Dynamics of

Low-order
Processes
 Attendance Quiz!
What is the transfer function for the
following, if X3 is the output and X1 and X2
are the deviation variables? What is the
steady-state gain? Liquid A Liquid B
1 2

Product
First-Order Processes

𝑌 𝑠 𝐾
=
𝑈(𝑠) 𝜏𝑠 + 1
Denominator is a 1st order polynomial
First-Order
Processes
Responses to different inputs
0, 𝑡≤0
• Step Input 𝑢 𝑡 =
𝑀, 𝑡 ≥ 0

• Ramp Input
0, 𝑡 ≤ 0
𝑢 𝑡 =
𝑎𝑡, 𝑡 ≥ 0 0, 𝑡<0
• Rectangular Pulse 𝑢 𝑡 = ℎ,
0,
0 ≤ 𝑡 < 𝑡𝑤
𝑡 ≥ 𝑡𝑤
• Sinusoidal input 𝑢 𝑡 = 𝐴 sin 𝜔𝑡
• Impulse Input 𝑢 𝑡 = 𝛿(𝑡)

• Random Input (fluctuation)

Attendance Quiz!
1) Graph u(t), 2) Express in terms of step function U,
and 3) find the inverse Laplace transform
0, 𝑡≤0
• Step Input 𝑢 𝑡 =
𝑀, 𝑡 ≥ 0

• Ramp Input
0, 𝑡 ≤ 0
𝑢 𝑡 =
𝑎𝑡, 𝑡 ≥ 0 0, 𝑡<0
• Rectangular Pulse 𝑢 𝑡 = ℎ,
0,
0 ≤ 𝑡 < 𝑡𝑤
𝑡 ≥ 𝑡𝑤
• Sinusoidal input 𝑢 𝑡 = 𝐴 sin 𝜔𝑡
• Impulse Input 𝑢 𝑡 = 𝛿(𝑡)

• Random Input (fluctuation)

 1st Order Process – Step
Response
𝑌 𝑠 𝐾
=
𝑈 𝑠 =
𝑀 𝑈(𝑠) 𝜏𝑠 + 1
𝑠

𝐾𝑀
𝑌 𝑠 =
𝑠 𝜏𝑠 + 1

𝑦 𝑡 = 𝐾𝑀 1 − 𝑒 −𝑡/𝜏
A complex process response to a unit step input is approximated as a first
order transfer function.
1. What is the steady-state gain?
2. What is the time constant?

10
t
 1st Order Process – Ramp
Response
𝑌 𝑠 𝐾
=
𝑎 𝑈(𝑠) 𝜏𝑠 + 1
𝑈 𝑠 =
𝑠2

𝐾𝑎
𝑌 𝑠 = 2
𝑠 𝜏𝑠 + 1

𝑡
−
𝑦 𝑡 = 𝐾𝑎𝜏 𝑒 𝜏 − 1 + 𝐾𝑎𝑡
 1st Order Process – Sinusoidal
Response
𝑌 𝑠 𝐾
=
𝑈 𝑠 =
𝐴𝜔 𝑈(𝑠) 𝜏𝑠 + 1
𝑠2 + 𝜔2

𝐾𝐴𝜔
𝑌 𝑠 =
𝜏𝑠 + 1 𝑠 2 + 𝜔 2

𝐾𝐴 −
𝑡
𝑦 𝑡 = 2 2 𝜔𝜏𝑒 − 𝜔𝜏 cos 𝜔𝑡 + sin 𝜔𝑡
𝜏
𝜔 𝜏 +1

𝐾𝐴𝜔𝜏 𝐾𝐴
𝑦 𝑡 = 𝑒 −𝑡/𝜏 + sin 𝜔𝑡 + 𝜙
𝜔2 𝜏2 +1 𝜔2𝜏2 +1 𝜙 = − tan−1 𝜔𝜏
 Integrating Processes
What are the transfer functions
H’(s)/Q3’(s)? What is h(t) if product flow
increases by a step of magnitude M?
Liquid A
1

Product
SECOND ORDER
PROCESSES
 Recall:
𝑒 𝑖𝑥 + 𝑒 −𝑗𝑥 𝑒 𝑗𝑥 − 𝑒 −𝑗𝑥
cos 𝑥 = sin 𝑥 =
2 2𝑗

𝑒 𝑥 + 𝑒 −𝑥
cosh 𝑥 = cos 𝑗𝑥 =
2

𝑒 𝑥 − 𝑒 −𝑥
𝑗 sinh 𝑥 = sin 𝑗𝑥 = 𝑗
2
Form
𝐾1 𝐾2
𝐺 𝑠 =
𝜏1 𝑠 + 1 𝜏2 𝑠 + 1
𝐾
𝐺 𝑠 = 2 2
𝜏 𝑠 + 2𝜁𝜏𝑠 + 1
 Attendance Quiz!
What is K, τ, and Ϛ in terms of their
counterparts?
𝐾1 𝐾2
𝐺 𝑠 =
𝜏1 𝑠 + 1 𝜏2 𝑠 + 1
𝐾
𝐺 𝑠 = 2 2
𝜏 𝑠 + 2𝜁𝜏𝑠 + 1
Form

𝐾1 𝐾2
𝐺 𝑠 =
𝜏1 𝑠 + 1 𝜏2 𝑠 + 1
𝐾
𝐺 𝑠 = 2 2
𝜏 𝑠 + 2𝜁𝜏𝑠 + 1
 Step Response
Overdamped, Ϛ>1
𝑌(𝑠) 𝐾
𝐺 𝑠 = = 2 2
𝑋(𝑠) 𝜏 𝑠 + 2𝜁𝜏𝑠 + 1

𝜁2 − 1 𝜁 𝜁2 − 1
𝑦 𝑡 = 𝐾𝑀 1 − 𝑒 −𝜁𝑡/𝜏 cosh 𝑡 + sinh 𝑡
𝜏 2
𝜁 −1 𝜏
 Step Response
Critically damped, Ϛ=1
𝑌(𝑠) 𝐾
𝐺 𝑠 = = 2 2
𝑋(𝑠) 𝜏 𝑠 + 2𝜁𝜏𝑠 + 1

1 −𝑡/𝜏
𝑦 𝑡 = 𝐾𝑀 1 − 1 + 𝑒
𝜏
 Step Response
Underdamped, Ϛ<1
𝑌(𝑠) 𝐾
𝐺 𝑠 = = 2 2
𝑋(𝑠) 𝜏 𝑠 + 2𝜁𝜏𝑠 + 1

1 − 𝜁2 𝜁 1 − 𝜁2
𝑦 𝑡 = 𝐾𝑀 1 − 𝑒 𝜁 𝑡/𝜏 cos 𝑡 + sin 𝑡
𝜏 1−𝜁 2 𝜏
 Overshoot (OS) = a/b
Decay Ratio (DR) = c/a = (a/b)2

If not underdamped, tr = time from 10% to 90%

 Sinuisoidal Input
𝑌(𝑠) 𝐾
𝐺 𝑠 = = 2 2
𝑋(𝑠) 𝜏 𝑠 + 2𝜁𝜏𝑠 + 1
𝐾𝐴
𝑦 𝑡 ≈ sin 𝜔𝑡 + 𝜙
1 − 𝜔𝜏 2 2 + 2𝜁𝜔𝜏 2

Output Amplitude = Â

ARN = Â/KA
TIME DELAY
Recall: Second
Shifting Theorem
Approximations
Taylor

𝜃 2 𝑠2 𝜃 3 𝑠3 𝜃 4 𝑠4 𝜃 5 𝑠5
𝑒 −𝜃𝑠 = 1 − 𝜃𝑠 + − + − +⋯
2! 3 4! 5!
−𝜃𝑠
1 1
𝑒 = 𝜃𝑠 ≈
𝑒 1 + 𝜃𝑠
Approximations
1/1 Padé

𝜃
1− 𝑠
𝑒 −𝜃𝑠
≈ 2
𝜃
1+2𝑠
Approximations
2/2 Padé

𝜃 𝜃 2 𝑠2
1− 𝑠+
𝑒 −𝜃𝑠
≈ 2 12
𝜃 𝜃 2 𝑠2
1+ 𝑠+
2 12
Approximations
1/1 Padé

Plot the poles and zeros of the 1/1

Padé approximation
𝜃
1− 𝑠
𝑒 −𝜃𝑠
≈ 2
𝜃
1+2𝑠
Time delay approximation of a step response
G1 = 1/1 Padé
G2 = 2/2 Padé
Step response of a first-order + time delay process
G1 = 1/1 Padé
G2 = 2/2 Padé
Upload your plots to UVLE:
UVLE
LAST NAME_2ndOrderwND

Plot and compare

Second-order process
• Overdamped w/o numerator dynamics
• Overdamped w/ numerator dynamics
– τa << τ 1 or τ2
– τa < τ 1 or τ2
– τa = τ1 or τ2
– τa > τ1 or τ2
– τa >> τ1 or τ2
• Overdamped w/ numerator dynamics using time delay approximation
MORE COMPLEX
PROCESSES
 imaginary

Poles
real

• Roots of denominator
polynomial
 imaginary

Poles
real

• What are the roots of the

following?
𝐾
𝐺 𝑠 =
𝑠 𝜏1 𝑠 + 1 𝜏22 𝑠 2 + 2𝜁𝜏2 𝑠 + 1
 imaginary

Zeros
real

• Roots of numerator
polynomial
 imaginary

Zeros
real

• Plot the poles and zeros of the

following:
𝐾 𝜏𝑎 𝑠 + 1
𝐺 𝑠 =
𝜏1 𝑠 + 1 (𝜏2 𝑠 + 1)
τa = 8
τ1 = 4
τ2 = 1
 imaginary

real
-0.5
-2 -1 1
-1
What is the
transfer function?
What are the
expected
responses?
Plot the poles and
zeros

𝐾 −0.1𝑠 + 1
𝐺 𝑠 =
5𝑠 + 1 3𝑠 + 1 0.5𝑠 + 1
EMPIRICAL MODELS
Empirical Models

• Linear Regression
• First Order Plus Time Delay
• Smith’s Method (Second Order Plus
Time Delay)
First Order Plus Time
Delay
Smith’s Method
Second Order Plus Time
Delay
Other Models

• Responses to non-step inputs

• Neural Network Models
• Soft sensors (values are calculated)
• Discrete-time dynamic models
 Develop the Transfer Functions
Tr’/T1’ and Tr’/Q’
Assume
Liquid A 1. Perfect mixing
2. Equal inlet and outlet flows
1
3. Density and heat capacity are constant
4. Negligible heat losses
T1 5. Instantaneous heating
w1 Tr
ρ Energy Balance
Vr 𝑑 𝑉𝑟 𝜌𝐶 𝑇𝑟 − 𝑇𝑟𝑒𝑓
T3 = Δ 𝑤𝐶(𝑇 − 𝑇𝑟𝑒𝑓 + 𝑄
𝑑𝑡
w3
Mass Balance
𝑑 𝜌𝑉𝑟
3 = 𝑤1 − 𝑤3
𝑑𝑡
Heating Element Energy Balance
Product 𝑚𝑒 𝐶𝑒
𝑑𝑇𝑒
= 𝑄 − ℎ𝑒 𝐴𝑒 𝑇𝑒 − 𝑇
𝑑𝑡

Q Instead of knowing the heat,

we know the heating element convection
Develop the Transfer Functions
Q4’/Q1’ and Q4’/Q3’

1 3

2 4
 Find CA2’(s)/CA1’(s) and
CA2’(s)/Q’(s)
A -> B
First order kinetics
1
Recall Arrhenius Equation k = Ae-Ea/RT