Page 2 of 3

Re: Restart feature in 2.1.17

Posted: Tue Sep 22, 2015 12:10 pm
by Tian
Hi,

Just a small question about restarting, I have add a line

Code: Select all

set rfw 1 test
in the input, however I have got the error massage
Fatal error in function StoreComposition:

Unable to open file for writing
I am using version 2.1.23. What might be wrong? Thank you.

Re: Restart feature in 2.1.17

Posted: Thu Sep 24, 2015 10:15 am
by Jaakko Leppänen
Looks like Serpent is unable to write the restart file. Have you checked the permission in the directory where the calculation is run? Is it possible that you are running out of disk space?

Re: Restart feature in 2.1.17

Posted: Thu Mar 24, 2016 1:26 pm
by Andy_Turner
Can this feature be used to restart from an interrupted calculation? e.g. like in MCNP, doing 'mcnp r=old_runtpe c'

Re: Restart feature in 2.1.17

Posted: Fri Mar 25, 2016 12:15 am
by Jaakko Leppänen
Yes, basically you can go back to the last completed burnup step (provided that the writing of the restart file was enabled).

Re: Restart feature in 2.1.17

Posted: Tue Mar 29, 2016 12:34 pm
by Andy_Turner
Jaakko Leppänen wrote:Yes, basically you can go back to the last completed burnup step (provided that the writing of the restart file was enabled).
Sorry I meant during a normal transport run, whether it is possible to continue a neutron transport calculation run that was terminated prematurely, or perhaps you want to run more histories for example...

Re: Restart feature in 2.1.17

Posted: Tue Mar 29, 2016 6:02 pm
by Jaakko Leppänen
Unfortunately, no.

Re: Restart feature in 2.1.17

Posted: Thu Mar 31, 2016 12:39 pm
by Andy_Turner
Jaakko Leppänen wrote:Unfortunately, no.
Oh okay. Feature request? ;)

I understand Serpent2 was designed with an ethos on scalability, to run on large clusters. If there is indeed a goal to do large-scale fixed source problems (once variance reduction is implemented) then we would be talking about using 100's of cores for many weeks. In these cases you would definitely need a binary dump and capability to restart/extend/complete a job from last dump (redundancy against node failure, flexibility, practicality, difficult to know how many NPS is needed before you commence the run etc).

Thanks

Andy

Re: Restart feature in 2.1.17

Posted: Thu Jul 07, 2016 7:57 pm
by satkinson
Hi Jaakko,

The restart feature seems to only work on bu's greater than 0.1 MWd/kgU. In this case restarting with 0.177 works, however 0.088 does not and it produces the error below.

Restart file "restartfile" contains the following burnup points:

0 : BU = 0.000 MWd/kgu, time = 0.00000E+00 days
1 : BU = 0.088 MWd/kgu, time = 1.00000E+00 days (closest point)
2 : BU = 0.177 MWd/kgu, time = 2.00000E+00 days


***** Thu Jul 7 17:40:05 2016

Input error:

Burnup point 0.088 MWd/kgU not found in restart file

Cheers,

Satkinson

Re: Restart feature in 2.1.17

Posted: Thu Jul 07, 2016 10:47 pm
by Jaakko Leppänen
Do you get this error with 2.1.26 or with some older update?

A quick solution might be to use the burn time instead of burnup (negative entry in the rfr card).

You can also use "set rfr idx 1" which should give you the first burnup point (this feature has probably not been described anywhere).

In any case, the routine not finding a burnup point < 0.1 is clearly a problem that should be fixed.

Re: Restart feature in 2.1.17

Posted: Fri Jul 08, 2016 11:39 am
by satkinson
Hi Jaakko,

This was found using 2.1.26.

Thanks for the work around.

Satkinson