NASTRAN file translation 1

[ianstifle]

My colleauges and I have been evaluating Algor's ability to translate NASTRAN models in order to collaborate with one of our subcontractors on future projects. The tech support people at Algor have translated a model for us, and allowed us to attempt to run the model ourselves so that we may compare results with that which our subcontractor produces in NASTRAN. However, we have yet to be able to run the model successfully. The strangest thing that happens is that we will get displacement results, but no stress results. This baffles even the Algor support personnel, so I come here hoping for any suggestions you may have as to why this happens, and how to fix it.

 

[GBor]

It has been a while since I used this particular feature, but I was very successful with it in the past. Are you building the model in Algor's FEMPRO, exporting it to a Nastran .nas File, running it through the Nastran processor and reading the results with FEMPRO? Or are you reading the Nastran file and running it through the Algor processor?

 

[ianstifle]

GBor, the model has been created using FEMAP, exported as a NASTRAN file (.nas), and we are now importing that file into Algor for processing.

 

[GBor]

What type of analysis are you running?

 

[ianstifle]

static stress with linear material models

 

[GBor]

I just ran a simple 10" x 10" x 0.05" plate element with a 1000 pound load (yes, this is very high). I got answers in Algor, so I exported to a .nas file. When I reimported, the boundary conditions and loads had been stripped from the file (they didn't show up in the model), but the plate element had changed to a pshell1 element, so the translation had taken place. When I reapplied the boundary conditions and load, I got the same results as I had earlier...stress and deflections both existed.

You may want to insure your loads and boundary conditions are being properly translated.

Who are you working with at Algor?

 

[ianstifle]

The big question that we are currently trying to answer is exactly what you asked, whether or not the model is translated properly, including geometry, load, boundary conditions, etc. Right now we do not have the capability of performing this translation, so are working with algor reps (marwan, and james herzing) to demonstrate to us that it is worth our $$ to purchase the add on module.

The model shows all of the boundary conditions and loads translated correctly, but still does not return stress results for unknown reasons. The reps we are working with claim to be able to run the model successfully, so we are trying to determine where things are going wrong.

 

[GBor]

How much are they asking for the add-on?

 

[ianstifle]

something like $2400

 

[GBor]

Hmmmm...as a small business, I understand your concern. It would pay for itself the first time or two that you don't have to rebuild a model, but if you can't get it to work, it's useless. It seems as though they could give you a 30 day license for a "test drive". I would be willing to give it a shot if you would like to contact me 251-232-3723 and I'll give you an e-mail address.

 

[ianstifle]

Funny you should mention a 30 day trial, our very used-car-salesman-like sales representative, gave us that option a while back, but only if we cut the purchase req. within the next 24 hours. This happened when our funding manager was out of town, so was impossible, besides the fact that we didn't take too kindly to being forced into something in this manner. We are continuing to work with their support people, if the situation becomes more dire, I will contact you. Thanks for the help.

 

[GBor]

I contacted the Algor sales manager to see if there was something about which, at a minimum, he could inquire. Quite frankly, to treat a current customer with this attitude really aggrevates me. I've always been very pleased with my sales associates (first Pat LaLonde who is no longer with Algor, then Ray Deldin for several years, and now Michael Miller). All of them have been very kind and "top notch" on both product knowledge and customer support.

Anyway, best wishes on your plight. Feel free to contact me at any time by phone or by e-mail. You can find my e-mail address at my VERY new website:

http://www.borowskiengineering.com

Garland

 

[ianstifle]

Whatever strings you pulled at Algor seem to be working, because we are receiving a 15 day trial of the nastran translation module to evaluate. I spoke with an Ed Simmons today and he seems very interested in getting to the bottom of our issue. Thanks again!

 

[GBor]

Hope everything works out.

 

[ttoole]

I have had similar issues with this. Our company has purchased this add-on as well. The biggest issue was that anything from NASTRAN that was complex (MPCs, RBE3 elements, temperature dependance, composites, etc.) read into Algor produces a model that would often not run or would give a much different result. Going the other way to NASTRAN was OK because Algor covers the basic stuff common to most FEA programs like shell elements, properties, etc. This does not due us any good because our vendor has FEMAP and could read a basic Algor file anyway. I think the main problem is that Algor does not have a lot of the specialized features Nastran does so if someone has those in one of there models, Algor will just ignore it. They make it seem like they have a Nastran equivalent but this is really only a basic translator. I know COSMOS has one as well but they do not call it CSM/Nastran.

 

[GBor]

The question that I've run in to isn't what Algor does or doesn't do, but rather how does FEMAP interpret the output. Algor doesn't have the "specialized" elements because they are just modifications of existing element types. I think they would prefer that the analyst know what they are doing instead of handing a seemingly easy-to-use product into the wrong hands.

I've had success with the Nastran translator, although most of my models for this application have admittedly been pretty basic.

As an aerospace engineer, I'm suprised to see you even have access to Algor. I thought Nastran was the "weapon of choice" in the aerospace industry. Anyway, I wouldn't take a shot at a specific software package until I was certain that the limitation is in the software instead of my ability to use it or the translators.

Anyway, as I said in the other post, let me know if I can help in any way.

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.

http://www.borowskiengineering.com

 

[ttoole]

Garland,

Thanks! When you say FEMAP interpreting the output, do you mean reading in the nastran input file or bulk data deck? The companies we deal with have nastran with FEMAP or PATRAN and expect the models to import without a hitch. I was considering getting a license of FEMAP so we can import Algor and export nastran. Management is not happy because we have a bunch of $$$$ in Algor already and the ALG product which is not living up to our expectations. A single license of FEMAP is more than the Algor support fees. I personally have spent a lot of time learning Algor. It was not as easy to pick it up as I thought it would be but that is water under the bridge now. I can do most of what I need with it, however. Others using it are not as supportive.

Yes, basic stuff works with the ALG translator but we never get basic models! The dynamics models have more rigid elements than elastic ones sometimes.

We originally went with Algor because another company we were working with had it. That relationship is long since gone. I do see a lot of other companies with Nastran, however, that is why we got the ALG module but I feel I am fighting an uphill battle. On one hand I wish we just got nastran but we have Algor now and we have paid for it and support for the next year.

 

[GBor]

I certainly don't know everything, but here's a "for instance":

NENastran used to have a 2-D plain strain element (I understand they have taken this out of the latest release -- don't know why -- I had some pretty good success with it, but I understand they wanted to do some additional verification...anyway) when I brought an Algor model into FEMAP and attempted to export it into Nastran, it came through as plate elements. This is because MSC Nastran doesn't have a 2-D plain strain element. I double checked the Algor export ASCII file and it clearly stated 2-D plain strain. It wasn't Algor causing the problem, it was FEMAP import of a Nastran file.

I then had to redefine the elements as 2-D plain strain and send it to NENastran for processing. This worked fine, but that initial import was a problem.

I like having access to FEMAP because I like to "clean up" these little problems before I send the files to clients and they try and run them, but I say again, this isn't an Algor error OR an NENastran error...it has to do with FEMAP.

There are no "perfect" translators, but I have been successful writing a Nastran output file from the Algor pre-processor, processing it through NENastran and MSC Nastran, and post-processing in FEMAP or Algor. It's been a while since I've run these comparisons, but I wouldn't expect it to get worse...just better.

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.

http://www.borowskiengineering.com