Noisy and implausible capacity fade

[laka7552]

Hi all,

I am using PyBaMM to set up a DFN with degradation mechanisms (SEI growth, LI plating, particle cracking).
I want to evaluate certain thermal management strategies (so different battery temperature) in a discharging cycle.
Initially, I am running a charging discharging test to determine the initial capacity.
Then I am reinitializing the model, applying the current at the given temperature to discharge the battery and then recharge it using the same capacity test than before.

When I look at LLI or LAM, the values seem plausible.
However, the capacity fade calculated from the two tests, is not consistent with the LAM and LLI.
I have also tried doing multiple cycles (up to 10), it seems to improve the reliability but not significantly

Any suggestions what could be the issue here?
Thanks!