Physics Revision Notes Thermodynamics

Page 1 of 20

Dynamics Rigid Body Dynamics

Introduction

A rigid body is a many-particle system in which the distance between

particles is fixed. The location of all particles is described by 6
coordinates 3 spatial and 3 angular.

The velocity is determined by v, the velocity of the CoM and w , the

angular velocity.

The basic two equations of angular motion are

= F
MR
0

The centre of mass moves as if it were a single particle under the

action of a force F0.
J = G0

The rate of change of angular momentum is equal to the total applied

couple.

Other basic equations:

o The velocity v of a particle at a distance r from an axis around
which a rotation at speed w is happening is

v = w r
o For similar reasons:
dJ
= w J
dt

o Angular speeds are additive. To if frame 1 is rotating with

w1 wrt 2 with respect to frame 2, which is rotating with w2 wrt 3

with respect o frame 3, then
w1 wrt 3 = w1 wrt 2 + w2 wrt 3

Relating J and w

If the body is rotating at w , the total angular momentum is given by

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 2 of 20

J = r p
= r m(w r )
= m r 2w - (w r )r
= m r 2w - (wx x + wyy + wz z )r

In detail

m(y 2 + z 2 )
- mxy
- mxz

2
2

J = - mxy
m(x + z ) - myz w

- mxz
m(x 2 + y 2 )
- myz

J = Iw
[The non-diagonal elements are fairly easy to derive. The diagonal ones
should actually have x2 + y2 + z2, because one of the terms is always
In other words, J is

knocked out by the second term in the sum].

proportional to w , but not necessarily parallel to it.

The off-axes elements are rather hard to understand they correspond

to the fact that looking at a particle at a given instant, its impossible
to tell exactly around which axis its moving.

Also, we can find the kinetic energy

T = 12 m [(w r ) (w r )]
= 12 m [w r (w r )]

T = 21 w J

The couple is then given by

G = J = w J

Note that I must be specified with its origin and with its set of axes.

Properties of I

I is a symmetric tensor. It therefore has three real eigenvalues and

three perpendicular eigenvectors.

With respect to the eigenvector basis:

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 3 of 20

I = I 2

I
3

J a = I a wa
[No sum]
T = 12 I a wa2

[Sum]

The eigenvector axes are called the principal axes, and the Is are called
the principal moments of inertia.

An alternative way to think of this is that the principal axes are ones
around which objects are happy to rotate without any torque being
applied.

In w -space, surfaces of constant T form an ellipsoid, with axes of

length I a-1/ 2 . Also, in w -space:
grad T = I a wa = J
So J is perpendicular to surfaces of constant T at w .

We can classify the principal axes as follows:

o Spherical tops all the I are equal, and J = I w , with I scalar.

The body is isotropic with the same I about any axis (eg: sphere,

cube).
o Symmetrical tops I 1 = I 2 I 3 . e3 axis is unique, but e1 and e2

are any two mutually perpendicular vectors perpendicular to e3

(eg: lens, cigar).
o Asymmetrical tops all Is different, and axes are unique.

Consider any two Is:

I 1 + I 2 = m(y 2 + z 2 + x 2 + z 2 ) = I 3 + 2 mz 2 I 3

So no I can be larger than the sum of the other two. Furthermore, if

z = 0 for every particle (ie: if we have a lamina), then

I 3 = I1 + I 2

Consider an axis at a distance a away from a principal axis and parallel

to it, and let r be the distance of each particle from the principal axis.

Then:
I = m(r + a ) (r + a ) = I 0 + Ma 2 + 2

mr )
(

a = I 0 + Ma 2

= 0 when r measured
relative to C of M

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 4 of 20

This is the Parallel Axis Theorem, where each vector is considered to

be a projection in a plane perpendicular to the axes.

Two Basic Problems

You whack it what happens? Steps for solution:

o Define principal axes with a sensible origin.
o Calculate an expression for J in terms of the impulse:
J =

t dt = r

F dt = rB F dt = rB P

Where B is the point at which the whack occurred, and rB can

be taken out of the integral because the whack is assumed to be
instantaneous.
o Work out an expression for J in terms of w , using the moments

of inertia.
o Equate the two expressions for J.
o Work out the motion of the CM using standard linear mechanics.
o NOTE: The obvious origin to use is the CM, but other origins
can be used subject to the provisos above for using t = J . So a

pivot, for example, is fine to use.

You apply a torque whats the frequency of rotation?

o Define principal axes with a sensible origin (eg: the CM see

above).
o Find an expression for w in these axes (with unknown

magnitude), and find a corresponding expression for J, using the

principal moments of inertia.
o Find dL / dt = w L .
o Calculate the torque (= r F ) and equate it with dL / dt .

Free Motion Eulers Equation

Free precession is a situation in which F = 0 and G = 0. In such a

case, J is constant. w is constant if J is along one of the principal axes,

but otherwise, it will change direction, and perhaps even magnitude.

We use the Euler Equations to analyse this problem.

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 5 of 20

The rate of change of angular momentum vector in the principal-axes

frame (which is rotating around with the body) and the lab frame are
related by

dJ
dJ

=
+ w J
dt lab dt PA
Now, lets assume that a couple G is being applied in the lab frame.

We know that
dJ
G=

dt lab

Therefore, using the equations above:

dJ
G=
+ w J
dt PA

Finally, we note that in the principal axes frame, J = (I 1w1, I 2w2, I 3w3 ) .
Therefore, casting both sides of this equation into the principal axes
frame only

t1 = I 1w 1 + (I 3 - I 2 )w3w2
And similarly with any cyclic permutation of indices.

A few notes:
o All the quantities in this equation are measured with respect to

the body frame (which is moving). This is the advantage of

these equations all we have to consider is the forces that the
body feels.
o The two terms of the RHS refer to two types of ways J can

change because it can change in the body frame and also

because the body frame is itself rotating.

Free Motion Examples

FREE SYMMETRIC TOP

o For a symmetrical top (I1 = I2 = I) which is free in space (ie: no
torque) the Euler Equations become
I w 1 + (I 3 - I )w3w2 = 0
I w 2 + (I - I 3 )w1w3 = 0
I 3w 3 = 0

o The last equation implies w3 is constant. Lets define

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 6 of 20

I3 - I
w3
I
Then the general solution of the first two equations becomes:
W=

cos[Wt + f ]
w1
= A

cos[Wt + f ]
w2

o Interpretation from the body frame

In the body frame, w1 and w2 seem to form a circle in

the xy plane, with frequency W . How high that circle is
depends on w3 .

L could be above w (if I3 > I an oblate top) or below

w (if I3 < I a prolate top).

o Interpretation from the fixed lab frame

In that case, the Euler Equations are useless, because

they deal with the body frame, so we express things from
scratch, but in terms of the body frames:
w = (w1x1 + w2x2 ) + w3x3
L = I (w1x1 + w2x2 ) + I 3w3x3

w=

L
- Wx3
I

With W defined as above.

This linear relationship between w , L and x3 implies

that they are in the same plane.

Furthermore, the rate of change of x3 is w x3 , because

it only changes as a result of the rotation. So

L
dx3 L
= - Wx3 x3 = x3
I

I
dt
This is equivalent to x3 rotating at a frequency L / I .

It turns out that we can interpret w as follows

L
I

Motion of body about

its own axis

Motion of body around L

w=

Wx3

HEAVY SYMMETRIC TOP

o Here, we must define the Euler angles as follows

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

f

Page 7 of 20
x3

x2

o The total angular velocity is then given by

y x3

Rotation of top

w=

+ qx1 + fz

Motion of top itself

Which can be expressed in terms of the body-frames only:

w = y x3 + qx1 + f (x3 cos q + x2 sin q )

w = (y + f cos q ) x3 + (f sin q ) x2 + qx1

o

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 8 of 20

Dynamics Normal Modes

Introduction

A normal mode of a system is an oscillation that has a single frequency.

All the more general oscillations of the system can be expressed as

superpositions of these normal modes.

General approach

Consider a system defined by generalised coordinates qi and acted on

by forces Fi, moving in a potential well U(x), and moving elastically.

The kinetic energy, T, is then given by

T =
Where

x (q )
j

1
2

2
x j (qi )

is the Cartesian coordinate of the jth part of the

system, taken about an equilibrium, where all the x j are 0. Expanding

about that equilibrium:

x (q ) = x (q
j

i ,eq

)+

x j
q i

x j

x (q ) q
j

eq

qi

qi +

i eq

And so:
T =

1
2

M q q

ij i j

= 21 qT Mq

Where
M ij = m

r
qi

eq

r
q j

eq

Consider the potential energy, about a point of equilibrium (ie: a

minimum in U) at which all the qi are chosen to be 0.
U
1 d2U
U (x ) = U 0 +
qi +
qi Eq
2 dx j dx i

qiq j +
x0

0 since at a minimum

1
Kijqiq j +
2
1
U (x ) = U 0 + qT Kq
2

U (x ) = U 0 +

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 9 of 20

The total energy is then

1
1
M ijqiq j + K ijqiq j

2
2
dE
1
= 2qi (M ijqj + K ijq j ) = 0
dt
2
dE
= qi (M ijqj + K ijq j ) = 0
dt
[Non rigorous argument] the equations of motion are then:
E =U0 +

M q + K q
ij j

ij j

=0

Mq + Kq = 0

If we seek normal modes of the form q (t ) = Qe i wt , we get:

(K - w 2M )Q = 0
Non-trivial solutions only exist if
det (K - w 2M ) = 0

This defines the w 2 normal mode frequencies.

In practice, the steps are:

o Find the K and M matrices by writing them out in terms of the

variables of the system, and comparing with

1
T = 12 qT Mq
U = U 0 + qT Kq
2
Both matrices must be symmetric.
o Use the determinant method above.

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 10 of 20

Dynamics Elasticity
Introduction

Hookes Law states that

Strain

F
Dl
=E
A
l

Stress

Where
o F is the force applied to a block of material over an area A.
o

Dl is the extension of the block in the direction of F.

o l is the original, relaxed length of the block in that direction.

o E is the Youngs Modulus of the material.

Furthermore, it states that

Dw
Dl
= -s
w
l
Where Dw is the length of the block in any direction perpendicular to
that of l.

For an isotropic material, E and s are all we need to define the elastic
properties of the material.

Since these equations are all linear, the principle of superposition

applies. If we have several forces, the displacements will be the sum of
the displacements with the forces acting individually.

Uniform Strain the Bulk Modulus

Consider a rectangular block in a pressure tank, say, with identical

stress p on every face.

Consider one direction the change in length Dl in that direction is

given by
Due to pressure
in that direction

Due to pressure in
other directions

p
p
p
+ s +s
E
E
E
Dl
1 - 2s
=p
l
E
The problem is symmetrical, so the value will be the same for all
Dl
=
l

directions.

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 11 of 20

Now, consider the change in volume

DV
Dx Dy Dz
=
+
+
V
x
y
z

We therefore have

DV
1 - 2s
p
= -3
V
E
We can then define the bulk modulus
K =

E
3(1 - 2s)

Such that the change of volume as a result of the stress p is

p = -K

DV
V

Shear Strain the Shear Modulus

Consider a cube with face area A and with shear forces acting on it

F
A
F

F
d

B
F

If cut the cube along the diagonals A and B, we find that

o There is a stretch normal to A, of magnitude F 2 .
o There is a compression normal to B, of magnitude F 2 .

And each of these diagonal faces has area A 2 .

The lengthening of the diagonal d will therefore be equal to the

lengthening of d in the following case:

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 12 of 20

F 2
F 2

F 2
F 2
From above, this is given by:
Dd
1F 2
1F 2
=
+s
d
E A 2
E A 2
Dd
1+s F
=
d
E A
By symmetry, the other diagonal is shortened by the same amount.

It is often useful to have this as a function of the twist angle:

Dd
F

d
d
q

From this diagram, it is (reasonably) clear that

d = Dd 2

d= 2

Therefore
d
Dd 2
Dd
2(1 + s) F
=
=2
=
l
l
d
E
A
We therefore define the shear modulus as
q

m=

E
2(1 + s)

Such that

g = mq
Where g is the shear stress = F/A.

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 13 of 20

Formal Definitions

Stress
o Defined in terms of force/unit area transmitted across planes

in the medium.
o Requires a tensor. We define

t ij

Force in the i

On a plane perpendicular

direction

to the j axis

o We can then show that the force on any arbitrary area element

is
F = t dS

o The tensor must be symmetric consider a small cube side dx.

Because the cube must be in equilibrium, the forces on it are as

follows:

Sxy
Syx

Syx
Sxy

The net couple on the cube is

(Sxy - Syx ) dx
But there must be no torque on the cube, or itd spin! So
S xy = Syx

o The stress tensor is diagonal for suitable choices of axes.

o The stress in a solid material is therefore described by a tensor
field.

Strain
o When a material is put under strain, a point (x , y, z ) in it is

moved to a point (x + X , y + Y , z + Z ) .
o The derivatives of these X, Y and Z contain information about

the strain.
o As we saw before, its worth considering two kinds of strain

For the normal strains, we define:

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

exx =

X
x

Page 14 of 20
eyy =

Y
y

ezz =

Z
z

For example, if we consider stress perpendicular to the x

direction in a cube initially of side Dx , itll increase by

exx Dx :
exx Dx

Dx

Now, for the shear stresses, consider

X

Dy

X
y

Y
x

Dx

[The expression for the angles are tricky to see but

consider that X is the change in x] We then simply
define
Y X
exy = eyx = 12
+

x
y

This ensures that if the block simply rotates (ie:

Y / y = X / x ), these strains are 0.
o So in general, we define

eij =

1
2

(X j / x i + Xi / x j )

The ith coordinate of a

point in the material

will have changed by eijD, assuming that

the jth coordinate of that point from the
origin is D.

So, for example

X = exx x + exyy + exz z

o The tensor is also symmetric, due to the exy = eyx condition.

o If the strains are non-homogenous, we sit down and cry.

The relation between them

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 15 of 20

o Each component of e is related to each component of t this

gives, overall, a fourth-rank tensor of elasticity relating the two:

tij = C ijklekl

(Using the summation convention).

o It looks like there are 92 = 81 coefficients in C, and that 81
numbers are therefore required to define the elastic properties of

a material! However, we note that since S and e are symmetry,

we must be able to swap ij and kl in C without changing a thing,
so there can be at most 36 different coefficients.
o If the material is isotropic, though, C must be completely frameindependent. As such, we must be able to express it in terms of

the tensor dij . There are only two ways of doing this that are
also invariant under i j and l k , and so

C ijkl = l (dij dkl ) + m (dik djl + dil djk )

So an isotropic material only requires two constants (E and s ,
for example). And we have
Sij = lekk dij + 2meij

Examples Statics

Thin tube in torsion

o Consider a thin tube being twisted an angle f

Dr

L
o We first note that
rf
l
o Next, consider a small square (dotted above) and its deformation
q=

as a result of the twist:

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 16 of 20

From the previous result:

F
= mq
Dr
rf
Dr
F =m
L
o This force contributes a torque Dt to the rod
r 2f
Dr
L
o Considering these bits around the whole rod, so that 2pr ,
Dt = rF = m

we get
t = 2pm

r 3Dr
f
L

Wire in torsion

For a wire, we simply integrate the above from r = 0 to the total

radius, giving
pr 4
t=m
f
2L

Can under pressure

o Consider a can of thickness t with closed ends with an internal
pressure p.
o Let the tangential stress in the walls be tq , and consider half

the can

tq

tq

t
The forces (= stress area) must balance, so
t q 2t = p 2r
pr
tq =
t
o Let the axial stress in the walls be sz , and consider one of the
ends. By the same logic as above

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 17 of 20

tz 2prt = p pr 2
pr
2t

tz =

Bent beam
o Consider a beam of length L, held in a bent position.
o We only consider longitudinal strains (valid for small deflections

and thin beams).

o Clearly, the bits at the top of the beam will be stretched, while

those at the bottom will be compressed. Somewhere in between,

therell be a neutral surface neither stretched nor compressed.
o Consider a small segment length of the bent beam:
q

+ D

y
Neutral surface

o The amount of stretching and compression at any point is

proportional to the distance from the neutral surface, y. The
constant of proportionality is /R . As such

D
y
= Strain =

R
o Clearly, therell be forces to the left above the neutral surface,

and vice versa. We therefore have

DF
D
=E
DA

E
DF = y DA
R
o The total torque produced about the neutral line is given by
t =
=

Cross
section

y dF

E
2
Cross y dA

R section
B=

EI
R

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 18 of 20

o Now, consider a beam loaded with weights given by W(x),

where W is the force per unit length. Consider the statics of a

small segment of the beam:

S + dS

dx
S

dF
Notes:

Due to the bending moment, some vertical forces are

produced. Ignoring products of infinitesimal quantities,
we can write, at that point
S dx = d B
EI
= S dx
R
[Effectively, were saying that due to the dS needed to
B=

balance dF, the bending moment must change]

The downwards loading force needs to be balanced by a

difference in the upwards stress
d S = dF = W d x
S =W

o Now, for small deflections

y = 1/ R

o As such, we can conclude

EIy = W (x )

o Boundary conditions for various cases are as follows

At a free end, S and B are clearly 0, and so y = y = 0 .

At a cantilevered end, y and y are given (usually 0).

o Finding y is then simply a question of solving that differential

equation. However, there are a few tricky points

All forces must be considered when writing down W(x),

including reactions at contacts. Most often, W will be a

series of d -functions.

Sign conventions:

Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 19 of 20

Downwards W positive.

The resulting y obtained is downwards positive,

because the way the radius of curvature is specified.

However, be very careful sometimes, the

convention appears to be reversed because the bar
curves downwards, and so -1/ R = y .

Dont worry too much about boundary conditions for y

just integrate d -functions from 0 to L (for a free end,
this is fully justified). Remember that therell often be a

d -functions at the very end of the range, which might

help satisfy the boundary conditions.

From then on, boundaries are just provided. Just also

remember to make the y , y and y continuous.

The couple provided by a cantilever can simply be

worked out by evaluating B = EIy at that point.

It is sometimes easier to simply write down y , the

bending moment from physical considerations.

o The Euler Strut is a beam buckled between two walls:

If we take y upwards, then the bending moment on any point is

B = -Fy
F
y
EI

F
y = A sin x
EI

Applying the boundary condition that y = 0 at x = L:

y = -

p 2EI
L2
This is independent of displacement (but only while y = 1/ R
F=

holds).
o The Reciprocity Theorem states that

The deflection at Q due to a load at P is the

same as the deflection at P due to the same
load at Q
Daniel Guetta, 2008

Physics Revision Notes Thermodynamics

Page 20 of 20

To prove, say PPQ means the deflection at P due to the load at

Q. Consider loading first P and then Q. The energy stored is
P

P
E = F PP + QQ + PPQ
2

The same result must be applied the other way round, so

PPQ = PQP

Dynamics of Rigid Bodies

Consider a small volume V of the material. It will have both external

forces acting on it (eg: gravity) and internal forces (eg: elastic stresses).
Fext + Fint =

rr dV

Every small particle in the volume experiences the external force,

though, so Fext is given by a volume integral.
Fint =

Define this = f

(-fext + rr)dV

Fint =

f dV

On the other hand, only the particles at the edge of the volume
experience the elastic force from surrounding media, and so Fint is given
by an area integral

fint dA =

f dV

We have, however, defined that the force in the x-direction, say, is

dFx = (S xx i + S xy j + S xzk ) dA

And so, taking only the x component of the integral above

(S
A

xx

i + S xy j + S xzk ) dA =

Using the Divergence Theorem on the LHS

S
S xy S
V xxx + y + zxz dV =

fx dV

fx dV

Removing the volume integrals (because this is true for any volume):
fi = Sij / x j

(Using the summation convention).

Now, using tij = lekk dij + 2meij (isotropic material), we obtain

f = (l + m)( u ) + m2u

Where u is the internal displacement in the solid.

Daniel Guetta, 2008