FEAP User Forum

FEAP => Input File Issues => Topic started by: Bruce Wayne on April 27, 2021, 05:55:06 AM

Title: LAYEr option in BLOCk command.
Post by: Bruce Wayne on April 27, 2021, 05:55:06 AM

Hello FEAP Users,

I am trying to get a structure of alternating layers in x-direction of two materials using LAYEr option in BLOCk command(as given in p37 in user manual) but this error is being thrown.

 *WARNING* Length allocation for:TEMP1 Length =           0
  * ERROR * Partition           1  NEQ =           0

I tried using r/x1/1 for dir_b but none seem to work. Can you help me out?

I am attaching the Inputfile.(I am trying a Hill-Mandel Homogenization on an RVE containing alternate layers of materials 1 and 2.)

Thanks in advance.

Title: Re: LAYEr option in BLOCk command.
Post by: Prof. S. Govindjee on April 27, 2021, 09:30:15 PM

Please always state which version of the code you are using.

Title: Re: LAYEr option in BLOCk command.
Post by: Prof. S. Govindjee on April 27, 2021, 10:21:42 PM

I checked in version 8.6.  The code looks to be a bit buggy.
I was able to make your input file work by switching to bilinear elements, manually computing the number of elements/nodes, and fixing an error you had in the input file.

Code: [Select]

feap **Hill for RVE of alternate layers of 2 Materials **
81 64 2 2 2 4
and

Code: [Select]

block
  cart d d
  layer 1 ! layer of material 1 and 2 along direction r (x1 or x)
    1 2 1 2 1 2 1 2
  quad 4
  1 0 0
  2 L 0
  3 L L
  4 0 L

The relevant code is program/blkgen.f and this helps to understand what FEAP is trying to do.  Note that nr, ns, nt are then number of nodal divisions that you specified with the CART d d line in your block; so in your case nr = 8 and ns = 8 too.

Code: [Select]

110   errck = tinput(layer,1,td,5)
      if(errck) go to 110
!     Layer data

      if(pcomp(layer(1),'laye',4)) then
        dlayer = nint(td(1))
        if(dlayer.eq.1) then
          nlay = nr
        elseif(dlayer.eq.2) then
          nlay = ns
        elseif(dlayer.eq.3) then
          nlay = nt
        endif
        errck = palloc(111,'TEMP1',nlay,1)

        j = 1
        do while(j.le.nlay)
120       errck = tinput(layer,0,td,16)
          if(errck) go to 120
          do i = j,min(j+15,nlay)
            mr(np(111)+i-1) = nint(td(i-j+1))
          end do ! i
          j = j + 16
        end do ! while
        go to 110

Note that the direction for LAYEr needs to be {1,2, or 3}.  Then the code performs an polling input until you have input data for the number of nodal divisions in the chosen direction, 8 in your case.
The reading happens 16 divisions at a time.

We'll have a look to see what can be done.  If you want to accelerate the process, have a look at program/blkgen.f and then program/sblke.f where the material numbers are set using the data that you have entered in the block command.  Note that in sblke.f the direction of the layers in in dlayer and the data is in the array ilr.  Most likely the error is in the indexing into ilr.

Title: Re: LAYEr option in BLOCk command.
Post by: Prof. S. Govindjee on April 27, 2021, 10:45:03 PM

Here is a not so ideal patch to make the quadratic case work.

In the definition of the block use

Code: [Select]

  layer 1
    1 1 2 2 1 1 2 2Thus you are entering two numbers for each element (really one per nodal division).  Then in program/sblke.f near lines 177 and 187 make the following changes

Code: [Select]

!177
ma = ilr(j)  ! ilr(inc*j-inc+1)

!187
 ma = ilr(i)  ! ilr(inc*i-inc+1)

These are quick hacks to get the code to function but are not fully robust.  In particular, the automatic counting feature will still not work so for the quadratic case you will need to state

Code: [Select]

feap **Hill for RVE of alternate layers of 2 Materials **
81 16 2 2 2 9

Title: Re: LAYEr option in BLOCk command.
Post by: Bruce Wayne on April 28, 2021, 04:18:25 AM

Thanks Professor , It is working now