Learning Forum

[abhnv_01]

With my two phase VOF simulation for a closed domain, I observed that I am loosing mass. 

Mesh is refined and time step is of the order of 0.0001 sec.

What can be the reason for loss of mass in the system.

[Rob]

Convergence and mesh resolution would be the first things to check. Are both phases fixed density?

[abhnv_01]

Liquid phase - Function of temperature

Vapor phase - Ideal gas 

Convergence criteria:

Continuity - 1e-05

Velocity - 1e-04

Energy - 1e-07

 

[Rob]

And the mesh resolution?

[abhnv_01]

[abhnv_01]

 

My Mesh quality 

1. Avg Element quality – 0.9977

2. Avg Aspect ratio – 1.0419

3. Avg Skewness – 0.001083

 

[Rob]

If you're wanting to model VOF phase change, ie you need to track the bubbles use a lot more cells. You're looking at putting 5-10 cells across each bubble, or the film if it's surface boiling. 

[abhnv_01]

But, using a lot more cells will also increase my computational time. 

Except the cell resolution, can there be any other reason for loss of mass inside the system?

[Rob]

Using more cells will cost compute and time, yes, but without running a finer mesh you won't know if the results are mesh dependent. 

Accuracy is likely reason for mass loss. Whether that's convergence, cell resolution or solver related is a different question, and you'd need to run more models to change solver mesh & parameters to figure that out. 

[abhnv_01]

Should we say that with accurate and correct model one can get zero mass loss/gain?

Or zero mass loss/gain for a closed domain is an ideal case?

[Rob]

In an ideal case all mass is conserved. In most cases some mass is lost eventually, but if the domain is well resolved spacially and with time the loss should be negligable. 

[abhnv_01]

With time step of 0.0001 sec(e-04) and after running 600,000 timesteps (=60sec), mass lost is approximately 0.3%, can this be considered negligible or the value is significant?

[Rob]

I'd consider it negligable, but you're the one deciding that based on your work. You also need to see if it's 0.3% mass liquid or vapour, and how that impacts the domain volume. 

[abhnv_01]

Thanks Rob for your help. 

Highly appreciate it.

[walaakm10]

Hello, I wonder, how can check the mass loss on my model.

[Rob]

Volume integrals, and then report the mass of a phase/species to a monitor plot. 

[walaakm10]

Many many thanks for your reply, I did as you told me, and I found this result. 

So, now to calculate the mass loss, should I subtract it from the assumed inlet mass? or what?

[Rob]

I don't know what you've set up so can't comment. 

[walaakm10]

ok, I got it. But could you please just give me a description of what this number represents? this one I got from "Volume integrals<< mass of a phase"water"? 

Is it equivalent to the mass of water in the system/model after the simulation?

[Rob]

It's the mass of water in the domain at the instant you ran the report. 

[walaakm10]

many thanks

[walaakm10]

would someone help me in evaluating the mass loss? I mean, how much the mass loss could be accepted? I got 30%, is this acceptable?

[Rob]

30% seems very high, so convergence may not be good. You need to review the post processing to see what's going on, and whether mass is lost due real flow patterns or poor convergence.