Very high or negative pressures
 

Dear All,

I am modeling a pipe with dynamic mesh. One end is a piston, the other end is an outlet/outflow.

Case setup: k-epsilon incompressible: air density = 2.33 kg/m^3 = constant.
Operating pressure = 265kPa at piston; also tried 0 Pa.
I tried the open pipe end as outlet as well as outflow.

The first step of the motion ends always with wrong static pressures. Next to piston they are very high (>1E+6). If the open end is an outflow, then the pressure there becomes negative.

Please help me correct the errors in this case.

*switch to pressure-outlet
*check your units (scale)
*what kind of method are you using for moving mesh (layering/remeshing/...)
display the grid while plotting the pressure, and check if the high pressures aren't linked with bad cells

Hello mAx,

thank you for your time.

* switched to pressure outlet, still calculated static/absolute pressure at piston is > 1E+7
* my pressure unit is Pa
* I am using layering, it is a hex mesh on a rectangular pipe
* equiangle/equivolume skew is < 0.2

*check the scale: grid/scale
*how fast is your piston moving?
*do you see any discontinuity in the fluid domain (display pressure)?
*does the fluid flows out?

* grid/scale: all scales = 1, unit = [meter]
* piston moves with 32 m/s
* no discontinuity: pressure is 265kPa at outlet and grows smoothly to 5E+7 at piston
* mass flux at outlet converged to -8.5kg/s
Pipe/duct has rectangular cross section, 0.3m x 0.1m, length ~20m

post a picture with pressure and grid.
did you check the mesh motion?

Yes, the mesh motion is fine.

If the time step is dt=1E-4, mass flux at outlet is -8.5kg/s, pressure at piston is 5E+7Pa

If the time step is dt=1E-6, mass flux at outlet is -600kg/s, pressure at piston is 3E+11Pa

This is totally bogus. Something seems to be conceptually wrong.

disable the MDM, switch to steady solver.
Change your moving piston to massflow inlet and give the 8.5kg/s.
Iterates and check the pressure distribution.
are you working 2d or 3d?

Apologies for the delayed reply.
Solution:

start the simulation with ideal-gas on, run for sufficient number of time steps. Switch to incompressible.

It works now.

Many thanks, mAx for the useful analysis; it has helped me tackle the problem from many aspects.

Cheers,
-jpo