Poiseuille Flow Setups are not stable when switched to PST

[svchb]

Using PST results in an unstable setup compared to TVF:

Switching to ConsistentShiftingSun2019 from TVF is all that is required to reproduce this.
The boundary layers are unstable because there seems to be some kind of disturbance in the fluid particles near the wall boundary, which evolves into the shown wavy pattern which should not be present.

[efaulhaber]

Please provide more details. What is wrong here, what's the expected result, and how exactly can we reproduce this?

@LasNikas can you take a look, please? Is this a known problem? Is this already fixed in one of your open PR? Maybe #1043?

[efaulhaber]

Note that the Poisseuille flow 3D example added in #1075 should probably be changed to use PST once this issue is closed.

[LasNikas]

I think there's something about free-surface handling that causes problems with PST but not with TVF.
I'll take a closer look at that, though.

TransportVelocityFormulation(; background_pressure)

ParticleShiftingTechnique()

ConsistentShiftingSun2019

[LasNikas]

@efaulhaber I'm confused now. Are 2. and 3. not identical?

1. (works):

trixi_include(@__MODULE__, joinpath(examples_dir(), "fluid", "poiseuille_flow_2d.jl"), 
shifting_technique=ParticleShiftingTechniqueSun2017(integrate_shifting_velocity=true, momentum_equation_term=MomentumEquationTermSun2019()))

2. (works):

trixi_include(@__MODULE__, joinpath(examples_dir(), "fluid", "poiseuille_flow_2d.jl"), 
shifting_technique=ParticleShiftingTechniqueSun2017(integrate_shifting_velocity=true, momentum_equation_term=nothing))

3. (that doesn't work):

trixi_include(@__MODULE__, joinpath(examples_dir(), "fluid", "poiseuille_flow_2d.jl"), 
shifting_technique=ConsistentShiftingSun2019(; modify_continuity_equation=false, second_continuity_equation_term=nothing, momentum_equation_term=nothing))

1.

2.

3.