Simscape model, not matching theoretical one.

Hello,

I have a problem to replicate on Simscape, a theoretical model. By comparing the bode diagram of the theoretical model (left) and the simscape model (right), we can see that the simscape model has a strange behavior in high frequencies.

The output is defined as y(t) = theta(t) + alpha(t)

I suspect the problem comes from the "Revolute1" that I use as a rotational spring in the Multibody part , but I'm not sure and I don't know how I should modify it.

I put in the image the parameters of Revolute1 and the last solid and the mechatronic part, just in case the problem come from there.

Answer 1

[Vasco Lenzi]

Hi Kamyar

as far as I can tell you are not closing the kinematic loop between multibody and the electric motor, so the rotational spring in the revolut joint is not considered as load into the 1D network.

Please take a look at this example that showcase how to connect Simscape 1D and Multibody:

https://ch.mathworks.com/matlabcentral/fileexchange/37636-simscape-multibody-multiphysics-library

Usually the kinematic loop is closed like this to give you an idea:

[Kamyar Taher]

Hi Vasco, thank you for your response,

I tried first to use the interface of your example, but I systematically had this error.

So I tried to use this other interface that is used in the video you send me and that seems to be equivalent. I have lot more coherent result, but there is still a little mistake and I can't figure what it could be? Is there a probleme of using inertia in mechanical part and in Multibody like this?

[Vasco Lenzi]

Hi Kamyar

I forgot to mention that when using 1D to 3D connection, you should have all the inertia on Multibody side if possible. This makes it a bit more challenging to model, because you need to put the gearbox on the Multibody side, and add to the model two gears with the correct inertia representing the two toothed wheels. You can find some gears to use in this example:

https://www.mathworks.com/help/releases/R2020b/physmod/sm/ug/using-the-common-gear-block-cardan-gear-mechanism.html

or on the Simscape Multibody Parts Library, that I suggest you to download:

https://ch.mathworks.com/matlabcentral/fileexchange/36536-simscape-multibody-parts-library

Hope this helps

Vasco

[Kamyar Taher]

Hi Vasco,

As you suggested, I added the gearbox to the multibody side.

For more detail on this system: I put all masses, product of inertia etc.. to zero and kept only the inertia around Z axes, for "spur_gear_iso_6_RIGID", "Module_inertie" and "tige_vis_RIGID2". There's a gear ratio of 70, and a damping and a stiffness in Revolute4. (edit: I inverted the name of angular speed and Motor input, but they are connected on the right link)

Unfortunately, the same strange behavior still occurs, this is the bode diagram with y= theta + alpha as output, I tried to inverse the sign of alpha (in blue), but it's still not correct. And unfortunately, I can't find an example projet with a rotational spring on multibody. Any idea from where this problem come from?

And here is what happen if i remove the Revolute4 and "tige" in Blue.

In red I removed everything expect Module_inertie, Revolute4 and "tige" and I actuated Revolute4 in this case. So separately I think they behaving correctly.

Best,

Kamyar

[Vasco Lenzi]

Modeling the torsional spring in the Jint Revolute is correct, and it does seems to work correctly if isolated.

Are you linearizing at the right equilibrium point? It may be a problem on how the linearization is setup, eg if you are linearizing at the wrong time you capture the wrong dynamics.

[Kamyar Taher]

I tried modeling the spring and damper like this and I get the correct behavior now. I think I was misinterpreting how the damper and spring were acting. Thank you for your help!

Kamyar