Flow3D Mesh Assessment

bfox48 · April 25, 2012, 17:59

I am new to using CFD software and am building my first model in FLOW3D. I had a question regarding assessment of mesh quality. It doesnt appear flow3d has any built in diagnostic tools regarding the quality of a finished mesh... or can someone tell me if I am not looking in the right place. In addition I am trying to model a situation where I have a wide natural channel that constricts laterally into a structure with a large width constriction and transition to a steep bedslope (~0.1). Any general recommendations on how to transition my mesh sizes at this abrupt transition?

My plan was to have a single coarse mesh block covering the entire spatial domain, then nest a block with cells half the size of the coarse mesh. Then possibly nesting 1 or 2 more blocks around the structure where there will be large gradients. This would effectively reduce my mesh size by a factor of 8 at each of the nested interfaces.

Any thoughts?

JBurnham · April 25, 2012, 18:22

The general approach sounds right. Use fixed planes in your outer blocks to force the mesh to match gridlines with the extents of the nested blocks. Make sure the nested block sizes are 1/2 the larger block sizes, and that the extents are even multiples of the nested and outer block cell sizes. The idea is to get all the gridlines perfectly aligned: even a little bit of variation will cause some momentum diffusion, maybe negligible, maybe not. The closer to cubes your cells are, the higher the quality of the mesh, but up to 3:1 aspect ratios are often considered acceptable. Avoid placing mesh boundaries (like the nested block extents) in regions of high gradients: put them out before the wake structures begin to form.

bfox48 · April 25, 2012, 18:32

Thanks...

Also one more dumb but simple question. When looking at the simulation dashboard and the graphs of the different variables there are no units. Can someone tell me what these are in... and most important is percentage in decimal for or whole percents (0.5 vs 50%)?

JBurnham · April 25, 2012, 18:36

No prescribed units in FLOW-3D. All input and output variables are in the same length/mass/time system. If you input density as the value that corresponds to units of kg/m^3, then your length units must be m, and your viscosity units must be kg/m/s. Pressure outputs will then be kg/m/s^2 = Pa, and turbulent mixing length will be reported in m. Same holds true for ft/slugs/s and cm/g/s systems. On the dashboard, when % is listed on the vertical axis, 0.5 means 0.5% (half a percent).

April 25, 2012, 17:59	Flow3D Mesh Assessment	#1
bfox48 New Member brian Join Date: Apr 2012 Posts: 25 Blog Entries: 1 Rep Power: 14	I am new to using CFD software and am building my first model in FLOW3D. I had a question regarding assessment of mesh quality. It doesnt appear flow3d has any built in diagnostic tools regarding the quality of a finished mesh... or can someone tell me if I am not looking in the right place. In addition I am trying to model a situation where I have a wide natural channel that constricts laterally into a structure with a large width constriction and transition to a steep bedslope (~0.1). Any general recommendations on how to transition my mesh sizes at this abrupt transition? My plan was to have a single coarse mesh block covering the entire spatial domain, then nest a block with cells half the size of the coarse mesh. Then possibly nesting 1 or 2 more blocks around the structure where there will be large gradients. This would effectively reduce my mesh size by a factor of 8 at each of the nested interfaces. Any thoughts?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[ICEM] surface mesh merging problem	everest	ANSYS Meshing & Geometry	44	April 14, 2016 06:41
[snappyHexMesh] non-smooth mesh	Svensson	OpenFOAM Meshing & Mesh Conversion	11	January 18, 2012 09:13
[ICEM] Problem making structural mesh on a surface	froztbear	ANSYS Meshing & Geometry	1	November 10, 2011 08:52
[snappyHexMesh] snappyHexMesh won't work - zeros everywhere!	sc298	OpenFOAM Meshing & Mesh Conversion	2	March 27, 2011 21:11
basic of mesh refinement	arya	CFX	4	June 19, 2007 12:21

April 25, 2012, 18:22		#2
JBurnham Senior Member Jeff Burnham Join Date: Apr 2010 Posts: 204 Rep Power: 17	The general approach sounds right. Use fixed planes in your outer blocks to force the mesh to match gridlines with the extents of the nested blocks. Make sure the nested block sizes are 1/2 the larger block sizes, and that the extents are even multiples of the nested and outer block cell sizes. The idea is to get all the gridlines perfectly aligned: even a little bit of variation will cause some momentum diffusion, maybe negligible, maybe not. The closer to cubes your cells are, the higher the quality of the mesh, but up to 3:1 aspect ratios are often considered acceptable. Avoid placing mesh boundaries (like the nested block extents) in regions of high gradients: put them out before the wake structures begin to form.

April 25, 2012, 18:32		#3
bfox48 New Member brian Join Date: Apr 2012 Posts: 25 Blog Entries: 1 Rep Power: 14	Thanks... Also one more dumb but simple question. When looking at the simulation dashboard and the graphs of the different variables there are no units. Can someone tell me what these are in... and most important is percentage in decimal for or whole percents (0.5 vs 50%)?

April 25, 2012, 18:36		#4
JBurnham Senior Member Jeff Burnham Join Date: Apr 2010 Posts: 204 Rep Power: 17	No prescribed units in FLOW-3D. All input and output variables are in the same length/mass/time system. If you input density as the value that corresponds to units of kg/m^3, then your length units must be m, and your viscosity units must be kg/m/s. Pressure outputs will then be kg/m/s^2 = Pa, and turbulent mixing length will be reported in m. Same holds true for ft/slugs/s and cm/g/s systems. On the dashboard, when % is listed on the vertical axis, 0.5 means 0.5% (half a percent).