3D Rotational Analysis

by
Jesús Dapena

Department of Kinesiology
Indiana University
Bloomington, IN 47405
USA

Notes for a tutorial given at the 19th Congress of the International Society of
Biomechanics, Dunedin, New Zealand, 2003.

Tel. 1-812-855-8407
E-mail: [email protected]
Web: http://www.indiana.edu/~sportbm/home.html
 1

2D translation = easy.

3D translation = easy to make the transition, just add one more dimension.

2D rotation = harder, probably because (1) moment of inertia is variable; (2) vectors are

less intuitive than in translation.

3D rotation = the hardest, because it’s not such a simple transition from 2D as it is in

translation.
 2

Most interesting parameters:

Translation:

0th derivative level = linear location

1st derivative level = linear velocity, linear momentum

2nd derivative level = linear acceleration, force

Rotation:

0th derivative level = angular location

1st derivative level = angular velocity, angular momentum

2nd derivative level = angular acceleration, torque

 3

Angular momentum: The equation “H = I ω” is not always valid. It’s only good for a

rigid object rotating in 2D.

Let’s consider a very general situation: an amoeba. For a single particle i, the angular

momentum HOi about a reference point O can always be computed correctly by:

H Oi = mi (ri " v i )

dri
H Oi = mi (ri " )
dt

(ri " dri )

H Oi = mi
dt
!

!
 4

Graphical concept of cross product

Cross product of two vectors:

a×b=c

Magnitude of the cross product:

c = a b sin θ

The magnitude of the cross product is equal to two times the area of the triangle formed

by the two vectors:

This holds whether the two vectors are drawn with their tails together or with one vector

starting at the tip of the other:

 5

Back to angular momentum ...

Going back to our amoeba, we can now see that the angular momentum of the particle of

material is directly proportional to the area swept per unit of time by the particle relative

to the reference point:

HOi = 2 ⋅ mi ⋅ amount of area swept per second

This allows us to compare graphically the amounts of angular momentum of different

particles:
 6

To calculate the angular momentum of the whole system (the whole amoeba!), we would

need to use the equation:

n
H O = # mi (ri " v i )
i=1

A problem with using this equation is that it would require us to compute the angular
! momentum values of an infinite number of individual particles. So we are lucky that we

don’t study amoebas!

The human body can be considered a series of rigid segments that can rotate relative to

each other at joints. So the human body is something intermediate between a single rigid

block and an amoeba. We will deal with it later.

 7

For now, we will continue developing our equations, to obtain an expression for the

angular momentum of a rigid object.

If all the points of the amoeba were only able to move in circles about point O, we would

have:

n
H O = # mi ri " v i
i=1
n
= # mi ri " $ i " ri
i=1

and thus,
! n
2
H O = # mi ri " i
i=1

If all particles also have the same angular velocity vector, we have that:
! n
2
H O = (# mi ri )"
i=1

HO = I ω
! This is the traditional formula for angular momentum. We have seen that it can only be

applied to a rigid object rotating in 2D.

 8

Angular momentum needs to be calculated about a given point

Angular momentum needs to be calculated about either:

(a) an inertial point, i.e., a point that is static or that translates at constant linear velocity

or

(b) the center of mass of the system. This is valid even if the center of mass does not

have constant linear velocity.

Usually, (b) is the best choice.

 9

Calculation of the angular momentum of the human body about its own center of

mass

The human body is considered to be composed of rigid segments that rotate relative to

each other.

Each segment has two angular momentum elements:

remote angular momentum: associated with the motion of the segment’s c.m. about the

c.m. of the whole body

local angular momentum: associated with the motion of the segment about its own c.m.

remote angular momentum = shank mass ⋅ 2 ⋅ area swept per second

(i.e. like amoeba particle)

local angular momentum = shank moment of inertia about own c.m. ⋅ shank angular
velocity
(i.e. like rigid object)
 10

The local terms of angular momentum are usually much smaller than the remote terms.

So if you ignore the local terms, you will have some error but not very much. This is

very useful for qualitative analysis, because the remote terms can be used to make

estimates of the angular momentum of each body segment:

In the tutorial, we will look at the angular momentum of each segment during the

airborne motion of a hitch-kick style (“bicycling”) long jump.

In 2D analysis, clockwise areas indicate an angular momentum vector pointing into the

blackboard; counterclockwise areas indicate an angular momentum vector pointing out

from the blackboard toward us.

 11

3D Rotation

Qualitative analysis of angular momentum

There are two possible ways to make a qualitative analysis of angular momentum in a 3D

motion:

(1) Try to make the analysis directly in 3D. This involves visualizing 3D vectors in a

multitude of directions. It’s difficult to do, and generally not a good idea.

(2) Look at the 3D motion as three separate projected 2D motions, by viewing in

succession from three different directions perpendicular to each other. Each of these

orthogonal directions will give us information about the angular momentum vector

component in the direction along the line of sight.

 12

Quantitative analysis of angular momentum

Those are the methods that we would use for an intuitive, qualitative approach. To make

accurate calculations with a computer, we would use other methods, as we will see next.

Calculating the remote terms of angular momentum is very simple, both in 2D and 3D

analysis. The formula:

n
H O = # mi (ri " v i )
i=1

would be applied to the linear location and linear velocity of the center of mass of each
! segment relative to the whole body center of mass.

In a true 2D motion, calculating the local terms of angular momentum would also be

quite simple. We would just need to multiply each segment’s centroidal moment of

inertia by the segment’s angular velocity about its own center of mass. This is the simple

formula H = I ω.

However, in 3D motion there is a problem for the calculation of the local terms of

angular momentum. The formula H = I ω does not work in 3D motion. We will see

why.
 13

Both objects in the drawing above are rotating about a vertical axis, and the angular

velocity vector ω is the same for both.

In the view from the positive Z direction, the particles that constitute the two objects

sweep the same amounts of counterclockwise areas, and in the view along the X axis no

areas are swept at all in either case at the instant shown in the drawing. However, in the

view along the Y axis the object on the left is sweeping no area, while the object on the

right is sweeping counterclockwise areas when viewed from the negative Y direction:

This implies that the object on the left has only a positive Z component of angular

momentum, while the object on the right has a positive Z component and a negative Y

component of angular momentum:

Therefore, in the object on the right the angular velocity and angular momentum vectors

do not point in the same direction. Consequently, H ≠ I ω.

 14

It is clear that the problem stems from the fact that the mass on one side of the angular

velocity vector is higher up along the vector than the mass on the other side:

This asymmetry is what makes the angular momentum be misaligned with the angular

velocity. The angular momentum and angular velocity vectors would be aligned if the

masses on either side were set more evenly:

Physicists have found a way to quantify this asymmetry, and to use it to calculate correct

values for the angular momentum of a rigid object in 3D.

 15

Moments of inertia:
n n
I XX = " dm(Yi2 + Z 2i ) I YY = " dm(X 2i + Z 2i )
i=1 i=1

n
I ZZ = " dm(X 2i + Yi2 )
! !i=1

! of inertia:
Products
n n
I XY = "$ dm(X i # Yi ) I XZ = "$ dm(X i # Z i )
i=1 i=1

n n
I YX = "$ dm(Yi # X i ) I YZ = "$ dm(Yi # Z i )
! i=1 ! i=1

n n
I ZX = "$ dm(Z i # X i ) I ZY = "$ dm(Z i # Yi )
! i=1 ! i=1

! (NOTE: In some books!

there are no minus signs before the sigmas in the definitions of the products of

inertia. Such a definition is valid too, but it requires changes also in other equations. Be sure not to mix

the two alternative conventions!)

 16

Let’s look at the IXZ product of inertia:

We see that the products of inertia give us good information about the “asymmetry” of

the object.
 17

Physicists like to show the moments of inertia and the products of inertia together in an

ordered array:

"I XX I XY I XZ &
$ $
#I YX I YY I YZ '
$I I ZY I ZZ $(
% ZX

This is called the tensor of inertia. It’s often designated simply as {I} .
!

The formula to calculate the angular momentum of a rigid!object using the moments and

products of inertia is the following:

H= IXX ω X + IXY ω Y + IXZ ω Z

+ IYX ω X + IYY ω Y + IYZ ω Z

+ IZX ω X + IZY ω Y + IZZ ω Z

where ωX, ωY and ωZ are the three components of the ω vector.

This can be expressed as the product of two matrices:

"I XX I XY I XZ & "* X &

$ $ $ $
H = #I YX I YY I YZ ' ) #* Y '
$I I ZY I ZZ $( $% * Z $(
% ZX

#" X '
% %
! where $" Y ( is the angular velocity vector, which can be designated simply as {"} .
%" %
& Z)

!
!
 18

So in the end we have that H = {I} {ω}, where {I} and {ω} are actually matrices.

Note that this is not the same as H = I ω, which would not produce the correct results,

because it is missing the terms involving the products of inertia.

Having said all this, it IS possible to calculate the angular momentum of a rigid object in

3D without using the products of inertia!

But before we learn how to do this we need to define the concept of principal axes.
 19

Principal axes
 20

In objects that have symmetry, the principal axes are easy to find. For instance, in the

human trunk:

We can use principal axes to avoid having to use the products of inertia.
 21

Calculation of local angular momentum in 3D without using products of inertia

Let’s consider a cylinder with elliptical cross-section to represent the human trunk.

The XYZ axes shown here are not principal axes. They are parallel to the axes of the

reference frame attached to the ground.

The standard equation to calculate angular momentum is:

"I XX I XY I XZ & "* X &

$ $ $ $
H = #I YX I YY I YZ ' ) #* Y '
$I I ZY I ZZ $( $% * Z $(
% ZX

! We will now show a temporary non-rotating reference frame abc such that axes a, b and c

happen to coincide instantaneously with the principal axes of the cylinder:

 22

The angular velocity of the cylinder will first be calculated in terms of the XYZ reference

frame attached to the ground, using standard film analysis or video analysis procedures.

Then, we can choose to express the angular velocity of the cylinder in terms of the abc

reference frame,. If we do this, we will have:

"I aa I ab I ac & "* a &

$ $ $ $
H = #I ba I bb I bc ' ) #* b '
$I I cb I cc $( $% * c $(
% ca

But, since axes a, b and c are principal axes, the products of inertia are zero. Therefore:
! "I aa 0 0 & "* a &
$ $ $ $
H = #0 I bb 0 ' ) #* b ' = Iaa ω a + Ibb ω b + Icc ω c
$0 0 I cc $( $% * c $(
%

For instance, if:

! Iaa = 1.4 Ibb = 1.6 Icc = 0.5

and if:

ω a = 27 ua ω b = 31 ub ω c = 45 uc

where ua, ub and uc are the unit vectors of the abc reference frame, we would have:

H = (1.4 ⋅ 27) ua + (1.6 ⋅ 31) ub + (0.5 ⋅ 45) uc

= 37.8 ua + 49.6 ub + 22.5 uc Kg ⋅ m2/s

 23

So that would be the angular momentum value that we were looking for. But there is a

slight problem: This angular momentum vector is expressed in terms of the abc reference

frame instead of the XYZ reference frame, which is inconvenient. But this problem is

actually very easy to solve.

Remember that the angular velocity vector was initially expressed in terms of the XYZ

reference frame. When we project this vector onto the a, b and c axes, the ω a, ω b and ω c

vectors that we get will each be expressed still in terms of the XYZ reference frame. For

instance, we could have:

ω a = 27 ua = 19.2 uX + 16.9 uY + 8.8 uZ (for example!)

ω b = 31 ub = -19.7 uX + 23.8 uY - 2.8 uZ (for example!)

ω c = 45 uc = -13.7 uX – 6.4 uY + 42.4 uZ (for example!)

where uX, uY and uZ are the unit vectors of the XYZ reference frame

In such case, the value of the angular momentum vector would be:

H= 1.4 ⋅ (19.2 uX + 16.9 uY + 8.8 uZ)

+ 1.6 ⋅ (-19.7 uX + 23.8 uY - 2.8 uZ)

+ 0.5 ⋅ (-13.7 uX – 6.4 uY + 42.4 uZ)

= -11.5 uX + 58.5 uY + 29.0 uZ Kg ⋅ m2/s

So, in effect, we are using H = Iaa ω a + Ibb ω b + Icc ω c, but with ω a, ω b and ω c expressed

in terms of the XYZ reference frame, which is a trivial thing to do, as we have seen.
 24

That is how we would calculate the local angular momentum of the trunk about its own

c.m.

The other segments are usually considered cylinders of circular (instead of elliptical)

cross-section. So for them the longitudinal axis is clearly defined, but there are infinite

possible orientations for the other two axes. Any two axes will do, as long as they are

both perpendicular to the longitudinal axis and perpendicular to each other. This

ambiguity allows us to calculate the local terms of angular momentum of these segments

with even greater ease than for the trunk.

We will pick the a or the b axis to be in the plane defined by ω and by the longitudinal

axis c. For instance, let’s assume that we choose to make the a axis be in that plane:

In this case, ω b will be zero. Therefore:

H = Iaa ω a + Icc ω c

where ω a and ω c will be expressed in terms of the XYZ reference frame.

 25

Angular momentum of the whole body:

14
H= " mi (ri × vi) ← remote terms
i=1

13
+ " ILONGi ω TWi + ITRANSVi ω SOMi ← non-trunk local terms
! i=1

+ ILONG14 ω TW14 + IML14 ω SOM-ML14 + IAP14 ω SOM-AP14 ← trunk local term

!
 26

References

For the representation of angular momentum by areas swept, see:

Hopper, Bernard J. The mechanics of human movement. American Elsevier, 1973.

This excellent book is out of print, but you may find it at a library. Or you can purchase a

used copy through Internet. For instance, try at:

http://www.amazon.com/exec/obidos/subst/books/misc/bibliofind.html

or at:

http://www.abebooks.com

For the computation of 3D angular momentum, see:

Dapena, J. A method to determine the angular momentum of a human body about

three orthogonal axes passing through its center of gravity. Journal of

Biomechanics 11:251-256, 1978.