[vinerm]

Wall shear stress should not be zero. There is something wrong either with the setup or with postprocessing. Calculate area averaged value of wall shear stress using Reports > Surface Integral.

[Damaja]

It says zero here.


Edit: Ok I think that uis because of symmetry of the obstacle wall. changed that, get a value now woth compute, its around 50. Will calculate a transient simulation.

[vinerm]

By the very definition, symmetry has all flux equal to 0, i.e., stress is 0 as well. Symmetry is implemented as gradient of each field with respect to the normal of the symmetry boundary being equal to 0.

[Damaja]

Thanks for that. So if my walls, which form the tube in which the body is, are in symmetry mode, that is not correct, right? Or is it okay, if I only wanna see that the fluid does when it flows around the body?


Here are the wall shear stress diagrams. So i think, these are alright? Still searching for the reason why powerlaw creates no vortex and the carreau model does crete a vortex.

[vinerm]

The walls of the domain around the obstacle are alright if you use symmetry. However, the walls of the obstacle should not be symmetric.

[Damaja]

That is the comparison of the Re numbers of powerlaw and carreau. They differ just slightly. Could it cause the vortex?

edit: Left upper one is carrea, right upper one powerlaw. So the fluid flows slightly faster around the body for the carreau model, or the viscosity is lower.

[vinerm]

Actually, for both of these, system is in transition. That implies, there should be unstable vortex generation at the back for both of these. Are you running these cases as transient? Both have to be run in transient to predict the vortex generation.

[Damaja]

Yes, it is transient in both cases.

[Damaja]

I wanna add, because of the non-newtonian fluid, I rather want the carreau model to look like the powerlaw model. So that there is no vortex street behind the body.

[vinerm]

Fluent will predict the flow behavior as per the conditions. However, if you look at the averaged field, these might look similar. If the shear stress and Re are same for both viscosity models, then there should not be vortex in one case while absent in other. This could also be numerical artifact. Ensure that the numerical settings are exactly same.

[Damaja]

Alright, I will check this in the evening and let you know. Thanks for your support!

[Damaja]

Hello again, I found a flaw!

edit: I used 2nd order upwind in methods/transient formualtion for carreau simulation and 1st order upwind in transient formulation for powerlaw.


Turns out, both simulations don't create a vortex for 1st order, and they do create a vortex both for 2nd order. I wonder why this makes a huge difference for the simulation. Is this typical?


So do you know, what is correct now? 2nd order upwind, or 1st order upwind?

[vinerm]

Second order is always more accurate. First order discretization has numerical diffusion that causes the momentum to diffuse, hence, the vortex do not appear. First order is good to start with because it is stable but less accurate. Final solution should always be determined using at least second order.

[Damaja]

Understood! That solved my problem. Thank you, for your continuous support!

[vinerm]

Nice to know that it worked.