ME 370

Vibration of Mechanical
Systems
Instructor: Dr. A. Scott Lewis ARL Science Park Building 814-865-0962
(Science Park Road)
E-Mail: [email protected]

Ordinary Differential Equations

LECTURE 4 2
Lecture Content
• Solving ODEs
• Introduction

• Review of Solution techniques for ODEs in the Time Domain

• Solution by Direct Integration
• Separation of Variables
• Trial Solution Method
• Homogenous and Particular Solution

3
SOLVING ORDINARY DIFFERENTIAL 4

EQUATIONS
Objective: Understand time responses
k c Output:
position of the mass
m 𝑦𝑦 𝑡𝑡
f (t )
Impulse Sine
𝑦𝑦𝑎𝑎𝑎𝑎𝑎𝑎 𝑡𝑡
Mechanical System
(Mass-spring-damper)
Step System
Input Response

y(t)
Pulse 𝑦𝑦 𝑡𝑡
Mathematical Model
(Mass-spring-damper) Input
Output
Parabola

Ramp
𝑚𝑚𝑦𝑦̈ 𝑡𝑡 + 𝑐𝑐 𝑦𝑦̇ 𝑡𝑡 + 𝑘𝑘𝑘𝑘 𝑡𝑡 = 𝑓𝑓 𝑡𝑡
with 𝑦𝑦 0 = 𝑎𝑎; 𝑦𝑦̇ 0 = 𝑏𝑏

5
 ODE Solution Techniques

• Analytical Methods
• Solution in Time Domain
• Several techniques available
• Applicable to limited number of systems
• Might be difficult for higher order system (3rd and above)

• Laplace Method (s-domain)

• Requires to be converted back into time domain
• Simple even for high order systems

• Numerical Tools (later in the Semester)

• Used in simulation software like Matlab/Simulink
6
Solution by Direct Integration
• With some 1st order equations, we can isolate the derivative
on the left as:
𝑑𝑑𝑑𝑑
= 𝑓𝑓(𝑡𝑡)
𝑑𝑑𝑑𝑑

• In this case, we can integrate bot side to obtain:

𝑥𝑥 𝑡𝑡 𝑡𝑡
� 𝑑𝑑𝑑𝑑 = � 𝑓𝑓(𝑡𝑡) � 𝑑𝑑𝑑𝑑
𝑥𝑥 𝑡𝑡0 𝑡𝑡0
Or:
𝑡𝑡
𝑥𝑥 𝑡𝑡 = 𝑥𝑥 𝑡𝑡0 + � 𝑓𝑓(𝑡𝑡) � 𝑑𝑑𝑑𝑑
𝑡𝑡0
7
Solution by Direct Integration
• Example: Find the solution of the following ODE:
𝑑𝑑𝑑𝑑
= 6𝑡𝑡 2 with 𝑥𝑥 0 = 5
𝑑𝑑𝑑𝑑

• Solution (system response):

𝑡𝑡
𝑥𝑥 𝑡𝑡 = 𝑥𝑥 𝑡𝑡0 + � 𝑓𝑓(𝑡𝑡) � 𝑑𝑑𝑑𝑑
𝑡𝑡0
𝑡𝑡 𝑡𝑡
𝑥𝑥 𝑡𝑡 = 5 + � 6𝑡𝑡 2 � 𝑑𝑑𝑑𝑑 = 5 + 2𝑡𝑡 3 � = 5 + 2𝑡𝑡 3
0 0

8
Separation of Variables
• If the ODE has the following form:
𝑥𝑥̇ = 𝑔𝑔 𝑡𝑡 � 𝑓𝑓(𝑥𝑥)

• One can solve it by separating the variables 𝑥𝑥 and 𝑡𝑡 as follows

𝑑𝑑𝑑𝑑
= 𝑔𝑔(𝑡𝑡) � 𝑑𝑑𝑑𝑑
𝑓𝑓 𝑥𝑥

• And integrate both sides to obtain:

𝑥𝑥 𝑡𝑡 𝑡𝑡
𝑑𝑑𝑑𝑑
� = � 𝑔𝑔(𝑡𝑡) � 𝑑𝑑𝑑𝑑
𝑥𝑥 0 𝑓𝑓 𝑥𝑥 0

9
Separation of Variables - Example
• Use separation of variables to solve the following problem:

𝑥𝑥̇ + 2𝑥𝑥 = 20 with 𝑥𝑥 0 =3

Solution (system response)
x =20 - 2 x ⇒ f ( x) =20 − 2 x and g (t ) =
1
As a result ,
t
x (t ) dx
∫3 20 − 2 x ∫0
= 1dt

Solving
x (t )
1 t 1 1
− ln(20 − 2 x) t 0 ⇒ − ln(20 − 2 x) + ln(20 − 2(3)) =
= t −0
2 3 2 2
Solve for x
 20 − 2 x 
ln(20 − 2 x) − ln(14) = −2t ⇒ ln  =−2t
 14 
20 − 2 x
or = e −2t ⇒ x = 10 − 7e −2t 10
14
Alternate Approach- Homogeneous and Particular Solution

x + 2 x =
20
So
x(t ) xh (t ) + x p (t )
=
xh (t ) is the solution to x + 2 x = 0 ⇒ s + 2 = 0 ⇒ xh (t ) = ce −2t
x p (t ) = A ⇒ x p (t ) =
0, plug into differential equation
0 + 2 A = 20 ⇒ A = 10
Therefore,
x(t ) = xh (t ) + x p (t ) = ce −2t + 10
Now use initial condition to solve for c
x(0) =c + 10 =3 ⇒ c =−7
x 10 − 7e −2t
=
11
Trial Solution Method
• Find the solution of the following equation:
𝑚𝑚𝑦𝑦̈ 𝑡𝑡 + 𝑐𝑐𝑦𝑦̇ 𝑡𝑡 + 𝑘𝑘𝑘𝑘 𝑡𝑡 = 𝑓𝑓 𝑡𝑡 with 𝑦𝑦 0 = 𝑎𝑎; 𝑦𝑦̇ 0 = 𝑏𝑏
• Approach consists of assuming a trial function:

𝑦𝑦 𝑡𝑡 = 𝐶𝐶 + 𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠
1. Plug in the candidate solution into the ODE
2. Use initial conditions to eliminate the coefficients

• Useful in case of common ODEs whose solution forms are

known from experience

• Note:
• the coefficients s are called characteristic roots 12
• Works also if m is zero (first order system)
Trial Solution Method – Example 1
• Consider the following ODE:
𝑦𝑦̈ + 7𝑦𝑦̇ + 10𝑦𝑦 = 20 with 𝑦𝑦 0 = 5 and 𝑦𝑦̇ 0 = 3

1. Find a solution candidate:

𝑦𝑦 𝑡𝑡 = 𝐶𝐶 + 𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠
2. Plug in into the ODE
𝑦𝑦̇ 𝑡𝑡 = 𝐷𝐷𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠
2 𝑠𝑠𝑠𝑠 𝐷𝐷𝑠𝑠 2 𝑒𝑒 𝑠𝑠𝑠𝑠 + 7 𝐷𝐷𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠 + 10 𝐶𝐶 + 𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠 = 20
𝑦𝑦̈ 𝑡𝑡 = 𝐷𝐷𝑠𝑠 𝑒𝑒
3. Group all the terms with the exponential
𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠 𝑠𝑠 2 + 7𝑠𝑠 + 10 + 10𝐶𝐶 = 20
The only way the above equation is true:
20 𝑠𝑠1 = −2 13
2
𝐶𝐶 = and 𝑠𝑠 + 7𝑠𝑠 + 10 = 0 𝑠𝑠2 = −5
10
Two Distinct Real Roots
Trial Solution Method – Example 1
• Consider the following ODE:
𝑦𝑦̈ + 7𝑦𝑦̇ + 10𝑦𝑦 = 20 with 𝑦𝑦 0 = 5 and 𝑦𝑦̇ 0 = 3

4. Plug the values of C and s back in the candidate solution:

𝑦𝑦 𝑡𝑡 = 2 + 𝐷𝐷1 𝑒𝑒 −2𝑡𝑡 + 𝐷𝐷2 𝑒𝑒 −5𝑡𝑡
5. Use initial conditions to determine the constants:
𝑦𝑦 0 = 2 + 𝐷𝐷1 + 𝐷𝐷2 = 5
𝐷𝐷1 = 6
𝑦𝑦̇ 𝑡𝑡 = −2𝐷𝐷1 𝑒𝑒 −2𝑡𝑡 − 5𝐷𝐷2 𝑒𝑒 −5𝑡𝑡 = 3
𝐷𝐷2 = −3
𝑦𝑦̇ 0 = −2𝐷𝐷1 − 5𝐷𝐷2 = 3
6. The solution is:
𝑦𝑦 𝑡𝑡 = 2 + 6𝑒𝑒 −2𝑡𝑡 − 3𝑒𝑒 −5𝑡𝑡
14
Solutions in Time Domain
15

Example 1
Example 2
Example 3

10
y(t)

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Time [s]

• The type of characteristic roots determines the shape of the 15

system response.
Summary
• Solution of lower order ODEs (1st and 2nd) can be obtained
easily.
• If the order of the system is higher, this approach will lead to
tedious math manipulations.
• In addition, the solution depends on the form of the input
function.

16
ODE Solution Techniques
• Laplace Transform:
• Equation of motion is transferred into the Laplace domain (s-
domain, frequency domain)
• ODE becomes an algebraic equation in the s-domain
• Gives insight into properties of Linear Time Invariant (LTI)
systems
• Controller synthesis and analysis simpler in the frequency domain

17
MORE EXAMPLES –WITH SOLUTION 18
Trial Solution Method – Example 2
• Consider the following ODE:
5𝑦𝑦̈ + 20𝑦𝑦̇ + 20𝑦𝑦 = 28 with 𝑦𝑦 0 = 5 and 𝑦𝑦̇ 0 = 8

1. Find a solution candidate:

𝑦𝑦 𝑡𝑡 = 𝐶𝐶 + 𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠
2. Plug in into the ODE
𝑦𝑦̇ 𝑡𝑡 = 𝐷𝐷𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠
2 𝑠𝑠𝑠𝑠 5𝐷𝐷𝑠𝑠 2 𝑒𝑒 𝑠𝑠𝑠𝑠 + 20 𝐷𝐷𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠 + 20 𝐶𝐶 + 𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠 = 28
𝑦𝑦̈ 𝑡𝑡 = 𝐷𝐷𝑠𝑠 𝑒𝑒
3. Group all the terms with the exponential
𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠 5𝑠𝑠 2 + 20𝑠𝑠 + 20 + 20𝐶𝐶 = 28
The only way the above equation is true:
28 −20 ± 202 − 202 19
𝐶𝐶 = and 5𝑠𝑠 2 + 20𝑠𝑠 + 20 = 0 𝑠𝑠1,2 =
20 10
𝑠𝑠1,2 = −2
Two Identical Real Roots
Trial Solution Method – Example 2
• Consider the following ODE:
5𝑦𝑦̈ + 20𝑦𝑦̇ + 20𝑦𝑦 = 28 with 𝑦𝑦 0 = 5 and 𝑦𝑦̇ 0 = 8

4. Plug the values of C and s back in the candidate solution:

7
𝑦𝑦 𝑡𝑡 = + 𝐷𝐷1 + 𝐷𝐷2 𝑡𝑡 𝑒𝑒 −2𝑡𝑡
5
5. Use initial conditions to determine the constants:
7
𝑦𝑦 0 = + 𝐷𝐷1 = 5 18
5 𝐷𝐷1 =
𝑦𝑦̇ 𝑡𝑡 = −2 𝐷𝐷1 + 𝐷𝐷2 𝑡𝑡 𝑒𝑒 −2𝑡𝑡 + 𝐷𝐷2 𝑒𝑒 −2𝑡𝑡 5
𝑦𝑦̇ 0 = −2 𝐷𝐷1 + 𝐷𝐷2 = 8 76
𝐷𝐷2 =
5
6. The solution is:
20
7 18 76
𝑦𝑦 𝑡𝑡 = + + 𝑡𝑡 𝑒𝑒 −2𝑡𝑡
5 5 5
Trial Solution Method – Example 3
• Consider the following ODE:
𝑦𝑦̈ + 16𝑦𝑦 = 144 with 𝑦𝑦 0 = 5 and 𝑦𝑦̇ 0 = 12

1. Find a solution candidate:

𝑦𝑦 𝑡𝑡 = 𝐶𝐶 + 𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠
2. Plug in into the ODE
𝑦𝑦̇ 𝑡𝑡 = 𝐷𝐷𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠
2 𝑠𝑠𝑠𝑠 𝐷𝐷𝑠𝑠 2 𝑒𝑒 𝑠𝑠𝑠𝑠 + 16 𝐶𝐶 + 𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠 =144
𝑦𝑦̈ 𝑡𝑡 = 𝐷𝐷𝑠𝑠 𝑒𝑒
3. Group all the terms with the exponential
𝐷𝐷𝑒𝑒 𝑠𝑠𝑠𝑠 𝑠𝑠 2 + 16 + 16𝐶𝐶 = 144
The only way the above equation is true:
144 21
𝐶𝐶 = =9 and 𝑠𝑠 2 + 16 = 0 𝑠𝑠1,2 = ±4𝑖𝑖
16
Two Imaginary Roots
Trial Solution Method – Example 3
• Consider the following ODE:
𝑦𝑦̈ + 16𝑦𝑦 = 144 with 𝑦𝑦 0 = 5 and 𝑦𝑦̇ 0 = 12

4. Plug the values of C and s back in the candidate solution:

𝑦𝑦 𝑡𝑡 = 9 + 𝐷𝐷1 𝑒𝑒 4𝑖𝑖𝑖𝑖 + 𝐷𝐷2 𝑒𝑒 −4𝑖𝑖𝑖𝑖
• Use Euler’s Formula:
𝑒𝑒 4𝑖𝑖𝑖𝑖 = cos 4𝑡𝑡 + 𝑖𝑖 sin 4𝑡𝑡
𝑒𝑒 −4𝑖𝑖𝑖𝑖 = cos 4𝑡𝑡 − 𝑖𝑖 sin 4𝑡𝑡
𝑥𝑥 𝑡𝑡 = 9 + 𝐷𝐷1 + 𝐷𝐷2 cos 4𝑡𝑡 + 𝐷𝐷1 − 𝐷𝐷2 𝑖𝑖 sin 4𝑡𝑡
• The solution must be real, hence D1 and D2 must be complex conjugate
𝐷𝐷1 = 𝛼𝛼 + 𝑖𝑖𝑖𝑖 𝐷𝐷1 + 𝐷𝐷2 = 𝛼𝛼 + 𝑖𝑖𝑖𝑖 + 𝛼𝛼 − 𝑖𝑖𝑖𝑖 = 2𝛼𝛼 = 𝐵𝐵1
𝐷𝐷2 = 𝛼𝛼 − 𝑖𝑖𝑖𝑖 𝐷𝐷1 − 𝐷𝐷2 𝑖𝑖 = 𝛼𝛼 + 𝑖𝑖𝑖𝑖 − 𝛼𝛼 + 𝑖𝑖𝑖𝑖 = −2𝛽𝛽 = 𝐵𝐵1 22

𝑦𝑦 𝑡𝑡 = 9 + 𝐵𝐵1 cos 4𝑡𝑡 + 𝐵𝐵2 sin 4𝑡𝑡

Trial Solution Method – Example 3
• Consider the following ODE:
𝑦𝑦̈ + 16𝑦𝑦 = 144 with 𝑦𝑦 0 = 5 and 𝑦𝑦̇ 0 = 12

5. Use initial conditions to determine the constants:

𝑦𝑦 0 = 9 + 𝐵𝐵1 = 5
𝐵𝐵1 = −4
𝑦𝑦̇ 𝑡𝑡 = −4𝐵𝐵1 sin 4𝑡𝑡 + 4𝐵𝐵2 cos 4𝑡𝑡
𝐵𝐵2 = 3
𝑦𝑦̇ 0 = 4𝐵𝐵2 = 12

6. The solution is:

𝑦𝑦 𝑡𝑡 = 9 − 4 cos 4𝑡𝑡 + 3 sin 4𝑡𝑡

23
Alternate Approach- Homogeneous and Particular Solution
Consider the following ODE
y + 2=
 y + 5 y 100 with
= y (0) 5 and=
y (0) 12
Solve for hom ogeneous part
y + 2 y + 5 y = 0 ⇒ s 2 + 2 s + 5 = 0

−2 ± 22 − 4(5) −2 ± −16
Roots are solved by quadratic equation : = =−1 ± j 2
2 2
s1 =−1 + j 2 Complex Roots
s 2 =−1 − j 2 ⇒ yh (t ) =e − t (c1 cos 2t + c2 sin 2t )
Solve for particular solution
y p (t ) =
A ⇒ y p (t ) =
y p (t ) =
 0
5 A = 100 ⇒ A = 20 ⇒ y p (t ) = 20
As a result
y (t ) = yh (t ) + y p (t ) = e − t (c1 cos 2t + c2 sin 2t ) + 20
Now apply initial conditions
y (0) =c1 + 20 =⇒
5 c1 = −15
y (t ) = e − t (−2c1 sin 2t + 2c2 cos 2t ) + e − t (c1 cos 2t + c2 sin 2t )(−1)
y (0) = 2c2 − c1 = 12 ⇒ c2 = (12 + c1 ) / 2 = (12 − 15) / 2 = −1.5

24
Finally,
y (t ) =yh (t ) + y p (t ) =e − t (−15cos 2t − 1.5sin 2t ) + 20