Merge branch 'main' into joss

demos/demo_references.bib

@@ -0,0 +1,11 @@
+@article{gupta2006calculation,
+  title={Calculation for the cathode surface concentrations in the electrochemical reduction of {CO}₂ in {KHCO}₃ solutions},
+  author={Gupta, N and Gattrell, M and MacDougall, B},
+  journal={Journal of applied electrochemistry},
+  volume={36},
+  number={2},
+  pages={161--172},
+  year={2006},
+  publisher={Springer},
+  doi={10.1007/s10800-005-9058-y}
+}

demos/flow_reactor/flow_reactor.py.rst

@@ -0,0 +1,311 @@
+Flow reactor for CO2 electrolysis
+=================================
+
+We consider a flowing bicarbonate elecrolyte with CO2 reduction on a flat plate
+electrode.
+
+.. image:: flow_reactor.png
+   :width: 400px
+   :align: center
+
+
+We consider simplified CO2 - KHCO3 homogeneous reactions:
+
+.. math::
+
+   \ce{CO2} + \ce{OH-} \xrightleftharpoons[k_1^b]{k_1^f} \ce{HCO3-}
+
+   \ce{HCO3-} + \ce{OH-} \xrightleftharpoons[k_2^b]{k_2^f} \ce{CO3^2+} + \ce{H2O}
+
+where :math:`k_i^f` and :math:`k_i^b` are the forward and backward rate constants, respectively.
+Here, we assume the dilute solution case where :math:`\ce{H2O}`
+concentration is not tracked. 
+
+We are solving for :math:`c_k`, the molar concentration of species
+:math:`k\in\{\ce{CO2}, \ce{HCO3-}, \ce{CO3^2+}, \ce{OH-},
+\ce{K+}\}`,
+as well as :math:`\Phi_2`, the ionic potential.
+For each species :math:`k`, the conservation of mass is given by
+
+.. math::
+
+   \nabla \cdot \mathbf{N}_k = R_k,
+
+where :math:`\mathbf N_k` and :math:`R_k` are the flux and homogeneous
+reactions for species :math:`k`, respectively The flux term is given by the
+Nernst-Planck equation, which includes diffusion, advection, and
+electromigration:
+
+.. math::
+
+   \mathbf N_k = -D_k \nabla c_k + c_k \mathbf u - z_k F \frac{D_k}{RT} c_k \nabla \Phi_2,
+
+where :math:`D_k` is the diffusion coefficient of species :math:`k`, and
+:math:`z_k`, its charge number, :math:`\mathbf{u}` is the velocity, :math:`F`
+is the Faraday constant, :math:`R` is the ideal gas constant, :math:`T` is the
+temperature. We assume shear flow such that the velocity field is given by
+:math:`\mathbf{u} = (\dot \gamma y, 0)`, where :math:`\dot\gamma` is the shear rate and
+:math:`y` is the distance from the electrode.
+
+We assume the electroneutrality approximation such that
+
+.. math::
+
+        \sum_k z_k c_k = 0.
+
+The are five charge-transfer reactions at the surface of the electrode
+
+.. math::
+
+   \ce{CO2} + \ce{H2O} + 2\ce{e-} \rightleftharpoons \ce{HCOO-} + \ce{OH-}
+
+   \ce{CO2} + \ce{H2O} + 2\ce{e-} \rightleftharpoons \ce{CO} + 2\ce{OH-}
+
+   \ce{CO2} + 6\ce{H2O} + 8\ce{e-} \rightleftharpoons \ce{CH4} + 8\ce{OH-}
+
+   2\ce{CO2} + 8\ce{H2O} + 12\ce{e-} \rightleftharpoons \ce{C2H4} + 12\ce{OH-}
+
+   2\ce{H2O} + 2\ce{e-} \rightleftharpoons \ce{H2} + 2\ce{OH-}
+
+Note that we ignore the transport of products other than :math:`\ce{OH-}`. For each
+reaction :math:`j`, we assume a constant current density :math:`i_j`, such that
+the normal flux at the electrode surface is given by
+
+.. math::
+
+    \mathbf{N}_k \cdot \mathbf n = -\sum_j \frac{s_{k,j} i_j}{n_j F},
+
+where :math:`s_{k,j}` is the stoichiometry coefficient for the species
+:math:`k` in reaction :math:`j`, and :math:`n_j` is the number of electrons
+transferred. The partial current density is given by :math:`i_j= \mathrm{cef}_j
+i_\mathrm{tot}`, where :math:`\mathrm{cef}_j` is the current efficiency of
+reaction :math:`j` and :math:`i_\mathrm{tot}` is the total current. 
+
+At the inlet, the boundary condition for each species :math:`k` is
+
+.. math::
+
+    \mathbf N_k \cdot \mathbf n = c_{k,\mathrm{bulk}} \mathbf u \cdot \mathbf n,
+
+where :math:`c_{k,\mathrm{bulk}}` is the bulk concentration, and at the outlet
+
+.. math::
+
+    \mathbf N_k \cdot \mathbf n = c_k \mathbf u \cdot \mathbf n.
+
+At the top boundary, i.e. the bulk, the ionic potential is set to a reference
+value :math:`\Phi_2=0`, and a no-flux condition is set for all species. A no
+flux-condition is also set on the walls with no electrode, as well as on the
+electrode for the species that are not reacting there.
+
+The species parameters, homogeneous bulk reactions, and the constant-rate
+charge-transfer kinetics are taken from :cite:`gupta2006calculation`.
+For this demo, we will use SI units. EchemFEM has no notion of units and the
+user simply must provide consistent units.
+
+We import the required packages::
+
+    from firedrake import *
+    from echemfem import EchemSolver, RectangleBoundaryLayerMesh
+
+Then, we create our own child class of :py:class:`EchemSolver <echemfem.EchemSolver>`::
+
+    class GuptaSolver(EchemSolver):
+        def __init__(self):
+        
+We now define a rectangular mesh with added refinement in a thin boundary layer
+along the electrode boundary using :py:func:`RectangleBoundaryLayerMesh <echemfem.utility_meshes.RectangleBoundaryLayerMesh>`. The
+channel is of width 1 mm and length 5 mm. We have 100 elements in the length
+direction. In the width direction, there are 50 elements in the first
+micrometer and 50 elements for the rest. The local refinement is meant to
+capture the thin :math:`\ce{OH-}` boundary layer. The electrode is on the
+bottom boundary, which has boundary marker 3::
+
+            Ly = 1e-3 # m
+            Lx = 5e-3 # m
+            mesh = RectangleBoundaryLayerMesh(100, 50, Lx, Ly, 50, 1e-6, boundary=(3,))
+
+For numerical reasons, we do not want the electrode boundary to be right next
+to the inlet and outlet regions. We will have 1 mm of inactive wall before and
+after the electrode. One easy way to do this for Neumann boundaries is to
+define an indicator function for the electrode using spatial coordinates and
+conditional given by Firedrake/UFL::
+
+            x, y = SpatialCoordinate(mesh)
+            active = conditional(And(x >= 1e-3, x < Lx-1e-3), 1., 0.)
+
+We create a list with parameters for all species. Each entry in the list is a
+dictionary containing the name, the diffusion coefficient, the bulk
+concentration, and charge number of a species::
+
+            conc_params = []
+
+            conc_params.append({"name": "CO2",
+                                "diffusion coefficient": 19.1e-10,  # m^2/s
+                                "bulk": 34.2,  # mol/m3
+                                "z": 0,
+                                })
+
+            conc_params.append({"name": "HCO3",
+                                "diffusion coefficient": 9.23e-10,  # m^2/s
+                                "bulk": 499.,  # mol/m3
+                                "z": -1,
+                                })
+
+            conc_params.append({"name": "CO3",
+                                "diffusion coefficient": 11.9e-10,  # m^2/s
+                                "bulk": 7.6e-1,  # mol/m3
+                                "z": -2,
+                                })
+
+            conc_params.append({"name": "OH",
+                                "diffusion coefficient": 52.7e-10,  # m^2/s
+                                "bulk": 3.3e-4,  # mol/m3
+                                "z": -1,
+                                })
+
+            conc_params.append({"name": "K",
+                                "diffusion coefficient": 19.6E-10,  # m^2/s
+                                "bulk": 499. + 7.6e-1 + 3.3e-4,  # mol/m3
+                                "z": 1,
+                                })
+
+Similarly, we provide a list containing the parameters for the homogeneuous
+reactions. Each entry is a dictionary for one reaction containing: a dictionary
+with the stoichiometry of all reactants (negative) and all products (positive),
+and rate constants::
+
+            homog_params = []
+
+            homog_params.append({"stoichiometry": {"CO2": -1,
+                                                   "OH": -1,
+                                                   "HCO3": 1,
+                                                   },
+                                 "forward rate constant": 5.93, # m3/mol/s
+                                 "backward rate constant": 1.34e-4 # 1/s
+                                 })
+
+            homog_params.append({"stoichiometry": {"HCO3": -1,
+                                                   "OH": -1,
+                                                   "CO3": 1,
+                                                   },
+                                 "forward rate constant": 1e5, # m3/mol/s
+                                 "backward rate constant": 2.15e4 # 1/s
+                                 })
+
+For convenience, we write a function that will return the current density
+function for each charge-transfer reaction. To have zero-flux on the inactive
+walls, we multiply by our indicator function. Since the currents are constant,
+the argument is unused::
+
+            def current(cef):
+                j = 50. # A/m2
+                def curr(u):
+                    return cef * j * active
+                return curr 
+
+We create a list for the parameters of the charge-transfer reactions where each
+entry is a dictionary associated with a reaction. For the key ``"reaction"`` we
+provide a function that returns the partial current density of the reaction.
+The stoichiometry is provided similarly to ``homog_params``. We also provide
+the number of electrons transferred and name the boundary where the reaction
+happens::
+
+            echem_params = []
+
+            echem_params.append({"reaction": current(0.1), # HCOO
+                                 "stoichiometry": {"CO2": -1,
+                                                   "OH": 1
+                                                   },
+                                 "electrons": 2,
+                                 "boundary": "electrode",
+                                 })
+
+            echem_params.append({"reaction": current(0.05), # CO
+                                 "stoichiometry": {"CO2": -1,
+                                                   "OH": 2
+                                                   },
+                                 "electrons": 2,
+                                 "boundary": "electrode",
+                                 })
+
+            echem_params.append({"reaction": current(0.25), # CH4
+                                 "stoichiometry": {"CO2": -1,
+                                                   "OH": 8
+                                                   },
+                                 "electrons": 8,
+                                 "boundary": "electrode",
+                                 })
+
+            echem_params.append({"reaction": current(0.2), # C2H4
+                                 "stoichiometry": {"CO2": -2,
+                                                   "OH": 12
+                                                   },
+                                 "electrons": 12,
+                                 "boundary": "electrode",
+                                 })
+
+            echem_params.append({"reaction": current(0.4), # H2
+                                 "stoichiometry": {"OH": 2
+                                                   },
+                                 "electrons": 2,
+                                 "boundary": "electrode",
+                                 })
+
+Most physical parameters that are not associated with species or reaction are
+passed through the ``physical_params`` argument in a dictionary. The ``"flow"`` key
+is given a list of the desired transport mechanisms. The other parameters are
+physical constants described above::
+
+            physical_params = {"flow": ["diffusion", 
+                                        "electroneutrality",
+                                        "migration",
+                                        "advection"],
+                               "F": 96485.3329,  # C/mol
+                               "R": 8.3144598,  # J/K/mol
+                               "T": 273.15 + 25.,  # K
+                               }
+
+The parameters and the mesh are passed to the initiator of parent class
+:py:class:`EchemSolver <echemfem.EchemSolver>`. The optional argument ``family``
+sets the finite element space; here, ``"CG"`` stands for continuous Galerkin,
+which is usually the fastest option, whereas ``"DG"`` stands for discontinuous Galerkin, which can provide more stability for advection-dominated cases::
+
+            super().__init__(conc_params, physical_params, mesh, family="CG",
+                             echem_params=echem_params,
+                             homog_params=homog_params)
+
+The boundary conditions are set through this abstract method, which the user always needs to specify. Note that the name ``"electrode"`` only has meaning because it is provided in ``echem_params``. Also note that the "natural" boundary condition for this finite element formulation is no-flux, so it is the default boundary condition::
+
+        def set_boundary_markers(self):
+            self.boundary_markers = {"inlet": (1,),
+                                     "outlet": (2,),
+                                     "bulk": (4,),
+                                     "electrode": (3,),
+                                     }
+
+Similarly, the velocity field is set through an abstract method::
+
+        def set_velocity(self):
+            _, y = SpatialCoordinate(self.mesh)
+            self.vel = as_vector([1.91*y, Constant(0)])  # m/s
+
+Finally, we create our solver object, set up the solver, and run it::
+
+    solver = GuptaSolver()
+    solver.setup_solver()
+    solver.solve()
+
+The solution fields can be visualized by opening ``results/collection.pvd``
+using Paraview. For example, the :math:`\ce{CO2}` solution field:
+
+.. image:: CO2_solution_field.png
+   :width: 600px
+   :align: center
+
+This demo can be found as a script :download:`here <flow_reactor.py>`
+
+.. rubric:: References
+
+.. bibliography:: demo_references.bib
+   :filter: docname in docnames

docs/Makefile

@@ -14,6 +14,17 @@ help:
 
 .PHONY: help Makefile
 
+.PHONY: copy_demos
+
+source/demos: copy_demos
+
+copy_demos:
+	install -d source/demos
+	cp ../demos/*/*.rst ../demos/*/*.png ../demos/demo_references.bib source/demos
+	for file in source/demos/*.py.rst; do pylit $$file; done
+
+html: Makefile copy_demos
+	@$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
 # Catch-all target: route all unknown targets to Sphinx using the new
 # "make mode" option.  $(O) is meant as a shortcut for $(SPHINXOPTS).
 %: Makefile

docs/source/_static/references.bib

@@ -0,0 +1,37 @@
+@article{Taylor1973,
+  title={A numerical solution of the {N}avier-{S}tokes equations using the finite element technique},
+  author={Taylor, Cedric and Hood, Paul},
+  journal={Computers \& Fluids},
+  volume={1},
+  number={1},
+  pages={73--100},
+  year={1973},
+  publisher={Elsevier}
+}
+
+@misc{Ladyzhenskaya1963,
+  title={The mathematical theory of incompressible viscous flows},
+  author={Ladyzhenskaya, O},
+  year={1963},
+  publisher={Gordon and Breach, New York}
+}
+
+@article{Brezzi1974,
+  title={On the existence, uniqueness and approximation of saddle-point problems arising from {L}agrangian multipliers},
+  author={Brezzi, Franco},
+  journal={Publications des s{\'e}minaires de math{\'e}matiques et informatique de Rennes},
+  number={S4},
+  pages={1--26},
+  year={1974}
+}
+
+@article{Babuvska1971,
+  title={Error-bounds for finite element method},
+  author={Babu{\v{s}}ka, Ivo},
+  journal={Numerische Mathematik},
+  volume={16},
+  number={4},
+  pages={322--333},
+  year={1971},
+  publisher={Springer}
+}

docs/source/conf.py

@@ -19,6 +19,7 @@
     'sphinx.ext.intersphinx',
     'sphinx.ext.napoleon',
     'sphinx.ext.mathjax',
+    'sphinxcontrib.bibtex',
 ]
 
 mathjax3_config = {
@@ -36,6 +37,9 @@
 
 templates_path = ['_templates']
 
+#  -- Options for sphinxcontrib.bibtex ------------------------------------
+bibtex_bibfiles = ['demos/demo_references.bib', '_static/references.bib']
+
 # -- Options for HTML output
 
 html_theme = 'sphinx_rtd_theme'