Вы находитесь на странице: 1из 6

Daniel Wiese Dynamics Note Sheet 1

Single Particle
m
O
B
r
Om
f
v
r
OB
r
Bm
Figure 1: Point mass m under action of force f. Point O is xed in inertial space, and
point B is a general point, not necessarily xed in inertial space.
p = mv
f =
dp
dt

O
= r
Om
f
h
O
= r
Om
p

O
=
d
dt
h
O

B
=
O
r
OB
f
h
B
= h
O
r
OB
p

B
=
d
dt
h
B
+ v
B
p
Kinematics of Rigid Bodies
C
O
B
P
r
BC
r
CP
v
C
v
B
Figure 2: Rigid body. O is inertially xed in space, and point B is a general point, which
can be moving, about which we may take moments. Point C is the body center of mass,
and point P is some other point xed on the body
Note that the vector notation r
BC
means a vector from point B to point C.
P = mv
C

ext
O
=
d
dt
H
O

ext
B
=
d
dt
H
B
+ v
B
P
H
B
= [I]
B

F
ext
=
d
dt
P
H
B
= H
C
+ r
BC
P
H
B
= H
O
+ r
BO
P
v
P
= v
C
+ r
CP
For non-holonomic (rolling coin):
ext
B
=
d
dt
H
C
+ r
BC

d
dt
P
To nd principal axes
Iprincipal I = 0
Impulse
Take linear and angular momentum principle for a rigid body
F
ext
=
d
dt
P

ext
B
=
d
dt
H
B
+ v
B
P
and separate them and integrate them over a short period of time

t=0
+
t=0

F
ext
dt =

P(0
+
)
P(0

)
dP

t=0
+
t=0

ext
B
dt =

H
B
(0
+
)
H
B
(0

)
dH
B
+

t=0
+
t=0

v
B
Pdt
P =

t=0
+
t=0

F
ext
dt
Remember when integrating from t = 0

to t = 0
+
constant forces like gravity integrate to
zero
Work and Energy Principles
KE of rigid body rotating about CM:
T =
1
2
Mv
2
G
+
1
2
_
Ix
2
x
+ Iy
2
y
+ Iz
2
z
_
Finding Center of Mass and Moment of Inertia
P is a general point.
Parallel axis theorem: I
P
= I
C
+ Mh
2
[I]
P
= [I]
C
+ M
_
_
b
2
+ c
2
ab ac
ab a
2
+ c
2
bc
ac bc a
2
+ b
2
_
_
Moments of inertia Ix =

m
(y
2
+ z
2
)dm
Products of inertia Ixy =

V
xydV
Ix =

V
(y
2
+ z
2
)dV
x CM: xcm =

i
Airi

i
Ai
y CM: zcm =

m
zdm

m
dm
z CM: zcm =

V
zdV

V
dV
Areas, Volumes, Centroids, Moments of Inertia
Asphere = 4r
2
Vsphere =
4
3
r
3
Vcone =
1
3
r
2
h
Through axis of rot. Icylinder,z =
1
2
mr
2
Through center Icylinder,x,y =
1
12
m(3r
2
+ h
2
)
Rod length L about end: Irod,end =
1
3
mL
2
Rod length L about center: Irod,center =
1
12
mL
2
Sphere radius r: Isphere =
2
5
mr
2
Cone Icone,z =
3
10
mr
2
Cube thru cent l(x), w(y), h(z) Icube,x =
1
12
(w
2
+ h
2
)
Axes at tip of cone Icone,x,y =
3
80
m(4r
2
+ h
2
)
Axes at base of cone Icone,x,y =
3
20
mr
2
+
1
10
mh
2
)
through center of hoop Ihoop,z = mr
2
Centroid of cone up from base z =
h
4
Lagranges Method to nd EOM
1. Identify number of generalized coordinates and any generalized forces
2. Choose generalized coordinates 1, 2, . . .
3. Find kinetic energy T and potential energy V in terms of these generalized coordinates
4. Assemble Lagrangian
L = T V
5. Express generalized forces in terms of the generalized coordinates
W = Fxx
W = jj
When nding generalized forces which require solving x in terms of , some-
times it is easiest to nd velocities, then cancel dt and make dx into x and d
into .
When breaking force F into components Fx, dont forget the sign
Measure springs deections from static equilibrium and gravity wont appear in
equations of motion
When nding kinetic energy of rigid bodies, place coordinate system at CG and
such that it is a set of principal axes, then the moment of inertia is about the CG
not the physical point of rotation.
6. Evaluate
L


j
and
L
j
7. Use the formula
d
dt
_
L


j
_

L
j
= j
That gives us the equations of motion
Daniel Wiese Dynamics Note Sheet 2
Stability Analysis of Discrete Systems
1. Use Lagranges method to get EOM
2. Identify steady motions
If j does not show up explicitly in the Lagrangian L, it is ignorable, or cyclic.
Set

ignorable = constant
If j does show up in the Lagrangian L, it is non-ignorable. Set
non-ignorable = s = constant
3. Linearize the equations of motion. The form is
[M] x + [K]x = 0
From here there are two options
4. Solve for the natural frequencies and mode shapes
Let x = sx
Solve
_
[M]s
2
+ [K]
_
x = 0 det
_
[K]
2
i
[M]
_
= 0
This is an eigenvalue problem where the eigenvalues are the natural frequencies,
and the eigenvectors are the mode shapes
Alternatively This way requires [M] and [K]
5. Guess as many modes {a}i as possible
6. Use orthogonality to verify guessed modes, and nd new modes
{a}

i
[M]{a}j = 0
{a}

i
[K]{a}j = 0
The orthogonality condition comes from left multiplying
_
[K]
2
i
[M]
_
{a}i = 0
for two cases with i and j by two modes which are orthogonal, {a}

j
and {a}

i
.
7. Use Rayleigh quotient to nd i

2
i
=
{a}

i
[K]{a}i
{a}

i
[M]{a}i
which comes from
_
[K]
2
i
[M]
_
{a}i = 0 and left multiplying by {a}

i
For the system [M]{ x} + [K]{x} = {F} sin t, after we nd the modes, we can put the
modes into a matrix [] and use this matrix to come up with a new system with vector {u},
where the mass and spring matrix are diagonal. Let {x} = []{u}. Plugging this in we get
[]

[M][]
.
[M]
D
{ u} + []

[K][]
.
[K]
D
{u} = []

{F} sin t
Derivations for Continuous Systems
Wave Equation for a String
String with mass/length under tension T. So mass of a little piece is dm = dx. String has
length s with angle on left side, +

x
dx on the right side, and the angle is small, so the
string is approximately length dx.
Conservation of momentum in x-direction: the string does not move in the x-direction
T(x + dx) cos
_
+

x
dx
_
T(x) cos() = 0
Expanding cos
_
+

x
dx
_
cos
_
+

x
dx
_
= cos() cos
_

x
dx
_
sin() sin
_

x
dx
_
Substituting this in
T(x + dx)
_
cos() cos
_

x
dx
_
sin() sin
_

x
dx
__
T(x) cos() = 0
Divide both sides by dx and take the limit as dx 0
T(x) = T = constant
Conservation of momentum in y-direction:
T(x + dx) sin
_
+

x
dx
_
T(x) sin() =
d
2
y
dt
2
dx
Expanding sin
_
+

x
dx
_
sin
_
+

x
dx
_
= sin() cos
_

x
dx
_
+ cos() sin
_

x
dx
_
Plugging in, and using the fact that T(x) = T we have
T
_
sin() cos
_

x
dx
_
+ cos() sin
_

x
dx
__
T sin() =
d
2
y
dt
2
dx
Using small angle approximations we get
T
_
sin() +

x
dx
_
T sin() =
d
2
y
dt
2
dx
Simplifying, we get
T

x
=
d
2
y
dt
2

Using small angle assumption again where tan() =
dy
dx
we have
T

2
y
x
2
=
d
2
y
dt
2

Can add forcing as
d
2
y
dt
2
= T

2
y
x
2
+ f(x, t)
Euler-Bernoulli Beam Equation
Beam with mass/length A, with internal shear force Q, bending moment M
b
, height
y(x, t). Square piece of block with shear force Qdown on left side, Q +
Q
x
dx pointing up
on the right side, and moments M
b
going up on the left side and M
b
+
M
b
x
dx going up on
the right side. The constitutive law for a bending beam relates moment to curvature as
M
b
= EI

2
y
x
2
Conservation of angular momentum about right side gives
M
b
+
M
b
x
M
b
+ Qdx = 0
gives
Q =
M
b
x
Substituting the constitutive law
Q = EI

3
y
x
3
Conservation of linear momentum in y-direction gives
(Adx)

2
y
t
2
=
Q
x
dx
Evaluating
Q
x
using the expression for Qderived using conservation of angular momentum
Q
x
= EI

4
y
x
4
Substituting in
A

2
y
t
2
= EI

4
y
x
4
Self adjoint means solution is separable.
Longitudinal Displacement (Stretching) of a Rod
A

t
2
= EA

x
2
Axial Displacement (Twisting) of a Shaft
J

2

t
2
= GJ

2

x
2
Solving Continuous Systems
Solving String Problems with Forcing
To solve the forced response, always solve the unforced problem rst.
1. Write down governing equation

2
y
t
2
= T

2
y
x
2
+ f(x, t)
2. Propose a solution of the form
y(x, t) = a(x) cos(t)
where the time varying harmonic function matches that of the forcing function (including
frequency)
3. Plut this solution in, and obtain a simplied ODE in a(x)

2
a(x) cos(t) = T
d
2
a
dx
2
cos(t) + f(x, t)
Take for example the forcing function to be f(x, t) = F0 cos(t) then

2
a(x) = T
d
2
a
dx
2
+ F0
rearranging
d
2
a
dx
2
+

2
T
a(x) = F0
using
2
=

2
T
d
2
a
dx
2
+
2
a(x) = F0
4. Find the homogeneous solution a
h
(x) and particular solution ap(x). Propose the ho-
mogeneous solution
a(x) = Ae
Bx
which has second derivative
d
2
a
dx
2
= B
2
Ae
Bx
plugging in we get
B = i
So the solutions are
a1(x) = A1e
ix
a2(x) = A2e
ix
giving
a
h
(x) = C1 cos(x) + C2 sin(x)
Daniel Wiese Dynamics Note Sheet 3
5. Solution to forced equation should be constant
6. Form the total solution by adding the homogeneous and particular solutions, and apply
BCs
Answering the question
Determine the steady-state vibration just means nd y(x, t)
Identifying resonances is to nd values of where the solution blows up
Solving String Problems with a Mass on them
The governing equation for this is the same as a regular string, but the solution is not valid
across mass. Will need to use two solutions, one valid on each side of the mass.
1. Write down governing equation

2
y
t
2
= T

2
y
x
2
2. Consider solving this equation between the left end at x = L and the mass, and then
between the mass and the right end at x = L. The general form of the solution is
y
L
(x, t) = a
L
(x) sin(t)
y
R
(x, t) = a
R
(x) sin(t)
Leads to
y
L
(x, t) = C
L
sin(x +
L
) sin(t)
y
R
(x, t) = C
R
sin(x +
R
) sin(t)
where
2
=

2
T
. Should get
L
= L and
R
= L
3. Apply 4 boundary conditions: each end of the string, and the matching conditions at the
mass
y
L
(xm, t) = y
R
(xm, t)
and the following, which comes from linear momentum of mass in y-direction
M

2
y
t
2

xm
= T
_
y
R
x

xm

y
L
x

xm
_
Where either y
L
(x, t) or y
R
(x, t) can be used to evaluate the second derivative. Use
y
R
(x, t).
4. Arranging these two conditions in matrix form
_
_
sin((xm L)) sin((xm + L))
M
2
sin((xm L))
Tcos((xm L)) Tcos((xm L))
_
_
_
C
R
C
L
_
=
_
0
0
_
5. Special case is when the mass is in the middle of the string at xm. This reduces the
above by taking the determinant to
sin(L)
_
M
2
sin(L) 2Tcos(L)
_
= 0
This equation can be solved to nd the frequencies n from each n.
6. Mode shapes are those when the mass is stationary
Solving Beam Problems
1. Write down governing equation
A

2
y
t
2
= EI

4
y
x
4
2. Propose following solution solution for beam problems. Can show such a separable
solution works.
y(x, t) = a(x) sin(t)
The second and fourth derivatives of general solution to beam equation are

2
y
t
2
=
2
a(x) sin(t)
d
4
y
dx
4
=
d
4
a(x)
dx
4
sin(t)
3. Plugging the proposed solution into the governing equation we get
d
4
a
dx
4

4
a(x) = 0 where
4
=
A
2
EI
4. The general solution to this ODE is
a(x) = C1 sin x + C2 cos x + C3 sinh x + C4 cosh x
5. Apply boundary conditions to solve. This may reduce the number of constants, e.g.
C2 = C4 = 0. Then put the remaining equations into matrix form, and solve for the
constants, either by row operations or by taking the determinant
Damping Problems
W is energy loss/cycle, V is peak potential energy (of whole system), is the loss factor.
Some formulas are:
W =
2

0
f
d
dx
where, for a linear dashpot
Linear dashpot: f
d
= c x
where x is the compression of the damper. This gives
W =
2

0
c
dx
dt
dx
W =
2

0
c
_
dx
dt
_
2
dt
The loss factor is calculated as
=
W
2V
To solve damping problems
1. Do lagrangian to get EOM and nd the natural frequencies and mode shapes assuming
there is no damping
2. Propose a solution of the form
x(t) = a sin(t)
3. Use modes to break this into components
4. Differentiate this solution to get
dx
dt
and plug into the integral to evaluate W.
Rigid Symmetric Body EOM
= 1e
1
+ 2e
2
+ 3e
3
H
c
= I11e
1
+ I22e
2
+ I33e
3

ext
=
d
dt
H
c
d
dt
H
c
= I1( 1e
1
12e
3
+ 13e
2
)
+ I2( 2e
2
+ 21e
3
23e
1
)
+ I3( 3e
3
31e
2
+ 32e
1
)
Set this equal to
ext
and group components together to get EOM. Can simplify with symmetry.
e.g. I1 = I2
1 = I1 1 + 23(I3 I2)
2 = I2 2 + 13(I1 I3)
3 = I3 3 + 12(I2 I1)
Euler Angles
Start with coordinate systemC
XY Z
. Rotate about Z and get C
abc
axes, then rotate about
a and get Cxyz, and nally rotate about z to get C123 axes, which are the body xed axes.
=

e
Z
+

e
x
+

e
3
= 1e
1
+ 2e
2
+ 3e
3
where
1 =

cos +

sin sin
2 =

sin cos

sin
3 =

+

cos
Torque-Free Precession
When there is no torque acting on the system, angular momentum principle
d
dt
H
C
= 0 tells
us that H
C
is constant, and since it is a vector this means its magnitude and direction are
constant. So, we can choose the coordinate system C
XY Z
such that the Z axis is aligned
with H
C
.
H
C
= H
C
e
Z
= H
C
(sin e
y
+ cos e
z
)
= H
C
(sin sin e
1
+ sin cos e
2
+ cos e
3
)
Compare this expression for H
C
to the earlier one
H
C
= I11e
1
+ I22e
2
+ I33e
3
Components have to match exactly. This gives

= H
C
_
sin
2

I1
+
cos
2

I2
_

= H
C
_
1
I1

1
I2
_
sin cos sin

= H
C
cos
_
1
I3

sin
2

I1

cos
2

I2
_
If the body has symmetry, say about 3 axis, then I1 = I2 = I and we can see from
these expressions that

= s = constant, that

=
H
C
I
= = constant, and

=
[(I I3)/I3]cos s = constant. This is torque-free precession.
Spinning Top with Euler Angles
Use LaGrange to get EOM, neglecting v
C
Using Euler angles for in Lagrange =

e
Z
+

e
x
+

e
3
Describe in terms of
Cxyz with e
Z
= sin e
z
+ cos e
y
e
3
= e
z
=

sin e
z
+

cos e
y
+

e
x
+

e
z
Do Lagrange, identify from equation I3z = constant z

Daniel Wiese Dynamics Note Sheet 4


Torque-Free Motion of a Rigid Body
= e
Z
+

e
z
= sin se
y
+ (

+ cos s)e
z
H
C
= Iye
y
+ I3ze
z
= Isin se
y
+ I3(

+ cos s)e
z
= I(sin se
y
+ cos s ez)
= Ie
Z
If we evaluate
d
dt
H
C
we can solve for the rate of change of the angular rates as
1 =
I2 I3
I1
23
2 =
I3 I1
I2
31
3 =
I1 I2
I3
12
We have with no torque that H
C
= I3(

+

) is constant and = (

+

)e
z
. Choose H
C
to be parallel to e
Z
. Use Euler equations to examine stability of steady rotation. = 3e
3
.
So 1 and 2 are basically constant, giving
1 +
(I1 I3)(I2 I3)
I1I2

2
3
1 = 0
and this is stable when (I1 I3)(I2 I3) > 0 and unstable when (I1 I3)(I2 I3) < 0.
So stable when I3 is either a maximum or minimum moment of inertia.
General Math Stuff
Taylor series
f(x) f(a) +
f

(a)
1!
(x a) +
f

(a)
2!
(x a)
2
+ . . .
Using this for sine and cosine, small angles
cos(x) = 1
1
2
x
2
sin(x) = x
x
3
6
Identities
sin(u v) = sin ucos v cos usin v
cos(u v) = cos ucos v sin usin v
e
ix
= cos x + i sin x
sinh(x) =
1
2
(e
x
e
x
)
cosh(x) =
1
2
(e
x
+ e
x
)
sin(x) =
1
2i
(e
ix
e
ix
)
cos(x) =
1
2
(e
ix
+ e
ix
)
sinh(x) + cosh(x) = e
x
d
dx
sinh(x) = cosh(x)
d
dx
cosh(x) = sinh(x)
sinh(0) = 0 cosh(0) = 1
Integrals

sin
2
(ax)dx =
x
2

sin(2ax)
4a

cos
2
(ax)dx =
x
2
+
sin(2ax)
4a
Writing sum of sin and cos as sin with a phase shift
A1 sin(x)+A2 cos(x) =
_
A
2
1
+ A
2
2
_
_
_
_
_
_
_
A1
_
A
2
1
+ A
2
2
.
cos
sin(x) +
A2
_
A
2
1
+ A
2
2
.
sin
cos(x)
_
_
_
_
_
_
_
Can check that sin
2
+ cos
2
= 1. Then use identity sin(x +) = sin(x) cos +
cos(x) sin giving
A1 sin(x) + A2 cos(x) = A3 sin(x + )
sin
cos
=
A2
A1
= tan
1
A2
A1
and A3 =
_
A
2
1
+ A
2
2
Daniel Wiese Dynamics Note Sheet 5
Wave Equation on String
This page gives an outline of the general procedure to derive the equations of motion, propose
a general solution, and solve for constants using boundary and initial conditions (here we
assume the boundary conditions are both ends xed, and zero initial conditions) in order to get
the mode shapes and natural frequencies.
Physical assumptions: homogenous string A = constant, the string is perfectly elastic (no
resistance to bending), the tension is way more than gravity, and string only vibrates perfectly
up and down.
Non dispersive waves: anything that obeys the wave equation, e.g. a string, the speed of wave
propagation is constant and independent of frequency. All energy travels the same speed inde-
pendent of frequency.
T
A
is the wave speed.
_
T
A
is phase velocity. In beams non dispersive
waves, the high frequency waves travel ahead of the lower frequency waves.
1. Derive governing equation
(a) Momentum in x-direction gives T(x) is constant
(b) Do momentum in the y-direction
(c) Use small angles: sin( +

x
dx) = +

x
dx and tan() =
The governing equation is T

2
y
x
2
= A

2
y
t
2
2. Propose a general separable solution y(x, t) = a(x)f(t)
(a) Rearrange the governing equation as C
2
2
y
x
2
=

2
y
t
2
where C
2
=
T
A
and
propose f(t) = Ae
int
giving y(x, t) = a(x)Ae
int
and plug in
(b) The governing equation becomes C
2

2
a
x
2
+
2
n
a(x) = 0 note: C is al-
ways positive
(c) Propose a(x) = Be
ix
and get n = C
(d) The total solution is then y(x, t) = Be
i
n
C
x
Ae
int
which can be
decomposed into sine and cosine as y(x, t) = (B1 sin(x) +
B2 cos(x))(A1 sin(nt) + A2 cos(nt))
3. Apply boundary and initial conditions to get the constants
(a) Apply boundary conditions y(x = 0, t) = y(x = L, t) = 0 gives B2 = 0
and B1 sin(
n
C
L) = 0 so
n
C
L = n where n = 1, 2, 3 . . . . So n =
Cn
L
. The solution becomes y(x, t) = B1 sin(
n
C
x)(A1 sin(nt) +
A2 cos(nt))
(b) Apply initial conditions y(x, t = 0) = 0 gives A2 = 0 reducing solu-
tion to y(x, t) = B1 sin(
n
C
x)A1 sin(nt) or by combining the constants
y(x, t) = Cn sin(
n
C
x) sin(nt)
4. Now we have the governing equation, now we see if it is self-adjoint if it satises the
following conditions
i)

uAvdx =

vAudx
ii)

v
_
T

2
x
2
_
udx =

u
_
T

2
x
2
_
vdx
The rst condition is satised automatically, since u and v (in our case a(x) and f(t)
commute. We show that the second condition holds by doing integration by parts twice.

L
0
ai
.
u
_
T

2
x
2
_
ajdx
.
dv
= ai
.
u
_
T

x
(aj)
_
.
v

L
0
+

L
0
T

x
(aj)
.
v
dai
dx
dx
.
du
one more integration by parts

L
0
dai
dx
.
u
T

x
(aj)dx
.
dv
=
dai
dx
.
u
(Taj)
.
v

L
0
+

L
0
Taj
.
v
d
2
ai
dx
2
dx
.
du
gives

L
0
ai
_
T

2
x
2
_
ajdx = ai
_
T

x
(aj
_

L
0

_
dai
dx
(Taj)
_

L
0
+

L
0
aj
_
T

2
x
2
(ai)
_
dx
and since we evaluate the rst two terms on the right hand side at x = 0 and x = L,
the boundary conditions dictate that ai = aj = 0 here, thus proving the system is
self-adjoint. Self-adjointness depends on the boundary conditions.
5. Now we use the self adjoint property to show that the modes are orthogonal, where aj
and ai are orthogonal functions if they satisfy

L
0
ajaidx = 0
(a) Start with the governing equation for the spatial function for two different
modes ai and aj, where i = j.
C
2

2
a
x
2
+
2
n
a(x) = 0
(b) Since the governing equation equals zero, we can post-multiply each of these
expressions by the other mode, sum them, and it is still zero.
_
C
2

2
ai
x
2
+
2
n
ai(x)
_
aj +
_
C
2

2
aj
x
2
+
2
n
aj(x)
_
ai = 0
(c) Expand this expression and integrate from 0 to L. Using the self-adjoint prop-
erty which we just showed, we get
1
C
2
_

2
i

2
j
_

L
0
aiajdx = 0
(d) Since the natural frequencies corresponding to each of these modes is different,
the integral must be zero, satisfying the denition and showing the modes are
orthogonal.
6. To nd the i-th modal mass and i-th modal stiffness, write down the governing spatial
differential equation T

2
a
x
2
+A
2
n
a(x) = 0 for a mode ai and multiply both sides
by ai and dx, rearrange and integrate from 0 to L.
ai
_
T

2
ai
x
2
_
dx + A
2
n
a
2
i
(x)dx = 0
giving the i-th modal mass and i-th modal stiffness as

2
n
A

L
0
a
2
i
(x)dx
.
m
i

ij
=

L
0
ai
_
T

2
ai
x
2
_
dx
.
k
i

ij
And we can divide and solve for
R
, which are regular modes? which gives the Rayleigh
quotient.

2
R
=

L
0
ai
_
T

2
a
i
x
2
_
dx
A

L
0
a
2
i
(x)dx
=
kiij
miij
Modal Decomposition
When the string problem is forced, the governing equation is
A

2
y
t
2
T

2
y
x
2
= f0(x) sin t
We propose the same separable solution as before, where we make explicit that there are an
innite number of solutions, indexed by i, and the total solution is the sum
y(x, t) =

i
ai(x)fi(t)
Plugging this into the governing equation we get

i
_
Aai(x)

2
fi
t
2
fi(t)T

2
ai
x
2
_
= f0(x) sin t
Left multiply by ai and integrate across the beam from 0 to L

i
_

L
0
Aa
2
i
(x)

2
fi
t
2
dx

L
0
ai(x)T

2
ai
x
2
fi(t)dx
_
=

L
0
aif0(x) sin tdx
giving

i
_
_
_
_
_
_

2
fi
t
2

L
0
Aa
2
i
(x)dx
.
m
i

ij
+fi(t)

L
0
ai(x)T

2
ai
x
2
dx
.
k
i

ij
_
_
_
_
_
_
=

L
0
aif0(x) sin tdx
giving

2
fi
t
2
miij + kiijfi(t) = sin t

L
0
aif0(x)dx
Solving for fi(t)
kifi(t) = sin t

L
0
aif0(x)dx

2
fi
t
2
mi
fi(t) =

L
0
ai(x)f0(x)dx
ki
sin t

2
f
i
t
2
mi
ki
Boundary Conditions
Rollers at End with Spring
T
dy
dx
= ky
Daniel Wiese Dynamics Note Sheet 6
Euler-Bernoulli Beam Equation
Proposing Separable Solution
Given the governing PDE for a bending beam
A

2
y
t
2
= EI

4
y
x
4
we propose a solution of the form
y(x, t) = a(x)y(t)
Evaluating the necessary derivatives given this solution form we get

2
y
t
2
= a(x)
d
2
f
dt
2
and

4
y
x
4
=
d
4
a
dx
4
f(t)
substituting in
Aa(x)
d
2
f
dt
2
= EI
d
4
a
dx
4
f(t)
which can be separated as
A
1
f(t)
d
2
f
dt
2
= EI
1
a(x)
d
4
a
dx
4
= constant
And so now we can solve each ODE separately now. This leads to a solution of the form
y(x, t) = a(x) sin (t)
Plugging this back into the governing equation, we reduce the equation to an ODE and then
only have to nd a(x)
d
4
a
dx
4

4
a(x) = 0 where
4
=
A
2
EI
Self-Adjointness
Now we have the governing equation for beams, now we see if it is self-adjoint if it satises
the following conditions
i)

uAvdx =

vAudx
ii) EI

u
d
4
v
dx
4
dx = EI

v
d
4
u
dx
4
dx
where the integrals are evaluated from one end of the beam to the other, usually 0 to L. The
rst condition is trivial, and we can show the second condition holds by applying integration by
parts. Additionally, from doing this integration, we also nd the following relationship

L
0
v
d
4
u
dx
4
dx =

L
0
d
2
u
dx
2
d
2
v
dx
2
dx
Orthogonality
To show orthogonality of the modes, start with the spatial governing equation for two modes
ai and aj with i = j
EI
d
4
ai
dx
4
A
2
ai(x) = 0
EI
d
4
aj
dx
4
A
2
aj(x) = 0
Left multiply the rst equation by aj and the second by ai. Integrate across the beam (from 0
to L) and subtract
EI

L
0
aj
d
4
ai
dx
4
dxA
2

L
0
ajaidxEI

L
0
ai
d
4
aj
dx
4
dx+A
2

L
0
aiajdx = 0
Use self-adjointness to cancel out terms, giving
A
2

L
0
aiajdx = A
2

L
0
ajaidx
Thus showing the modes are orthogonal.
Finding i-th Modal Mass and Stiffness
To nd the i-th modal mass and stiffness, again use the spatial governing ODE for mode ai
EI
d
4
ai
dx
4
A
2
ai(x) = 0
left multiply by ai, and integrate across the beam from 0 to L
EI

L
0
ai
d
4
ai
dx
4
dx A
2

L
0
a
2
i
(x)dx = 0
use the additional property from self-adjointness to replace the fourth derivative as the product
of two second derivatives
EI

L
0
_
d
2
ai
dx
2
_
2
dx
.
k
i

ij

2
A

L
0
a
2
i
(x)dx
.
m
i

ij
= 0
where ki and mi are the i-th modal stiffness and mass, respectively. From this we can nd
as

2
=
kiij
miij
Boundary Conditions
Free End No moment, no shear.
Fixed or Clamped End Displacement and slope are zero.
Pinned End No displacement, no moment. Remember M
b
= EI

2
y
x
2
so for pinned end
this means the second derivative must be zero.
Point Mass at End No moment, external shear due to mass boundary condition, from con-
servation of linear momentum in y-direction.
Q = m

2
y
t
2
Applied moment From M
b
= EI

2
y
x
2
, the boundary condition due to Mapplied is

2
y
x
2
=
1
EI
Mapplied
Forced Beam Response
To solve the forced response, always solve the unforced problem rst.
y(x, t) =

i
ai(x)fi(t)

i
_
EI
d
4
ai
dx
4
fi + Aai
d
2
fi
dt
2
_
= f0 sin(t)

L
0
aj
_

i
_
EI
d
4
dx
4
fi + Aai
d
2
fi
dt
2
_
= f0 sin(t)
_
dx
using orthogonality of modes (ith modal mass and stiffness?)
kjfj + mj
d
2
fj
dt
2
=

L
0
ajf0 sin(t)dx
The solution is
fj(t) =

L
0
aj(x)dx
mj(
2
j
)
f0 sin(t) + Cj sin(jt + Bj)
Find Cj and Bj by applying initial conditions.