I have some issues when simulating model.

[gyuyoung-hwang]

Hi Dr. Khalik,

At first, I really appreciated sharing your greatest model. It helps me to understand how p2d model does work, along with your papers.

Here, I have some questions while simulationg some cases with your model.

For instance,

When I opened 'discharge_charge_cycle.m' as an example code and run that code, then I changed the value of p.Crate_charge (0.5C) with high C-rate (2C, 3C) under 1C discharge rate, I found simulation didn't work or showed .

Specifically, charge stage after discharging didn't happen. I guess it would be caused by electrolyte concentration because I found phie has a large deviated values when converting charge mode. (Actually, after high rate discharge, electrolyte concentration at the end of the positive region has almost zero value so that it causes problem for the unstable convergence.)

Also, when I changed 'p.set_simp' (simplification) with [1,1,1,1,0,0] (using previous state values), the simulation cannot work more often, compared with the set value of [2,2,2,2,0,0]. For example, in the case of the simulation under 1C-discharge/1C-charge, charge stage kept going with [2,2,2,2,0,0]. In the other hand, charge stage didnt continue after discharging with [1.1.1.1,0,0]

Along with it, 1C-rate discharging took a short time less than 3600s (about 3300s with [1,1,1,1,0,0] but 2700s with [2,2,2,2,0,0]).

However, C-rataes under 0.5C shows ideal results ( e.g., 0.5C takes almost 7200 sec)

I think the parameters are fitted well for the default case (0.5C charging and discharging). Is it right?

Here, I want to ask you how simplfication and C-rate make such siginifant difference.

Considering thoese facts, if I changed some parmeters such as particle radius or diffusivity and do not introduce simplification (i.e., use [1 1 1 1 0 0], can I handle such issusues and get more accurate results ?

Thank you for reading this issues.

---

After I uploaded this issues, I changed electrolyte diffusivity with the one from 'parameter_LS'. As a result, 2C charge stage right after 2C discharge could start to be converged. Thank you.

However, I have other question, the temperature change during cell operation can be observed but some coefficient including Diffusivity didn't consider temperature dynamics. Could you please check this part, together?

[Zuan-Khalik]

Hi Dr. Khalik,

At first, I really appreciated sharing your greatest model. It helps me to understand how p2d model does work, along with your papers.

Here, I have some questions while simulationg some cases with your model.

For instance,

When I opened 'discharge_charge_cycle.m' as an example code and run that code, then I changed the value of p.Crate_charge (0.5C) with high C-rate (2C, 3C) under 1C discharge rate, I found simulation didn't work or showed .

Specifically, charge stage after discharging didn't happen. I guess it would be caused by electrolyte concentration because I found phie has a large deviated values when converting charge mode. (Actually, after high rate discharge, electrolyte concentration at the end of the positive region has almost zero value so that it causes problem for the unstable convergence.)

I have also run this simulation, and you are indeed correct, the limitation in this case is the electrolyte concentration coming close to 0. What you could do to improve convergence, is to take a lower sampling time. For example, if I choose all 3 dt's as 1, then the simulation progresses further, until cs reaches cs_max in the negative electrode at some point. Generally, when you faces these limitations, playing around with the sampling time can be helpful.

Also, when I changed 'p.set_simp' (simplification) with [1,1,1,1,0,0] (using previous state values), the simulation cannot work more often, compared with the set value of [2,2,2,2,0,0]. For example, in the case of the simulation under 1C-discharge/1C-charge, charge stage kept going with [2,2,2,2,0,0]. In the other hand, charge stage didnt continue after discharging with [1.1.1.1,0,0]

This is also expected, as applying various simplifications changes the model dynamics. Hence, it could be that when you apply less simplifications, you may hit model limitations sooner than with more simplifications. In this case, when I choose lower sampling times, the simulation runs completely (albeit slower, of course).

Along with it, 1C-rate discharging took a short time less than 3600s (about 3300s with [1,1,1,1,0,0] but 2700s with [2,2,2,2,0,0]).

This is due to the definition of the C-rate. The way I defined C-rate is using the EMF capacity of the battery (the amount of charge that the battery can charge/discharge with an infinitely small current). However, commonly 1 C-rate is defined as the current that charges (or discharges) the battery in 1 hour (when starting from a particular voltage until a threshold voltage is reached. Since there are resistive effects at play, the capacity corresponding to this 1 C-rate is less than the EMF capacity. This is why with the definition that I have used for 1 C, the battery discharges to 2.7 V in less than an hour.

Why you see such a large difference with the different simplifications, is that in this case, when applying more simplifications, the battery shows a smaller degree of diffusivity, and therefore you see a higher overpotential at the voltage. This higher overpotential causes the battery to reach 2.7V sooner.

However, C-rataes under 0.5C shows ideal results ( e.g., 0.5C takes almost 7200 sec)

At lower C-rates, there is also a lower overpotential, and so the effect that I described above decreases.

I think the parameters are fitted well for the default case (0.5C charging and discharging). Is it right?

I have taken the parameters from literature, and I am not really sure for which case they have been fit exactly. This particular set of parameters does show a low rate of diffusivity, which causes model limitations to show up sooner.

Here, I want to ask you how simplfication and C-rate make such siginifant difference.
Generally, the simplifications becomes less valid as the current increases, or more specifically, as the battery is reaching its limits. For more details on this, I would like to refer you to the paper that this toolbox is based on. If you have any specific questions after that, I would be happy to answer them.

Considering thoese facts, if I changed some parmeters such as particle radius or diffusivity and do not introduce simplification (i.e., use [1 1 1 1 0 0], can I handle such issusues and get more accurate results ?

Actually, [1 1 1 1 0 0] is pretty much the non-simplified case, so you should take that as the baseline. However, no matter if you take simplifications or not, with a high enough C-rate, you will hit the limitations of the model with any set of parameters, although it varies how soon you hit these limitations with different sets of parameters.

Thank you for reading this issues.

After I uploaded this issues, I changed electrolyte diffusivity with the one from 'parameter_LS'. As a result, 2C charge stage right after 2C discharge could start to be converged. Thank you.

However, I have other question, the temperature change during cell operation can be observed but some coefficient including Diffusivity didn't consider temperature dynamics. Could you please check this part, together?

I can try to help you here, which coefficient do you mean exactly?

[gyuyoung-hwang]

Dear Dr.Khalick

I don’t know how to express the great thanks for your answers.

I’m so appreciated your answer, again

I’ll try to run model agian along with your help and guidence.

After that, I’ll leave some comments again with great thanks.

Have a nice day!

sincerely.

[Zuan-Khalik]

You're welcome, glad I could help! Feel free to ask any other questions you may have. I'll try to do my best to answer to them as quickly as possible :)