Difference between MCNP and FLUKA in the tritium production of water

Physics, Transport and Magnetic Fields
1

Dear experts,

When I calculate the tritium production of thermal neutrons impinge water, I found the results between MCNP and FLUKA has a difference about an order of magnitude. I assume that the deuterium in natural water would explain this gap. I adjust the atom fraction of water, but the result didn’t occur some difference.
The question is that I want to know how FLUKA deal with the multi_capture reaction like H(n,g)D,D(n,g)T.

image
image842×157 20.4 KB

3

Hi Yongguang,

thank you for your question. In FLUKA it is not possible to simulate a chain of multiple reaction after each other. In other words, in pure H, you would be able to simulate only H(n,g)D.
Adding deuterium to water, as you mentioned, would allow you to simulate D(n,g)T interactions.

Could you perhaps share the inputfile, the *.out file and your scorings? I can check if there anything weird in there.

Cheers,
Daniele

4

Hi Daniele @ dcalzola,

Thank you for your answer. Here is my inputfiles and out file. I used the default H and O element to define water in the water_cube_1.inp while added extra D element in the water_cube_2.inp. But it seem the result is quite close.
water_cube_2.lis (1.7 KB)
water_cube_2.inp (4.4 KB)
water_cube_1.lis (1.7 KB)
water_cube_1.inp (3.2 KB)

Cheers,
Yongguang

5

Hi Yongguang,

Your files seem correct. Let me try to assess the situation with you:

  1. If my calculations are correct, your beam contains 1.8E22 thermal neutrons.
  2. After your irradiation time, you expect 4.32E20 H(n,γ)D reactions in your sample. I agree with that. However, there may be a typo in your table—should your yield for D(n,γ)T be 4.2E-9 [nuclide/cm³/pr]?

Aside from a non-negligible Monte Carlo uncertainty (8.9%), I do not see anything significantly wrong with your simulation. However, I still encourage you to use the pointwise treatment for neutrons (FLUKA libraries for low-energy neutron pointwise interaction cross sections | The official CERN FLUKA website).

To investigate potential issues with the D(n,γ)T cross section, I recommend performing a simulation where, instead of natural water, you directly use heavy water (D₂O) as the medium.

Now, regardless of any differences you may observe between the two codes, let me highlight what I believe to be a more fundamental issue:

  1. You start the simulation with 1 gram of water as the material. In this gram of water, you have 6.7E22 H and D atoms, of which only 1E19 are D atoms.
  2. After the total irradiation time, you have produced an additional 4.32E20 D atoms. This means your material composition changes significantly during irradiation. Even with a perfect cross-section, this would lead to a severe underestimation of tritium production. The only way to address this issue is to split the calculation into several sub-simulations.

Let me know if this helps.

Cheers,
Daniele

6

Hi Daniele,

First of all, thank you for pointing out my typo in my table. It actually is 4.2E-9[nuclide/cmc/pt] while 6.91E-6 is the yield of 17O.

Then, about the D(n,γ)T corss section issue, I simulated thermal neutrons injected the pure heavy water to produce tritium before a couple days ago and found the gap of FLUKA and MCNP is small in the total amount of the tirtium where the MCNP’s reslut is 1.11E9 and FLUKA’s result is 1.03E9. It may proves the D(n,γ)T cross section correct.

As for the material compsition problem, I have derivated a simplest theoretic formula with taking into account it and neglecting the effect of neutron scattering and energy loss. The value of tritium comes to 5.782E15 . I should reconsider the effect of material compsition.

Lastly, you said that In FLUKA it is not possible to simulate a chain of multiple reaction after each other, so is it reasonal for me to assume it can’t produce tritium if I use pure H element to simulate? But the fact is that it really appears. So may be the element H will contain the 0.003% D in FLUKA.

So appreciate your massage and comments.

Cheers,
Yongguang

7

Hi Yongguang,

The value of tritium comes to 5.782E15

This is indeed quite far away from your calculation in both codes.

Lastly, you said that In FLUKA it is not possible to simulate a chain of multiple reaction after each other, so is it reasonal for me to assume it can’t produce tritium if I use pure H element to simulate?

This is indeed correct. In your geometry, you selected natural Hydrogen as material (and cross section). You can find more references about the materials and cross sections:

  1. Materials: 5.2. Pre-defined materials — FLUKA Manual. In particular, the material composition assumed for Hydrogen is the NIST one (Atomic Weights and Isotopic Compositions for Hydrogen), slightly different than the one you picked.

  2. Cross sections (with the multi-group treatment that you choose): 10.1.3.2. Available cross sections — FLUKA Manual

Cheers,
Daniele

8

Hi Daniele，

Thank you for your help. I would try other ways to solve this question.

Cheers,
Yongguang