Different filters giving different results for she same score

User Support
1

So, out of curiosity I started playing with different filters for scoring flux across an AP1000 assembly when I found that the values differ in each of the filters.

I tried 2D meshfilter, 3D meshfilter and distributed_cell. Though It was expected the results would not going to be same for meshfilters and distributed_cell. But why they are not same among the meshfilters.

I made a 2D meshfilter of (200,200,1) and a 3D meshfilter of (200,200,200). Then took a slice along Z axis for the 3D mesh but as per my comprehension, I could not get the same value as the 2D mesh. What should be the reason? Which approach is better taking a 2D mesh or 3D then slicing?

I run the simulation for 200 batches with 59 inactive batches and 100k particles

Oh I also used an axial 1D mesh for axial peaking factor calculation and again I compare that value with the value of a legendre filter (order 10) being applied on the Z axis. And the value seems off a noticeable amount. So, which approach should be preferable?

Thanks in advance :smiley:

2

Hi @Allamanda,

Welcome to the forum and thanks for posting!

To answer this question thoroughly we’ll need a good deal of additional information about the model. But generally, a distributed cell filter is going to give you results that correspond to cell region boundaries as opposed to the mesh boundaries, so unless the model’s geometric cells are exactly aligned with the mesh elements, I’d expect that the results will be different than those of the tally using a mesh filter, yes.

As for the 2D vs. 3D mesh, the additional dimension in the 3D mesh is going to separate the tallies into 200 additional regions along Z, splitting up the statistical contributions that are all attributed to the same x,y location into 200 additional bins. So in this case I’d also expect that the 2D and 3D mesh will give different results.

Overall, these tally approaches are valid for different purposes and it really depends on what information you’re trying to extract from the simulation. Can you shed some light on what kind of analysis you’re trying to do?

-Patrick