Repository navigation
Allow arbitrary IAM function in pvlib.iam.marion_diffuse #2049
Description
Is your feature request related to a problem? Please describe.
I am working on an application using pvlib.iam.marion_diffuse, and I would like to inject my own IAM function (from interpolated IEC 61853-2 data) as a parameter to pvlib.iam.marion_diffuse.
Describe the solution you'd like
Add the option for passing a callable to pvlib.iam.marion_diffuse instead of the existing strings for the builtin IAM model functions. (The existing strings would still be accepted.)
Also add a function to create IAM functions as a PCHIP interpolation of data such as from IEC 61853-2. Note that this solution can also be made to add support for passing the existing pvlib.iam.interp function (whose cubic option is slightly different that pchip) to pvlib.iam.marion_diffuse.
Describe alternatives you've considered
Rolling my own interpolant each time, and then calling the lower-level pvlib.iam.marion_integrate for each of the three components.
Additional context
This is a followup to #1980, but without any concern to improving the diffuse-horizontal integral.
Activity
adriesse commentedon May 13, 2024
Have you considered adding the pchip option into
pvlib.iam.interprather than creating a new function?markcampanelli commentedon May 13, 2024
@adriesse
I am open to this consolidation. Do you think folks would be open to
pvlib.iam.interpreturning acallablethat could then be passed to the newpvlib.iam.marion_diffuse(or used for other calculations)?If I understand correctly, this avoids some overhead of recreating the interpolating function on each usage, although it would break the existing interface to
pvlib.iam.interp.adriesse commentedon May 14, 2024
I think Marion only has one call to the model.
markcampanelli commentedon Jul 1, 2024
I noticed that the builtin
interpandschlickAOI modifiers didn't have full support inModelChainlike the other builtins, so I'm trying to add that here too.echedey-ls commentedon Jul 1, 2024
Feel free to break down these issues into different PRs, doing too many things in one big PR makes the review process more time consuming - and reviewers more hesitant to review.