I have a closing door, that moves from angle 75 to angle 5. Torque acting on that door at each angle is known -of course differs at each angle- I want to calculate the time required for that door to close, knowing that it starts with angular velocity $\omega(t=0) = 0$ and angular acceleration $\alpha(t=0) = 0$ at angle 75°.

I tried to calculate angular acceleration at each angle as: $$\alpha = \tau / I,$$ where $I$ is the moment of inertia of the door.

Acceleration is different at each angle. From instantaneous acceleration and knowing the distance between each calculation positions ($\Delta angle = (\pi/180)$) I could get angular speed at each position as: $$\omega_\text{final}^2 = \omega_\text{initial}^2+2\,\Delta (\text{angle})\,\alpha_\text{average}$$

Then I calculate ($\Delta time$) for each as: $$(\Delta t) = (\omega_\text{final}-\omega_\text{initial})/\alpha_\text{average}.$$

Then I get the summation of all ($\Delta t$) to get the total time required for the door to close from 75 to 5 degrees but it was completely different from the actual measured time :(.

Is there any thing wrong about that calculation procedure? What is the right way to calculate time for such problems?

  • $\begingroup$ If the applied force was different than what was calculated, it could cause the difference. And how different is "totally different" that you measured? $\endgroup$
    – Kyle Kanos
    Jan 15, 2015 at 13:25
  • $\begingroup$ no , it is the same applied force , cuz i'm validating my results and comparing them with the valid model , time was 0.68 in the valid model , but in my model it was 0.079 seconds !! $\endgroup$ Jan 15, 2015 at 14:16
  • $\begingroup$ Based on your starting conditions, $\alpha$ must become non-zero at $\ang{75}$ at some time, t>0. Do you know what that time is? Otherwise, the door never moves. $\endgroup$
    – Bill N
    Jan 15, 2015 at 21:21

1 Answer 1


If you know the torque and hence the acceleration as a function of position then

$$ \frac{{\rm d}^2\theta}{{\rm d}t^2} =\frac{{\rm d}\omega}{{\rm d}t} = I^{-1} \tau(\theta) $$

$$ =\frac{{\rm d}\omega}{{\rm d}\theta} \omega = I^{-1} \tau(\theta)$$

$$ \int \omega \,{\rm d} \omega = \int I^{-1} \tau(\theta) \, {\rm d} \theta $$

$$ \frac{1}{2} \omega^2 +C = \int I^{-1} \tau(\theta) \, {\rm d} \theta $$

Now you have the calculated speed as a function of angle given some initial conditions and a numerical integral

$$ \omega(\theta) =\sqrt{2 \left( \ldots \right) } $$

A second integral is needed to get to the time needed to reach a certain speed.

$$ t = \int \frac{1}{\omega(\theta)}\,{\rm d} \theta $$


So if at one instant you measure/calculate the rotational speed and acceleration as $\omega_1$ and $\alpha_1$ then you can estimate the speed at the next step $\omega_2 = \omega_1 + \Delta \omega$ if you know the acceleration change $\alpha_2 = \alpha_1 + \Delta \alpha$. By assuming linear acceleration curve between the points you arrive at

$$ \frac{1}{2} \omega^2 - \frac{1}{2} \omega_1^2 = \int \limits_{\theta_1}^\theta \alpha(\theta)\,{\rm d}\theta $$ which simplifies to

$$ \frac{1}{2} \omega^2 - \frac{1}{2} \omega_1^2 = \alpha_1 (\theta-\theta_1) + \frac{\Delta \alpha}{2 \Delta \theta} (\theta-\theta_1)^2 $$

$$ \frac{\Delta \omega}{2} (2 \omega_1+\Delta \omega) = \alpha_1 \Delta \theta + \frac{\Delta \alpha}{2} \Delta \theta $$

$$ \boxed{ \Delta \omega = \sqrt{\omega_1^2+\Delta \theta (2 \alpha_1 + \Delta \alpha)} -\omega_1 }$$

This can also be expressed as

$$ \omega(\theta) = \sqrt{ \omega_1^2 + 2 \alpha_1 (\theta-\theta_1) + \frac{(\theta-\theta_1)^2 \Delta \alpha}{\Delta \theta}} $$

The time for the step $\omega_1 \rightarrow \omega_1 + \Delta \omega$ is estimated by

$$ \Delta t = \int \limits_{\theta_1}^{\theta} \frac{1}{\omega(\theta)}\,{\rm d}\theta $$

$$ \Delta t = \sqrt{\frac{\Delta \theta}{\Delta \alpha}} \ln \left(1+ \frac{\sqrt{\Delta \alpha}(\Delta \omega+\sqrt{\Delta \alpha \Delta \theta}}{\alpha_1 \sqrt{\Delta \theta}+\omega_1 \sqrt{\Delta \alpha}} \right) $$

The above is simplified a bit if you consider the time constant $T = \sqrt{\frac{\Delta \theta}{\Delta \alpha}}$. When $\Delta \alpha \neq 0$ and $T \neq \infty$ then $$ \boxed{ \Delta t = T \ln \left(1+ \frac{\Delta \theta + T \Delta \omega}{T (\omega_1 + T \alpha_1)} \right) }$$

Otherwise when $\Delta \alpha=0$ and $T=\infty$ $$\boxed{\Delta t = \frac{\Delta \omega}{\alpha_1}}$$

  • $\begingroup$ unfortunately , torque is not a function of position , it is measured at each position individually :( $\endgroup$ Jan 15, 2015 at 18:12
  • $\begingroup$ As long as you have torque data as function of position then the above applies. Just replace the integrals with discrete sums. $\endgroup$ Jan 15, 2015 at 20:25
  • $\begingroup$ so , this procedure is not valid :( $\endgroup$ Jan 17, 2015 at 14:44
  • $\begingroup$ I think it is. Torque is measured at each position and thus is treated like function. $\endgroup$ Jan 17, 2015 at 22:19
  • $\begingroup$ See edits above for more details on how to do this with a time series. $\endgroup$ Jan 17, 2015 at 23:05

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.