Discrepancy of K-eff by different "opti" mode

Report all good and bad behavior here
Post Reply
hartanto
Posts: 15
Joined: Sun Oct 21, 2012 11:28 am
Security question 1: No
Security question 2: 92
Location: University of Sharjah

Discrepancy of K-eff by different "opti" mode

Post by hartanto » Mon Aug 10, 2015 5:41 pm

Dear Serpent team,

I am currently working on a research reactor core design using Serpent 2.1.24. We are doing equilibrium core search and the "set opti" card is considered to reduce memory consumption.

The problem is the k-effs with different "opti" mode are very different. For your information, I used the ENDF/B-VII.1 library (converted from MCNP6 library to Serpent library) and we generated our own S(a,b) library based on ENDF/B-VII.1 too. In the input, the tms option is used to treat the temperature-dependent cross-sections of all materials. The new s(a,b) on-the-fly interpolation capability is also used for the water and beryllium reflector.

These are the k-eff results at clean core with different "opti":

Code: Select all

Opti_1_res.m:ABS_KEFF                  (idx, [1:   2]) = [  1.19948065E+00 0.00043 ];
Opti_2_res.m:ABS_KEFF                  (idx, [1:   2]) = [  1.11664997E+00 0.00051 ];
Opti_3_res.m:ABS_KEFF                  (idx, [1:   2]) = [  6.09881387E-01 0.00073 ];
Opti_4_res.m:ABS_KEFF                  (idx, [1:   2]) = [  1.11565342E+00 0.00039 ];
 
I believe "opti=4" provides the correct answer.

Thanks.

Donny

User avatar
Tuomas Viitanen
Posts: 65
Joined: Thu Mar 25, 2010 1:50 pm
Security question 2: 0
Location: VTT Technical Research Centre of Finland, Espoo

Re: Discrepancy of K-eff by different "opti" mode

Post by Tuomas Viitanen » Tue Aug 11, 2015 5:10 pm

Hi,

This problem seems awkwardly familiar to me: I experienced similar problems just before making the last fixes to the TMS routines for the 2.1.24 update. In fact, TMS did not work correctly in the lower optimization modes in many of the previous updates, but those problems should have been fixed in version 2.1.24.

I re-ran my own TMS / OTF S(a,b) test cases with Serpent 2.1.24, but failed to reproduce the problem. Could you send me your input file by e-mail? (tuomas.viitanen (a) vtt.fi)

Best regards,

Tuomas

User avatar
Tuomas Viitanen
Posts: 65
Joined: Thu Mar 25, 2010 1:50 pm
Security question 2: 0
Location: VTT Technical Research Centre of Finland, Espoo

Re: Discrepancy of K-eff by different "opti" mode

Post by Tuomas Viitanen » Wed Aug 12, 2015 4:29 pm

There is a bug in the new on-the-fly S(a,b) interpolation capability that I was unable to find in my test calculations.

The new capability is working correctly only in optimization modes 2 and 4. I'm looking for a way to get it working also in optimization modes 1 and 3.

Tuomas

User avatar
Tuomas Viitanen
Posts: 65
Joined: Thu Mar 25, 2010 1:50 pm
Security question 2: 0
Location: VTT Technical Research Centre of Finland, Espoo

Re: Discrepancy of K-eff by different "opti" mode

Post by Tuomas Viitanen » Fri Aug 14, 2015 1:07 pm

A solution to this issue is reported in

http://ttuki.vtt.fi/serpent/viewtopic.p ... b3c4#p6036

Thanks to Donny for reporting this problem!

Best regards,

Tuomas

novako
Posts: 21
Joined: Sun Feb 15, 2015 1:01 pm
Security question 1: No
Security question 2: 92
Location: Prague, Czech Republic

Re: Discrepancy of K-eff by different "opti" mode

Post by novako » Tue Apr 26, 2016 5:13 pm

Hi,
I have similar issue while calculating few -group cross section for several burnsteps and branch cases. I used opti 1 and keff was 9.70421E-01. When I tried opti 4, keff reached ..... for the first burn step. Opti 2 seems to behave correct as well with keff 1.12967E+00 . Due to high memory requirements of opti 4, I use opti 2 now for my cross section generation. I found this error while comparing Serpent and SCALE results. SCALE keff 1.12131774
I have not used above mentioned local code fix, I am planning to do it in few weeks.

novako
Posts: 21
Joined: Sun Feb 15, 2015 1:01 pm
Security question 1: No
Security question 2: 92
Location: Prague, Czech Republic

Re: Discrepancy of K-eff by different "opti" mode

Post by novako » Tue Apr 26, 2016 9:07 pm

Sorry,
It was old calculation calculated with old serpent version.
Ondrej

Post Reply