[Andrey.M.]

Hello
1st case is a pipeline with 2 phase flow - gas and liquid. Volume of Fluid approach, Geo-Reconstruct. The problem is since certain time moment solution significantly slows down. Iterations goes with the same speed, but time spending on "Updating solution at time level" is immense.
Mesh is pretty coarse and my only guess is that solver needs much time for Geo-Reconstruct algorythm to determine interface shape. Solver meets difficulties due to coarse mesh which leads to smeared volume fraction in some locations. Small features of the flow such as droplets and small bubbles are not of interest - that's the reason why mesh is made not fine enough to represent interface of small features when they occur.
Is my assumption right?
What are another causes?
How would you recommend to deal with that in order to preserve solution speed as it were?

2nd case is a 3-phase flow in a separator, Euler approach, transient. It's not a surprise that in Euler we face a convergence difficulties. Is there some rule of thumb to improve it? Such as cell size and particle size ratio? What approaches usually used?
There is also something interesting in timestep convergence dependance. Usually to improve solution stability one decreases timestep, but in that case we faced that smaller timestep causes poor convergence and large timestep causes better. How could it be explained?

Thank you in advance!