Академический Документы
Профессиональный Документы
Культура Документы
By
Inno Odira
c Copyright by Inno Odira, 2018
Table of Contents
Bibliography 6
REFERENCES 6
iii
1
0.1 Natural Frequencies and Mode Shapes
Example:
Consider a simple 2DOF mass- spring system Fig. 1
For m1
For m2
In Matrix form:
" #" # " #" # " #
m1 0 ẍ1 k1 + k2 −k2 x1 0
+ = (1)
0 m2 ẍ2 −k2 k2 + k3 X2 0
| {z } | {z } | {z } | {z } | {z }
M Ẍ K X 0
In a familiar form
M Ẍ + KX = 0 (2)
Just like the scalar case, we can assume a solution:
" #
X̄ 1
X(t) = X̄eiωt = eiωt (3)
X̄2
2
Differentiate (3)and substitute into the equation of motion (2)
" #!
(k1 + k2 ) − ω 2 m1 −k2
det(A) = 0 → det([K−ω 2 M ]) = 0 → det =0
(k2 + k3 ) − ω 2 m2
−k2
(5)
This gives the characteristic polynomial i.e. Characteristic Equation. The roots of this
characteristic equation defines the natural frequencies (= (nat freq)2 )
Note: From Controls, solutions to this are also called the poles of the system. Their
location in the s-plane tells us about the system response.
Example Parameters
m1 = m2 = 1kg, k1 = k2 = k3 = 4N/m
Characteristic Equation
ω4 − 16ω 2 + 48 = 0
Solve this equation for ω 2 (it is quadratic in ω 2 )
p
2 16 ± 162 − 4(48)
ω1,2 =
2
3
Giving
We can now use ω12 and ω22 (The eigenvalues) to solve for X̄1 and X̄2 (the eigenvectors) in
the equation [K − ω 2 M ]X̄ = 0
This results in
4x̄1 − 4x̄2 = 0
−4x̄1 + 4x̄2 = 0
Note that the two equations have linearly dependent rows and columns hence Det(.) = 0
Resulting in
" #
1
x̄1 = x̄2 → X̄1 =
1
So, the part of the response from this eigenvalue/ eigenvector pair is
" # " #
1 1
X1 (t) = a1 eiω1 t + a2 e−iω1 t ≡ X1 (t) = b1 X̄1 cos(ω1 t − φ1 )
1 −1
Considering ω22 first: i.e. ω22 = 12
4
" # " #! " # " #
8 −4 1 0 x̄1 0
− ω22 =
−4 8 0 1 x̄2 0
" #" # " #
−4 −4 x̄1 0
=
−4 −4 x̄2 0
This results in
−4x̄1 − 4x̄2 = 0
−4x̄1 − 4x̄2 = 0
Resulting in
" #
1
x̄1 = −x̄2 → X̄2 =
−1
So, the part of the response from this eigenvalue/ eigenvector pair is
" # " #
1 1
X2 (t) = a3 eiω2 t + a4 e−iω2 t ≡ X2 (t) = b2 X̄2 cos(ω2 t − φ2 )
−1 1
The total solution is then
X(t) = b1 X̄1 cos(ω1 t − φ1 ) + b2 X̄2 cos(ω2 t − φ2 )
Note: The undetermined coefficients b1 , b2 , φ1 and φ2 are determined from the initial
conditions just like the 1DOF
" # " #
1 1 √
X(t) = b1 cos(2t − φ1 ) + b2 cos( 12t − φ2 )
1 −1
" # " #
1 √ 1 √
Ẋ(t) = −2b1 sin(2t − φ1 ) − 12b2 sin( 12t − φ2 )
1 −1
Substitute in the initial conditions
" # " # " #
1 1 1
X(0) = = b1 cos(−φ1 ) + b2 cos(−φ2 ) (6)
0 1 −1
" # " # " #
1 1 √ 1
Ẋ(0) = = −2b1 sin(−φ1 ) − 12b2 sin(−φ2 ) (7)
0 1 −1
5
These are equivalent to Four simultaneous equations
b1 = b2 = 0.5mm and φ1 = φ2 = 0
The total solution becomes
" # " #
1 1 √
X(t) = o.5 cos(2t) + .5 cos( 12t)
1 −1
Notes on Eigenvectors.
2. For this reason you can think about eigenvectors as mode shapes
3. If you excite a system at exactly a mode-shape, only that mode shows up in the
response.
4. Ortherwise, all modes show up with relative contributions depending on the excitation.
REFERENCES