Fuel Assembly Burn-up calculation K-eff value

User Support
1

Hi all,

I am trying to investigate a 17*17 PWR fuel assembly depletion problem. The depletion time is one refuelling cycle which is 18 months. The depletion step size was divided have a very short step (1 hour) at the beginning of the calculation, and a bit longer steps (one month)*18 at the later part of cycle time. The results of K-eff values decrease but have some point waving. The k-eff will increase at some steps.

I can’t figure out why it looks like that. In my mind, the k-eff value should smoothly decrease. The k-eff values in each step and plots would be attached in this mail.
Please contact me if you have any idea why that could happen and how to optimise that.

Thanks,

Yiming

results.png
results.png389×611 13.2 KB

setting.png
setting.png1120×775 104 KB

K-effs.png

2

Hye Yiming

Have you tried any other integrator beside predictor integrator?

3

Hi Yiming,

Looking at the uncertainties on your k value, I don’t think this is purely a statistical artifact. I do find it interesting that your fuel assembly is pretty subcritical, but that shouldn’t be causing this significant of an issue.

Without more information on your model, my initial reaction would be to check the convergence of your fission source for the fresh model using the Shannon entropy setting. I would also like to repeat Khurrum’s comment on using other time-integration schemes (CE/CM, CE/LI, etc.) The predictor is conditionally stable, meaning sufficiently large time steps can cause instabilities and inaccuracies. If you are able to use one of the other integrators, please reach out with an update.

Regards,

Andrew

4

Hi Andrew, Khurrum,

Thanks for your advice. I will try the other integrators and update the results here.

Thanks,
Yiming

Andrew Johnson <1drew.e.johnson@gmail.com> 于2020年7月7日周二 下午12:23写道：

5

Hi,
After the above discussion, I rebuild the 17*17 nuclear assembly model geometry with 3D pitch universes Rectangular Lattices function. Now the results of eigenvalue calculation look good, even with the predictor integrator. I also used the CE/CM and CE/LI integrator as well. The results look very close. I put the calculation results in this email. Hope that could be helpful.

Meanwhile, I have another question. Due to the application of ‘diff_burnable_mats’ argument, there are many fuel materials in the fuel assembly model. But it brings some challenge to the analysis of the results. The first required parameter in the openmc.deplete.ReactionRates [Python API is the local_mats(list of str) Material ID. But there are maybe some materials listed in the material.XML file. Is there any method to get the reaction rate or atom number in all the burnable materials at the same time? Thanks for your advice.

Best,
Yiming

Zhong Yiming <zhongyming123@gmail.com> 于2020年7月7日周二 下午3:25写道：

predictor-keff.png

CELI-keff.png

CECM-Keff.png

6

is there any burnable nuclides, like Gd, in your case?

7

Hi Tianxiong,

In this case, there are no Gadolinium rods inside the fuel assembly. I also did a burn-up calculation of the reactor core contains the Gd rods. The initial k-eff will be even lower.

Best,
Yiming

8

@YimingZ Currently there is no function in the Python API for getting a reaction rate / atom density for all burnable materials at the same time. If you have a list of all the materials / material IDs, you could use a Python list comprehension though to get all rates / densities. Something like:

mat_ids = [1, 2, 3, 9, 13, 21]  # ... or whatever the material IDs are
results = openmc.deplete.ResultsList.from_hdf5(...)
u235_densities = [
    results.get_atoms(str(mat_id), "U235")
    for mat_id in mat_ids
]
1 Like