Problem using 'restart' to continue transient analysis

[hustzjy312]

Dear all,

When I use 'restart-save' command to continue the transient analysis with solid element using Newmark time integrator (considering finite deformation), it works well on 'Hubble' (Personal computer, Max OS X 10.6.8, Compiled with 32-bit although the system is 64-bit). However, the results divergence after 3 time steps when I run the same the same problem with same input file on 'Sugar' (Cluster, Linux, Compiled with 64-bit).

In addition, I have successfully used 'restart-save' command to continue the transient analysis with finite deformation beam element using Newmark time integrator on both 'Hubble' and 'Sugar'.

Therefore, I wondering whether there is some bugs with 'restart' command so that it is not compatible with finite deformation solid element.

Could you please give me any suggestion?

Thanks.

Yang

[FEAP_Admin]

Are you writing restart files on one machine and then trying to use them on a different machine?  This can be tricky since the restart files are binary files.  The file that controls the writing and reading of the restart data is program/restrt.f.  That is the place to look to start to debug.

If the problem you are seeing is related to writing the restart and reading it on the same machine, can you please provide the simplest possible input file that shows the error?

Also what version of the code are you using?

[hustzjy312]

My problem is related to writing the restart and reading it on the same machine. The input files are attached. Let me briefly introduce my procedures.

There are two input files 'IP1' and 'IP2'. These two input files are the same except the 'restart' and 'save' part. Firstly I run the input file 'IP1' and obtain the restart file 'RP1001.01'. Secondly I copy the restart file 'RP1001.01' and rename it as 'RP2001.01'. Thirdly I run the input file 'IP2' using the restart file 'RP2001.01'.
 
The version of my FEAP is Feap83.

Thanks,

Yang

[FEAP_Admin]

I have verified your problem on my 64-bit Mac.  There is probably a pointer error in the save or the restart subroutines.
I'll have a look.

[FEAP_Admin]

There is an error in program/restrt.f:

To fix the error change

Code: [Select]

       setvar = walloc( 28,'F0   ',nneq, 2, ios) to

Code: [Select]

       setvar = walloc( 28,'F0   ',4*nneq, 2, ios)
This fixed the problem on my machine (why you did not see it on your 32bit machine is hard to understand).