Understimating neutron flux

Hey everyone,

I am using GEANT4 simulation for finding neutron fluxes from various targets when electron beam strikes them. I see a pattern that the value of neutron flux that I am getting through GEANT4 is less than the literature suggests (various work done and published in renowned journals).
I am told MC simulations tend to underestimate the absolute values of particle fluxes sometimes by a power of 10. I am not very convinced. Does anyone else face this issue?
I am using QGSP_BIC_HP physics list and for material definition I am using NIST data.

Any help would be much appreciated.

Hello,

we have so far only one model of electro-nuclear interation and only one cross section, which are used in all Physics Lists. Secondary neutron transport a bit different between QGSP_BIC and QGSP_BIC_HP. Would you be able to show neutron spectra after a target this known data and Geant4 prediction?

VI

Hi,

my problem actually seems to be related to this thread, so I’m posting here:

I am just starting to perform simulations in Gate / Geant4 including Neutrons. Concerning that, I performed a simulation with 15MeV photons impinging on a thin (10µm) target. The simulation in Gate / Geant4 yields about a factor 3 less neutrons than a simple calculation using the respective crosssection. So far I tested W and Pb.

In a publication „Quantification of the validity of simulations based on Geant4 and FLUKA for photo-nuclear interactions in the high energy range“ it is stated that Geant4 underestimates Neutron production. Is this really true? I have been using both the QGSP_BIC_HP and the QGSP_BERT_HP physics list with the same result in terms of the Neutron production rate.

Best regards

Johannes

Hi,

are there any suggestions about how to improve the accuracy of simulations with neutrons in Geant4?

Any help / suggestion would be greatly apperciated!

Best regards

Johannes

The factor of three was observed in this bug report:

https://bugzilla-geant4.kek.jp/show_bug.cgi?id=1680

Could you please try the analysis described in that report. With example Hadr03, count the number of photonuclear interactions and the number of photoneutrons. The ratio for your energy and material should be close to 1.

with the latest development version, the ratio is rather ~1/2 …
gamma.mac.txt (304 Bytes)

With the environment variable G4CASCADE_CHECK_PHOTONUCLEAR set, there is a check if the photonuclear reaction produced only gammas. If so the reaction products are resampled until the check passes. This is limited to incident particles with energy < 50 MeV. The deexcitation code otherwise doesn’t handle the giant dipole resonance.

The code is here:
https://geant4.kek.jp/lxr/source/processes/hadronic/models/cascade/cascade/src/G4InuclCollider.cc#L249

I ignored the existence of this environment variable …
Why this feature is only optional ? why it is not done by default ?
And why an environment variable and not an usual C++ function or flag ?

Indeed. Let’s discuss with hadronics people. I opened a bug report about the missing documentation.
https://bugzilla-geant4.kek.jp/show_bug.cgi?id=2229

Additionally, according to Dennis’s comments in bugzilla 1680, we should try LEND data set. This should be used if G4EmExtraPhysics is registered and the data is downloaded. My attempts today were unsuccessful. I see it is deactivated by default in G4EmExtraPhysics. Note the link below is out of data–follow the link in Geant4 download page.

from G4EmExtraPhysics:
Low Energy Nuclear Data (LEND) model for gamma nuclear interactions.\n The LEND model needs data files and they are available from ftp://gdo-nuclear.ucllnl.org/GND_after2013/GND_v1.3.tar.gz.\n Please set the environment variable G4LENDDATA to point to the directory named v1.3 extracted from the archive file.

The answer to the original question is to investigate LEND. In Hadr03 PhysicsList.cc, use
RegisterPhysics(new G4EmExtraPhysics())
and remove the others. Otherwise LEND looks to be part of the main physics lists like QGSP_BIC etc.

Download the files and set G4LENDDATA envvar.

In the macro set
/physics_lists/em/LENDGammaNuclear true

Results look reasonable (nearly all interactions produce >= 1 neutron, but I haven’t looked up the expected values).

The run is 100000 gamma of 15 MeV through 100 m of G4_W (density: 19.3 g/cm3 )

Process calls frequency:
photonNuclear= 100000

MeanFreePath: 38.911 cm ± 38.831 cm massic: 750.98 g/cm2
CrossSection: 0.0257 cm^-1 massic: 133.16 um2/mg
crossSection per atom: 406.5 mbarn

Verification: crossSections from G4HadronicProcessStore:
photonNuclear= 133.64 um2/mg 407.99 mbarn
total= 133.64 um2/mg 407.99 mbarn

List of nuclear reactions:

                     gamma + W180 --> N gamma or e- + W180:       1   Q = 0.013293 eV 
                           gamma + W180 --> neutron + W179:     123   Q = -13.545 MeV
                         gamma + W182 --> 2 neutron + W180:     638   Q = -14.787 MeV
                     gamma + W182 --> N gamma or e- + W182:     184   Q = 0.0056906 eV 
                           gamma + W182 --> neutron + W181:   25936   Q = -13.552 MeV
                         gamma + W183 --> 2 neutron + W181:     926   Q = -14.527 MeV
                     gamma + W183 --> N gamma or e- + W183:      42   Q = 0.0017366 eV 
                           gamma + W183 --> neutron + W182:   13700   Q = -13.068 MeV
         gamma + W184 --> N gamma or e- + 2 neutron + W182:    9586   Q = -8.5215 MeV
         gamma + W184 --> N gamma or e- + 2 neutron + W183:    5644   Q = -8.9998 MeV
           gamma + W184 --> N gamma or e- + neutron + W183:   14564   Q = -10.583 MeV
  gamma + W184 --> N gamma or e- + proton + neutron + W183:       1   Q = -3.5572 MeV
                         gamma + W186 --> 2 neutron + W184:   19646   Q = -13.391 MeV
                     gamma + W186 --> N gamma or e- + W186:      42   Q = 0.0014322 eV 
                           gamma + W186 --> neutron + W185:    8967   Q = -11.665 MeV

                      number of gamma or e- (ic): N = 2 --> 16

List of generated particles:
W179: 123 Emean = 8.8966 keV ( 557.04 eV --> 31.703 keV)
W180: 639 Emean = 1.9281 keV ( 35.52 eV --> 7.4462 keV)
W181: 26862 Emean = 8.5172 keV ( 8.2666 eV --> 39.994 keV)
W182: 23470 Emean = 11.189 keV ( 2.4015 eV --> 130.01 keV)
W183: 20251 Emean = 13.76 keV ( 0.0014155 eV --> 51.927 keV)
W184: 19646 Emean = 9.3703 keV ( 19.486 eV --> 44.554 keV)
W185: 8967 Emean = 18.74 keV ( 147.5 eV --> 43.913 keV)
W186: 42 Emean = 959.21 eV ( 58.98 eV --> 2.4659 keV)
e-: 428 Emean = 164.39 keV ( 18.476 keV --> 846.91 keV)
gamma: 82746 Emean = 1.0653 MeV ( 20.708 eV --> 14.874 MeV)
neutron: 136171 Emean = 1.4782 MeV ( 25.885 eV --> 18.27 MeV)
proton: 1 Emean = 5.7388 MeV ( 5.7388 MeV --> 5.7388 MeV)

but there are many warnings:
-------- WWWW ------- G4Exception-START -------- WWWW -------
*** G4Exception : had012
issued by : G4HadronicProcess:CheckResult()
Warning: Bad energy non-conservation detected, will re-sample the interaction
Process / Model: photonNuclear / LENDorBERTModel
Primary: gamma (22), E= 15, target nucleus (74, 184)
E(initial - final) = -169446 MeV.

*** This is just a warning message. ***
-------- WWWW -------- G4Exception-END --------- WWWW -------

Hi Daren,

Thanks for the informations.
GammaNuclearPhysics is a PhysicsConstructor for Hadr03 (and others).
Attached, a variant which uses LEND. I know that it compiles, but I did not yet try to execute it.

GammaNuclearPhysics.cc.txt (3.1 KB)

Hi Maire and Daren,

many thanks for your comments. I tested the Hadr03 example with the LEND-variant of GammaNuclearPhysics.cc. The photoneutron yield is comparable to what I expect from my calculation.

Best regards
Johannes

If the problem is understood we may close this discussion.