Orthogonal matrices and Euler angles, the angular velocity and its additivity, the velocity and acceleration fields, the instantaneous axis of rotation, the centre of mass, the tensor of inertia, Steiner's theorem and the principal axes.
Sources: Notes.pdf, chapters 9 and 10; Lectures 7 and 8.
Central idea of the topic
A rigid body is kinematically the same object as a reference frame attached to it: its whole velocity field follows from the velocity of one body point and the angular velocity $\boldsymbol\omega_{B0}$, its attitude follows from three Euler angles, and its instantaneous motion is always a screw about the instantaneous axis of rotation and slip. Its mass distribution enters the mechanics only through three objects built from the zeroth, first and second moments of that distribution: the total mass $m$, the centre of mass $G$ and the tensor of inertia $\boldsymbol{\mathcal I}_O$, which changes basis like any second-order tensor, has an orthonormal principal basis, and moves from $G$ to any point $O$ by Steiner's theorem.
This note follows Notes.pdf Chapter 9 (Kinematics of rigid bodies, Eqs. 9.1 to 9.24) and Chapter 10 (Geometry of masses and inertia properties, Eqs. 10.1 to 10.48), and the decks of Lecture 7 (Rigid body kinematics, Slides 1 to 14) and Lecture 8 (Inertia properties, Slides 1 to 22), in the order of the notes. Earlier results that are used are cited, not re-derived: the change of basis of Chapter 1 (Eqs. 1.6 to 1.23), the tensor product (Eqs. 1.34 to 1.48), the double vector product (Eq. 1.37), and the Coriolis theorem with the composition laws of Chapter 3 (Eqs. 3.8 to 3.27). See also Concept - Vector Bases Rotation Matrices and Poisson Theorem, Concept - Attitude of a Reference Frame and Euler Angles and Concept - Angular Velocity Angular Acceleration and Coriolis Theorem.
principal moments and principal directions of inertia
1. The rigid body model and body-fixed reference frames
The point particle model ignores the dimensions of an object, its mass distribution and its orientation (attitude) in space. The rigid body model keeps the last two: a rigid body is a body of finite, indeformable geometry (Notes.pdf, Sec. 9; [Lecture 7, Slide 3]). Like the point particle, it is an idealisation: all materials deform somewhat under load, but in many Flight Mechanics problems these small deformations do not affect the motion.
The defining kinematic constraint is that the distance between any two body points stays constant in time ([Lecture 7, Slide 3]). The first consequence is that angles between body segments are also constant.
Rigidity keeps distances and anglesLaw: rigid body constraint; definition of the angle through the dot productSource: [Lecture 7, Slide 3]; Notes.pdf, Eq. 1.24
For three body points $A$, $B$, $C$ the constraint states that $\lvert\mathbf{BA}\rvert$, $\lvert\mathbf{BC}\rvert$ and $\lvert\mathbf{AC}\rvert$ are constant. The angle $ABC$ follows from them:
$$\begin{aligned}
\lvert\mathbf{AC}\rvert^2 &= (\mathbf{BC}-\mathbf{BA})\cdot(\mathbf{BC}-\mathbf{BA}) \quad &\text{(since $\mathbf{AC} = \mathbf{AB}+\mathbf{BC} = \mathbf{BC}-\mathbf{BA}$)} \\
&= \lvert\mathbf{BC}\rvert^2 + \lvert\mathbf{BA}\rvert^2 - 2\,\mathbf{BA}\cdot\mathbf{BC} \quad &\text{(distributive and commutative dot product, Eqs. 1.25-1.26)} \\
\mathbf{BA}\cdot\mathbf{BC} &= \lvert\mathbf{BA}\rvert\,\lvert\mathbf{BC}\rvert\cos\widehat{ABC} \quad &\text{(definition of the dot product, Eq. 1.24)} \\
\cos\widehat{ABC} &= \frac{\lvert\mathbf{BA}\rvert^2+\lvert\mathbf{BC}\rvert^2-\lvert\mathbf{AC}\rvert^2}{2\,\lvert\mathbf{BA}\rvert\,\lvert\mathbf{BC}\rvert} \quad &\text{(eliminate the dot product between both lines)}
\end{aligned}$$
Units: the cosine is a ratio $[\mathrm{L^2}]/[\mathrm{L^2}] = [1]$
Meaning: a rigid body keeps every length and every angle, so a triad of vectors drawn on it keeps its shape: it can only translate and rotate.
A body-fixed reference frame $S_B$ is built from three non-collinear body points $P$, $Q$, $R$: the origin is $P$ and the vectors $\mathbf{PQ}$, $\mathbf{PR}$ and their vector product are attached to the body (Notes.pdf, Sec. 9; [Lecture 7, Slide 4]). They need not be orthonormal, but we always choose a right-handed orthonormal basis built from them, as below. The choice of origin and basis is not unique.
Figure 1. Body-fixed frame $S_B$ built from the non-collinear body points $P$, $Q$, $R$: $\mathbf i_B$ along $\mathbf{PQ}$, $\mathbf k_B$ normal to the plane $PQR$ and $\mathbf j_B = \mathbf k_B\times\mathbf i_B$, with the fixed frame $S_0$ at $O$..
An orthonormal right-handed body basis from three body pointsLaw: normalisation and vector product; right-hand ruleSource: Notes.pdf, Sec. 9; [Lecture 7, Slide 4]
$$\begin{aligned}
\mathbf i_B &= \frac{\mathbf{PQ}}{\lvert\mathbf{PQ}\rvert} \quad &\text{(unit vector along the first body segment)} \\
\mathbf k_B &= \frac{\mathbf{PQ}\times\mathbf{PR}}{\lvert\mathbf{PQ}\times\mathbf{PR}\rvert} \quad &\text{(normal to the plane $PQR$; non-zero because the points are not collinear)} \\
\mathbf j_B &= \mathbf k_B\times\mathbf i_B \quad &\text{(completes the triad; $\mathbf k_B \perp \mathbf i_B$, so $\lvert\mathbf j_B\rvert = 1$)} \\
\mathbf i_B\times\mathbf j_B &= \mathbf i_B\times(\mathbf k_B\times\mathbf i_B) = (\mathbf i_B\cdot\mathbf i_B)\mathbf k_B - (\mathbf i_B\cdot\mathbf k_B)\mathbf i_B = \mathbf k_B \quad &\text{(double vector product, Eq. 1.37: the basis is right-handed)}
\end{aligned}$$
Units: each vector is a ratio $[\mathrm L]/[\mathrm L]$ or $[\mathrm{L^2}]/[\mathrm{L^2}]$: dimensionless, as unit vectors must be
Meaning: the three vectors are built only from body segments, whose lengths and angles never change, so the basis moves with the body and rotates with its angular velocity.
Two observations make the rigid body and its body-fixed frame interchangeable (Notes.pdf, Sec. 9; [Lecture 7, Slide 4]):
$S_B$ keeps its orientation relative to the body, so both rotate with the same angular velocity $\boldsymbol\omega_{B0}$: knowing the attitude of the body and knowing the attitude of $S_B$ are the same thing.
For an observer on $S_B$ every body point is at rest. Every point of the Euclidean space attached to $S_B$ can be treated as a body point, even outside the physical body.
Rigid body and reference frame
From the kinematic point of view a rigid body is fully equivalent to a reference frame: its shape and dimensions can be ignored and it can be replaced by $S_B$. Conversely, any reference frame can be thought of as a rigid body with a Euclidean space attached to it ([Lecture 7, Slide 4]).
2. Velocity and acceleration fields of a rigid body
Let $P$ be the origin of $S_B$ and $A$ any other body point (Fig. 9.1 of the notes). The composition laws of relative motion, Eqs. 3.26 and 3.27, are applied with the moving frame $S_B$; the relative velocity and acceleration of a body point in $S_B$ are zero by the definition of body point ([Lecture 7, Slide 5]).
Velocity of a body point from the composition of velocitiesLaw: composition of velocities with the moving frame $S_B$; $A$ is at rest in $S_B$Source: Notes.pdf, Eq. 9.1; Notes.pdf, Eq. 3.26; [Lecture 7, Slide 6]
$$\begin{aligned}
\mathbf v_0^A &= \mathbf v_0^P + \mathbf v_B^A + \boldsymbol\omega_{B0}\times\mathbf r_B^A \quad &\text{(Eq. 3.26 with the moving frame $S_B$ of origin $P$)} \\
\mathbf v_B^A &= \mathbf 0 \quad &\text{($A$ is a body point: it does not move in $S_B$)} \\
\mathbf v_0^A &= \mathbf v_0^P + \boldsymbol\omega_{B0}\times\mathbf r_B^A \quad &\text{(substitute the zero relative velocity)}
\end{aligned}$$
Meaning: Eq. 9.1 is the general composition law with its relative term crossed out: a body point is dragged by the frame, it never moves relative to it.
Acceleration of a body point from the composition of accelerationsLaw: composition of accelerations with the moving frame $S_B$; $A$ is at rest in $S_B$Source: Notes.pdf, Eq. 9.2; Notes.pdf, Eq. 3.27; [Lecture 7, Slide 6]
$$\begin{aligned}
\mathbf a_0^A &= \mathbf a_B^A + \mathbf a_0^P + \boldsymbol\alpha_{B0}\times\mathbf r_B^A + \boldsymbol\omega_{B0}\times(\boldsymbol\omega_{B0}\times\mathbf r_B^A) + 2\boldsymbol\omega_{B0}\times\mathbf v_B^A \quad &\text{(Eq. 3.27 with the moving frame $S_B$)} \\
\mathbf a_B^A &= \mathbf 0,\quad \mathbf v_B^A = \mathbf 0 \quad &\text{($A$ is at rest in $S_B$ at every instant)} \\
2\boldsymbol\omega_{B0}\times\mathbf v_B^A &= \mathbf 0 \quad &\text{(no relative velocity: no Coriolis term)}
\end{aligned}$$
Meaning: of the five terms of Eq. 3.27 only the drag terms survive: the acceleration of the origin, the Euler term and the centripetal term.
Velocity and acceleration fieldsLaw: position of $A$ in $S_B$ written as the body segment from the originSource: Notes.pdf, Eq. 9.3; Notes.pdf, Eq. 9.4; [Lecture 7, Slide 6]
$$\begin{aligned}
\mathbf r_B^A &= \mathbf{PA} \quad &\text{($P$ is the origin of $S_B$)} \\
\mathbf v_0^A &= \mathbf v_0^P + \boldsymbol\omega_{B0}\times\mathbf{PA} \quad &\text{(substitute in Eq. 9.1)} \\
\mathbf a_0^A &= \mathbf a_0^P + \boldsymbol\alpha_{B0}\times\mathbf{PA} + \boldsymbol\omega_{B0}\times(\boldsymbol\omega_{B0}\times\mathbf{PA}) \quad &\text{(substitute in Eq. 9.2)}
\end{aligned}$$
Units: velocity $[\mathrm{L\,T^{-1}}]$ and acceleration $[\mathrm{L\,T^{-2}}]$ in every term
Meaning: the motion of every body point is the motion of an arbitrarily chosen body point $P$ plus a rotation about $P$; these two formulas are the velocity field and the acceleration field of the rigid body.
To know the whole velocity field at an instant it is enough to know the velocity of one body point and $\boldsymbol\omega_{B0}$ (Notes.pdf, Sec. 9.1.1). The field has the following properties.
Pure translationLaw: velocity field with zero angular velocitySource: Notes.pdf, Eq. 9.3; [Lecture 7, Slide 6]
$$\begin{aligned}
\boldsymbol\omega_{B0} &= \mathbf 0 \quad &\text{(hypothesis)} \\
\mathbf v_0^A &= \mathbf v_0^P + \mathbf 0\times\mathbf{PA} = \mathbf v_0^P \quad &\text{(Eq. 9.3; the vector product with the zero vector vanishes)}
\end{aligned}$$
Meaning: all points have the same velocity: the body moves, at that instant, in pure translation.
All body points share the velocity component along the angular velocityLaw: dot product of the velocity field with $\mathbf e_\omega$; triple product with two parallel vectorsSource: Notes.pdf, Eq. 9.5; [Lecture 7, Slide 6]; Notes.pdf, Eq. 1.33
$$\begin{aligned}
\mathbf e_\omega &= \frac{\boldsymbol\omega_{B0}}{\omega_{B0}} \quad &\text{(unit vector of the angular velocity, $\omega_{B0} \neq 0$)} \\
\mathbf v_0^A\cdot\mathbf e_\omega &= \mathbf v_0^P\cdot\mathbf e_\omega + (\boldsymbol\omega_{B0}\times\mathbf{PA})\cdot\mathbf e_\omega \quad &\text{(dot-multiply Eq. 9.3 by $\mathbf e_\omega$)} \\
(\boldsymbol\omega_{B0}\times\mathbf{PA})\cdot\mathbf e_\omega &= \omega_{B0}\,(\mathbf e_\omega\times\mathbf{PA})\cdot\mathbf e_\omega = 0 \quad &\text{(a vector product is perpendicular to both factors)}
\end{aligned}$$
Meaning: the component of the velocity along the angular velocity, the slip speed, is the same for every body point; when it is zero the motion is a pure rotation.
Parallel and perpendicular parts of the velocityLaw: decomposition of a vector along a unit vector and in the plane normal to itSource: Notes.pdf, Eq. 9.6; [Lecture 7, Slide 6]; Notes.pdf, Eq. 1.57
$$\begin{aligned}
\mathbf v_{0\parallel} &= (\mathbf v_0^A\cdot\mathbf e_\omega)\,\mathbf e_\omega = v_{0\parallel}\,\mathbf e_\omega \quad &\text{(projection on $\mathbf e_\omega$; identical for all points by Eq. 9.5)} \\
\mathbf v_{0\perp}^A &= \mathbf v_0^A - \mathbf v_{0\parallel} \quad &\text{(remainder, in the plane normal to $\boldsymbol\omega_{B0}$)} \\
\mathbf v_{0\perp}^A\cdot\mathbf e_\omega &= v_{0\parallel} - v_{0\parallel} = 0 \quad &\text{(check: the remainder has no component along $\mathbf e_\omega$)}
\end{aligned}$$
Meaning: the slip velocity is common to the whole body; only the perpendicular part changes from point to point.
Equal projections along the segment joining two pointsLaw: dot product of the velocity field with $\mathbf{PA}$Source: Notes.pdf, Eq. 9.7; [Lecture 7, Slide 7]
$$\begin{aligned}
\mathbf v_0^A\cdot\mathbf{PA} &= \mathbf v_0^P\cdot\mathbf{PA} + (\boldsymbol\omega_{B0}\times\mathbf{PA})\cdot\mathbf{PA} \quad &\text{(dot-multiply Eq. 9.3 by $\mathbf{PA}$)} \\
(\boldsymbol\omega_{B0}\times\mathbf{PA})\cdot\mathbf{PA} &= 0 \quad &\text{(the vector product is perpendicular to $\mathbf{PA}$)}
\end{aligned}$$
Units: $[\mathrm{L\,T^{-1}}][\mathrm L] = [\mathrm{L^2\,T^{-1}}]$ on both sides
Meaning: any two body points have the same velocity projection on the line that joins them; otherwise their distance would change and the body would not be rigid.
Figure 2. Equal projections: $\mathbf v_0^A\cdot\mathbf{PA}=\mathbf v_0^P\cdot\mathbf{PA}$. The perpendicular parts of the velocities differ, the parts along the line $AP$ coincide ($p$), so the distance $AP$ stays constant..
The velocities of three non-collinear body points $A$, $C$, $D$ also fix the field, through Eqs. 9.12 and 9.13 ([Lecture 7, Slide 7] writes them with $B$ and $C$). The data must be compatible, or the object is not a rigid body.
Field from three non-collinear points and compatibility of the dataLaw: velocity field between pairs of body pointsSource: Notes.pdf, Eq. 9.12; Notes.pdf, Eq. 9.13; [Lecture 7, Slide 7]
$$\begin{aligned}
\mathbf v_0^C &= \mathbf v_0^A + \boldsymbol\omega_{B0}\times\mathbf{AC} \quad &\text{(Eq. 9.3 from $A$ to $C$: three scalar equations)} \\
\mathbf v_0^D &= \mathbf v_0^A + \boldsymbol\omega_{B0}\times\mathbf{AD} \quad &\text{(Eq. 9.3 from $A$ to $D$: three more scalar equations)} \\
(\mathbf v_0^C-\mathbf v_0^A)\cdot\mathbf{AC} &= 0,\quad (\mathbf v_0^D-\mathbf v_0^A)\cdot\mathbf{AD} = 0,\quad (\mathbf v_0^D-\mathbf v_0^C)\cdot\mathbf{CD} = 0 \quad &\text{(Eq. 9.7 for the three pairs: conditions on the data)}
\end{aligned}$$
Meaning: six scalar equations for the three unknown components of $\boldsymbol\omega_{B0}$; they have a solution only when the nine measured components satisfy the three equal-projection conditions, which is what "compatible data" means.
What fixes the instantaneous velocity field
Completely determining the instantaneous velocity field of a rigid body is equivalent to knowing $\boldsymbol\omega_{B0}$ and the velocity of one body point ([Lecture 7, Slide 7]). Other equivalent data sets (Notes.pdf, Sec. 9.1.3): a point of the instantaneous axis, $\boldsymbol\omega_{B0}$ and the slip speed; the velocities of three non-collinear points; or the velocities of two points plus one extra independent condition.
3. Instantaneous axis of rotation and slip, and helical motion
In a pure rotation (zero slip speed) the instantaneous axis of rotation is the locus of body points with zero velocity; it is the line parallel to $\boldsymbol\omega_{B0}$ through any one such point $I$ (Notes.pdf, Sec. 9.1.2).
Condition for a point of the instantaneous axisLaw: velocity field from the known point $A$; double vector productSource: Notes.pdf, Eq. 9.8; Notes.pdf, Eq. 9.9; Notes.pdf, Eq. 1.37; [Lecture 7, Slide 9]
$$\begin{aligned}
\mathbf v_0^I &= \mathbf v_0^A + \boldsymbol\omega_{B0}\times\mathbf{AI} = \mathbf 0 \quad &\text{(Eq. 9.3 from $A$ to $I$; $I$ has zero velocity)} \\
\boldsymbol\omega_{B0}\times\mathbf v_0^A + \boldsymbol\omega_{B0}\times(\boldsymbol\omega_{B0}\times\mathbf{AI}) &= \mathbf 0 \quad &\text{(cross-multiply on the left by $\boldsymbol\omega_{B0}$)} \\
\boldsymbol\omega_{B0}\times(\boldsymbol\omega_{B0}\times\mathbf{AI}) &= (\boldsymbol\omega_{B0}\cdot\mathbf{AI})\,\boldsymbol\omega_{B0} - (\boldsymbol\omega_{B0}\cdot\boldsymbol\omega_{B0})\,\mathbf{AI} \quad &\text{(Eq. 1.37 with $\mathbf A = \mathbf B = \boldsymbol\omega_{B0}$, $\mathbf C = \mathbf{AI}$)}
\end{aligned}$$
Units: $[\mathrm{T^{-1}}][\mathrm{L\,T^{-1}}] = [\mathrm{T^{-2}}][\mathrm L] = [\mathrm{L\,T^{-2}}]$ in every term
Meaning: Eq. 9.9 is satisfied by every point of the axis; the middle term is what distinguishes one of them from another.
The foot of the perpendicular from A to the axisLaw: choice of $I$ with $\mathbf{AI}$ perpendicular to the angular velocitySource: Notes.pdf, Eq. 9.10; [Lecture 7, Slide 9]
$$\begin{aligned}
\boldsymbol\omega_{B0}\cdot\mathbf{AI} &= 0 \quad &\text{(choose $I$ as the foot of the perpendicular from $A$, Fig. 9.3)} \\
\boldsymbol\omega_{B0}\times\mathbf v_0^A - \omega_{B0}^2\,\mathbf{AI} &= \mathbf 0 \quad &\text{(the middle term of Eq. 9.9 vanishes)} \\
\mathbf{AI} &= \frac{\boldsymbol\omega_{B0}\times\mathbf v_0^A}{\omega_{B0}^2} \quad &\text{(divide by $\omega_{B0}^2 \neq 0$)} \\
\mathbf v_0^{I'} &= \mathbf v_0^I + \boldsymbol\omega_{B0}\times(s\,\boldsymbol\omega_{B0}) = \mathbf v_0^I \quad &\text{(any $I'$ with $\mathbf{II'} = s\,\boldsymbol\omega_{B0}$ has the same velocity: the axis)}
\end{aligned}$$
Meaning: the axis is the line through $I$ parallel to $\boldsymbol\omega_{B0}$; any other choice $I'$ on it keeps the middle term of Eq. 9.9 and makes the computation longer.
Figure 3. The point $I$ is the foot of the perpendicular from $A$ to the instantaneous axis (IAR), so $\mathbf{AI}\perp\boldsymbol\omega_{B0}$; another point $I'$ of the axis makes $\mathbf{AI'}$ longer and the second term of Eq. 9.9 does not vanish..
Remarks of the notes (Notes.pdf, Sec. 9.1.2): the axis can change in time (hence "instantaneous"), and the ruled surfaces it sweeps, seen from $S_0$ and from $S_B$, help to picture the motion; the point $I$ and even the whole axis may lie outside the physical body, but it is still a point of the Euclidean space attached to $S_B$; and the axis depends on the frame from which it is computed.
In the general case the slip speed is not zero. The instantaneous axis of rotation and slip is then the locus of points with minimum velocity: the points whose velocity has no perpendicular part, so that it equals the slip velocity ([Lecture 7, Slide 8]).
Instantaneous axis of rotation and slipLaw: velocity field with the slip velocity imposed; the slip velocity is parallel to the angular velocitySource: Notes.pdf, Eq. 9.11; Notes.pdf, Eq. 9.10; [Lecture 7, Slide 9]
$$\begin{aligned}
\mathbf v_0^I &= \mathbf v_0^A + \boldsymbol\omega_{B0}\times\mathbf{AI} = \mathbf v_{0\parallel} \quad &\text{(the perpendicular part of the velocity of $I$ vanishes)} \\
\boldsymbol\omega_{B0}\times\mathbf v_0^A + \boldsymbol\omega_{B0}\times(\boldsymbol\omega_{B0}\times\mathbf{AI}) &= \boldsymbol\omega_{B0}\times\mathbf v_{0\parallel} = \mathbf 0 \quad &\text{(cross-multiply by $\boldsymbol\omega_{B0}$; $\mathbf v_{0\parallel}$ is parallel to it)} \\
\mathbf{AI} &= \frac{\boldsymbol\omega_{B0}\times\mathbf v_0^A}{\omega_{B0}^2} \quad &\text{(same steps as Eq. 9.10 with $\mathbf{AI}$ perpendicular to $\boldsymbol\omega_{B0}$)}
\end{aligned}$$
Meaning: the same formula locates the axis with or without slip; with zero slip speed it is just the instantaneous axis of rotation.
The general instantaneous motion of a rigid body is therefore a pure rotation about the instantaneous axis plus a pure translation along it with the slip velocity: a screw, or helical, motion (Notes.pdf, Sec. 9.1.3; [Lecture 7, Slide 8]).
Figure 4. Screw (helical) motion, Fig. 9.4. Left: in a plane normal to $\boldsymbol\omega_{B0}$ (out of the paper) the perpendicular velocities $\mathbf v_{0\perp}=\boldsymbol\omega_{B0}\times\mathbf{IA}$ are tangent to circles about the axis and grow with $d$. Right: on top of them all points share the slip velocity $\mathbf v_{0\parallel}$..
The screw (helical) velocity fieldLaw: velocity field from a point $I$ of the axis; magnitude of a vector product of perpendicular vectorsSource: Notes.pdf, Sec. 9.1.3; Notes.pdf, Eq. 9.6; [Lecture 7, Slide 8]
$$\begin{aligned}
\mathbf v_0^A &= \mathbf v_0^I + \boldsymbol\omega_{B0}\times\mathbf{IA} \quad &\text{(Eq. 9.3 from $I$ to $A$)} \\
&= \mathbf v_{0\parallel} + \boldsymbol\omega_{B0}\times\mathbf{IA} \quad &\text{(Eq. 9.11)} \\
\mathbf v_{0\perp}^A &= \boldsymbol\omega_{B0}\times\mathbf{IA} \quad &\text{(compare with Eq. 9.6)} \\
\lvert\mathbf v_{0\perp}^A\rvert &= \omega_{B0}\,d_A \quad &\text{(choose $I$ as the foot of the perpendicular from $A$: $\mathbf{IA} \perp \boldsymbol\omega_{B0}$, $d_A = \lvert\mathbf{IA}\rvert$)}
\end{aligned}$$
Meaning: in each plane normal to the angular velocity the perpendicular velocities are tangent to circles centred on the axis and grow linearly with the distance to it (Fig. 9.4); on top of them every point has the same slip velocity.
Extension: the helix of a steady screw motion
The notes and slides call the general motion a screw. When the axis is fixed and the angular velocity and slip speed are constant (an assumption added here), each body point describes a helix about the axis; the box below shows it. This is an extension, not course content.
A body point in steady screw motion describes a helixLaw: position about a fixed axis; time derivative with the chain ruleSource: Notes.pdf, Sec. 9.1.3 (extension)
Take the axis as the line through $I$ along $\mathbf e_\omega$, and two fixed unit vectors $\mathbf e_1$, $\mathbf e_2$ normal to it with $\mathbf e_1\times\mathbf e_2 = \mathbf e_\omega$; $d$ is the distance of the point to the axis.
Three simultaneous measurements are made of the motion of a rigid airplane: the centre of mass $G$, a point $A$ on the wing and a point $C$ on the vertical tail, with the positions and velocities of the table below (basis $\mathcal B_0$, metres and metres per second). Slide 11 is left blank for the classroom solution; the questions answered here are the ones the deck prepares in Slides 7 and 9: the data are checked for compatibility, the angular velocity of the airplane is found, and then the slip velocity and the instantaneous axis of rotation and slip.
Figure 5. Rigid airplane of Lecture 7 (sketch, not to scale): velocities of $G$, $A$, $C$, the angular velocity $\boldsymbol\omega_{B0}=\mathbf i+\mathbf j+\mathbf k$ and the instantaneous axis of rotation and slip through $I$, where the velocity is the slip velocity $\mathbf v_{0\parallel}=2\mathbf i+2\mathbf j+2\mathbf k$..
Compatibility: equal projections for the three pairsLaw: equal-projection property of the rigid velocity fieldSource: Notes.pdf, Eq. 9.7; [Lecture 7, Slide 7]; [Lecture 7, Slide 10]
Units: $[\mathrm{L\,T^{-1}}][\mathrm L]$: $0$, $-16$ and $-5\ \mathrm{m^2/s}$
Meaning: the three measurements are compatible with a rigid body, so an angular velocity that reproduces them exists.
Velocity field from G to ALaw: velocity field of the rigid body; vector product in componentsSource: Notes.pdf, Eq. 9.3; Notes.pdf, Eq. 1.42; [Lecture 7, Slide 7]
Units: $[\mathrm{T^{-1}}][\mathrm L] = [\mathrm{L\,T^{-1}}]$, so each component of the angular velocity is in $\mathrm{rad/s}$
Meaning: one pair of points fixes the angular velocity only up to its component along $\mathbf{GA}$: the first two equations repeat each other, which is the compatibility of the pair.
Velocity field from G to CLaw: velocity field of the rigid body; vector product in componentsSource: Notes.pdf, Eq. 9.12; Notes.pdf, Eq. 1.42; [Lecture 7, Slide 7]
$$\begin{aligned}
\mathbf v_0^C - \mathbf v_0^G &= (5-4,\ -3-2,\ 4-0) = (1,\,-5,\,4)\ \mathrm{m/s} \quad &\text{(left-hand side)} \\
\boldsymbol\omega_{B0}\times\mathbf{GC} &= \begin{vmatrix} \mathbf i & \mathbf j & \mathbf k \cr \omega_x & \omega_y & \omega_z \cr -4 & 0 & 1 \end{vmatrix} \quad &\text{(Eq. 1.42)} \\
&= (\omega_y - 0)\,\mathbf i + (-4\omega_z - \omega_x)\,\mathbf j + (0 + 4\omega_y)\,\mathbf k \quad &\text{(expand along the first row)} \\
\omega_y &= 1,\qquad -4\omega_z - \omega_x = -5,\qquad 4\omega_y = 4 \quad &\text{(equate components)} \\
\omega_x &= 5 - 4\omega_z = 5 - 4 = 1 \quad &\text{(use $\omega_z = 1$ from the pair $G$, $A$)} \\
2\omega_x + \omega_y &= 2 + 1 = 3 \quad &\text{(check of the third equation of the pair $G$, $A$)}
\end{aligned}$$
Units: $\mathrm{rad/s}$, that is $[\mathrm{T^{-1}}]$
Meaning: the second pair removes the indeterminacy of the first, and every one of the six scalar equations is satisfied: the data are consistent.
Slip velocity of the airplaneLaw: common component of the velocity along the angular velocitySource: Notes.pdf, Eq. 9.5; Notes.pdf, Eq. 9.6; [Lecture 7, Slide 6]
Meaning: the slip speed is not zero, so the instantaneous motion of the airplane is a genuine screw motion, not a pure rotation.
Instantaneous axis of rotation and slip of the airplaneLaw: foot of the perpendicular from $G$ to the axisSource: Notes.pdf, Eq. 9.10; Notes.pdf, Eq. 9.11; [Lecture 7, Slide 9]
$$\begin{aligned}
\boldsymbol\omega_{B0}\times\mathbf v_0^G &= \begin{vmatrix} \mathbf i & \mathbf j & \mathbf k \cr 1 & 1 & 1 \cr 4 & 2 & 0 \end{vmatrix} = (0-2)\,\mathbf i + (4-0)\,\mathbf j + (2-4)\,\mathbf k \quad &\text{(Eq. 1.42)} \\
\mathbf{GI} &= \frac{-2\mathbf i + 4\mathbf j - 2\mathbf k}{3} \quad &\text{(Eq. 9.10 with $A$ replaced by $G$, $\omega_{B0}^2 = 3$)} \\
\boldsymbol\omega_{B0}\times\mathbf{GI} &= \tfrac13\begin{vmatrix} \mathbf i & \mathbf j & \mathbf k \cr 1 & 1 & 1 \cr -2 & 4 & -2 \end{vmatrix} = \tfrac13\left[(-2-4)\,\mathbf i + (-2+2)\,\mathbf j + (4+2)\,\mathbf k\right] \quad &\text{(check, Eq. 1.42)} \\
\mathbf v_0^I &= (4\mathbf i + 2\mathbf j) + (-2\mathbf i + 2\mathbf k) = 2\mathbf i + 2\mathbf j + 2\mathbf k \quad &\text{(Eq. 9.3 from $G$ to $I$: equals the slip velocity)}
\end{aligned}$$
Meaning: the airplane screws about a line through $I$, about $1.6\ \mathrm m$ from $G$, parallel to $\mathbf i+\mathbf j+\mathbf k$; along it every point moves only with the slip velocity, the minimum speed in the body.
5. Points of a rigid body
Equations 9.3 and 9.4 hold only when both points are body points: points that move with the body and have zero velocity and acceleration in $S_B$ ([Lecture 7, Slide 5]; Notes.pdf, Sec. 9.1.4). Often a body point is fixed by a geometric condition (the centre of a rectangular plate). Sometimes it must be defined as the body point that instantaneously coincides with a geometric point $A$ that does not belong to the body; the notes write it $A(B)$ and the slides $A(b)$. Both have the same position, but not the same velocity or acceleration.
A geometric point and the body point on top of itLaw: composition of velocities for $A$ (not a body point) and velocity field for $A(B)$ (a body point)Source: Notes.pdf, Eq. 9.14; Notes.pdf, Eq. 9.15; Notes.pdf, Eq. 3.26; [Lecture 7, Slide 5]
$$\begin{aligned}
\mathbf r_0^A &= \mathbf r_0^{A(B)} \quad &\text{(by definition $A(B)$ lies on $A$ at this instant)} \\
\mathbf v_0^A &= \mathbf v_0^P + \mathbf v_B^A + \boldsymbol\omega_{B0}\times\mathbf{PA} \quad &\text{(Eq. 3.26 with moving frame $S_B$; $A$ may move in $S_B$)} \\
\mathbf v_0^{A(B)} &= \mathbf v_0^P + \boldsymbol\omega_{B0}\times\mathbf{PA} \quad &\text{(Eq. 9.3: $A(B)$ is a body point, $\mathbf v_B^{A(B)} = \mathbf 0$)} \\
\mathbf v_0^A - \mathbf v_0^{A(B)} &= \mathbf v_B^A \quad &\text{(subtract the two lines)}
\end{aligned}$$
Meaning: plugging the velocity of a non-body point into the field equation is wrong: the difference is exactly the velocity of $A$ relative to the body; the same holds, mutatis mutandis, for the accelerations.
The body point that lies on $A$ changes as the body moves, so $A(B)$ names a different material point at each instant. The notation is used only where ambiguity is possible; in every problem each point must be declared as a body point or not. Points outside the physical body that move rigidly with $S_B$ are also body points (Notes.pdf, Sec. 9.1.4).
6. Orthogonal rotation matrices
Since a rigid body is replaced by $S_B$, its attitude is the attitude of the basis $\mathcal B_B$ with respect to $\mathcal B_0$, described by a change of basis matrix (Notes.pdf, Sec. 1.5). This section collects the properties of those matrices that the Euler angles use. The notes state the orthogonality and the unit determinant without proof (Eqs. 1.14 and 1.15); the two boxes below prove them from the orthonormality and the handedness of the bases.
Change of basis matrixLaw: components of one basis in the other; uniqueness of componentsSource: Notes.pdf, Eq. 1.8; Notes.pdf, Eq. 1.12; Notes.pdf, Eq. 1.13
$$\begin{aligned}
\mathbf i_1 &= \ell_{x1,x0}\,\mathbf i_0 + \ell_{x1,y0}\,\mathbf j_0 + \ell_{x1,z0}\,\mathbf k_0,\ \ \text{and likewise for } \mathbf j_1,\ \mathbf k_1 \quad &\text{(change of basis, Eq. 1.8)} \\
\mathbf b &= b_{x1}\mathbf i_1 + b_{y1}\mathbf j_1 + b_{z1}\mathbf k_1 \quad &\text{(the same vector written in $\mathcal B_1$, Eq. 1.6)} \\
b_{x0} &= \ell_{x1,x0}\,b_{x1} + \ell_{y1,x0}\,b_{y1} + \ell_{z1,x0}\,b_{z1},\ \ \text{and likewise for } b_{y0},\ b_{z0} \quad &\text{(substitute Eq. 1.8 and collect the coefficient of $\mathbf i_0$, Eqs. 1.9-1.11)}
\end{aligned}$$
Units: the entries are direction cosines, dimensionless $[1]$
Meaning: column $n$ of the matrix is the $n$-th vector of the new basis written in the old one; the vector itself is not moved, only described from another basis.
The change of basis matrix is orthogonalLaw: orthonormality of $\mathcal B_1$; dot product in componentsSource: Notes.pdf, Eq. 1.14; Notes.pdf, Eq. 1.40
$$\begin{aligned}
\big([{}_0R_1]^T[{}_0R_1]\big)_{mn} &= [\mathbf u_m]_0^T\,[\mathbf u_n]_0 \quad &\text{(row $m$ of the transpose is column $m$; $\mathbf u_1,\mathbf u_2,\mathbf u_3 = \mathbf i_1,\mathbf j_1,\mathbf k_1$)} \\
&= \mathbf u_m\cdot\mathbf u_n \quad &\text{(dot product in components, Eq. 1.40)} \\
&= \delta_{mn} \quad &\text{($\mathcal B_1$ is orthonormal)} \\
[{}_1R_0] &= [{}_0R_1]^{-1} \quad &\text{(inverse operation, by definition)}
\end{aligned}$$
Meaning: inverting a change of basis costs nothing: transpose the matrix; so the rows of the matrix are the old basis vectors written in the new basis.
The determinant of a rotation matrix is oneLaw: determinant as a triple product; right-handed basisSource: Notes.pdf, Eq. 1.15; Notes.pdf, Eq. 1.43
$$\begin{aligned}
\det[{}_0R_1] &= \det [{}_0R_1]^T = [\mathbf i_1]_0\cdot\big([\mathbf j_1]_0\times[\mathbf k_1]_0\big) \quad &\text{(rows of the transpose are $\mathbf i_1,\mathbf j_1,\mathbf k_1$; Eq. 1.43)} \\
\mathbf j_1\times\mathbf k_1 &= \mathbf i_1 \quad &\text{($\mathcal B_1$ is right-handed)} \\
\det[{}_0R_1] &= \mathbf i_1\cdot\mathbf i_1 = 1 \quad &\text{($\mathbf i_1$ is a unit vector)}
\end{aligned}$$
$$\boxed{\det[{}_0R_1] = 1}$$
Units: dimensionless
Meaning: the matrix maps a right-handed basis onto a right-handed basis; a left-handed one would give minus one. With the orthogonality it means $[{}_0R_1]$ is a rotation: it has the eigenvalue one, whose eigenvector is the axis of the equivalent single rotation (Notes.pdf, Sec. 1.5.1).
Composition of changes of basisLaw: chaining two changes of basisSource: Notes.pdf, Eq. 1.16; Notes.pdf, Eq. 1.23
$$\begin{aligned}
[\mathbf b]_0 &= [{}_0R_1]\,[\mathbf b]_1 \quad &\text{(from $\mathcal B_1$ to $\mathcal B_0$)} \\
[\mathbf b]_1 &= [{}_1R_2]\,[\mathbf b]_2 \quad &\text{(from $\mathcal B_2$ to $\mathcal B_1$)} \\
[\mathbf b]_0 &= [{}_0R_1]\,[{}_1R_2]\,[\mathbf b]_2 \quad &\text{(substitute the second line into the first)} \\
\det\big([{}_0R_1][{}_1R_2]\big) &= 1\cdot 1 = 1,\quad \big([{}_0R_1][{}_1R_2]\big)^T[{}_0R_1][{}_1R_2] = [{}_1R_2]^T[U][{}_1R_2] = [U] \quad &\text{(the product is again a rotation matrix)}
\end{aligned}$$
$$\boxed{[{}_0R_2] = [{}_0R_1]\,[{}_1R_2]}$$
Units: dimensionless
Meaning: a sequence of rotations is a product of matrices written from the first to the last; the product does not commute, so the order of the rotations matters.
7. Euler angles of type 3-1-3
The attitude of $S_B$ with respect to $S_0$ is given by three angles; when they are three successive simple rotations they are called Euler angles (Notes.pdf, Sec. 9.2; [Lecture 7, Slide 12]). The standard set is the 3-1-3 sequence of Fig. 9.5: precession about the third axis, nutation about the new first axis, spin about the new third axis. Each simple rotation is drawn looking down its axis, which then points out of the paper, as recommended in Notes.pdf, Sec. 1.5.2.
Figure 6. Precession $\psi$ about $\mathbf k_0=\mathbf k_1$ (out of the paper $\odot$): $\mathbf i_1$ and $\mathbf j_1$ in $\mathcal B_0$ from the shaded right triangles (hypotenuse $1$)..
Precession: simple rotation about the third axisLaw: simple rotation by $\psi$ about $\mathbf k_0$, positive by the right-hand ruleSource: Notes.pdf, Eq. 9.16; Notes.pdf, Eq. 1.20; [Lecture 7, Slide 12]
View with $\mathbf k_0$ out of the paper (⊙), $\mathbf i_0$ to the right and $\mathbf j_0$ up. The positive sense of $\psi$ is counterclockwise in this view (thumb along $\mathbf k_0$, fingers curl from $\mathbf i_0$ towards $\mathbf j_0$).
$$\begin{aligned}
\mathbf i_1 &= \cos\psi\,\mathbf i_0 + \sin\psi\,\mathbf j_0 \quad &\text{(right triangle with hypotenuse $\lvert\mathbf i_1\rvert = 1$: adjacent leg along $\mathbf i_0$, opposite leg along $\mathbf j_0$)} \\
\mathbf j_1 &= -\sin\psi\,\mathbf i_0 + \cos\psi\,\mathbf j_0 \quad &\text{($\mathbf j_0$ turned counterclockwise leans towards $-\mathbf i_0$: opposite leg $\sin\psi$ along $-\mathbf i_0$)} \\
\mathbf k_1 &= \mathbf k_0 \quad &\text{(the rotation axis does not move)} \\
\det[{}_0R_1] &= \cos^2\psi + \sin^2\psi = 1 \quad &\text{(expand along the third row)} \\
[{}_0R_1][{}_0R_1]^T &= \begin{bmatrix} \cos^2\psi+\sin^2\psi & \cos\psi\sin\psi-\sin\psi\cos\psi & 0 \cr \sin\psi\cos\psi-\cos\psi\sin\psi & \sin^2\psi+\cos^2\psi & 0 \cr 0 & 0 & 1\end{bmatrix} = [U] \quad &\text{(row by row products)}
\end{aligned}$$
Meaning: the columns are $\mathbf i_1$, $\mathbf j_1$, $\mathbf k_1$ in $\mathcal B_0$; the only minus sign sits where the rotated second axis acquires a component along the negative first axis.
Figure 7. Nutation $\theta$ about $\mathbf i_1=\mathbf i_2$ (out of the paper $\odot$, line of nodes): $\mathbf j_2$ and $\mathbf k_2$ in $\mathcal B_1$ from the shaded right triangles..
Nutation: simple rotation about the new first axisLaw: simple rotation by $\theta$ about $\mathbf i_1$, positive by the right-hand ruleSource: Notes.pdf, Eq. 9.17; Notes.pdf, Eq. 1.18; [Lecture 7, Slide 12]
View with $\mathbf i_1$ out of the paper (⊙), $\mathbf j_1$ to the right and $\mathbf k_1$ up (since $\mathbf j_1\times\mathbf k_1 = \mathbf i_1$). Positive $\theta$ is counterclockwise.
$$\begin{aligned}
\mathbf i_2 &= \mathbf i_1 \quad &\text{(the rotation axis does not move)} \\
\mathbf j_2 &= \cos\theta\,\mathbf j_1 + \sin\theta\,\mathbf k_1 \quad &\text{(hypotenuse $1$; adjacent leg along $\mathbf j_1$, opposite leg along $\mathbf k_1$)} \\
\mathbf k_2 &= -\sin\theta\,\mathbf j_1 + \cos\theta\,\mathbf k_1 \quad &\text{($\mathbf k_1$ turned counterclockwise leans towards $-\mathbf j_1$)} \\
\det[{}_1R_2] &= \cos^2\theta + \sin^2\theta = 1 \quad &\text{(expand along the first row)} \\
\mathbf j_2\cdot\mathbf k_2 &= -\cos\theta\sin\theta + \sin\theta\cos\theta = 0 \quad &\text{(columns orthogonal; each has unit norm)}
\end{aligned}$$
Meaning: the nutation tilts the third axis away from $\mathbf k_0$ by the angle $\theta$ about the line of nodes $\mathbf i_1$.
Figure 8. Spin $\varphi$ about $\mathbf k_2=\mathbf k_B$ (out of the paper $\odot$): $\mathbf i_B$ and $\mathbf j_B$ in $\mathcal B_2$ from the shaded right triangles..
Spin: simple rotation about the new third axisLaw: simple rotation by $\varphi$ about $\mathbf k_2$, positive by the right-hand ruleSource: Notes.pdf, Eq. 9.18; Notes.pdf, Eq. 1.20; [Lecture 7, Slide 12]
View with $\mathbf k_2$ out of the paper (⊙), $\mathbf i_2$ to the right and $\mathbf j_2$ up. Positive $\varphi$ is counterclockwise.
$$\begin{aligned}
\mathbf i_B &= \cos\varphi\,\mathbf i_2 + \sin\varphi\,\mathbf j_2 \quad &\text{(hypotenuse $1$; adjacent leg along $\mathbf i_2$, opposite leg along $\mathbf j_2$)} \\
\mathbf j_B &= -\sin\varphi\,\mathbf i_2 + \cos\varphi\,\mathbf j_2 \quad &\text{($\mathbf j_2$ turned counterclockwise leans towards $-\mathbf i_2$)} \\
\mathbf k_B &= \mathbf k_2 \quad &\text{(the rotation axis does not move)} \\
\det[{}_2R_B] &= \cos^2\varphi + \sin^2\varphi = 1 \quad &\text{(expand along the third row)}
\end{aligned}$$
Meaning: the spin turns the body about its own symmetry-like axis $\mathbf k_B$; the matrix has the same form as the precession one.
Composed matrix of the 3-1-3 sequenceLaw: composition of changes of basis; matrix product row by columnSource: Notes.pdf, Eq. 9.19; Notes.pdf, Eq. 1.16
Abbreviations: $c_\psi$ and $s_\psi$ stand for the cosine and sine of $\psi$, and likewise for $\theta$ and $\varphi$.
Meaning: the third column gives the body axis in the fixed basis, with $\mathbf k_B\cdot\mathbf k_0 = \cos\theta$: the nutation is the angle between the two third axes.
Inverse of the composed matrixLaw: orthogonality; transpose of a productSource: Notes.pdf, Eq. 9.20; Notes.pdf, Eq. 1.14
$$\begin{aligned}
[{}_BR_0] &= [{}_0R_B]^{-1} = [{}_0R_B]^T \quad &\text{(Eq. 1.14)} \\
&= \big([{}_0R_1][{}_1R_2][{}_2R_B]\big)^T \quad &\text{(Eq. 9.19)} \\
&= [{}_2R_B]^T[{}_1R_2]^T[{}_0R_1]^T \quad &\text{(the transpose of a product reverses the order)}
\end{aligned}$$
Meaning: to pass components from $\mathcal B_0$ to $\mathcal B_B$ undo the rotations in reverse order, each by its transpose (the same matrix with the opposite angle).
The angular velocity follows from the additivity of angular velocities, Eq. 3.14 (Notes.pdf, Sec. 9.2). Each simple rotation contributes the rate of its angle along its own axis; the first box shows why, for the precession.
Angular velocity of a simple rotationLaw: time derivative of the basis vectors with the chain rule; Poisson formulaSource: Notes.pdf, Eq. 9.16; Notes.pdf, Eq. 3.3; Notes.pdf, Eq. 1.63
Meaning: the three Poisson formulas $d\mathbf u/dt = \boldsymbol\omega_{10}\times\mathbf u$ hold with this vector; in the same way the nutation gives $\dot\theta\,\mathbf i_1$ and the spin $\dot\varphi\,\mathbf k_B$.
Angular velocity of the body from the Euler angle ratesLaw: additivity of angular velocitiesSource: Notes.pdf, Eq. 9.21; Notes.pdf, Eq. 3.14
$$\begin{aligned}
\boldsymbol\omega_{B0} &= \boldsymbol\omega_{B2} + \boldsymbol\omega_{20} \quad &\text{(Eq. 3.14 with the intermediate frame $S_2$)} \\
\boldsymbol\omega_{20} &= \boldsymbol\omega_{21} + \boldsymbol\omega_{10} \quad &\text{(Eq. 3.14 with the intermediate frame $S_1$)} \\
\boldsymbol\omega_{10} &= \dot\psi\,\mathbf k_0,\quad \boldsymbol\omega_{21} = \dot\theta\,\mathbf i_1,\quad \boldsymbol\omega_{B2} = \dot\varphi\,\mathbf k_B \quad &\text{(simple rotations about their own axes)}
\end{aligned}$$
Meaning: the angular velocity is the vector sum of the three rates, each along a different axis of a different basis; before computing with it, all terms must be written in one basis.
Body components of the angular velocityLaw: inverse changes of basis of Eqs. 9.17 and 9.18Source: Notes.pdf, Eq. 9.22; Notes.pdf, Eq. 9.23; Notes.pdf, Eq. 9.24
Meaning: these are the components a gyroscope fixed on the body would measure; for zero nutation the first two vanish and the third is the sum of precession and spin rates.
Angular acceleration from the body componentsLaw: the angular acceleration is the same derivative in $S_0$ and in $S_B$; product and chain rulesSource: Notes.pdf, Eq. 3.9; Notes.pdf, Sec. 9.2
Meaning: only for the angular velocity of $S_B$ itself may the body components be differentiated directly; in any other moving frame the Coriolis theorem adds a term.
8. Other Euler sequences, singularities and equivalent sets
Twelve sequences are possible, with the only rule that two successive rotations are never about the same axis (Notes.pdf, Sec. 9.2.1):
First axis 1
First axis 2
First axis 3
1-2-1, 1-2-3, 1-3-1, 1-3-2
2-1-2, 2-1-3, 2-3-1, 2-3-2
3-1-2, 3-1-3, 3-2-1, 3-2-3
Use
Sequence
First angle
Second angle
Third angle
Standard (notes)
3-1-3
precession $\psi$
nutation $\theta$
spin $\varphi$
Orbit of a satellite in an Earth-centred inertial equatorial frame
3-1-3
right ascension of the ascending node $\Omega$
inclination $i$
argument of the pericentre $\omega$
Flight Mechanics
3-2-1
yaw $\psi$ (guiñada)
pitch $\theta$ (cabeceo)
roll $\varphi$ (alabeo)
The notes only name the 3-2-1 sequence. Its matrices follow directly from the simple rotations of Eqs. 1.18 to 1.20, as below: the yaw is the same rotation as the precession (Eq. 9.16, same view), the pitch is about the new second axis and the roll about the new first axis.
Figure 9. Pitch $\theta$ about $\mathbf j_1=\mathbf j_2$ (out of the paper $\odot$): $\mathbf k_1$ to the right, $\mathbf i_1$ up, since $\mathbf k_1\times\mathbf i_1=\mathbf j_1$; the shaded triangles give $\mathbf k_2$ and $\mathbf i_2$..
Pitch: simple rotation about the new second axisLaw: simple rotation by $\theta$ about $\mathbf j_1$, positive by the right-hand ruleSource: Notes.pdf, Eq. 1.19; Notes.pdf, Sec. 9.2.1
View with $\mathbf j_1$ out of the paper (⊙), $\mathbf k_1$ to the right and $\mathbf i_1$ up (since $\mathbf k_1\times\mathbf i_1 = \mathbf j_1$). Positive $\theta$ is counterclockwise, turning $\mathbf k_1$ towards $\mathbf i_1$.
$$\begin{aligned}
\mathbf k_2 &= \cos\theta\,\mathbf k_1 + \sin\theta\,\mathbf i_1 \quad &\text{(hypotenuse $1$; adjacent leg along $\mathbf k_1$, opposite leg along $\mathbf i_1$)} \\
\mathbf i_2 &= \cos\theta\,\mathbf i_1 - \sin\theta\,\mathbf k_1 \quad &\text{($\mathbf i_1$, pointing up, turned counterclockwise leans towards $-\mathbf k_1$)} \\
\mathbf j_2 &= \mathbf j_1 \quad &\text{(the rotation axis does not move)} \\
\det[{}_1R_2] &= \cos\theta\cos\theta - \sin\theta(-\sin\theta) = 1 \quad &\text{(expand along the second row)}
\end{aligned}$$
Meaning: for a rotation about the second axis the minus sign moves below the diagonal, because the cyclic order is $z \to x$ (the pair $\mathbf k_1$, $\mathbf i_1$ is the right-handed pair in this view); a positive pitch turns the first axis towards the negative third axis, which is nose-up in aircraft axes, where the third axis points down.
Figure 10. Roll $\varphi$ about $\mathbf i_2=\mathbf i_B$ (out of the paper $\odot$): $\mathbf j_B$ and $\mathbf k_B$ in $\mathcal B_2$ from the shaded triangles (same as the nutation)..
Roll: simple rotation about the new first axis and composed 3-2-1 matrixLaw: simple rotation by $\varphi$ about $\mathbf i_2$; composition of changes of basisSource: Notes.pdf, Eq. 1.18; Notes.pdf, Eq. 1.16; Notes.pdf, Sec. 9.2.1
View with $\mathbf i_2$ out of the paper (⊙), $\mathbf j_2$ to the right and $\mathbf k_2$ up. With the abbreviations of the 3-1-3 box:
Meaning: the first column is the body axis $\mathbf i_B$ (the nose) in $\mathcal B_0$: its heading is the yaw and its elevation out of the $x_0 y_0$ plane is fixed by the pitch alone.
Extension: body components of the angular velocity for yaw, pitch and roll
The notes give the body components only for the 3-1-3 sequence (Eqs. 9.22 to 9.24). The same steps applied to the 3-2-1 sequence give the box below; it is an extension.
Angular velocity in body axes for the 3-2-1 sequenceLaw: additivity of angular velocities; inverse changes of basisSource: Notes.pdf, Eq. 3.14; Notes.pdf, Eq. 9.21 (extension to the 3-2-1 sequence)
Meaning: with all angles small the body rates reduce to the three angle rates; at a pitch of a right angle the yaw and roll rates enter only through their difference, which is the singularity below.
Every choice of Euler angles has singular attitudes: when the second rotation aligns the first and third rotation axes, the first and third angles are not defined separately, only their sum or difference (Notes.pdf, Sec. 9.2.2; [Lecture 7, Slide 12]). The definition is chosen so that the singularity is rarely met (an aircraft is seldom completely nose-up or nose-down); it is the "gimbal lock" of three-gimbal systems. Quaternions avoid it but are beyond the course.
Singularity of the 3-1-3 anglesLaw: composed matrix at zero nutation; addition formulasSource: Notes.pdf, Sec. 9.2.2; Notes.pdf, Eq. 9.19; [Lecture 7, Slide 12]
$$\boxed{\theta = 0:\ [{}_0R_B] = [{}_0R_1(z,\psi+\varphi)],\qquad \theta = \pi:\ [{}_0R_B] \text{ depends only on } \psi-\varphi}$$
Units: dimensionless
Meaning: with the nutation at zero or a half turn the first and third axes coincide, so precession and spin rotate about the same line and cannot be separated.
Singularity of the yaw, pitch and roll anglesLaw: composed 3-2-1 matrix at a pitch of a right angle; difference formulasSource: Notes.pdf, Sec. 9.2.2
$$\begin{aligned}
c_\theta &= 0,\qquad s_\theta = 1 \quad &\text{(pitch of a right angle)} \\
-s_\psi c_\varphi + c_\psi s_\varphi &= \sin(\varphi-\psi),\qquad s_\psi s_\varphi + c_\psi c_\varphi = \cos(\varphi-\psi) \quad &\text{(sine and cosine of a difference)} \\
[{}_0R_B] &= \begin{bmatrix} 0 & \sin(\varphi-\psi) & \cos(\varphi-\psi) \cr 0 & \cos(\varphi-\psi) & -\sin(\varphi-\psi) \cr -1 & 0 & 0 \end{bmatrix} \quad &\text{(substitute in the composed 3-2-1 matrix)}
\end{aligned}$$
$$\boxed{\theta = \tfrac{\pi}{2}:\ [{}_0R_B] \text{ depends only on } \varphi - \psi}$$
Units: dimensionless
Meaning: nose straight up (or down, for a pitch of three right angles): yaw and roll turn about the same vertical line and only their difference is defined.
Any attitude also has several equivalent sets of Euler angles ([Lecture 7, Slide 12]). The slide illustrates this for the 3-1-3 sequence with a pair of sets; the second set of that pair, as printed, does not give the same attitude.
Erratum in the source
[Lecture 7, Slide 12] gives the sets with precession $\pi/2$, nutation $\pi/2$, spin $0$ and with precession $-\pi/2$, nutation $\pi/2$, spin $\pi$ as the same attitude. The box below shows they are different attitudes; the set equivalent to the first is precession $-\pi/2$, nutation $-\pi/2$, spin $\pi$ (the general rule adds a half turn to the first and third angles and changes the sign of the nutation). The note keeps the corrected set.
Check of the equivalent 3-1-3 sets of Slide 12Law: third and first columns of the composed 3-1-3 matrixSource: Notes.pdf, Eq. 9.19; [Lecture 7, Slide 12]
Meaning: two bases with the same first and third vectors have the same second one ($\mathbf j_B = \mathbf k_B\times\mathbf i_B$), so the corrected sets describe one attitude; the printed second set turns the body axis to the opposite side.
9. Degrees of freedom of a rigid body
The motion of a rigid body is the translation of one body point plus a rotation about it (Eq. 9.3). Without constraints the point needs three coordinates (commonly the centre of mass $G$, defined in Section 11) and the attitude three more, such as the Euler angles (Notes.pdf, Sec. 9.3; [Lecture 7, Slide 13]).
Six degrees of freedom of a free rigid bodyLaw: counting coordinates minus independent rigidity constraintsSource: Notes.pdf, Sec. 9.3; [Lecture 7, Slide 3]; [Lecture 7, Slide 13]
$$\begin{aligned}
3N &\quad &\text{(coordinates of $N$ point particles, the body seen as a collection of particles)} \\
9 - 3 &= 6 \quad &\text{(three non-collinear points: nine coordinates, three fixed mutual distances)} \\
3 - 3 &= 0 \quad &\text{(each further point: three coordinates fixed by its three distances to the first three points)} \\
3 + 3 &= 6 \quad &\text{(equivalently: position of one point plus three attitude angles)}
\end{aligned}$$
$$\boxed{\text{free rigid body: } 6 \text{ degrees of freedom}}$$
Units: a count, dimensionless
Meaning: the rigidity constraint reduces the $3N$ coordinates of the particles to six; given them the position of every point is known without ambiguity.
Solving the kinematics of a rigid body ([Lecture 7, Slide 13])
A free rigid body is equivalent to a reference frame and has six degrees of freedom. A common choice of coordinates is three for the position of a body point $A$ (later, the centre of mass $G$) and three for the attitude (Euler angles), with $S_B$ defined from them. Solving the kinematics of the body means giving, as functions of the six coordinates and their derivatives, the position, velocity and acceleration of one point, $\mathbf r_0^A$, $\mathbf v_0^A$, $\mathbf a_0^A$; the attitude, $\mathbf i_B$, $\mathbf j_B$, $\mathbf k_B$; and the angular velocity and acceleration, $\boldsymbol\omega_{B0}$, $\boldsymbol\alpha_{B0}$. Other choices may suit a given problem better.
10. Mass distributions, their moments and second-order tensors
The point particle needs a single inertia property, its mass. A rigid body also keeps its spatial mass distribution $\mathcal D$, which governs its rotational dynamics: a metal bar is harder to set spinning about an axis across it than about its own long axis ([Lecture 8, Slide 7]). The inertia properties defined in this part of the note are the total mass, the centre of mass and the moments, products and tensor of inertia; they are built from the zeroth, first and second moments of the mass distribution and are valid for any material system, rigid or not (Notes.pdf, Sec. 10). A superscript naming the body is added when more than one body is present.
Second-order tensors and the tensor product
The tensor of inertia is a second-order tensor, so the lecture opens with the few tensor operations it needs ([Lecture 8, Slides 3-6]); the notes collect them in Chapter 1 (Notes.pdf, Eqs. 1.34 to 1.48).
Object
Meaning
$\mathcal T$
a second-order tensor: a linear map that, dot-multiplied with a vector, gives a vector, and with a vector on each side gives a scalar
$[\mathcal T]_{00}$
its $3\times 3$ matrix of components in the tensor basis $\mathcal B_0\otimes\mathcal B_0$
$\mathbf u\otimes\mathbf v$
tensor product (dyad) of two vectors; the slides sometimes omit the symbol and write $\mathbf u\mathbf v$
$\boldsymbol{\mathcal U}$
unit tensor, identity matrix of components in every orthonormal basis
symmetric tensor
a tensor whose component matrix equals its transpose
Contractions of a second-order tensor in matrix formLaw: components of vectors and tensors in the basis $\mathcal B_0$; tensor productSource: [Lecture 8, Slide 3]; [Lecture 8, Slide 4]; Notes.pdf, Eq. 1.36; Notes.pdf, Eq. 1.48
A vector is a column of three components and a tensor a square matrix of nine, both in $\mathcal B_0$.
$$\boxed{\mathbf a\cdot\mathcal T\cdot\mathbf b = [\mathbf a]_0^{T}[\mathcal T]_{00}[\mathbf b]_0,\qquad \mathbf a\cdot(\mathbf u\otimes\mathbf v)\cdot\mathbf b = (\mathbf a\cdot\mathbf u)(\mathbf v\cdot\mathbf b)}$$
Units: the dimension of a contraction is the product of the dimensions of its factors; for a tensor of inertia in $\mathrm{kg\,m^2}$ and two unit vectors the scalar is in $\mathrm{kg\,m^2}$
Meaning: a dyad is a pair of vectors "waiting" for a vector on its left and on its right; the dot is what tells a contraction from a tensor product, so it is never omitted.
The lecture also states the two facts about symmetric tensors that Section 15 uses: their components change basis with the change of basis matrix on both sides ([Lecture 8, Slide 5]), and there is always an orthonormal basis of eigenvectors in which the matrix is diagonal, the principal directions of the tensor ([Lecture 8, Slide 6]).
Moments of a mass distribution
Symbol
Meaning
$m_i$, $\mathbf r_0^{P_i}$
mass and position of the particle $P_i$ of a discrete system, $i$ from $1$ to $N$
$dm$, $\mathbf r_0^{P}$
mass element of a continuous distribution and position of its integration point $P$
$\lambda$, $\sigma$, $\rho$
linear ($\mathrm{kg/m}$), area ($\mathrm{kg/m^2}$) and volume ($\mathrm{kg/m^3}$) densities
$d\ell$, $dA$, $dV$
length, area and volume elements
homogeneous body
a body whose density is uniform in space
Figure 11. Discrete (left) and continuous (right) mass distributions; the moments of the distribution sum or integrate $\mathbf r_0^P$ weighted by the mass (Notes.pdf Fig. 10.1)..
Zeroth, first and second moments of a mass distributionLaw: definition of the $n$-th moment of a mass distribution about $O$Source: Notes.pdf, Eq. 10.1; Notes.pdf, Eq. 10.2; Notes.pdf, Eq. 10.3; Notes.pdf, Eq. 10.4; [Lecture 8, Slide 8]
Units: $[\mathrm M]$ (a scalar), $[\mathrm{M\,L}]$ (a vector) and $[\mathrm{M\,L^2}]$ (a second-order tensor): $\mathrm{kg}$, $\mathrm{kg\,m}$, $\mathrm{kg\,m^2}$
Meaning: the three moments are a scalar, a vector and a tensor; the next sections show that they are the total mass, the mass times the position of the centre of mass, and the second-moment part of the tensor of inertia. Discrete masses add sums of the form of Eq. 10.1 to every integral.
The hard part of these integrals is writing the element of length, area or volume in terms of the differentials of the coordinates that describe the body, and then the limits of every coordinate. For uncommon coordinates one starts from Cartesian coordinates and multiplies by the Jacobian of the change of variables (Notes.pdf, Sec. 10.1). In the spherical coordinates of the notes the angle $\phi$ is a latitude, measured from the plane $Oxy$, not from the axis $Oz$.
Volume element in cylindrical and spherical coordinatesLaw: change of variables in a triple integral; the Jacobian determinantSource: Notes.pdf, Eq. 10.5; [Lecture 8, Slide 8]; [Lecture 8, Slide 17]
Cylindrical coordinates: $x$, $y$, $z$ written with $R$, $\theta$, $z$.
$$\begin{aligned}
x &= R\cos\theta,\quad y = R\sin\theta,\quad z = z \quad &\text{(definition of the coordinates)} \\
J_{\mathrm{cyl}} &= \begin{vmatrix} \cos\theta & -R\sin\theta & 0 \cr \sin\theta & R\cos\theta & 0 \cr 0 & 0 & 1 \end{vmatrix} \quad &\text{(columns: partial derivatives with respect to $R$, $\theta$, $z$)} \\
&= R\cos^2\theta + R\sin^2\theta = R \quad &\text{(expand along the third row; $\cos^2\theta+\sin^2\theta = 1$)}
\end{aligned}$$
Spherical coordinates with latitude: the projection of $\mathbf r_0^P$ on the plane $Oxy$ is the leg $r\cos\phi$ adjacent to $\phi$ in the right triangle whose hypotenuse is $r$; the opposite leg $r\sin\phi$ is the height $z$.
$$\begin{aligned}
x &= r\cos\phi\cos\theta,\quad y = r\cos\phi\sin\theta,\quad z = r\sin\phi \quad &\text{(project the leg $r\cos\phi$ on the axes $Ox$, $Oy$)} \\
J_{\mathrm{sph}} &= \begin{vmatrix} \cos\phi\cos\theta & -r\cos\phi\sin\theta & -r\sin\phi\cos\theta \cr \cos\phi\sin\theta & r\cos\phi\cos\theta & -r\sin\phi\sin\theta \cr \sin\phi & 0 & r\cos\phi \end{vmatrix} \quad &\text{(columns: derivatives with respect to $r$, $\theta$, $\phi$)} \\
&= \sin\phi\,\big(r^2\sin^2\theta\sin\phi\cos\phi + r^2\cos^2\theta\sin\phi\cos\phi\big) + r\cos\phi\,\big(r\cos^2\phi\cos^2\theta + r\cos^2\phi\sin^2\theta\big) \quad &\text{(expand along the third row)} \\
&= r^2\cos\phi\,\big(\sin^2\phi + \cos^2\phi\big) = r^2\cos\phi \quad &\text{(use $\sin^2\theta+\cos^2\theta = 1$ twice)}
\end{aligned}$$
Units: $[\mathrm L]^3$ in the three forms: $R\,dR\,dz$ carries $[\mathrm L^3]$ with $d\theta$ dimensionless, and $r^2\,dr$ carries $[\mathrm L^3]$ with $d\theta\,d\phi$ dimensionless
Meaning: the factors $R$ and $r^2\cos\phi$ are the stretching of the coordinate box into a true volume; with the latitude convention the factor is $\cos\phi$, positive for $\phi$ between minus and plus a right angle.
two parts of a composite body, their masses and centres of mass
Figure 12. The centre of mass $G$ is the mass-weighted average of the positions $\mathbf r_0^P$ of all the mass elements..
Total mass and position of the centre of massLaw: definitions of the total mass and of the centre of massSource: Notes.pdf, Eq. 10.6; Notes.pdf, Eq. 10.7; Notes.pdf, Eq. 10.8; [Lecture 8, Slide 9]
$$\begin{aligned}
m &= \mathcal M_{O,0}^{\mathcal D} = \int_{\mathcal D} dm \quad &\text{(total mass: zeroth moment, Eq. 10.6)} \\
m\,\mathbf r_0^G &= \mathcal M_{O,1}^{\mathcal D} = \int_{\mathcal D}\mathbf r_0^P\,dm \quad &\text{(G: the point that, holding the whole mass, keeps the first moment, Eq. 10.7)} \\
\mathbf r_0^G &= \frac{1}{m}\int_{\mathcal D}\mathbf r_0^P\,dm \quad &\text{(divide by the scalar $m$)} \\
x_G &= \frac{1}{m}\int_{\mathcal D}x\,dm,\quad y_G = \frac{1}{m}\int_{\mathcal D}y\,dm,\quad z_G = \frac{1}{m}\int_{\mathcal D}z\,dm \quad &\text{(components in $\mathcal B_0$: constant unit vectors leave the integral)}
\end{aligned}$$
Units: $[\mathrm M]$ for $m$; $[\mathrm{M\,L}]/[\mathrm M] = [\mathrm L]$ for $\mathbf r_0^G$
Meaning: the centre of mass is the mass-weighted average of the positions of all the mass elements: the quotient of the first and the zeroth moments.
The centre of mass does not depend on the origin chosen to compute it: moving the origin from $O$ to $O'$ adds the same vector to every position and to $\mathbf r_0^G$, because the weights add up to $m$.
Symmetry and the centre of mass
When the mass distribution has a plane or an axis of symmetry, the centre of mass lies on it (Notes.pdf, Sec. 10.3). "Symmetric" means that the reflection in the plane (or the half-turn about the axis) maps the body onto itself and every mass element onto a mirror element of the same mass.
Figure 13. Mirror elements of a plane of symmetry have opposite signed distances to it: $G$ lies on the plane and every product of inertia that involves the plane vanishes..
First symmetry theorem: the centre of mass lies on a plane or axis of symmetryLaw: pairing of mirror mass elements in the first momentSource: Notes.pdf, Sec. 10.3; Notes.pdf, Eq. 10.8
Take the plane of symmetry as $Oyz$: the element $dm$ at $(x, y, z)$ has a mirror element $dm$ at $(-x, y, z)$.
$$\begin{aligned}
\int_{\mathcal D} x\,dm &= \int_{\mathcal D^{+}} x\,dm + \int_{\mathcal D^{-}} x\,dm \quad &\text{(split the body into its halves with positive and negative $x$)} \\
\int_{\mathcal D^{-}} x\,dm &= \int_{\mathcal D^{+}} (-x)\,dm \quad &\text{(map each element of the negative half to its mirror: same mass, opposite $x$)} \\
\int_{\mathcal D} x\,dm &= \int_{\mathcal D^{+}} x\,dm - \int_{\mathcal D^{+}} x\,dm = 0 \quad &\text{(the two halves cancel)} \\
x_G &= \frac{1}{m}\int_{\mathcal D} x\,dm = 0 \quad &\text{(Eq. 10.8, first component)}
\end{aligned}$$
For an axis of symmetry $Oz$ the half-turn sends $(x, y, z)$ to $(-x, -y, z)$, and the same pairing cancels both $\int x\,dm$ and $\int y\,dm$.
$$\boxed{\text{plane of symmetry } Oyz \Rightarrow x_G = 0;\qquad \text{axis of symmetry } Oz \Rightarrow x_G = y_G = 0}$$
Units: $\int x\,dm$ in $[\mathrm{M\,L}]$, $x_G$ in $[\mathrm L]$; the result is a zero length
Meaning: spotting a plane or axis of symmetry fixes one or two coordinates of $G$ before any integral is computed.
Composite bodies and holes
The lecture recommends to "divide and conquer": break a complex geometry into simpler parts, add their contributions, and treat holes by subtraction ([Lecture 8, Slide 17]). All the inertia properties are integrals over $\mathcal D$, so they are additive over disjoint parts.
Figure 14. Divide and conquer: a composite body is the sum of simple parts, and a hole is a part of negative mass ([Lecture 8, Slide 17])..
Centre of mass of a composite body and of a body with a holeLaw: additivity of the integral over disjoint domains; definition of the centre of massSource: [Lecture 8, Slide 17]; Notes.pdf, Eq. 10.6; Notes.pdf, Eq. 10.8
$$\begin{aligned}
m &= \int_{\mathcal D_1} dm + \int_{\mathcal D_2} dm = m_1 + m_2 \quad &\text{(the body is the union of the disjoint parts $\mathcal D_1$ and $\mathcal D_2$)} \\
m\,\mathbf r_0^G &= \int_{\mathcal D_1}\mathbf r_0^P\,dm + \int_{\mathcal D_2}\mathbf r_0^P\,dm \quad &\text{(first moment, split the same way)} \\
&= m_1\,\mathbf r_0^{G_1} + m_2\,\mathbf r_0^{G_2} \quad &\text{(Eq. 10.7 applied to each part)} \\
\mathbf r_0^G &= \frac{m_1\,\mathbf r_0^{G_1} + m_2\,\mathbf r_0^{G_2}}{m_1 + m_2} \quad &\text{(divide by the total mass)} \\
\mathbf r_0^G &= \frac{m_1\,\mathbf r_0^{G_1} - m_2\,\mathbf r_0^{G_2}}{m_1 - m_2} \quad &\text{(a hole $\mathcal D_2$ cut out of a full body $\mathcal D_1$: a part of negative mass)}
\end{aligned}$$
$$\boxed{\mathbf r_0^G = \frac{\sum_k m_k\,\mathbf r_0^{G_k}}{\sum_k m_k},\qquad \text{holes enter with } m_k < 0}$$
Meaning: each part acts as a point particle of its own mass at its own centre of mass; the same additivity holds for the moments, products and tensors of inertia, provided they are taken about the same point and in the same basis.
12. Moments and products of inertia
The moments and products of inertia come from the second moment of the mass distribution and govern rotational dynamics. They have dimensions $[\mathrm{M\,L^2}]$ and SI unit $\mathrm{kg\,m^2}$ (Notes.pdf, Sec. 10.4). The moment of inertia of a body about an axis is the "resistance" it offers to changes of its angular velocity about that axis ([Lecture 8, Slide 10]).
Symbol
Meaning
$d$
distance from the mass element to the point, line or plane considered
$I_r^b$
moment of inertia of the body $b$ about the axis $r$ (the slides' notation)
$I_O$
moment of inertia about the point $O$ (polar moment of inertia)
$I_x$, $I_y$, $I_z$
moments of inertia about the axes $Ox$, $Oy$, $Oz$
$I_{xy}$, $I_{xz}$, $I_{yz}$
moments of inertia about the planes $Oxy$, $Oxz$, $Oyz$
$\pi_1$, $\pi_2$; $d_1$, $d_2$
two orthogonal planes and the signed distances of the mass element to them
$P_{xy,xz}$, $P_{xy,yz}$, $P_{xz,yz}$
products of inertia with respect to pairs of coordinate planes
Figure 15. Distances used in the moments of inertia about a point, an axis and a plane, and the signed distances of the products of inertia..
Moments of inertia about a point, the coordinate axes and the coordinate planesLaw: definition of the moment of inertia; Pythagoras in $\mathcal B_0$Source: Notes.pdf, Eq. 10.9; Notes.pdf, Eq. 10.10; Notes.pdf, Eq. 10.11; Notes.pdf, Eq. 10.12; [Lecture 8, Slide 10]
The element $dm$ sits at $\mathbf r_0^P = x\,\mathbf i + y\,\mathbf j + z\,\mathbf k$.
$$\begin{aligned}
I &= \int_{\mathcal D} d^2\,dm \quad &\text{(definition, Eq. 10.9)} \\
d_O^2 &= \lvert\mathbf r_0^P\rvert^2 = x^2+y^2+z^2 \quad &\text{(distance to the point $O$)} \\
d_{Ox}^2 &= y^2+z^2,\quad d_{Oy}^2 = x^2+z^2,\quad d_{Oz}^2 = x^2+y^2 \quad &\text{(distance to an axis: hypotenuse of the two coordinates normal to it)} \\
d_{Oxy}^2 &= z^2,\quad d_{Oxz}^2 = y^2,\quad d_{Oyz}^2 = x^2 \quad &\text{(distance to a plane: the coordinate normal to it)}
\end{aligned}$$
Units: $[\mathrm L^2][\mathrm M] = [\mathrm{M\,L^2}]$, in $\mathrm{kg\,m^2}$
Meaning: mass far from the axis weighs with the square of its distance; the integrand is never negative, so no moment of inertia is negative.
Relation between the moments about the point, the axes and the planesLaw: add the definitions of Eqs. 10.10 to 10.12Source: Notes.pdf, Eq. 10.13; [Lecture 8, Slide 10]
$$\begin{aligned}
I_{xy}+I_{xz}+I_{yz} &= \int_{\mathcal D}(z^2+y^2+x^2)\,dm = I_O \quad &\text{(sum of the three plane moments, Eq. 10.12)} \\
I_x+I_y+I_z &= \int_{\mathcal D}\big[(y^2+z^2)+(x^2+z^2)+(x^2+y^2)\big]\,dm \quad &\text{(sum of the three axis moments, Eq. 10.11)} \\
&= \int_{\mathcal D} 2\,(x^2+y^2+z^2)\,dm = 2\,I_O \quad &\text{(each square appears twice; Eq. 10.10)}
\end{aligned}$$
Meaning: the moment about a point is the sum of the moments about three orthogonal planes through it, and half the sum of the moments about three orthogonal axes through it; it often saves one integral.
Erratum in the source
The notes state that the moment of inertia "is always positive" (Notes.pdf, Sec. 10.4) and [Lecture 8, Slide 12] calls the tensor of inertia "definite-positive". The integrand is never negative, so the correct statement is "never negative" (positive semi-definite): the moment vanishes when all the mass lies on the axis. Check: for a thin straight wire along $Ox$ every element has $y$ and $z$ equal to zero, so the integrand of $I_x$ vanishes everywhere and $I_x$ is zero, although the wire has mass. For any body that is not contained in a line the moments are strictly positive and the slide's statement holds.
Products of inertia
Products of inertia with respect to the coordinate planesLaw: definition of the product of inertia with signed distancesSource: Notes.pdf, Eq. 10.14; Notes.pdf, Eq. 10.15; [Lecture 8, Slide 11]
$$\begin{aligned}
P_{\pi_1,\pi_2} &= \int_{\mathcal D} d_1\,d_2\,dm \quad &\text{(definition for two orthogonal planes, Eq. 10.14)} \\
d_{Oxy} &= z,\quad d_{Oxz} = y,\quad d_{Oyz} = x \quad &\text{(signed distance to a coordinate plane: the coordinate along its normal)} \\
P_{xy,xz} &= \int_{\mathcal D} z\,y\,dm,\quad P_{xy,yz} = \int_{\mathcal D} z\,x\,dm,\quad P_{xz,yz} = \int_{\mathcal D} y\,x\,dm \quad &\text{(the three pairs of coordinate planes)}
\end{aligned}$$
Units: $[\mathrm L][\mathrm L][\mathrm M] = [\mathrm{M\,L^2}]$, in $\mathrm{kg\,m^2}$
Meaning: the signs of the distances are kept, so a product of inertia can be negative; $P_{xy,xz}$ measures the coupling between rotations about $Oy$ and $Oz$ ([Lecture 8, Slide 11]), the two axes normal to its planes.
Second symmetry theorem: a plane of symmetry cancels its products of inertiaLaw: pairing of mirror mass elements in the product integralSource: Notes.pdf, Sec. 10.4; [Lecture 8, Slide 11]
Let $Oyz$ be a plane of symmetry: the element $dm$ at $(x, y, z)$ has a mirror element $dm$ at $(-x, y, z)$.
$$\begin{aligned}
P_{xz,yz} &= \int_{\mathcal D^{+}} y\,x\,dm + \int_{\mathcal D^{+}} y\,(-x)\,dm = 0 \quad &\text{(the mirror element has the same $y$ and the opposite $x$)} \\
P_{xy,yz} &= \int_{\mathcal D^{+}} z\,x\,dm + \int_{\mathcal D^{+}} z\,(-x)\,dm = 0 \quad &\text{(same $z$, opposite $x$)} \\
P_{xy,xz} &= \int_{\mathcal D^{+}} z\,y\,dm + \int_{\mathcal D^{+}} z\,y\,dm = 2\int_{\mathcal D^{+}} zy\,dm \quad &\text{(neither factor changes sign: no cancellation)}
\end{aligned}$$
Units: $\mathrm{kg\,m^2}$; the result is a zero product of inertia
Meaning: every product of inertia that involves a plane of symmetry vanishes; the product of the other two planes survives in general. Checking the symmetries before integrating saves most of the work.
Flat bodies
For a planar body of negligible thickness in the plane $Oxy$ every mass element has zero $z$ coordinate (Notes.pdf, Sec. 10.5, below Eq. 10.21).
Figure 16. For a flat body the distance to $Oz$ is the hypotenuse of the legs $x$ and $y$, so $I_z$ equals $I_x + I_y$..
Perpendicular axis relation of a flat bodyLaw: Eqs. 10.10, 10.11 and 10.15 with the coordinate normal to the body set to zeroSource: Notes.pdf, Eq. 10.13; Notes.pdf, Eq. 10.22; Notes.pdf, Eq. 10.24
Meaning: the moment about the axis normal to a flat body equals the sum of the moments about two orthogonal in-plane axes through the same point, so only two in-plane integrals are needed; the plane of the body acts as a plane of symmetry for the products.
13. Worked examples: wire, plate, half disk and half sphere
The notes illustrate the definitions with four homogeneous bodies of Fig. 10.2, which the lecture repeats as its last four slides (Notes.pdf, Sec. 10.5; [Lecture 8, Slides 18-21]). The notes leave the moments of inertia of the half disk and of the half sphere as exercises; they are solved here and marked as extensions.
Half-circumference wire
Half-circumference wire (Notes.pdf, Sec. 10.5; [Lecture 8, Slide 18])
Example
A homogeneous wire of linear density $\lambda$ has the shape of a half circumference of radius $R$ centred at $O$ in the plane $Oxy$, on the side of positive $y$. The polar angle $\theta$, measured from $Ox$ towards $Oy$, runs from zero to $\pi$ along the wire. Find its mass, its centre of mass, its moments of inertia about the coordinate axes and its products of inertia.
Figure 17. Half-circumference wire, $\theta$ from $0$ to $\pi$; $G$ lies on the axis of symmetry at $2R/\pi$ (Notes.pdf Fig. 10.2(a))..
Mass element and total mass of the wireLaw: linear density; arc length of a circle; Eq. 10.6Source: Notes.pdf, Eq. 10.16; Notes.pdf, Eq. 10.17; Notes.pdf, Eq. 10.18; [Lecture 8, Slide 18]
$$\begin{aligned}
x &= R\cos\theta,\quad y = R\sin\theta \quad &\text{(parametric equation, Eq. 10.16; $\theta$ from $0$ to $\pi$)} \\
d\ell &= \sqrt{\Big(\tfrac{dx}{d\theta}\Big)^2+\Big(\tfrac{dy}{d\theta}\Big)^2}\,d\theta = \sqrt{R^2\sin^2\theta+R^2\cos^2\theta}\,d\theta = R\,d\theta \quad &\text{(arc length; derivatives of the parametrization)} \\
dm &= \lambda\,d\ell = \lambda R\,d\theta \quad &\text{(Eq. 10.17)} \\
m &= \int_0^{\pi}\lambda R\,d\theta = \lambda R\,\big[\theta\big]_0^{\pi} = \lambda R\,(\pi - 0) \quad &\text{(antiderivative $\theta$; Barrow's rule with both limits)}
\end{aligned}$$
Meaning: $G$ lies on the axis of symmetry $Oy$, as the first symmetry theorem predicts, at about $0.64R$ from $O$: off the wire itself.
Moments of inertia of the wire about Oz and OxLaw: Eq. 10.11; half-angle identitiesSource: Notes.pdf, Eq. 10.20; Notes.pdf, Eq. 10.21; [Lecture 8, Slide 18]
$$\begin{aligned}
I_z &= \int_{\mathcal D}(x^2+y^2)\,dm = \int_0^{\pi}R^2(\cos^2\theta+\sin^2\theta)\,\lambda R\,d\theta = \lambda R^3\big[\theta\big]_0^{\pi} = \pi\lambda R^3 \quad &\text{(every point is at distance $R$ from $Oz$; Barrow)} \\
I_x &= \int_{\mathcal D}y^2\,dm = \lambda R^3\int_0^{\pi}\sin^2\theta\,d\theta \quad &\text{($z$ is zero; $y = R\sin\theta$)} \\
\sin^2\theta &= \tfrac12(1-\cos 2\theta) \quad &\text{(half-angle identity)} \\
\int_0^{\pi}\sin^2\theta\,d\theta &= \Big[\tfrac{\theta}{2}-\tfrac{\sin 2\theta}{4}\Big]_0^{\pi} = \Big(\tfrac{\pi}{2}-\tfrac{\sin 2\pi}{4}\Big)-\Big(0-\tfrac{\sin 0}{4}\Big) = \tfrac{\pi}{2} \quad &\text{(antiderivative; Barrow with both limits)} \\
I_x &= \tfrac{\pi}{2}\lambda R^3 = \tfrac12\,(\pi\lambda R)\,R^2 \quad &\text{(write with $m$)}
\end{aligned}$$
Meaning: all the mass of the wire is at the same distance $R$ from $Oz$, so $I_z$ is that of a ring; because the wire is planar, $I_z$ is also the polar moment $I_O$.
Moment about Oy and products of inertia of the wireLaw: perpendicular axis relation of a flat body; second symmetry theoremSource: Notes.pdf, Eq. 10.22; Notes.pdf, Eq. 10.23; [Lecture 8, Slide 18]
Meaning: the integral confirms what symmetry predicts: $Oyz$ is a plane of symmetry of the wire, so $P_{xz,yz}$ vanishes, and $Oxy$ contains the wire.
Homogeneous rectangular plate
Homogeneous plate (Notes.pdf, Sec. 10.5; [Lecture 8, Slide 19])
Example
A homogeneous plate of mass $M$, length $L$ along $Ox$, width $W$ along $Oy$ and negligible thickness is centred at $O$. Find its moments and products of inertia in $S_0$, and the product of inertia with respect to the planes $Ax_1z_1$ and $Ay_1z_1$ of a frame $S_1$ with origin at a corner $A$ of the plate.
Figure 18. Homogeneous plate centred at $O \equiv G$, and the frame $S_1$ at the corner $A$ used for Eq. 10.27 (Notes.pdf Fig. 10.2(b))..
Note on the source figure
The notes refer to the frame $S_1$ "as shown in Fig. 10.2(b)", but the figure does not draw it. The limits of Eq. 10.27 (from zero to $L$ and from zero to $W$, with a positive result) fix it: origin $A$ at a corner of the plate, $Ax_1$ along the side of length $L$ and $Ay_1$ along the side of width $W$, both pointing into the plate, parallel to $Ox$ and $Oy$. The figure above draws it that way, and the Steiner check of Section 16 confirms the value.
Surface density, mass element and centre of mass of the plateLaw: area density of a homogeneous body; first symmetry theoremSource: Notes.pdf, Sec. 10.5; Notes.pdf, Eq. 10.24; [Lecture 8, Slide 19]
$$\begin{aligned}
\sigma &= \frac{M}{LW} \quad &\text{(homogeneous: mass divided by area)} \\
dm &= \sigma\,dA = \frac{M}{LW}\,dx\,dy \quad &\text{(Cartesian area element)} \\
x_G &= 0,\quad y_G = 0 \quad &\text{($Oyz$ and $Oxz$ are planes of symmetry)}
\end{aligned}$$
$$\boxed{dm = \frac{M}{LW}\,dx\,dy,\qquad G \equiv O}$$
Units: $\sigma$ in $[\mathrm M]/[\mathrm L^2] = \mathrm{kg/m^2}$; $dm$ in $\mathrm{kg}$
Meaning: two planes of symmetry through $O$ put the centre of mass at the centre of the plate.
Moments of inertia of the plate about its central axesLaw: Eq. 10.11 as a double integral over the rectangleSource: Notes.pdf, Eq. 10.24; Notes.pdf, Eq. 10.25; Notes.pdf, Eq. 10.26; [Lecture 8, Slide 19]
Units: $[\mathrm M][\mathrm L^2]$, in $\mathrm{kg\,m^2}$
Meaning: the moment about $Ox$ depends only on the width $W$, the spread of the mass away from $Ox$; the long side $L$ makes $I_y$ the larger one when $L$ exceeds $W$.
Products of inertia of the plate in S0 and at the corner ALaw: second symmetry theorem; Eq. 10.15 as a double integralSource: Notes.pdf, Eq. 10.27; Notes.pdf, Sec. 10.5; [Lecture 8, Slide 19]
$$\begin{aligned}
P_{xy,xz} &= P_{xy,yz} = 0 \quad &\text{(the plate lies in $Oxy$)} \\
P_{xz,yz} &= 0 \quad &\text{($Oyz$ and $Oxz$ are planes of symmetry through $O$)} \\
P_{x_1z_1,y_1z_1} &= \int_0^{W}\int_0^{L}\frac{M}{LW}\,x\,y\,dx\,dy = \frac{M}{LW}\int_0^{W}y\Big[\frac{x^2}{2}\Big]_0^{L}dy \quad &\text{(coordinates measured from the corner $A$; inner integral in $x$)} \\
&= \frac{M}{LW}\,\frac{L^2}{2}\Big[\frac{y^2}{2}\Big]_0^{W} = \frac{M}{LW}\,\frac{L^2}{2}\,\frac{W^2}{2} \quad &\text{(Barrow with both limits, twice)}
\end{aligned}$$
Meaning: seen from the corner the whole plate lies in the quadrant of positive $x_1$ and $y_1$, so the product is positive; the planes through $A$ are no longer planes of symmetry. The products with the plane $Ax_1y_1$ stay zero because the plate lies in it.
Homogeneous half disk
Half disk (Notes.pdf, Sec. 10.5; [Lecture 8, Slide 20])
Example
A homogeneous half disk of radius $R$ and area density $\sigma$ lies in the plane $Oxy$, centred at $O$, on the side of positive $y$. Find its mass and its centre of mass (notes), and its moments of inertia about the coordinate axes (left as an exercise in the notes).
Figure 19. Half disk with the polar element $r\,dr\,d\theta$; $G$ lies on the axis of symmetry at $4R/(3\pi)$ (Notes.pdf Fig. 10.2(c))..
Mass element and mass of the half diskLaw: polar coordinates in the plane; area densitySource: Notes.pdf, Eq. 10.28; Notes.pdf, Eq. 10.29; Notes.pdf, Eq. 10.30; [Lecture 8, Slide 20]
$$\begin{aligned}
x &= r\cos\theta,\quad y = r\sin\theta \quad &\text{(Eq. 10.28; $r$ from $0$ to $R$, $\theta$ from $0$ to $\pi$)} \\
dA &= r\,dr\,d\theta \quad &\text{(Jacobian $r$ of the polar coordinates, Section 10 with $z$ fixed)} \\
dm &= \sigma\,r\,dr\,d\theta \quad &\text{(Eq. 10.29)} \\
m &= \sigma\int_0^{\pi}\int_0^{R}r\,dr\,d\theta = \sigma\Big[\frac{r^2}{2}\Big]_0^{R}\big[\theta\big]_0^{\pi} = \sigma\,\frac{R^2}{2}\,\pi \quad &\text{(the limits are constant: the integral factorizes; Barrow)}
\end{aligned}$$
Meaning: $G$ is at about $0.42R$ from $O$ on the axis of symmetry, closer to $O$ than for the wire ($0.64R$), because the half disk keeps mass near the centre.
Extension: moments of inertia of the half disk (exercise of the notes)Law: Eq. 10.11 in polar coordinates; perpendicular axis relationSource: Notes.pdf, Sec. 10.5; Notes.pdf, Eq. 10.11; [Lecture 8, Slide 20]
Meaning: these are the values of a full disk per unit mass: each half of a disk carries half the mass and half the moment about any diameter or about the normal axis through the centre.
Homogeneous half sphere
Half sphere (Notes.pdf, Sec. 10.5; [Lecture 8, Slide 21])
Example
A solid homogeneous half sphere (hemisphere) of radius $R$ and volume density $\rho$ has its flat face on the plane $Oxy$, centred at $O$, and lies on the side of positive $z$. Find its mass and its centre of mass (notes), and its moments and products of inertia about $O$ (left as an exercise in the notes). The spherical coordinates are those of Section 10: radius $r$, longitude $\theta$ measured in $Oxy$ from $Ox$, latitude $\phi$ measured from $Oxy$.
Figure 20. Half sphere with the latitude $\phi$ measured from $Oxy$ (so the volume element is $r^2\cos\phi\,dr\,d\theta\,d\phi$); $G$ lies on the axis at $3R/8$ (Notes.pdf Fig. 10.2(d))..
Erratum in the source
Eq. 10.32 of the notes reads "dm equals r squared cos phi dr dtheta dphi": the density is missing. The right-hand side is the volume element of Eq. 10.5, in cubic metres, so it cannot be a mass; Eq. 10.33, in the next line, does carry $\rho$. The note writes the mass element with $\rho$. The same paragraph refers to the half sphere as Fig. 10.2(c); it is Fig. 10.2(d).
Mass element and mass of the half sphereLaw: volume density; spherical volume element of Eq. 10.5Source: Notes.pdf, Eq. 10.5; Notes.pdf, Eq. 10.32; Notes.pdf, Eq. 10.33; [Lecture 8, Slide 21]
$$\begin{aligned}
dm &= \rho\,dV = \rho\,r^2\cos\phi\,dr\,d\theta\,d\phi \quad &\text{(Eq. 10.32 with the density restored)} \\
m &= \rho\int_0^{R}r^2\,dr\int_0^{\pi/2}\cos\phi\,d\phi\int_0^{2\pi}d\theta \quad &\text{(limits: $r$ to $R$, upper half $\phi$ from $0$ to $\pi/2$, full turn in $\theta$; the integral factorizes)} \\
\int_0^{R}r^2\,dr &= \Big[\frac{r^3}{3}\Big]_0^{R} = \frac{R^3}{3},\quad \int_0^{\pi/2}\cos\phi\,d\phi = \big[\sin\phi\big]_0^{\pi/2} = 1,\quad \int_0^{2\pi}d\theta = 2\pi \quad &\text{(Barrow with both limits)} \\
m &= \rho\,\frac{R^3}{3}\cdot 1\cdot 2\pi \quad &\text{(product of the three factors)}
\end{aligned}$$
Meaning: $G$ lies on the axis of symmetry, below mid-height, because most of the volume of the hemisphere is near its flat face.
Extension: moments and products of inertia of the half sphere about O (exercise of the notes)Law: Eqs. 10.10, 10.11 and 10.13 in spherical coordinates; second symmetry theoremSource: Notes.pdf, Sec. 10.5; Notes.pdf, Eq. 10.13; [Lecture 8, Slide 21]
$$\begin{aligned}
I_O &= \int_{\mathcal D}r^2\,dm = \rho\Big[\frac{r^5}{5}\Big]_0^{R}\big[\sin\phi\big]_0^{\pi/2}\big[\theta\big]_0^{2\pi} = \frac{2\pi\rho R^5}{5} \quad &\text{(distance to $O$ is $r$; Barrow three times)} \\
I_z &= \int_{\mathcal D}(x^2+y^2)\,dm = \rho\int_0^{R}r^4\,dr\int_0^{\pi/2}\cos^3\phi\,d\phi\int_0^{2\pi}d\theta \quad &\text{(distance to $Oz$ is the leg $r\cos\phi$)} \\
\int_0^{\pi/2}\cos^3\phi\,d\phi &= \int_0^{1}(1-u^2)\,du = \Big[u-\frac{u^3}{3}\Big]_0^{1} = \frac23 \quad &\text{($u = \sin\phi$, $du = \cos\phi\,d\phi$, limits $0$ and $1$; Barrow)} \\
I_z &= \rho\,\frac{R^5}{5}\,\frac23\,2\pi = \frac{4\pi\rho R^5}{15} \quad &\text{(product of the factors)} \\
I_x &= I_y = \tfrac12\,(2I_O - I_z) = \tfrac12\Big(\frac{4\pi\rho R^5}{5}-\frac{4\pi\rho R^5}{15}\Big) = \frac{4\pi\rho R^5}{15} \quad &\text{(Eq. 10.13; $I_x = I_y$ by the axial symmetry)} \\
P_{xz,yz} &= P_{xy,yz} = 0,\qquad P_{xy,xz} = 0 \quad &\text{($Oyz$ and $Oxz$ are planes of symmetry; each product involves one of them)}
\end{aligned}$$
$$\boxed{I_x = I_y = I_z = \tfrac25\,mR^2,\qquad I_O = \tfrac35\,mR^2,\qquad \text{all products about } O \text{ zero}}$$
Units: $[\mathrm{M\,L^{-3}}][\mathrm L^5] = \mathrm{kg\,m^2}$; with $m$ from the previous box the ratios are dimensionless
Meaning: about $O$ the hemisphere has the moments of a full sphere per unit mass: the mirror half below $Oxy$ would add an equal contribution. The tensor about $O$ is a multiple of the unit tensor, so every axis through $O$ is principal.
14. The tensor of inertia
The moments and products of inertia appear in the dynamics of material systems as the components of one second-order tensor, the tensor of inertia (Notes.pdf, Sec. 10.6). The notes define it directly; Chapter 11 shows that it is the tensor that turns the angular velocity of a rigid body into its angular momentum and kinetic energy (Notes.pdf, Sec. 11.2), which is why it "describes the rotational inertia" of the body ([Lecture 8, Slide 12]).
Symbol
Meaning
$\boldsymbol{\mathcal I}_O$
tensor of inertia of the material system about the point $O$, origin of $S_0$
$[\boldsymbol{\mathcal I}_O]_{00}$
its components in $\mathcal B_0\otimes\mathcal B_0$
$\ell$, $\mathbf u$
a line through $O$ and its unit vector
$I_\ell$
moment of inertia about the line $\ell$
$\mathbf u_1$, $\mathbf u_2$
unit normals of two orthogonal planes through $O$
Definition and components of the tensor of inertiaLaw: definition of the tensor of inertia; components of the unit tensor and of a dyadSource: Notes.pdf, Eq. 10.35; Notes.pdf, Eq. 10.36; Notes.pdf, Eq. 10.37; Notes.pdf, Eq. 10.38; [Lecture 8, Slide 12]
$$\begin{aligned}
\boldsymbol{\mathcal I}_O &= \int_{\mathcal D}\Big[(r_0^P)^2\,\boldsymbol{\mathcal U} - \mathbf r_0^P\otimes\mathbf r_0^P\Big]\,dm \quad &\text{(definition, Eq. 10.35)} \\
\big[(r_0^P)^2\,\boldsymbol{\mathcal U}\big]_{00} &= \begin{bmatrix} x^2+y^2+z^2 & 0 & 0 \cr 0 & x^2+y^2+z^2 & 0 \cr 0 & 0 & x^2+y^2+z^2 \end{bmatrix} \quad &\text{(Eq. 10.36: $(r_0^P)^2$ times the identity)} \\
\big[\mathbf r_0^P\otimes\mathbf r_0^P\big]_{00} &= \begin{bmatrix} x^2 & xy & xz \cr xy & y^2 & yz \cr xz & yz & z^2 \end{bmatrix} \quad &\text{(Eq. 10.37: column times row, Eq. 1.48)} \\
[\boldsymbol{\mathcal I}_O]_{00} &= \int_{\mathcal D}\begin{bmatrix} y^2+z^2 & -xy & -xz \cr -xy & x^2+z^2 & -yz \cr -xz & -yz & x^2+y^2 \end{bmatrix}dm \quad &\text{(subtract entry by entry; the basis is constant, so the integral acts on each entry)} \\
&= \begin{bmatrix} I_x & -P_{xz,yz} & -P_{xy,yz} \cr -P_{xz,yz} & I_y & -P_{xy,xz} \cr -P_{xy,yz} & -P_{xy,xz} & I_z \end{bmatrix} \quad &\text{(identify each entry with Eqs. 10.11 and 10.15)}
\end{aligned}$$
Units: every entry in $[\mathrm{M\,L^2}]$, $\mathrm{kg\,m^2}$
Meaning: the diagonal holds the moments of inertia about the coordinate axes and the off-diagonal entries the products of inertia with a minus sign; the entry in row $x$ and column $y$ is minus the product of the two planes normal to $Ox$ and $Oy$. The matrix is symmetric, and its second-moment part is $\mathcal M_{O,2}^{\mathcal D}$.
Figure 21. The distance $d$ from $dm$ to the line $\ell$ is the leg of the right triangle with hypotenuse $\mathbf r_0^P$ and adjacent leg $\mathbf r_0^P\cdot\mathbf u$..
Moment of inertia about any line through OLaw: Pythagoras for the distance to a line; contraction of the unit tensor and of a dyadSource: Notes.pdf, Eq. 10.39; Notes.pdf, Eq. 1.36; [Lecture 8, Slide 13]
The vector $\mathbf r_0^P$ is the hypotenuse of a right triangle whose leg along $\ell$ is the projection $\mathbf r_0^P\cdot\mathbf u$ and whose other leg is the distance $d$ to the line.
$$\begin{aligned}
d^2 &= (r_0^P)^2 - (\mathbf r_0^P\cdot\mathbf u)^2 \quad &\text{(Pythagoras in that right triangle)} \\
\mathbf u\cdot\boldsymbol{\mathcal U}\cdot\mathbf u &= \mathbf u\cdot\mathbf u = 1,\qquad \mathbf u\cdot(\mathbf r_0^P\otimes\mathbf r_0^P)\cdot\mathbf u = (\mathbf u\cdot\mathbf r_0^P)(\mathbf r_0^P\cdot\mathbf u) \quad &\text{(unit tensor; Eq. 1.36)} \\
d^2 &= \mathbf u\cdot\Big[(r_0^P)^2\,\boldsymbol{\mathcal U} - \mathbf r_0^P\otimes\mathbf r_0^P\Big]\cdot\mathbf u \quad &\text{(write both terms as contractions)} \\
I_\ell &= \int_{\mathcal D}d^2\,dm = \mathbf u\cdot\Big\{\int_{\mathcal D}\Big[(r_0^P)^2\,\boldsymbol{\mathcal U} - \mathbf r_0^P\otimes\mathbf r_0^P\Big]dm\Big\}\cdot\mathbf u \quad &\text{(Eq. 10.9; the constant $\mathbf u$ leaves the integral)}
\end{aligned}$$
Units: $[\mathrm{M\,L^2}]$ times two dimensionless unit vectors: $\mathrm{kg\,m^2}$
Meaning: the six independent components of the tensor give the moment of inertia about every line through $O$; with $\mathbf u$ equal to $\mathbf i$ the formula returns the first diagonal entry, $I_x$. Since every $d^2$ is never negative, the tensor is positive semi-definite ([Lecture 8, Slide 12]; see the erratum of Section 12).
Product of inertia of any two orthogonal planes through OLaw: signed distance to a plane through $O$; orthogonality of the normalsSource: Notes.pdf, Eq. 10.40; Notes.pdf, Eq. 10.14; Notes.pdf, Eq. 1.36
$$\begin{aligned}
d_1 &= \mathbf r_0^P\cdot\mathbf u_1,\qquad d_2 = \mathbf r_0^P\cdot\mathbf u_2 \quad &\text{(signed distance: projection on the normal of each plane)} \\
\mathbf u_1\cdot\boldsymbol{\mathcal U}\cdot\mathbf u_2 &= \mathbf u_1\cdot\mathbf u_2 = 0 \quad &\text{(the planes are orthogonal, so are their normals)} \\
-\mathbf u_1\cdot\boldsymbol{\mathcal I}_O\cdot\mathbf u_2 &= -\int_{\mathcal D}\Big[(r_0^P)^2\cdot 0 - (\mathbf u_1\cdot\mathbf r_0^P)(\mathbf r_0^P\cdot\mathbf u_2)\Big]dm = \int_{\mathcal D}d_1\,d_2\,dm \quad &\text{(Eq. 10.35 contracted; Eq. 1.36)}
\end{aligned}$$
Meaning: reversing one normal reverses the sign of the product (Notes.pdf, Sec. 10.6, footnote 17); with $\mathbf u_1$ equal to $\mathbf j$ and $\mathbf u_2$ equal to $\mathbf i$ the formula returns $P_{xz,yz}$, the entry of the matrix with its minus sign removed.
Third symmetry theorem: two planes of symmetry make the tensor diagonalLaw: second symmetry theorem applied to two coordinate planesSource: [Lecture 8, Slide 13]; Notes.pdf, Sec. 10.4; Notes.pdf, Eq. 10.38
$$\boxed{\text{two coordinate planes of symmetry} \Rightarrow [\boldsymbol{\mathcal I}_O]_{00} \text{ diagonal: } \mathbf i,\ \mathbf j,\ \mathbf k \text{ are principal}}$$
Units: $\mathrm{kg\,m^2}$
Meaning: each coordinate plane appears in two of the three products, and two different planes together appear in all three, so two planes of symmetry cancel every product. The plate and the half sphere of Section 13 have two planes of symmetry through $O$; the half disk has one and lies in $Oxy$, which cancels the other two products; all three have diagonal tensors about $O$.
Extension: moment of the rectangular plate about its diagonalLaw: moment about a line through $O$ from the tensor of inertiaSource: Notes.pdf, Eq. 10.39; Notes.pdf, Eq. 10.25; Notes.pdf, Eq. 10.26
The diagonal of the plate of Section 13 passes through $G \equiv O$ with unit vector along $(L, W, 0)$; the tensor about $O$ is diagonal by the third symmetry theorem.
$$\begin{aligned}
[\mathbf u]_0 &= \frac{1}{\sqrt{L^2+W^2}}\begin{bmatrix} L \cr W \cr 0 \end{bmatrix} \quad &\text{(diagonal direction, normalised)} \\
I_\ell &= \frac{1}{L^2+W^2}\begin{bmatrix} L & W & 0 \end{bmatrix}\begin{bmatrix} \tfrac{1}{12}MW^2 & 0 & 0 \cr 0 & \tfrac{1}{12}ML^2 & 0 \cr 0 & 0 & \tfrac{1}{12}M(L^2+W^2) \end{bmatrix}\begin{bmatrix} L \cr W \cr 0 \end{bmatrix} \quad &\text{(Eq. 10.39 in matrix form)} \\
&= \frac{1}{L^2+W^2}\Big(\tfrac{1}{12}MW^2L^2 + \tfrac{1}{12}ML^2W^2\Big) \quad &\text{(row times matrix times column)}
\end{aligned}$$
Meaning: for a square plate, with $L$ and $W$ both equal to $a$, the result is $Ma^2/12$, the same as about $Ox$: the in-plane moment of a square about its centre is the same for every direction.
15. Change of basis of the tensor and principal axes of inertia
A tensor and a vector can only be contracted when both are written in consistent bases: if the tensor is in $\mathcal B_0\otimes\mathcal B_0$, the vector must be in $\mathcal B_0$ (Notes.pdf, Sec. 10.7). A change of basis changes the component matrix, never the tensor, just as for a vector.
Symbol
Meaning
$\mathcal B_1$
a rotated orthonormal basis $\{\mathbf i_1,\mathbf j_1,\mathbf k_1\}$
$[{}_0R_1]$
change of basis matrix from $\mathcal B_1$ to $\mathcal B_0$: column $n$ holds vector $n$ of $\mathcal B_1$ in $\mathcal B_0$ (Section 6)
$[\boldsymbol{\mathcal I}_O]_{11}$
components of the same tensor in $\mathcal B_1\otimes\mathcal B_1$
principal directions (unit eigenvectors) and principal moments (eigenvalues)
Components of the tensor of inertia in a rotated basisLaw: components of a second-order tensor as double contractions; change of basis of vectorsSource: Notes.pdf, Eq. 10.41; [Lecture 8, Slide 5]; [Lecture 8, Slide 13]; Notes.pdf, Eq. 1.13; Notes.pdf, Eq. 1.14
Write the vectors of $\mathcal B_1$ as $\mathbf b_1$, $\mathbf b_2$, $\mathbf b_3$ (that is $\mathbf i_1$, $\mathbf j_1$, $\mathbf k_1$).
$$\begin{aligned}
\big([\boldsymbol{\mathcal I}_O]_{11}\big)_{mn} &= \mathbf b_m\cdot\boldsymbol{\mathcal I}_O\cdot\mathbf b_n \quad &\text{(entry $mn$ in $\mathcal B_1\otimes\mathcal B_1$: contract with the $m$-th and $n$-th basis vectors)} \\
&= [\mathbf b_m]_0^{T}\,[\boldsymbol{\mathcal I}_O]_{00}\,[\mathbf b_n]_0 \quad &\text{(evaluate the scalar with every factor in $\mathcal B_0$, Section 10)} \\
[\mathbf b_n]_0 &= \text{column } n \text{ of } [{}_0R_1] \quad &\text{(definition of the change of basis matrix)} \\
[\boldsymbol{\mathcal I}_O]_{11} &= [{}_0R_1]^{T}\,[\boldsymbol{\mathcal I}_O]_{00}\,[{}_0R_1] \quad &\text{(the nine entries at once: rows of $[{}_0R_1]^{T}$ times columns of $[{}_0R_1]$)} \\
[\boldsymbol{\mathcal I}_O]_{00} &= [{}_0R_1]\,[\boldsymbol{\mathcal I}_O]_{11}\,[{}_0R_1]^{T} \quad &\text{(invert: $[{}_0R_1]^{-1} = [{}_0R_1]^{T}$, Section 6)}
\end{aligned}$$
Units: $[{}_0R_1]$ is dimensionless, so both matrices are in $\mathrm{kg\,m^2}$
Meaning: the rule holds for every second-order tensor; the slides write the left factor as $[{}_1R_0]$, which is the same matrix because a rotation matrix is orthogonal. A symmetric matrix stays symmetric, since the transpose of the product is the product of the transposes in reverse order.
Principal axes of inertia
There is always a rotated basis in which the tensor of inertia is diagonal; its directions are the principal axes (directions) of inertia and its diagonal entries the principal moments of inertia (Notes.pdf, Sec. 10.7.1; [Lecture 8, Slide 13]). Because the tensor is symmetric, the spectral theorem of linear algebra guarantees that they exist and that the principal moments are real ([Lecture 8, Slide 6]).
Figure 22. About the corner $A$ the principal axes of a square plate are the diagonal ($I_1 = Ma^2/12$) and the perpendicular in-plane direction ($I_2 = 7Ma^2/12$), a rotation of $\pi/4$ about $\mathbf k_1$..
Principal moments and directions as an eigenvalue problemLaw: diagonal form of the tensor; eigenvalue problem of a symmetric tensorSource: Notes.pdf, Eq. 10.42; Notes.pdf, Eq. 10.43; [Lecture 8, Slide 6]; [Lecture 8, Slide 13]
$$\begin{aligned}
[\boldsymbol{\mathcal I}_O]_{11} &= \begin{bmatrix} I_1 & 0 & 0 \cr 0 & I_2 & 0 \cr 0 & 0 & I_3 \end{bmatrix}_{11} \quad &\text{(principal basis $\mathcal B_1$ of vectors $\mathbf e_1$, $\mathbf e_2$, $\mathbf e_3$, Eq. 10.42)} \\
\boldsymbol{\mathcal I}_O\cdot\mathbf e_i &= I_i\,\mathbf e_i \quad &\text{(column $i$ of a diagonal matrix: Eq. 10.43, no sum over $i$)} \\
\big([\boldsymbol{\mathcal I}_O]_{00} - I_i\,\mathbb 1\big)[\mathbf e_i]_0 &= \mathbf 0 \quad &\text{(the same equation in $\mathcal B_0$; $\mathbb 1$ is the identity matrix)} \\
\det\big([\boldsymbol{\mathcal I}_O]_{00} - I\,\mathbb 1\big) &= 0 \quad &\text{(a non-zero solution needs a singular matrix: cubic in $I$)} \\
I_i &= \mathbf e_i\cdot\boldsymbol{\mathcal I}_O\cdot\mathbf e_i \geq 0 \quad &\text{(Eq. 10.39 with the unit vector $\mathbf e_i$: each principal moment is the moment about its axis)}
\end{aligned}$$
Units: the eigenvalues $I_i$ in $\mathrm{kg\,m^2}$; the eigenvectors are unit vectors, dimensionless
Meaning: the three roots of the cubic are the principal moments and the unit solutions of each linear system the principal directions; symmetries of the body often give them without solving anything ([Lecture 8, Slide 13]).
Principal directions of different principal moments are orthogonalLaw: symmetry of the tensor of inertiaSource: [Lecture 8, Slide 6]; Notes.pdf, Sec. 10.7.1; Notes.pdf, Eq. 10.38
Units: $\mathrm{kg\,m^2}$ times a dimensionless dot product
Meaning: principal directions of distinct principal moments are automatically orthogonal; when two principal moments coincide, every direction of their plane is principal and an orthonormal pair is chosen in it. Either way the principal basis is orthonormal, and it is chosen right-handed.
Extension: principal axes of a square plate about a cornerLaw: Eq. 10.11 and Eq. 10.15 from the corner; eigenvalue problem; change of basis of the tensorSource: Notes.pdf, Eq. 10.27; Notes.pdf, Eq. 10.41; Notes.pdf, Eq. 10.43
The plate of Section 13 with $L$ and $W$ both equal to $a$, about its corner $A$ in the frame $S_1$ (coordinates $x$, $y$ from the corner, $dm$ equal to $M\,dx\,dy/a^2$).
$$\begin{aligned}
I_{x_1} &= \frac{M}{a^2}\int_0^{a}\int_0^{a}y^2\,dx\,dy = \frac{M}{a^2}\,\big[x\big]_0^{a}\Big[\frac{y^3}{3}\Big]_0^{a} = \tfrac13 Ma^2 \quad &\text{(Eq. 10.11 with $z$ zero; Barrow)} \\
I_{y_1} &= \tfrac13 Ma^2,\qquad I_{z_1} = I_{x_1} + I_{y_1} = \tfrac23 Ma^2,\qquad P_{x_1z_1,y_1z_1} = \tfrac14 Ma^2 \quad &\text{(same steps with $x^2$; flat body; Eq. 10.27)} \\
[\boldsymbol{\mathcal I}_A]_{11} &= Ma^2\begin{bmatrix} \tfrac13 & -\tfrac14 & 0 \cr -\tfrac14 & \tfrac13 & 0 \cr 0 & 0 & \tfrac23 \end{bmatrix} \quad &\text{(Eq. 10.38: the other two products vanish, plate in $Ax_1y_1$)} \\
\Big(\tfrac13 - \tfrac{I}{Ma^2}\Big)^2 - \tfrac{1}{16} &= 0 \quad &\text{(characteristic equation of the in-plane block; $\mathbf k_1$ is already principal)} \\
\frac{I}{Ma^2} &= \tfrac13 \mp \tfrac14 = \tfrac{1}{12},\ \tfrac{7}{12} \quad &\text{(square root of both sides)} \\
\mathbf e_1 &= \tfrac{1}{\sqrt2}(\mathbf i_1+\mathbf j_1),\quad \mathbf e_2 = \tfrac{1}{\sqrt2}(-\mathbf i_1+\mathbf j_1),\quad \mathbf e_3 = \mathbf k_1 \quad &\text{(solve each system: $\tfrac{1}{12}$ gives equal components, $\tfrac{7}{12}$ opposite ones)} \\
[{}_1R_2] &= \frac{1}{\sqrt2}\begin{bmatrix} 1 & -1 & 0 \cr 1 & 1 & 0 \cr 0 & 0 & \sqrt2 \end{bmatrix},\quad \det = \tfrac12(1+1) = 1,\quad [{}_1R_2][{}_1R_2]^{T} = \mathbb 1 \quad &\text{(columns: principal vectors; a rotation of $\pi/4$ about $\mathbf k_1$, which points out of the paper)} \\
[{}_1R_2]^{T}[\boldsymbol{\mathcal I}_A]_{11}[{}_1R_2] &= Ma^2\,\mathrm{diag}\Big(\tfrac{1}{12},\ \tfrac{7}{12},\ \tfrac23\Big) \quad &\text{(Eq. 10.41: the off-diagonal entries cancel)}
\end{aligned}$$
$$\boxed{I_1 = \tfrac{1}{12}Ma^2 \text{ along the diagonal},\qquad I_2 = \tfrac{7}{12}Ma^2,\qquad I_3 = \tfrac23 Ma^2 \text{ along } \mathbf k_1}$$
Units: $\mathrm{kg\,m^2}$; the rotation matrix is dimensionless
Meaning: about the corner the axes along the sides are not principal, because the product of inertia is non-zero; the principal axes are the diagonal of the square through $A$, which is a line of symmetry of the plate, and the line perpendicular to it. The smallest moment is about the diagonal, along which the mass lies closest. The sum of the in-plane principal moments, $\tfrac23 Ma^2$, equals $I_3$, as a flat body requires.
16. Change of reference point: Steiner's theorem
The tensor of inertia depends on the point about which it is taken. Steiner's theorem relates the tensor about any point $O$ to the tensor about the centre of mass $G$ (Notes.pdf, Sec. 10.8; [Lecture 8, Slides 14-15]).
Symbol
Meaning
$S_G = \{G;\mathcal B_0\}$
frame with origin at $G$ and the same basis as $S_0$ (parallel to it)
$\mathbf r_G^P$
position of the mass element relative to $G$, components $x_G^P$, $y_G^P$, $z_G^P$
$\mathbf r_0^G$
position of $G$ in $S_0$, components $x_0^G$, $y_0^G$, $z_0^G$
$\boldsymbol{\mathcal I}_G$
tensor of inertia about $G$
$I_{z_G}$, $I_{z_0}$
moments of inertia about the parallel axes $Gz_G$ and $Oz_0$
Figure 23. The frame $S_G$ at the centre of mass is parallel to $S_0$, and $\mathbf r_0^P = \mathbf r_0^G + \mathbf r_G^P$ (Notes.pdf Fig. 10.3)..
The first moment about the centre of mass vanishesLaw: definition of the centre of mass; linearity of the integralSource: Notes.pdf, Eq. 10.44; Notes.pdf, Eq. 10.45; Notes.pdf, Eq. 10.46; [Lecture 8, Slide 14]
$$\boxed{\int_{\mathcal D}\mathbf r_G^P\,dm = \mathbf 0\ \Rightarrow\ \text{every term linear in } \mathbf r_G^P \text{ integrates to zero}}$$
Units: $[\mathrm{M\,L}]$ for the first moment, $[\mathrm{M\,L^2}]$ for the two tensor integrals
Meaning: measured from $G$ the mass distribution has no first moment; this is the property that makes $G$ the natural point for inertia.
Steiner's theorem for the tensor of inertiaLaw: definition of the tensor of inertia at $O$ and at $G$; expansion of the square and of the dyadSource: Notes.pdf, Eq. 10.47; Notes.pdf, Eq. 10.48; [Lecture 8, Slide 14]; [Lecture 8, Slide 15]
Units: $[\mathrm{M\,L^2}]$ on both sides: $m$ in $\mathrm{kg}$ times $(r_0^G)^2$ in $\mathrm{m^2}$
Meaning: the tensor about $O$ is the tensor about $G$ plus the tensor about $O$ of a point particle of mass $m$ placed at $G$.
Steiner's theorem in componentsLaw: Eqs. 10.36 to 10.38 applied to a single particle of mass $m$ at $G$Source: Notes.pdf, Eq. 10.48; Notes.pdf, Eq. 10.38; [Lecture 8, Slide 15]
$$\begin{aligned}
m\Big[(r_0^G)^2\boldsymbol{\mathcal U} - \mathbf r_0^G\otimes\mathbf r_0^G\Big]_{00} &= m\begin{bmatrix} (y_0^G)^2+(z_0^G)^2 & -x_0^Gy_0^G & -x_0^Gz_0^G \cr -x_0^Gy_0^G & (x_0^G)^2+(z_0^G)^2 & -y_0^Gz_0^G \cr -x_0^Gz_0^G & -y_0^Gz_0^G & (x_0^G)^2+(y_0^G)^2 \end{bmatrix} \quad &\text{(Eq. 10.38 with the integral replaced by one particle)} \\
I_{x} &= I_{x_G} + m\big[(y_0^G)^2+(z_0^G)^2\big] \quad &\text{(diagonal entry $xx$)} \\
P_{xz,yz} &= P_{x_Gz_G,y_Gz_G} + m\,x_0^G\,y_0^G \quad &\text{(entry $xy$ with its minus sign removed)}
\end{aligned}$$
Meaning: all the terms must be in the same tensor basis, which is why $S_G$ is taken parallel to $S_0$ (Notes.pdf, Sec. 10.8); if the tensor about $G$ is known in another basis, it is rotated first with Eq. 10.41 ([Lecture 8, Slide 15]).
From a point A to a point B: Steiner's theorem twiceLaw: Steiner's theorem applied from $G$ to each of the two pointsSource: Notes.pdf, Sec. 10.8; [Lecture 8, Slide 15]
$$\begin{aligned}
\boldsymbol{\mathcal I}_A &= \boldsymbol{\mathcal I}_G + m\Big[(r_A^G)^2\boldsymbol{\mathcal U} - \mathbf r_A^G\otimes\mathbf r_A^G\Big] \quad &\text{(Eq. 10.48 with $O$ replaced by $A$; $\mathbf r_A^G$ is $\mathbf{AG}$)} \\
\boldsymbol{\mathcal I}_G &= \boldsymbol{\mathcal I}_A - m\Big[(r_A^G)^2\boldsymbol{\mathcal U} - \mathbf r_A^G\otimes\mathbf r_A^G\Big] \quad &\text{(first step: from $A$ down to $G$)} \\
\boldsymbol{\mathcal I}_B &= \boldsymbol{\mathcal I}_G + m\Big[(r_B^G)^2\boldsymbol{\mathcal U} - \mathbf r_B^G\otimes\mathbf r_B^G\Big] \quad &\text{(second step: from $G$ up to $B$)}
\end{aligned}$$
Meaning: the theorem only links a point to the centre of mass; the particle term computed with $\mathbf{AB}$ directly would be wrong, because the cross terms vanish only when the integrals are taken from $G$.
Check of Eq. 10.27 with Steiner's theoremLaw: Steiner's theorem in components from the centre of the plate to its cornerSource: Notes.pdf, Eq. 10.48; Notes.pdf, Eq. 10.25; Notes.pdf, Eq. 10.26; Notes.pdf, Eq. 10.27
For the plate of Section 13, $G$ is at the centre; seen from the corner $A$, with axes along the sides, $\mathbf r_A^G$ has components $L/2$, $W/2$, $0$.
Meaning: the direct integral from the corner and Steiner's theorem agree; with $L$ and $W$ both equal to $a$ they also reproduce the corner tensor used for the principal axes in Section 15.
Parallel axis theorem
Often only the moment of inertia about an axis $Gz_G$ through $G$ is known and the moment about a parallel axis $Oz_0$ is needed, as in planar problems, where only the moment about the axis normal to the plane of motion enters (Notes.pdf, Sec. 10.8.1; [Lecture 8, Slide 16]).
Figure 24. Parallel axis theorem: $I_{z_0}$ exceeds $I_{z_G}$ by $m$ times the squared distance between the axes (Notes.pdf Fig. 10.4)..
Parallel axis theoremLaw: Eq. 10.11 with the coordinates split at $G$; Eq. 10.44Source: Notes.pdf, Sec. 10.8.1; Notes.pdf, Eq. 10.44; [Lecture 8, Slide 16]
Units: $\mathrm{kg\,m^2}$: $m$ times the square of the distance between the two axes
Meaning: the bracket is the squared distance between the parallel axes, so the moment about an axis through $G$ is the smallest of all the moments about axes parallel to it, wherever $O$ is. It is the $zz$ entry of Steiner's theorem.
Extension: central moments of the wire and of the half sphereLaw: parallel axis theorem used from $O$ down to $G$Source: Notes.pdf, Sec. 10.8.1; Notes.pdf, Eq. 10.19; Notes.pdf, Eq. 10.20; Notes.pdf, Eq. 10.34
$$\begin{aligned}
I_{z_G}^{\text{wire}} &= I_z - m\Big(\tfrac{2R}{\pi}\Big)^2 = mR^2\Big(1 - \tfrac{4}{\pi^2}\Big) \quad &\text{(Eq. 10.20; $G$ at distance $2R/\pi$ from $Oz$, Eq. 10.19)} \\
I_{x_G}^{\text{sphere}} &= I_x - m\Big(\tfrac{3R}{8}\Big)^2 = mR^2\Big(\tfrac25 - \tfrac{9}{64}\Big) = \tfrac{128-45}{320}\,mR^2 \quad &\text{($I_x$ from Section 13; $G$ at height $3R/8$, Eq. 10.34, so at that distance from $Ox$)} \\
I_{z_G}^{\text{sphere}} &= I_z = \tfrac25 mR^2 \quad &\text{($G$ lies on $Oz$: zero distance between the axes)}
\end{aligned}$$
Meaning: both central moments are smaller than those about $O$, as the theorem requires; the subtraction always goes from a known moment towards $G$, never between two points that are both away from $G$.
Computing the inertia properties of a rigid body ([Lecture 8, Slide 17])
The mass, centre of mass and tensor of inertia need triple integrals. Look for symmetries first: they place $G$ and cancel products of inertia. Divide complex geometries into simple parts and add their contributions, about the same point and in the same basis; treat holes by subtraction. Write the volume element in the coordinates that fit the body (Cartesian, cylindrical or spherical, Eq. 10.5) with explicit limits for every coordinate. Take the tensor about $G$, move it to any other point with Steiner's theorem, rotate it with the change of basis matrix, and diagonalise it to get the principal axes and moments.