Re: [sparta-users] ERROR: Collision cell volume is zero

[<pi...@rt...>]

Thank you Savio,

I will try your suggestion of changing the surface resolution of my 
geometry right away.

The modification of the code you suggest sounds like a more pragmatic 
approach however. I am curious to know if you could make your 
modification available. Having the number of deleted particles available 
though the "stats" command would be ideal. I am curious to know how you 
have implemented that.

Pierre

On 2024-09-19 15:50, Poovathingal, Savio wrote:

> Pierre,
> 
> It happens sometimes with axis-symmetric simulations especially with 
> when the ratio of grids (high density) to surface elements is small 
> (small grids and large elements). It's a general floating-point issue 
> where the surface cannot be infinitely watertight. Essentially, 
> occasionally, a particle tunnels through inside the body. Our solution 
> has been to tag the molecule for deletion but we also count/keep track 
> of it. If we are deleting 1 particle in many timesteps, it's okay.  If 
> we delete too much, we know it's a bad simulation. We reset our 
> surface-element resolution.
> 
> If you go into collide.cpp, you can hack it a little bit for your need. 
> Alternatively, make surface elements of different resolutions for 
> different densities, which could fix the problem.
> 
> Savio
> 
> -------------------------
> 
> From: pi...@rt... <pi...@rt...>
> Sent: Thursday, September 19, 2024 4:19 AM
> To: Sparta Users <spa...@li...>
> Subject: [sparta-users] ERROR: Collision cell volume is zero
> 
> You don't often get email from pi...@rt.... Learn why 
> this is important [1]
> 
> CAUTION: External Sender
> 
> Dear SPARTA users/developers,
> 
> I am having the following error with many of my simulations:
> 
> _ERROR on proc 18: Collision cell volume is zero 
> ~/workspace/sparta/src/collide.cpp:481)_
> _ _
> 
> The cases model an Earth reentry from 110 km to 72 km for a 1-meter 
> diameter sphere using a 2D axisymmetric computation. All my simulations 
> use the same geometry file, but the free-stream conditions change. I 
> experience no issues from 110 km down to 86 km - the less challenging 
> cases. However, the problem starts at 82 km and below.
> 
> The error occurs randomly after a few hundred thousand iterations (at 
> 82 km, 78 km), and sooner - after just a few tens of thousands of 
> iterations - at 72 km, during the "run" command. Occasionally, 
> adjusting a single parameter, like FNUM by a small factor, resolves the 
> issue or causes it to appear sooner or later.
> 
> I have noticed that the problem is more likely to occur with a large 
> number of particles (approximately 1e8) and, on average, arises sooner 
> with more particles. This suggests to me that the error might be caused 
> by a rare, unhandled event in the code. However, I hope I am wrong and 
> that there is another way to resolve the issue.
> 
> Has anyone else encountered this problem, and if so, how did you fix 
> it?
> 
> Thank you in advance and best regards,
> Pierre


Links:
------
[1] https://aka.ms/LearnAboutSenderIdentification