About results of using pvlib.irradiance.get_total_irradiance

[JING LIU]

Dear Sir or Madam,

I am trying to get effective irradiance by  considering incident angle correction first. But confused in the accuracy of using code 'pvlib.irradiance.get_total_irradiance'.

The poa_global I get from the code has error (-100W/m2 to 100W/m2) with my measued poa irradiance.

I am confused in whether the results I get from the code is accurate or is there some other solutions in transpose the measured POA irradiance, GHI, DNI, DHI to effective irradiance when count on AOI modification? Or is the error is acceptable?

My python code and raw data are enclosed. Could anyone help me check it?

Thanks in advance.

Regards,

Jing

[kevina...@gmail.com]

Very few points show error as large as 100 W/m2: 50% of values are within +/- ~10 W/m2 and 90% are within +/- ~30 W/m2.  I think that's pretty decent accuracy overall.

You could try using a different model (instead of 'king') and see if the results improve, but I doubt it will help much with these large-error points.  The problem is not the model, but the fact that the sensors are momentarily experiencing different conditions, and so it might be cloudy according to the DNI sensor but sunny according to the POA sensor, or vice versa.  The attached plot shows an example where the POA sensor reports a dip, but the DNI and DHI are stable.  The transposition model cannot be expected to be accurate in the case when the irradiance measurements are inconsistent with each other.

error = poa_irr['poa_global'] - df['POAirradiance']
fig, axes = plt.subplots(2, 1, sharex=True)
df[['DNI', 'DHI', 'POAirradiance']].plot(ax=axes[0])
error.plot(ax=axes[1])
axes[1].set_ylabel('Difference (modeled - measured) [W/m2]')

Best,

Kevin