Mechanism Simulations Software 3

[subsearobot]

Hi folks!
I am looking into multibody simulation packages. I am unfamiliar with the specifics of these types of packages, and after a quick google, this is the place I come to speak with the experts... because most of what i found was written by Marketing folks...

Specifically, I've been asked to look at Adams Multibody Dynamics Simulation Solution, NPD solutions Kinematic and Dynamic Analysis Overview, and Solidworks Simulation. We are an Autodesk Inventor house (which is hard to admit) so I suspect that Nastran may also be an option. Other references woudl be helpful.

The problem: We have a fairly complex device that includes several mechanisms. We have built hundreds of functional prototypes via high resolution 3d printing to refine the details for fit and function.

Before we move into DFM for injection molding, new management has asked me to review some simulation packages for feasibility as to whether they could accelerate our analysis, and thusly decrease our time to market. We have hand driven gearboxes, mechanisms and some snap features.

I have experience with FEA, but not multibody kinematic analysis. The idea up until now has been to simply utilize nonlinear FEA with Nastran.

Can anyone speak to the benefits of using an intermediate simulator before embarking on detailed FEA? It seems like it woudl slow me down, but that's only due to my ignorance at this point...

thanks, I appreciate your thoughts.

 

[rb1957]

I don't think NASTRAN has a kinematics package; possibly MARC which has been bought I believe by MSC.

There is some kinematics you can do inside Inventor.

another day in paradise, or is paradise one day closer ?

 

[subsearobot]

rb1957, interesting.
DO you have any knowledge of Marc? MSC also sells Adams, which appears to be lower capability, but similar. Any thoughts woudl be helpful (especially what I will not glean from marketing literature). (For example, I very much prefer Solidworks. not because it is "better" than Inventor, but because Inventor requires me to make about 4 clicks to get the same feature as SW in one click. all those clicks add up into a slower design package...) I imagine simulation packages inspire similar user impressions.

cheers

 

[rb1957]

how do you spell "carpal tunnel" ?

I'm surprised you think Adams has a "lower capability" ... I thought it was the benchmark s/ware for mechanisms ?

Most (all?) CAE packages come with a mechanisms capability (even Inventor). Which is better than the others, sorry don't know (as a structures guy I don't use them) but there are plenty of places online offering advice.

my advice ... if someone (sales-ish) is offering you a s/ware, and demonstrating it ... ask them to get out of the seat and drive it yourself.

another day in paradise, or is paradise one day closer ?

 

[ThomasH]

Hi

I don't think NASTRAN is suitable for kinematics. I work with NX Nastran (and have used MSC) and while simple things may be possible I would not be comfortable with the results. I have no experience from Marc but it is a nonlinear FEM-code so there is no kinematics there either,as far as I know.

My assumption would be Adams, if we are talking MSC. From what I have heard in the past Adams is good. But I have no first-hand experience.

Regardless of what you plan to buy I would start with a test. And I mean a real test, not a preprepared demo from the vendor.

Edit: I just noticed on the MSC website that thare is a package of Nastran with the label "motion". It seems to be Nastran with some type od Adams integration. Possibly something that can be of interest?

Thomas

 

[IceBreakerSours]

Finite element machinery is NOT suitable for doing rigid body kinematics. Assuming it has typical features required in such analyses, you can do rigid body kinematics simulations with any FE code that can let you take inertia into account. However, it will be a poor use of resources. This is why large commercial analysis code vendors (ANSYS, MSC, Dassault Systemes, ..) end up buying such solvers.

*********************************************************
Are you new to this forum? If so, please read these FAQs:

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083

 

[GregLocock]

"We have hand driven gearboxes, mechanisms and some snap features."

I use ADAMS most days of the week. Of the various MBD (multi body dynamics) packages I have tried it is not the easiest to learn, or the most robust, or the cheapest, or the most flexible. But it is at the sweet spot for those criteria for my application. As an example I knocked up a model of an automotive differential in about a day, but it didn't use gears, it used equations to link the speed of each gear pair. I could have modelled each gear pair instead as a contact problem, but that wasn't what I needed and run times would have been considerably extended. I could have replaced the rigid bodies used in a contact problem with linear FEA models, again the run times would have gone up. I don't know if ADAMS would directly link a non linear FEA model, there's no reason why not, but I've never done it. It might have to be a co-sim. If you can swing a demo copy of ADAMS/View then the 'latch' tutorial will give you a pretty good idea of whether the level of simulation is adequate.

I had a look at Solidworks Simulation but it looks as though they are not talking about mechanism design (linkages and so on) as much as FEA.

You may be able to use whatever simplified MBD software is embedded in your CAD, how will you know that it is wrong (it will be) but good enough? Which leads me to the main point...

Your actual problem is that the learning curve for MBD is steep, and proving that you have a valid model is often tricky. GIGO applies even more so than in linear FEA.

There are many packages out there for MBD, unfortunately it would take about a month to evaluate each one for a given application. Of the freebies Project Chrono, Askoh Freecad, and Modelica have stood the test of time. I have not got to the point of USING any of them, but at first sight Freecad is the easiest to learn. Of the others I'd add Altair Motionview, LMS Virtual.Lab Motion, Dymola and various matlab or simulink toolboxes or Simulink, but there are dozens of others as well.

Cheers

Greg Locock


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

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

 

[Mustaine3]

For basic kinematic analysis I would look toward Solidworks Motion when Solidworks is available in the company. Otherwise the already mentioned Adams or Simpack are professional packages for complex multi body dynamics.

 

[rb1957]

don't forget your Inventor ... it has some capability for mechanisms.

another day in paradise, or is paradise one day closer ?

 

[subsearobot]

Greg,
thanks- you hit my concerns head on. GIGO is my biggest concern, especially considering the learning curve, and time pressures.

My impression is that the MBD solution (if I am doing it- small company) would be slower than closing our eyes and crossing our fingers, but long term, there is of course benefit to a thorough analysis.

Once we define potential MBD studies, we may realize that a consultant is the answer...

 

[rb1957]

"we may realize that a consultant is the answer" ... I hear Dogbert's tail wagging ...

another day in paradise, or is paradise one day closer ?

 

[GregLocock]

I think rb1957 has a good idea with low barriers to entry. Use Inventor until you can't believe the results. Then do some real life tests to quantify the errors, then figure out if Inventor/hand calcs (remember those?) is good enough and just your model needs improving, or if you really need to got a standalone package, or a consultant. Can you concentrate on just one subsystem out of the whole assembly? Is one aspect the most troublesome? The idea of building an all-singing all-dancing sim is great marketing, but of the ADAMS jobs I do that aren't whole vehicles, few have more than 10 parts. For instance a model of a car going through a square edge pothole, a destructive test, needs only a pothole, a wheel, a tire, and the suspension arms. We did that one to back out the loads in all the suspension arms as the whole thing tore itself out of the car. I had high speed cameras to measure the gross motion of the wheel and car body, and a few strain gages, and a load cell in the pothole. Then I screwed around with the model of the bit where the tire hit the load cell, to get a good match to the limited data I had, and then cascaded to loads out from all the joints.

You are attempting to do the last bit without doing the whole physical testing that led up to it, tricky.

Cheers

Greg Locock


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

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

 

[GregLocock]

You may wonder why we bothered, rather than just strain gaging every arm. That was the idea originally, but then the strain gages did what what strain gages do, fail during the test.

Cheers

Greg Locock


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

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

 

[subsearobot]

Greg, I concur. thanks! this is what I was expecting to find.

We have a new round of upper management, and a long over run project. The project is typical- much more complicated than anticipated due to strict size constraints resulting in a ton of re-designs, iterations and then new mechanisms followed by marketing hemming and hawing. for several years! They are hoping to throw some sims at it, and accelerate time to market... It's a long fun story, worthy of a couple beers....

 

[richardl63]

You can do motion simulations in Inventor and get forces at any time step to feed back into FEA for the assembly. If you have many 3D contacts, it can quickly get bogged down, but other than that it works well. If your mechanisms are 2D planar, a very simple but very powerful tool I use sometimes is SAM.

http://www.artas.nl

Richard Lee, PE
Lee Engineering Inc

http://www.leeengineeringinc.com

 

[mronlinetutor]

When you are dealing with dynamic simulation links, it is no longer a FEA problem. It is a robotic problem. You have to find the force and moment on each link member using dynamics solved by the robotic program and then design the link member by a structural FEA program. one of the robotic program I know but never used is Simulink by Maple. Maple is a Mathcad program that can solve differential equations and tensors.

website for Simulink is:

https://www.mathworks.com/products/simulink.html

disclaimer: all calculations and comments must be checked by senior engineers before they are taken to be acceptable.