Solver conditioning with L**2 #3918
-
|
In the basic_DFN model implementation, I can see that the potential equations are conditioned with L**2 .
I have tried with and without this conditioning and I see the results dont vary too much. I understand the conditioning is to help thewith speedup, but is it mandatory to keep this term in the solution? Is this dimensionally correct ? Just wondering, thanks ! |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 2 replies
-
|
Hi @veera-visw , L_x should be a significant value, without it, i think, will affect the conservation of the equation. Thus, it should change the result. I suggest you present an example code for others' convenience for checking this. |
Beta Was this translation helpful? Give feedback.
-
|
I have included this term in my equations now. I dont want to reflect yet on why the absence gave similar results. But I am happy with the results I get from including the conditioning term. Does the magnitude of this value matter? Right now, I am using the total length of the 1D domain as my conditioner. |
Beta Was this translation helpful? Give feedback.
-
|
The value has no effect on the results, but including it helps the numerical solver to converge more easily since it makes the overall magnitude of the equation closer to 1 (otherwise it would be very large) |
Beta Was this translation helpful? Give feedback.
Hi @veera-visw , L_x should be a significant value, without it, i think, will affect the conservation of the equation. Thus, it should change the result. I suggest you present an example code for others' convenience for checking this.