open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello all,

Excuse my lack of knowledge about Finite Element numerical schemes, but are there 2nd order accurate FEM schemes available in TELEMAC-2D (space and time)? Some of the FVM schemes explicitly state their order of accuracy (e.g. HLLC 1st-order, WAF 2nd-order), but the orders of accuracy for T2D FEM schemes hasn't been so obvious. For example, when using the FEM schemes for velocity advection (e.g. scheme 1 vs. 5) would MOC and PSI mass-conservative both be considered 1st order in time and space?

Thank you in advance for your help.

-Chris

Hello,

Well all is 1st order in space, except the PSI distributive scheme, and all is 1st order in time unless you have the implicitation coefficients equal to 0.5, which is the verge of numerical stability. In 3D being really second order in time with a moving mesh is very difficult. We are working on advection distributive schemes 2nd order in time and space (hopefully a first example in the next version). However our tests show that even Finite Volumes schemes which are officially 2nd order loose this property on non structured grids, and, given their CFL limitation, are much more diffusive than 1st order schemes (you can do the comparisons in Telemac-2D). So the order in time and space is something to be considered with caution.

Regards,

Jean-Michel Hervouet