Adams, SimMechanics, MapleSim 1

[FeX32]

Hello,

This is a general discussion question for everyone. I had a discussion the other day about multi-body dynamics software packages with a colleague. The 3 main packages in competition are Adams, SimMechanics, and MapleSim.
Presume, an R&D company that does work in robotics and mechatronics needs to speed up the dynamic optimization process of their new products. (From manual programming, to a package)
Based on experience, which software package is the most appropriate?




Fe

 

[GregLocock]

Are you trying to simulate the dynamic performance of the mechanism, or get the programming/control side right? ADAMS could do both, but is probably better at the first than the second.

Sorry I haven't used either of the other two, another alternative would be matlab.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[FeX32]

Thanks Greg,

Yes I agree that Matlab is a good alternative. Although, this is what I usually do. I derive the dynamical relations manually and then program it into Matlab usually using one of its numerical solvers to solve it.
The reason for investigating the software packages is due to some new robots complexity makes deriving dynamical equations by hand very time consuming (if they can be done at all).
On example is the stewart platform. There are others that are more complex.
Oh, and it would involve dynamic performance and control. So you are saying that Adams is not very friendly for the controls side. Thanks.



Fe

 

[GregLocock]

A stewart platform is a bit of an odd example since it is fairly easy to calculate in matlab.

I'd have no hesitation about an adams model of one, it would work fine- I'd guess a day or so to set one up from scratch. The nice thing is that including compliances and friction are easy.

ADAMS does have a control method, in that you can write a dll that is called at every time step by the ADAMS solver, and this returns variables to the Solver. So for example we run an ADAMS model of a car, which then talks to a real ABS module, which tells the ADAMS model what brake pressure to use on each brake in the model. Alternatively we can use a black box program that is supplied to us to replace the physical component.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[FeX32]

Thanks for your comments.

Yes you are right. Stewart platform is not the best example. The inverse dynamics has been readily solved using the principle of virtual work. The forward dynamic model does exist too. There are many other mechanisms that fall into my category mentioned though.
Regardless, it would take less time to model say a stewart platform in Adams than to write the code by hand (even for an experienced coder).
You have touched on another point. The friction. 99% of the time to derive these relations we must assume friction is null, and assume a rigid body model.
Does Adams have a flexible link model? I just found out SimMechanics does.


Fe

 

[GregLocock]

You can model compliances as discrete 6 dof (non linear) springs, or as non linear beam element FEA, or as flexible bodies (ie use an FEA program to build a compliance matrix). So in a suspension we use those for bushings, stabiliser bars, and the subframe or body, respectively.

Or again, you can write an external subroutine in c or fortran or matlab and hook into it. That's how proprietary tire models work, for example.









Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[FeX32]

Understandable. Thanks Greg.


Fe

 

[GregLocock]

A couple of other suggestions- Altair MotionSolve and MSC Nastran 4D.

I'm pretty sure all these vendors will give you a month long trial - it might be as simpe as asking them if they've ever done robotics sims.



I'd also be surprised if LMS didn't have something suitable.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[FeX32]

Thanks. I never heard of Altair.
I will look into them. I do have trial versions of SimMechanics and MapleSim.
I don't have one yet for Adams.
So far I prefer SimMechanics over MapleSim. Seems its more convient to extract inertial properties during simulation in order to optimize the dynamic manipulability.

I will also look at Nastran 4D. Thanks for your help.




Fe