[Bug]: Half-cell, particle distribution DFN fails during charge experiments

[Sergio-PY]

PyBaMM Version

23.5

Python Version

3.9.18

Describe the bug

Hello,

When i create a half-cell DFN model and I try to include the particle distribution for the positive electrode, it wouldn't take any charge procedure. It only discharges, but the charge always fails.
When i do exactly the same but without including the particle distribution everything works fine.

Steps to Reproduce

import pybamm

Create the model using the particle size distribution option as well as the half cell configuration.

options = {"particle size": "distribution","working electrode": "positive" }
MPDFN_model = pybamm.lithium_ion.DFN(options=options)

Import half cell NMC parameters

param_nmc = pybamm.ParameterValues("Xu2019")

Create size distribution for the positive particles

param_nmc_MP=pybamm.get_size_distribution_parameters(param_nmc, electrode="positive")

Load a simple experiment

experiment = pybamm.Experiment(["Discharge at C/25 until 3.5 V", "Charge at C/25 until 4.2 V"])

Load everything (regardless of the solver used the result and error is the same)

MP_sim_pOCV = pybamm.Simulation(MPDFN_model, parameter_values=param_nmc_MP, solver=pybamm.CasadiSolver(atol=1e-3, rtol=1e-3, mode="safe"), experiment=experiment)

Simulate. The error always arise in the next step.

MP_sol_pOCV = MP_sim_pOCV.solve()

The error always gives a problem when charging, when i just discharge is ok.

Relevant log output

---------------------------------------------------------------------------
NotImplementedError                       Traceback (most recent call last)
Cell In[7], line 3
      1 # MP-DFN
      2 MP_sim_pOCV = pybamm.Simulation(MPDFN_model, parameter_values=param_nmc_MP, solver=pybamm.CasadiSolver(atol=1e-3, rtol=1e-3, mode="safe"), experiment=experiment)
----> 3 MP_sol_pOCV = MP_sim_pOCV.solve()

File ~\.conda\envs\PyBaMM_env\lib\site-packages\pybamm\simulation.py:725, in Simulation.solve(self, t_eval, solver, check_model, save_at_cycles, calc_esoh, starting_solution, initial_soc, callbacks, showprogress, **kwargs)
    723 npts = max(int(round(dt / op_conds.period)) + 1, 2)
    724 try:
--> 725     step_solution = solver.step(
    726         current_solution,
    727         model,
    728         dt,
    729         npts=npts,
    730         save=False,
    731         **kwargs,
    732     )
    733 except pybamm.SolverError as error:
    734     if (
    735         "non-positive at initial conditions" in error.message
    736         and "[experiment]" in error.message
    737     ):

File ~\.conda\envs\PyBaMM_env\lib\site-packages\pybamm\solvers\base_solver.py:1165, in BaseSolver.step(self, old_solution, model, dt, npts, inputs, save)
   1163         model.y0 = old_solution.all_ys[-1][:, -1]
   1164     else:
-> 1165         _, concatenated_initial_conditions = model.set_initial_conditions_from(
   1166             old_solution, return_type="ics"
   1167         )
   1168         model.y0 = concatenated_initial_conditions.evaluate(
   1169             0, inputs=model_inputs
   1170         )
   1172 set_up_time = timer.time()

File ~\.conda\envs\PyBaMM_env\lib\site-packages\pybamm\models\base_model.py:576, in BaseModel.set_initial_conditions_from(self, solution, inplace, return_type)
    574 if isinstance(var, pybamm.Variable):
    575     try:
--> 576         final_state = solution[var.name]
    577     except KeyError as e:
    578         raise pybamm.ModelError(
    579             "To update a model from a solution, each variable in "
    580             "model.initial_conditions must appear in the solution with "
    581             "the same key as the variable name. In the solution provided, "
    582             f"'{e.args[0]}' was not found."
    583         )

File ~\.conda\envs\PyBaMM_env\lib\site-packages\pybamm\solvers\solution.py:510, in Solution.__getitem__(self, key)
    507     return self._variables[key]
    508 else:
    509     # otherwise create it, save it and then return it
--> 510     self.update(key)
    511     return self._variables[key]

File ~\.conda\envs\PyBaMM_env\lib\site-packages\pybamm\solvers\solution.py:470, in Solution.update(self, variables)
    468         model._variables_casadi[key] = var_casadi
    469     vars_casadi.append(var_casadi)
--> 470 var = pybamm.ProcessedVariable(
    471     vars_pybamm, vars_casadi, self, cumtrapz_ic=cumtrapz_ic
    472 )
    474 # Save variable and data
    475 self._variables[key] = var

File ~\.conda\envs\PyBaMM_env\lib\site-packages\pybamm\solvers\processed_variable.py:105, in ProcessedVariable.__init__(self, base_variables, base_variables_casadi, solution, warn, cumtrapz_ic)
    102     self.initialise_2D()
    103 else:
    104     # Raise error for 3D variable
--> 105     raise NotImplementedError(
    106         "Shape not recognized for {} ".format(base_variables[0])
    107         + "(note processing of 3D variables is not yet implemented)"
    108     )

NotImplementedError: Shape not recognized for 48230.0 * y[0:12000] (note processing of 3D variables is not yet implemented)

[Sergio-PY]

Ok, i have seen that the error is general for all the particle distribution implementations. No DFN + particle distribution works when it has to do a charge process. I imagine that is a well-known issue and there are probably people working on that.

[rtimms]

Thanks for raising. The issue is with post-processing of 3D variables. Duplicate of #1549