Features of GPUSPH

[gpusph]

Features of GPUSPH

[agnesLeroy]

Hi there,

Thank you for your interest in GPUSPH. To answer your questions, it is not possible to use variable particle size in version 4.1, and the upcoming version 5.0 will not include this feature either. The surface tension code as described in Wei et al. 2017 is not available in an official release yet, and it will not be part of 5.0. Maybe 5.1, or a later release. The mooring system was modeled using an analytical solution for non-elastic mooring lines. If you need the corresponding code, please let us know, we can send it to you. There is indeed a limit on the number of floating objects: the maximum is currently 16.

Best regards,
Agnès

[Dongxueyang]

Hi agnesLeroy:
You see there is indeed a limit on the number of floating objects:the maximum is currently 16.
You mean I can only generate 16 floatings in a model. Such as only 16 spheres or cubics? And I can not find any example about floating in the pragram. Could you send me a simple example to further study?
Thanks a lot.
dong

[agnesLeroy]

Hello,

Yes, at the moment you can only include 16 different floating bodies (spheres, cubes, etc.).
There are several examples in GPUSPH that include floating bodies: for example, you can look at BuoyancyTest (it uses dynamic boundaries) or CompleteSaExample (it uses semi-analytical boundaries).

Cheers,
Agnès

[Dongxueyang]

Thanks a lot . But why the limit number is 16? If I want to simulate sand it seems impossible.
What facter is related to the limit number?

[agnesLeroy]

Hello,

For simulations involving very small objects we prefer to create models (for example, for sand we have an ngoing work on multi-phase non-Newotnian flows, that shall be merged into next soon). Otherwise, the particle size in the simulation is so small that computational times become too high, even on GPU.

It is true we need to relax that limit, I'm not sure how complicated this is.

Cheers,
Agnès

[Dongxueyang]

Thanks for your reply soon.
I have another question. Is there any non-Newotnian flows examples in the program, and when will the next version will be released?

[agnesLeroy]

Hello,

At the moment there is no example of a non-Newtonian flow: this feature is still at development stage and when it will be merged into next there will be an associated example. I can let you know about it when it happens. The version 5.0 is planned to be released next spring, but before that we're making the stabilised developments public on the branch next, so you can try it out before the version release if you need to.

Best regards,
Agnès