Simulation with active scalar field

[ayush-9s]

Hi,

I am trying to simulate variable viscosity incompressible flows using xcompact3d. I have modified the transeq.f90 file to include the viscosity variation into the momentum equation rhs. The viscosity variation is governed by a scalar field(concentration) which is being solved for as well. I am getting fringe like patterns in the solution scalar distribution, while the velocity and pressure distributions do not have any such patterns. I wanted to ask if anyone knows why these kinds of fringes might appear in the solution.

I have attached the concentration(fig 1) and x-velocity(fig 2) distributions for reference. I was giving a steep concentration distribution (zero in the jet, and one everywhere else) for this case. Can this be a problem?

Following is the plot of all the variables at a particular axial location in the centerplane.

This is a circular jet flow case I am testing, and it was working for incompressible constant viscosity simulations without any problems.

Any help will be highly appreciated.

Thanks,
Ayush.

[mathrack]

Hello, if the concentration field is discontinuous, the finite difference schemes can produce oscillations. I would suggest one of the following solutions :

• Use a smooth field
• Change the derivative operator to properly compute the derivative at discontinuities

EDIT : try to input a discontinuous signal and look at the first and second derivative

[slaizet]

Hi,
To complement Cedric' comment, it looks like you are using a zero velocity co-flow, which might not be a good idea. The shape of the oscillations (aligned with the mesh) suggests that you might want to increase your spatial resolution.
Best,
Sylvain

[ayush-9s]

Hello,

Thanks a lot for the responses. I have tried using a more smooth concentration distribution, and it is reducing the fringes appearing in the solution. I will try with a more resolved grid and small co-flow as well.

Thanks again,
Ayush.