chaining the time resolution

[akariwrynn]

Hi,
I would like to measure the voltage and the current at very low time domains for impedance spec. However, after some time voltage seems to be constant for small time changes like:
V(t=0.00097889)
Out[33]: array(4.18440764)

V(t=0.000980453)
Out[34]: array(4.18440764)

V(t=0.000982017)
Out[35]: array(4.18440764)

V(t=0.000983581)
Out[36]: array(4.18440764)
V(t=0.000984363)
Out[39]: array(4.18440787)
V(t=0.000985927)
Out[37]: array(4.18522475)

V(t=0.000985145)
Out[38]: array(4.18522475)

How can I increase the time resolution or fix this problem?

Thanks.

[YannickNoelStephanKuhn]

Could you please post the way you invoke the solver? The simplest way to increase the time resolution depends on that.

[akariwrynn]

Sure,
i=0
while i< len(Array): # Simulation

params['Current function [A]'] = deneme(amplitude,freq[i])
sim=pybamm.Simulation(model,parameter_values=params)
sim.solve(Array[i]) #solving the simulation for the time domain that we specified
solution=sim.solution
V=solution["Terminal voltage [V]"]
C=solution["Current [A]"]
for e in Array[i]:
Voltage_gen.append(V(t=e))
Current_gen.append(C(t=e))
i+=1
freq is the frequencies that I want to look and deneme is the function that turned to function to sin function and finally array is the time required for each frequency.
Thanks.

[YannickNoelStephanKuhn]

This should work:

dt_max_in_seconds = np.min(np.abs(Array[1:] - Array[:-1]))
sim.solve(Array[i], dt_max=dt_max_in_seconds * model.timescale_eval)

[akariwrynn]

Hi again,
It says 'DFN' object has no attribute 'timescale_eval' so, what should I do?

Sorry for bothering too much

[Tyssel]

You may try to replace model.timescale_eval by

pybamm.ParameterValues().evaluate(model.timescale)

[YannickNoelStephanKuhn]

Seems like the documentation is outdated then: https://github.com/pybamm-team/PyBaMM/blob/develop/pybamm/solvers/casadi_solver.py#L47
Seems like one of the developers needs to help here.

[akariwrynn]

So we wait then

[rtimms]

dt_max is already in seconds, so assuming Array is in seconds you can just do

dt_max = np.min(np.abs(Array[1:] - Array[:-1]))
sim.solve(Array[i], dt_max=dt_max)

The default is dt_max = 0.01* model.timescale_eval and model.timescale_eval only exists once the simulation has been set up.

I can see why this is confusing though. We should improve the docs.