I am new to cfd. I am modelling blood flow through a pipe with a valve at the centre. Assuming the flow is laminar is the boundary layer resolution and the number of prism layers important. I could find specifications(y+ ranges) for turbulent flow but none for laminar flow.would greatly appreciate any help

in my opinion, y+ is only for turbulence flow, not for laminar flow.

Hi,

Laminar flows are strongly influenced by the boundary layer, but the boundary layer is much thicker than in turbulent flow so a coarser mesh can sometimes resolve it. Depends on exactly what Reynolds number you are at. Inflation layers are still useful until you are doing very low Reynolds numbers, ie Stokes flow.

Blood has strong non-newtonian properties so you may well need a fine boundary mesh to resolve these non-linear properties. I have never modelled blood so am no expert on it but you should check it out.

Glenn Horrocks

If you assume for a moment that you have a constant viscosity (I know you haven't but go with it), then since the flow is laminar you basically have a Hagen-Poiseuille flow, i.e. a parabolic velocity profile. Therefore (IMHO) you should use a power law grid distribution so that you have equal increments of velocity across each cell. For example, if u_wall=0 and u_max=1 m/s, and you use 20 cells across the channel width, then each cell should be at the position that correlates to velocity profile increments of 0.1 m/s. This will give you the refinement you need at the walls. If you have problems understanding this rather wordy explanation, draw a parabola on some graph paper(y is position, x is velocity) and mark off the y positions that correlate with equal increments in x. Good luck.

Thank you Frank, Glenn & Jim for ur suggestions

[Trudix]

Hi guys, even though this post is pretty old allready, I have a similar question. I'm new to CFD and I'm trying to model a microchannel device. According to Glen Horrocks: "Inflation layers are still useful until you are doing very low Reynolds numbers, ie Stokes flow." To which magnitude of Re Inflation-layers make sense? Since I'm dealing with really low Re, should I just use my regular tedrahedron-mesh near the walls. Is a mesh refinement near the walls requiered? Thanks a lot for your help. It's very much appreciated.

[ghorrocks]

You should consider why inflation layers are recommended in the first place. They are useful in high Re flows as the flow variables change rapidly in the boundary layer region and a fine mesh is required to capture this rapid change so the flow is accurate. So the mesh resolution you require depends on how rapidly the variables change.

For microfluidic devices you rarely get boundary layers forming and you are more likely to have laminar flow, possibly fully developed. In fully developed flow the variables change smoothly across the entire section - which means there is no rapid change near the wall, so no inflation layers are required. But a laminar flow which is not fully developed may require additional resolution at the wall to capture the start up behaviour.

As always, it is problem dependant. So do a mesh sensitivity study and find out in your case.

[Trudix]

Thanks a lot for your help ghorrocks!