B1 fundamental mode calculation in SERPENT

 Posts: 2
 Joined: Tue Sep 28, 2010 11:16 pm
 Security question 2: 0
 Location: Montréal, Québec
B1 fundamental mode calculation in SERPENT
Hi everyone,
I would be interested to know if anybody is using the B1 iteration mode in SERPENT?
I am working on the derivation of an actual B1 model for Monte Carlo calculations (yet based
on a multigroup form) and it would be great to have some idea of whether the SERPENT model is effective, and what kind
of Monte Carlo lattice calculations would typically require a regular leakage model.
For the time being, comparisons between SERPENT (B1 iteration mode) and deterministic B1 calculations
with the DRAGON code give me some differences, but it is likely to be due to differences in the models themselves.
So let me know if you have nonrestricted references on this topic, or comparison between SERPENT and other lattice codes such as CASMO, HELIOS or whatever, or simply your opinion on the relevance of Monte Carlo leakage models for criticality calculations!
Nicolas
I would be interested to know if anybody is using the B1 iteration mode in SERPENT?
I am working on the derivation of an actual B1 model for Monte Carlo calculations (yet based
on a multigroup form) and it would be great to have some idea of whether the SERPENT model is effective, and what kind
of Monte Carlo lattice calculations would typically require a regular leakage model.
For the time being, comparisons between SERPENT (B1 iteration mode) and deterministic B1 calculations
with the DRAGON code give me some differences, but it is likely to be due to differences in the models themselves.
So let me know if you have nonrestricted references on this topic, or comparison between SERPENT and other lattice codes such as CASMO, HELIOS or whatever, or simply your opinion on the relevance of Monte Carlo leakage models for criticality calculations!
Nicolas
 Jaakko Leppänen
 Site Admin
 Posts: 2134
 Joined: Thu Mar 18, 2010 10:43 pm
 Security question 2: 0
 Location: Espoo, Finland
 Contact:
Re: B1 fundamental mode calculation in SERPENT
Nicolas,
The B1 mode in Serpent is based on some ideas without real theoretical foundation, and it should be seen more like an experimentation. In general, the results are not very consistent with deterministic critical spectrum calculations.
There is another critical spectrum calculation mode under development, based on the solution of multigroup B1 equations, and it will be available in the next update. The nice thing about this method is that it is based on a very similar approach to what is used in deterministic codes and it should give results that are very consistent with deterministic lattice transport calculations.
The B1 mode in Serpent is based on some ideas without real theoretical foundation, and it should be seen more like an experimentation. In general, the results are not very consistent with deterministic critical spectrum calculations.
There is another critical spectrum calculation mode under development, based on the solution of multigroup B1 equations, and it will be available in the next update. The nice thing about this method is that it is based on a very similar approach to what is used in deterministic codes and it should give results that are very consistent with deterministic lattice transport calculations.
 Jaakko

 Posts: 2
 Joined: Tue Sep 28, 2010 11:16 pm
 Security question 2: 0
 Location: Montréal, Québec
Re: B1 fundamental mode calculation in SERPENT
Jaakko,
It is exactly what I have done.
I use the same routines for deterministic B1 calculations into a (multigroup) Monte Carlo criticality calculation, to get diffusion coefficients and to introduce a pseudoleakage reaction (d(g)b2) in each g group. Criticality MC shall converge to the fundamental mode in this case, identically to deterministic calculations.
It is not yet fully validated and I will check my programming with the SERPENT one. When do you expect to release the 1.14 version?
I am preparing a paper documenting this approach (as it has surprisingly never been proposed, to my knowledge), so let me know if you are interested in a joint one, in the case of you haven't already done one!
Nicolas
It is exactly what I have done.
I use the same routines for deterministic B1 calculations into a (multigroup) Monte Carlo criticality calculation, to get diffusion coefficients and to introduce a pseudoleakage reaction (d(g)b2) in each g group. Criticality MC shall converge to the fundamental mode in this case, identically to deterministic calculations.
It is not yet fully validated and I will check my programming with the SERPENT one. When do you expect to release the 1.14 version?
I am preparing a paper documenting this approach (as it has surprisingly never been proposed, to my knowledge), so let me know if you are interested in a joint one, in the case of you haven't already done one!
Nicolas
 Jaakko Leppänen
 Site Admin
 Posts: 2134
 Joined: Thu Mar 18, 2010 10:43 pm
 Security question 2: 0
 Location: Espoo, Finland
 Contact:
Re: B1 fundamental mode calculation in SERPENT
The B1 fundamental mode calculation and group constant generation in criticality spectrum was implemented in update 1.1.14. The method is invoked by the "fum" option (Sec. 5.16 in the User's Manual) and it is applied to a set of separately generated group constants (diffusion coefficient, absorption, fission and fission production cross sections and the scattering matrix), named B1_<etc> in the _res.m output file.
The method works by generating homogenized reaction cross sections on an arbitrary multigroup structure, defined by the "fum" option. After the transport simulation is completed, this data is used to form the B1 equations. The solution yields the criticality spectrum, which is then used to rehomogenize the cross sections.
The method was developed with the help of FZD in Germany and it has been tested to some extent by comparison to HELIOS calculations. You should expect differences no larger than few percent when compared to deterministic lattice codes in LWR geometries (when using cross sections based on the same evaluation). The calculation should work with multiple universes in the "set gcu" card, but I haven't tried running any such cases yet.
The next step is to apply the spectrum correction to transmutation cross sections in the burnup calculation mode. I'll start working on that soon.
The method works by generating homogenized reaction cross sections on an arbitrary multigroup structure, defined by the "fum" option. After the transport simulation is completed, this data is used to form the B1 equations. The solution yields the criticality spectrum, which is then used to rehomogenize the cross sections.
The method was developed with the help of FZD in Germany and it has been tested to some extent by comparison to HELIOS calculations. You should expect differences no larger than few percent when compared to deterministic lattice codes in LWR geometries (when using cross sections based on the same evaluation). The calculation should work with multiple universes in the "set gcu" card, but I haven't tried running any such cases yet.
The next step is to apply the spectrum correction to transmutation cross sections in the burnup calculation mode. I'll start working on that soon.
 Jaakko
Re: B1 fundamental mode calculation in SERPENT
Hello Jaakko,
A few comments on the B1 fundamental mode calculation, or rather the documentation of the same.
As you state above the B1corrected parameters are named B1_<etc>, meaning having the prefix B1_ in front of the uncorrected parameter. This goes for all parameters except B1_SCATTXS, which, if I understand this correctly, is the corrected value of GTRANSFXS. Maybe the parameter should be renamed B1_GTRANSFXS to avoid misinterpretations?
The same parameter has, according to the manual, 4*G^2 values in the output. It should be changed to 2*G^2.
The parameter B1_BUCKILNG is misspelled in the table of section 6.1.29, page 88.
Section 5.16, line 5. The word criticality is misspelled.
I just read your and Fridman's paper in ANE (ANE 38, pp 13991405) and your results are very encouraging!
All the best,
Calle Berglöf, KTH
A few comments on the B1 fundamental mode calculation, or rather the documentation of the same.
As you state above the B1corrected parameters are named B1_<etc>, meaning having the prefix B1_ in front of the uncorrected parameter. This goes for all parameters except B1_SCATTXS, which, if I understand this correctly, is the corrected value of GTRANSFXS. Maybe the parameter should be renamed B1_GTRANSFXS to avoid misinterpretations?
The same parameter has, according to the manual, 4*G^2 values in the output. It should be changed to 2*G^2.
The parameter B1_BUCKILNG is misspelled in the table of section 6.1.29, page 88.
Section 5.16, line 5. The word criticality is misspelled.
I just read your and Fridman's paper in ANE (ANE 38, pp 13991405) and your results are very encouraging!
All the best,
Calle Berglöf, KTH

 Posts: 34
 Joined: Thu Sep 22, 2011 5:27 pm
Re: B1 fundamental mode calculation in SERPENT
Dear all,
I have also read your ANE (38, pp 13991405) paper and I m currently trying to do some similar comparison involving SERPENT, CASMO5 and PARCS for UO2 PWR type lattice. I came to the same conclusions for the comparison of the homogenized parameters at the assembly level when the infinite spectrum is considered, i.e. that the cross sections agreement between CASMO5 and SERPENT is good (<1% relative difference) but that SERPENT underpredicts the diffusion coefficient by about 20%. However, when I turn on the fundamental mode calculation, the discrepancy becomes even worse, ~ 25 % underprediction...
Here is how I turned on the B1 calculation in my input:
=======
set nfg 2 0.625e6 % set up of macro group
ene egrid 1 1.E10 0.625E6 15.0 % set up of the energy grid
set fum egrid % set up of fundamental mode calculation
=======
The value of the CASMO5 predicted diffusion coefficient for the fast group in on par with the values reported in the ANE paper (1.40 against 1.45 in the ANE paper)...the issue lays in the diffusion coefficient predicted by SERPENT (1.10 with infinite spectrum, 1.06 with the critical spectrum).
Any idea what I m doing wrong?
Did anybody compare CASMO4 / CASMO5 homogenized parameters with SERPENT ones? if so, could you point me to some reference?
Thanks a lot!
Mathieu
PS: I put the entire input in attachement
I have also read your ANE (38, pp 13991405) paper and I m currently trying to do some similar comparison involving SERPENT, CASMO5 and PARCS for UO2 PWR type lattice. I came to the same conclusions for the comparison of the homogenized parameters at the assembly level when the infinite spectrum is considered, i.e. that the cross sections agreement between CASMO5 and SERPENT is good (<1% relative difference) but that SERPENT underpredicts the diffusion coefficient by about 20%. However, when I turn on the fundamental mode calculation, the discrepancy becomes even worse, ~ 25 % underprediction...
Here is how I turned on the B1 calculation in my input:
=======
set nfg 2 0.625e6 % set up of macro group
ene egrid 1 1.E10 0.625E6 15.0 % set up of the energy grid
set fum egrid % set up of fundamental mode calculation
=======
The value of the CASMO5 predicted diffusion coefficient for the fast group in on par with the values reported in the ANE paper (1.40 against 1.45 in the ANE paper)...the issue lays in the diffusion coefficient predicted by SERPENT (1.10 with infinite spectrum, 1.06 with the critical spectrum).
Any idea what I m doing wrong?
Did anybody compare CASMO4 / CASMO5 homogenized parameters with SERPENT ones? if so, could you point me to some reference?
Thanks a lot!
Mathieu
PS: I put the entire input in attachement
 Attachments

 uo2_asy_crit.zip
 SERPENT input deck
 (1.04 KiB) Downloaded 1140 times
 Jaakko Leppänen
 Site Admin
 Posts: 2134
 Joined: Thu Mar 18, 2010 10:43 pm
 Security question 2: 0
 Location: Espoo, Finland
 Contact:
Re: B1 fundamental mode calculation in SERPENT
The problem might be related to using only two energy groups for the B1 calculation. Try redefining energy grid "egrid" with something like 70 groups or so.
 Jaakko

 Posts: 34
 Joined: Thu Sep 22, 2011 5:27 pm
Re: B1 fundamental mode calculation in SERPENT
Just to make sure I m understanding that stuff correctly, the B1 cross sections will be collapsed to the group structure defined in the nfg card (in my case 2 groups) using the fine energy group structure defined in the ene card?
Thanks for your help!
Mathieu
Thanks for your help!
Mathieu
 Jaakko Leppänen
 Site Admin
 Posts: 2134
 Joined: Thu Mar 18, 2010 10:43 pm
 Security question 2: 0
 Location: Espoo, Finland
 Contact:

 Posts: 34
 Joined: Thu Sep 22, 2011 5:27 pm
Re: B1 fundamental mode calculation in SERPENT
I m now performing the B1 calculation with 70 energy groups (cas70 structure). The diffusion coefficient is much better with a max. relative error compared to CASMO5 equal to about 5% (in the fast group). I m still above the 2% reported in the ANE paper.
One of the potential reason for such a difference is the method used in CASMO5 which differs significantly from the one in SERPENT and HELIOS.
Could you confirm that in the ANE paper, the diffusion coefficients reported are the direct byproducts of the B1 calculation (and not coming for the transport corrected cross section)?
Thank you!
Mathieu
One of the potential reason for such a difference is the method used in CASMO5 which differs significantly from the one in SERPENT and HELIOS.
Could you confirm that in the ANE paper, the diffusion coefficients reported are the direct byproducts of the B1 calculation (and not coming for the transport corrected cross section)?
Thank you!
Mathieu