Professional Documents
Culture Documents
2 Rot Kin 1 PDF
2 Rot Kin 1 PDF
(MEAM535)
Chapter 2
Michael A. Carchidi
February 13, 2017
The following notes are written by Dr. Michael A. Carchidi at the University
of Pennsylvania for the purpose of teaching the contents covered in the MEAM535
- Advanced Dynamics Course.
Sections 1, 2 and 3 are meant as a review for those students that require a
review of some basic ideas in undergraduate dynamics. For those that do not
require this review, you may skip to Section 4.
A body is called rigid if the distance between any two points that make up the
body is fixed in time, even when the body is in motion. A drop of water would
not be considered rigid whereas a steel ball would be considered rigid. A rigid
body is called linear if all points of the body lie on a line. An example of a linear
rigid body would be a thin straight rod. A rigid body is called planar if all points
of the body lie on a plane. Examples of planar rigid bodies are thin disks, rings,
and any linear rigid body such as a rod. A ball (solid or hollow) would not be a
linear nor planar rigid body.
Ideal constructs are not unusual in the study of dynamics (or science in general)
and they are used as approximations to real physical constructs in order to simply
the mathematics that is required to describe the science. For example, other ideal
constructs that are used in the study of dynamics are: massless springs, massless
and frictionless pulleys and frictionless surfaces, to name just a few. If you are
familiar with the study of thermodynamics, you may recognize that a Carnot cycle
is an ideal construct. The student should try to think of other ideal constructs in
dynamics and other areas of science and engineering.
Ideal constructs are use as models for real physical objects because the math-
ematics needed to describe an ideal object is much simpler than that needed to
describe real objects. Of course, the representation of any real physical object
by an ideal construct is valid only when this representation has a small effect on
the comparison between the theory and experiments. For example, it is common
to model the earth as a point particle when predicting properties of its elliptical
path around the sun, since the internal motions of the earth (i.e., its rotation
about its axis), does not have a large effect on this part of the earth’s motion, and
comparison between the model calculations and the actual measurements show
this to be true. On the other hand, we could not model the earth as a point
particle when it comes to calculating weather patterns, since here the rotation of
the earth about its axis has a huge effect on the weather.
Thus, at least as far as much dynamics is concerned, we can still model many
objects as point particles, or we can model ideal springs and ideal pulleys as
2
massless and frictionless and we may also model many bodies, such as billiard
balls and steel blocks, as being rigid. For example, if two billiard balls were to
collide, both moving at reasonably slow speeds so as not to effect the shape of
each ball, then the rigid body assumption would be valid in studying the effects
of the collision.
Degrees of Freedom
Because the distance between any two points on a rigid body does not change
in time, it makes the analysis of rigid bodies much simpler than the analysis of a
non-rigid body, such as a drop of water. In dynamics, we assume that the rigid-
body approximation is valid and we shall see that the motion of a rigid body in
space has only 6 degrees of freedom, as oppose to a non-rigid body that has an
infinite number of degrees of freedom. The study of non-rigid bodies is the topic
of Continuum Mechanics and Fluid Mechanics.
The six degrees of freedom commonly used to describe the motion of a rigid
body in space consists of the three coordinates needed to describe the position of
the body’s center of mass (xG , yG , zG ) and the body’s orientation about its center
of mass, which (we shall see later) can be described using three angles (α, β, γ)
known as Euler Angles.
For a special type of rigid body motion, known as planar motion, we shall see
that only three degrees of freedom will be needed to describe its motion. These
will consists of a translational motion of one point on the body (usually the body’s
center of mass), which requires two degrees of freedom (xG , yG ), and the rotational
motion of the body about an axis perpendicular to the planar motion of the body
(θ) which requires only one degree of freedom.
It is important that the student not confuse planar motion of a rigid body
with a planar rigid body. A planar rigid body is one in which all points of the
body, at any instant in time, lie on a plane. This means that the thickness of the
body has a small effect on the body’s motion through space and so it is ignored. A
disk, ring, or a even a tennis racket can be viewed as a planar body, even though
its motion could be far from planar. For example, think about the non-planar
3
motion of a coin when it is flipped into the air.
On the other hand, a rigid body undergoing planar motion means that any
given point on the body moves along a fixed plane throughout the body’s motion.
A steel ball, which is not a planar rigid body, undergoes planar motion when
rolling without slipping, down an incline because any given point on the ball
moves along a plane parallel to the ball’s direction of motion as the ball rolls
down the incline. A more detailed discussion of these ideas will be discussed later
in the course. For now, let’s consider the type of qualitative discussion that you
would find in an undergraduate course in dynamics. If you are already familiar
with this, you may skip to Section 5.
A particle’s motion is called planar if all points along the path of the particle
lie on a fixed plane. You have seen in basic physics that the projectile motion of a
point particle is an example of planar motion since the path of motion lies along a
plane determined by the particle’s initial velocity vector (v0 ) and its acceleration
vector (a), as shown in the figure below.
A rigid body is said to undergo planar motion if any given point on the body
is restricted to move along a fixed plane. Note that for this type of motion, any
4
two points on the rigid body will either: (a) move along the same plane, like in
the case of a ring rolling down an incline or (b) they will move along two different,
but parallel planes, like in the case of a ball rolling down an incline. The collection
of these parallel planes (even if it is just a single plane) will be called the planes
of motion for the rigid body, and all these different planes will remain parallel to
each other throughout the motion. A collection of planes that are parallel to each
other are also all perpendicular to some single fixed unit vector n b . For example,
all 20 floors of a level 20-story building are a collection of 20 parallel planes all
perpendicular to a single fixed unit vector n b that points straight up from the
ground.
The figure below shows a disk (or a ring) rolling down a fixed incline.
This is an example of planar motion since all points on the disk (or ring) are
moving along paths that lie on the plane of the disk, which is the same as the
plane of the paper that the figure is drawn on. In this case there is a single plane
of motion and this plane is the plane of the paper and this plane is perpendicular
to a unit vector coming out of (or going into) the page.
If a hollow ball (a non-planar object) rolls down a fixed incline (as shown
below), then only those points on the ball that actually touch the incline sometime
5
during the motion, move along the plane of the paper.
Other points on the ball move along different planes but all these different planes
are parallel to the plane of the paper and are all perpendicular to the same fixed
unit vector coming out of (or going into) the page. Thus, a ball (hollow or solid)
rolling down a fixed incline is also an example of rigid body planar motion, but
unlike the rolling disk (or ring) that has a single plane of motion, the ball has
an infinite number of planes of motion, all parallel to the plane of the paper and
all perpendicular to the same fixed unit vector coming out of (or going into) the
page.
Once again, it is important that the student understands the different between
a non-planar and a planar rigid body and the difference between non-planar and
planar rigid-body motion. A non-planar rigid body (such as the hollow ball above)
or a planar rigid body (such as the disk above) could both exhibit planar motion
while rolling down a fixed incline. In addition, a non-planar rigid body (such
as a rocket) or a planar rigid body (such as a tennis racket) could both exhibit
non-planar motion. A non-planar rocket launched into space exhibits non-planar
motion. In addition, think about the complicated non-planar motion that could
result after throwing a tennis racket (a planar rigid body) into the air.
If one imagines a rigid body undergoing planar motion, we find that there
are three basic types of planar motion that are possible. These are: (a) purely
6
b (a unit vector
translational motion only, (b) purely rotational motion only about n
parallel to the planes of motion), and (c) a combination of these. We now discuss
each of these in a qualitative way.
For purely translational planar motion of a rigid body, all points on the body
move along paths that are parallel to each other. For example, the figure below
shows a rigid rod moving in the plane of the paper along a family of parallel curves
(shown dotted in the figure) by translating to the right and up along these curves.
Throughout this type of motion, the rigid body (the rod here) remains parallel to
its initial orientation. Thus, to track this type of motion, we need only the initial
orientation of the body and the motion of just one point on the body (usually
7
taken to be the body’s center of mass) as shown in the figure below.
To describe the orientation of a rigid body undergoing planar motion, one may
simply pick any two distinct points on the body, (e.g., the body’s center of mass
point G and any other point P), and draw the vector (PG) connecting these two
points. Then consider any constant unit vector u b that does not lie along the line
PG and is parallel to all the planes of the motion. The angle that PG makes with
b can then be used to describe the orientation of the body.
u
For example, the figure below uses P as the top of the rod, so that the line
GP points along the rod, and the figure uses u b (dotted in the figure) as the fixed
direction ”to the right” and parallel to (or in this case lying in) the plane of motion
(i.e., the plane of paper here). The angle between the vector GP (along the rod)
and u b is then used to describe the orientation of the rod and note that this angle
8
does not change as the rod translates.
For purely translation motion of a planar rigid body (like the rod above), once
the position of the body’s center of mass (G) is known, the position of any other
point on the body (P) is determined from the body’s orientation and the position
of this point as measured from the body’s center of mass point. For example, in
the rod example above. If the orientation angle is θ and the length of the rod is L
and if the center of mass of the rod is located at (xG , yG ), then, using rectangular
coordinates, the position of the top right part of the rod is
rtop right/O = rtop right/G + rG/O
where rA/B is one common notation giving the position of A as measured by B.
This results in
µ ¶
L L
(xtop right , ytop right ) = xG + cos(θ), yG + sin(θ)
2 2
9
and the position of the bottom left part of the rod is
resulting in
µ ¶
L L
(xbottom left , ybottom left ) = xG − cos(θ), yG − sin(θ) .
2 2
Purely rotational planar motion of a rigid body occurs when every point of the
body undergoes circular motion about centers that lie on a fixed axis of rotation.
The figure below shows a rigid rod rotating about an axis passing through the
origin (O) and perpendicular to the planes of motion and coming out or (or going
into) the page. The rod’s center of mass moves along the thin solid circle.
The dotted lines coming from the origin show the position of the rod’s center of
mass at two points during the rotation. Note that for the special case of a planar
rigid body undergoing purely rotational motion about a fixed axis (like that in
the figure above), all points on the rigid body move along concentric circles with
the common center on the axis of rotation and in the plane of the body.
10
General Planar Motion of a Rigid Body
From undergraduate dynamics you should recall that all instantaneous planar
motion of a rigid body can be viewed as a pure translation followed by a pure
rotation, or as a pure rotation followed by a pure translation, whichever you prefer.
Consider the instantaneous planar motion of a rod shown in the figure below.
The figures below show the same motion of the rod at two points in time as
represented by a pure translation (of the bold rod on the left to the thin rod on
the right) followed by a pure rotation (of the thin right on the right to the bold
rod on the right).
A pure translation
followed by a pure rotation
The motion is from left to right
11
We may also represent this as a pure rotation (of the bold rod on the left to the
thin rod on the left) followed by a pure translation (of the thin rod on the left to
the bold rod on the right).
A pure rotation
followed by a pure translation
The motion is from left to right
They both start with the same position and orientation (i.e., the bold line on the
left pointing up and to the right) and they both lead to the same position and
orientation at the end (the bold line on the right pointing up and to the left), as
shown below.
12
3. Planar Motion of a Rigid Body: Quantitative Discussion of Velocity
Let N be some reference frame, and let A and B be any two points that are
fixed on a rigid body undergoing purely translational motion. Then using the
relative motion equation for the points A and B, we may write that
N
vB = N
v A + A vB (1)
where N vB and N vA are the velocities of points A and B, as measured by the
reference frame N, and A vB is the velocity of point B as measured by point A.
Remember that when it comes to velocities, we need only specify the frame in
which the measurements are being made. We need not specify who in this frame is
doing the measurement since everyone in the frame will measure the same velocity.
To see this, suppose that observers O and O0 are in the frame N, which means
that O and O0 are both fixed in N. If O and O0 are both measuring the velocity
of a point P moving in N. e.g., both observers are standing on the ground (frame
N) and watching a car (point P) moving by, then
O 0 O0 O0
vP = O
vO + vP = vP (2)
since both O and O0 being fixed in N means the velocity of O0 as measured by O
0
is zero, i.e., O vO = 0.
Now if A rB is the vector pointing from point A on the rigid body to point B
on the rigid body, then
A A B
A B d( r )
v = (3)
dt
is the rate of change of point B as measured by an observer fixed at point A. Since
the body is rigid, the distance between points A and B on the body is constant and
13
so |A rB | is fixed. Consider the figure below which shows a rigid body translating
along the dashed line shown. The dotted lines show the vector A rB connecting
points A and B on the body during the translation.
It is clear that if only purely translational motion is present, as in this case, the
direction of A rB is also fixed and thus the vector A rB is not changing in time as
the body translates and so we must have A vB = 0. Thus we see that
N
vB = N
vA (4)
which shows that under purely translational motion, all points on the rigid body
move with the same velocity, as measured in the reference frame N.
Again, let N be some reference frame and let A and B be two points that
are fixed on a rigid body R that this time is undergoing purely rotational motion
about some fixed axis O passing through point A and perpendicular to the planes
of motion. In addition, let N ωR be the angular velocity of this rigid body about
O, as measured in N. Note that N ωR is measured by the reference frame N that is
watching the body rotate about the fixed axis passing through point A and this
vector N ω R points along the axis O, and we shall assume that this vector satisfies
the following right-hand rule.
14
The Right-Hand Rule For Pure Rotations About An Axis
If the fingers of the right hand are curled to follow the rotational motion of
the rigid body (as viewed in N), then the thumb of this hand will point in the
direction of N ω R . Thus a rigid body rotating counterclockwise in the xy plane will
have
N R
ω = +ωk b with ω>0 (5)
while a rigid body rotating clockwise in the xy plane will have
N b
ω R = −ω k with ω>0 (6)
b of course, points along the positive z axis, and ω is the angular speed of
where k,
the rotation (a scalar). Note that in both cases we may simply write
N b
ωR = ωk (7)
and call ω the angular velocity (not angular speed) and allow ω to be positive for
counterclockwise motion and negative for clockwise motion.
If A rB is the vector pointing from point A on the rigid body to point B on the
rigid body, then
A A B
A B d( r )
v = (8)
dt
is the rate of change of point B as measured by an observer fixed at point A. Since
the length of A rB (which is the distance between points A and B on the rigid body)
is fixed for a rigid body, and since the rigid body is undergoing purely rotational
motion, an observer fixed at point A must see point B moving along a circle having
its center at point A and having its radius equal to the distance from B to the axis
passing through A. Moreover, point B must be moving around its circle of motion
at the angular rate |ω|. This is illustrated in the figure below where point A is the
center of the figure and the rigid body is an ellipsoid rotating counterclockwise
15
about an axis perpendicular to the page and passing through point A.
It should be clear under this type of motion that point A sees all points on the
rigid body moving along circles all having their centers on the axis of rotation and
having their radii equal to the distance from these points to the axis of rotation,
and since the body is assumed to be rigid, these points must all be moving around
their respective circles at the same angular rate |ω|. If they did not, the shape of
the rigid body would have to change with every rotation and this is not allowed
because the body is assumed to be rigid.
Thus, the point B on the rigid body must move at an angular rate |ω| along a
circle having radius
A B
R = |A rB sin(ϕ)| (9)
and center on the axis of rotation, where ϕ is the angle between A rB and the axis
b be a unit vector (so that |b
of rotation. If we let n n| = 1) pointing along the axis
of rotation and satisfying the above right-hand rule, then we may write the above
radius as
A B
R = |A rB || sin(ϕ)| = |b
n||A rB || sin(ϕ)| = |b
n × A rB | (10)
The speed of this point as measured by point A must then be
A B A
v = RB |ω| = |ω||b
n × A rB | = |ωb
n × A rB |
16
and since the direction of motion of point B (as measured by A) must be tangent
to the circle that B moves along and in the direction of rotation, and since ω = ωb
n,
we may write that
A B
v = |A vB | = |N ωR × A rB | (11)
and, in fact
A
A B d(A rB ) N
v = = ω R × A rB . (12)
dt
We want to determine the velocity of the center of the rod (R) as measured by
the fixed xyz frame. Choosing A as the origin and G as the center of the rod, we
have µ ¶
O G L L b
r = D + cos(ϕ) b er + sin(ϕ)k
2 2
17
where ber is the polar unit vector pointing radially outward from the z axis. We
also have pure rotation about the z axis with
O b
ωR = ωk
Note that
L 4
D+ cos(ϕ) = 3 + cos(60◦ ) = 4 m
2 2
is the radius of the circle followed by the center of the rod as the rod rotates about
the z axis, and this is shown as the larger dotted path in the figure above. ¥
Although it is not true that all finite planar motion of a rigid body (R) can
be viewed as a pure translation followed by a pure rotation (or vice versa), it
is true that all instantaneous planar motion of a rigid body can be viewed as a
pure translation followed by a pure rotation (or vice versa). Since velocity is an
instantaneous rate of change, we may say that the instantaneous velocity of a rigid
body undergoing planar motion can be written as a sum of a translational part
and the rotational part. Thus we may put Equation (3) into Equation (1), and
18
get the complete velocity of a point B on a rigid body undergoing planar motion
as
N B
v = N vA + A vB = N vA + N ω R × A rB (13)
where N vB , N vA and N ω R are all measured from the reference frame N, and A rB
and
A B
v = N ωR × A rB
are measured from point A. Note that observers in N see point A of the planar
rigid body translating with velocity N vA , and they also see the rigid body rotating
with angular velocity ω about an axis perpendicular to the rigid body and passing
through the point A. Note that purely translation motion is equivalent to having
N R
ω = 0 while purely rotational motion about a fixed axis passing through point
A is equivalent to having N vA = 0.
A rod of length L = 4 meters is thrown in the air so that its center follows
the path of a projectile with an initial speed of v0 = 3 m/s and initial angle of
ϕ = 30◦ , and the rod rotates counterclockwise with constant angular speed of
ω = 2 rad/sec about an axis passing through its center and perpendicular to the
plane of the rod’s motion as shown in the figure below.
We want to determine the velocity of the top of the rod at time t (as measured
by the ground N), assuming that the rod is rotating counterclockwise and the rod
19
started out in the vertical orientation. We start with the equation
N
vT = N
vG + N ω R × G rT
with
b L L
N
ωR = ωk and G T
r = cos(ωt + ϕ)bi + sin(ωt + ϕ)bj
2 2
This last equation follows since at time t, the rod, which starts at an angle ϕ
would have rotated counterclockwise by ωt. In addition, since the center of the
rod exhibits projectile motion, we have
N
vG = v0 cos(ϕ)bi + (v0 sin(ϕ) − gt)bj.
Thus
N
vT = v0 cos(ϕ)bi + (v0 sin(ϕ) − gt)bj
µ ¶
b L b L b
+(ω k) × cos(ωt + ϕ)i + sin(ωt + ϕ)j
2 2
= v0 cos(ϕ)bi + (v0 sin(ϕ) − gt)bj
Lω ³ b b b b
´
+ cos(ωt + ϕ)(k × i) + sin(ωt + ϕ)(k × j)
2
= v0 cos(ϕ)bi + (v0 sin(ϕ) − gt)bj
Lω ³ b b
´
+ cos(ωt + ϕ)j + sin(ωt + ϕ)(−i)
2
or simply
µ ¶
1
N
v T
= v0 cos(ϕ) − ωL sin(ωt + ϕ) bi
2
µ ¶
1
+ v0 sin(ϕ) − gt + ωL cos(ωt + ϕ) bj
2
and this gives the velocity of the top of the rod as measured by the inertial frame
(the ground). ¥
20
Example #3 - Rolling Without Slipping
When a uniform rigid object of radius R rolls without slipping, its center of
mass moves with a speed v and the body rotates with angular speed ω about
an axis passing through the center of mass G and perpendicular to the object’s
planes of motion. Let G be the center of mass of the rolling object and let B be
the point on the object that is momentarily in contact with the floor. Suppose
the object is moving along the floor (F) from left to right. Then
F b
ωR = −ω k with ω>0 and b
k
v = vbi
F G
where bi
points to the right (positive x). Now it is clear that since point B is only rotating,
as seen by G, we must have
G b × (−Rbj) = −Rωbi.
vB = F ω R × G rB = (−ωk)
Then using the velocity transformation equation, we may write that the velocity
of point B as measured by the floor (F) is
F
vB = F vG + G vB = −Rωbi + vbi = (v − Rω)bi.
resulting in v = Rω. If you are not convinced that a point on the rolling object
that is in contact with floor is fixed in F at this instant in time, think of tire tracks
in mud. If the wheel rolls in the mud without slipping, these tracks are very clear
and they nicely show the tread of the tires. Only if there is slippage do the tracks
become smeared.
21
Therefore, if a uniform rigid body of radius R rolls without slipping so that
its center moves with speed v and it rotates with angular speed ω about an axis
passing through the center and perpendicular to the planes of the motion, it must
be true that
v = Rω (Rolling Without Slipping) (14a)
We may differentiate both sides of this with respect to time and also write
a = Rα (Rolling Without Slipping) (14b)
where a = dv/dt is the acceleration of the object’s center of mass and α = dω/dt
is the angular acceleration of the rolling object about the axis passing through its
center and perpendicular to the planes of the motion. More shall be said about
angular acceleration later in this note.
A wheel of radius b rolls to the right without slipping along a level floor so that
its center (point G) moves with constant velocity v (as measured by an observer
O fixed to the floor). Consider a point P that is a distance a < b from the center
of the wheel. If at time t = 0, this point is directly above the center of the wheel,
determine (in terms of a, b and v) the speed of this point at some later time t, as
measured by O. A picture of point P at some later time t is as shown below.
We assume that the wheel rolls without slipping with constant velocity v along a
horizontal floor and point P is a distance a < b from the center of the wheel so that
22
at time t = 0, this point is directly above the center of the wheel. Letting O ω W be
the angular velocity (as measured by O) of the wheel about an axis perpendicular
to the wheel and passing through its center, we must have
O b
ω W = −ω k with ω>0 and b
k
points to the right (positive x). Assuming that the center of the wheel is at the
point (0, b) at time t = 0, then, the position of point P as measured by the center
of the wheel (point G) is just
r = a sin(ωt)bi + a cos(ωt)bj
G P
where bj
or simply
O
vP = (v + aω cos(ωt))bi − aω sin(ωt)bj.
Then p
O P
v = |O v P | = (v + aω cos(ωt))2 + (−aω sin(ωt))2
which reduces to p
O P
v = v2 + 2avω cos(ωt) + a2 ω 2 .
Of course, since the wheel has radius b, and since its center moves with speed v
and it rolls without slipping, we must also have v = bω, and so
O P
p
v = v 2 + 2av(v/b) cos(vt/b) + a2 (v/b)2
or simply p
O P
v =v 1 + 2(a/b) cos(vt/b) + (a/b)2 .
23
By integrating the velocity
O
vP = (v + aω cos(ωt))bi − aω sin(ωt)bj.
0 0 10 20 30 40
0 0 10 20 30 40
24
Note the sharp corners in this graph at the times when point P hits the floor. This
is because the velocity of P at this moment in time is zero and so no well-defined
tangent to the graph can exist here. ¥
Remember that the tangent to a graph of the form x = x(t), y = y(t) is the
unit vector
v x(t)bi + y(t)bj
b=
v =p
|v| x2 (t) + y 2 (t)
which is not defined when the velocity v is zero. For example, the graph below
shows the two-dimensional motion
1 1 1 3
x(t) = t3 − t2 , y(t) = t3 + 2t − t2
3 2 3 2
for 0 ≤ t ≤ 2 seconds and starting at the origin.
0.8
0.6
0.4
0.2
This has ẋ(t) = t2 − t and ẏ(t) = t2 − 3t + 2, which are both zero at t = 1 second
and this occurs at the point (−1/6, 5/6). This point is shown in the graph as a
sharp corner.
25
Example #5 - A Simple Gear Train
If the 3-m radius disk rotates at ω1 = 8 rad/sec, determine the resulting angular
speed of the 2-m radius disk.
To solve this we let G1 , G2 and G3 be the centers of the three disks (from left
to right) and we let A be the point of contact between the first and second disks
and we let B be the point of contact between the second and third disks. The
key to solving these types of problems is to focus on these points of contact and
to compute the velocities of these points using two different approaches and then
setting them equal. Toward this end, we have
G1
vA = ω1 × G1 A
r so that G1 A
v = ω1 G1 rA = ω1 R1 .
Also
G2
vA = ω2 × G2 A
r so that G2 A
v = ω2 G2 rA = ω2 R2 .
26
Since we also have
G1
vA = G1
vG 2 + G2
vA and G1
vG 2 = 0
G1
we see that vA = G2
vA and so G1 A
v = G2 A
v , resulting
ω1 R1 = ω2 R2 .
Also
G3
vB = ω3 × G3 B
r so that G3 B
v = ω3 G3 rB = ω3 R3 .
Since we also have
G2
vA = G2
vG 3 + G3
vA and G2
vG 3 = 0
G2
we see that vB = G3
vB and so G2 B
v = G3 B
v , resulting
ω2 R2 = ω3 R3 .
In total we have
ω1 R1 8(3)
ω1 R1 = ω2 R2 = ω3 R3 so that ω3 = =
R3 2
or ω3 = 12 rad/sec. ¥
27
the figure below.
x0 = R1 + 2R2 , y0 = 0.
If the smaller disk rolls about an axis passing through its center (and perpendicular
to the page) at a constant counterclockwise rate of ω as seen by an observer fixed
to the center of this disk, determine (in terms of ω, R1 and R2 and time t), the x
28
and y coordinate of point P at some later time t, as shown below.
To solve this, we let O be the center of the larger fixed disk and we get G be
the center of the smaller rotating disk and we let P be the point on the outer edge
of the smaller disk that we want to track. Then we have
O
vP = O
vG + G vP = ω × G rP + ω 0 × O rG
where ω0 is the angular velocity of the center (G) of the smaller disk about an axis
passing through O and perpendicular to the page and ω is the angular velocity
of the smaller disk about an axis through G and perpendicular to the page. Note
that in getting O vG , you may think of the figure below which shows an imaginary
rigid rod of length R1 + R2 from O to G and this rod is rotating about an axis
29
passing through point O with angular speed |ω0 |.
In order to relate ω and ω0 , let us focus on the point (Q) that is common to both
disks. Since the smaller disk rolls without slipping around the larger disk, which
is fixed, we must have
O
vQ = O
vG + G vQ = ω × G rQ + ω0 × O rG .
But point Q is in contact with the larger disk, which is not moving (as seen by
O) and so O vQ = 0. Thus we have
ω × G rQ + ω0 × O rG = 0 or ω × G rQ = −ω0 × O rG
Note that the 90◦ is due to the fact that the vectors G rQ and O rG lie in the plane
of the page and ω and ω0 are perpendicular to the plane of the page. Thus we
have
ω = ωkb and ω0 = ω0kb
30
with |G rQ | = R2 and |rG/O | = R1 + R2 , and so
µ ¶
0 0 R2
ωR2 = ω (R1 + R2 ) or ω = ω.
R1 + R2
Going back to
O
vP = O
vG + G vP = ω × G rP + ω 0 × O rG
we now have
O G
r = (R1 + R2 ) cos(ω 0 t)bi + (R1 + R2 ) sin(ω0 t)bj
and
G P
r = R2 cos(ωt)bi + R2 sin(ωt)bj
and so
O
vP = ω G P 0
³ × ´ r ³+ ω × r
O G
´
b b
= ωk × R2 cos(ωt)i + R2 sin(ωt)j b
³ ´ ³ ´
+ ω0 k b × (R1 + R2 ) cos(ω0 t)bi + (R1 + R2 ) sin(ω 0 t)bj
³ ´
= ωR2 cos(ωt)(k b × bi) + sin(ωt)(kb × bj)
³ ´
0 0 b b 0
+ω (R1 + R2 ) cos(ω t)(k × i) + sin(ω t)(k × j) b b
³ ´
= ωR2 cos(ωt)bj + sin(ωt)(−bi)
³ ´
+ω 0 (R1 + R2 ) cos(ω0 t)bj + sin(ω 0 t)(−bi)
or simply
O
vP = −(ωR2 sin(ωt) + ω 0 (R1 + R2 ) sin(ω 0 t))bi
+(ωR2 cos(ωt) + ω 0 (R1 + R2 ) cos(ω 0 t))bj.
or
O
vP = −ωR2 (sin(ωt) + sin(ω0 t))bi + ωR2 (cos(ωt) + cos(ω 0 t))bj.
31
This leads to
O P
ẋ = −ωR2 sin(ωt) − ωR2 sin(ω 0 t)
and
O P
ẏ = ωR2 cos(ωt) + ωR2 cos(ω 0 t)
and integrating these (and requiring that O xP (0) = R1 + 2R2 and O y P (0) = 0)
give
O P ω
x = R2 cos(ωt) + 0 R2 cos(ω0 t)
ω
and
O P ω
y = R2 sin(ωt) + 0 R2 sin(ω0 t).
ω
But µ ¶
0 R2 ω
ω = ω so that R2 = R1 + R2
R1 + R2 ω0
and so µ ¶
O P R2 ωt
x (t) = R2 cos(ωt) + (R1 + R2 ) cos
R1 + R2
and µ ¶
O P R2 ωt
y (t) = R2 sin(ωt) + (R1 + R2 ) sin .
R1 + R2
Suppose we have trace out the motion of P using R1 = R2 = 2 meters and ω = π
rad/sec, then we get
µ ¶
O P πt
x (t) = 2 cos(πt) + 4 cos
2
and µ ¶
O P πt
y (t) = 2 sin(πt) + 4 sin
2
Plots of µ ¶
O P πt
ẋ (t) = −2π sin(πt) − 2π sin
2
and µ ¶
O P πt
ẏ (t) = 2π cos(πt) + 2π cos
2
32
over one period of T = 4 seconds are shown in the figure below.
10
0 0 1 2 3 4
t
-5
A plot of path of P over one period of T = 4 seconds is shown in the figure below.
Note the sharp corner at the time (t = 2 sec) when both ẋP/O and ẏP/O are zero.
You can see this motion very nicely by taking two quarters with one fixed and the
other allowed to roll around this one.
33
Suppose we also trace out the motion of P using R1 = 5 meters and R2 = 2
meters and ω = π rad/sec, then we get
µ ¶
O P 2πt
x (t) = 2 cos(πt) + 7 cos
7
and µ ¶
O P 2πt
y (t) = 2 sin(πt) + 7 sin
7
Plots of µ ¶
O P 2πt
ẋ (t) = −2π sin(πt) − 2π sin
7
and µ ¶
O P 2πt
ẏ (t) = 2π cos(πt) + 2π cos
7
over one period are shown in the figure below.
10
0 0 2 4 6 8 10 12 14
t
-5
34
A plot of path of P over one period of T = 14 seconds is shown in the figure below.
Note the sharp corners at the times when both O ẋP and O ẏ P are zero.
and µ ¶
O P 4πt
y (t) = 4 sin(πt) + 6 sin .
6
Plots of µ ¶
O P 2πt
ẋ (t) = −4π sin(πt) − 4π sin
3
and µ ¶
O P 2πt
ẏ (t) = 4π cos(πt) + 4π cos
3
35
over one period are shown in the figure below.
20
10
0 0 1 2 3 4 5 6
t
0
A plot of path of P over one period of T = 6 seconds is shown in the figure below.
Note the sharp corners at the time (t = 3 sec) when both O ẋP and O ẏ P are zero.
36
seconds of the motion
and this time, the motion is not periodic and the path of P will√eventually fill the
entire region between the two circles with radii 5 m and 5 + 2 2 meters.
Plots of O ẋP and O ẏ P versus time t are shown below and one can see the many
37
times in which both O ẋP and O ẏ P are both zero.
8
6
4
2
-2 0 0 2 4 6 8 10 12 14
t
-4
-6
-8
In general, the motion of P is periodic (resulting in a closed orbit for P) when the
ratio of R1 to R2 is a rational number, and the period is obtain by taking
2π 2πR2 2π 2π(R1 + R2 )
T1 = = and T2 = =
ω ωR2 ωR2 /(R1 + R2 ) ωR2
and setting
2π
T = lcm(R2 , R1 + R2 )
ωR2
where lcm(a, b) equals the smallest integer that both a and b divide into, known
as the least common multiple of a and b. For ω = π rad/sec, R1 = 5 m and R2 = 2
m, we get
2π
T = lcm(2, 7) = lcm(2, 7) = 14 seconds.
2π
√
If the ratio of R1 to R2 is not a rational number, such as 2, then the orbit of P
is not closed and it will eventually fill the entire space between the circles r = R1
and r = R1 + 2R2 . ¥
Let us now consider two rigid bodies A and B, with standard reference triads
(SRTs)
a1 , b
{b a2 , b
a3 } and {bb1, b
b2, b
b3 },
38
attached to them, respectively. Suppose that B is rotating with some angular
velocity as seen by observers fixed in A. We represent this angular velocity by
A B
ω and say that it is the angular velocity of B, as measured by A. Of course, if
A sees B rotating with angular velocity A ωB , it stands to reason that B sees A
rotating with angular velocity B ω A , and that
B
ωA = − A ωB . (15)
A
Assuming That ω B is Constant and P is Perpendicular A
ωB
Suppose for now that there exist a point O that is fixed in both A and B. This
can be used to represent an origin that is common to both A and B. Next, let P
be any vector that is fixed in B so that
B
dP
= 0.
dt
Once again, we stress that we write this as
B
dP dP
and not simply
dt dt
since we must make it clear that it is B who is measuring the change in P with
respect to time t. Since B does not see P change in time, we must have
B
dP
=0
dt
however, A does see P change with time (since A sees B rotating and P is fixed
in B and so A must see P rotating with B) and our goal is to compute
A
dP
dt
which is the rate in which P changes in time, as measured by A.
39
that is fixed in B, by writing the expansion
b1 + p2 b
P = p1 b b 2 + p3 b
b3 with bk
pk = P · b
a1 , b
{b a2 , b
a3 },
P = q1 b
a1 + q2 b
a2 + q3 b
a3 with qk = P · b
ak
{b b2 , b
b1, b b3 },
for k = 1, 2, 3, then A and B must find that rk equals pk . In addition, since time
is absolute in both A and B, the rate of change in rk (or pk ) as measured by A
must be the same as the rate of change in pk (or rk ) as measured by B. We may
thus represent this rate of change by
dpk
= ṗk
dt
without having to write it as
A B A B
dpk dpk dpk dpk dpk
or since = = = ṗk .
dt dt dt dt dt
40
which is the rate in which P changes in time, as measured by A, let us first
consider the simple case where: (i) A ω B is constant, (ii) the tail of P is located at
point O, and (iii) P is perpendicular to A ω B , as indicated in the following figure.
It is clear then that the tail of P remains fixed on point O but tip of P moves in a
circle of radius |P| about an axis containing A ω B and passing through the center
of this circle, and if θ is the angle of rotation (as measured by A), then the vector
to emphasize the fact that it is observers in A who are taking the limit. The
length of
∆P = P(θ + ∆θ) − P(θ)
41
is simply the length of the cord subtended by the angle ∆θ, as shown in the
following figure.
This is approximately the same as the arc length along the circle subtended by
the angle ∆θ, (for small ∆θ) and this is just the radius of the circle transverse by
the tip of P (i.e., |P|) times ∆θ, so that in the limit as ∆θ goes to zero, observers
is A find that ¯ ¯ ¯ ¯
¯ A P(θ + ∆θ) − P(θ) ¯ ¯ A dP ¯
¯ lim ¯=¯ ¯
¯∆θ→0 ∆θ ¯ ¯ dθ ¯ = |P|.
Then, by the chain rule, we have
A
dP A dP A dθ A
dP A B
= = | ω |
dt dθ dt dθ
which then is a vector of magnitude
¯A ¯ ¯A ¯
¯ dP ¯ ¯ dP ¯ A B
¯ ¯ ¯ ¯ A B
¯ dt ¯ = ¯ dθ ¯ | ω | = |P|| ω |
and direction tangent to the circle transverse by the tip of P. That makes this
vector perpendicular to both P and A ωB , and in fact this vector must be given by
A
ω B × P,
A
Assuming That ω B is Constant and P is NOT Perpendicular to A ω B
Suppose next that: (i) A ωB is still constant and (ii) the tail of P is still fixed
at O, but (iii) P is not necessarily perpendicular to A ω B . Once again, the tip of
P moves in a circle, but this time the radius of this circle is
|P| sin ϕ
A
where ϕ is the angle between P and ω B . This in indicated in the following
diagram.
P = OC + CP
CP = OC − P
43
and from Equation (16a), we may write
A
d(CP) A
= ω B × (CP)
dt
or
A
d(OC − P) A
= ω B × (OC − P)
dt
or
A
d(OC) A dP
− = A ωB × OC − A ω B × P.
dt dt
But OC is along the vector A ω B so that
A
ω B × OC = 0
and since OC points along the vector A ω B , it is not changed by the rotation and
so we also have
A
d(OC)
= 0.
dt
Thus we see that
A
dP
0− = 0 − A ωB × P
dt
or simply
A
dP
= A ωB × P (16b)
dt
which is the same as Equation (16a) and so now Equation (16b) is true for any
vector whose tail is located at point O and A ωB is constant.
A
The General Case for dP/dt When A ω B Is a Constant
Finally, now assume that: (i) A ωB is constant, but (ii) P is now any vector
fixed in body B, with its tail not necessarily located at point O. We may draw
a vector from point O to the tail of P, which we call Ptail , and we may draw a
vector from point O to the tip of P, which we call Ptip as shown in the following
44
figure.
Then
A
dP A d(Ptip − Ptail ) A
dPtip A dPtail
= = − .
dt dt dt dt
But each of Ptip and Ptail begin at point O and so from Equation (16b), we have
A A
dPtip A dPtail
= ωB × Ptip and = A
ωB × Ptail
dt dt
and so
A
dP A
= ωB × Ptip − A ωB × Ptail = A
ωB × (Ptip − Ptail ) = A
ω B × P.
dt
Thus we see that
A
dP
= A ωB × P (16c)
dt
is now true for any vector that is fixed in the rigid body B and any constant
rotation A ω B .
45
A
Assuming That ω B is Constant: Kane’s Formula for A
ωB
Since Equation (16c) is true for any vector P that is fixed in B, we may say
that
A b1
db A b2
db A b3
db
= A b1
ωB × b , = A b2
ωB × b , = A b3
ωB × b
dt dt dt
since the standard reference triad
B = {b b2, b
b1, b b3}
b b2 × b
b1 = b b3 , b2 = b
b b3 × b
b1 , b3 = b
b b1 × b
b2
and so
A b
A b 1 = (A ωB ) · (b
ωB · b b2 × b
b 3 ) = (A ωB × b b 3 = db2 · b
b2) · b b3
dt
and
A b
A b 2 = (A ωB ) · (b
ωB · b b3 × b
b 1 ) = (A ωB × b b 1 = db3 · b
b3) · b b1
dt
and
A b1
db
A b 3 = (A ωB ) · (b
ωB · b b1 × b
b 2 ) = (A ωB × b
b1) · b
b2 = b2
·b
dt
which means we have
à ! à ! à !
A b A b A b
A B d b2 b3 bb1 + db 3 b1 bb2 + db 1 b2 bb3
ω = ·b ·b ·b (17)
dt dt dt
which is known as Kane’s formula for A ωB . Although this derivation assumes that
A B
ω is constant, we now show that it is, in fact, valid in general.
46
The General Case When A ωB is Not Constant: Kane’s Formula for A
ωB
Using the product rule for the derivative of a dot product, we note that for
any vector V,
V · V = |V|2
and so for any reference frame we have
A
Back to Kane’s Formula for ωB
Since each vector in the SRT for B has constant magnitude of 1, even when
measured by A, we may say that
A b1
db A b2
db A b3
db
b1
⊥b , b2
⊥b , b3
⊥b
dt dt dt
and so we must have
A b1
db b1 = 0
·b
dt
resulting in
b 1 + a12 b
(a11 b b2 + a13 b b 1 = a11 (b
b3) · b b1 · b b2 · b
e1 ) + a12 (b b 1 ) + a13 (b
b3 · b
b1 )
= a11 (1) + a12 (0) + a13 (0) = a11 = 0
48
and so, for example,
A b3
db A b b2 )
d(b1 × b A b
db1 b A b
= = b 1 × db2
× b2 + b
dt dt dt dt
or
A b3
db b 2 + a13 b
= (a12 b b3 ) × b b2 + b
b 1 × (a21 bb 1 + a23 b
b3 )
dt
b2 × b
= a12 (b b 2 ) + a13 (b b3 × bb 2 ) + a21 (b
b1 × b
b1 ) + a23 (b
b1 × b
b3)
= a12 (0) + a13 (−b b 1 ) + a21 (0) + a23 (−b b2)
since
b b 2 = −b
b3 × b b1 and b1 × b
b b3 = −b
b2
and of course
b1 × b
b b1 = b
b2 × b
b 2 = 0.
Thus we have
b3
db A
b 1 − a23 b
= −a13 b b2
dt
and setting this equal to the result in Equation (20c),
A b3
db b1 + a32 b
= a31 b b2,
dt
then leads to −a13 = a31 and −a23 = a32 . Similarly we find that −a21 = a12 . Thus
we see that
A b
db1 b2 + a13 b
b3
= a12 b (21a)
dt
and
A b
db2 b 1 + a23 b
b3
= −a12 b (21b)
dt
and
A b
db3 b 1 − a23 b
b2.
= −a13 b (21c)
dt
If we define a vector
A B
ω = ω1 b b1 + ω2 bb2 + ω3 bb3 (22)
49
having components
Using the cross product, it is easy to see that this set of equations is the same as
A b1 ³
db ´
b 1 + ω2 b
= ω1 b b 2 + ω3 b
b3 × b
b1
dt
and
A b2 ³
db ´
b 1 + ω2 b
= ω1 b b 2 + ω3 b
b3 × b
b2
dt
and
A b3 ³
db ´
b 1 + ω2 b
= ω1 b b 2 + ω3 b b3
b3 × b
dt
which says that
A bk
db
= A bk
ωB × b for k = 1, 2, 3. (24)
dt
Note that from Equations (23a,b,c), we have
A b2
db A b
ω1 = b 3 = − db3 · b
·b b2
dt dt
50
and
A b3
db A b
ω2 = b 1 = − db1 · b
·b b3
dt dt
and
A b1
db A b
ω3 = b 2 = − db2 · b
·b b1
dt dt
and so Equation (22) reads
à ! à ! à !
A b A b A b
A B db 2 b b1 + db 3 b b2 + db 1 b b3,
ω = · b3 b · b1 b · b2 b (25a)
dt dt dt
which is the same as Equation (17) and the derivation of Kane’s equation for A ω B
is complete.
A
The General Case for dP/dt When A ω B Is Not a Constant
If P now is any vector fixed in B, then we may expand P in terms of the SRT
and write
b1 + p2 b
P = p1 b b 2 + p3 b
b3
and hence
à ! à ! à !
A A A b A A b A A b
dP dp1 b db1 dp2 b db2 dp3 b db3
= b1 + p1 + b2 + p2 + b3 + p3
dt dt dt dt dt dt dt
µA A A
¶
dp1 b dp2 b dp3 b
= b1 + b2 + b3
dt dt dt
b1 ) + p2 (A ω B × b
+p1 (A ω B × b b 2 ) + p3 (A ωB × b
b3 )
µA A A
¶
dp1 b dp2 b dp3 b b1 + p2 b
b 2 + p3 b
b3 )
= b1 + b2 + b3 + A ωB × (p1 b
dt dt dt
or simply µA ¶
A A A
dP dp1 b dp2 b dp3 b
= b1 + b2 + b3 + A ω B × P.
dt dt dt dt
Since P is fixed in B, it must be true that p1 , p2 and p3 , which are the components
of P with respect to the SRT B (that is fixed in B), must be constants. Since A
must measure the same components, we must have
A B A B A B
dp1 dp1 dp2 dp2 dp3 dp3
= =0 , = =0 , = = 0.
dt dt dt dt dt dt
51
In other words, the only reason why P is changing, as far as A is concerned, is
because A measures the SRT
n o
B= b b2, b
b1, b b3
a1 , b
A = {b a2 , b
a3 }
P = q1 b
a1 + q2 b
a2 + q3 b
a3 ,
and then A sees P changing with time not because the b ak ’s are changing with
time, but because the qk ’s are changing with time, and in fact A would write in
this case that
A
dP A dq1 A
dq2 A
dq3
= b
a1 + b
a2 + b
a3 . (26a)
dt dt dt dt
However, if A chooses to represent P in terms of the SRT
n o
B= b b2, b
b1, b b3
(just like B would find) and now A measures P as changing with time not because
A measures the pk ’s as changing with time, but because A measures the b bk ’s as
changing with time, and in fact A would write in this case that
A A b A b A b
dP db1 db2 db3
= p1 + p2 + p3 . (26b)
dt dt dt dt
Since Equations (26a,b) give the same left-hand-sides, their right-hand-sides must
be the same as well and so
A A A A b A b A b
dq1 dq2 dq3 db1 db2 db3
b
a1 + b2 +
a b
a3 = p1 + p2 + p3 .
dt dt dt dt dt dt
52
The representation in terms of the SRT
n o
B= b b2, b
b1, b b3
turns out to be more convenient in the derivation of the main result that
A
dP A
= ω B × P.
dt
Anyway you slice it, we still come up with the conclusion that for any vector
P that is fixed in B, we must have
A
dP A
= ωB × P (27)
dt
where A ωB is defined as in Equation (25a). What is nice about Equation (27) is
that it makes no mention of either SRT
n o
A = {b a1 , b
a2 , b
a3 } or b b b
B = b1 , b2 , b3
and so it is true as a vector equation for any SRTs that A and B choose to use.
The only requirement is that P is fixed in B.
A
Kane’s Other Formula for ωB
We end this section by deriving what we shall refer to as Kane’s other Formula
for A ω B . Using the equation
1n b b 1 + (b
b2 × X) × b
b 2 + (b
b3 × X) × bb3
o
X= (b1 × X) × b
2
A
with X = ωB , we have
A B 1n b A B b b A B b b A B b
o
ω = (b1 × ω ) × b1 + (b2 × ω ) × b2 + (b3 × ω ) × b3 .
2
But
A b
b1 = − db1
b1 × A ωB = − A ωB × b
b
dt
53
b2 and b
with similar results for b b 3 . Thus we have
( )
A b A b A b
A B 1 db 1 b1 − db 2 b2 − db 3 b3 .
ω = − ×b ×b ×b
2 dt dt dt
or simply ( )
1 b A A b A b
A B
ω = b 1 × db1 + b
b b2 × db2 + b
b3 × db3 (25b)
2 dt dt dt
which we shall refer to as Kane’s other formula for A ω B .
We have seen that for a rigid body such as A, we may choose to represent
vectors in terms of a SRT that is fixed in A, or in terms of a SRT that is fixed in
any other rigid body B, and how these representations are related was the subject
of the geometric rotations given in an earlier chapter.
This section can be skipped without any lost in continuity. Suppose that A
and B are two SRTs fixed in the rigid bodies A and B, respectively, and suppose
that P is a fixed vector in the rigid body B. Then the matrix B {P} contains the
components of P with respect to the SRT B and these are constant (as measured
in both A and B). Using the rotation matrix from B to A, we may write that
A
{P} = [A RB ]B {P}
and differentiating (as measured in A) this equation with respect to time t, we
have
¡ ¢
A A A ¡ ¢ A d[A RB ] B A
d B {P}
d {P} d A B A
= [ RB ] {P} = {P} + [ RB ]
dt dt dt dt
or ¡ ¢
A
A A
d {P} A A
d[ RB ] B d B {P}
= {P} since =0
dt dt dt
But
B
{P} = [B RA ]A {P} = [A RB ]T A {P}
and so we have
A
µA ¶
d A {P} d[A RB ] A
= [ RB ]T A
{P} = [A SB ]A {P} (28a)
dt dt
54
with
A
d[A RB ] A
[A SB ] ≡ [ RB ]T . (28b)
dt
Since we know that
[A RB ][A RB ]T = [I]
we must have
A
d¡A ¢ A d[I]
[ RB ][A RB ]T = = [0]
dt dt
and hence
A
d[A RB ] A A A
d[ RB ]T
[ RB ]T + [A RB ] = [0]
dt dt
or
[A SB ] + [A SB ]T = [0]
which means that the matrix [A SB ] defined in Equation (28b) is skew symmetric.
From this we see that if Sij are the components of the [A SB ], then Sij = −Sji
resulting in Sii = 0 and the form of [A SB ] must be
⎡ ⎤ ⎡ ⎤
0 −S21 S13 0 −s3 s2
[A SB ] = ⎣ S21 0 −S32 ⎦ or simply [A SB ] = ⎣ s3 0 −s1 ⎦ .
−S13 S32 0 −s2 s1 0
X3 A
d[A RB ]ik A X
3 A bk )
ai · b
d(b
= [ RB ]jk = (b bk )
aj · b
k=1
dt k=1
dt
(A × B) · (C × D) = (A · C)(B · D) − (A · D)(B · C)
55
we have
à ! à ! à !
A bk
db bk
db A ³ ´ A b
dbk
ai × b
(b aj ) · bk
×b = b
ai · bk ) − b
aj · b
(b bk
ai · b b
aj ·
dt dt dt
or à !
X
3 A bk
db
ai × b
Sij − Sji = (b aj ) · bk
×b .
k=1
dt
A
But since [ SB ] is skew symmetric we have Sij = −Sji and so
à !
X3 A b
dbk b
ai × b
Sij + Sij = (b aj ) · × bk
k=1
dt
or à !
1 X3 A bk
db
ai × b
Sij = (b aj ) · bk
×b . (29b)
2 k=1
dt
Using Kane’s other equation for A ωB in Equation (25b), namely
( )
A b A b A b
A B 1 b db1 b db2 b db3
ω = b1 × + b2 × + b2 ×
2 dt dt dt
aj ) · (A ωB )
ai × b
Sij = −(b
or simply
Sij = (A ωB ) · (b
aj × b
ai ). (30)
Writing these out in detail we have
⎡ A B ⎤
( ω ) · (b a1 ) (A ωB ) · (b
a1 × b a1 ) (A ω B ) · (b
a2 × b a3 × b
a1 )
[A SB ] = ⎣ (A ω B ) · (b a2 ) (A ωB ) · (b
a1 × b a2 ) (A ω B ) · (b
a2 × b a2 ) ⎦
a3 × b
A B A B A B
( ω ) · (b a1 × b
a3 ) ( ω ) · (b a2 × b
a3 ) ( ω ) · (b a3 × b
a3 )
56
or ⎡ ⎤
0 (A ωB ) · (−ba3 ) (A ω B ) · (ba2 )
[A SB ] = ⎣ (A ω B ) · (b
a3 ) 0 (A ω B ) · (−ba1 ) ⎦
A B A B
( ω ) · (−b a2 ) ( ω ) · (b a1 ) 0
or simply
⎡ ⎤
0 −(A ω B ) · b
a3 (A ωB ) · b
a2
[A SB ] = ⎣ (A ω B ) · b
a3 0 a1 ⎦ .
−(A ω B ) · b
A B A B
−( ω ) · b a2 ( ω ) · b a1 0
By writing
A
ω B = (A ω B )1 b
a1 + (A ωB )2 b
a2 + (A ω B )3 b
a3
we finally have
⎡ ⎤
0 −(A ω B )3 (A ω B )2
[A SB ] = ⎣ (A ωB )3 0 −(A ω B )1 ⎦ . (31)
A B A B
−( ω )2 ( ω )1 0
P = (P · b a1 + (P · b
a1 )b a2 + (P · b
a2 )b a3 )b
a3
we have
A
µA ¶ µA ¶ µA ¶
dP dP dP dP
= ·b
a1 ba1 + ·b
a2 ba2 + ·b
a3 ba3
dt dt dt dt
so that ½A ¾
A A
d {P} A dP
=
dt dt
57
and putting this into Equation (32) finally yields
½A ¾
A dP
= A {A ωB × P}
dt
or
A
dP
= A ωB × P
dt
for any vector P that is fixed in B, just like we showed earlier in this chapter.
We have shown so far that for two rigid bodies A and B, the equation
A
dP A
= ωB × P
dt
is true provided that the vector P is fixed in B. Suppose we now allow for the
possibility that the vector P is moving in B. By writing P as
b1 + p2 b
P = p1 b b 2 + p3 b
b3
A b1
db A b2
db A b3
db
= A b1
ωB × b , = A b2
ωB × b , = A b3 ,
ωB × b
dt dt dt
58
and putting these into the above expression for A dP/dt, we get
A B
dP dP b 1 + p2 (A ωB ) × b
b 2 + p3 (A ω B ) × b
b3
= + p1 (A ω B ) × b
dt dt
B
dP b 1 + p2 b
b2 + p3 b
b3)
= + (A ω B ) × (p1 b
dt
or simply
A
dP B dP A B
= + ω × P. (33a)
dt dt
where A ωB is given by either Equation (25a), which we repeat here
à ! à ! à !
A b A b A b
A B db 2 b b1 + d b 3 b b2 + db 1 b b3 ,
ω = · b3 b · b1 b · b2 b (33b)
dt dt dt
This is a key equation involving the kinematics of rigid-body motion and may
other equations will follow from it. For this reason, it is very important that the
student understands all that this equations teaches.
Suppose that
a1 , b
A = {b a2 , b ex , b
a3 } = {b ey , b
ez }
are the unit vectors in a rectangular coordinate frame that is fixed in a rigid body
A and suppose that the SRT
B = {b b2, b
b1, b b3}
59
and r r
b2 = 1−t 1+t
b b a2 −
a1 + 0b b
a3
2 2
and r r r
b3 = − 1 + t 4 1−t
b b
a1 + b
a2 − b
a3
6 6 6
where the variable t takes on values in the interval −1 < t < +1. Then we have
A b1
db 1 1
= p b a2 − p
a1 + 0b b
a3
dt 2 3(1 + t) 2 3(1 − t)
and
A b2
db 1 1
=− p b a2 − p
a1 + 0b b
a3
dt 2 2(1 − t) 2 2(1 + t)
and
A b3
db 1 1
=− p b a2 + p
a1 + 0b b
a3 ,
dt 2 6(1 + t) 2 6(1 − t)
and using Kane’s equation for A ωB ,
à ! à ! à !
A b A b A b
A B db 2 b3 b b1 + db 3 b1 bb2 + db 1 b2 bb3,
ω = ·b ·b ·b
dt dt dt
we find that
A b2
db A b3
b3 = p 1
·b ,
db b1 = 0
·b
dt 12(1 − t2 ) dt
and
A b1
db b2 = p 1
·b
dt 6(1 − t2 )
and so à ! à !
A 1 b1 + 0b
b2 + 1 b3.
ωB = p b p b
12(1 − t2 ) 6(1 − t2 )
or just √
b1 + 2b
b b3
A B
ω = p .
2 3(1 − t2 )
60
Now if P is a vector, such as
P = tbb3
then, putting in the above expression for bb 3 in terms of ba1 , b
a2 , and b
a3 , we may
write that r r r
1+t 4 1−t
P = −t b
a1 + t b
a2 − t b
a3
6 6 6
and so computing A dP/dt directly, we have
Ãr ! Ãr ! Ãr !
A
dP d 1+t d 4 d 1−t
=− t b
a1 + t b
a2 − t b
a3
dt dt 6 dt 6 dt 6
which becomes
à ! r à !
A
dP 2 + 3t 4 −2 + 3t
=− p b
a1 + b
a2 + p b
a3 .
dt 2 6(1 + t) 6 2 6(1 − t)
Computing A dP/dt using Equation (33a), we have
⎤ ⎡
A B b b2 b
b1 b b3
dP dP A B b3 + p 1 √
= + ω ×P=b det ⎣ 1 0 2 ⎦
dt dt 2 3(1 − t2 ) 0 0 t
or
A
dP b b2
tb
= b3 − p .
dt 2 3(1 − t2 )
b 2 and b
To see that these are equal we replace b b3 in terms of b
a1 , b
a2 , and b
a3 , and
get
à r r r !
A
dP 1+t 4 1−t
= − b
a1 + b
a2 − b
a3
dt 6 6 6
Ãr r !
t 1−t 1+t
− p b
a1 + 0ba2 − b
a3
2 3(1 − t2 ) 2 2
which reduces to
à ! r à !
A
dP 2 + 3t 4 −2 + 3t
=− p b
a1 + b
a2 + p b
a3 .
dt 2 6(1 + t) 6 2 6(1 − t)
and agrees with the results from the direct calculation. ¥
61
7. Simple Angular Velocities
A rigid body B has a simple angular velocity as measured in A when there exist
b whose orientation as seen in both A and B is constant (independent
a unit vector n
of time). In such a case, the vector nb can be easily identified by inspection, and
by Euler’s theorem from previous chapters, this vector must be along the axis of
rotation. Thus the angular velocity of B as measured by A must be of the form
A
A dθ
ωB = A
ωB n
b where A
ωB = (34)
dt
is the time derivative of a scalar (angular) quantity as measured by A. Note that
B
ωA = − A ωB
we see that body A is pinned to body B via a revolute joint and therefore B ωA is
clearly a simple angular velocity
A dθ1 b
ωB = d3
dt
62
where db 3 is a unit vector pointing out of the page and θ1 is the angle indicated in
the figure. Thus we see that if two bodies are pinned via a revolute joint, there is
a simple angular velocity relating their relative rotational motion. However, we
must point out that the two bodies in question need not be pinned together for
the angular velocity between them to be simple. For example the following figure
shows that A and C are not pinned together, yet the motion of C as seen in
D is a combination of rotation and translation. However, the angular motion is
described by the variable θ2 and
A dθ2 b
ωC = d3
dt
where db 3 is a unit vector pointing out of the page and θ2 is the angle indicated in
the figure, thereby making this a simple angular velocity.
In general, the angular velocity vector between two rigid bodies cannot be
written as the derivative of a scalar quantity times a vector, i.e., it’s not simple.
63
For example, the following figure
shows a rolling wheel. Rigid body A is the ground, rigid body B is the axial of
the wheel and rigid body C is the wheel. It should be clear that
A
ωB and B
ωC
where nb1 points upward (perpendicular to the ground) and where n b 2 points along
the axis of the wheel (i.e., perpendicular to the plane of the wheel), but A ωC is
not a simple angular velocity since the motion of C as seen by A is such that there
is no vector fixed in C that also remains fixed in A. In other words, the axis of
rotation (guaranteed by Euler’s Theorem) is constantly changing with time.
Suppose once again that A and B are rigid bodies with A fixed and B exhibiting
planar-like motion, as seen by A. A rigid body B exhibits planar-like motion as
seen by another rigid body A if there exist a SRT fixed in B
b1, b
B = {b b2, b
b3}
which has one of its vectors constant in time as seen by A. Without any lost in
generality, let’s suppose that it’s the third vector in B that is constant in time as
64
seen by A. Using Kane’s formula for A ω B ,
à ! à ! à !
A b A b A b
A B db 2 b b1 + d b 3 b b2 + db 1 b b3
ω = · b3 b · b1 b · b2 b
dt dt dt
we find that
à ! à !
A b2
db ³ ´ A b
db
A
ωB = b3 b
·b b1 + 0 · b
b1 bb2 + 1 b2 b
·b b3
dt dt
or à ! à !
A b2
db A b
db
A
ωB = b3 b
·b b1 + 1 b
· b2 bb3. (35)
dt dt
b2 · b
But b b 3 = 0 so that
A b2 · b
d(b b3) A b
db2 b A b
db3 b
= · b3 + · b2 = 0
dt dt dt
yielding
b2
db A A b
·bb 3 = − db3 · bb 2 = −0 · b
b 2 = 0.
dt dt
Putting this into Equation (35) finally leads to
A b1
db
A b3
ωB = ωb with ω≡ b2 ,
·b (36)
dt
which shows that A ω B has a constant direction and hence it is a simple angular
velocity.
65
Example #8: Planar-Like Motion of a Rigid Body
The figure below shows an object (disk, ring or sphere) rolling down a fixed
incline.
Let A, B and C be three rigid bodies. The addition theorem for angular
velocities states that
A C
ω = A ωB + B ωC . (37)
66
To prove this we consider any vector P that is fixed in C. Then using Equation
(33a) with bodies B and C, we have
B C
dP dP B C
= + ω × P = B ωC × P
dt dt
and using Equation (33a) with bodies A and B, we have
A
dP B dP A B
= + ω × P = B ωC × P + A ωB × P
dt dt
or
A
dP
= (A ω B + B ωC ) × P.
dt
But using Equation (33a) with bodies A and C, we also have
A C
dP dP A C A
= + ω ×P= ωC × P
dt dt
and so comparing these two equations for A dP/dt, we have
A
ω C × P = (A ωB + B ω C ) × P
for all vectors fixed in C, and hence for any constant vector in a SRT fixed in C.
This can only be true if
A C
ω = A ωB + B ωC .
Note that by setting A = C, this also implies that
A
ωA = A
ωB + B ωA = 0
so that
B
ωA = − A ωB
as expected. Of course, Equation (37) extends to many rigid bodies A, B, C, . . .,
Y, and Z, by writing
A
ωZ = A
ωB + B ωC + C ωD + D ωE + · · · + Y ωZ . (38)
67
us consider the angular acceleration between two rigid bodies and the addition
theorem for angular accelerations.
68
does not (in general) apply to angular accelerations since we need not have
A
ω B × B ωC = 0.
Equation (41) is the correct result. Let’s now consider an example using the
rolling wheel.
The figure below shows the side view of a rolling wheel which rolls without slip-
ping along a circle in the xy plane by use of an axial that rotates counterclockwise
about the positive z axis.
ex , b
A = {b ey , b
ez }
in the ground frame (A). The unit vector b ex is pointing out of the page in this
figure, b
ey is pointing to the right in this figure and b
ez is pointing up in this figure,
so that these are the usual units vectors in rectangular coordinates.
Let’s call the axial, body B and a good choice of unit vectors that are fixed in
B are the usual cylindrical unit vectors
er , b
B = {b eθ , b
ez }
69
where A and B are related via the following table,
b
ex b
ey b
ez b
ex b
ey b
ez
b
er b
er · b
ex b
er · b
ey b
er · b
ez b
e cos(θ) sin(θ) 0
= r
b
eθ b b
eθ · ex b b
eθ · ey b b
eθ · ez b
eθ − sin(θ) cos(θ) 0
b
ez b
ez · b
ex b
ez · b
ey b
ez · b
ez b
ez 0 0 1
where θ is the counterclockwise angle the arm makes with the unit vector b
ex . This
says, for example, that
b
er = cos(θ)b
ex + sin(θ)b
ey + 0b
ez = cos(θ)b
ex + sin(θ)b
ey
b er − sin(θ)b
ex = cos(θ)b eθ + 0b er − sin(θ)b
ez = cos(θ)b eθ
reading across the first column, and so on. Because of the axial’s motion. The
angular velocity of the wheel (let’s call this body C) as seen by an observer moving
with the axial (let’s call this body B) is
B
ω C = −ϕ̇b
er = −ϕ̇ cos(θ)b
ex − ϕ̇ sin(θ)b
ey
where ϕ is the angle covered by the wheel, as indicated in the following figure
showing the view looking perpendicular to the plane of the wheel. In this view,
the wheel is rolling to the right and the angle ϕ is indicated. In addition, the
axial of the wheel is behind the wheel.
70
In this figure b
er is a unit vector pointing out of the page, beθ is a unit vector
pointing counterclockwise tangent to the wheel and b ez is a unit vector pointing
straight up. Now the angular velocity of the arm (body B) as seen by an observer
fixed at the origin O (let’s call this body A) is
A
ω B = θ̇b
ez .
Then the angular velocity of the wheel (body C) as measured by A (where point
O is fixed) is
A C
ω = A ωB + B ω A = (θ̇bez ) + (−ϕ̇b
er )
or
A
ω C = −ϕ̇ cos(θ)b
ex − ϕ̇ sin(θ)b
ey + θ̇b
ez ,
which is not a simple angular velocity, but yet it was expressible as a sum of two
simple angular velocities. The angular acceleration of C as measured by A (where
point O is fixed) is then
A A C
A C d( ω )
α =
dt
which leads to
A
αC = (−ϕ̈ cos(θ) + ϕ̇θ̇ sin(θ))b
ex − (ϕ̈ sin(θ) + ϕ̇θ̇ cos(θ))b
ey + θ̈b
ez
As a check to this, we note that the angular acceleration of the wheel as seen by
an observer moving with the arm is
B
αC = −ϕ̈b
er
while the angular velocity of the arm as seen by an observer fixed at the origin O
is
A B
α = θ̈bez .
Then using Equation (41) we have
A
αC = A αB + B αC + A ω B × B ωC
= θ̈b
ez + (−ϕ̈b ez ) × (−ϕ̇b
er ) + (θ̇b er )
ez − ϕ̈b
= θ̈b ez × b
er − ϕ̇θ̇(b er )
ez − ϕ̈b
= θ̈b er − ϕ̇θ̇b
eθ
or
A
αC = −ϕ̈b
er − ϕ̇θ̇b
eθ + θ̈b
ez .
71
But
b
er = cos(θ)b
ex + sin(θ)b
ey and b
eθ = − sin(θ)b
ex + cos(θ)b
ey
and so
A
αC = −ϕ̈(cos(θ)bex + sin(θ)b
ey )
−ϕ̇θ̇(− sin(θ)b
ex + cos(θ)b ey ) + θ̈b
ez
or
A
αC = (−ϕ̈ cos(θ) + ϕ̇θ̇ sin(θ))b
ex − (ϕ̈ sin(θ) + ϕ̇θ̇ cos(θ))b
ey + θ̈b
ez
and this agrees with the result earlier. You see how
A
αC = A
αB + B αC
would not give the correct answer since it would not give the ϕ̇θ̇ terms, so be
careful! ¥
72
Now using the fact that
bk
db A
= A ωB × b bk
dt
for k = 1, 2, 3, the last three terms in the above expression become
à ! à ! à !
A b b1 A b b2 A b b3
db2 b A db db3 b A db db1 b A db
· b3 + · b1 + · b2
dt dt dt dt dt dt
à ! à !
db2 b ³ A B b ´
A b
db3 b ³ A B b ´
A b
= · b3 ω × b1 + · b1 ω × b2
dt dt
à !
db1 b ³ A B b ´
A b
+ · b2 ω × b3
dt
(Ã ! Ã ! Ã ! )
A b A b A b
d b2 b3 bb1 + db 3 b1 b b2 + db 1 b2 bb3
= A ωB × ·b ·b ·b
dt dt dt
A
= ωB × A ωB = 0
then
A b 1 + (A αB )2 b
αB = (A αB )1 b b 2 + (A αB )3 b
b3 , (43b)
with
A
d(A ωB )k
(A αB )k = (43c)
dt
for k = 1, 2, 3. In fact when A ω B is represented as
A
ωB = (A γ B )1 b
a1 + (A γ B )2 b
a2 + (A γ B )3 b
a3 , (43d)
73
then we also have
A
αB = (A µB )1 b
a1 + (A µB )2 b
a2 + (A µB )3 b
a3 , (43e)
with
A
d(A γ B )k
(A µB )k = (43f)
dt
for k = 1, 2, 3.
a1 , b
A = {b a2 , b ex , b
a3 } = {b ey , b
ez }
which are fixed unit vectors pointing in the positive x, y and z directions, respec-
tively, and consider a turntable rotating with angular speed ω about the z axis as
body B so that
B = {b b2 , b
b1, b b3 } = {b
er , b
eθ , b
ez }.
Then, since b b2 involve counterclockwise rotations about b
b 1 and b b3 = b
a3 , we must
have
b 1 = cos(θ)b
b a1 + sin(θ)b
a2 and b2 = − sin(θ)b
b a1 + cos(θ)b
a2 .
and the relationships between A and B are summarized in the following table.
b
a1 b
a2 b
a3 b
a1 b
a2 b
a3
b1
b b1 · b
b a1 b1 · b
b a2 b1 · b
b a3 b1
b cos(θ) sin(θ) 0
b2 b b b2 · b = b2
b b2 · b
a1 b2 · b
a2 b a3 b − sin(θ) cos(θ) 0
b3
b b
b3 · b
a1 b
b3 · b
a2 b3 · b
b a3 b3
b 0 0 1
Now suppose that P is a point that is moving outward in the radial direction as
74
seen by B, as shown in the following figure.
The position vector of P as represented by the standard triad in B (on that rotates
with the turntable) is given by
b1 .
P = sb
Then A ωB = θ̇b
a3 = ωba3 = ω bb 3 , and
A
dP B dP A B b 1 + (θ̇b
b3 ) × (sb
b 1 ) = ṡb
b 1 + sθ̇(b
b3 × b
b1 )
= + ω × P = ṡb
dt dt
so that
A
dP b 1 + sθ̇b
b2.
= ṡb (44)
dt
Using the above table and putting in the expressions
b 1 = cos(θ)b
b a1 + sin(θ)b
a2 and b2 = − sin(θ)b
b a1 + cos(θ)b
a2 ,
this leads to
A
dP
a1 + sin(θ)b
= ṡ (cos(θ)b a2 ) + sθ̇ (− sin(θ)b
a1 + cos(θ)b
a2 )
dt
so that
A
dP
= (ṡ cos(θ) − sθ̇ sin(θ))b
a1 + (ṡ sin(θ) + sθ̇ cos(θ))b
a2 . (45)
dt
We may get Equation (45) directly by noting that
b 1 = s cos(θ)b
P = sb a1 + s sin(θ)b
a2
75
and so
A
dP
= (ṡ cos(θ) − sθ̇ sin(θ))b a1 + (ṡ sin(θ) + sθ̇ cos(θ))b a2 .
dt
Let us also compute
A 2 A
µ ¶ B µA ¶
dP d A dP d dP A B
A
dP
2
= = + ω ×
dt dt dt dt dt dt
B ³ ´
d b b2 + (θ̇b b3 ) × (ṡb
b 1 + sθ̇b
b2)
= ṡb1 + sθ̇b
dt
= s̈b b 1 + (ṡθ̇ + sθ̈)b
b 2 + ṡθ̇(b
b3 × bb 1 ) + sθ̇2 (b
b3 × b
b2 )
= s̈b b 1 + (ṡθ̇ + sθ̈)b
b 2 + ṡθ̇b
b 2 + sθ̇2 (−bb1)
and so
A 2
dP b 1 + (2ṡθ̇ + sθ̈)b
b2 .
= (s̈ − sθ̇2 )b (46)
dt2
Putting in the expressions
b1 = cos(θ)b
b a1 + sin(θ)b
a2 and b2 = − sin(θ)b
b a1 + cos(θ)b
a2
then gives
A 2
dP
= (s̈ − sθ̇2 ) (cos(θ)b
a1 + sin(θ)b
a2 )
dt2
+(2ṡθ̇ + sθ̈) (− sin(θ)ba1 + cos(θ)b
a2 )
or simply
A 2
dP
2
= ((s̈ − sθ̇2 ) cos(θ) − (2ṡθ̇ + sθ̈) sin(θ))b
a1
dt
((s̈ − sθ̇2 ) sin(θ) + (2ṡθ̇ + sθ̈) cos(θ))b
a2 .
This also follows from Equation (45),
A
dP
= (ṡ cos(θ) − sθ̇ sin(θ))b
a1 + (ṡ sin(θ) + sθ̇ cos(θ))b
a2 .
dt
and writing
A 2
dP
2
= (s̈ cos(θ) − ṡθ̇ sin(θ) − ṡθ̇ sin(θ) − sθ̈ sin(θ) − sθ̇2 cos(θ))b
a1
dt
+(s̈ sin(θ) + ṡθ̇ cos(θ) + ṡθ̇ cos(θ) + sθ̈ cos(θ) − sθ̇2 sin(θ))ba2
76
or simply
A 2
dP
2
= ((s̈ − sθ̇2 ) cos(θ) − (2ṡθ̇ + sθ̈) sin(θ))b
a1
dt
+((s̈ − sθ̇2 ) sin(θ) + (2ṡθ̇ + sθ̈) cos(θ))ba2
so that ⎡ ⎤
A A − sin(θ) − cos(θ) 0
d[ RB ] ⎣
= cos(θ) − sin(θ) 0 ⎦
dθ
0 0 1
and then
⎡ ⎤⎡ ⎤T
A A − sin(θ) − cos(θ) 0 cos(θ) − sin(θ) 0
d[ RB ] A
[ RB ]T = ⎣ cos(θ) − sin(θ) 0 ⎦ ⎣ sin(θ) cos(θ) 0 ⎦
dθ
0 0 0 0 0 1
which reduces to ⎡ ⎤
A A 0 −1 0
d[ RB ] A
[ RB ]T = ⎣ 1 0 0 ⎦
dθ
0 0 0
so that
A
d[A RB ] A dθ A d[A RB ] A A A
d[ RB ] A
[A SB ] = [ RB ]T = [ RB ]T = ω [ RB ]T
dt dt dθ dθ
or
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
0 −1 0 0 −ω 0 0 −(A ω B )3 (A ω B )2
[A SB ] = ω ⎣ 1 0 0 ⎦ = ⎣ ω 0 0 ⎦ = ⎣ (A ω B )3 0 −(A ω B )1 ⎦
A B A B
0 0 0 0 0 0 −( ω )2 ( ω )1 0
77
resulting in
⎧ ⎫
⎨ 0 ⎬
A A B A
{ ω }= 0 and hence ω B = ωb
a3
⎩ ⎭
ω
which agrees with the result obtained in Example #10 (Part I). ¥
Consider two reference frames A and B fixed to rigid bodies A and B, respec-
tiively and consider two points P and Q as shown in the figure below.
b3
P BrP
b2
B
PrQ
ArP
b1
a3 ArB BrQ
ArQ
Q
A
a2
a1
Points P and Q as measured by A and B
78
Taking the derivative of both sides of this equation with repect to time t (as
measured in A) we have
A A
d A Q A P d B Q B P
( r − r )= ( r − r ).
dt dt
Next, suppose that A ωB and A αB are the angular velocity and angular acceleration,
respectively, of B as meausred by A. Then since
A
A d(A rP ) A
d(A rQ )
vP = and A
vQ = , (47a)
dt dt
are the velocities of P and Q, respectively, as measured by A, we have
A
A d B Q B P
vQ − A vP = ( r − r )
dt
and using
A
dP B dP A B
= + ω ×P
dt dt
with P = B rQ − B rP , we have
B
A d B Q B P
vQ − A vP = ( r − r ) + A ω B × (B rQ − B rP )
dt
or
A
vQ − A vP = B vQ − B vP + A ωB × P rQ
where
B
B d(B rP ) B B Q
d( r )
vP = and B Q
v = , (47b)
dt dt
are the velocities of P and Q, respectively, as measured by B and P rQ = PQ is
the vector pointing from P to Q, as measured by either A or B, since
P Q
r = B rQ − B rP = A Q
r − A rP
79
The Accelerations of P and Q in A
We can further differentiate (in A) both sides of this Equation (48) and get
A
d(A vQ − A vP ) A
d( B vQ − B vP ) A d(A ω B × PQ)
= +
dt dt dt
tom get
A
A Q d( B vQ − B vP ) A d(A ω B ) A
d(PQ)
a − A aP = + × PQ + A ω B × . (49)
dt dt dt
where
A
A Pd(A vP ) A Q
A A Q
d( v )
a = and a = , (47c)
dt dt
are the accelerations of P and Q, respectively, as measured by A. But using
A
dP B dP A B
= + ω ×P
dt dt
with P = B vQ − B vP , we have
A
d( B vQ − B vP ) B
d( B vQ − B vP ) A B
= + ω × (B vQ − B vP )
dt dt
B Q
= a − B aP + A ωB × (B vQ − B vP )
where
B
B Pd(B vP ) B Q
B B Q
d( v )
a = and a = , (47d)
dt dt
are the accelerations of P and Q, respectively, as measured by B. Putting this into
Equation (49), we have
A Q
a − A aP = B Q
a − B aP + A ωB × (B vQ − B vP )
A A B A
d( ω ) d(PQ)
+ × PQ + A ωB × .
dt dt
We also have
A
A d(A ω B )
αB =
dt
and
A A
d(PQ) d(B rQ − B rP ) B
d( B rQ − B rP ) A B
= = + ω × (B rQ − B rP )
dt dt dt
B Q
= v − B vP + A ω B × (B rQ − B rP )
80
and so
A Q
a − A aP = B Q
a − B aP + A ωB × (B vQ − B vP )
+ A αB × PQ + A ωB × ( B vQ − B vP + A ω B × (B rQ − B rP ))
or simply
A Q
a − A aP = B Q
a − B aP + A ωB × (B vQ − B vP )
+ A αB × PQ + A ωB × ( B vQ − B vP ) + A ωB × (A ω B × PQ)
term is called the centripetal acceleration and is independent of the angular ac-
celeration.
or
A
ωB × (A ωB × PQ) = (A ωB · PQ)(A ωB ) − |A ω B |2 PQ.
In the special case when A ω B ⊥PQ, we get A ω B · PQ = 0, and
A
ω B × (A ω B × PQ) = −|A ωB |2 PQ. (50)
81
13. A Few More Examples
We end this chapter by presenting a few more examples involving the kine-
matics of rigid-body motion.
The figure below shows the top view of a small disk of radius R that rotates
counterclockwise with angular speed ω2 about an axis passing through its center
and perpendicular to the disk. Connected to the center (C) of this small disk is an
arm of length L. One end of the arm is connected to the origin O while the other
is connected to the center of the smaller disk. This arm rotates counterclockwise
about the z axis (coming out of the page) with angular speed ω1 . Note that the
angular speed ω2 is measured by the arm A.
A person, starting at the center of the small disk, walks outward away from the
center of the disk, at a constant speed of v as measured by someone fixed to the
center (C) of the disk. Determine the velocity and acceleration of the person as
measured by O, when the person reaches the edge of the disk
To solve this we let B be the disk, A be the arm and O be the ground in which
point O is fixed. We also define
ex , b
O = {b ey , b
ez } , a1 , b
A = {b a2 , b b1, b
a3 } and B = {b b2 , b
b3 }
as the SRTs in O, A and B, respectively, and these are related via the following
82
tables.
b
ex b
ey b
ez b
ex b
ey b
ez
b
a1 b
a1 · b
ex b
a1 · b
ey b
a1 · b
ey b
a cos(θ) sin(θ) 0
= 1
b
a2 b
a2 · b
ex b
a2 · b
ey b
a2 · b
ey b
a2 − sin(θ) cos(θ) 0
b
a3 b
a3 · b
ex b
a3 · b
ey b
a3 · b
ey b
a3 0 0 1
and
b
a1 b
a2 b
a3 b
a1 b
a2 b
a3
b1
b b
b1 · b
a1 b
b1 · b
a2 b
b1 · b
a3 b1
b cos(ϕ) sin(ϕ) 0
b2 b2 · b b2 · b b2 · b = b2
b b a1 b a2 b a3 b − sin(ϕ) cos(ϕ) 0
b3
b b3 · b
b a1 b3 · b
b a2 b3 · b
b a3 b3
b 0 0 1
and
b
ex b
ey b
ez b
ex b
ey b
ez
b1
b b
b1 · b
ex b
b1 · b
ey b
b1 · b
ez b1
b cos(θ + ϕ) sin(θ + ϕ) 0
b2 b b b2 · b = b2
b b2 · b
ex b2 · b
ey b ez b − sin(θ + ϕ) cos(θ + ϕ) 0
b3
b b
b3 · b
ex b
b3 · b
ey b3 · b
b ez b3
b 0 0 1
Then
O
ω A = ω1b
ez and A
ω B = ω2b
ez
so that
O
ωB = O
ωA + A ω B = (ω1 + ω2 )b
ez .
Then, letting P be the person walking along B and C be the center of the disk
(which is the outer end of the arm). We may assume without any lost in generality
b 1 direction, as seen by B since this
that P walks with constant speed v along the b
is radially outward as seen by B. Thus we have
O
vP = O vC + B vP + O ω B × C rP
= (O ω A × O rC ) + (B vP ) + (O ω B × C rP )
= (ω1bez ) × (Lb b1 + ((ω1 + ω2 )b
a1 ) + v b b1)
ez ) × (Rb
or
O
vP = Lω1 b b 1 + R(ω1 + ω2 )b
a2 + vb b2 .
If we want to express this completely in terms of reference frames in O, we use
the tables above and write
O
vP = Lω1 (− sin(θ)b
ex + cos(θ)b
ey ) + v(cos(θ + ϕ)b
ex + sin(θ + ϕ)b
ey )
+R(ω1 + ω2 )(− sin(θ + ϕ)b ex + cos(θ + ϕ)b
ey )
83
or
O
vP = (v cos(θ + ϕ) − R(ω1 + ω2 ) sin(θ + ϕ) − Lω1 sin(θ))b
ex
+(v sin(θ + ϕ) + R(ω1 + ω2 ) cos(θ + ϕ) + Lω1 cos(θ))bey
where θ is the rotation angle of the arm, ϕ is the rotation angle of the disk. Now
we have ω1 = θ̇ and ω2 = ϕ̇. Thus we have
O
vP = (v cos(θ + ϕ) − R(θ̇ + ϕ̇) sin(θ + ϕ) − Lθ̇ sin(θ))b
ex
+(v sin(θ + ϕ) + R(θ̇ + ϕ̇) cos(θ + ϕ) + Lθ̇ cos(θ))bey
with
O C
a = O αA × O rC + O ωA × (O ωA × O rC )
ez ) × (L cos(θ)b
= (θ̈b ex + L sin(θ)b
ey )
+(θ̇bez ) × ((θ̇b
ez ) × (L cos(θ)b
ex + L sin(θ)bey ))
2
= −Lθ̈ sin(θ)b ex + Lθ̈ cos(θ)bey + Lθ̇ (bez × (cos(θ)b
ey − sin(θ)bex ))
2
= −Lθ̈ sin(θ)b ex + Lθ̈ cos(θ)bey − Lθ̇ (cos(θ)bex + sin(θ)b
ey )
2 2
= (−Lθ̈ sin(θ) − Lθ̇ cos(θ))b ex + (Lθ̈ cos(θ) − Lθ̇ sin(θ))bey
and
C P
a =0
since v = constant,
and
O
αB × C rP = (θ̈ + ϕ̈)b b 1 ) = R(θ̈ + ϕ̈)b
ez × (Rb b2
= R(θ̈ + ϕ̈)(− sin(θ + ϕ)b ex + cos(θ + ϕ)b ey )
= −R(θ̈ + ϕ̈) sin(θ + ϕ)b ex + R(θ̈ + ϕ̈) cos(θ + ϕ)b
ey ,
84
and
O
ω B × (O ω B × C rP ) = ((θ̇ + ϕ̇)b
ez ) × (((θ̇ + ϕ̇)b
ez ) × (Rbb1)
= R(θ̇ + ϕ̇)2 (b b 2 ) = R(θ̇ + ϕ̇)2 (−b
ez × b b1 )
= R(θ̇ + ϕ̇)2 (− cos(θ + ϕ)b
ex − sin(θ + ϕ)bey )
2 2
= −R(θ̇ + ϕ̇) cos(θ + ϕ)b ex − R(θ̇ + ϕ̇) sin(θ + ϕ)b
ey .
which reduces to
O P
a = {−Lθ̈ sin(θ) − Lθ̇2 cos(θ) − 2v(θ̇ + ϕ̇) sin(θ + ϕ)
−R(θ̈ + ϕ̈) sin(θ + ϕ) − R(θ̇ + ϕ̇)2 cos(θ + ϕ)}b ex
+{Lθ̈ cos(θ) − Lθ̇2 sin(θ) + 2v(θ̇ + ϕ̇) cos(θ + ϕ)
+R(θ̈ + ϕ̈) cos(θ + ϕ) − R(θ̇ + ϕ̇)2 sin(θ + ϕ)}b ey .
gives the velocity of P only when P reaches the edge of the disk. Therefore we
cannot simply take
O O P
d( v )
dt
O P
to compute a . We must allow for the equation to be more general where R is
replaced by r (some arbitrary point between the center and the edge of the disk)
and ṙ = v. If we allow this, then
O
vP = (−Lθ̇ sin(θ) + v cos(θ + ϕ) − r(θ̇ + ϕ̇) sin(θ + ϕ))bex
(Lθ̇ cos(θ) + v sin(θ + ϕ) + r(θ̇ + ϕ̇) cos(θ + ϕ))b
ey .
85
and then
O
O P d(O vP )
a =
dt
leads to
O P
a = {−Lθ̈ sin(θ) − Lθ̇2 cos(θ) − v(θ̇ + ϕ̇) sin(θ + ϕ) − ṙ(θ̇ + ϕ̇) sin(θ + ϕ)
−r(θ̈ + ϕ̈) sin(θ + ϕ) − r(θ̇ + ϕ̇)2 cos(θ + ϕ)}b
ex
2
+{Lθ̈ cos(θ) − Lθ̇ sin(θ) + v(θ̇ + ϕ̇) cos(θ + ϕ) + ṙ(θ̇ + ϕ̇) cos(θ + ϕ)
+r(θ̈ + ϕ̈) cos(θ + ϕ) − r(θ̇ + ϕ̇)2 sin(θ + ϕ)}b
ey .
The figure below shows the side view of a rolling wheel of radius R (body C),
which rolls without slipping along a circle in the xy plane by use of an arm of
length L (Body B) that rotates counterclockwise about the positive z axis.
86
Let us fixed the unit vectors
ex , b
A = {b ey , b
ez }
in the ground frame (A). The unit vector b ex is pointing out of the page in this
figure, b
ey is pointing to the right in this figure and b
ez is pointing up in this figure,
so that these are the usual units vectors in rectangular coordinates.
Let’s call the axial, body B and a good choice of unit vectors that are fixed in
B are the usual cylindrical unit vectors
er , b
B = {b eθ , b
ez }
b
ex b
ey b
ez b
ex b
ey b
ez
b
er b
er · b
ex b
er · b
ey b
er · b
ez b
e cos(θ) sin(θ) 0
= r
b
eθ b
eθ · b
ex b
eθ · b
ey b
eθ · b
ez b
eθ − sin(θ) cos(θ) 0
b
ez b
ez · b
ex b
ez · b
ey b
ez · b
ez b
ez 0 0 1
where θ is the counterclockwise angle the arm makes with the unit vector b
ex . This
says, for example, that
b
er = cos(θ)b
ex + sin(θ)b
ey + 0b
ez = cos(θ)b
ex + sin(θ)b
ey
b er − sin(θ)b
ex = cos(θ)b eθ + 0b er − sin(θ)b
ez = cos(θ)b eθ
reading across the first column, and so on. Because of the axial’s motion. The
angular velocity of the wheel (let’s call this body C) as seen by an observer moving
with the axial (let’s call this body B) is
B
ω C = −ϕ̇b
er = −ϕ̇ cos(θ)b
ex − ϕ̇ sin(θ)b
ey
while the angular velocity of the arm (body B) as seen by an observer fixed at
the origin O (let’s call this body A) is
A
ω B = θ̇b
ez .
87
Let’s call the center of the wheel, which is the right end of the arm, point G. Then
the velocity of the contact point with the floor (call if F) as measured by O (body
A) is
O
vF = O vG + A ω C × G rF
= A ω B × O rG + (A ω B + B ωC ) × G rF
ez ) × (Lb
= (θ̇b er + Rb
ez ) + (−ϕ̇b ez ) × (−Rb
er + ϕ̇b ez )
= Lθ̇beθ − Rϕ̇b
eθ = (Lθ̇ − Rϕ̇)beθ ,
and since the wheel rolls without slipping, this must be zero and we have
Lθ̇ = Rϕ̇.
We also have
A
ω C × (A ωC × G rF )
= A ω C × ((−ϕ̇b ez ) × (−Rb
er + θ̇b ez ))
= (−ϕ̇b ez ) × (−Rϕ̇b
er + θ̇b eθ )
2
= Rϕ̇vber + Rϕ̇ b ez
and so
O F O G
a = er − Rϕ̈b
a + Rϕ̇θ̇b er + Rϕ̇2b
eθ + Rϕ̇θ̇b ez
88
or
O F O G
a = a + 2Rϕ̇θ̇b eθ + Rϕ̇2b
er − Rϕ̈b ez .
The acceleration of point G (on body B) as measured by point O (on body A) is
O G
a = A αB × O rG + A ω B × (A ω B × O rG )
ez ) × (Lb
= (θ̈b er + Rb ez ) × ((θ̇b
ez ) + (θ̇b ez ) × (Lb
er + Rb
ez ))
2
= Lθ̈b ez × b
eθ + Lθ̇ (b eθ )
or
O G
a = −Lθ̇2b
er + Lθ̈b
eθ .
The acceleration of point F as measured by O is then
O F
a = −Lθ̇2b
er + Lθ̈b
eθ + 2Rϕ̇θ̇b eθ + Rϕ̇2b
er − Rϕ̈b ez
or
O F
a = (2Rϕ̇θ̈ − Lθ̇2 )b eθ + Rϕ̇2b
er + (Lθ̈ − Rϕ̈)b ez .
But
Lθ̇ = Rϕ̇ so that Lθ̈ = Rϕ̈
and so
O F
a = θ̇(2Rϕ̇ − Rϕ̇)b eθ + Rϕ̇2b
er + (0)b er + Rϕ̇2b
ez = Rϕ̇θ̇b ez
or µ ¶
O F Rϕ̇
a = Rϕ̇ er + Rϕ̇2b
b ez
L
or µ ¶
O F 2 R
a = Rϕ̇ er + b
b ez .
L
er , b
Since {b eθ , b
ez } forms a SRT, we have
O F
√ p
a = O aF · O aF = Rϕ̇2 (R/L)2 + 12
or p
O F
a = Rϕ̇2 1 + (R/L)2
and this gives the acceleration of the point on the wheel that is in contact with
the floor.
89
A Check to these Calculations - Getting An Arbitrary Position First
Then
O
O d(O rE ) O
d(b
er ) O
d(b
eθ )
vE = =L − Rϕ̇ cos(ϕ)b
eθ − R sin(ϕ) + Rϕ̇ sin(ϕ)b
ez
dt dt dt
since b
ez is fixed in frame A (in which O is the fixed origin). This leads to
O
vE = L(A ω B × b eθ − R sin(ϕ)(A ω B × b
er ) − Rϕ̇ cos(ϕ)b eθ ) + Rϕ̇ sin(ϕ)b
ez
ez × b
= L(θ̇b er ) − Rϕ̇ cos(ϕ)b
eθ − R sin(ϕ)(θ̇bez × beθ ) + Rϕ̇ sin(ϕ)b
ez
eθ − Rϕ̇ cos(ϕ)b
= Lθ̇b eθ − R sin(ϕ)(−θ̇ber ) + Rϕ̇ sin(ϕ)b ez
which reduces to
O
vE = Rθ̇ sin(ϕ)b
er + (Lθ̇ − Rϕ̇ cos(ϕ))b
eθ + Rϕ̇ sin(ϕ)b
ez .
or
O
vF = (Lθ̇ − Rϕ̇)b
eθ
90
and this is zero when Lθ̇ = Rϕ̇, which agrees with our earlier results. To check
our calculation of O aF , we compute
O
O E d(O vE ) O
d(b
er )
a = er + Rθ̇ϕ̇ cos(ϕ)b
= Rθ̈ sin(ϕ)b er + Rθ̇ sin(ϕ)
dt dt
O
d(beθ )
+(Lθ̈ − Rϕ̈ cos(ϕ) + Rϕ̇2 sin(ϕ))b eθ + (Lθ̇ − Rϕ̇ cos(ϕ))
dt
2
+(Rϕ̈ sin(ϕ) + Rϕ̇ cos(ϕ))b ez .
or
O E
a = Rθ̈ sin(ϕ)b
er + Rθ̇ϕ̇ cos(ϕ)b ez × b
er + Rθ̇ sin(ϕ)(θ̇b er )
2
+(Lθ̈ − Rϕ̈ cos(ϕ) + Rϕ̇ sin(ϕ))b ez × b
eθ + (Lθ̇ − Rϕ̇ cos(ϕ))(θ̇b eθ )
2
+(Rϕ̈ sin(ϕ) + Rϕ̇ cos(ϕ))b ez
= Rθ̈ sin(ϕ)b
er + Rθ̇ϕ̇ cos(ϕ)ber + Rθ̇2 sin(ϕ)b
eθ
2
+(Lθ̈ − Rϕ̈ cos(ϕ) + Rϕ̇ sin(ϕ))b eθ + (Lθ̇ − Rϕ̇ cos(ϕ))(−θ̇ber )
2
+(Rϕ̈ sin(ϕ) + Rϕ̇ cos(ϕ))b ez
which reduces to
O E
a = (Rθ̈ sin(ϕ) + 2Rθ̇ϕ̇ cos(ϕ) − Lθ̇2 )b
er
2 2
+(Lθ̈ − Rϕ̈ cos(ϕ) + R(ϕ̇ + θ̇ ) sin(ϕ))b
eθ
2
+(Rϕ̈ sin(ϕ) + Rϕ̇ cos(ϕ))b ez .
which reduces to
O F
a = (2Rθ̇ϕ̇ − Lθ̇2 )b eθ + Rϕ̇2b
er + (Lθ̈ − Rϕ̈)b ez .
But since Lθ̇ = Rϕ̇ for all time, we also (after differentiating with respect to time)
have Lθ̈ = Rϕ̈ and so the above expression reduces to
O F R R2
a = (2R ϕ̇2 − L 2 ϕ̇2 )b eθ + Rϕ̇2b
er + (0)b ez
L L
91
or µ ¶
R2 2
O F R
a = er + Rϕ̇2b
ϕ̇ b ez = Rϕ̇2 er + b
b ez
L L
and this agrees with the result obtained above.
Note that we could have made the calculation of O aE a little simpler by putting
Lθ̇ = Rϕ̇ into the expression for O vE before taking its derivative, which is valid
since Lθ̇ = Rϕ̇ is true for all time. If we do this we get
O
vE = Rθ̇ sin(ϕ)ber + (Lθ̇ − Rϕ̇ cos(ϕ))b
eθ + Rϕ̇ sin(ϕ)b
ez
R
= R ϕ̇ sin(ϕ)b er + (LRϕ̇/L − Rϕ̇ cos(ϕ))beθ + Rϕ̇ sin(ϕ)b
ez
L
R2
= er + Rϕ̇(1 − cos(ϕ))b
ϕ̇ sin(ϕ)b eθ + Rϕ̇ sin(ϕ)b
ez .
L
Then
O
O E d(O vE ) R2 R2 2 R2 O
d(ber )
a = = ϕ̈ sin(ϕ)ber + ϕ̇ cos(ϕ)b er + ϕ̇ sin(ϕ)
dt L L L dt
O
d(b
e θ)
+Rϕ̈(1 − cos(ϕ))b eθ + Rϕ̇2 sin(ϕ)beθ + Rϕ̇(1 − cos(ϕ))
dt
+Rϕ̈ sin(ϕ)b ez + Rϕ̇2 cos(ϕ)b ez
R2 R2 2 R2
= ϕ̈ sin(ϕ)b
er + ϕ̇ cos(ϕ)b
er + ez × b
ϕ̇ sin(ϕ)(θ̇b er )
L L L
+Rϕ̈(1 − cos(ϕ))b eθ + Rϕ̇2 sin(ϕ)beθ + Rϕ̇(1 − cos(ϕ))(θ̇b ez × b eθ )
2
+Rϕ̈ sin(ϕ)b ez + Rϕ̇ cos(ϕ)b ez
2 2
R R 2 R2
= ϕ̈ sin(ϕ)b
er + ϕ̇ cos(ϕ)b
er + ϕ̇ sin(ϕ)(θ̇b
eθ )
L L L
+Rϕ̈(1 − cos(ϕ))b eθ + Rϕ̇2 sin(ϕ)beθ + Rϕ̇(1 − cos(ϕ))(−θ̇b er )
2
+Rϕ̈ sin(ϕ)b ez + Rϕ̇ cos(ϕ)b ez
or
µ ¶
O E R2 R2 2
a = ϕ̈ sin(ϕ) + ϕ̇ cos(ϕ) − Rθ̇ϕ̇(1 − cos(ϕ)) b er
L L
µ ¶
2 R2
+ Rϕ̈(1 − cos(ϕ)) + Rϕ̇ sin(ϕ) + θ̇ϕ̇ sin(ϕ) beθ
L
+R(ϕ̈ sin(ϕ) + ϕ̇2 cos(ϕ))bez .
92
Putting in θ̇ = Rϕ̇/L, this reduces to
µ 2 ¶
O E R R2 2 R2 2
a = ϕ̈ sin(ϕ) + ϕ̇ cos(ϕ) − ϕ̇ (1 − cos(ϕ)) ber
L L L
³ ´
+ Rϕ̈(1 − cos(ϕ)) + Rϕ̇2 sin(ϕ) + Rθ̇2 sin(ϕ) b eθ
+R(ϕ̈ sin(ϕ) + ϕ̇2 cos(ϕ))b
ez .
or
µ ¶
O E R2 2R2 2 R2 2
a = ϕ̈ sin(ϕ) + ϕ̇ cos(ϕ) − ϕ̇ b er
L L L
³ ´
+ Rϕ̈(1 − cos(ϕ)) + R(ϕ̇2 + θ̇2 ) sin(ϕ) beθ
+R(ϕ̈ sin(ϕ) + ϕ̇2 cos(ϕ))b
ez ,
The Big Advantage in Using the Formulas for Velocity and Acceleration
The student should note that there are two big advantages in using the for-
mulas for velocity
A Q
v = A vP + B vQ + A ωB × P rQ
and acceleration
A Q A P
a = a + B aQ + A αB × P rQ
+2(A ωB × B vQ ) + A ω B × (A ω B × P rQ ).
over computing the position first and then differentiating. The first is that it is
usually harder to figure out the geometry in getting position, since the position of
an arbitrary point must be determined and not just the specific point in question.
In this example, we had to determine the position of an arbitrary point on the
93
edge of the wheel and not just the point of contact between the edge of the wheel
and the floor, which was just
O F
r = Lb er .
in this example, which is much simpler than having to get
O E
er − R sin(ϕ)b
r = Lb eθ + R(1 − cos(ϕ))b
ez .
first. Second, all the derivatives are already taken in the above two expressions
and so no first and second derivative of O rE need be computed.
Determine the velocity of the free end of the smaller link (point P) as measured
by point O. To solve this we let
O
ω1 = θ̇b
ez
be the angular velocity of link 1 as measured by O and
1
ω2 = −ϕ̇b
eθ
94
be the angular velocity of link 2 as measured by link 1. Then
O
ω2 = O
ω1 + 1 ω2 = θ̇b
ez + (−ϕ̇b
eθ )
which yields
O
vP = (θ̇b
ez ) × (L1be ) + (−ϕ̇b ez ) × (L2 cos(ϕ)b
eθ + θ̇b er + L2 sin(ϕ)b
ez )
¯ r ¯
¯ b
er b
eθ b
ez ¯
¯ ¯
eθ + ¯
= L1 θ̇b ¯ 0 −ϕ̇ θ̇ ¯
¯
¯ L2 cos(ϕ) 0 L2 sin(ϕ) ¯
= L1 θ̇b er + L2 θ̇ cos(ϕ)b
eθ + (−L2 ϕ̇ sin(ϕ)b eθ + L2 ϕ̇ cos(ϕ)b
ez )
or just
O
vP = −L2 ϕ̇ sin(ϕ)b
er + θ̇(L1 + L2 cos(ϕ))b
eθ + L2 ϕ̇ cos(ϕ)b
ez .
or just
O P
r = (L1 + L2 cos(ϕ))b
er + L2 sin(ϕ)b
ez .
Then
O
O d(O rP ) O
db
er
vP = = −L2 ϕ̇ cos(ϕ)b
er + (L1 + L2 cos(ϕ)) + L2 ϕ̇ cos(ϕ)b
ez
dt dt
or
O
vP = −L2 ϕ̇ cos(ϕ)b
er + θ̇(L1 + L2 cos(ϕ))b
eθ + L2 ϕ̇ cos(ϕ)b
ez
which agrees with the result above. ¥ For practice, the student should compute
the acceleration of P as measured by O using Equation (51) and the result for
O P
v .
95
Example #14: Throwing A Baton
A uniform rigid rod of length L is thrown into the air with an initial center-
of-mass speed v0 in a direction θ0 with respect to the positive horizontal. The
rod is also given an angular velocity ω in the counterclockwise direction with its
orientation horizontal at the start of the motion. The center-of-mass of the rod
will follow the path of a projectile and throughout its motion, the rotation of the
rod is about an axis perpendicular to the plane of the projectile as shown below
If the orientation of the rod at the end of its motion is to be the same as its initial
orientation (as shown above), show that ω, v0 and θ0 must be chosen to satisfy
the equation
v0 ω sin(θ0 ) = nπg
where n is a positive integer. Plot on the same scale, the motion of each end of
the rod (along with its center) from beginning to end using the values: L = 2 m,
v0 = 16 m/sec, g = 9.8 m/sec2 , θ0 = 76◦ , n = 2, and ω as computed using the
equation above. Assume no air resistance throughout the motion and constant g.
We first want to show that if the orientation of the rod at the end of its motion
is to be the same as its initial orientation (as shown above), then ω, v0 and θ0
96
must be chosen to satisfy the equation
ωv0 sin(θ0 )
= nπ
g
where n is a positive integer. Toward this end, we note that the position of the
center of the rod, as measured form the launch point O) is at
µ ¶
O C O C 1 2
( x , y ) = v0 t cos(θ0 ), v0 t sin(θ0 ) − gt
2
so that
O P
x = v0 t cos(θ0 ) + r cos(ωt)
and
O P 1
y = v0 t sin(θ0 ) − gt2 + r sin(ωt).
2
The velocity of this point can be computed using either
O P
ẋ = v0 cos(θ0 ) − rω sin(ωt)
and
O P
ẏ = v0 sin(θ0 ) − gt + rω cos(ωt),
or
O
vP = O vC + O ωrod × C rP
= v0 cos(θ0 )bi + (v0 sin(θ0 ) − gt)bj + (ω k)
b × (r cos(ωt)bi + r sin(ωt)bj)
resulting in
O
vP = (v0 cos(θ0 ) − rω sin(ωt))bi + (v0 sin(θ0 ) − gt + rω cos(ωt))bj,
which is the same result. The center of the rod hits the ground at time T where
1 2v0 sin(θ0 )
v0 T sin(θ0 ) − gT 2 = 0 resulting in T =
2 g
97
and it will be located at the point
At this moment we want the right end of the rod to be located at the point D+L/2
and so we must have
L L
cos(ωT ) = or cos(ωT ) = 1
2 2
which leads to ωT = 2nπ, so that
gnπ
ω= or v0 ω sin(θ0 ) = nπg
v0 sin(θ0 )
and so the first part of the problem is complete. To address the plotting part of
the problem, we note that the left end of the rod has coordinates
O L 1
x = v0 t cos(θ0 ) − L cos(ωt)
2
and
O L 1 1
y = v0 t sin(θ0 ) − gt2 − L sin(ωt)
2 2
and the right end has coordinates
O R 1
x = v0 t cos(θ0 ) + L cos(ωt)
2
and
O R 1 1
y = v0 t sin(θ0 ) − gt2 + L sin(ωt)
2 2
A plot of the motion for the ends of the rod, using the values: L = 2 m, v0 = 16
m/sec, g = 9.8 m/sec2 , θ0 = 76◦ and n = 2 for that
gnπ (9.8)(2)π
ω= = ' 3.97 rad/sec
v0 sin(θ0 ) 16 sin(76◦ )
98
is shown in the figure below
for 0 ≤ t ≤ T , with
v0 cos(θ0 ) − rω sin(ωt) = 0
and
v0 sin(θ0 ) − gt + rω cos(ωt) = 0
and solving each of these for r, we get
v0 cos(θ0 ) gt − v0 sin(θ0 )
r= and r=
ω sin(ωt) ω cos(ωt)
and so
v0 cos(θ0 ) gt − v0 sin(θ0 )
=
ω sin(ωt) ω cos(ωt)
99
or
v0 cos(ωt) cos(θ0 ) + v0 sin(ωt) sin(θ0 ) = gt
or
v0 cos(ωt − θ0 ) = gt
gives the value of t and either
v0 cos(θ0 ) gt − v0 sin(θ0 )
r= or r=
ω sin(ωt) ω cos(ωt)
gives the point on the rod having zero velocity. These are the corner points in the
above plot. ¥
100