Control Theory and Self-Regulation (TCA - 22134)

Lecture 2: M2.1 Mathematical modeling of dynamic systems

Outline
o Module M2: Mathematical modeling and response analysis of
dynamic systems
• M2.1: Mathematical modeling of dynamic systems
• M2.2: Response analysis of dynamic systems

2
Module M2

Module M2: Mathematical modeling and response

analysis of dynamic systems

Contents
• M2.1: Mathematical modeling of dynamic systems
• M2.2: Response analysis of dynamic systems
3
M.2.1: Mathematical modeling of dynamic systems
o Model (in control engineering): set of differential equations
that describe the dynamic behavior of a process.

Two general methods for obtaining models:

Goal of this 1. By using principles of the underlying physics of the

lecture
process.

2. By testing a prototype, measuring its response to inputs

and using the data to construct the model. (This
approach is sometimes called system identification.)

4
Module M2.2

Module M2.1: Mathematical modeling of

dynamic systems

Contents
• M2.1.1: Dynamics of mechanical systems
• M2.1.2: Modeling of electronic circuits
• M2.1.3: Modeling of heat flow systems

5
M2.1.1: Dynamics of mechanical systems
o Newton's second law is the cornerstone for obtaining the dynamic
equations of any mechanical system:
𝐅 = 𝑚𝐚
vector acceleration of each body
vector sum of all forces applied to with respect to an inertial
each body, newtons (N) reference frame, m/sec2

mass of the body, kg (Note we use here the symbol sec for
seconds, instead of s. This is to avoid
confusion with the s of the Laplace
transform.)
𝑑𝐯 𝑑! 𝐱
𝐚= = ≡ 𝐚 = 𝐯̇ = 𝐱̈
𝑑𝑡 𝑑𝑡 !

6
Example 1: translational motion with friction

We want to calculate the speed (𝑣) of a car. The engine imparts a

force 𝑢 as shown and we assume that the friction retarding the
motion of the car is proportional to the car’s speed with a
proportionally constant b.
𝐅 = 𝑚𝐚

𝑢 − 𝑏𝑥̇ = 𝑚𝑥̈

𝑏 𝑢
𝑥̈ + 𝑥̇ =
𝑚 𝑚

𝑏 𝑢
(free body diagram) 𝑣̇ + 𝑣=
𝑚 𝑚
For computing the solution to differential equations, we will use the Laplace transform (M2.2) which, in essence
assumes that for stimuli (input) of the form 𝑢 = 𝑈! 𝑒 "# the solution (output) is also of the form 𝑣 = 𝑉! 𝑒 "# , therefore,
𝑣̇ = 𝑠𝑉! 𝑒 "# and… 1
𝑏 1 𝑉!
𝑏
𝑣̇ + 𝑣 =
𝑢
𝑠+ 𝑉! 𝑒 "# = 𝑈! 𝑒 "# = 𝑚
𝑚 𝑚 𝑚 𝑚 𝑈! 𝑠 + 𝑏
𝑚 7
Example 1: translational motion with friction
1
𝑉(𝑠) 𝑚 We will later see that this expression of the differential is the transfer
= function and will be used extensively to design feedback controllers.
𝑈(𝑠) 𝑠 + 𝑏
𝑚
𝑏 𝑢
Note that, in effect, we have substituted 𝑠 for 𝑑 ⁄𝑑𝑡 in 𝑣̇ + 𝑣=
𝑚 𝑚

The transfer function characterizes the dynamics of a system and it can be used, for
instance, to prescribe the system to MATLAB to obtain time responses.

(force =500 N, mass = 1000 kg, drag coefficient = 50 N.sec/m)

8
Example 2: translational motion with flexible and damped coupling
spring

(rearrangement)

damper

(𝑑 ⁄𝑑𝑡 → 𝑠)

(free body diagrams) (written in matrix format)

(Cramer’s rule)

Transfer function
(rearrangement)

9
Example 3: simple rotational motion (Satellite attitude model)
Application of Newton's second law to one-dimensional rotational system:

𝑀 = 𝐼𝛼
sum of all external moments about
the center of mass of a body (i.e., Angular acceleration of the body,
rad/sec2
torque magnitude, τ ), N·m

body’s moment of inertia, kg·m2

attitude (orientation)
Transfer function
Disturbance moment
on the satellite (𝑑 ⁄𝑑𝑡 → 𝑠)

Note that the output of this system, 𝜃 , results from

Jets for controlling integrating the sum of the input torques twice; this type of
attitude system is often refereed to as a double-integrator plant

10
Example 4: rotational motion with flexible and damped coupling

𝐼) 𝜃̈) = 𝑇* − b 𝜃̇) − 𝜃̇+ − 𝑘 𝜃) − 𝜃+

𝐼𝛼 = 𝑀
𝐼+ 𝜃̈+ = 0 − b 𝜃̇+ − 𝜃̇) − 𝑘 𝜃+ − 𝜃)

(rearrangement)

𝐼) 𝜃̈) + b 𝜃̇) − 𝜃̇+ + 𝑘 𝜃) − 𝜃+ = 𝑇*

𝐼+ 𝜃̈+ + b 𝜃̇+ − 𝜃̇) + 𝑘 𝜃+ − 𝜃) = 0

(free body diagrams of the satellite and the instrument package)

11
Example 5: pendulum
𝑀 = 𝐼𝜃̈

𝑀 = 𝑇$ − 𝑇% = 𝑇$ − 𝐹% 𝑙 = 𝑇$ − 𝑚𝑔 sin 𝜃 𝑙

Applied 𝑇$ − 𝑚𝑔𝑙 sin 𝜃 = 𝐼𝜃̈

torque
(rearrangement)

𝐅! 𝑔 𝑇$
𝜃̈ + sin 𝜃 = &
𝑙 𝑚𝑙
(linearization) sin 𝜃 ≅ 𝜃

𝑔 𝑇$
𝜃̈ + 𝜃 = &
𝑙 𝑚𝑙
(𝑑 ⁄𝑑𝑡 → 𝑠)

Transfer function
1
𝛩𝑠 &
= 𝑚𝑙 𝑔
𝑇$ 𝑠 𝑠& + 𝑙 12
Example 5: pendulum

The Simulink block diagram representing the linear equation:

(180/π)
𝑔 𝑇$
𝜃̈ + 𝜃 = &
𝑙 𝑚𝑙

(g = 9.81 m/sec2, l = 1 m, m = 1 kg)

The Simulink block diagram representing the nonlinear equation:

𝑔 𝑇$
𝜃̈ + sin 𝜃 = &
𝑙 𝑚𝑙
13
Example 5: pendulum

14
M2.1.1: Dynamics of mechanical systems
o Summary to develop equations of motion for rigid bodies (steps)
1. Assign variables such as 𝑥 and 𝜃 that are necessary and sufficient to describe an arbitrary
position of the object.
2. Draw a free-body diagram of each component. Indicate all forces acting on each body and
their reference directions. Also indicate the accelerations of the center of mass with respect
to an inertial reference for each body.
3. Apply Newton’s second law in translation and/or rotation form.
4. Combine the equations to eliminate internal forces (simplification).
5. The number of independent equations should equal the number of unknowns.

15
Module M2.1

Module M2.1: Mathematical modeling of

dynamic systems

Contents
• M2.1.1: Dynamics of mechanical systems
• M2.1.2: Modeling of electronic circuits
• M2.1.3: Modeling of heat flow systems

16
M2.1.2: Modeling of electronic circuits
Recap of key concepts

§ In a circuit composed of linear components (e.g., resistances,

capacitances, inductances and ideal amplifiers) all voltages and
current are linearly related.

§ The basic equations for electric circuits are the Kirchhoff’s laws:
§ KCL: the algebraic sum of currents leaving a junction or node
equals the algebraic sum of currents entering that node.
§ KVL: the algebraic sum of all voltages taken around a closed
path in a circuit is zero.

§ For complex circuits we can use the node method:

1. Select one node as the reference (ground).
2. Write equations for the selected unknowns using KCL at each node.
3. Use element equations and KVL to express currents.

17
Example 1: Bridged T filter
𝑣' = 𝑣(
𝑣, = 𝑣!

1 0 = 𝑖+( + 𝑖)' + 𝑖*&

2 𝑖)' = 𝑖)& + 𝑖*'

3 𝑖)& + 𝑖*& = 0

𝑣( − 𝑣& 𝑣& − 𝑣! 𝑑𝑣&

2 − + + 𝐶' =0
𝑅' 𝑅& 𝑑𝑡
𝑣& − 𝑣! 𝑑 𝑣( − 𝑣!
3 + 𝐶& =0
𝑅& 𝑑𝑡
(𝑑⁄𝑑𝑡 → 𝑠)

𝑉! (𝑠)
𝑉( (𝑠)
Transfer function
18
Example 2: Circuit with a current source
1 𝑖 𝑡 = 𝑖- + 𝑖'

2 𝑖- + 𝑖' = 𝑖& + 𝑖,

𝑑𝑣'
𝑖' = 𝐶'
𝑑𝑡
𝑑𝑖-
𝑣) = 𝑖- 𝑅 𝐿 = 𝑣' − 𝑣)
𝑑𝑡
𝑑𝑣&
𝑖& = 𝐶&
𝑑𝑡
𝑣&
𝑖, =
𝑅&

𝑑𝑖-
𝐿 = 𝑣' − 𝑖- 𝑅
𝑑𝑡
𝑑𝑣'
𝐶' = 𝑖(𝑡) − 𝑖-
𝑑𝑡
𝑑𝑣& 𝑣&
𝐶& = 𝑖(𝑡) −
𝑑𝑡 𝑅& 19
Example 3: Ideal operational amplifier

Symbol

Circuit model

Ideal opamp → →

20
Example 4: Opamp adder (summer)

𝑣. − 𝑣/ = 0

𝑖' + 𝑖& = −𝑖01#

𝑣' 𝑣& 𝑣01#

(If not represented, we must + =−
assume the non-inverting 𝑅' 𝑅& 𝑅2
terminal, +, is connected to
ground)
𝑅2 𝑅2
𝑣01# = − 𝑣' + 𝑣&
𝑅' 𝑅&

21
Example 5: Opamp integrator

𝑣. − 𝑣/ = 0 #
𝑣(3 𝑑𝑣01# 1
+𝐶 =0 𝑣01# = − H 𝑣(3 𝜏 𝑑𝜏 + 𝑣01# 0
𝑖)(3 + 𝑖01# =0 𝑅(3 𝑑𝑡 𝑅(3 𝐶 !

(𝑑 ⁄𝑑𝑡 → 𝑠)

1 𝑉(3 𝑠
𝑉01# 𝑠 = −
𝑠 𝑅(3 𝐶
Transfer function

22
Module M2.1

Module M2.1: Mathematical modeling of

dynamic systems

Contents
• M2.1.1: Dynamics of mechanical systems
• M2.1.2: Modeling of electronic circuits
• M2.1.3: Modeling of heat flow systems

23
M2.1.3: Modeling of heat flow systems

Heat energy flows through substances at a rate proportional to the temperature difference
across the substance:
1
𝑞= 𝑇 − 𝑇!
𝑅 "
heat-energy flow, joules
per second (J/sec) temperature, K

thermal resistance, K/J·sec

The net heat-energy flow into a substance affects the temperature of the substance according
to:
𝑑𝑇 1
= 𝑞
𝑑𝑡 𝐶 Thermal capacity, J/K

24
Example 1: heat flow from a room

1
𝑞( = 𝑇 − 𝑇5
𝑅( 4

1 1 1 1
𝑇̇5 = 𝑞' + 𝑞& = + 𝑇4 − 𝑇5
𝐶 𝐶 𝑅' 𝑅&
Room with all but two sides insulated

Normally the material properties are given and these have to be scaled according to the
mass and geometry:

1. The specific heat (per mass unit), 𝑐8 , is simply converted into heat capacity by 𝐶 = 𝑚𝑐8
2. The thermal conductivity, 𝑘, is simply converted into thermal resistance by:
1 𝑘𝐴
=
𝑅 𝑙
𝐴 is the cross-sectional area and 𝑙 is the length of the heat-flow path. 25
Example 2: thermal flow through two masses

𝐶' 𝑇'̇ = 𝑢 − 𝐻' 𝑇' − 𝐻6 𝑇' − 𝑇&

Heat input
𝐶& 𝑇̇& = 𝐻6 𝑇' − 𝑇& −𝐻& 𝑇&

𝑇) , 𝑇+ temperature difference of masses with respect to ambient

𝐻) , 𝐻+ thermal conductances (1⁄𝑅) from masses to ambient
𝐻, thermal conductance from mass 1 to mass 2

(𝑑 ⁄𝑑𝑡 → 𝑠)

Cramer’s rule

𝑇& 𝑠 𝐻6
=
𝑈𝑠 𝐶' 𝑠 + 𝐻6 + 𝐻' 𝐶& 𝑠 + 𝐻6 + 𝐻& − 𝐻6&
26
Module M2.1
o Summary
• Mathematical modeling of the system to be controlled is the first step in analyzing and
designing the required system controls.

• In this lecture, we have reviewed some analytical models for representative dynamic
systems. In particular, we have developed models for simple mechanical, electronic and
heat flow systems.

• Despite we usually linearize systems in order to simplify the analysis, it is always a good
idea to numerically simulate the system taking into account the non-linearities that can be
present during the real performance.

• Note: apart from analytical modeling, it is also possible to obtain a mathematical model using
experimental data exclusively, or in addition to the analytical solution. However, advanced
methods for numerical modeling or system identification are out of the scope of this course.

27