Criticality "Keff" result mismatch with MCNP4C model

Validation and Verification
1

Hello, I have a TRIGA reactor model, 95 fuel rods and 5 fuel follower rods and 1 air follow control rods. I have two cases of Criticality study.

  1. When the control rods are not inserted. So the “fuel section top” of the control rods is in align with the fuel rods fuel section top.
  2. When some portion of the control rods (23cm) are inserted in the core. So, some section of the control rods control section is in align with the fuel rods fuel section top.

Now, on both cases the material composition is the same, just in the second case, 6 rods are 23cm lower in z-axis. My issue is in the criticality calculation In the first case I am getting a combine Keff of 1.067, But I am supposed to get a Keff of 1.077, verified by the experiment and MCNP4C model. Now, in the second case with the control rod inserted position the Keff, I am getting is 0.9975, which is same as the experiment(1.00) and MCNP4C(0.99749).

So, the Keff I am getting in the first case has a significant difference with the experiment and MCNP4C analysis, But not in the second case. Is this because of the continuous energy cross section library for MCNP4C is ENDF/B-VI and in my case ENDF/B-VII.1? Can anyone give me a pointer what might be the reason in Keff difference the first case? Any guide is appriciated. Thanks.

2

One possibility is that the Monte Carlo simulations are not statistically converged. How many batches and particles per batch were used for the OpenMC and MCNP calculations? What are the reported uncertainties on the keff values?

3

Hello, Dr. In MCNP4C the estimated statistical error 1σ was reduced below 0.03% upon 3000 cycles of iteration with 3000 particles per cycle. I tried that with openmc with 3000baches 50 inactive batches and 3000 particles per batch and the result is the same as before. I even tried increasing the number of particles to 30000. The result hasn’t improve. Still, in both cases, I have 0.01 difference with MCNP4C and the experimental value. Anything Else I can check?

Thank you, Dr.

4

Hi Sharif,

If your MCNP4C calculations are using ENDF/B-VI and your OpenMC calculations are using ENDF/B-VII.1, there should be no expectation that the results will match exactly. The only way to get a fair comparison is to use the same cross section library for both codes.

Best regards,
Paul

5

Hello, Dr, paul. How to get ENDF/B VI for openmc? I was using openmc-get-nndc-data command which by default download version VII.1.

6

Assuming you are using the 66c cross sections from MCNP, something like this script should work:
https://gist.github.com/paulromano/5ba4cde5b33e35366074531186ae46d6

Note that it doesn’t include thermal scattering data, but you can use the openmc.data.ThermalScattering class to convert S(a,b) tables from ACE files as well.

One other caveat – it looks like Be9 from ENDF/B-VI.6 uses an ACE secondary energy distribution law that OpenMC does not support; hopefully you don’t have any Be9 in your model.

7

Hello, Dr Paul. I am not working with MCNP. I am only working with OpenMC. The MCNP result is from a published work. Can you guide me how I can download the ENDF/B-VI cross section library for OpenMC to use? Thanks.