.. _benchmark-spe11c:

SPE11C
======

SPE11C is the three-dimensional field-scale benchmark case. The configurations
compare Cartesian and corner-point grids, convective dissolution, and a model
with more than 100 million cells.

Configurations
--------------

``r1_Cart_50m-50m-10m``
   Cartesian grid with nominal cell dimensions of 50 m, 50 m, and 10 m. The
   lateral boundaries use 1 m cells.

``r2_cp_50m-50m-8mish``
   Corner-point grid with nominal lateral dimensions of 50 m and an approximate
   vertical cell size of 8 m. The lateral boundaries use 1 m cells.

``r3_cp_50m-50m-8mish_convective``
   The ``r2`` grid with the convective-dissolution model enabled for facies 2
   and 5.

``r4_cp_8m-8mish-8mish``
   Fine corner-point grid with an x cell size of 8 m and approximate y and z
   cell sizes of 8 m. The lateral boundaries use 1 m cells.

Run the cases
-------------

Run these commands from the ``benchmark/spe11c`` directory:

.. code-block:: console

   pyopmspe11 -i r1_Cart_50m-50m-10m.toml -o r1_Cart_50m-50m-10m -m all -g all -r 168,100,120 -t 0,5,10,15,20,25,30,35,40,45,50,75,100,150,200,250,300,350,400,450,500,600,700,800,900,1000 -w 0.1
   pyopmspe11 -i r2_cp_50m-50m-8mish.toml -o r2_cp_50m-50m-8mish -m all -g all -r 168,100,120 -t 0,5,10,15,20,25,30,35,40,45,50,75,100,150,200,250,300,350,400,450,500,600,700,800,900,1000 -w 0.1
   pyopmspe11 -i r3_cp_50m-50m-8mish_convective.toml -o r3_cp_50m-50m-8mish_convective -m all -g all -r 168,100,120 -t 0,5,10,15,20,25,30,35,40,45,50,75,100,150,200,250,300,350,400,450,500,600,700,800,900,1000 -w 0.1
   pyopmspe11 -i r4_cp_8m-8mish-8mish.toml -o r4_cp_8m-8mish-8mish -m all -g all -r 168,100,120 -t 0,5,10,15,20,25,30,35,40,45,50,75,100,150,200,250,300,350,400,450,500,600,700,800,900,1000 -w 0.1

Large-scale SPE11C case
-----------------------

The ``r4`` case contains more than 100 million cells. Running it required
improvements to OPM Flow, including `METIS support
<https://github.com/OPM/opm-grid/pull/738>`_, and scalable pyopmspe11
preprocessing and postprocessing. The published run used 1024 processes on the
`NORCE <https://www.norceresearch.no/>`_ HPC cluster.

.. image:: ../figs/pyopmspe11c100Mcells.gif
   :alt: CO2 molar fraction in the 100-million-cell SPE11C model
   :align: center

The animation shows liquid-phase CO2 molar fraction in
`ParaView <https://www.paraview.org/>`_.

Results
-------

Performance data
++++++++++++++++

.. figure:: ../figs/benchmark_spe11c_performance.png
   :alt: SPE11C performance benchmark results
   :align: center

Sparse data
+++++++++++

.. figure:: ../figs/benchmark_spe11c_sparse_data.png
   :alt: SPE11C sparse benchmark results
   :align: center

Spatial maps
++++++++++++

.. figure:: ../figs/massfractc.png
   :alt: SPE11C liquid-phase CO2 mass-fraction maps
   :align: center

Well 1 location correction
--------------------------

In the submitted SPE11C results, Well 1 was located approximately 100 m above
the benchmark depth of 300 m.

The corrected comparison below uses an additional case without dispersion on
the reporting grid. This case, referred to as ``r5``, was generated by setting
``dispersion`` to zero in the ``r1`` configuration. It is shown instead of the
large ``r4`` model.

.. figure:: ../figs/benchmark_spe11c_sparse_data_corrected.png
   :alt: SPE11C sparse data with the corrected Well 1 location
   :align: center

Continue
--------

* See :doc:`../configuration_file` for configuration-variable definitions.
* Follow the :doc:`../tutorial` for a guided introductory workflow.
* Use :doc:`plopm-visualization` to reproduce the simulation-grid maps.
* See :doc:`../convergence` for the SPE11B grid-refinement study.

.. button-ref:: benchmark
   :ref-type: ref
   :color: primary
   :outline:

   Back to the benchmark gallery