Unable to run custom experiments with pouch cell model

[SimMLPhys]

I’ve been following the example code for the pouch cell model from the PyBaMM documentation: Pouch Cell Model. I am trying to simulate CCCV cycling using string-based instructions with pybamm.Experiment. However, I encounter a NotImplementedError when running the simulation. Is there a better way to simulate CCCV cycling if the string-based experimental instructions don't work for this model yet?

For instance, using the following experiment setup:

experiment = pybamm.Experiment( [ "Discharge at C/10 for 10 hours or until 3.0 V", "Rest for 1 hour", "Charge at C/10 for 5 hours", ] )

I receive the error:

---------------------------------------------------------------------------
NotImplementedError                       Traceback (most recent call last)
Cell In[6], line 21
     10 experiment = pybamm.Experiment(
     11     [
     12         "Discharge at C/10 for 10 hours or until 3.0 V", 
   (...)
     15     ]
     16 )
     18 sim = pybamm.Simulation(dfn, parameter_values=param, var_pts=var_pts,experiment = experiment)
---> 21 solution_dfn=sim.solve(
     22             solver=pybamm.CasadiSolver(mode="safe")
     23         )

File ~\AppData\Roaming\Python\Python311\site-packages\pybamm\simulation.py:702, in Simulation.solve(self, t_eval, solver, save_at_cycles, calc_esoh, starting_solution, initial_soc, callbacks, showprogress, inputs, t_interp, **kwargs)
    697 t_eval, t_interp_processed = step.setup_timestepping(
    698     solver, dt, t_interp
    699 )
    701 try:
--> 702     step_solution = solver.step(
    703         current_solution,
    704         model,
    705         dt,
    706         t_eval,
    707         t_interp=t_interp_processed,
    708         save=False,
    709         inputs=inputs,
    710         **kwargs,
    711     )
    712 except pybamm.SolverError as error:
    713     if (
    714         "non-positive at initial conditions" in error.message
    715         and "[experiment]" in error.message
    716     ):

File ~\AppData\Roaming\Python\Python311\site-packages\pybamm\solvers\base_solver.py:1222, in BaseSolver.step(self, old_solution, model, dt, t_eval, npts, inputs, save, t_interp)
   1220     model.y0 = old_solution.all_ys[-1][:, -1]
   1221 else:
-> 1222     _, concatenated_initial_conditions = model.set_initial_conditions_from(
   1223         old_solution, return_type="ics"
   1224     )
   1225     model.y0 = concatenated_initial_conditions.evaluate(0, inputs=model_inputs)
   1227 set_up_time = timer.time()

File ~\AppData\Roaming\Python\Python311\site-packages\pybamm\models\base_model.py:809, in BaseModel.set_initial_conditions_from(self, solution, inplace, return_type)
    807 if isinstance(var, pybamm.Variable):
    808     try:
--> 809         final_state = solution[var.name]
    810     except KeyError as e:
    811         raise pybamm.ModelError(
    812             "To update a model from a solution, each variable in "
    813             "model.initial_conditions must appear in the solution with "
    814             "the same key as the variable name. In the solution provided, "
    815             f"'{e.args[0]}' was not found."
    816         ) from e

File ~\AppData\Roaming\Python\Python311\site-packages\pybamm\solvers\solution.py:547, in Solution.__getitem__(self, key)
    544     return self._variables[key]
    545 else:
    546     # otherwise create it, save it and then return it
--> 547     self.update(key)
    548     return self._variables[key]

File ~\AppData\Roaming\Python\Python311\site-packages\pybamm\solvers\solution.py:507, in Solution.update(self, variables)
    505         model._variables_casadi[key] = var_casadi
    506     vars_casadi.append(var_casadi)
--> 507 var = pybamm.ProcessedVariable(
    508     vars_pybamm, vars_casadi, self, cumtrapz_ic=cumtrapz_ic
    509 )
    511 # Save variable and data
    512 self._variables[key] = var

File ~\AppData\Roaming\Python\Python311\site-packages\pybamm\solvers\processed_variable.py:110, in ProcessedVariable.__init__(self, base_variables, base_variables_casadi, solution, warn, cumtrapz_ic)
    107     return self.initialise_2D()
    109 # Raise error for 3D variable
--> 110 raise NotImplementedError(
    111     f"Shape not recognized for {base_variables[0]}"
    112     + "(note processing of 3D variables is not yet implemented)"
    113 )

NotImplementedError: Shape not recognized for 24983.2619938437 * y[2:4098](note processing of 3D variables is not yet implemented)

Entire code for reference:

import pybamm
import json
import matplotlib.pyplot as plt
import numpy as np
import scipy.interpolate as interp
import time as time
dfn = pybamm.lithium_ion.DFN(
    {"current collector": "potential pair", "dimensionality": 1, "thermal": "x-lumped"},
    name="1+1D DFN",
)
param = dfn.default_parameter_values

param.update(
    {
        "Negative particle diffusivity [m2.s-1]": 3.9 * 10 ** (-14),
        "Positive particle diffusivity [m2.s-1]": 10 ** (-13),
        "Negative current collector surface heat transfer coefficient [W.m-2.K-1]": 10,
        "Positive current collector surface heat transfer coefficient [W.m-2.K-1]": 10,
        "Negative tab heat transfer coefficient [W.m-2.K-1]": 10,
        "Positive tab heat transfer coefficient [W.m-2.K-1]": 10,
        "Edge heat transfer coefficient [W.m-2.K-1]": 10,
        "Total heat transfer coefficient [W.m-2.K-1]": 10,
    },
    check_already_exists=False,
)

npts = 16
var_pts = {
    "x_n": npts,
    "x_s": npts,
    "x_p": npts,
    "r_n": npts,
    "r_p": npts,
    "z": npts,
}
experiment = pybamm.Experiment(
    [
        "Discharge at C/10 for 10 hours or until 3.0 V", 
        "Rest for 1 hour", 
        "Charge at C/10 for 5 hours",
    ]
)

sim = pybamm.Simulation(dfn, parameter_values=param, var_pts=var_pts,experiment = experiment)


solution_dfn=sim.solve(
            solver=pybamm.CasadiSolver(mode="safe")
        )

[Santiagopeacely]

I think it's because the solver Casadi stores intermediate variables, and currently pybamm doesn't support handling 3D variables, so errors occur. The IDAKLUSolver solver can solve it, but it cannot access 3D variables。
2 + 1D Pouch Cell Model|An error occurred while viewing the negative potential variable #4606

[SimMLPhys]

I think it's because the solver Casadi stores intermediate variables, and currently pybamm doesn't support handling 3D variables, so errors occur. The IDAKLUSolver solver can solve it, but it cannot access 3D variables。 2 + 1D Pouch Cell Model|An error occurred while viewing the negative potential variable #4606

Are there any workarounds? Either to simulate multiply charge-discharge cycles or to somehow access the 3D variables? Perhaps I can define the current function as a function of time to change signs after every...3 hours when cycling at C/3 or something like that.

[SimMLPhys]

Looks like this is a known issue #1549 #3552 and Valentin mentions some workarounds in #1549