FEA Software

[DavidJ]

FEA Software

Having been using AD for a short while now, I have a job coming up that will need FEA capability beyond ALGOR DesignCheck (shrink fits, temperature effects). This will be a very expensive addition for me, so I'm keen to make a good choice.

Does anyone have experience with Plassotech 3g.author?

Can anyone recommend other packages?

I'm interested in:-

* Easy to use, and get sensible results from
* Features / speed
* Value for money

Any input appreciated.

 

[MagneO]

Hello!

http://www.calculix.de/

The Calculix software would probably fit all criteria except easy to use. This is guesswork since I have not actually used the software, simply downloaded it.

Best Regards,

MagneO

 

[DavidJ]

Thanks for reminding me about calculix - I tried it some time ago and found it almost unfathomable. There are also packages like LISA and a few others (several for Linux) which seem to have great capabilities, but either can't import STEP, or are just completely non-obvious to use (and little or no support available when you get stuck). Having said that, documentation on some of the more expensive packages isn't always great!

I know the basics of stress analysis, but am not experienced enough to feel confident with packages that assume you know lots. So in this case, I'm prepared to pay for the combination of functionality and ease of use as it will allow me to do work I otherwise would have to decline.

 

[marc]

Algor is the one!

I have been an Algor user since 1998 and have been nothing but pleased. When I found Alibre and began to use the .stp format, Algor had a problem with that, but we fixed it by putting Alibre and Algor together. This has resulted in an absolutely seamless connection between the two programs. I have not used an .igs file in ages for translation into Algor. You may buy Algor in different levels of capability to fit your application. If you don't need Mechanical Event Simulation or Non-Linear analysis, simply but the Linear Static stuff.

Marc A. Meadows, P.E.
Meadows Analysis & Design, LLC
Fort Worth, Texas

 

[fitzbond]

Marc,

You might be the person I need to talk to my partner and I have had a difficult time in the last couple of years using Alibre & Algor. Do you have any tips about importing assemblies, models that use boolean features and configuration. We have had all kinds of problems getting models into Algor it seems now we have to create two models one for design and the other for analysis and on top of that we have use SW as the import tool by exporting from Alibre are either a SAT or STEP then importing into Algor from SW which I hate.

Anything you could offer would be appreciated.

 

[erezler]

Since you considered Algor, I will assume that your analysis needs are linear structural and thermal. If this is the case and you are willing to learn Nastran, I would recommend the solver from TMG: http://www.macnealgroup.com.

TMG was started by Dick MacNeal who was the orginal developer of the MSC.Nastran code used by the Aero and Auto firms today.

 

[DavidJ]

That looks interesting - I've sent an enquiry in for details. I may need to add non-linear materials in future to cope with local yielding. There are certainly huge variations in ease of use and capability in available systems!!

I don't expect I'll find everything I'd like in one package that I can afford.

My real limitation is my own lack of experience with FEA - so ease of use probably comes top of the list.

Thanks again for the input.

 

[cherkey]

James,
Using Alibre and Algor here....assemblies and Boolean Subtract. Make sure that the assembly and the part being used for the Boolean operation have the same origin location. It may look right in Alibre but once you import it into Algor, the assembly and the part that had the Boolean operation on it come in positioned on their own origin.

 

[DavidJ]

Ed,

I contacted macneal - they have suggested a slightly different product for my experience level, 3dmpact http://www3dmpact.com . Pricing looks interesting compared to others I've considered, I'll try the demo and see how it is for ease of use.

 

[fitzbond]

Re:

James,
Using Alibre and Algor here....assemblies and Boolean Subtract. Make sure that the assembly and the part being used for the Boolean operation have the same origin location. It may look right in Alibre but once you import it into Algor, the assembly and the part that had the Boolean operation on it come in positioned on their own origin.

Thanks Cherkey that's good to know I however have run into parts that are fully anchored in Alibre but it Algor it comes it detached the I will use just to get the part to remain in its place. Then there configuration when I export the current active configuration once imported into Algor it will choose a different configuration or combination of configuration which blows apart the whole assembly because of the mismatched configuration. That’s when I use SW as the go between and have to create a FEA model which in turn is twice the work just to get a model that I can analysis.

 

[erezler]

David, the thanks goes to you.

I use to work for Pedro's orginal company (MARC Analysis Research) before it got bought out by MSC.Software in '99. Always wondered about what Pedro was up to other than retirement. Now I know: he started a new FEA company (http://www.3dmpact.com) not 10 mintues from me.

Ed

 

[MagneO]

Hello!

Is http://www.3dmpact.com an operating company? The web-site appears to be inactive, latest update in 2004?

Best Regards,

MagneO

 

[DavidJ]

I had the same thought - so I asked at Macneal, they claim the software is being developed and supported.

I'd be interested if anyone else can shed light on this, as I find it a bit odd that a software company doesn't bother to maintain its web site...

 

[MagneO]

Hello again David,

If you find time to do some testing, please let us know how it worked. At least I would be interested in knowing about a (low cost?) FEM software with active support.

Best Regards,

MagneO

Edit:

Is this http://www.3dfeva.com/ the same company?

 

[DavidJ]

Now pay attention, this gets confusing!

mpact is a licensed version of 3dfeva.
3dfeva is no longer supported.
The people behind 3dfeva now have AMPS (http://www.ampstech.com)

The feedback form and support e-mail address on the mpact web site don't work, but you can download the software.

The 3dfeva support address does work (but not the feedback form) - they did answer to let me know about AMPS.

AMPS web site looks pretty, but I don't understand it.

I will take a quick look at 3dmpact just to see if it lives up to the claims - I've also asked Macneal about long term viability of this product.

I'll report back but it may take a while....

 

[Anonymous]

I too checked out the mpact software. I'm really leary of it because it's default is to mesh shells with triangles and solids with tets. These elements are notoriously stiff and should only be used in areas of non-interest. I work for a large aerospace company as a stress analyst and we use Nastran/Patran for linear statics. We use Abaqus for non-linear and composites. If we had our way we would use Abaqus for everything because of it's deck format and awesome documentation, but some customers require the use of Nastran. These are both really expensive programs but are worth it. I looked at the $995 package from Macneal and that looks to be the best deal for trustworthy FEA with all of the standard elements. I'm not sure what kind of pre/post processor that they offer. I do know that programs like Algor, Cosmos, and Pro/Mechanica do not allow enough control of the model. They don't offer enough MPC's and don't allow realistic enough representations of the system. IDEAS Model Solution is actually really good, but I don't know how much it costs or how much longer it will be supported. We do most of our meshing/model generation with I-DEAS and export an Abaqus or Nastran deck. Also, with deck editing capability you can add all kinds of features to your model that the pre/post processor may not be capable of generating. So, even if the Macneal Nastran pre/post processor is not that good you can still probably do everything you need to in the deck once the mesh, materials, and property sets are created. You can always add your RBE's by hand. One option to look into is the Algor/Nastran product. You might be able to get Algor's Pre/Post processor and Macneals Nastran to have a really good set of tools. I hope this helps.

 

[Cameraman]

I've used Algor since 1992 (when it came on red and blue 3.5" floppies that you had to store in a long tray ... ah, the memories) and feel that it is in a nice niche between the "big boys" and the "toys". The current package that I have is a basic linear stress package (and I don't need it very often in any case), but in a previous life I had the full non-linear package with kinematics, heat transfer, etc. The company is good to work with and they have a solid product ... and not just "for the money", either ... I had a chance to perform some significant hands-on comparisons with ANSYS, and Algor definitely held its own. And the new interface with AD works quite well for me, which saves a lot of time.

With all of that being said, I must offer two clarifying remarks:

First, regarding Fitzbond's lamentation above, I would note that it is common practice to create "special" 3D models specifically for performing an FEA ... there are a number of reasons for this, but the most common reason is speed ... by simplifying the model and removing features that have no affect on the analysis, I can cut my analysis time often by an order of magnitude.

Second, I must admit that my relatively recent experience extends only to Algor and ANSYS ... in the past 10-15 years, I have found no reason to look beyond Algor and have therefore not done so. I look forward to hearing about the other packages being discussed herein (for possible future reference).

Regards,
Greg

 

[DavidJ]

Finally got 3dMPACT loaded - import of a model I'd already tried in other systems did not give proper solid gemetry, so I could not go much further.

It seems quite a few FEA systems import via IGES, and this doesn't always work well. MPACT filled in quite a lot of holes in the component I tried to import.

I think I'm maybe having IGES related problems with another package I'm looking at.

 

[DavidJ]

Status of my LIMITED evaluation so far - looking at several products currently available for a similar cost (some via special time limited offers)

ALGOR (Linear Static module) - works seamlessly with Alibre, reporting tools excellent and versatile. Expandable later (non-linear etc). Handling interference fits is a bit laborious (an issue for me). Support seems very good from my exposure to them.

Plasotech 3g.author - really easy to use, fast and meshes nicely. Very easy to get started with. Great features for design optimisation, but these don't work with Alibre. Very easy to deal with interference fits. Not able to be expanded to non-linear materials (a potential issue for some of my work). Reporting tools not very versatile, software seems a little unstable.

ROSHAZ - This seems very capable, and can be expanded to non-linear materials at no cost (by working with the Calculix solver) - unfortunately I haven't yet been able to successfuly generate a solid from imported CAD file - so can't check out the features / usability. Maybe an IGES issue, or perhaps I'm just stupid.

NEiFusion (Designer)- Sounds excellent, great CAD file import capabilities. A new product which combines 3D core from SolidWorks (drawings etc removed) with Noran's Nastran L series solvers. Unfortunately in the UK the 30 day evaluation is not being made available 'to keep costs down' so I can't easily check if this system will do what I want. Expansion to non-linear, is reasonably priced - and there is an 'Analyst' package (more expensive) which allows use with other solvers.

I'm also trying to look at AMPS - the documentation suggests this has advanced capabilities which I don't understand. I haven't as yet been able to find anything to allow be to judge ease of use.


There it is - summary so far:-
The guys at ALGOR are trying hard to be helpful. Plassotech is best so far for ease of use (IMHO), if only the design optimisation tools worked with Alibre and/or later expansion to non-linear materials was possible. IGES based import seems to be a serious limitation of some systems (not all support STEP). Many tools are probably very capable, but are let down by ease of use for a relative novice like me. There are some good options available at pirces which are still significant for a one man business, but are not ridiculously high.

I'll report back if I find a way to look properly at NEiFusion.

PS I've also found that vendors tend to structure pricing differently for the US or Europe - US prices are lower but typically don't include the first year of maintenance (20-25% of price).

 

[DavidJ]

Update - by tweaking some IGES export settings in Alibre, I did get one of my test parts to load into MPACT. But then I couldn't fathom how to get much further, and the package seemed to stall when meshing. Too many different program modules to interact with in my opinion.

Same IGES tweaks got the model into Roshaz successfully - but again, I couldn't realy get a lot further.

In both cases, this is more likely due to my level of expertise than it is any error in the package - but I think it suggests that these are perhaps not for beginners (in contrast to 3g.author and ALGOR where the inteface is pretty intuitive, and default settings at least get you started).