Dear FLUKA experts,
I used two different BIASING setups for my calculations. The BIASING card was used. With the first bias setting, the 21-card outputs a result of 2.8823E+07 (error, 3.678) for 99Mo activity and 4.0631E+07 (error, 0.2253 ) for tritium activity. With the second bias setting, the 21-card output results: 99Mo activity of 2.6017E+07 (error, 3.806 ), and tritium activity of 4.0758E+07 (error, 0.4865 ).The 99Mo results are highly biased, and although the errors are close in the output cards (3.678 and 3.806), the actual results are highly biased, by about 11% (( 2.8823-2.6017)/2.6017 = 0.11). But the tritium results are again almost identical ((4.0758-4.0631)/4.0631=0.0035). This is very strange and I’m not sure what this is asking? I hope an expert can help me. And with the second bias setting, after I increased the number of cycles and continued the calculation, the error for 99Mo increased instead (from 3.763 to 3.806), while the error for tritium decreased (from 0.5076 to 0.4865). Again, very strange.
Attached are the flair file and part of the output file for the first BIASING setup and the second BIASING setup, respectively.
The first BIASING setting：
W .flair (12.7 KB)
W _21_tab.lis (69.1 KB)
W _22_tab.lis (69.1 KB)
The second BIASING setting：
W .flair (12.4 KB)
W _21_tab.lis (71.8 KB)
W _22_tab.lis (71.8 KB)
Thank you very much for your help and answer! Looking forward to your reply. Wish you the best of luck in everything.
Very thankful!!!
Lin
Dear Lin,
thank you for your question. I will come back to you as soon as possible.
Cheers,
Giuseppe
Dear @gmazzola ,
Thank you very much. Looking forward to your reply.
Wish you all the best!
Lin
Dear Lin,
assuming that you are trying to look at the production of 99Mo coming from the 98Mo(n,gamma) channel (which I guess it represents the main contribute), then the real problem of your input comes from the missing biasing on the mean free path of electronuclear reaction, and, consequently, to a low number of neutrons generated in your simulation. Same reasoning is valid also for tritium production.
As described in note 3 in the manual entry of the PHOTONUC card, it is recommended to use PHOTONUC in combination with LAM-BIAS.
From a quick check, I got that, running 600k particles, with your input I get a value for 3H after 10h of 4.3836E+07 Bq with error 6.645%. Running the same input adding the LAM-BIAS card, the activation of 3H is 4.0293E+07 Bq with error 0.6571%.
An example of the use of the LAM-BIAS to decrease the mean free path of electronuclear reaction:
Considering now the fluctuation you see in the values and errors of your simulations. The reason is that, since the number of neutrons is low due to the missing LAM-BIAS card, the simulation’s results, even running a large number of primaries ( 125M? ), are far from convergence.
Hope this help you,
Giuseppe
Dear @gmazzola ,
Thank you very much for your reply. But in my model I don’t care about 98Mo(n,gamma)99Mo because I don’t set 98Mo in my target material.I describe the physical process briefly. The whole model is shown in figure 1.First my incident particle is electron . The electron reacts with the tungsten target (the gray part) and produces secondary photons, which continue to react with the tungsten target through photonuclear reaction to produce secondary neutrons. The unreacted secondary photons and the newly generated secondary neutrons pass through the polyethylene moderating material (green part), so the neutrons are moderated. Finally the secondary neutrons and secondary photons irradiate the molten salt target material (blue part) and the secondary photons and secondary neutrons react with the uranium in the molten salt to produce 99Mo. This is my understanding of the physical process, so please bear with me and correct me if there are any problems. You said that the number of neutrons is too small, in fact there are a lot of neutrons, more than 10^10 n/s.You also mentioned the PHOTONUC card and the LAM-BIAS card, and the error of the 3H was reduced from 6.645% to 0.6571% with the use of these two cards.
But sorry I still don’t understand why? As I said earlier, the 99Mo results were 2.8823E7 (error, 3.678%), and 2.6017E7 (error, 3.806%) at different BIASING settings. The errors are all below 5% (in Monte Carlo simulations, it is generally accepted that results with errors below 5% are more credible). So the errors of 3.678% and 3.806% are plausible. Both results are plausible, but why are these two results 10% different.
If one result has an error of 15% and one result has an error of 20%, then these two results differ by 10%. I see no problem with that. But now the two plausible results don’t match.
I am using BIASING cards, i.e. particle splitting and roulette, which is an unbiased BIASING setup. And the values I set are in line, 5^(-1)<R<5.
You suggested a new way of setting the bias, but I still don’t understand why my previous settings were problematic?
Thanks.
Lin
Dear Lin,
you are right, I have confused the material set for GH3535 with the one for the FUEL region (N.B. in your input the FLiBeZrU material is defined using atomic composition, is this what you really wanted?). Still, I assume the presence of neutron produced by photonuclear reactions is relevant for the generation of 99Mo in the fuel. Therefore, my suggestion to optimize your biasing setup with the LAM-BIAS card is still valid.
Regarding the difference you see in the Mo production. FLUKA gives back an estimation of the error for a specific result. Additionally, the error represents the 1-sigma of the Gaussian distribution of the expectation value around the true expectation value (check some slides for more details, if needed). Based on this, I would say the results you get with the two different biasing are not erroneous, they simply include each other in a 3-sigma interval. In conclusion, both biasing are not problematic and this is even reinforced from the correlated values you get for the 3H.
My suggestion for the LAM-BIAS is to eventually further optimized the biasing effect in your simulation in order to reduce even more the, already small, statistical uncertainty for 99Mo.
Cheers,
Giuseppe
Dear @gmazzola ,
Thank you very much for your help. I’ve understood a little bit. I will study the PPT you recommended. Thanks again for your help. Wish you all the best.
Lin