[max.kozak]

Hi all,

I am running a Non-Newtonian simulation using a Herschel-Bulkley model. I am wanting to run over a shear range from 0 to 148s^-1. The simulation will only run it from 0.0001 to 148s^-1, but there are a few areas where the shear is actually about 1e^-6 s^-1. However, whenever I lower the range to this value, I do not get the solution to converge. Does anyone have any experience with this, or any ideas regarding how to accurately run the simulation and capture the results where shear rate is this low? Thanks for any advice regarding this.

[ghorrocks]

This is a common problem with non-Newtonian models as they often have divide by zero and similar issues at zero shear strain rate. So I think you will find you cannot run it down to zero SSR as it is not defined. If you look at the CFX implementation it has a maximum and minimum SSR so you can avoid these problems by artificially changing the SSR in these regions to a value where it is defined. Obviously this affects the result, the question is whether it is significant or not.

So I would recommend you do a sensitivity study on the minimum SSR setting and see how low you need to go.

[max.kozak]

Thanks ghorrocks. I'm reading about using a CEL, and I'm wondering if you think this would be a good option? Perhaps it would be reasonable to say if the shear strain rate is less than the minimum CFX is defined for, then the dynamic viscosity value is some constant value? In reality, I'm not sure if this would affect the results too significantly, but I was thinking it could work around the obviously very flawed results I currently have.

Thanks

[ghorrocks]

The Herschel-Bulkley model in CFX has max and min thresholds built in. There is no need to use CEL to do this. Just use the built in functions. But you will need to do a sensitivity analysis to check that the values you are using stop the excessive values which lead to numerical failure, but are wide enough to allow accurate modelling of your case.

[max.kozak]

Hi Glenn,

It seems the values I am using with the built-in range are not stopping the excessive error. I'm wondering, if instead of directly applying the Herschel-Bulkley model I create my own expression, that I could work around this issue?

Is there any way to create an expression where I say,
Dynamic Viscosity=some function at a certain range, but when shear strain rate falls out of the range, say below 0.001s^-1 (where I was having problems), then the Dynamic Viscosity=some value? I was thinking an if or step function could work around this??

[ghorrocks]

If the built in limits are not stopping the error then tighten them further. It is far preferable to use the built in stuff rather than generate your own function, as using discontinuous functions for any material property almost always causes grief.