Boundary Constraints in Tutorial not being enforced?

[codebro634]

Hey,

I am currently working through the tutorial pages and have a problem understanding the boundary conditions. I am currently stuck at the tutorial dealing with the Navier-Stokes equations for Cylinder Flow (https://fenicsproject.org/pub/tutorial/html/._ftut1009.html). Why is it that the inflow nodes defined by

inflow = ‘near(x[0], 0)’

with the inflow profile

inflow_profile = (‘4.01.5x[1]*(0.41 - x[1]) / pow(0.41, 2)’, ‘0’)

at first do not have the values of the inflow profile? What I noticed is that it that the inflow-node values start at 0 (of course, this is the default initialization value) and then quickly converge to the inflow-profile after a few time steps. Why does the employed DirichletBoundary conditions not hard-set these values such that they are enforced within the first time step?

[dokken]

Please note that Dolfin is deprecated, and that you should consider moving to dolfinx.

The new DOLFINx solver is now recommended over DOLFIN

In April 2022 we released versions of the new FEniCS Project components; DOLFINx (v0.4.1) , FFCx (v0.4.2), Basix (v0.4.1) and UFL (2022.1.0.post0). These projects are hosted under our GitHub account at FEniCS Project · GitHub We now recommend that all new projects begin in the new DOLFINx problem solving environment, rather than the old DOLFIN problem solving environment. We consider DOLFINx to be superior to the old DOLFIN, particularly in terms of performance, scalability, API consistency and…

For DOLFINx Ive made a similar tutorial.
https://jsdokken.com/dolfinx-tutorial/chapter2/ns_code2.html

@codebro634

inflow_profile = (‘4.01.5x[1]*(0.41 - x[1]) / pow(0.41, 2)’, ‘0’)

at first do not have the values of the inflow profile? What I noticed is that it that the inflow-node values start at 0 (of course, this is the default initialization value) and then quickly converge to the inflow-profile after a few time steps. Why does the employed DirichletBoundary conditions not hard-set these values such that they are enforced within the first time step

There is very little value in setting these, as all those dofs at the boundary are strongly enforced in the first tentative solve. You would only get a slight difference in the region between the boundary and the first element inside the boundary…

[codebro634]

Thank you for the quick response! To check my understanding: The boundary condition is only enforced in the intermediate solution but not at the final solution? Therefore, it is not a bug but rather a consequence of the technique that was employed?

[dokken]

Yes, that is correct. For more details on splitting schemes, I recommend Splitting schemes — Oasisx which explains consequences of the schemes

[codebro634]

Awesome, thank you so much that really helped!