Error at edge between normal-tangential and cartesian BCs

[lionel]

Hi Elmer,

The first picture shows my model in which the lateral BCs are defined in the cartesian coordinate system and the bottom BC is defined in the Normal-Tangential system. The direction of the flow is from left to right and I represented the velocities below 20m/a.

The problem is when the angle between these both surfaces (bottom and lateral) is different than 90°, the velocities at the edge are set to zero.

issue_BCs.png

(163.26 KiB) Not downloaded yet

Here the bed surface is the sum between a linear downsloping bed and a gaussian function. When the lateral slope is not modified because out of the influence of the gaussian curve, there is not problem with the velocities, whereas there is when this lateral slope is changed.
I tested also with various bed, and each time I have not a 90° angle between both BCs, I observe the same behavior.

I attached my .sif file also

main.sif

(12.56 KiB) Downloaded 262 times

Apparently there was a lot of changes in the Normal-Tangential stuff in the past, but I use one of the latest version of elmer, in Vuori.

Cheers
Lionel

[lionel]

Hi Elmer,

I solved my problem by replacing the condition imposed for the normal velocity (Velocity 2 = Real 0.0) by a huge Slip coefficient (Slip Coefficient 2 = Real 1.0e20) in the lateral BC (the one defined in cartesian coordinate system).

Regards
Lionel