.. _meshing_tutorial:

Meshing CAD models
==================

.. include:: ../icons.inc

In this tutorial we will create meshes for SMITER from surface CAD
models extracted beforehand. Unless we continue from
:ref:`geom_tutorial` tutorial we need to ensure that study files are
available for reading with

.. code-block:: console

    make -C study/tutorial

We open :file:`study/mesh1.hdf` study that contains CAD geometry with
``FW4`` and ``PANELS`` from which we will create target and shadow
meshes.  Before we go to meshing we activate the Geometry module with
a click on |icon_geometry_module| button from the toolbar. We setup
visibility of both models by clicking on |icon_visiblemesh| icons and
:guilabel:`Fit All` to get ready for meshing.

Mesh for the target is usually required to be higher density that the
shadow as we want better wetted pattern on the target. It is always
good practice to find the correct mesh density by firstly trying rough
meshes that are still acceptable from past experience and then doubling
the density by recomputing the meshes.

Meshing FW4
~~~~~~~~~~~

Now we can move to Mesh module by clicking |icon_mesh_module| in the
main toolbar. While being in *Mesh* module we again select ``FW4``
geometry and set visibility by |icon_visiblemesh| and :guilabel:`Fit
All` and the ``FW4`` panel appears in the Mesh module.

.. image:: images/mesh_1.png
    :align: center

From the menu we select :menuselection:`Mesh --> Create mesh` or
|icon_mesh_init| button the meshing toolbar where we setup dialog with
the :guilabel:`Name` to ``target`` and :guilabel:`Mesh type` to
``Triangular``. For :guilabel:`Algorithm` we select
:guilabel:`Triangle: Mefisto` as shown in the following image.

.. image:: images/mesh_2.png
    :align: center

Before pressing :guilabel:`A&pply and Close` we click on
:guilabel:`Assign a set of hypotesses` and within
:guilabel:`Hypothesis Construction` dialog change :guilabel:`Name` to
``Max Size Target`` and :guilabel:`Length` to ``20``. This will create
:guilabel:`Hypothesis` for :guilabel:`Wire discretisation`
:guilabel:`Algorithm` in :guilabel:`1D` tab.

At :guilabel:`A&pply and Close` and :menuselection:`Mesh --> Compute`
from the menu or |icon_mesh_compute| button from the meshing toolbar
we get newly created mesh statistics stating 18545 triangles created
ifor the ``target`` mesh. After :guilabel:`Close` and inpecting
:guilabel:`Object Browser` and zooming mesh we obtain the follwoing
image.

.. image:: images/mesh_3.png
    :align: center

Creating shadow mesh
~~~~~~~~~~~~~~~~~~~~

For creating the ``panels`` mesh we again :menuselection:`Mesh -->
Create mesh` or |icon_mesh_init| with ``PANELS`` selected from the
:guilabel:`Geometry`. The only difference is that we :guilabel:`Assign
a set of hypotheses` for :guilabel:`2D: Automatic Triangulation` with
``Max Size Shadow`` :guilabel:`Name` and :guilabel:`Length` set to
``50`` that created arounf 165000 triangles shadow mesh with some
errors in shape evaluation that seems OK.

.. image:: images/mesh_4.png
    :align: center

We will now create rotated panels with menu selection
:menuselection:`Modification --> Transformation --> Rotation`, where
we :guilabel:`Select whole mesh...` named ``panels``. Rotation axis
and point of rotation needs to be entered as shown in the folowing
image.

.. image:: images/mesh_5.png
    :align: center

With :guilabel:`Preview` enabled we :guilabel:`Copy Elements` with
:guilabel:`Angle` ``-40`` degrees of rotation that gets created in the
study after :guilabel:`A&pply and Close`.

Translating target
~~~~~~~~~~~~~~~~~~

In order to avoid intersection of the *target* with the *shadow*
panels due do different meshing density and numerical errors we
translate the *target* panel by less that 0.01 mm inwards. From menu
:menuselection:`Modification --> Transformation --> Translation` we
open a dialog for translation. The relative translation vector can be
easily determined by firstly clicking on the center of the panel in
"two-point" mode and then dividing that point position by 1000 or so
to get small translation in the correct direction inwards.

.. image:: images/mesh_6.png
    :align: center

With this operation we have now prepared all meshes necessary to
compute the case and can save the study somewhere. Mesh ``panels``
will serve as a *shadow* and ``target_translated`` will be a *target*
for SMITER.

Computing the case
==================

In this tutorial we will create SMITER case with meshes prepared in
:ref:`meshing_tutorial` tutorial. Unless we continue from
:ref:`meshing_tutorial` tutorial we need to ensure that study files
are available for reading with

.. code-block:: console

    make -C smiter/salome/study/tutorial

We open :file:`study/mesh2.hdf` study that contains first wall CAD
geometry with corresponding *target* and *shadow* meshes.

.. image:: images/mesh_7.png
    :align: center

We change the following :guilabel:`beqparameters` in associated
*Eqdsk* meshes by right clicking on them and selecting :guilabel:`Edit
ctl` to:

 beq_bdryopt=9
    Required for limiter case. We set this for meshes shadow and target.
 beq_fldspec=3
    Mapping equilibrium field to cylindrical (axisymmetric). Set for meshes
    wall, shadow and target.
 beq_bdryopt=3
    Set this for wall mesh :file:`LIMITER-554.eqdsk`. It calculates the value
    of flux at plasma boundary while also including the wall silhouette.
 beq_nzetap=1
    Set for shadow and target.

For hierarhical data structure (HDS) we need to increase "bin size"
:guilabel:`Hdsgen` CTL :guilabel:`hdsgenparameters`

 limit_geoobj_in_bin=200
   Default value was 20 objects (triangles).


See Val-F11-fw246t-fw4t for similar test case parameters or press
:guilabel:`Help` button when setting parameters.

After right clicking on :guilabel:`Compute case` the following results
appear automatically.

.. image:: images/mesh_8.png
    :align: center

When comparing the wetted pattern with NF55 benchmark case we see
similarity in the shape. Maximal heat flux should be proportional to
power loss :math:`1.76 MW/ 7.5 MW \cdot 2.2 MW/m^2=0.52 MW/m^2`, while
here we observe :math:`Q_{max}=1.0 MW/m^2`

The study is available as :file:`study/tutorial/fwd1.hdf`.

The difference is due to small shadow. If we increase the shadow by
rotating left and dight we get :math:`Q_{max}=0.47` which is close to the
estimation and available as :file:`study/tutorial/fwd2.hdf` and shown in the
following figure.

.. image:: images/mesh_9.png
    :align: center