FEAP User Forum

FEAP => Input File Issues => Topic started by: crepes on December 03, 2018, 01:30:56 AM

Title: Fully coupled problem with partition command
Post by: crepes on December 03, 2018, 01:30:56 AM

Hello,

Imagine a problem with 3 degrees of freedom.

Should an input file which includes

Code: [Select]

part
1 1 1
as the only partition with e.g.

Code: [Select]

      PART,,1
      loop,, 10
        tang,,1
      next
yield the same solution as an input file without any partition stuff?

Or maybe another question: Should they have the same tangent matrix?

Kind regards

Title: Re: Fully coupled problem with partition command
Post by: Prof. S. Govindjee on December 03, 2018, 06:37:07 AM

I believe so.  See http://feap.berkeley.edu/wiki/index.php/Partitions for some additional information on partitions.

Title: Re: Fully coupled problem with partition command
Post by: Prof. R.L. Taylor on December 03, 2018, 10:31:28 AM

Try it and let us know!

Title: Re: Fully coupled problem with partition command
Post by: crepes on December 04, 2018, 09:55:11 AM

Dear Professor Taylor, Dear Professor Govindjee,

Thank you very much for your replies.

Quote from: Prof. R.L. Taylor on December 03, 2018, 10:31:28 AM

Try it and let us know!

I tried it recently with a built-in feap-element and both the variants (with PART 1 1 1, and additionally without any partition commands) behaved the same. I had done it before with different results, but I think I just had messed something up with the input file.

Quote from: Prof. S. Govindjee on December 03, 2018, 06:37:07 AM

I believe so.  See http://feap.berkeley.edu/wiki/index.php/Partitions for some additional information on partitions.

The link is very helpful. I  did not take into account that I might need to do some changes in my user element when using part command. But still when I had a look at the routines solid2d.f, partpt.f, pmacr6.f and other routines that are called, it seems to me that changes regarding PART are optional inside a user element. Maybe just for saving computation time.

Am I right? If not, I should start a second attempt to understand what's going on.

Title: Re: Fully coupled problem with partition command
Post by: Prof. R.L. Taylor on December 05, 2018, 11:26:44 AM

You are correct about adding 'part' checks in an element.  Partitions only assemble the dof's that are active.  So even if you compute everything in an element the inactive parts are not used.

Title: Re: Fully coupled problem with partition command
Post by: crepes on December 05, 2018, 11:15:30 PM

Thank you very much for this information