Is it Helpful?

Are you after FEA or fatigue? Fatigue in plastic is harder to do simply because the data is harder to come by. Fatigue in aluminum is well understood and documented. Obviously we are talking about high cycle fatigue and so aluminum's somewhat non-linear stress strain curve is not an issue.

TOP
CSWP, BSSE

"Node news is good news."

Even the limited static analysis of the demo gave me a small bump in comfort.

I used it to highlight a potential problem area, and sent a copy to my boss, with a gentle suggestion that the full FEA package would be nice to have.

My boss got yelled at for allowing me to waste my time on such fancy analysis.

Sigh.


Mike Halloran
Pembroke Pines, FL, USA

In a bitter irony, I had done the analysis on a Saturday, on my own time, but that didn't mitigate my offense.



Mike Halloran
Pembroke Pines, FL, USA

There are cheaper programs out there but then it becomes an issue of how accurate the answer is.

Click to expand...

Generally the accuracy of the answer is a function of the user's savvy. All the codes use essentially the same methods. You can ask how accurate a given element is, but in Simulation you don't have much of a choice aside from Draft or Normal.

Using FEA in conjunction with a test of the actual part is the best way to check "accuracy" and all the assumptions you made along the way.

As far as the Cosmos products are concerned, if you dig a little you can find verification problems and examples. This is generally true of any published code.

TOP
CSWP, BSSE

"Node news is good news."

Thanks for everyone's answers.

I am going ahead to upgrade to Professional.

I spoke to Abaqus as well and for me to upgrade to Professional will cost 6K but to get Abaqus would be 16K.

Abaqus said the difference is in the way the various contacts are handled. Most FEA programs are not that good.

For me who is a design engineer and not a FEA specialist and the fact that I am not building a plane but more basic items.

The combination of Cosmos and my own intuition and prototypes should give me a solution which I have confidence in.

Just to divert and have a rant.

In the future Engineers and other technically minded people will be the ones who will solve the problems generated by the accountants.

You might want to consider the life cost of Abaqus vs Professional. Since Abaqus is standalone you don't have to upgrade every year. Abaqus is a pretty mature code and not likely to have the kind of feature creep in Professional. If you go with Professional you do have to upgrade. It is the 6k plus the yearly subscription cost which IIRC would be about 1.6k. In 6 years you pay for the Abaqus in subscription fees.

Second, fatigue runs may take some time. With a standalone you just run it on three of your four processors and SW on the other. With Professional you are locked out till the job is done.

The Abaqus sales person was right about the handling of contact. Of course if you do fatigue, I believe that will use linear non-contact runs. Don't underestimate those little details.

Also, tech support for Abaqus will likely be different in some ways. You might ask about that in the FEA forum.

I've used both. 16k is pretty inexpensive.

TOP
CSWP, BSSE

"Node news is good news."

I remember being told that in the more recent SolidWorks releases (either 2009 or 2010) the simulation can run on an isolated set of cores while you continue to do modelling work in the same SolidWorks session utilizing the remaining cores. I believe by default it will run over all cores, but can be restricted to fewer in order to allow you to continue working.

I would verify claims like that.

I have a preference for standalone which is why I went with DesignStar and Cosmos/M a long time ago. My reason is twofold. First I know I can run as a completely separate process, perhaps even on a different machine. Second, I have a permanent record of what I modeled. In Simulation, I believe that when you change the model in SW you have no record of what you ran the FEA on other than the results which don't give the dimensions back to you.

In Cosmos/M as in Abaqus, NENastran and I believe Ansys, the analysis is saved as a script which can easily be tweaked, especially if you are running shells, plates or beams. Cosmos/M has a scripting language that can be used to make parametric models that find optimum dimensions for example or run through a bunch of possible materials.


TOP
CSWP, BSSE

"Node news is good news."

I'm waiting for my corporation to get around to upgrading our SW seats so I can verify the core isolation capability (among many other points).

I don't understand your comment of having a permanent record of what you modeled. I admittedly don't have much experience with analysis programs other than Simulation & Flow Simulation. How does your program of choice handle geometry iterations while maintaining model definition for each? In Simulation it's not automatic that you save a geometry set with each analysis run, but it can be done if you need that - and most people don't bother, only saving those runs they think they'll need.

Simulation can also goal-seek to find optimum dimensions. Not sure offhand if it can vary material as part of this process.

In my world you want to have a record of the geometry that goes along with the results and can't be changed inadvertently.

What you have to be careful about is not changing the geometry in SW after you have completed a run and then using that run to justify the geometry.

TOP
CSWP, BSSE

"Node news is good news."