BEAMS3D: Removed beam multi-pass option #345
Conversation
There was a problem hiding this comment.
The deposition works as expected, changing the allocation between deposited/port/shinethrough particles slightly for the tangential beams at high energy. However, I would propose fixing the bug by moving the 'end do' statement of the multi-pass loop to the end of the function, before the return statement. From what I can tell, this recovers the desired behavior:
Left: old multipass, middle: this PR, right: old multipass with the proposed fix
|
@kudav if you appove the merge, we can base a new pull request with your proposed modification for testing. |
|
Thanks for addressing this issue! It looks like the middle image still has one particle from the yellow Q6/8 beam and about 5 particles from the blue Q3 beam which have escaped. So there's still something funny happening. |
|
I agree, and I dont want to rule out a bug with the wall collision algorithm, however note in the figure above that the red contours mark the plasma and not the wall, which in this case is a detailed 3D wall and not drawn. I think singular particles can pass through the ports (which are not closed) opposite to the NBI. |
The pull request removes the multi-pass beam option from BEAMS3D. The multipass option was implemented by @kudav to deal with ASDEX-U tangential beams which 'scrapped' the inner stack. Here the beamline would allow some particles which are not ionized on the first pass to reenter the plasma from the inboard side. Unfortunately the implementation failed to properly catch the wall in some cases for shine-through particles. In particular @cfe316 demonstrated that particles for more radial beams would often find themselves on the other side of the torus before ionizing or being detected as wall hits. For this reason the modifications are being removed.
The ability to catch re-entrant particles is a worthwhile modificaiton to the code but must be done more rigerously. An issue related to this topic with a proposed test problem (not ASDEX-U) will be generated to properly develop and test such a feature.
This should close #332