Driven torsional oscillator equation setup

5 replies · 2K views · last reply
zenterix
Messages
774
Reaction score
84
Homework Statement
A torsional oscillator comprises a cylinder with moment of inertia ##I## hanging from a light rod with torsional spring constant ##k##. The cylinder also experiences a drag torque equal to ##-\mu\dot{\theta}## when moving with angular velocity ##\dot{\theta}##. The top of the rod is driven with angular displacement ##\phi(t)=\phi_0\cos{\omega t}##.

Find the steady-state solution for ##\theta(t)##.

Plot the amplitude ##A(\omega)## and phase ##\delta\omega## of your solution for ##\theta(t)## in (1) as a function of ##\omega##.
Relevant Equations
For your plot assume that the natural frequency of oscillation of the system is ##\omega_0=1## and plot three curves on the same plot with ##\frac{\mu}{l}=0.25, 1##, and 2. Label your curves to distinguish the three cases.
The "Vibrations and Waves" problem-solving course on MIT OCW has a section on driven harmonic oscillators which can be seen here.

I would like to do the first of the two problems. Unfortunately, there are two issues

1) The latex is not rendering on that website (relatively minor issue, I think I can get parse it).

2) The problem is about a type of oscillator that I am unfamiliar with.

I'd like to show here how I understood the problem to be set up and then how I set up the equations.

My question is if this set up and equations are correct.

I've stated the problem above as I understood it from the website.

Here is how I set it up mathematically.

Diagram of a torsional oscillator: a horizontal bar at the top with a red circular arrow and the label 'simple point rotating' and angle phi, a vertical wire hanging from it to a flat disk below with a green angle theta marked from the center.


$$\vec{\tau}=(-k(\theta-\phi)-\mu\dot{\theta})\hat{k}=I\ddot{\theta}\hat{k}\tag{1}$$

$$I\ddot{\theta}+\mu\dot{\theta}+k\theta=k\phi\tag{2}$$

$$\ddot{\theta}+\frac{\mu}{I}\dot{\theta}+\frac{k}{I}\theta=\frac{k}{I}\phi_0\cos{\omega t}\tag{3}$$

In words, the torque experienced by the cylinder has two components. One is due to the angle difference between the cylinder and the rod relative to the equilibrium position, in which this angle difference is zero. This is the ##-k(\theta-\phi)## term.

The other component is drag which is always in the opposite direction to the angular velocity.

If this is correct, this boils down to the same problem as a driven simple pendulum or a driven RLC circuit.
 
Assuming the problem is set up correctly, a solution is

$$\theta_p(t)=A(\omega)\cos{(\omega t-\alpha)}\tag{4}$$

$$\alpha=\text{arg}(\omega_0^2-\omega^2+i\gamma\omega)\tag{5}$$

$$=\text{arg}\left ( \frac{k}{I}-\omega^2+i\frac{\mu\omega}{I} \right )\tag{6}$$

$$A(\omega)=\frac{f}{\sqrt{(\omega_0^2-\omega^2)^2+\gamma^2\omega^2}}\tag{7}$$

$$=\frac{\frac{k\phi_0}{I}}{\sqrt{\left ( \frac{k}{I}-\omega^2 \right )^2+\frac{\mu^2}{I^2}\omega^2}}\tag{8}$$
 
zenterix said:
If this is correct, this boils down to the same problem as a driven simple pendulum or a driven RLC circuit.
Precisely. Only the symbols are different.
 
The problem then asks us to plot amplitude and phase for different values of ##\gamma=\frac{\mu}{I}##.

##\gamma## represents the amount of damping in the system.

We should expect that as we increase ##\gamma## the system at some point becomes overdamped.

Scenario 1: ##\frac{\mu}{I}=0.25##

Amplitude

Plot of amplitude against omega from 0 to 5 for a driven oscillator: the curve starts near 1 at omega 0, peaks near 4 at omega 1, then decays toward 0.

Phase
Plot of a phase curve against ω from 0 to 5: the vertical axis runs 0 to 3, the curve rises slowly, climbs steeply near ω = 1, then levels off near 3.



Scenario 2: ##\frac{\mu}{I}=1##

Amplitude

Graph of a normalized amplitude curve against ω from 0 to 5, starting at 1.0, rising to a peak just above 1.0 near ω of 0.7, then falling steeply and flattening toward 0 by ω of 5.

Phase
Graph of a curve rising from 0 at omega = 0, steepest near omega = 1, and leveling off near 2.9 by omega = 5; horizontal axis labelled omega, vertical axis unlabelled.


Scenario 3: ##\frac{\mu}{I}=2##

Amplitude
Graph of a decreasing curve from 1.0 at ω = 0 toward about 0.04 at ω = 5, with the horizontal axis labelled ω and the vertical axis unlabelled.

Phase
Line graph of a rising curve, plotted from x = 0 to 5 on the omega axis and 0 to about 2.75 on the vertical axis; it starts at the origin, rises steeply, then levels off toward about 2.75.
 
So how do we interpret these results?

The natural frequency of the system is ##\omega_0=\frac{k}{I}=1##. The smaller the damping term ##\gamma## is, the more significant is the phenomenon of resonance which we observe as the peak in amplitude.

Note that this occurs at a driving frequency smaller than the natural frequency

$$\omega_{max}=\sqrt{\omega_0^2-\frac{\gamma^2}{2}}$$

When ##\gamma=2## this max occurs at ##\omega=0##.

What about the phase?

The higher the driving angular frequency, the more out of phase the cylinder.

The limiting phase is ##\pi##. In the plots from my previous post, this isn't clear because I only plotted to ##\omega=5##, but all three have ##pi## as their asymptote.

The speed at which they reach this limiting phase is different.

Here are some plots of all three phases together

Graph of phase angle against driving angular frequency omega from 0 to 5, with three curves starting at 0 and rising toward about 3 (pi), all crossing at omega = 1 and phase about pi/2.

Graph of phase angle against driving angular frequency omega from 0 to 20, with three curves for different damping values rising from 0 toward about 3.1 and levelling off.

Graph of phase angle against driving angular frequency omega from 0 to 60, with three curves (red, blue, green) rising steeply near omega 0 to 1 and approaching about 3.14 (pi) as omega grows; the red curve reaches the plateau fastest, green slowest.


Since

$$\alpha=\text{arg}(\omega_0^2-\omega^2+i\gamma\omega)=\tan^{-1}{\frac{\gamma\omega}{\omega_0^2-\omega^2}}$$

Then when ##\omega=\omega_0## the phase is always ##\pi/2##.

Here we see what happens to the imaginary part

Graph of three curves against ω from 0 to 2 with the vertical axis from -6 to 5, each with a vertical asymptote near ω = 1, rising steeply on the left and falling on the right.


For ##\omega<\omega_0## we can see that we're in the first quadrant of the complex plane and a higher ##\gamma## means the imaginary part grows faster. ##\alpha## is initially increases relatively faster as a function of ##\omega## and is always above the ##\alpha## for a smaller ##\gamma##.

For ##\omega>\omega_0## we are in the second quadrant of the complex plane and as we increase ##\omega## a higher ##\gamma## means the imaginary part is always more negative than the imaginary part for a smaller ##\gamma##. Since the real part is the independent of ##\gamma## this means that the argument is always smaller for the larger ##\gamma##.

But this is just math. What about the physics?
 
zenterix said:
What about the physics?
Perhaps some of the physics pops up when you consider limiting cases
In electronics and in control engineering one often looks at log-log plots

##\ ##