# What is the limitation/caution of using EWF with DPM model?

Member Posts: 23

Hello,

I am using EWF and DPM to simulate air/water annular flow in a tube. I am using a small region in front of an annular tube to inject liquid water. Here I notice that the inlet velocity and thickness of liquid film are changing abruptly, even film velocity is going negative. I changed the droplet diameter, mesh number in the axial and radial direction, injection wall length, but nothing worked. Would you tell me what from the below may be responsible for it?

1) Thick liquid film formed on the wall. I knew from some documents that the EWF model depends on film thickness. For some cases, we can use EWF, and for some cases not. Should/ shouldn't I use EWF for shown thickness and velocity?

2) High air velocity/ Reynolds number

3) Geometry/ Mesh Configuration

4) Droplet diameter

I am using UDF for calculating the Entrainment rate. Even without entrainment /droplet injection/ UDF, film thickness and velocity are showing abnormal trends.

Here are film thickness and velocity at two flow times-

Sincerely,

Tagged:

• UKPosts: 11,770Forum Coordinator
• Posts: 117Member

Hello @Rob ,

Previously I was getting negative film velocity at the inlet of the injection section. I have increased the injection tube length from 0.2 m to 0.3m, and not getting negative velocity at the inlet of the injection section. But, my primary concern is the annular section. I am getting both positive and negative film velocity at the inlet of this section. However, I am expecting a positive, smooth velocity and thickness at the annular section. Here I notice that film thickness reaches a maximum value of 2.75mm whereas tube diameter is only 9.4mm, and film velocity is becoming negative from a small distance of the annular inlet section. I have confusion with this peak value because 5.5 mm (2.75+2.75 )of 9.4mm tube is covered by film and air with a velocity of 18.9m/s flows through the rest 3.9 (9.4-5.5=3.9)mm section.

Here I am using UDF for the entrainment rate. I even tried this simulation without UDF but film velocity and thickness patterns are almost the same(velocity is positive and negative both).

The injection section is just to inject liquid to get a certain liquid/film flow rate at the inlet of the annular section. I simulated the injection section only and got that film velocity and film thickness are 3.47703 m/s and 0.000353 m respectively for a film flow rate of 0.03622 kg/s at the end of the injection section/inlet of the annular section.

At the starting of the simulation film thickness and velocity pattern are smooth (flow time 0.33s), but change randomly with the passes of time. See another image at 0.98s.

Moreover, I am using PISO solver and I find that residual is very sensitive with momentum relaxation factor. If I decrease the momentum relaxation factor to 0.0001, the residual of the continuity equation goes to the order of 10e-6. Can I claim it as the converged solution?

Regards,

• GermanyPosts: 7,893Forum Coordinator

Key monitors are not changing much: what about convergence every time step?

• UKPosts: 11,770Forum Coordinator

Remember the film is "thin" so it could be 1m thick in your model and still not affect the gas velocity.

• Posts: 117Member

Hello @DrAmine ,

The residual of continuity equation ranges between 2.28e-4 and 6.92e-6. Can I tell it converged solution? Remember I am using PISO solver with a momentum relaxation factor of 0.0001. Can you suggest anything so that I will get a positive film velocity and a stable solution? Currently, I am getting positive film velocity as well as negative film velocity at the inlet of the annular section.

Regards,

• Posts: 117Member

Hello @Rob ,

In a previous comment, you wrote "In the first layer or few of cells adjacent to the inlet the flow needs to adjust from the assigned value to the calculated value." Will you explain how can I implement this?

I even tried the simulation with EWF (applied liquid mass flux at the injection wall) and airflow at the inlet of the injection wall. But, it also fails(negative film velocity).

• UKPosts: 11,770Forum Coordinator

Don't put the film immediately next to the boundary, and don't take data from cells immediately next to a change in boundary conditions. In a wind tunnel experiment would you measure the free stream velocity in the fan volute or downstream once the flow has fully developed?

• GermanyPosts: 7,893Forum Coordinator

PISO SIMPLE or whatever: deep converging results should be independent on the solver. Be aware that EWF is for thin films if the film is thick: do not rely on the results :)

• Posts: 117Member

Hello @DrAmine ,

Would you tell me when should I consider film as thin or thick? If you tell me the thickness limit for both cases, it will be easy to understand me.