|
|
interMixingFoam uses a combined limiter to guarantee phase consistency which is appropriate in this case because only two of the phases are immiscible. Can you provide a patch and an example where your suggested change is needed? For cases in which several phases are immiscible a full multiphase limiter is needed as used in multiphaseInterFoam. |
|
|
modified.png (33,440 bytes) |
|
|
original.png (39,156 bytes) |
|
|
|
|
|
In my testcase I am simulating the impact of a drop of one fluid (shown as white in the contour figures) with a thin layer of another fluid miscible with the drop's fluid (shown as red in the contour figures). The inmiscible fluid is air (shown as blue in the contour figures). When the fluxes are computed using the limiters of the second phase alpha2, there exists diffusion between one of the miscible fluids and air which shouldn't occur (as shown in figure original.png). When modifying the solver to compute the fluxes using: alphaPhi1 = alphaPhi1BD + lambda1*alphaPhi1 alphaPhi2 = alphaPhi2BD + lambda2*alphaPhi2 Where lambda1 and lambda2 are the limiters of each phase, the diffusion between the inmiscible and the miscible phase disappears (as seen in figure modified.png). I don't know if this change will reduce the accuracy or stability of the solver in more complex testcases but it seems that in this testcase the original computation of the fluxes produces unphysical results. I'll attach a proposed modification of the alphaEqns.H file of the interMixingFoam solver which calculates the fluxes using the limiter of each phase. |
|
|
How many test cases have you validated this change on? Do you see any phase consistency and conservation issues from having inconsistent limiters between the phases? |
|
|
Our team just started investigating these kind of flows and we don't have quite yet defined a set of testcases that would allow us to analyze phase consistency and mass conservation issues. I just tryed this new implementation in the tutorial interMixingFoam/damBreak (refining its mesh) and it seems that the new way of calculating the fluxes allows to preserve better the interface between the miscible and inmiscible phases. I don't see major mass conservation or phase consistency issues in the new implementation although we haven't performed simulations in complex testcases. I will let you know in this bug report if we are able of providing a set of testcases that could help you decide which is the correct implementation. Moreover, if you have any suggestions for these testcases we would be willing to perform some validation of the results. |
|
|
No feedback from reporter |
- Anonymous
