I'm working on simulation of bubble by using LSM. To solve level set function, centered difference scheme or ENO can be used to discretize the convection term (according to Chang, et al 124, 449-464, JCP, 1996). But in my case, the results are pretty different, could any one give suggestion to determine which way is more realistic?

My thought is that if I refine the grids, the scheme leading to small different results is more realistic, does this more sense?

Thanks for your attention!

Wu

most of cases, ENO is better than centeral scheme when dealing with convection terms.

Thanks, when I use central scheme, the bubble grows, then shrink, then grows (oscillation duo to evaporation and condensation). After refine, these oscillation disappear.

If using ENO, no oscillation comes out. The problem is that in reality, oscillation may appear.

Bubble oscillation should come from surface tension, not from the convection discretization scheme.

Sorry for not making myself clear. In my case, the bubble is put at the heating surface. The oscillation mentioned is referred to the shrinking and expanding of the bubble base, the change of the bubble volume is related to heat transfer. Using ENO scheme didn't show the base oscillation. What I want to know is if the different schemes of convection term may give different topology.

Unfornately, it does change the toplogy with different convection scheme. Especially, some critical situation such as pinch-off and dramatic topological transitions, but in lot of cases, those could be generated by overall numerical code, so you'd better test yours with established test problems first.