Highlighted

NX Nastran SOL 402 versus SOL 601: Hyperelasticity

Creator
Creator

We have been informed by Siemens PLM that, starting with 2020, we cannot further use the ADINA based Nastran Advanved Nonlinear Solver (SOL601/701). Instead we have to change to the SAMCEF based solvers SOL 401 and SOL 402. In the first place, we are unhappy with this decision because we had acquired some knowledge in SOL 601 over time, which would then be useless. Second, the question arrises, whether the SAMCEF based solution really can compete with SOL 601.

 

Hyperelastic materials is one important issue to us. So the first thing we did, was to compare SOL 601 and SOL 402 (SOL 401 does not support hyperelasticity) for this sort of nonlinear static problems. In short: the outcome was not encouraging.

 

Performance: For realistic models, SOL 402 seems to run nearly one order of magnitude slower than SOL 601 for this kind of simulations.

 

Documentation: Not yet really available and not yet correct. For instance, SOL 402 does output Second Piola-Kirchhoff stresses in contrast to Cauchy stresses mentioned in the release guide.

 

Reliability: SOL 402 seems to produce spurious stress results for larger strains and nearly compressible material. Even SOL 106 performs better.

 

The complete comparison is attached to this post. Any ideas from the SAMCEF gurus?

 

Regards

Thomas

16 REPLIES 16

Re: NX Nastran SOL 402 versus SOL 601: Hyperelasticity

Solution Partner Phenom Solution Partner Phenom
Solution Partner Phenom

Dear Thomas,

Very sorry to hear your conclusions of SOL402 vs SOL601, I have read your report with attention and seems excellent, well done (by the way, you mention ion Fig. 1 to mesh with linear hexaedral elements, but the pictures shows tetrahedral elements, and I hope to be quadratic TET10 elements), I am following with interest the replacement of SOL601 by the new Nonlinear MultiStep Structural & Kinematic SOL401/402 solvers because I have customers with FEMAP & NX NASTRAN under maintenance with SOL601 that will have to do the swich as well soon, and I want the best for all of us, end users & VARs, then my worries. Please contact with SIEMENS GTAC guys, they are excellent for support.

The new FEMAP V2019.1 will feature soon a new version of SIMCENTER NASTRAN, hopefully the software will be released in next May 2019 (now we are finishing Beta2), Femap guys are working harder in the integration of the new SOL401/402 solvers in FEMAP doing a great job, then I hope to see very positive changes in the next release 2019.1 soon, they are very good, very professional.

I have requested several times to receive training materials for SOL401/402 solvers and the answer received is that the Nastran group will have training materials around the middle of this year, so let's see what will have.

The important message for me is that I am sure SIEMENS will invest all resources in the development & improvement of SOL401/402 solvers, and soon we will have a nonlinear solver ready to compete at the same level or superior with the most important nonlinear codes of the market, OK?.

Best regards,
Blas.

Blas Molero Hidalgo, Ingeniero Industrial, Director
IBERISA • 48004 BILBAO (SPAIN)
WEB: http://www.iberisa.com
Blog Femap-NX Nastran: http://iberisa.wordpress.com/

Re: NX Nastran SOL 402 versus SOL 601: Hyperelasticity

Creator
Creator

Dear Blas,

 

thank you for your kind feedback. Your are right: Fig. 1 shows TET10 elements. In most cases, we can use linear hex elements for hyperelastic materials. This is the reason for confining the reliability tests to them.

 

There are more worrying issues with SOL 402, which have to be investigated in more detail. My present impression is, there is still some way to go for a competitive nonlinear code. Maybe, the switch was too early.

 

Kind regards

Thomas

 

CES Grauer GmbH

Stuttgart (Germany)

https://ces-grauer.de/

Re: NX Nastran SOL 402 versus SOL 601: Hyperelasticity

Phenom
Phenom

I don't have enough experience with such material but your document is interesting despite not being able to ask "probing" questions. Have you taken this matter with SIEMENS as I 'd like to think that they would like to go through every points you raised to provide explanation.

If the current SOL601 works for you then may there is no need to upgrade! If indeed there are issues, developement time scale are not to your liking (The standard answer is usually "it will be fixed in next release") them maybe you ought to have a good look at competitive nonlinear code. There are some well known, very stable out there

Production: NX.CAE 10.0.2.6
Development: VB.NET (amateur level !)

Re: NX Nastran SOL 402 versus SOL 601: Hyperelasticity

Pioneer
Pioneer

@tg_s

indeed, following @selex_ct answer, you could maybe directly ask ADINA (SIEMENS integrates ADINA as the SOL601/701).
Here is a statement on their website:

"ADINA R & D would like to keep NX Nastran Advanced Non-Linear (SOL 601/701) customers very satisfied and continuously served with valuable solution capabilities. Hence, please note:

The ADINA solver can be used with Simcenter 3D and Femap pre/post, so NX Nastran customers can seamlessly migrate from SOL 601/701 to ADINA without changing their workflow."

For our NX Nastran Advanced Non-Linear Customers

 

best regards.

Cyril

Re: NX Nastran SOL 402 versus SOL 601: Hyperelasticity

Pioneer
Pioneer

I also consider that NX SIMCENTER 3D documentation as a whole needs a lot of improvement.

Especially, detailed, applied examples such as provided in ANSYS "ANSYS Mechanical APDL Technology Demonstration Guide" should be available to the users of NX (SIMCENTER 3D).

 

best regards,

Cyril

Re: NX Nastran SOL 402 versus SOL 601: Hyperelasticity

Creator
Creator

I also did some comparison between SOL601 and SOL401 for the arc-length method (SOL402 not yet include arc-length  method while it is included in  samcef ). For my test , the calculated buckling loads from the two solver are almost the same(with about 0.3% difference). However, SOL401 takes quite longer time than SOL601. For test case 1,SOL401 takes 536 seconds vs 83seconds in SOL601. For test case 2, SOL401 takes 1611 seconds vs 234seconds in SOL601. 

Now I have access to a trial license for samcef environment and I will try to test the arc-length method in samcef.

If there is still large gap between samcef and SOL601, I will try to get SOL601 from Adina in future. 

By the way , it looks like that Adina offer a big discount for SOL601/701 user.

 

Re: NX Nastran SOL 402 versus SOL 601: Hyperelasticity

Pioneer
Pioneer
I personnaly think that the people at SAMTECH are doing a great job. The integration of SAMCEF solver in NX and NASTRAN surely is not a easy task, but the contacts I had with them makes me confident that they care about us, users.

I contacted the Product Development Manager for Simcenter Nastran SOL402 via linkedin, they are working on your question. For the precision of the results it may only be due to a parameter (reduce MATHE K to a much smaller value); and the performance problem may be due the same problem.

I understand that you may be unhappy to have to change tools and habits; but it has been know for some years now that the ADINA solution would be dropped sometime. We have been unhappy when Model Solution in Ideas was dropped in favor of NXNastran but not we are quite used to it.

best regards,
Cyril

Re: NX Nastran SOL 402 versus SOL 601: Hyperelasticity

Creator
Creator

Cyril,

 

thank you for contacting Nastran development. Great to hear they are working on it.

 

By the way, it is the other way round: SOL402 is working best, when the MATHE K value is very large. If one does not enter a K value, SOL 601 does adjust it in a way that  Poisson ratio is 0,499. For this ratio SOL 402 produces those large spurious transverse stresses. This was my first test and therefore I stumbled over the problem.

 

On the other hand, there are materials with finite compressibility. For them SOL 402 is not yet working at all.

 

Zongfang, buying an upgrade to ADINA would be an option. The discount is attractive.

 

Regards,

Thomas

 

Re: NX Nastran SOL 402 versus SOL 601: Hyperelasticity

Pioneer
Pioneer

@tg_sglad this helped you, and to know that in fact SOL402 is "better" than SOL601!

 

From what you say, it may have been more a problem due to the lack of information on how to use the software (or difficulty to find infos in the documentation) than a problem with SOL402.

Therefore, don't forget to contact them directly next time; I work with SOL402 and whenever I have a question or a submit a problem they are very responsive (with solutions to my inquiries!). I would say, SAMCEF guys are like MAYA guys, very responsive and knowledgeable.

By the way, I'm sure you can find problems with other solvers too ... (be it Adina, Ansys or whatever).

Best regards,

Cyril