Adams acceleration magnitude

Hello,

I am simulating an experimental gearbox in Adams. When complaring the gearbox vibrations (accelerations) between the experimental measurements and the Adams requests, there is a huge difference in magnitudes. The units in Adams are in mm, but I always make sure to turn them into m so as to compare them to the experimental measurements. I am using the default Adams solver settings, except from the fact that I use multithreading as PREFERENCES/, NTHREADS = 4.

I have also read in the documentation that the I3 integrator (which is the default|) may not calculate the accelerations properly, but this great a difference in magnitude shouldn't be related to that.

Are there any other things I should look into that might help mitigate my problem?

P.S.

I am using flexible bodies (the gearbox included) in order to allow for some flexibility in the model.

Thanks,

Josef

0 Alexis Del Rio over 3 years ago

Silly question, when you changed from mm to m, did you update the error to accomodate?
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Josef K. over 3 years ago

I am not sure what you mean, but I might have not stated my question well. When I say that I changed the units from mm to m, I mean after exporting the request files. To clarify, the units in Adams are in mm, I run the simulation and get a request file containing the accelerations, which I then import to a different sofware for post processing (units change and such)

Sorry if my initial question caused confusion

Josef
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Jesper Slattengren over 3 years ago

Default settings for the integrator, specifically if you are looking at accelerations are not sufficient.

Before doing anything else, you need to iterate to good settings of error and HMAX. First keep decreasing error until displacements converge, then decrease hmax until accelerations converge. You'll be surprised how big difference this make.

Even better is to start with SI2 if you're interested in accelerations. You still need to go through the iterations to make sure the solution accuracy has converged, but it should be easier with SI2.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Josef K. over 3 years ago

Hello,

Unfortunately the SI2 formulation requires huge amounts of time for this model (using SI2 instead of I3, without limiting the max iterations of SI2 took 3 days of simulation instead of 15 minutes).

I do want to observe convergance but I do not know where from. Is there a visulization tool in Adams? How will I know if my displacements/accelerations have converged?

Thanks,

Josef
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Jesper Slattengren over 3 years ago
There is a reason why SI2 takes so long time, it makes more accurate simulations.

But back to I3 (or try HHT, particularly if there is contacts involved).

Run a simulation with default values (in A/Car ERROR=0.01 and HMAX=output step).

Decrease to ERROR = 0.001 (which is Adams default). Run with a new name.

Plot some typical displacements: angle of a wheel or nodal displacement of you have flex parts.

Continue decreasing ERROR until two consecutive runs have acceptable difference (here is where engineering judgement comes in). Then select the larger of those two values for error.

Save that run (with the larger error), delete all other runs from the session (makes it easier to maintain.

Set HMAX=output_step/10 and run a new run.

Compare a typical acceleration between the two runs.

Repeat until convergence.

This is a procedure that we always used to teach new users back in the 90's (when I was responsible for trainings and support). But I guess this is not included in training material or documentation anymore.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Josef K. over 3 years ago

Thank you, this is immensly helpful really!

Regards,

Josef
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Alexis Del Rio over 3 years ago

Playing with the ERROR/HMAX... was part of the ADM703 training I attended 4 years ago
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Josef K. over 3 years ago

Silly question, but I assumed that the units Adams exports data in are the same as the model units, correct?
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Jesper Slattengren over 3 years ago

Yes, they are. Except for angles. Adams solver always works in radians.

You can bypass this by creating your own requests of course and convert to whatever you need.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel