|
| 1 | +""" |
| 2 | +Varying Model Components in ModelChain |
| 3 | +====================================== |
| 4 | +
|
| 5 | +This example demonstrates how changing modeling components |
| 6 | +within ``pvlib.modelchain.ModelChain`` affects simulation results. |
| 7 | +
|
| 8 | +Using the same PV system and weather data, we create two |
| 9 | +ModelChain instances that differ only in their temperature |
| 10 | +model. By comparing the resulting cell temperature and AC |
| 11 | +power output, we can see how changing a single modeling |
| 12 | +component affects overall system behavior. |
| 13 | +""" |
| 14 | + |
| 15 | +# %% |
| 16 | +# Varying ModelChain components |
| 17 | +# ------------------------------ |
| 18 | +# |
| 19 | +# Below, we create two ModelChain objects with identical system |
| 20 | +# definitions and weather inputs. The only difference between them |
| 21 | +# is the selected temperature model. This highlights how individual |
| 22 | +# modeling components in ``ModelChain`` can be swapped while keeping |
| 23 | +# the overall workflow unchanged. |
| 24 | + |
| 25 | +import pvlib |
| 26 | +import pandas as pd |
| 27 | +import numpy as np |
| 28 | +import matplotlib.pyplot as plt |
| 29 | + |
| 30 | +# %% |
| 31 | +# Define location |
| 32 | +# --------------- |
| 33 | +# |
| 34 | +# We select Tucson, Arizona, a location frequently used in pvlib |
| 35 | +# examples due to its strong solar resource and available TMY data. |
| 36 | +latitude = 32.2 |
| 37 | +longitude = -110.9 |
| 38 | +location = pvlib.location.Location(latitude, longitude) |
| 39 | + |
| 40 | +# %% |
| 41 | +# Generate clear-sky weather data |
| 42 | +# -------------------------------- |
| 43 | +# |
| 44 | +# We generate clear-sky irradiance using pvlib and create a |
| 45 | +# varying air temperature profile instead of using constant |
| 46 | +# values. |
| 47 | +times = pd.date_range( |
| 48 | + "2019-06-01 00:00", |
| 49 | + "2019-06-07 23:00", |
| 50 | + freq="1h", |
| 51 | + tz="Etc/GMT+7", |
| 52 | +) |
| 53 | + |
| 54 | +# Clear-sky irradiance |
| 55 | +clearsky = location.get_clearsky(times) |
| 56 | + |
| 57 | +# Create a simple daily temperature cycle |
| 58 | +temp_air = 20 + 10 * np.sin(2 * np.pi * (times.hour - 6) / 24) |
| 59 | + |
| 60 | +weather_subset = clearsky.copy() |
| 61 | +weather_subset["temp_air"] = temp_air |
| 62 | +weather_subset["wind_speed"] = 1 |
| 63 | + |
| 64 | +# %% |
| 65 | +# Define a simple PV system |
| 66 | +# ------------------------- |
| 67 | +# |
| 68 | +# To keep the focus on the temperature model comparison, |
| 69 | +# we define a minimal PV system using the PVWatts DC and AC models. |
| 70 | +# These models require only a few high-level parameters. |
| 71 | +# |
| 72 | +# The module DC rating (pdc0) represents the array capacity at |
| 73 | +# reference conditions, and gamma_pdc describes the power |
| 74 | +# temperature coefficient. |
| 75 | +# |
| 76 | +# For the temperature model parameters, we use the sapm values |
| 77 | +# for an open-rack, glass-glass module configuration. These |
| 78 | +# parameters describe how heat is transferred from the module |
| 79 | +# to the surrounding environment. |
| 80 | +module_parameters = dict(pdc0=5000, gamma_pdc=-0.003) |
| 81 | +inverter_parameters = dict(pdc0=4000) |
| 82 | + |
| 83 | +temperature_model_parameters = ( |
| 84 | + pvlib.temperature.TEMPERATURE_MODEL_PARAMETERS["sapm"] |
| 85 | + ["open_rack_glass_glass"] |
| 86 | +) |
| 87 | + |
| 88 | +system = pvlib.pvsystem.PVSystem( |
| 89 | + surface_tilt=30, |
| 90 | + surface_azimuth=180, |
| 91 | + module_parameters=module_parameters, |
| 92 | + inverter_parameters=inverter_parameters, |
| 93 | + temperature_model_parameters=temperature_model_parameters, |
| 94 | +) |
| 95 | + |
| 96 | +# %% |
| 97 | +# ModelChain using the sapm temperature model |
| 98 | +# -------------------------------------------- |
| 99 | +# |
| 100 | +# First, we construct a ModelChain that uses the sapm |
| 101 | +# temperature model. All other modeling components remain |
| 102 | +# identical between simulations. |
| 103 | +# |
| 104 | +# This ensures that any differences in the results arise |
| 105 | +# solely from the temperature model choice. |
| 106 | +temperature_model_parameters_sapm = ( |
| 107 | + pvlib.temperature.TEMPERATURE_MODEL_PARAMETERS["sapm"] |
| 108 | + ["open_rack_glass_glass"] |
| 109 | +) |
| 110 | + |
| 111 | +system_sapm = pvlib.pvsystem.PVSystem( |
| 112 | + surface_tilt=30, |
| 113 | + surface_azimuth=180, |
| 114 | + module_parameters=module_parameters, |
| 115 | + inverter_parameters=inverter_parameters, |
| 116 | + temperature_model_parameters=temperature_model_parameters_sapm, |
| 117 | +) |
| 118 | + |
| 119 | +mc_sapm = pvlib.modelchain.ModelChain( |
| 120 | + system_sapm, |
| 121 | + location, |
| 122 | + dc_model="pvwatts", |
| 123 | + ac_model="pvwatts", |
| 124 | + temperature_model="sapm", |
| 125 | + aoi_model="no_loss", |
| 126 | +) |
| 127 | + |
| 128 | +mc_sapm.run_model(weather_subset) |
| 129 | + |
| 130 | +# %% |
| 131 | +# ModelChain using the Faiman temperature model |
| 132 | +# ---------------------------------------------- |
| 133 | +# |
| 134 | +# Next, we repeat the same simulation but replace the |
| 135 | +# temperature model with the Faiman model. |
| 136 | +# |
| 137 | +# No other system or weather parameters are changed. |
| 138 | +# This illustrates how individual components within |
| 139 | +# ModelChain can be varied independently. |
| 140 | +temperature_model_parameters_faiman = dict(u0=25, u1=6.84) |
| 141 | + |
| 142 | +system_faiman = pvlib.pvsystem.PVSystem( |
| 143 | + surface_tilt=30, |
| 144 | + surface_azimuth=180, |
| 145 | + module_parameters=module_parameters, |
| 146 | + inverter_parameters=inverter_parameters, |
| 147 | + temperature_model_parameters=temperature_model_parameters_faiman, |
| 148 | +) |
| 149 | + |
| 150 | +mc_faiman = pvlib.modelchain.ModelChain( |
| 151 | + system_faiman, |
| 152 | + location, |
| 153 | + dc_model="pvwatts", |
| 154 | + ac_model="pvwatts", |
| 155 | + temperature_model="faiman", |
| 156 | + aoi_model="no_loss", |
| 157 | +) |
| 158 | + |
| 159 | +mc_faiman.run_model(weather_subset) |
| 160 | + |
| 161 | +# %% |
| 162 | +# Compare modeled cell temperature |
| 163 | +# --------------------------------- |
| 164 | +# |
| 165 | +# Since module temperature directly affects DC power |
| 166 | +# through the temperature coefficient, differences |
| 167 | +# between temperature models can propagate into |
| 168 | +# performance results. |
| 169 | + |
| 170 | +# %% |
| 171 | +fig, ax = plt.subplots(figsize=(10, 4)) |
| 172 | +mc_sapm.results.cell_temperature.plot(ax=ax, label="SAPM") |
| 173 | +mc_faiman.results.cell_temperature.plot(ax=ax, label="Faiman") |
| 174 | + |
| 175 | +ax.set_ylabel("Cell Temperature (°C)") |
| 176 | +ax.set_title("Comparison of Cell Temperature") |
| 177 | +ax.legend() |
| 178 | +plt.tight_layout() |
| 179 | + |
| 180 | +# %% |
| 181 | +# Compare AC power output |
| 182 | +# ------------------------ |
| 183 | +# |
| 184 | +# Finally, we compare the resulting AC power. In this case, the |
| 185 | +# differences in temperature modeling lead to small |
| 186 | +# differences in predicted energy production. |
| 187 | + |
| 188 | +# %% |
| 189 | +fig, ax = plt.subplots(figsize=(10, 4)) |
| 190 | +mc_sapm.results.ac.plot(ax=ax, label="SAPM") |
| 191 | +mc_faiman.results.ac.plot(ax=ax, label="Faiman") |
| 192 | + |
| 193 | +ax.set_ylabel("AC Power (W)") |
| 194 | +ax.set_title("AC Output with Different Temperature Models") |
| 195 | +ax.legend() |
| 196 | +plt.tight_layout() |
0 commit comments