Process Dynamics & Control

Complete Tutorial Solutions

Detailed Solutions with Step-by-Step Reasoning

Contents

1 Tutorial 1: Feedback & Feedforward Control Concepts 2

1.1 Problem 1: True/False Statements on Control . . . . . . . . . . . . . . . . . . . . 2
1.2 Problem 2: Distillation Column Control . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Problem 3: Heat Exchanger Control . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Tutorial 2: Dynamic Modeling 6

2.1 Problem 4: Stirred-Tank Mixing (Constant Volume) . . . . . . . . . . . . . . . . 6
2.2 Problem 1 (Page 5): Two Interacting Tanks . . . . . . . . . . . . . . . . . . . . . 7
2.3 Problem 2 (Page 6): Sealed Tank with Gas . . . . . . . . . . . . . . . . . . . . . 8
2.4 Problem 3 (Page 7): Two-Compartment Reactor . . . . . . . . . . . . . . . . . . 9

3 Tutorial 3: Laplace Transforms 11

3.1 Problem 1 (Page 8): Pulse Function . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2 Problem 2 (Page 9): Functions of Time in ODE Solution . . . . . . . . . . . . . . 12
3.3 Problem 3 (Page 10): Staircase Function Laplace Transform . . . . . . . . . . . . 13

4 Tutorial 4: Inverse Laplace Transforms (PFE) 14

4.1 Problem 1 (Page 11): Inverse Laplace Transforms . . . . . . . . . . . . . . . . . . 14
4.2 Problem 2 (Page 12): Integro-Differential Equation . . . . . . . . . . . . . . . . . 16
4.3 Problem 3 (Page 13): Second-Order ODE – Functions of Time . . . . . . . . . . 17

5 Tutorial 5: Transfer Functions 18

5.1 Problem 1 (Page 14): First-Order Transfer Function . . . . . . . . . . . . . . . . 18
5.2 Problem 2 (Page 15): Pressure Sensor Alarm . . . . . . . . . . . . . . . . . . . . 19
5.3 Problem 3 (Page 16): First-Order Plus Dead Time . . . . . . . . . . . . . . . . . 20

6 Tutorial 6: State-Space Models & Transfer Functions 21

6.1 Problem 1 (Page 17): Two Interacting Tanks – State Space . . . . . . . . . . . . 21
6.2 Problem 2 (Page 18): Tank with Square-Root Outflow . . . . . . . . . . . . . . . 23

7 Tutorial 7: First-Order System Response 24

7.1 Problem 1 (Page 19): Stirred-Tank Heater . . . . . . . . . . . . . . . . . . . . . . 24
7.2 Problem 2 (Page 20): Chemostat Linearization . . . . . . . . . . . . . . . . . . . 25

8 Tutorial 8: Higher-Order Transfer Functions 27

8.1 Problem 1 (Page 21): Transfer Function Analysis . . . . . . . . . . . . . . . . . . 27
8.2 Problem 2 (Page 22): Thermometer with Ramp Input . . . . . . . . . . . . . . . 28
8.3 Problem 3 (Page 23): Critically Damped Second-Order System . . . . . . . . . . 29

1
Process Dynamics & Control – Tutorial Solutions 2

Tutorial 1: Feedback & Feedforward Control Concepts

Problem 1: True/False Statements on Control
(a) “Feedforward and feedback control require a measured variable.”

Answer
TRUE. Both feedforward and feedback control strategies rely on measurements. In
feedback control, the controlled variable (output) is measured and compared to the set
point. In feedforward control, the disturbance variable is measured before it affects
the process. Without measurements, neither strategy can function.

(b) “For feedforward control, the measured variable is the variable to be controlled.”

Answer
FALSE. In feedforward control, the measured variable is the disturbance (the input
disturbance entering the process), not the controlled variable. The idea of feedforward
is to measure the disturbance before it affects the output and take preemptive corrective
action. It is in feedback control where the controlled variable itself is measured.

(c) “Feedback control theoretically can provide perfect control if the process model used to
design the control system is perfect.”

Answer
FALSE. Even with a perfect process model, feedback control cannot provide perfect
control (zero deviation from set point) because it is inherently reactive—it requires an
error (deviation from set point) to exist before corrective action is taken. There will
always be a transient deviation after a disturbance occurs.
It is feedforward control that can theoretically provide perfect control if the process
model and disturbance model are perfect, because it acts before the disturbance affects
the output.

(d) “Feedback control takes corrective action for all types of process disturbances, both known
and unknown.”

Answer
TRUE. This is one of the major advantages of feedback control. Since it measures the
actual output (controlled variable), any disturbance—whether known, unknown, mea-
sured, or unmeasured—that causes a deviation from the set point will be detected and
corrected. The controller does not need to know the source or nature of the disturbance.

(e) “Feedback control is superior to feedforward control.”

Process Dynamics & Control – Tutorial Solutions 3

Answer
FALSE. Neither is universally superior. Each has advantages and disadvantages:
Feedback advantages: handles unknown disturbances, does not require a process model,
simpler to implement.
Feedforward advantages: acts before the disturbance affects the output (proactive), can
theoretically achieve perfect control.
In practice, the best strategy is often a combined feedforward-feedback control sys-
tem that leverages the strengths of both.
Process Dynamics & Control – Tutorial Solutions 4

Problem 2: Distillation Column Control

Given: A binary distillation column with feed F (mole fraction z), distillate D (mole fraction
y), bottoms B (mole fraction x), and reflux R. The objective is to control distillate composition
y despite disturbances in feed flow rate F .
(a) Feedback Control:

Answer
In feedback control, we measure the controlled variable y (distillate composition) using
the composition analyzer and compare it with the set point ysp .
Strategy: Measure y → Compare with ysp → Controller adjusts the reflux flow rate R
(manipulated variable).
Control loop:

1. The composition analyzer measures y.

2. The controller computes the error: e = ysp − y.

3. Based on the error, the controller adjusts the reflux flow rate R via a control valve.

4. Increasing R increases the purity of the distillate (increases y); decreasing R de-
creases y.

(b) Feedforward Control:

Answer
In feedforward control, we measure the disturbance variable F (feed flow rate) and take
preemptive action before the disturbance affects y.
Strategy: Measure F → Feedforward controller computes required adjustment → Ad-
just reflux flow rate R.
Control loop:

1. A flow sensor measures the feed flow rate F .

2. The feedforward controller, using a process model, computes the required change
in R to compensate for the change in F .

3. The control valve on the reflux line is adjusted accordingly.

The feedforward controller transfer function is derived from the process model relating
F to y and R to y.
Process Dynamics & Control – Tutorial Solutions 5

Problem 3: Heat Exchanger Control

Given: Heat exchanger with process stream (flow F , inlet temperature Ti , outlet temperature
T ) heated by saturated steam (flow Fst ). Objective: keep T = Tsp .
(i) Variable Identification:

Answer
ˆ Controlled Variable (CV): Outlet temperature of the process stream, T

ˆ Manipulated Variable (MV): Steam flow rate, Fst (adjusted via the steam
control valve)

ˆ Disturbance Variables (DVs):

– Process stream flow rate, F

– Process stream inlet temperature, Ti
– Steam supply pressure/temperature variations

(ii) Feedback and Feedforward Control Loops:

Answer
Feedback Control Loop:

1. Temperature sensor measures outlet temperature T .

2. Controller compares T with set point Tsp , computes error e = Tsp − T .

3. Controller output adjusts the steam valve to change Fst .

4. If T < Tsp , increase Fst ; if T > Tsp , decrease Fst .

Feedforward Control Loop:

1. Measure disturbance variables: flow rate F and/or inlet temperature Ti .

2. Feedforward controller uses a process model to compute the required steam flow
rate Fst to maintain T = Tsp .

3. Adjust steam valve preemptively when F or Ti changes.

Process Dynamics & Control – Tutorial Solutions 6

Tutorial 2: Dynamic Modeling

Problem 4: Stirred-Tank Mixing (Constant Volume)
Given: Constant-volume tank with two input streams (same liquid). Stream i has temperature
Ti and mass flow rate wi . Liquid properties are constant.
(a) Dynamic Model Derivation:
Overall Mass Balance:
dm
= w1 + w2 − w3 (1)
dt
Since the tank volume V is constant and density ρ is constant:
dm
m = ρV = constant =⇒ =0
dt
Therefore:
w1 + w2 = w3 (2)
Energy Balance: Assuming no heat losses, no shaft work, and negligible kinetic/potential
energy:
d(mCT3 )
= w1 CT1 + w2 CT2 − w3 CT3 (3)
dt
where C is the specific heat capacity (constant) and T3 is the outlet (= tank) temperature
(perfectly stirred).
Since m = ρV is constant and C is constant:
dT3
ρV C = w1 CT1 + w2 CT2 − w3 CT3 (4)
dt
Dividing by C:
dT3
ρV = w1 T1 + w2 T2 − w3 T3 (5)
dt
Degrees of Freedom Analysis:

ˆ Variables: w1 , w2 , w3 , T1 , T2 , T3 (6 variables)

ˆ Equations: 2 (mass balance + energy balance)

ˆ Known inputs (from upstream): w1 , T1 , w2 , T2 (4 known functions of time)

ˆ Unknowns: w3 , T3 (2 unknowns)

ˆ Degrees of freedom: 6 − 2 − 4 = 0 ✓ (system is exactly specified)

(b) Simplified Model:

Substituting w3 = w1 + w2 into the energy balance:
dT3
ρV = w1 T1 + w2 T2 − (w1 + w2 )T3 (6)
dt

dT3
ρV = w1 (T1 − T3 ) + w2 (T2 − T3 ) (7)
dt
This is a single first-order ODE in T3 (t) with known time-varying inputs w1 (t), w2 (t), T1 (t), T2 (t).
Process Dynamics & Control – Tutorial Solutions 7

Problem 1 (Page 5): Two Interacting Tanks

Given: Two tanks connected with flow between them at the bottom. Tank cross-sections A1 ,
A2 . Linear valves with resistances R2 and R3 . Input flow w1 into Tank 1.
(a) Dynamic Model:
Mass balance on Tank 1:
dh1
A1 ρ = w1 − w2 − w3 (8)
dt
Mass balance on Tank 2:
dh2
A2 ρ = w2 (9)
dt
Valve equations (linear valves):
For the valve between tanks (w2 is positive from Tank 1 to Tank 2):

ρ(h1 − h2 )
w2 = (10)
R2
For the exit valve from Tank 1:
ρ h1
w3 = (11)
R3
Answer
The complete model is:

dh1 ρ(h1 − h2 ) ρ h1
A1 ρ = w1 − − (12)
dt R2 R3
dh2 ρ(h1 − h2 )
A2 ρ = (13)
dt R2
ρ(h1 − h2 )
w2 = (14)
R2
ρ h1
w3 = (15)
R3

(b) Degrees of Freedom:

Answer
ˆ Variables: h1 , h2 , w1 , w2 , w3 (5 variables)

ˆ Equations: 4 (two mass balances + two valve equations)

ˆ Input variable: w1 (1 – must be specified as a function of time)

ˆ Output variables: h1 , h2 , w2 , w3 (4 outputs to be determined)

ˆ DOF: 5 − 4 = 1, which is satisfied by specifying w1 (t). ✓

Process Dynamics & Control – Tutorial Solutions 8

Problem 2 (Page 6): Sealed Tank with Gas

Given: Sealed tank with non-condensible gas above liquid. Cross-section A, height H, liquid
level h. Gas pressure Pg , ambient pressure Pa . Input flow qi , output through valve with square-
root relation.
Derivation:
Mass balance on liquid:
dh
= qi − q
A (16)
dt
Valve equation (square-root relation):
p
q = Cv Pg + ρgh − Pa (17)

where the driving force is the total pressure at the bottom minus ambient pressure.
Gas pressure (ideal gas, isothermal):
mg
Pg V g = RT = constant (18)
M
The gas volume is Vg = A(H − h), so:

mg RT ng RT
Pg = = (19)
M · A(H − h) A(H − h)

where ng = mg /M is the constant number of moles of gas.

Answer
Complete unsteady-state model:
s
dh ng RT
A = qi − C v + ρgh − Pa (20)
dt A(H − h)

ng RT
Pg = (21)
A(H − h)

Is operation independent of Pa ?
For the sealed tank: No, the operation is not independent of Pa . The outlet flow
depends on the pressure difference (Pg + ρgh − Pa ), so Pa directly affects the outflow rate
and hence the dynamic behavior.
For a system open to the atmosphere: The gas pressure Pg = Pa at the liquid sur-
face. The driving force becomes Pa + ρgh − Pa = ρgh. Thus the operation is independent
of Pa when the tank is open.
Process Dynamics & Control – Tutorial Solutions 9

Problem 3 (Page 7): Two-Compartment Reactor

Given: Two-compartment stirred-tank reactor. Feed: q0 , Ti . Compartment 1: volume V1 ,
temperature T1 . Compartment 2: volume V2 , temperature T2 , with cooling coil at Tc . The wall
allows heat transfer; reactor is insulated.
Assumptions: perfectly mixed, ρ and C constant, no heat losses to ambient.
(a) Dynamic Model (no reaction):
Note from the figure: Flow goes q0 → Compartment 1 → (via q1 ) → Compartment 2 → q2 .
Since both are completely full at all times, q0 = q1 = q2 = q (volumetric flow rates are equal).
Energy balance on Compartment 1:
dT1
ρCV1 = ρCq0 (Ti − T1 ) + U Aw (T2 − T1 ) (22)
dt
where U is the overall heat transfer coefficient and Aw is the area of the separating wall.
Energy balance on Compartment 2:
dT2
ρCV2 = ρCq0 (T1 − T2 ) − U Aw (T2 − T1 ) − Uc Ac (T2 − Tc ) (23)
dt
where Uc and Ac are the heat transfer coefficient and area for the cooling coil.

Answer

dT1
ρCV1 = ρCq0 (Ti − T1 ) + U Aw (T2 − T1 ) (24)
dt
dT2
ρCV2 = ρCq0 (T1 − T2 ) − U Aw (T2 − T1 ) − Uc Ac (T2 − Tc ) (25)
dt
Two ODEs, two unknowns (T1 , T2 ) for given inputs q0 (t), Ti (t), Tc (t).

(b) Degrees of Freedom:

Answer
ˆ Parameters: ρ, C, V1 , V2 , U, Aw , Uc , Ac (8 parameters)

ˆ Outputs: T1 , T2 (2 unknowns)

ˆ Inputs (known functions of time): q0 (t), Ti (t), Tc (t) (3 inputs)

ˆ Equations: 2 (energy balances)

ˆ DOF = 2 − 2 = 0 (with 3 inputs specified) ✓

(c) Modification for exothermic reaction A → B:

Process Dynamics & Control – Tutorial Solutions 10

Answer
We must add:

1. A reaction rate term to each energy balance: (−∆Hr )rj Vj where rj =

k0 e−E/RTj CA,j is the rate of reaction in compartment j.

2. Component material balances for species A in each compartment:

dCA,1
V1 = q0 (CA,i − CA,1 ) − r1 V1 (26)
dt
dCA,2
V2 = q0 (CA,1 − CA,2 ) − r2 V2 (27)
dt

3. The modified energy balances become:

dT1
ρCV1 = ρCq0 (Ti − T1 ) + U Aw (T2 − T1 ) + (−∆Hr )r1 V1 (28)
dt
dT2
ρCV2 = ρCq0 (T1 − T2 ) − U Aw (T2 − T1 ) − Uc Ac (T2 − Tc ) + (−∆Hr )r2 V2 (29)
dt
Process Dynamics & Control – Tutorial Solutions 11

Tutorial 3: Laplace Transforms

Problem 1 (Page 8): Pulse Function
Given: A pulse u(t) starts at height h at t = 0 and decreases linearly with slope −a until
u(tw ) = 0.
(a) Pulse width tw :
At t = 0: u(0) = h. The function decreases linearly with slope −a:

u(t) = h − at

Setting u(tw ) = 0:
h
h − atw = 0 =⇒ tw =
a

(b) Express u(t) as sum of simpler functions:

The pulse can be decomposed as:

u(t) = h · S(t) − a · t · S(t) − [h · S(t − tw ) − a · (t − tw ) · S(t − tw ) − atw · S(t − tw )] (30)

More cleanly, using unit step S(t) and ramp:

u(t) = (h − at) S(t) − (h − at) S(t − tw ) (31)

Or equivalently:

u(t) = h S(t) − a t S(t) − [−a(t − tw ) S(t − tw )]

u(t) = h S(t) − a t S(t) + a (t − tw ) S(t − tw )

Taking the Laplace transform:

h a a
U (s) = − 2 + 2 e−tw s (32)
s s s
Process Dynamics & Control – Tutorial Solutions 12

Problem 2 (Page 9): Functions of Time in ODE Solution

d2 y dy
Given: + 7 + 12y = 5u(t), with y(0) = y ′ (0) = 0.
dt2 dt
The characteristic equation is: s2 + 7s + 12 = (s + 3)(s + 4) = 0, giving roots s = −3 and
s = −4.
(a) u(t) = b e−2t :
b
U (s) = s+2 . The Laplace transform of y is:

5b
Y (s) =
(s + 2)(s + 3)(s + 4)

By partial fractions, the terms correspond to poles at s = −2, −3, −4.

Answer

Functions of time in y(t): e−2t , e−3t , e−4t

(b) u(t) = ct:

U (s) = sc2 . The Laplace transform of y is:

5c
Y (s) =
s2 (s + 3)(s + 4)

Poles at s = 0 (double), s = −3, s = −4.

By partial fractions: Y (s) = As + sB2 + C
s+3 + D
s+4

Answer

Functions of time in y(t): 1 (constant), t, e−3t , e−4t

Process Dynamics & Control – Tutorial Solutions 13

Problem 3 (Page 10): Staircase Function Laplace Transform

Given: From the figure, f (t) is:

ˆ 0 ≤ t < 1: f (t) = 1

ˆ 1 ≤ t < 2: f (t) = 2

ˆ 2 ≤ t < 3: f (t) decreases linearly from 2 to 0 (slope = −2)

ˆ t ≥ 3: f (t) = 0

Decomposition using step and ramp functions:

1 1
f (t) = 1 · S(t) + 1 · S(t − 1) − 2(t − 2) · S(t − 2) · + 2(t − 3) · S(t − 3) · − 2 · S(t − 3) · 0
1 1
(33)

Let me be more careful. Decompose step by step:

f (t) = S(t) + S(t − 1) − 2(t − 2) S(t − 2) + 2(t − 3) S(t − 3) (34)

Verification:

ˆ 0 < t < 1: f = 1 + 0 − 0 + 0 = 1 ✓

ˆ 1 < t < 2: f = 1 + 1 − 0 + 0 = 2 ✓

ˆ t = 2.5: f = 1 + 1 − 2(0.5) + 0 = 1 ✓(should be 2 − 2(0.5) = 1, the line from 2 to 0)

ˆ t = 3: f = 1 + 1 − 2(1) + 0 = 0 ✓

ˆ t > 3 (e.g., t = 4): f = 1 + 1 − 2(2) + 2(1) = 2 − 4 + 2 = 0 ✓

Laplace Transform:

1 e−s 2e−2s 2e−3s

F (s) = + − + (35)
s s s2 s2
Process Dynamics & Control – Tutorial Solutions 14

Tutorial 4: Inverse Laplace Transforms (PFE)

Problem 1 (Page 11): Inverse Laplace Transforms
s(s + 1)
(a) X(s) =
(s + 2)(s + 3)(s + 4)
Using partial fraction expansion:

s(s + 1) A B C
= + +
(s + 2)(s + 3)(s + 4) s+2 s+3 s+4

s(s+1) (−2)(−1) 2
A= (s+3)(s+4) s=−2 = (1)(2) = 2 =1
s(s+1) (−3)(−2) 6
B= (s+2)(s+4) s=−3 = (−1)(1) = −1 = −6
s(s+1) (−4)(−3) 12
C= (s+2)(s+3) s=−4 = (−2)(−1) = 2 =6

Answer

x(t) = e−2t − 6e−3t + 6e−4t

s+2
(b) X(s) =
(s + 1)2
Repeated root at s = −1:
s+2 A B
2
= +
(s + 1) s + 1 (s + 1)2

s+2
B = (s + 1)2 · (s+1)2 s=−1
= (−1 + 2) = 1
For A: Multiply both sides by (s + 1)2 : (s + 2) = A(s + 1) + B
Setting s = 0: 2 = A(1) + 1 =⇒ A = 1.

Answer
1 1
X(s) = +
s + 1 (s + 1)2
x(t) = e−t + t e−t

1
(c) X(s) =
s2
+s+1 √ √
−1± 1−4 −1±j 3
The denominator has complex roots: s = 2 = 2
√
So σ = 12 , ω = 23 .

2
Complete the square: s2 + s + 1 = s + 21 + 3
4
√
3
1 2 2
X(s) =  √  2 = √ ·  √ 2

2
s + 12 + 23 3 s + 1
2 + 3
2 2
Process Dynamics & Control – Tutorial Solutions 15

Answer
√ !
2 3
x(t) = √ e−t/2 sin t
3 2

(s + 1)
(d) X(s) = e−0.5s
s(s + 4)(s + 3)
s+1
First find the inverse of s(s+4)(s+3) (without the delay):

s+1 A B C
= + +
s(s + 4)(s + 3) s s+4 s+3

s+1 1
A= (s+4)(s+3) s=0 = 12
−3
s+1
B = s(s+3) = (−4)(−1) = −3 3
4 = −4
s=−4
s+1 −2
C = s(s+4) = (−3)(1) = 32
s=−3
Let g(t) = 121
− 34 e−4t + 32 e−3t .
The e −0.5s factor gives a time delay of 0.5:

Answer
 
1 3 2 −3(t−0.5)
x(t) = − e−4(t−0.5) + e S(t − 0.5)
12 4 3
where S(t − 0.5) is the unit step function shifted to t = 0.5.
Process Dynamics & Control – Tutorial Solutions 16

Problem 2 (Page 12): Integro-Differential Equation

Z t
d2 x dx
Given: + 4 + 3x = 2 e−τ dτ , with x(0) = x′ (0) = 0.
dt2 dt 0
(a) Find x(t):
Taking Laplace transforms (zero initial conditions):
1 1
s2 X(s) + 4s X(s) + 3X(s) = 2 · ·
s s+1
Rt F (s)
(Using L{ 0 f (τ ) dτ } = s and L{e−t } = 1
s+1 .)

2
(s2 + 4s + 3) X(s) =
s(s + 1)
2 2
X(s) = =
s(s + 1)(s + 1)(s + 3) s(s + 1)2 (s + 3)
Note: s2 + 4s + 3 = (s + 1)(s + 3).
Partial fraction expansion:
2 A B C D
= + + +
s(s + 1)2 (s + 3) s s + 1 (s + 1)2 s + 3

2 2
A= (s+1)2 (s+3) s=0
= 1·3 = 32
2 2 2
D= s(s+1)2 s=−3
= (−3)(−2) 2 = −12 = − 16
2 2
C= s(s+3) s=−1 = (−1)(2) = −1
For B: Multiply through by s(s + 1)2 (s + 3):

2 = A(s + 1)2 (s + 3) + Bs(s + 1)(s + 3) + Cs(s + 3) + Ds(s + 1)2

Comparing coefficients of s3 : 0 = A + B + D
2 1 4 1 1
B = −A − D = − + = − + = −
3 6 6 6 2

Answer

2/3 1/2 1 1/6

X(s) = − − 2
−
s s + 1 (s + 1) s+3
2 1 −t 1
x(t) = − e − te−t − e−3t
3 2 6

(b) Limiting value as t → ∞:

As t → ∞, all exponential terms vanish:

Answer

2
lim x(t) =
t→∞ 3
2
This can be verified using the Final Value Theorem: lims→0 sX(s) = lims→0 (s+1)2 (s+3)
=
2 2
1·3 = 3 ✓
Process Dynamics & Control – Tutorial Solutions 17

Problem 3 (Page 13): Second-Order ODE – Functions of Time

d2 y dy
Given: τ1 τ2 + (τ1 + τ2 ) + y = K u(t)
dt2 dt
The characteristic equation is τ1 τ2 s2 +(τ1 +τ2 )s+1 = 0, which factors as (τ1 s+1)(τ2 s+1) = 0.
Roots: s = −1/τ1 and s = −1/τ2 . Homogeneous solutions: e−t/τ1 and e−t/τ2 .
(a) u(t) = a S(t) (step of magnitude a):
U (s) = a/s. Poles of Y (s): s = 0, −1/τ1 , −1/τ2 .

Answer

Functions of time: 1 (constant), e−t/τ1 , e−t/τ2

(b) u(t) = b e−t/τ , τ ̸= τ1 ̸= τ2 :

U (s) = b/(s + 1/τ ). Poles: s = −1/τ, −1/τ1 , −1/τ2 .

Answer

Functions of time: e−t/τ , e−t/τ1 , e−t/τ2

(c) u(t) = c e−t/τ , τ = τ1 ̸= τ2 :

U (s) = c/(s + 1/τ1 ). This creates a repeated root at s = −1/τ1 .

Answer

Functions of time: e−t/τ1 , t e−t/τ1 , e−t/τ2

(d) u(t) = d sin(ωt), τ1 ̸= τ2 :

U (s) = dω/(s2 + ω 2 ). Poles: s = ±jω, −1/τ1 , −1/τ2 .

Answer

Functions of time: sin(ωt), cos(ωt), e−t/τ1 , e−t/τ2

(or equivalently sin(ωt + ϕ) for the sinusoidal part)
Process Dynamics & Control – Tutorial Solutions 18

Tutorial 5: Transfer Functions

Problem 1 (Page 14): First-Order Transfer Function
Y (s) d
Given: =
U (s) bs + c
(a) Steady-state gain:
The steady-state gain is G(0):

d d
Kss = G(0) = = (36)
b(0) + c c

(b) Bounded output for step input?

For a step of magnitude M : U (s) = M/s.

dM dM
Y (s) = =
s(bs + c) b · s(s + c/b)

The system has a pole at s = −c/b. The output is:

dM  
y(t) = 1 − e−(c/b)t
c

Answer
The output is bounded only if c/b > 0 (i.e., b and c have the same sign), so that the
exponential decays. If c/b < 0, the pole is in the right-half plane and y(t) → ∞ as t → ∞.
Therefore, the answer is No—the output is not bounded for all values of b, c, and d. It
requires c/b > 0 for stability.
Process Dynamics & Control – Tutorial Solutions 19

Problem 2 (Page 15): Pressure Sensor Alarm

Pm′ (s) 1
Given: ′
= , τ = 10 s.
P (s) 10s + 1
Initial: Pm = P = 35 psi. Step change: P goes from 35 to 50 psi at 1:30 PM. Alarm sounds
when Pm > 45 psi.
Solution:
In deviation variables: P ′ = 50 − 35 =
15 psi (step change).
The response: Pm ′ (t) = 15 1 − e−t/10

In actual variables: Pm (t) = 35 + 15 1 − e−t/10

Setting Pm = 45:  
45 = 35 + 15 1 − e−t/10
  2
10 = 15 1 − e−t/10 =⇒ 1 − e−t/10 =
3
 
1 t 1
e−t/10 = =⇒ − = ln = − ln 3
3 10 3
t = 10 ln 3 = 10 × 1.0986 = 10.99 seconds

Answer
t ≈ 11.0 seconds after the step change.
The alarm will sound at approximately 1:30:11 PM (about 11 seconds after 1:30 PM).
Process Dynamics & Control – Tutorial Solutions 20

Problem 3 (Page 16): First-Order Plus Dead Time

Y (s) 2e−s
Given: G(s) = =
U (s) 5s + 1
(a) Steady-state gain:
2e0
K = G(0) = = 2
5(0) + 1
(b) Time constant:
τ =5
(c) U (s) = 4/s, find y(∞):
By the Final Value Theorem:

2e−s 4 2·4
y(∞) = lim s · G(s) · U (s) = lim s · · = = 8
s→0 s→0 5s + 1 s 1
(d) For U (s) = 4/s, find y(10) and fraction of new steady state:
2e−s 4 8e−s
Y (s) = 5s+1 · s = s(5s+1)
y(t) = 8 1 − e−(t−1)/5 S(t − 1) (accounting for dead time of 1).

At t = 10:
 
y(10) = 8 1 − e−9/5 = 8 1 − e−1.8 = 8(1 − 0.1653) = 8 × 0.8347 = 6.678

Fraction of steady state: 6.678

8 = 0.835 (83.5%)
(e) U (s) = (1 − e−s )/s (unit rectangular pulse), find y(∞):
By the FVT:
2e−s 1 − e−s 2e−s (1 − e−s )
y(∞) = lim s · · = lim
s→0 5s + 1 s s→0 5s + 1
As s → 0: e−s → 1 and (1 − e−s ) → 0. We need L’Hôpital’s or series expansion. Since
1 − e−s ≈ s for small s:
2·1·s
y(∞) = lim = 0
s→0 1
The output returns to zero because the pulse is temporary.
(f ) u(t) = δ(t), U (s) = 1, find y(∞):
2e−s −s
Y (s) = 5s+1 . By FVT: y(∞) = lims→0 2se
5s+1 = 0.

Answer

y(∞) = 0 for the impulse input. The system is stable (pole at s = −1/5), so the output
decays to zero.
Process Dynamics & Control – Tutorial Solutions 21

Tutorial 6: State-Space Models & Transfer Functions

Problem 1 (Page 17): Two Interacting Tanks – State Space
Given: Two interacting tanks. Data: A1 = 1, A2 = 0.5, R1 = 0.5, R2 = 2, R3 = 1.
(From: A1 = 2A2 = 1 =⇒ A2 = 0.5; 4R1 = R2 = 2R3 = 2 =⇒ R1 = 0.5, R2 = 2, R3 = 1.)
Mass balances (linearized):
Tank 1:
dh1 h1 − h2 h1
A1 = q1 − −
dt R1 R3
Tank 2:
dh2 h1 − h2 h2
A2 = q2 + −
dt R1 R2
Substituting values:
dh1 h1 − h2 h1
= q1 − − = q1 − 2(h1 − h2 ) − h1 = q1 − 3h1 + 2h2 (37)
dt 0.5 1
dh2 h1 − h2 h2
0.5 = q2 + − = q2 + 2(h1 − h2 ) − 0.5h2 (38)
dt 0.5 2
dh2
= 2q2 + 4h1 − 4h2 − h2 = 2q2 + 4h1 − 5h2 (39)
dt
State-space
  form: 
ẋ = Ax + Bu, y = Cx
h1 q
With x = , u = 1 , y = h1 − h2 :
h2 q2

Answer
   
−3 2 1 0  
A= , B= , C = 1 −1
4 −5 0 2

Transfer function matrix: G(s) = C(sI − A)−1 B

 
s + 3 −2
sI − A =
−4 s + 5
det(sI − A) = (s + 3)(s + 5) − 8 = s2 + 8s + 15 − 8 = s2 + 8s + 7 = (s + 1)(s + 7)
 
−1 1 s+5 2
(sI − A) =
(s + 1)(s + 7) 4 s+3
  
1   s+5 2 1 0
G(s) = 1 −1
(s + 1)(s + 7) 4 s+3 0 2
1 1
C(sI − A)−1 =
   
s + 1 −s − 1 = 1 −1
(s + 1)(s + 7) (s + 7)
 
s+5 2
Wait, let me recompute: [1 −1] = [s + 5 − 4 2 − (s + 3)] = [s + 1 − (s + 1)]
4 s+3
So: C(sI − A)−1 = (s+1)(s+7)
s+1
[1 − 1] = s+71
[1 − 1]
Then:  
1 1 0 1
G(s) = [1 − 1] = [1 − 2]
s+7 0 2 s+7
Process Dynamics & Control – Tutorial Solutions 22

Answer
1 −2
 
G(s) = s+7 s+7

Y (s) 1 Y (s) −2
So: = and =
Q1 (s) s+7 Q2 (s) s+7
where Y = H1 − H2 .
Process Dynamics & Control – Tutorial Solutions 23

Problem 2 (Page 18): Tank with Square-Root Outflow

√
Given: A = 100 cm2 , qout = 3 h. Steady state: q̄i = 18 cm3 /s, h̄ = 36 cm. Tank capacity =
5 L = 5000 cm3 .
Mass balance:
dh √
A = qi − 3 h
dt
√
Linearization: Let f (h) = 3 h. Then:

df 3 3 3 1
= √ = √ = =
dh h̄ 2 h̄ 2 36 12 4

In deviation variables (h′ = h − h̄, qi′ = qi − q̄i ):

dh′ 1
A = qi′ − h′
dt 4
dh′ 1 ′
100 + h = qi′
dt 4
dh ′
400 + h′ = 4qi′
dt
Answer
Transfer function:
H ′ (s) 4 K
′ = =
Qi (s) 400s + 1 τs + 1

where K = 4 cm/(cm3 /s) and τ = 400 s.

Overflow time:
Maximum height: hmax = 5000 ′
100 = 50 cm. So hmax = 50 − 36 = 14 cm.
Step change: qi = 4 cm /s. New steady-state deviation: h′∞ = K · qi′ = 4 × 4 = 16 cm.
′ 3

Response: h′ (t) = 16(1 − e−t/400 )

Setting h′ (t) = 14:
14 7
14 = 16(1 − e−t/400 ) =⇒ 1 − e−t/400 = =
16 8
 
−t/400 1 1
e = =⇒ t = −400 ln = 400 ln 8 = 400 × 2.0794
8 8
Answer

t ≈ 832 seconds ≈ 13.9 minutes

Process Dynamics & Control – Tutorial Solutions 24

Tutorial 7: First-Order System Response

Problem 1 (Page 19): Stirred-Tank Heater
Given: w = 800 kg/h, V = 1.6 m3 , ρ = 980 kg/m3 , C = 4.2 kJ/(kg·°C).
Ti = 110°C, T = 140°C, Q̄ = 34.89 kW.
1/(wC)
Transfer function: T ′ (s) = (m/w)s+1 Q′ (s) + (m/w)s+1
1
Ti′ (s)
(a) Values of K and τ :
m = ρV = 980 × 1.6 = 1568 kg
K = wC1 1
= 800×4.2 1
= 3360 °C·h/kJ = 1.071 °C/kW (converting units appropriately)
m ρV 1568
τ = w = w = 800 = 1.96 h

Answer

K = 1.071 °C/kW, τ = 1.96 h

1.071 1
Transfer function: T ′ (s) = Q′ (s) + T ′ (s)
1.96s + 1 1.96s + 1 i

(b) Time for 99% response to 30% heater increase:

Step change: Q′ = 0.30 × 34.89 = 10.467 kW.
For a first-order system, 99% response occurs at t = 5τ (since 1 − e−5 ≈ 0.9933).
Actually, solving 1 − e−t/τ = 0.99:

e−t/τ = 0.01 =⇒ t = −τ ln(0.01) = τ × 4.605

t = 1.96 × 4.605 = 9.03 h

Answer

t ≈ 9.03 hours
New steady-state temperature change: ∆T = K × Q′ = 1.071 × 10.467 = 11.21°C.
Final temperature: T = 140 + 11.21 = 151.21°C.

(c) Time for outlet to go from 140 to 145°C when Ti jumps from 110 to 130°C:
Ti′ = 130 − 110 = 20°C (step change in inlet temperature).
Response: T ′ (t) = 20(1 − e−t/1.96 ) (gain from Ti to T is 1 for a first-order system).
New steady state: T = 140 + 20 = 160°C.
We need T ′ = 145 − 140 = 5°C:

5 = 20(1 − e−t/1.96 ) =⇒ 1 − e−t/1.96 = 0.25

e−t/1.96 = 0.75 =⇒ t = −1.96 ln(0.75) = 1.96 × 0.2877

Answer

t ≈ 0.564 h ≈ 33.8 minutes

Process Dynamics & Control – Tutorial Solutions 25

Problem 2 (Page 20): Chemostat Linearization

Given:

Ẋ = µ(S)X − DX (40)
Ṡ = −µ(S)X/YX/S + D(Sf − S) (41)

with µ(S) = Kµsm+S

S
.
Parameters: µm = 0.20 h−1 , Ks = 1.0 g/L, YX/S = 0.5 g/g.
Steady state: D̄ = 0.1 h−1 , X̄ = 4.5 g/L, S̄ = 1.0 g/L, S̄f = 10 g/L.
Linearization of Ẋ = µ(S)X − DX:
Let f1 (X, S, D) = µ(S)X − DX.
Partial derivatives at steady state:
∂f1 0.2 × 1
= µ(S̄) − D̄ = − 0.1 = 0.1 − 0.1 = 0 (42)
∂X 1+1
∂f1 dµ µm Ks 0.2 × 1 0.9
= X̄ = 2
X̄ = 2
× 4.5 = = 0.225 (43)
∂S dS S̄ (Ks + S̄) (2) 4
∂f1
= −X̄ = −4.5 (44)
∂D
Linearized: Ẋ ′ = 0 · X ′ + 0.225 S ′ − 4.5 D′
Linearization of Ṡ = −µ(S)X/YX/S + D(Sf − S):
Let f2 (X, S, D) = −µ(S)X/YX/S + D(Sf − S).

∂f2 µ(S̄) 0.1

=− =− = −0.2 (45)
∂X YX/S 0.5
∂f2 dµ X̄ 0.225 1
=− − D̄ = − · − 0.1 = −0.45 − 0.1 = −0.55 (46)
∂S dS YX/S 0.5 1
∂f2 dµ
Wait, let me redo this carefully. ∂S = −Y 1 dS X̄ − D̄
X/S
1 0.2×1
= − 0.5
× 4× 4.5 − 0.1 = −2 × 0.225 − 0.1 = −0.45 − 0.1 = −0.55
∂f2
∂D = S̄f − S̄ = 10 − 1 = 9
Linearized system:
 ′ 
0.225 X ′
   
Ẋ 0 −4.5
= + D′
Ṡ ′ −0.2 −0.55 S ′ 9

Taking Laplace transforms (zero ICs):

sX ′ (s) = 0.225 S ′ (s) − 4.5 D′ (s) (47)

′ ′ ′ ′
sS (s) = −0.2 X (s) − 0.55 S (s) + 9 D (s) (48)
′ ′
From eq. 1: X ′ (s) = 0.225 S (s)−4.5
s
D (s)

From eq. 2: S (s)(s + 0.55) = −0.2 X ′ (s) + 9 D′ (s)

′

Substituting:
0.2(0.225 S ′ − 4.5 D′ )
S ′ (s)(s + 0.55) = − + 9D′
s
0.045 S ′ 0.9 D′
S ′ (s + 0.55) + = + 9D′
s s
   
′ 0.045 ′ 0.9
S s + 0.55 + =D +9
s s
Process Dynamics & Control – Tutorial Solutions 26

s2 + 0.55s + 0.045 0.9 + 9s

S′ · = D′ ·
s s
9s + 0.9
S ′ (s) = 2 D′ (s)
s + 0.55s + 0.045
Now substituting back:
 
′ 1 9s + 0.9
X (s) = 0.225 · 2 − 4.5 D′ (s)
s s + 0.55s + 0.045

0.225(9s + 0.9) − 4.5(s2 + 0.55s + 0.045) ′

= D (s)
s(s2 + 0.55s + 0.045)
2.025s + 0.2025 − 4.5s2 − 2.475s − 0.2025 ′
= D (s)
s(s2 + 0.55s + 0.045)
−4.5s2 − 0.45s
= D′ (s)
s(s2 + 0.55s + 0.045)
−4.5s − 0.45
= D′ (s)
s2 + 0.55s + 0.045
−4.5(s + 0.1)
= D′ (s)
s2 + 0.55s + 0.045
Answer

X ′ (s) −4.5(s + 0.1)

′
= 2
D (s) s + 0.55s + 0.045
√
−0.55± 0.3025−0.18
The denominator factors: s2 + 0.55s + 0.045 = 0 =⇒ s = 2 =
√
−0.55± 0.1225 −0.55±0.35
2 = 2
Roots: s = −0.1 and s = −0.45.
Note that the zero at s = −0.1 cancels with one pole, giving the simplified form:

X ′ (s) −4.5 −10 −10

= = =
D′ (s) s + 0.45 (1/0.45)s + 1 2.22s + 1
Process Dynamics & Control – Tutorial Solutions 27

Tutorial 8: Higher-Order Transfer Functions

Problem 1 (Page 21): Transfer Function Analysis
240(1 − 4s)e−3s
Given: G(s) =
18s2 + 21s + 3
(a) Properties:
First, put the denominator in standard form by dividing by 3:

240(1 − 4s)e−3s 80(1 − 4s)e−3s

G(s) = =
3(6s2 + 7s + 1) (6s + 1)(s + 1)

Factoring: 6s2 + 7s + 1 = (6s + 1)(s + 1). Check: 6s2 + 6s + s + 1 = 6s(s + 1) + (s + 1) ✓.

Answer
240(1)
ˆ Gain: G(0) = 3 = 240
3 = 80

ˆ Time delay: θ = 3 (from e−3s )

ˆ Time constants: τ1 = 6, τ2 = 1 (from (6s + 1)(s + 1))

ˆ Poles: s = −1/6 and s = −1

ˆ Zero: From 1 − 4s = 0 =⇒ s = 1/4 (a RHP zero)

(b)(i) Inverse response?

Answer
Yes, the step response will exhibit inverse response. This is because there is a positive
(RHP) zero at s = +1/4. A RHP zero causes the initial response to move in the opposite
direction from the final steady-state value before eventually moving toward it.

(b)(ii) Oscillations?

Answer
No oscillations. The poles are at s = −1/6 and s = −1, both real and distinct. Oscilla-
tory behavior requires complex conjugate poles (i.e., the discriminant of the characteristic
equation must be negative). Here both poles are real, so the response is non-oscillatory
(overdamped).
Process Dynamics & Control – Tutorial Solutions 28

Problem 2 (Page 22): Thermometer with Ramp Input

Given: First-order thermometer: τ = 0.1 min, K = 1. Initial bath temperature: 25°C. Bath
temperature ramp: 1°/min.
′ (s)
Transfer function: TTm′ (s) 1
= 0.1s+1
Input: T ′ (t) = t (ramp of rate 1°/min), so T ′ (s) = 1/s2 .

′ 1
Tm (s) =
(0.1s + 1)s2
1 A B C
Partial fractions: (0.1s+1)s2
= s + s2
+ 0.1s+1
1
B = s2 · 0.1s+1 s=0 = 1
1
C = (0.1s + 1) · s12 s=−10 = 100 = 0.01
1 A 1 0.01
For A: (0.1s+1)s2 = s + s2 + 0.1s+1
Multiply by s, let s → ∞: 0 = A + 0.01/0.1 = A + 0.1, so A = −0.1.

′ 1 −t/0.1
Tm (t) = −0.1 + t + 0.01 · e = (t − 0.1) + 0.1 e−10t
0.1
The actual temperatures:

T (t) = 25 + t (49)
−10t −10t
Tm (t) = 25 + (t − 0.1) + 0.1 e = 24.9 + t + 0.1 e (50)

Difference: T (t) − Tm (t) = 0.1 − 0.1 e−10t = 0.1(1 − e−10t )

(a) Differences at specific times:
At t = 0.1 min:

T − Tm = 0.1(1 − e−1 ) = 0.1(1 − 0.3679) = 0.1 × 0.6321 = 0.0632°C

At t = 1.0 min:

T − Tm = 0.1(1 − e−10 ) ≈ 0.1(1 − 0.0000454) ≈ 0.1°C

Answer

At t = 0.1 min: T − Tm ≈ 0.063°C

At t = 1.0 min: T − Tm ≈ 0.1°C

(b) Maximum deviation:

Answer

The deviation T − Tm = 0.1(1 − e−10t ) is a monotonically increasing function that ap-

proaches 0.1 asymptotically.
The maximum deviation is 0.1°C = τ × (ramp rate) .
This maximum is approached as t → ∞ (i.e., once the transient e−10t has died out). In
practice, the deviation reaches 99% of its maximum at t = 5τ = 0.5 min.
Process Dynamics & Control – Tutorial Solutions 29

Problem 3 (Page 23): Critically Damped Second-Order System

Y (s) K
Given: =
U (s) (τ s + 1)2
(a) 5% settling time for step input of magnitude M :
For a step input: Y (s) = s(τKM
s+1)2
Inverse transform: y(t) = KM 1 − 1 + τt e−t/τ



At 95% of final value: 1 + tτs e−ts /τ = 0.05

Let x = ts /τ : (1 + x)e−x = 0.05

This transcendental equation must be solved numerically. Using iteration or numerical
methods: x ≈ 4.744.

Answer

ts ≈ 4.744 τ
(For comparison, a first-order system has ts ≈ 3τ for 5% settling.)

(b) Time lag for ramp input (K = 1):

For u(t) = at, U (s) = a/s2 :
a
Y (s) = 2
s (τ s + 1)2
The long-time asymptotic behavior (after transients decay): y(t) → a(t − 2τ ) as t → ∞
The input is u(t) = at. The lag is the time by which y(t) trails u(t):

Lag = t − (t − 2τ ) = 2τ

Answer

The output lags behind the input by 2τ .

K
This is a general result: for an nth-order system (τ s+1)n with K = 1 and a ramp input,

the asymptotic time lag is nτ . For our critically damped case (n = 2), the lag is 2τ .