See the fourth line in the "hourly simulation" integration test. Possibly not a big deal, but we should investigate the reason later.
There is already (a related?) critical discussion in the documentation on method Calc_ExtraterrestrialRadiation_V1.
My first guess is, that using the max function is too simple and causes inconsistency with the possibly sunshine duration:
flu.extraterrestrialradiation = max(
12.0
* 4.92
/ d_pi
* flu.earthsundistance
* (
(
(d_omega2 - d_omega1)
* modelutils.sin(der.latituderad)
* modelutils.sin(flu.solardeclination)
)
+ (
modelutils.cos(der.latituderad)
* modelutils.cos(flu.solardeclination)
* (modelutils.sin(d_omega2) - modelutils.sin(d_omega1))
)
),
0.0,
)