  wC (Ti  T )  w s H v (2)
subtract (2) from (1)

dT
V C  wC (T  T )  ( ws  ws )H v
dt
divide by wC

V  dT H v
 T T  ( ws  ws )
w dt wC
 Define deviation variables (from set point)

y  T T T is desired operating point

u  ws  ws ws (T ) from steady state
V dy H v H v V
 y  u note that  K p and  1
w dt wC wC w
dy
note when 0 y  K pu
dt
dy
1   y  K pu
dt
General linear ordinary differential equation solution: sum of exponential(s)
Suppose u  1 (unit step response)
  
t

y (t )  K p 1  e 1 
 
 
 Table 2.2. Degrees of Freedom Analysis
1. List all quantities in the model that are known constants (or parameters that
can be specified) on the basis of equipment dimensions, known physical
properties, etc.
2. Determine the number of equations NE and the number of process
variables, NV. Note that time t is not considered to be a process variable
because it is neither a process input nor a process output.
3. Calculate the number of degrees of freedom, NF = NV - NE.
4. Identify the NE output variables that will be obtained by solving the
process model.
5. Identify the NF input variables that must be specified as either disturbance
variables or manipulated variables, in order to utilize the NF degrees of
freedom.
Degrees of Freedom Analysis for the Stirred-Tank
Model:
3 parameters: V , ,C
4 variables: T , Ti , w, Q
1 equation: Eq. 2-36

Thus the degrees of freedom are NF = 4 – 1 = 3. The process variables are classified as:

1 output variable: T
3 input variables: Ti, w, Q

For temperature control purposes, it is reasonable to classify the three inputs as:

2 disturbance variables: Ti, w

1 manipulated variable: Q
Laplace Transform
Methods
revision
 Laplace Transforms
1. Standard notation in dynamics and control
(shorthand notation)

2. Converts mathematics to algebraic operations

3. Advantageous for block diagram analysis

4. Control system design
5.Stability analysis
6. Frequency response
Laplace Transform

L(f (t)) =  f (t)e-st dt
0

Example 1:

a  st   a a
L(a)=  ae dt   e   0     
-st

0
s 0
 s s
 
1  1
L(e )=  e e dt   e
-bt -bt -st -(b+s)t
dt  -e  ( b  s)t
 
0 0
b+s 0 s+b

 df  df -st
L(f )  L     e dt  sL(f) f(0)
 dt  0 dt

Usually define f(0) = 0 (e.g., the error)

Other Transforms
 d2 f   dφ  df
 
L 2  = L  where φ =
 dt   dt  dt
 sφ(s) - φ(0)
= ssF(s) - f(0)   f (0)
= s 2F(s)  sf(0)  f (0)
d nf
etc. for  e  j t - e  jt 
dt n L(sin ωt) = L  
 2j 
 e -jt  e  jt 
L(cos ωt) = L  ω
 2  = 2
1 1 1  s  ω2
=   
2  s  jω s  jω  Note:
1  s  jω s  jω 
=  2  
2  s  ω 2 s2  ω2 
s
= 2
s  ω2
Table 3.1 Laplace Transforms for Various Time-Domain Functionsa

f(t) F(s)
Table 3.1 Laplace Transforms for Various Time-Domain Functionsa

f(t) F(s)
Table 3.1 Laplace Transforms for Various Time-Domain
Functionsa (continued)

f(t) F(s)
Example 3.1
Solve the ODE,
dy
5  4y  2 y 0  1 (3-26)
dt
First, take L of both sides of (3-26),
5  sY  s   1  4Y  s  
2
s
Rearrange,
5s  2
Y s  (3-34)
s  5s  4 
Take L-1,
1 
5s  2 
y t   L  
 s  5s  4  
From Table 3.1 (line 11),
y  t   0.5  0.5e0.8t (3-37)
Example:

d3y d2y dy
3
 6 2  11  6 y  4
dt dt dt
y( 0 )= y( 0 )= y( 0 )= 0
system at rest (s.s.)

Step 1 Take L.T. (note zero initial conditions)

4
s Y(s)+ 6s Y(s)+11sY(s)  6Y ( s ) =
3 2

s
 Rearranging,

4
Y(s)=
( s 3  6s 2  11s  6) s

Step 2a. Factor denominator of Y(s)

s(s 3+6s 2+11s+6 )=s(s+1 )(s+2 )(s+3 )

Step 2b. Use partial fraction decomposition

4 α α α α
 1 2  3  4
s(s +1 )(s + 2 )(s + 3 ) s s  1 s  2 s  3

Multiply by s, set s = 0
4  α α α 
 α1  s  2  3  4 
(s +1 )(s + 2 )(s + 3 ) s 0  s  1 s  2 s  3  s 0
4 2
 α1 
1 2  3 3
 For a2, multiply by (s+1), set s=-1 (same procedure
for a3, a4)
2
α2  2, α3  2, α4  
3
2 2 2 2/3
Step 3. Take inverse of L.T. (Y(s)=  +  )
3s s 1 s  2 s3
2 2
y(t)=  2e  t  2e 2t  e 3t
3 3
2
t  y(t)  t  0 y (0)  0. (check original ODE)
3
You can use this method on any order of ODE,
limited only by factoring of denominator polynomial
(characteristic equation)

Must use modified procedure for repeated roots,

imaginary roots
 Factoring the denominator polynomial

1. 2
3s 2  4 s  1
1
3s  4 s  1  (3s  1)( s  1)  3( s  )( s  1)
2

3
Transforms to e-t/3, e-t

Real roots = no oscillation

2. 2 s
s2  s  1
1 3 1 3 1 3
s 2  s  1  (s   j )( s   j )  (s  )2  ( )2
2 2 2 2 2 2
0.5 t 3 3
Transforms to e sin t , e 0.5t cos t
2 2
Complex roots = oscillation
3 1
s
L1[ 3( 2 )( 2 )]
1 3 1 3
(s  )2  ( )2 (s  )2  ( )2
2 2 2 2

From Table 3.1, line 17 and 18

t
 3 3
y (t )  e (sin(
2
t )  3 cos( t ))
2 2
Let h→0, f(t) = δ(t) (Dirac delta) L(δ) = 1
h 1  st 1
F ( s )   e dt  (1  e  hs ) Use L’Hopital’s theorem
0 h hs
(h→0)
If h = 1, rectangular pulse input
Difference of two step inputs S(t) – S(t-1)

(S(t-1) is step starting at t = h = 1)

By Laplace transform
1 e s
F ( s)  
s s

Can be generalized to steps of different magnitudes

(a1, a2).
One other useful feature of the Laplace transform
is that one can analyze the denominator of the transform
to determine its dynamic behavior. For example, if
1
Y(s)= 2
s  3s  2
the denominator can be factored into (s+2)(s+1).
Using the partial fraction technique
α1 α2
Y(s)= 
s  2 s 1
The step response of the process will have exponential terms
e-2t and e-t, which indicates y(t) approaches zero. However, if
1 1
Y(s)= 2 
s  s  2 (s  1 )(s  2 )
We know that the system is unstable and has a transient
response involving e2t and e-t. e2t is unbounded for
large time. We shall use this concept later in the analysis
of feedback system stability.
Other applications of L( ):
A. Final value theorem
y()= lim sY(s) “offset”
s 0

Example 3: step response

1 a
Y(s) 
τs  1 s
a a
sY(s)  lim a
τs  1 s 0 τs  1

offset (steady state error) is a.

B. Time-shift theorem
y(t)=0 t < θ L  yt- =e -s
Y(s)
C. Initial value theorem

lim y(t)= lim sY(s)

t 0 s 

4s+2
For Y(s)=
s(s+1 )(s+2 )(s+3 )

y( 0 )= 0 by initial value theorem

1
y()= by final value theorem
3
Transfer Functions
 Transfer Functions
• Convenient representation of a linear, dynamic model.
• A transfer function (TF) relates one input and one output:

u (t ) y (t )
 system 
U ( s) Y (s)

The following terminology is used:

u y
input output
forcing function response
“cause” “effect”
Definition of the transfer function:
Let G(s) denote the transfer function between an input, x, and an
output, y. Then, by definition
Y ( s)
G( s) 
U ( s)
where:

Y ( s )  L y (t ) 
U ( s )  L u (t ) 

Development of Transfer Functions

Example: Stirred Tank Heating System
Figure 2.3 Stirred-tank heating process with constant holdup, V.
Recall the previous dynamic model, assuming constant liquid
holdup and flow rates:
dT
V C  wC Ti  T   Q (2-36)
dt
Suppose the process is at steady state:

0  wC Ti  T   Q (2)

Subtract (2) from (2-36):

 wC Ti  Ti   T  T     Q  Q 
dT
V C (3)
dt
But,

dT 
V C  wC Ti  T    Q (4)
dt

where the “deviation variables” are

T   T  T , Ti  Ti  Ti , Q  Q  Q

Take L of (4):
V  C  sT   s   T   0    wC Ti s   T   s   Q  s  (5)

At the initial steady state, T′(0) = 0.

Rearrange (5) to solve for

 K   1 
T  s     Q  s     Ti s  (6)
  s 1   s 1
where

1 V
K and  
wC w
 T (s)=G1(s)Q(s)  G2(s)Ti(s)


G1 and G2 are transfer functions and independent of the

inputs, Q′ and Ti′.
Note G1 (process) has gain K and time constant .
G2 (disturbance) has gain=1 and time constant .
gain = G(s=0). Both are first order processes.
If there is no change in inlet temperature (Ti′= 0),
then Ti′(s) = 0.
System can be forced by a change in either Ti or Q
(see Example 4.3).
Conclusions about TFs
1. Note that (6) shows that the effects of changes in both Q
and Ti are additive. This always occurs for linear, dynamic
models (like TFs) because the Principle of Superposition is
valid.
2. The TF model enables us to determine the output response to
any change in an input.
3. Use deviation variables to eliminate initial conditions for TF
models.
 Example: Stirred Tank Heater
K  0.05   2.0

T 
0.05
Q No change in Ti′
2s  1
Step change in Q(t): 1500 cal/sec to 2000 cal/sec
500

Q 
s
0.05 500 25
T  
2s  1 s s(2s  1)
What is T′(t)? From line 13, Table 3.1
 t / 25
T (t )  25[1  e ] 
 T ( s) 
s( s  1)
 t / 2
T (t )  25[1  e ]
Properties of Transfer Function Models

1. Steady-State Gain
The steady-state of a TF can be used to calculate the steady-
state change in an output due to a steady-state change in the
input. For example, suppose we know two steady states for an
input, u, and an output, y. Then we can calculate the steady-
state gain, K, from:

y2  y1
K (4-38)
u2  u1

For a linear system, K is a constant. But for a nonlinear

system, K will depend on the operating condition  u , y  .
Calculation of K from the TF Model:
If a TF model has a steady-state gain, then:

K  lim G  s  (14)
s 0

• This important result is a consequence of the Final Value

Theorem

• Note: Some TF models do not have a steady-state gain (e.g.,

integrating process)
2. Order of a TF Model
Consider a general n-th order, linear ODE:
dny dy n1 dy d mu
an n
 an1 n 1
 a1  a0 y  bm m 
dt dt dt dt
d m1u du
bm1 m 1
  b1  b0u (4-39)
dt dt

Take L, assuming the initial conditions are all zero. Rearranging

gives the TF:
m

Y s
i
b s i

G s   i 0
(4-40)
U s n
i
a s i

i 0
Two IMPORTANT properties (L.T.)

A. Multiplicative Rule

B. Additive Rule
Example 1:
Place sensor for temperature downstream from heated
tank (transport lag)

Distance L for plug flow,

L
Dead time 
V
V = fluid velocity

T(s) K1
Tank: G1 = =
U(s) 1+ 1s
Ts (s) K 2 e - s
Sensor: G 2 = = K 2  1,  2 is very small
T(s) 1 +  2s
(neglect)
Overall transfer function:
Ts Ts T K1K 2 e  s
   G 2  G1 
U T U 1  1s
Linearization of Nonlinear Models
• Required to derive transfer function.
• Good approximation near a given operating point.
• Gain, time constants may change with
• operating point.
• Use 1st order Taylor series.
dy
 f ( y, u ) (4-60)
dt
f f
f ( y, u )  f ( y, u )  ( y  y)  (u  u ) (4-61)
y y ,u
u y ,u

Subtract steady-state equation from dynamic equation

dy f f
 
y  u (4-62)
dt y s u s
Example 3:
q0: control,
qi: disturbance

dh
A  qi  q0 qi  q0 at s.s.
dt
dh
Use L.T. A  qi  q0
dt
AsH (s)  qi (s)  q0 (s) (deviation variables)

suppose q0 is constant

 H(s) 1 pure integrator (ramp) for


AsH ( s)  qi ( s), 
q  ( s) As step change in qi
i
If q0 is manipulated by a flow control valve,

q0  Cv h
nonlinear element

Figure 2.5

Linear model

R: line and valve resistance

linear ODE : eq. (4-74)

 if q0  CV h
dh
A  qi  Cv h
dt
Perform Taylor series of right hand side

dh 1
A  qi  h
dt R

R  2h 0.5
/ Cv
Dynamic Behavior
 Dynamic Behavior
In analyzing process dynamic and process control systems, it is important to know how
the process responds to changes in the process inputs.
A number of standard types of input changes are widely used for two reasons:

1. They are representative of the types of changes that occur

in plants.

2. They are easy to analyze mathematically.

1. Step Input

A sudden change in a process variable can be approximated by

a step change of magnitude, M:

The step change occurs at an arbitrary time denoted as t = 0.

• Special Case: If M = 1, we have a “unit step change”. We

give it the symbol, S(t).

• Example of a step change: A reactor feedstock is suddenly

switched from one supply to another, causing sudden
changes in feed concentration, flow, etc.
Example:
The heat input to the stirred-tank heating system in Chapter 2 is suddenly changed from
8000 to 10,000 kcal/hr by changing the electrical signal to the heater. Thus,

Q  t   8000  2000 S  t  , S t  unit step

and
Q  t   Q  Q  2000S  t  , Q  8000 kcal/hr

2. Ramp Input
• Industrial processes often experience “drifting
disturbances”, that is, relatively slow changes up or down
for some period of time.
• The rate of change is approximately constant.
We can approximate a drifting disturbance by a ramp input:

Examples of ramp changes:

1. Ramp a setpoint to a new value. (Why not make a step

change?)
2. Feed composition, heat exchanger fouling, catalyst
activity, ambient temperature.
3. Rectangular Pulse
It represents a brief, sudden change in a process variable:

URP
tw Time, t
h

Examples:

1. Reactor feed is shut off for one hour.

2. The fuel gas supply to a furnace is briefly interrupted.
 Other Inputs
4. Sinusoidal Input
Processes are also subject to periodic, or cyclic, disturbances.
They can be approximated by a sinusoidal disturbance:

0 for t  0
U sin  t   (5-14)
 A sin t  for t  0
where: A = amplitude, ω = angular frequency

A
U sin ( s )  2
s  2
Examples:

1. 24 hour variations in cooling water temperature.

2. 60-Hz electrical noise (in USA!)
For a sine input (1st order process)


U (s) 
s2   2
output is...

Kp  a0 a1s a2
Y(s)   2   2  2
s  1 s   2
s  1 s   s  2
2

By partial fraction decomposition,

K p  2
a0 
2  2  1
 K p 
a1 
2  2  1
K p
a2 
2  2  1
Inverting,

this term dies out for large t

K p  t 
Kp
y( t )  e  sin(t  )
  1
2 2
2  2  1
   arctan()
note:  is not a function of t but of  and .

For large t, y(t) is also sinusoidal,

output sine is attenuated by…

1
(fast vs. slow )
2  2  1
5. Impulse Input

• Here, U I  t     t  and U I ( s )  1
• It represents a short, transient disturbance.
• It is the limit of a rectangular pulse for tw→0 and h = 1/tw
Examples:
1. Electrical noise spike in a thermo-couple reading.
2. Injection of a tracer dye.

Here,

Y s  G s (1)

Second order process example, Example 4.2

y  T T u=Q-Q Ti fixed

mm e Ce d 2 y  m e Ce m e Ce m  dy 1
2
    y u
wh e A e dt  h e Ae wC w  dt wC

note when Ce  0, obtain 1st order equation

(simpler model)
Block Notation:

Composed of two first order subsystems (G1 and G2)

K 2nd order ODE model

G(s) = 2 2
 s  2s  1
(overdamped)

K
   1 2 G(s) =
1 2s 2 + (1   2 )s + 1
1   2
=  1 overdamped
2 1 2
 1 underdamped
    2 1
roots:  1 critically damped

 Second Order Step Change
a. Overshoot
a   
 exp  
b  1 2 
 

b. time of first maximum


tp 
1 2

c. decay ratio (successive maxima – not min.)

c  2  a2
 exp   
a  1 2  b 2
 

d. period of oscillation
2
p 
1 2
 K1 2  K 2 1
For inverse response 0 (6  21)
K1  K 2
 K
G (s) 
s ( 1s  1)( 2 s 2  2 s  1)

4 poles (denominator is 4th order polynomial)

More General Transfer Function Models
• Poles and Zeros:
• The dynamic behavior of a transfer function model can be
characterized by the numerical value of its poles and zeros.

• General Representation of a TF:

There are two equivalent representations:

m
i
b s i

G s  i 0
n
(4-40)
i
a s i

i 0
 G s 

bm  s  z1  s  z2 s  zm 
(6-7)
an  s  p1  s  p2   s  pn 

where {zi} are the “zeros” and {pi} are the “poles”.

• We will assume that there are no “pole-zero” cancellations. That

is, that no pole has the same numerical value as a zero.

• Review: n  m in order to have a physically realizable system.

 Time Delays
Time delays occur due to:

1. Fluid flow in a pipe

2. Transport of solid material (e.g., conveyor belt)
3. Chemical analysis
- Sampling line delay
- Time required to do the analysis (e.g., on-line gas
chromatograph)
Mathematical description:
A time delay, , between an input u and an output y results in the following expression:
θ

 0 for t  θ
y t    (6-27)
u  t  θ  for t  θ
  s Y (s)
G( s)  e 
U ( s)
H1/Qi has numerator dynamics (see 6-72)
 Approximation of Higher-Order Transfer
Functions
In this section, we present a general approach for approximating high-order transfer
function models with lower-order models that have similar dynamic and steady-
state characteristics.

In Eq. 6-4 we showed that the transfer function for a time delay can be expressed as
a Taylor series expansion. For small values of s,

e θ 0 s  1  θ 0 s (6-57)
• An alternative first-order approximation consists of the transfer
function,

θ 0 s 1 1
e   (6-58)
e θ0 s 1  θ0 s

where the time constant has a value of θ0 .

• These expressions can be used to approximate the pole or zero
term in a transfer function.
 1/1 Pade
approximation:
Feedback Controller:Example
Flow in

SP

LC
LT

Flow out

Slide by: Shimelis Kebede (PhD), AAU/AAiT/SCBE email:-

[email protected]
A conceptual block diagram of feedback
control loop of a chemical process

Slide by: Shimelis Kebede (PhD), AAU/AAiT/SCBE email:-

[email protected]
A feedback control law consists of the
following steps of action
Measurement of process output y, such as pressure, temperature,
flow, level, composition, using a suitable measuring device.
Comparison of process output and its desired setpoint (ysp) and
calculation of the deviation error є=(y-ysp)
Feeding the value of deviation error to the controller, which
calculates the appropriate control action (c) to be taken.
Feeding the value of the control action to the final control element,
that translates the control action into manipulated form of process
input u .

Slide by: Shimelis Kebede (PhD), AAU/AAiT/SCBE email:-

[email protected]
feedback control configuration
Process
Any equipment that serves the targeted physical/chemical operation of the plant is
termed as a process.
Measuring devices
Measuring Instruments or Sensors
Transmission lines
Measurement and/or control signals are carried through various transmission lines.
Controllers
a mathematical function block (P,PI, and PID)
Final Control Elements
The hardware component of a control loop

Slide by: Shimelis Kebede (PhD), AAU/AAiT/SCBE email:-

[email protected]
Dynamic behaviour of feedback controller

Slide by: Shimelis Kebede (PhD), AAU/AAiT/SCBE email:-

[email protected]
Slide by: Shimelis Kebede (PhD), AAU/AAiT/SCBE email:-
[email protected]
Simplified block diagram of feedback control
loop of a chemical process

Slide by: Shimelis Kebede (PhD), AAU/AAiT/SCBE email:-

[email protected]
Transfer Function for Controller

Proportional (P) controller

Proportional Integral (PI) controller

Proportional Integral Derivative (PID) controller

Slide by: Shimelis Kebede (PhD), AAU/AAiT/SCBE email:-

[email protected]
EXAMPLE
Consider Standard block diagram of a feedback control system with
the following transfer functions:

Show that the closed-loop system

Slide by: Shimelis Kebede (PhD), AAU/AAiT/SCBE email:-

[email protected]
Solution
consider a unit step change in set point,

Slide by: Shimelis Kebede (PhD), AAU/AAiT/SCBE email:-

[email protected]