Hi, does anyone have the experience handling the "negative speed of sound" in simulation?

Right now I am working on a compressible flow modeling. The convergence trouble arose because of some "bad" points of negative speed of sound near a pressure outlet, which is subsonic. And the temperature limiter is trying very hard to underflow the tolerance during the running, but unsuccessful. I checked the flow field and found that the "bad" points was a region of extremely high density. At first, I thought that may be the grid quality problem. But "bad" points were still there after I improved the grid.

The solver is WIND-US. Any idea?

Thank you.

Could you please check the temperature field. As you know, the speed of sound depends on the temperature

For the flow field, it's supposed to be with very small temperature gradient in the outlet region. I don't have any heat source or significant heat generation in the domain. And I don't need to specify the temperature at outlet as the flow properties are extrapolated from upstream except the pressure.

From my experience, "negative speed of sound" and "negative temperature" are two common warning messages in high-speed flow computation, regardless of solver. Has anyone ever handled this?

Hi,

Are you getting negetive speed of sound or negitive temperature. Because I dont see any change of getting -ve speed of sound as we calculate speed of sound by SQRT(gamma * R * T) and we allways take +ve value of it.

On other hand if you are getting -ve temperatures, that is because of having -ve pressure or -ve density which is an indication that your solution has blown up and the way to tackle this problem is to coursen your grid or use limiters in your schemes(for high speed flows) or use schemes which are more robust.

Thanks, Vinayender

Hi, Negative speed of sound yeah you can look it in this way From coding point of view when we write the basic equations in conservative form and find the characteristics we will end up with 3 characteristics namely u,u+c,u-c at the both the inlet and exit boundaries which essentially represents propagation of solutions in the directions.

Two points raised by your reply

1- Energy dissipation = (Kinematic viscosity)*(symmetric part of the rate of deformation tensor)*(symmetric part of the rate of deformation tensor)

Unless your fluid has Zero kinematic viscosity or the flow is almost static, there is always viscous energy dissipation.

How much, I would not judge it untill you give details about your simulation.

2- The speed of sound is calculated by the equation given by Vinayender in the case of considering your fluid to be represented by the ideal equation of state. If this is not the case, the equation used to calculate the speed of sound is:

speed of sound = square root of ( [partial] pressure /[partial] density {evaluated at constant entropy} )

Hope you get the picture, describing mathematical and tensor relations in words is not always easy.

Look, when you get unphysical results out of a computer simulation, it is always a good idea to have the basic definitions handy. Good Luck.

One last check before you reply, could you upload a picture of the internal energy field as given by your simulation