I am trying to solve the torque needed to rotate a rectangular plate of sides $a$ and $b$, about a diagonal with constant angular velocity $\omega$.

Euler equations are given by,

$$ I_1\dot{\omega}_1 + (I_2 - I_3)\omega_2\omega_3 = N_1, $$

$$ I_2\dot{\omega}_2 + (I_3 - I_1)\omega_3\omega_1 = N_2, $$

$$ I_3\dot{\omega}_3 + (I_1 - I_2)\omega_1\omega_2 = N_3, $$

where $I_1$, $I_2$ and $I_3$ are the principal moments of inertia of the rigid body, $\omega_1$, $\omega_2$ and $\omega_3$ are the angular velocities around the axes of these moments of inertia, and $N_i$ denotes the external torque applied along the axis of $\omega_i$ and $i$ = 1,2,3.


For this problem, suppose $a > b$. $$I_1 = \frac{M a^2}{12}, \quad I_2 = \frac{M b^2}{12}, \quad I_3 = \frac{M (a^2 + b^2)}{12}$$

$$\omega_1 = \frac{\omega b}{\sqrt{a^2+b^2}}, \quad \omega_2 = \frac{\omega a}{\sqrt{a^2+b^2}}, \quad \omega_3 = 0$$

We find from Euler equations that, $$N_1 = 0 = N_2, \text{ while } N_3 = \frac{Mab\omega^2}{12 (a^2+b^2)}(a^2 - b^2) $$.

But this implies if $a = b$, the torque needed is zero. How should we find this intuitive that it requires zero torque for a square hinged at opposite corners?

  • 2
    $\begingroup$ Is external torque required to rotate a body around a principal axis? $\endgroup$
    – eranreches
    Commented Nov 18, 2018 at 18:50
  • $\begingroup$ Right. We don't need an external torque to rotate the body around any of the principal axes. So, is it just that for a square, the inertia tensor is a diagonal matrix and for a rectangle, we should be considering non-zero $I_{xy}$ and $I_{yx}$? $\endgroup$ Commented Nov 18, 2018 at 18:54
  • $\begingroup$ Even in the rectangle case $I$ is diagonal. The non-zero torque is related to the fact that $\vec{L}$ and $\vec{\omega}$ don't point in the same direction. Therefore, $\vec{L}$ precesse around $\vec{\omega}$. $\endgroup$
    – eranreches
    Commented Nov 18, 2018 at 19:11
  • $\begingroup$ @eranreches Okay, now I get it. This torque is arising from $\vec{\omega} \times \vec{L}$ since they are not pointing in the same direction. And in the case of a square lamina, these vectors are parallel. $\endgroup$ Commented Nov 19, 2018 at 8:30
  • $\begingroup$ Exactly. See my full answer below. $\endgroup$
    – eranreches
    Commented Nov 19, 2018 at 12:13

1 Answer 1


Let me expand my comments. In the Lab frame you can write

$$\dfrac{{\rm d}\vec{L}}{{\rm d}t}=\vec{\omega}\times\vec{L}$$

where $\vec{L}$ is the angular momentum and $\vec{\omega}$ is the angular velocity. Now since $\vec{L}=\hat{I}\vec{\omega}$ and $\vec{\tau}=\dfrac{{\rm d}\vec{L}}{{\rm d}t}$, one has


Therefore, the torque vanishes iff $\vec{\omega}\parallel\hat{I}\vec{\omega}$, in other words iff there exists $\lambda$ such that


i.e. the rotation is about one of the principal axes.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.