BUG: Extend current in calcSeries to quad4

[mikofski]

• fixes "Resolution" of the I-V plots made with plotSys() changes with cell temperature #110 - bad resolution for IV curve at high temperature
• change Iforward range to be from 0 to meanIsc, instead of from Imin
  which was TOO SMALL, and use pvconst.Imod_pts which has closer spacing
  near zero than pvconst.pts
• add quad4 range from Imin to zero, where Imin is a negative current in
  the 4th quadrant, using pvconst.Imod_negpts which goes from 1 to 0,
  with closer spacing near zero
• concatenate Iquad4 before Iforward, and increase Vtot to same size
• also add new pvconst.Imod_pts_sq which has even closer spacing near
  zero to use with Ireverse, so there's more than 1 point before the
  breakdown voltage!
• add pvstr.Voc_mod which is a list of the sum of Voc for all cells in
  each module
• and add pvsys.Voc_str which is a list of the sum of the Voc for all
  modules in each string
• then use pvsys.Voc_str.max() instead of pvsys.Vstring.max() in
  calcParallel so there's not too many points in the 4th quadrant pull
  thing the curve way off, and messing up the resolution in the forward
  bias region (1st quadrant)
• add a test in test_pvconstants to test the minimum current close to
  max voc by comparing the IVP curve of pvsys with data from gh110.dat
  which contains the correct resolution for a curve with npts=101, the
  default at Tcell=288[K]
• also update test_pvcell.test_calc_series() with new data, but also
  compare the old data by interpolating with the new - we can delete the
  old data eventually right?
• still a few more bugs, tests to fix, so this is a WIP

[mikofski]

OK, fixed partial cross ties, but now total cross ties is broken

[mikofski]

@adambgnr and @chetan201 I think this is ready for review.

@adambgnr please bear in mind that when I started pvmismatch like 7 years ago, I was just learning Python so there's a lot of really bad code in here that should probably be fixed up, but I don't think this PR covers that.

This PR just fixes #110 the high temperature system, string, and module problem. The bug was caused by #73 a fix for #36 which solved the high temperature problem for voltage in the cell. This fix is the same approach, but at the module, string, and system level, and splits the IV curve into 3 regions instead of 2, when interpolating either voltage or current:

• quadrant 4
• forward bias
• reverse bias

Let me know if you have any questions, and please open separate issues if you want to address improving the code for readability or performance. Thanks!

[mikofski]

@chetan201 ready for review. @adambgnr feel free to add your comments, as you're the one who caught the bug.

[adambgnr]

@chetan201 ready for review. @adambgnr feel free to add your comments, as you're the one who caught the bug.

Thanks a lot for making this fix @mikofski! Indeed now the plots (and the underlying results) are nice and smooth regardless of the temperature. While testing, I have found one thing that I don't understand, see in the comments at npts().

[mikofski]

Here's what the different spacings are supposed to look like:

So using np.flip(Imod_negpts) would reverse the spacing to create tighter spacing near the max power point, and less spacing as Vsys -> Voc.

from pvmismatch import *
pvconst = pvconstants.PVconstants(npts=20)
pvsys = pvsystem.PVsystem(pvconst=pvconst)
f = pvsys.plotSys(fmt='+:')
f.show()

It looks like this:

I like it better. The benefit is that there are more points around the knee of the max power point which is kind of what we originally wanted so it's easier to determine the max power, and the only downside is that we may miss some curvature near Voc, but the user can increase npts if they want more detail there.

I also updated the plotSys(fmt='') to allow specification of plotting formats like "+:" used here, and I saved the new set of points in PVconstants as Vmod_q4pts. I noticed that Imod_pts isn't used anywhere anymore except to create Imod_pts_sq which is a really tight spacing to capture the reverse breakdown, which is very flat and fast for avalanche. Maybe there should be a ticket to remove this at some point, not sure if deprecation is required, since it should be private anyway. Also I added sharex and tight_layout() to the figures so they look nicer and zoom more easily.

[adambgnr]

It looks like this:

I like it better. The benefit is that there are more points around the knee of the max power point which is kind of what we originally wanted so it's easier to determine the max power, and the only downside is that we may miss some curvature near Voc, but the user can increase npts if they want more detail there.

I also updated the plotSys(fmt='') to allow specification of plotting formats like "+:" used here

I like it too! Also I like the new fmt='+:' option for plotSys()! Much more informative than just the line plot.

[chetan201]

@mikofski @adambgnr
Thanks for catching and fixing this. Since it changes the absolute value of the MPP, I would like to review a few cases before I approve. Thanks for your patience!

[mikofski]

@chetan201 - gentle nudge 😉