Current simulation #1639
Current simulation
#1639
-
|
Hello all, |
Beta Was this translation helpful? Give feedback.
Answered by
Saransh-cpp
Sep 7, 2021
Replies: 8 comments 21 replies
-
|
I think the first issue is related to You can check the version by running - |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
Thanks for the reply, I installed v0.4.0 and it worked. Does anyone have any idea about the second question? |
Beta Was this translation helpful? Give feedback.
Answer selected by
brosaplanella
-
|
If you run the drive cycle using |
Beta Was this translation helpful? Give feedback.
4 replies
-
|
@rtimms Thanks for your reply, Regarding CCCV cycling, do you mean that I should break my solution into 100 pieces if I want to run 50 cycles as I input my current? error |
Beta Was this translation helpful? Give feedback.
-
|
Not sure about the error, looks like a divide by zero somewhere. Re CCCV and drive cycles, you can now do drive cycles in experiments but it doesn't look like you can run a drive cycle in an experiment until a given voltage. What I was suggesting was to do CCCV, then use the solution as the initial conditions for a drive cycle simulation until a voltage cut off, then repeat N times. |
Beta Was this translation helpful? Give feedback.
-
|
Not sure I'm following your answer, maybe I wasn't clear enough. I need to simulate 50 cycles of my experiment and the voltage response includes CCCV charge and discharge. I extracted the input current but I can't simulate the first cycle and repeat it for 50 because the current dramatically changes with cycling. Now, does it mean I should break my solution into 100 pieces and initialize each CCCV based on the previous solution? Can the model follow the current and go to the CV model whenever reached the cutoff voltage? |
Beta Was this translation helpful? Give feedback.
-
|
Hi, sorry, my answer wasn't clear. If you are just doing CCCV charging with a constant current discharge in between you can prescribe an experiment like in this example. You can also include drive cycles in experiments like in this example, but there is an issue with event handling (e.g. stopping at a given voltage) in drive cycles when using the casadi solver which we are aware of (see #1902). If I understand correctly, you want to prescribe the current from data until you hit a voltage limit, then switch to doing CCCV charging? I think your options are 1) just use the current data from the experiment (assuming you have the current data for the CCCV part?) or 2) split the data so you have 50 "drive cycles" (sounds like you have a different driving current each cycle?) and then piece the solution together with steps of CCCV charging in between. Let me know if I have misunderstood what you're trying to do or if my answer isn't clear. A working example might be useful. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the clarification, as you mentioned, when I run the simulation with a defined cycling current and reach cutoff voltage, I run into an error which is probably the same event handling problem with Casadi solver. |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
By option 1 I meant can you just pass the current directly from the experiment (i.e. the text file you just shared) and see what the voltage response is? You could pass the whole thing like a drive cycle like in this example. If you are just doing steps of constant current you can do something like import pybamm
experiment = pybamm.Experiment(
[
(
"Discharge at C/5 for 10 hours or until 3.3 V (20 second period)",
"Rest for 1 hour (120 second period)",
"Charge at 1 A until 4.1 V",
"Hold at 4.1 V until 10 mA",
"Rest for 1 hour",
),
]
* 3
)
model = pybamm.lithium_ion.DFN()
sim = pybamm.Simulation(
model, experiment=experiment, solver=pybamm.CasadiSolver("fast with events")
)
sim.solve()
sim.plot(["Current [A]", "Terminal voltage [V]"])where you can control the resolution of each step by passing a period in brackets (the default is 60s). This produces |
Beta Was this translation helpful? Give feedback.
4 replies
-
|
Thanks for the reply, I already used the first option and it stops when the voltage reaches its cut-off value (safe mode). The question is simply: "how do you make the drive-cycle simulation go to CV mode when it reaches cut-off voltages?" |
Beta Was this translation helpful? Give feedback.
-
|
Once the issue with the casadi solver is fixed (#1902) you will be able to do something like # import drive cycle from file
drive_cycle = pd.read_csv(
"pybamm/input/drive_cycles/US06.csv", comment="#", header=None
).to_numpy()
# define experiment - you would need to define mutliple cycles with different drive cycles
experiment = pybamm.Experiment(
[
"Run US06 (A) until 3.5V",
"Charge at 1 A until 4.0 V",
"Hold at 4.0 V until 50 mA",
"Rest for 30 minutes",
],
drive_cycles={
"US06": drive_cycle,
},
)where the voltage cut off can be specified along with the drive cycle. You will need to have a drive cycle for each of the different currents you want to use and build up a big experiment. While you wait for this to get fixed you could try installing one of the optional DAE solvers (either If you just want to run ODE models (e.g. SPM or SPMe), another workaround would be to do run the drive cycle simulation using an ODE solver ( import pybamm
import pandas as pd
import os
os.chdir(pybamm.__path__[0] + "/..")
model = pybamm.lithium_ion.SPM()
param = model.default_parameter_values
drive_cycle = pd.read_csv(
"pybamm/input/drive_cycles/US06.csv", comment="#", header=None
).to_numpy()
timescale = param.evaluate(model.timescale)
current_interpolant = pybamm.Interpolant(
drive_cycle[:, 0], drive_cycle[:, 1], timescale * pybamm.t
)
# define drive cycle and voltage cut-off
param["Current function [A]"] = current_interpolant
param["Lower voltage cut-off [V]"] = 3.6
sim = pybamm.Simulation(model, parameter_values=param, solver=pybamm.ScipySolver())then use the solution of this as initial conditions to run a CCCV charge, then move on to the next drive cycle. You could set up a script to automate this. Re the divide by zero, you could try setting |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the clarification, I think there is still a misunderstanding when we use words like "drive cycle" or "experiment". Let's forget about these words because by drive cycle, I don't mean CC discharge and I don't want to have a CCCV charge after each drive cycle. Regarding the error, yes, I changed almost every parameter by using the Chen2020 format. These are the parameters I use in COMSOL and I'm pretty sure nothing is wrong with them. I got the error below with setting the logging level to debug. error |
Beta Was this translation helpful? Give feedback.
-
|
I also cannot run for various cycles to obtain the data from the simulated model and simulation ended just at 6 cycle. experiment = pybamm.Experiment( The information was something like this:- 2022-01-19 18:33:21,748 - [NOTICE] simulation.solve(991): Finish experiment simulation, took 782.065 ms Few doubts if @rtimms could you please clear, firstly why in the simulation are 500 cycles showing and not the 100(which I have coded). And why its not going trough each step in a cycle. Instead each cycle corresponds to step only( which should not happen since each step in experiment corresponds to a cycle. |
Beta Was this translation helpful? Give feedback.
-
Agreed, we are certainly getting our wires crossed!
So, just to be totally clear, do you want to do cycles of "CCCV charge+rest+CCCV discharge+rest" but with a different constant current in each cycle? Or are you trying to prescibe current as a function of time at some point? In both cases you will need to define each cycle, but you can automate this somewhat. E.g. if you had a number of different currents you wanted to discharge at for each cycle you could do the following import pybamm
# loop to build experiment
discharge_currents = [0.5, 1, 1.5]
cycles = []
for discharge_current in discharge_currents:
cycle = [
(
# use fstring to give current
f"Discharge at {discharge_current}A for 10 hours or until 3.3 V",
"Rest for 1 hour",
"Charge at 1 A until 4.1 V",
"Hold at 4.1 V until 10 mA",
"Rest for 1 hour",
),
]
# add to existing cycles
cycles += cycle
experiment = pybamm.Experiment(cycles)
# set up and solve sim
model = pybamm.lithium_ion.SPMe()
sim = pybamm.Simulation(model, experiment=experiment, solver=pybamm.CasadiSolver())
sim.solve()
# plot
sim.plot(["Current [A]", "Terminal voltage [V]"])which gives Re the error it looks like it is in setting the parameters in the rhs of the PDE for the positive particle concentration. @Saransh-cpp if you have time can you take a look at the error and see if it is related to the paramater path (I think you fixed a bug related to this recently?), thanks. |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
Thanks, I think now we are on the same page. I have the same current for each cycle but the duration of CC (and possibly CV) is different in each cycle but I think I can define the way you defined it. I need to extract all durations for each cycle which is still time taking. I wish I could just use the current vs time as the input of the model. |
Beta Was this translation helpful? Give feedback.
-
experiment = pybamm.Experiment(
[
(
"Charge at 1 A until 4.0 V",
"Hold at 4.0 V until 50 mA",
"Rest for 30 minutes",
"Run US06_A (A) until 3.7 V",
"Rest for 30 minutes",
)
]
* 100,
termination="80% capacity",
drive_cycles={
"US06_A": drive_cycle_current,
},
)
|
Beta Was this translation helpful? Give feedback.
0 replies
-
I think you can use For the error, can you share the script that you are using? |
Beta Was this translation helpful? Give feedback.
11 replies
-
|
Thanks. The diffusivity problem is now solved. I want to add that I could not use any format other than "example" to insert my SEI parameters because they lack some parameters. error |
Beta Was this translation helpful? Give feedback.
-
|
I’m not sure I follow the issue about “example”. I’m guessing you are putting your parameters in a csv? But you are free to include all the parameters the model needs by adding more values. Some parameter sets may not contain all of the values needed to run all model options. You can do Yes, second problem looks like initial conditions. Hard to say without digging deeper. You could try running a model that is only ODEs (like SPM or SPMe) where consistent initial conditions isn’t an issue, just to check the model is behaving as expected. |
Beta Was this translation helpful? Give feedback.
-
|
We are saying the same thing. For any other component or option, I guess we can use any of the available datasets to parameterize our model; however, for SEI, only the "example" dataset works because the rest of the available datasets do not have the required parameters. Maybe you can add this to the documentation that we need to manually add required parameters if we use datasets other than "example" because I could not find it anywhere. |
Beta Was this translation helpful? Give feedback.
-
|
We are actually planning to reformat how the parameters are supplied with pybamm. As you say, we want to make it clear that the sets we provide are there as examples from the literature and that they may not contain all the parameters needed to run all of the models/options. We also want to improve the documentation around users adding their own parameter sets. Any feedback around this is very much appreciated! Re: the error, it looks like to solver is failing at the first step. Could you try doing |
Beta Was this translation helpful? Give feedback.
-
|
Happy to hear about the parameter reformating. I tried |
Beta Was this translation helpful? Give feedback.
-
|
I just wanted to check in and see if the problem with installing Scikit.ODEs is resolved? I tried to install it via WSL (Ubuntu). The first line When I wanted to run the last line, I got the error below: error |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I think the first issue is related to
pybamm's version. In some older versions, theInterpolantclass used to take indata(having two columnsxandy), but in the latest versions the class takes inxandyseparately.You can check the version by running -