[znusth]

Hello everyone, I calculate microscale(about 20um) liquid dynamic mesh by cfx, but failed.
To find out the reason， I used to do the following attempts:
1、when I change the material to ideal gas(compressible), it calculated successful.
2、I set the liquid(oil) with compressible , just use the following formula ,but it also failed.
Pref = 100 [bar]
BulkMod = 2.2e9[Pa]
Density = 1002[kg/m^3]*(1 +(Absolute Pressure – Pref)/BulkMod)
3、 I simplify the model and change the geometric dimension from 20um to 10mm, it calculated successful. But it failed with the size of 20um.
4、 I use software fluent to calculate a simple model, use "pressure-base ",it can calculate incompressible liquid well, and finally it successful.
5、Increase timestep or decrease time step, but have no effect.
I DON'T KNOW WHY CALCULATE FAILED WHEN USE CFX.
more details in the pictures.The last two pictures is about 2way-fsi, but is related to this case, just the reason like it describe, but with no method in this case .1.jpg

2.jpg

3.jpg

4.jpg

5.jpg
THANK YOU VERY MUCH！
THANK YOU VERY MUCH！
THANK YOU VERY MUCH！
zhaunghui
2017.8.9

[ghorrocks]

With such massive changes in pressure likely, you are probably going to need VERY small time steps. Are you sure you tested time steps small enough?

[znusth]

Originally Posted by ghorrocks

With such massive changes in pressure likely, you are probably going to need VERY small time steps. Are you sure you tested time steps small enough?

many thanks!
when I use the ideal gas to calculate,I always set the timestep with 1e-6s,just 0.000001s, so I set it to 1e-6s when use oil . But the pressure is also large...So I consider if there's some control settings in CFX with this kind of cases...

[ghorrocks]

The acoustic velocity of water is around 1500m/s. The acoustic velocity of air is around 340m/s. This suggests as a first guess your water time step will need to be about 1/5 of the air time step. But this just a starting point, if that does not work go smaller again.

You will probably find it useful to start at a crazy small time step (1e-10s) and check that converges for the first few time steps. If that is OK then try 1e-9s, and keep going until you find the biggest time step size which converges.

[znusth]

Originally Posted by ghorrocks

The acoustic velocity of water is around 1500m/s. The acoustic velocity of air is around 340m/s. This suggests as a first guess your water time step will need to be about 1/5 of the air time step. But this just a starting point, if that does not work go smaller again.

You will probably find it useful to start at a crazy small time step (1e-10s) and check that converges for the first few time steps. If that is OK then try 1e-9s, and keep going until you find the biggest time step size which converges.

Thank you! Just now,when I change the timestep to 1e-10s,the effect is similar...The pressure is so large. Because for the liquid ,the coefficient of compressibility is so small...

[ghorrocks]

Is this in your micron sized duct? If so then you will need to go much smaller.

[znusth]

Originally Posted by ghorrocks

Is this in your micron sized duct? If so then you will need to go much smaller.

Yes， in the duct which height is 20um ，I ’ll try it later. Thank you!