Simulation 3d

[rf96]

Hi,

I would like to simulate this device:



Is it possible ? and is it difficult ?

Thanks

[benelot]

Please describe in more detail what kind of device this is. It is impossible to read any of the writings in the picture and I can not see what device the picture really shows. Is it some sort of pendulum?

[rf96]

Hi,

You couldn't click on the image for increase the size ?

It's not a pendulum. The black vertical axis is fixed. The red arm turns around the black axis, the angle is fixed. The red arm can only turn around the black axis. The orange disk can turn around itself, only.

I would like to know if a fixed point on the orange disk moves up/down when the red arm turns with no angular velocity for the orange disk. It's difficult to test in reality because there is friction. I can test at home but with an acceleration not with a fixed angular velocity for the red arm.

[benelot]

Sorry, I did not see that it is possible to enlarge the image, I was on the phone. Also I would not have been familiar with the notion, so thanks for the explanation. I think it is definitely possible to simulate this device, but I am not sure if the effect you want to see is simulated in Bullet, so you might see no movement when in reality it would move. Look at the Bullet Physics examples/ConstraintDemo on how to setup the constraints. You should be able to figure it out easily, there is a motorized one and a lot of different constraint types. Also give your thread a more descriptive title to make it more attractive to answer you. You can change it by editing your first post's subject.

[Basroil]

Hi,

You couldn't click on the image for increase the size ?

It's not a pendulum. The black vertical axis is fixed. The red arm turns around the black axis, the angle is fixed. The red arm can only turn around the black axis. The orange disk can turn around itself, only.

I would like to know if a fixed point on the orange disk moves up/down when the red arm turns with no angular velocity for the orange disk. It's difficult to test in reality because there is friction. I can test at home but with an acceleration not with a fixed angular velocity for the red arm.

In other words, it's a regulator?

It's certainly possible, though accuracy is another issue. It might be faster and more accurate to model the system directly, unless collisions with other objects is a concern.

[rf96]

No it is not a regulator. The red axis keep always the same angle, never move up or down.

There is no collision detection. Only a simple physics device. I don't know how to model it in physics with equation it's for that I would like to simulate.

[drleviathan]

Yes, Bullet could simulate that device. Since you'd be pursuing numerical accuracy you'd want to build Bullet with support for double precision to reduce floating point errors. The results may be sensitive to the details of how you set up the constraints. For example, the drawing shows the wheel's pivot connecting at the surface of the disk, not at the disk's center of mass -- this configuration would put extra torque that would work against the constraint which might affect the behavior of the simulation. You could imagine a scenario where the pivot is actually in the center of the disk with minimal torque and the Bullet's simulation of such a configuration might be different.

My physics intuition tells me that if you setup real device with a low friction bearing in the disk then yes it would spin under certain circumstances. In particular, if the device experienced a torque that accelerated its rotation about the vertical axis then the orange disk would gently spin: the outside edge would migrate against the torque and the inside edge would migrate with it. That is, if the red arm were spun up with upward torque (counter clockwise by right-hand-rule) then the spin of the disk would point downward parallel to the axis of the disk. In the steady state scenario where the angular velocity about the vertical axis is constant there will be no torque on the disk about the red axis -- the contraption must be accelerating in order for the disk to experience torque about its axis of symmetry.

The magnitude of the symmetric torque on the disk depends on the mass distribution, dimensions, and angles of the various parts and also the instantaneous torque about the vertical axis. The symmetric torque spinning the disk would be relatively weak compared to the other stresses on the device. Hence for a real (non-idealized) device the disk would quickly stop spinning about its axis.

There are a couple ways to reason through the direction of the disk's spin:

(1) Conservation of momentum: When the disk is at rest angular momentum is zero. When the disk is spun up (positive) by the red armature then the disk system would tend to conserve angular momentum by spinning the opposite direction. Since it can only spin about its axis, its reactive spin would point down (negative) along its axis.

(2) Simplification: Simplify the system such that the disk is two equal point masses connected to each other by a massless wire that is connected to the red axis at its center. Let one point be on the outermost edge of the disk (farthest from the vertical axis) and the other on the inside edge (closest to vertical axis). As the contraption is accelerated about the vertical axis the outer point must travel a larger circumference and must therefore experience a larger linear force to achieve the same orbit period. Since the perpendicular force produced by the red axis on the points is equal for both points there is a discrepancy of torques: the outer point will get too little and the inner will get too much so the outer point will lag behind and the inner will push forward about the orbit.

This problem could be solved analytically (for the idealized case: zero friction, thin disk, etc) on the back of an envelope using the principle of conservation of momentum (1). It could also be solved discretely by numerically integrating (with a computer) over the linear torque discrepancies of the disk volume (2).

[rf96]

Thanks for your help

With no friction. Someone said a point A fixed on the orange disk will move up/down like that:

direct link: http://s5.postimg.org/4cop9kg5z/image.png

Even the angular velocity of the red arm is constant the point A moves up/down.

direct link: http://s5.postimg.org/aewc024lz/image.png

I don't understood all your simplification(2).

If the point A fixed on the disk can move up when the angular velocity or the red arm is constant and if I apply a force F1 to the disk (a small force compared to the inertia of the disk), the red arm receives the force F2 and the disk will receive the force F3. The disk will receive a torque from F1/F3 and will increase its kinetic energy around itself. The red arm receives the force F2 but this can't give a torque to the red arm. If the force F1 comes from a springs for example, the spring will increase its potential energy. The sum of energy seems not to be constant.



Direct link: http://s5.postimg.org/lkqn10es7/image.png

Edit: Could you explain : "this configuration would put extra torque that would work against the constraint which might affect the behavior of the simulation. You could imagine a scenario where the pivot is actually in the center of the disk with minimal torque" Where is the extra torque ? Like that:

[drleviathan]

After thinking about this during my morning commute I decided that a conservation of momentum treatment just won't work. Although the system is driven by an external torque (momentum isn't conserved) I thought maybe the conservation rules could be applied along specific axes -- nope.

Then I looked at the Euler's equations and finally concluded that the disk just won't spontaneously spin about its axis of symmetry.

The system is constrained (by definition) and therefore torques on the bar and disk will be whatever they need to be to make the disk follow an orbit, but there won't be any torque about the disk's axis of symmetry... in the disk's reference frame. Yes, there will be torque about the axis of symmetry in the inertial frame of the stationary observer, but not in the disk's local frame.

[rf96]

I took an evaluation of Spaceclaim and I will try to simulate with it (I don't know Spaceclaim nor Bullet but a friend said Spaceclaim is easy for a beginner) I don't know if simulate is a good idea but I don't know how to modelize this device. I done a lot of experiments at home and I can look the point A moves up/down, but like there is friction the orange disk turns like the support with time. I would like to know if another torque appears to the red arm and can change the angular velocity of it. I think it's important to take a thickness different of 0 because the torque can appear with a thickness only. A guy say to me that the device is a Lagrangian non-autonomous with one degree of freedom with the constraint w0=w(t) and can be resolved with Lagrangian, I don't know this method, is it difficult to modelize it with a tickness ?

I think a video with a simulation of this device could help to have the model of the device, because if the point A moves up/down this could say the sum of energy of a closed device is not constant.

[rf96]

I tested on Ansys, the disk turns around itself even at start it don't turn and even there is no friction. The angular velocity of the disk is lower than the support so a point can move up

The video :

https://youtu.be/Pjc4dIf1aWI

[rf96]

Like I can have a point that moves up on the disk, I can add a spring :

http://s5.postimg.org/bjdkqhxav/image.png



The spring will increase its potential energy because its length increases.
The disk rotates around itself more, its kinetics energy increases.
The support don't receive a torque.

The sum of the energy of the closed device is not constant.

What do you think about that ?

[drleviathan]

I don't believe it.

[rf96]

So, in this case where is my error ?

[drleviathan]

The simple reason is that I don't believe in perpetual motion machines so I discard such notions without inspection.

However, in this specific case:

(1) Any spin of the disk about its axis, if any, could only be introduced when the system changes speed about the vertical axis -- that is, when there is an external torque on the system. Put another way: there would be no torque on the disk about its axis for the constant velocity state -- Euler's equations say this.

(2) The spring would actually counteract the rotation of the disk if it had any -- it would not push the disk faster.