Notes on NREL NSRDB PSM4

[Will Hobbs]

If you had not seen it, there are several variations of the NREL NSRDB PSM v4 now available: https://developer.nrel.gov/docs/solar/nsrdb/. pvlib.iotools has been updated to access PSM4 and those changes will be available in a future release. See https://github.com/pvlib/pvlib-python/issues/2326 and https://github.com/pvlib/pvlib-python/pull/2378. 

Notable differences: PSM4 includes 2023 (PSM3 ends at 2022) and appears to fix the "horns" issue explored in https://doi.org/10.1109/PVSC57443.2024.10748762 and elsewhere (see https://github.com/pvlib/pvlib-python/issues/2326#issuecomment-2577927005 for an illustration). 

I've just run some rough tests, and it looks like PSM4 can result in annual AC energy 1-3% lower than PSM3 (spot checking Birmingham, AL, and Denver, CO).

This is obviously not a rigorous comparison, but I thought folks here might find it interesting and/or get some ideas.

Will

Code:

import pvlib

import matplotlib.pyplot as plt

# NREL api info

api_key = 'DEMO_KEY' # <-- change this

email = 'your...@email.com' # <-- change this

def simple_backtracing_power(resource,nameplate_dc,nameplate_ac,

                             eta_inv_nom,gamma_pdc,dc_loss_fraction,

                             gcr,max_angle):

   

    # get solar position

    times = resource.index

    solar_position = pvlib.solarposition.get_solarposition(times, latitude, longitude)

    # array orientation

    orientation = pvlib.tracking.singleaxis(

        solar_position['apparent_zenith'],

        solar_position['azimuth'],

        max_angle=max_angle,

        backtrack=True,

        gcr=gcr,

    )

    # poa irradiance

    poa = pvlib.irradiance.get_total_irradiance(

            surface_tilt=orientation.surface_tilt,

            surface_azimuth=orientation.surface_azimuth,

            dni=resource['dni'],

            ghi=resource['ghi'],

            dhi=resource['dhi'],

            solar_zenith=solar_position['apparent_zenith'],

            solar_azimuth=solar_position['azimuth'])

    # cell temperature

    temp_cell = pvlib.temperature.faiman(

        poa_global=poa['poa_global'],

        temp_air=resource['temp_air'],

        wind_speed=resource['wind_speed'],

    )

    # dc power

    pdc = pvlib.pvsystem.pvwatts_dc(

        g_poa_effective=poa['poa_global'],

        temp_cell=temp_cell,

        pdc0=nameplate_dc,

        gamma_pdc=gamma_pdc,

        )

    pdc_inv = pdc * (1 - dc_loss_fraction)  # dc power into the inverter after losses

    # ac power

    pdc0 = nameplate_ac/eta_inv_nom

    power_ac = pvlib.inverter.pvwatts(pdc_inv, pdc0, eta_inv_nom)

    return power_ac

nameplate_dc = 135

nameplate_ac = 100

eta_inv_nom = 0.98

gamma_pdc = -0.0035

dc_loss_fraction = 0.1

gcr = 0.35

max_angle = 60

names_list = list(map(str,range(1998,2022+1)))

# first location

latitude, longitude = 33.5, -86.8

result=[]

for names in names_list:

    # get PSM3 resource data

    resource, metadata = pvlib.iotools.get_psm3(latitude, longitude, api_key, email, names, interval=30)

    # get PSM4 resource data

    resource4, metadata4 = pvlib.iotools.get_nsrdb_psm4_aggregated(

        latitude, longitude, api_key, email, year=names, time_step=30) # note that input parameter for year is now "year"

    power_ac_psm3 = simple_backtracing_power(resource,nameplate_dc,nameplate_ac,

                                eta_inv_nom,gamma_pdc,dc_loss_fraction,

                                gcr,max_angle)

    power_ac_psm4 = simple_backtracing_power(resource4,nameplate_dc,nameplate_ac,

                                eta_inv_nom,gamma_pdc,dc_loss_fraction,

                                gcr,max_angle)

   

    diff = (power_ac_psm3.sum()-power_ac_psm4.sum())/power_ac_psm3.sum()

    result.append(diff)

plt.bar(names_list,result)

plt.gcf().autofmt_xdate()

plt.ylabel('Relative Difference in Annual AC Energy, \n (psm3 - psm4)/psm3')

plt.title('BHM')

plt.show()

# second location

latitude, longitude = 39.7, -105.0

result=[]

for names in names_list:

    # get PSM3 resource data

    resource, metadata = pvlib.iotools.get_psm3(latitude, longitude, api_key, email, names, interval=30)

    # get PSM4 resource data

    resource4, metadata4 = pvlib.iotools.get_nsrdb_psm4_aggregated(

        latitude, longitude, api_key, email, year=names, time_step=30) # note that input parameter for year is now "year"

    power_ac_psm3 = simple_backtracing_power(resource,nameplate_dc,nameplate_ac,

                                eta_inv_nom,gamma_pdc,dc_loss_fraction,

                                gcr,max_angle)

    power_ac_psm4 = simple_backtracing_power(resource4,nameplate_dc,nameplate_ac,

                                eta_inv_nom,gamma_pdc,dc_loss_fraction,

                                gcr,max_angle)

   

    diff = (power_ac_psm3.sum()-power_ac_psm4.sum())/power_ac_psm3.sum()

    result.append(diff)

plt.bar(names_list,result)

plt.gcf().autofmt_xdate()

plt.ylabel('Relative Difference in Annual AC Energy, \n (psm3 - psm4)/psm3')

plt.title('DEN')

plt.show()