|
| 1 | +""" |
| 2 | +Use different Perez coefficients with the ModelChain |
| 3 | +==================================================== |
| 4 | +
|
| 5 | +This example demonstrates how to customize the ModelChain |
| 6 | +to use site-specific Perez transposition coefficients. |
| 7 | +""" |
| 8 | + |
| 9 | +# %% |
| 10 | +# The :py:class:`pvlib.modelchain.ModelChain` object provides a useful method |
| 11 | +# for easily constructing a PV system model with a simple, unified interface. |
| 12 | +# However, a user may want to customize the steps |
| 13 | +# in the system model in various ways. |
| 14 | +# One such example is during the irradiance transposition step. |
| 15 | +# The Perez model perform very well on field data, but |
| 16 | +# it requires a set of fitted coefficients from various sites. |
| 17 | +# It has been noted that these coefficients can be specific to |
| 18 | +# various climates, so users may see improved model accuracy |
| 19 | +# when using a site-specific set of coefficients. |
| 20 | +# However, the base :py:class:`~pvlib.modelchain.ModelChain` |
| 21 | +# only supports the default coefficients. |
| 22 | +# This example shows how the :py:class:`~pvlib.modelchain.ModelChain` can |
| 23 | +# be adjusted to use a different set of Perez coefficients. |
| 24 | + |
| 25 | +import pandas as pd |
| 26 | +from pvlib.pvsystem import PVSystem |
| 27 | +from pvlib.modelchain import ModelChain |
| 28 | +from pvlib.temperature import TEMPERATURE_MODEL_PARAMETERS |
| 29 | +from pvlib import iotools, location, irradiance |
| 30 | +import pvlib |
| 31 | +import os |
| 32 | +import matplotlib.pyplot as plt |
| 33 | + |
| 34 | +# load in TMY weather data from North Carolina included with pvlib |
| 35 | +PVLIB_DIR = pvlib.__path__[0] |
| 36 | +DATA_FILE = os.path.join(PVLIB_DIR, 'data', '723170TYA.CSV') |
| 37 | + |
| 38 | +tmy, metadata = iotools.read_tmy3(DATA_FILE, coerce_year=1990, |
| 39 | + map_variables=True) |
| 40 | + |
| 41 | +weather_data = tmy[['ghi', 'dhi', 'dni', 'temp_air', 'wind_speed']] |
| 42 | + |
| 43 | +loc = location.Location.from_tmy(metadata) |
| 44 | + |
| 45 | +#%% |
| 46 | +# Now, let's set up a standard PV model using the ``ModelChain`` |
| 47 | + |
| 48 | +surface_tilt = metadata['latitude'] |
| 49 | +surface_azimuth = 180 |
| 50 | + |
| 51 | +# define an example module and inverter |
| 52 | +sandia_modules = pvlib.pvsystem.retrieve_sam('SandiaMod') |
| 53 | +cec_inverters = pvlib.pvsystem.retrieve_sam('cecinverter') |
| 54 | +sandia_module = sandia_modules['Canadian_Solar_CS5P_220M___2009_'] |
| 55 | +cec_inverter = cec_inverters['ABB__MICRO_0_25_I_OUTD_US_208__208V_'] |
| 56 | + |
| 57 | +temp_params = TEMPERATURE_MODEL_PARAMETERS['sapm']['open_rack_glass_glass'] |
| 58 | + |
| 59 | +# define the system and ModelChain |
| 60 | +system = PVSystem(arrays=None, |
| 61 | + surface_tilt=surface_tilt, |
| 62 | + surface_azimuth=surface_azimuth, |
| 63 | + module_parameters=sandia_module, |
| 64 | + inverter_parameters=cec_inverter, |
| 65 | + temperature_model_parameters=temp_params) |
| 66 | + |
| 67 | +mc = ModelChain(system, location=loc) |
| 68 | + |
| 69 | +# %% |
| 70 | +# Now, let's calculate POA irradiance values outside of the ``ModelChain``. |
| 71 | +# We do this for both the default Perez coefficients and the desired |
| 72 | +# alternative Perez coefficients. This enables comparison at the end. |
| 73 | + |
| 74 | +# Cape Canaveral seems like the most likely match for climate |
| 75 | +model_perez = 'capecanaveral1988' |
| 76 | + |
| 77 | +solar_position = loc.get_solarposition(times=weather_data.index) |
| 78 | +dni_extra = irradiance.get_extra_radiation(weather_data.index) |
| 79 | + |
| 80 | +POA_irradiance = irradiance.get_total_irradiance( |
| 81 | + surface_tilt=surface_tilt, |
| 82 | + surface_azimuth=surface_azimuth, |
| 83 | + dni=weather_data['dni'], |
| 84 | + ghi=weather_data['ghi'], |
| 85 | + dhi=weather_data['dhi'], |
| 86 | + solar_zenith=solar_position['apparent_zenith'], |
| 87 | + solar_azimuth=solar_position['azimuth'], |
| 88 | + model='perez', |
| 89 | + dni_extra=dni_extra) |
| 90 | + |
| 91 | +POA_irradiance_new_perez = irradiance.get_total_irradiance( |
| 92 | + surface_tilt=surface_tilt, |
| 93 | + surface_azimuth=surface_azimuth, |
| 94 | + dni=weather_data['dni'], |
| 95 | + ghi=weather_data['ghi'], |
| 96 | + dhi=weather_data['dhi'], |
| 97 | + solar_zenith=solar_position['apparent_zenith'], |
| 98 | + solar_azimuth=solar_position['azimuth'], |
| 99 | + model='perez', |
| 100 | + model_perez=model_perez, |
| 101 | + dni_extra=dni_extra) |
| 102 | + |
| 103 | +# %% |
| 104 | +# Now, run the ``ModelChain`` with both sets of irradiance data and compare |
| 105 | +# (note that to use POA irradiance as input to the ModelChain the method |
| 106 | +# `.run_model_from_poa` is used): |
| 107 | + |
| 108 | +mc.run_model_from_poa(POA_irradiance) |
| 109 | +ac_power_default = mc.results.ac |
| 110 | + |
| 111 | +mc.run_model_from_poa(POA_irradiance_new_perez) |
| 112 | +ac_power_new_perez = mc.results.ac |
| 113 | + |
| 114 | +start, stop = '1990-05-05 06:00:00', '1990-05-05 19:00:00' |
| 115 | +plt.plot(ac_power_default.loc[start:stop], |
| 116 | + label="Default Composite Perez Model") |
| 117 | +plt.plot(ac_power_new_perez.loc[start:stop], |
| 118 | + label="Cape Canaveral Perez Model") |
| 119 | +plt.xticks(rotation=90) |
| 120 | +plt.ylabel("AC Power ($W$)") |
| 121 | +plt.legend() |
| 122 | +plt.tight_layout() |
| 123 | +plt.show() |
| 124 | +# %% |
| 125 | +# Note that there is a small, but noticeable difference from the default |
| 126 | +# coefficients that may add up over longer periods of time. |
0 commit comments