"Branching" geometry

Suggest new ideas here
Post Reply
kaartinen
Posts: 8
Joined: Wed Aug 01, 2018 7:53 am
Security question 1: No
Security question 2: 93
Location: Loviisa NPP

"Branching" geometry

Post by kaartinen » Fri Feb 26, 2021 11:21 am

I was wondering if anyone had ever developed or thought about a branching geometry.

By branching I mean for example a situation where a region of interest is far away or is hard to reach from the source (gamma calculations) and usually heavily involves the use of weight windows. When you have multiple different geometries that you would like to study for this region of interest usually you have to run multiple calculations. What if it was possible to set up a cell that has two or more different configurations like filling the cell with different parallel universes. When a particle enters this special branching cell, the simulation branches giving you two or more different sets of detectors (or tallys).

A concrete example of this would be testing different types of shielding configurations.

A picture to explain:
Image

Ville Valtavirta
Posts: 539
Joined: Fri Sep 07, 2012 1:43 pm
Security question 1: No
Security question 2: 92

Re: "Branching" geometry

Post by Ville Valtavirta » Fri Feb 26, 2021 2:58 pm

Hi,

I think it's an interesting idea and could certainly be useful in some situations. I think adding support on such a branching geometry in the transport simulation itself might require quite extensive changes, however.

I think currently the best approach to achieve something similar would be to set up a simple surface to surround the region of interest and use det ds SURF -1 and det df to save all particles that go through the surface to the region.

This source can then be used in multiple fixed source simulations focusing on the region of interest with different contents applied for the region in each case.

Since Serpent can handle the simulation in parts using this approach, it would certainly be possible to make it all happen internally, but considering the added value gained by a fully internal scheme versus the implementation effort, I think we will most likely rely on the current multi-part approach for the foreseeable future.

-Ville

Post Reply