Two Cylinders ALE (2D)

Two Cylinders ALE (2D)#

Description of the setup#

Figure 11 shows the geometry of the test case you will run in this tutorial.

It consists of the annulus region between two coaxial cylinders, with the inner cylinder having an imposed motion.

The inner cylinder has harmonic motion : \(U_x = A \sin(w \times t)\) where A is the amplitude of displacement and w the angular frequency of displacement.

Mesh of two concentric circles — Figure 11 Geometry of the test case#

Tutorial setup#

First, go to an empty directory and copy the base TrioCFD test case from which we will start: TwoCylindersALE_jdd1

triocfd -copy TwoCylindersALE_jdd1 && cd TwoCylindersALE_jdd1

Open the datafile TwoCylindersALE_jdd1.data in a text editor of your choice.

Modifying the test case#

Start by making the following changes in the file TwoCylindersALE_jdd1.data. The goal is to have a simulation fast enough so that you will be able to visualize what happens. The results will probably not be very valid physically.

In the Scheme_euler_implicit block:
- Set nb_pas_dt_max to a large number.
- Set facsec to 30 and facsec_max to 100.
- Set tmax to 2.
In the Post_processing of the problem:
- set format to lata.
- set dt_post to 0.1.
Change the amplitude and frequency of the displacement to 0.1 and 2 respectively:
- The formula is 0.1*2*cos(2*t).
- You have to make this change in two places:
  - The speed of the mesh boundary: Imposer_vit_bords_ALE.
  - The CircleA boundary_conditions of Navier_Stokes_standard_ALE.

Running and visualizing the simulation#

Now, you can run the calculation:

triocfd TwoCylindersALE_jdd1

It should run in a few minutes at most. If it takes too long, you may increase the facsec.

Once it is done, open with visit (you can do it before the end and reopen from visit to obtain new postprocessed timesteps).

visit -o TwoCylindersALE_jdd1.lata

Display the Mesh and the vector field VITESSE_SOM_dom, then start the time slider.

Over the total time (=2), you should see the inside circle go to the right then come back to the left.

The velocity field is probably not correct because we set a facsec way too high in our implicit scheme, in order to shorten the simulation for the sake of this exercise.