Reef3D stalling sometimes with BiCGStab solver

KevinMcTaggart · July 5, 2023, 13:13

I'm running Reef3D on Windows using Cygwin. Some cases, including some of the tutorials, will stall after a large number of time steps. I using the default N 10 14 BiCGStab (HYPRE Struct). Might a different solver or something else be able to prevent stalling? I'm running predictions of ship resistance.

valgrinda · July 7, 2023, 07:37

The solver should only stall in cases of numerical instabilities. If they are not present, then the problem is probably related to running on Windows. From user feedback, my impression is that most of the Linux on Windows solutions show the occcasional problem.

Which 6DOF solver are you using? Habe you started to use the direct forcing version?

KevinMcTaggart · July 7, 2023, 08:34

Yes, the numerical instabilities could quite possibly be caused by Cygwin.

For 6DOF, I'm using the default for X 13, which is 2 Continuous direct forcing.

Aside from the occasional stall, the code is working very well for computing steady flow around a ship, including free sinkage and trim.

valgrinda · July 7, 2023, 08:41

Ok, good to hear that you are using the direct forcing version. Right now, the model works really well for inertia driven cases, see here: https://www.dropbox.com/scl/fi/fr3m2...7cnrf0d6a&dl=0 . We are still working on and testing viscous flow dominated cases.

July 5, 2023, 13:13	Reef3D stalling sometimes with BiCGStab solver	#1
KevinMcTaggart New Member Kevin McTaggart Join Date: May 2018 Posts: 15 Rep Power: 8	I'm running Reef3D on Windows using Cygwin. Some cases, including some of the tutorials, will stall after a large number of time steps. I using the default N 10 14 BiCGStab (HYPRE Struct). Might a different solver or something else be able to prevent stalling? I'm running predictions of ship resistance.

July 7, 2023, 07:37		#2
valgrinda Super Moderator Hans Bihs Join Date: Jun 2009 Location: Trondheim, Norway Posts: 403 Rep Power: 19	The solver should only stall in cases of numerical instabilities. If they are not present, then the problem is probably related to running on Windows. From user feedback, my impression is that most of the Linux on Windows solutions show the occcasional problem. Which 6DOF solver are you using? Habe you started to use the direct forcing version? __________________ Hans Bihs Team REEF3D www.reef3d.com

July 7, 2023, 08:41		#4
valgrinda Super Moderator Hans Bihs Join Date: Jun 2009 Location: Trondheim, Norway Posts: 403 Rep Power: 19	Ok, good to hear that you are using the direct forcing version. Right now, the model works really well for inertia driven cases, see here: https://www.dropbox.com/scl/fi/fr3m2...7cnrf0d6a&dl=0 . We are still working on and testing viscous flow dominated cases. __________________ Hans Bihs Team REEF3D www.reef3d.com

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[foam-extend.org] simpleFoam examples crush in foam-extend-5.0	yeharav	OpenFOAM Installation	6	June 17, 2024 01:51
fluent divergence for no reason	sufjanst	FLUENT	2	March 23, 2016 17:08
3d vof	Smaras	FLUENT	2	February 19, 2013 07:58
Working directory via command line	Luiz	CFX	4	March 6, 2011 21:02
why the solver reject it? Anyone with experience?	bearcat	CFX	6	April 28, 2008 15:08

July 7, 2023, 08:34		#3
KevinMcTaggart New Member Kevin McTaggart Join Date: May 2018 Posts: 15 Rep Power: 8	Yes, the numerical instabilities could quite possibly be caused by Cygwin. For 6DOF, I'm using the default for X 13, which is 2 Continuous direct forcing. Aside from the occasional stall, the code is working very well for computing steady flow around a ship, including free sinkage and trim.