RE: Re: Deviations of flux caculation results by FLUKA and MCNP

From: Fasso, Alberto <fasso_at_slac.stanford.edu>
Date: Mon, 18 Mar 2013 04:25:12 -0700

2013 12:25:27 +0100 (CET)
X-Virus-Scanned: clamav-milter 0.97.6 at smtp2.mi.infn.it
X-Virus-Status: Clean
Sender: owner-fluka-discuss_at_mi.infn.it

Dear Walker,

A possible cause of disagreement, already pointed out by Paola, could be due to different cross section sets used by the two programs. FLUKA uses for Al ENDF/B-VIIR0: what did your MCNP version use?

Also, the energy of your source (14.1 MeV) corresponds to FLUKA energy group 15, between 13.84031 and 14.19068 MeV. But it is close to its upper limit. It could well be that part of the disagreement could be due to the difference between the cross sections averaged over the energy range of group 15 and the pointwise cross sections at 14.1 MeV, used by MCNP.
_Total_ Al cross sections as found in ENDF/B-VIIR0 vary only by 0.28 % between the two limits of the above energy range. But the difference between the _inelastic_ cross section at 13.8 and at 14.2 MeV is much larger (about 20 %). I didn't look at other cross sections. Keep in mind, however, that I am talking about interpolations.

In any case, I don't see many applications where a difference of 1.5 % could be of any importance (certainly not shielding calculations).
The evaluated cross sections used by both MCNP and FLUKA have certainly larger errors.

Finally, I noticed that you input 6248.61 cm3 as the volume of your cylinder shell, while it should be 6251.77 (40 pi (50^2 - 49.5^2)).
The difference is only of 0.05 %, but could partially contribute to the disagreement.

Best regards,

Alberto
________________________________________
From: walker [wdx456_at_mail.ustc.edu.cn]
Sent: Saturday, March 16, 2013 7:25 AM
To: paola.sala_at_mi.infn.it
Cc: 'fluka-discuss'
Subject: Re: Re: Deviations of flux caculation results by FLUKA and MCNP

Dear Paola,



I'm trying to solve the problem referred to about two months ago, as the original mail showed.



My first step is to test the program difference between FLUKA and MCNP, with single nuclide as well as simple geometry.

Conditions are as following:

---------------------------------------

Geometry: three concentric cylinders with a sphere black hole around.

Source: Isotropic 14.1MeV neutron source, filling inner cylinder of 'Geometry' mentioned above.

Physical quantity calculated: neutron volume flux.

Detect region: outer cylinder of 'Geometry' mentioned above.

Material: 13027.21c in MCNP and '27Al. Aluminium 27 SelfShielded, 296K' in flair.

---------------------------------------



Test results:

FLUKA: 2.4948E-05(flux), 0.04%(error)

MCNP: 2.5327E-05(flux), 0.07%(error)

Relative deviation: 1.5%.



Is the 1.5% deviation acceptable, please?



Besides, when the material changed to Fe, the deviation enlarged to 2.4% (FLUKA value smaller still). What about this diviation?



Input file of FLUKA and MCNP with material Al, was attached in this email.

Iím wondering if there is any negligence in the input that contributes to the deviation.





Best wishes

Walker









-----Original E-mail-----
From: "Paola Sala" <paola.sala_at_mi.infn.it>
Sent Time: 2013-1-31 15:14:13
To: wdx456_at_mail.ustc.edu.cn
Cc: paola.sala_at_mi.infn.it; fluka-discuss
Subject: Re: Re: Deviations of flux caculation results by FLUKA and MCNP



Hello

with 14 MeV neutrons in most materials one would expect something better than 10%, however

- self-shielding could make a big difference

- different choices of databases could make some difference

AND: if I understand correctly, you are testing ALL materials together in a complex geometry, therefore differences propagate themselves from one region to another, making it difficult to understand what and where really happens.

Regards

Paola



> 2013 14:23:03 +0100 (CET)

> X-Virus-Scanned: clamav-milter 0.97.6 at smtp2.mi.infn.it

> X-Virus-Status: Clean

> Sender: owner-fluka-discuss_at_smtp2.mi.infn.it<mailto:owner-fluka-discuss_at_smtp2.mi.infn.it>

>

>

> Hi Paola,

>

> The primary particle is neutron with energy 14MeV and the particle I'm

> scoring is also neutron.

> I didn't find any pattern in the diffferences(Really troublesome:£©).

>

> Maybe I should do more careful analysis to find out the true reason.

>

> Then,from your expert view,is it possible that 10% difference of FLUKA

> and MCNP flux results are the right results?(considering material lib

> and software differences)

>

> Thanks for your kindness!

>

> Best wishes,

> Walker

>

>

>> -----Original E-mail-----

>> From: "Paola Sala" <paola.sala_at_mi.infn.it> Sent Time: 2013-1-30

>> 15:39:26

>> To: walker <wdx456_at_mail.ustc.edu.cn>

>> Cc: fluka-discuss_at_fluka.org

>> Subject: Re: Deviations of flux caculation results by FLUKA and MCNP

>>

>> Dear Walker,

>> it is difficult to answer without more information:

>> What is your primary particle? Its energy? Which particles are you

>> scoring?

>> Do you see some pattern in the differences? For instance, do they

>> depend on the position , on the material, or other?

>> Regards

>> Paola

>> > Dear all,

>> >

>> >

>> >

>> > Recently, I was trying to calculate flux distribution of a model

>> > via

>> FLUKA

>> > and MCNP respectively (in order to verify the correctness of my

>> result).

>> >

>> >

>> >

>> > The model was made up of over 100 regions and the materials included :

>> C,

>> > V

>> > , Cr, Mn, Fe, Ta, W, H, He, Li, Pb, Ni, etc.

>> >

>> > About 50 regions' fluxes were scored.

>> >

>> >

>> >

>> > I tried several ways to decrease the relative error. For example,

>> > the

>> same

>> > volume values were input into FLUKA and MCNP for flux calculation;

>> > the primary particles were increased from 1.0e6 to 1.0e7 to 1.0e8;

>> > the selfshielded material in FLUKA was also chosen.

>> >

>> >

>> >

>> > However, the FLUKA and MCNP results seemed not to match ideally:

>> > about

>> 25

>> > regions had deviation less than 1%; about 20 ones had deviation

>> between 1%

>> > and 2 %; most of the left ones had deviation beside 10% with the

>> > max deviation 19%. Both the max statistical discrepancies of FLUKA

>> > and

>> MCNP

>> > are

>> > no bigger than 1%.

>> >

>> >

>> >

>> > I' m wondering if my calculation result is correct. What's the

>> reasonable

>> > flux deviation of FLUKA and MCNP? How about 10%?

>> >

>> >

>> >

>> > Looking forward to any suggestion.

>> >

>> >

>> >

>> >

>> >

>> > Best wishes

>> >

>> > Walker

>> >

>> >

>> >

>> >

>>

>>

>> Paola Sala

>> INFN Milano

>> tel. Milano +39-0250317374

>> tel. CERN +41-227679148

>>

>

>

>





Paola Sala

INFN Milano

tel. Milano +39-0250317374

tel. CERN +41-227679148
Received on Mon Mar 18 2013 - 13:46:01 CET

This archive was generated by hypermail 2.3.0 : Mon Mar 18 2013 - 13:46:04 CET