Discussion Forum

Hello!

I have simulated a three phase flow problem, consisting on a dense droplet that is aspirated into a tube filled with a light fluid. To do so, I have used the example in the COMSOL library "air bubble penetrates the phase boundary between water and oil":
-2D-axisymmetric model
-Creeping flow (neglecting inertia terms for all the phases; inlet: pressure ramp; outlet: pressure level)
-Three phase flow (water inside the tube: phase A; droplet: phase B; water outside the tube: phase C; inlet: phase C; and outlet).

Since the dimensions of my problem were a bit different from that of the COMSOL example, I had to modify the mobility factor M0 for the method to converge. However, I find that for certains values of the pressure ramp slope, the simulation doesn't converge (i.e. if the slope of the ramp is quite large (>5000 Pa/s), the droplet is aspirated; if the slope is small (5 Pa/s) , the droplet remains unaltered). To deal with this problem, so far I have tried unsuccesfully:
-Mofying M0 -> I use 2e-15 m3/s; I tried from 2e-13 to 2e-17 m3/s but the phases are not stable after some time.
-Modyfing the mesh
-Increasing the computation time for small slopes: if I put a final time grater that 1.2s the program doesn't find a consistent initial value. Why does this occur?

I would like to run the simulation with a slope of 5 Pa/s but I cannot find the way to do so. Does anyone know which is the critical parameter to increase the computation time? Is it M0 or the interfacial thickness perhaps? I really think it must be a small thing, since for slopes of >5000 Pa/s it works fine.

Thank you in advanced for any possible help!!
Blanca