Learning Forum

[cibi]

Hello everyone, 

I am trying to compute the platelet lysis index (PLI) in my simulations by solving this transport equation: 

where tau is the shear stress. 

My simulations are steady state, so if I have understood correctly from Ansys Fluent User Guide, this is the type of equation I am solving in Fluent with the diffusion coefficient equal to zero:

I have prepared the UDF for my source term:

To obtain convergence in my simulations I had to change the under relaxation factor of my UDS to 0.50.

After runnign my simulation I use CFD post to compute the power of my scalar (PLI = scalar 0 ^(1/0.77)).

I am expecting a PLI distribution similar to this:

But the resuls that I get are like this:

 

I also tried on a simplier geometry and I think that the results are stil wrong in terms of values but the colormap is more similar to what I am supposed to obtain:

I was wondering if there is something wrong in my code for the source term or if this might be related to something else?

Thank you in advance!

Claudia

[Rob]

What scalar diffusion did you set? How turbulent is the flow?

[cibi]

Thanks for replying.

I'm not exactly sure what you mean by scalar diffusion, however I disabled Inlet Diffusion and then set the diffusion coefficient as zero because I don't have a diffusion term in my equation.

My simulations are laminar

[Rob]

I'd set the Diffusion coefficient as a very (e-14 or so) small number rather than zero. Zero's can cause problems if you divide! 

How well refined is the mesh? What other settings did the paper(?) use or report? 

[cibi]

I have run a simulation with diffusion coefficient set to e-14 and indeed results look better. However, the values of my uds are still very small. 

This is with the local scale:

This is with user defined scale:

 

Regarding my mesh, I have got 3'060'421 elements and this is how it looks like:

In terms of the settings of the paper, simulations are unsteady state and laminar flow, no-flux (for PLI ) conditions were applied at the walls and PLI= 0  at the inlet.

I was wondering if there is something wrong in my code or something I am missing when adapting my equation to the one that Fluent solves?

Thank you very much!

[Rob]

Possibly, but that's not something I can really comment on. 

The mesh isn't great. It looks refined enough on the cross channels but it's very coarse down the longer sections. Have a look at decomposition (break up the geometry) and then sweep/multizone.