Abaqus versions/ results issue

[MNS747]

Dear All
I have come across this interesting issue. I created all my thermal models and simulated using Abaqus 6.7-1 version to see the displacement. I was very happy with my results until i installed Abaqus 6.10-EF2. The version 6.10-EF2 converted the database. After re-running the simulation, i am not getting same results as i used to have with the version 6.7-1, the coutours and the magnitude of the temperature being read is fine but i am getting a lower magnitude in expansion. Nothing was changed, just an upgrade of the 6.7-1 file to the latest version and the results does not match.
I uninstalled the 6.10-EF2 and installed Abaqus 6.12-2 and repeated which again resulted in the same and i completely dont understand what I am doing wrong. Is it the compiler issue?

The question arises, why should one go for an upgrade if this spoils results.
Has anyone had this issue?
I would appreciate if someone could guide

Regards

 

[hadiandfem]

Hey there! Do a trial test. You model a simple version of your problem in both version. Compare the results. I am happy to see the outcomes!

 

[MNS747]

Hi hadiandfem
thankyou for your reply. I have done this trial test using two solid rectangular bodies interconnected. Done on 6.7-1 and converted the database into 6.12-2. Both the temperature and expansion results are same on both versions.
This means that it works fine for a simple model but doesn't for complicated geometries and assemblies such as my original model.
I am sure that the conversion process is missing something but how to find it?
Any suggestions?

Regards

 

[burningDiesel]

But what exactly are you converting? .cae files? If you write the input file with your 6.7 version and then run it using Abaqus 6.12, what happens?

 

[MNS747]

Hi sdebock
Yes it is the .cae file. By 'converting' i meant the Abaqus 6.7-1 file when openned in Abaqus 6.12-2 or any later versions, the software asks to convert (upgrade) the file to make it compatible for the current new version. (This upgraded file then cannot be opened back on 6.7-1 as being an older software version). I havent tried the option you mentioned. Will do and update.

Regards

 

[MNS747]

An Update
I have tried writing the input file and ran it on Abaqus 6.12-2. The results have become more unrealistic.
The .cae file conversion from 6.7-1 to 6.12-2 worked with a simpler model but did not work with my model (it is an assembly of a machine tool where some parts are 0with complicated geometries however all parts are Solid parts i.e. no shells or wires etc).

any suggestions

 

[hadiandfem]

Hi, alright. We now know the degree of geometry non-linearity is important. Well, I suggest you that review the [highlight #3465A4]solver[/highlight] in both version. Is that a big difference between results?

 

[IceBreakerSours]

Open the inp generated from a successful run of the previous version of the CAE in the new CAE (i.e., import it as a model). Save it as your new CAE and run a job, and let us know what happens.

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

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

 

[MNS747]

Hi hadiandfem, I am not too sure how can i know about the solver, could you guide me please?

The results difference is:
Heat transfer simulation is accurate
Displacement simulation (where the temperature simulation file is read into the model) the thermal expansion is less i.e. one parameter reduced from 71um to 62um, the other parameter reduced from -32um to 26um. Temperature file being read is correct when i plot NT11.
I have checked the thermal expansion coefficients are not changed

IcebreakerSours
I have done that already and as i mentioned earlier in my post, the results have gone more unrealistic.

Regards

 

[IceBreakerSours]

Compare the two INPs side by side. And check Simulia's customer help website for any answers.

Also, can you upload your old CAE and jnl files?

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

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

 

[hadiandfem]

Well, Please read the manual (theory) in each version, and try to realize that how displacement can be determined where the model is subjected to thermal loading. I can guess the solver is probably upgraded in new version.

 

[MNS747]

Hi hadiandfem
Yes you are right, I have found out that there have been changes in the solvers since every upgrade such as from 6.5 to 6.7 and to 6.12 and so on. However i would not have expected the reduction in the numerical value to almost 10um in my case. This can however be corrected if i use a higher expansion coefficient (still within the accepted range of the material).

IceBreakerSours
Thanks for your reply, I have checked both inp files side by side and YES there are changes in them but only with Parts and node numbers in the ElSet. Nothing major apart from some spaces.
Unfortunately it is not possible to upload the file because it is too big. Howeveri greatly appreciate your inputs to the problem.

Regards

 

[IceBreakerSours]

I don't know if you have managed to solve your problem but perhaps you could use the solution suggested by ABAQUS on their customer support website (click here); I just happened to bump into it today.

Search for the question ID QA00000008488 entitled "Abaqus/CAE Job submission using a prior version of Abaqus". I imagine you are going to need to have both versions of ABAQUS on your machine but I am not sure.

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

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

 

[MNS747]

Hi IceBreakerSours
Thank you for your reply and aplogies couldnt get back to you as i was away. For the time being i have turned back to the version 6.7-1. What i have noticed is if i increase one of the thermal expansion coefficients (still within the legitimate range) the results seems fine. This is absolutely wierd however this way atleast i can match up with my old results. I will have a look at your suggested link and get back.

Regards