More problems with the smoothness of the Torque plot results

Carlos Fondevila

Hello again.

I continue having problems when I plot the torque of a simple model than a linear screw.

The model (attached) is simple I have a 200kg mass with a linear movement (STEP5) actuated by a rotary motor that transform the rotationl movement in linear.

When I plot the torque of the motor, I have the next:

 

With a RMS of torque of 4.956Nm, or a Maximum value of 49.787Nm which is totally crazy. If I make a zoom I can see slightly the real form of the torque applied, but this is not acceptable:

If I play with the maximum step size, and the Integrator tolerance, I can improve that, but is unaffordable to play with this values in each simulation, or each time that I change the mass o another parameters, and the ressults that I can get are not pettry good. If I do the study in Inspire, I don't have this type of problems.

Please, someone have a solution for this?

Christopher_Fadanelli

Hi Carlos,

 

I ran the model attached and my torque results are pretty smooth. 

 

 

Reattaching the model. I'm trying to think what could be the issue. Is the model saved in a non-local directory (onedrive for example)? Can you try loading the model I attached to this post and try again? Could you run the Cleanup batch file located in your installation directory. "C:\Program Files\Altair\2024\hwdesktop\utility\HWCleanup"?

Best,

 

Christopher Fadanelli

Chris Coker_20903

I took a quick look at the original model, just for fun.  One thing sticks out to me:  the mass of the "box" is 200kg, and the inertias are all 5 kg-mm^2.  I don't know if this is still the case with MotionSolve, but it used to be rather picky about having "reasonable" inertia values relative to the mass, based on the unique way it writes out and solves the equations of motion.  You might try bumping those inertia values up to 1e6, or 1e7 and see if that has any change on the robustness of the results.

One other thought when looking at the .log file.  The analysis is being run as a kinematic analysis, since it has zero degrees of freedom.  It's usually difficult to get trustworthy FORCES out of a purely kinematic solution.  One trick to add DOF to a kinematic model is to add a separate, unconstrained mass to the model.  The solver is then "forced" to solve the problem as a dynamic analysis.  I haven't used this approach often, but it might be useful in this specific case.

Carlos Fondevila

Christopher Fadanelli_22565 said:

Hi Carlos,

 

I ran the model attached and my torque results are pretty smooth. 

 

 

Reattaching the model. I'm trying to think what could be the issue. Is the model saved in a non-local directory (onedrive for example)? Can you try loading the model I attached to this post and try again? Could you run the Cleanup batch file located in your installation directory. "C:\Program Files\Altair\2024\hwdesktop\utility\HWCleanup"?

Best,

 

Christopher Fadanelli

Hi Christopher,

I tryed to run your model, in a local directory and the results are the same. 

I triyed also to run your model in another computer (a less powrful laptop) and the plot is correct!

I shoul'd say that in the computer where the model plots non correctly I work with the last version of Hpermesh and Motioview, in the laptop, where the model runs correctly I have the Hyperworks (the old name) 2022.3.

I want to update the version of Hypermesh in the lapton and see if the problem comes from the software version.

Which version do you have installed?

I tryed also restarting the Preferences of Motionview and the problem is continuing.

 

Carlos Fondevila

Carlos Fondevila said:

Hi Christopher,

I tryed to run your model, in a local directory and the results are the same. 

I triyed also to run your model in another computer (a less powrful laptop) and the plot is correct!

I shoul'd say that in the computer where the model plots non correctly I work with the last version of Hpermesh and Motioview, in the laptop, where the model runs correctly I have the Hyperworks (the old name) 2022.3.

I want to update the version of Hypermesh in the lapton and see if the problem comes from the software version.

Which version do you have installed?

I tryed also restarting the Preferences of Motionview and the problem is continuing.

 

Yes, I installed the last version of Hypermesh and the mechanical solvers, and I have the same problem with my second computer now.

I think that it's a problem of the nex versions (I tryed with 2024 and 2024.1 with the same results)...

Carlos Fondevila

Christopher Fadanelli_22565 said:

Hi Carlos,

 

I ran the model attached and my torque results are pretty smooth. 

 

 

Reattaching the model. I'm trying to think what could be the issue. Is the model saved in a non-local directory (onedrive for example)? Can you try loading the model I attached to this post and try again? Could you run the Cleanup batch file located in your installation directory. "C:\Program Files\Altair\2024\hwdesktop\utility\HWCleanup"?

Best,

 

Christopher Fadanelli

I did some trials, and I advice that the problem occurs with the 2024 and 2024.1 versions.

The 2023.1 plots the torque perfectly. May be I should to notificate this to Altair?

Thanks!

Carlos Fondevila

Chris Coker_20903 said:

I took a quick look at the original model, just for fun.  One thing sticks out to me:  the mass of the "box" is 200kg, and the inertias are all 5 kg-mm^2.  I don't know if this is still the case with MotionSolve, but it used to be rather picky about having "reasonable" inertia values relative to the mass, based on the unique way it writes out and solves the equations of motion.  You might try bumping those inertia values up to 1e6, or 1e7 and see if that has any change on the robustness of the results.

One other thought when looking at the .log file.  The analysis is being run as a kinematic analysis, since it has zero degrees of freedom.  It's usually difficult to get trustworthy FORCES out of a purely kinematic solution.  One trick to add DOF to a kinematic model is to add a separate, unconstrained mass to the model.  The solver is then "forced" to solve the problem as a dynamic analysis.  I haven't used this approach often, but it might be useful in this specific case.

Hi Chris,

Interesting reflexion, in this case I created a box, and I change its mass directly, without modify the inertia, the correct methodology is to adapt all the mechanical properties to avoid malfunction or inestability in the calculus.

Thanks!

Christopher_Fadanelli

Carlos Fondevila said:

I did some trials, and I advice that the problem occurs with the 2024 and 2024.1 versions.

The 2023.1 plots the torque perfectly. May be I should to notificate this to Altair?

Thanks!

Hi Carlos,

That is very interesting. I used 2024.0. 

I just tried 2024.1 and I noticed the issue when the solver was set to python, when it is set to xml the results are correct. Which solver did you try? In addition, did you have the run offline option enabled? I did for mine. I will check with the developers and see if they can reproduce the same issue. 

Best,

Christopher Fadanelli

 

 

Christopher_Fadanelli

Christopher Fadanelli_22565 said:

Hi Carlos,

That is very interesting. I used 2024.0. 

I just tried 2024.1 and I noticed the issue when the solver was set to python, when it is set to xml the results are correct. Which solver did you try? In addition, did you have the run offline option enabled? I did for mine. I will check with the developers and see if they can reproduce the same issue. 

Best,

Christopher Fadanelli

 

 

@Carlos Fondevila ,

 

Regarding this issue. Could you please have run offline enabled and solver type set to XML.

Thanks,

Christopher Fadanelli