PySAM DOE Data Prize Webinar

Modeling capabilities of PySAM and pvlib with use cases for DOE Data Prize datasets

Setting up PySAM case

1. Import PySAM modules

2. Load in weather data

3. Set up PV system design

4. Run model timestep-by-timestep

5. Compare results to measured plant data

6. Extra model features for monitoring, etc.

#Setting up PySAM
#bring in PV module
import PySAM as pysam
import PySAM.Pvsamv1 as pv
#bring in LCOE calculator for example
import PySAM.Lcoefcr as lcoe
import matplotlib.pyplot as plt
import pandas as pd #Data processing
import json
import datetime
import PySAM.ResourceTools

Weather Data

Option 1. Read from file

-Use SAM weather file formats such as that downloaded from NSRDB

Option 2: Read from dictionary

-solar_resource_data dictionary can store values for ‘dhi’, ‘ghi’, ‘dni’, ‘poa’, ‘tamb’, etc. for full year, partial year, or single time step simulations

DOE Data Prize Datasets

• System 9068 (SR_CO) system

• Single-axis tracking system

#Weather data
#Additional data access tools available at https://github.com/NREL/pvdaq_access
weather_data_prize = pd.read_csv('9068_irradiance_data.csv')
amb_weather_data = pd.read_csv('9068_environment_data.csv')
weather_data_prize["measured_on"] = pd.DatetimeIndex(pd.to_datetime(weather_data_prize["measured_on"], utc=False))
weather_data_prize = weather_data_prize.set_index("measured_on")
weather_data_prize.head()

amb_weather_data["measured_on"] = pd.DatetimeIndex(pd.to_datetime(amb_weather_data["measured_on"], utc=False))
amb_weather_data = amb_weather_data.set_index("measured_on")
amb_weather_data.head()

	weather_station_ambient_temperature_(c)_o_149727	thermocouple_pad_1_back-of-module_temperature_1_(c)_o_149728	thermocouple_pad_1_back-of-module_temperature_2_(c)_o_149729	thermocouple_pad_2_back-of-module_temperature_1_(c)_o_149730	thermocouple_pad_2_back-of-module_temperature_2_(c)_o_149731	wind_sensor_max_wind_speed_(m/s)_o_149732	wind_sensor_wind_direction_o_149733	wind_sensor_wind_direction_fast_o_149735	wind_sensor_wind_speed_(m/s)_o_149736	wind_sensor_wind_speed_fast_(m/s)_o_149737	...	inverter_module_2.4_temperature_t2_(c)_inv_149817	inverter_module_2.4_temperature_t3_(c)_inv_149818	inverter_1_max_igbt_temperature_(c)_inv_149819	inverter_1_min_igbt_temperature_(c)_inv_149820	inverter_1_derating_ambient_temperature_(c)_inv_149821	inverter_1_temperature_int_(c)_inv_149822	inverter_2__max_igbt_temperature_(c)_inv_149823	inverter_2_min_igbt_temperature_(c)_inv_149824	inverter_2_derating_ambient_temperature_(c)_inv_149825	inverter_2_temperature_int_(c)_inv_149826
measured_on
2017-08-23 17:45:00	0.0	39.20	39.20	38.6	39.8	0.0	0.0	0.0	0.0	0.0	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
2017-08-23 17:50:00	0.0	37.95	37.95	37.3	38.6	0.0	0.0	0.0	0.0	0.0	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
2017-08-23 17:55:00	0.0	36.60	36.60	35.9	37.3	0.0	0.0	0.0	0.0	0.0	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
2017-08-23 18:00:00	0.0	35.50	35.50	34.8	36.2	0.0	0.0	0.0	0.0	0.0	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
2017-08-23 18:05:00	0.0	34.65	34.65	34.1	35.2	0.0	0.0	0.0	0.0	0.0	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN

5 rows × 99 columns

#Weather data for 2019
weather_data_prize_2019 = weather_data_prize.loc['2019']
amb_weather_data_2019 = amb_weather_data.loc['2019']

weather_data_prize_2019.to_csv('2019_weather_data.csv')
amb_weather_data_2019.to_csv('2019_amb_data.csv')

PV System Design

Option 1 (recommended): set up PV system design in SAM GUI

• Set module, inverter specs

• Set modules per string, number of parallel strings

• Define tilt (or tracking), azimuth, GCR

• Specify system losses

• Once system design is set, ‘Generate Code’ to PySAM json

# create a new instance of the Pvsamv1 module
pv_model = pv.new()
# get the inputs from the JSON file from GUI
with open( 'System_9068_pvsamv1.json', 'r') as f:
        pv_inputs = json.load( f )

# iterate through the input key-value pairs and set the module inputs
for k, v in pv_inputs.items():
    if k != 'number_inputs':
        pv_model.value(k, v)

solar_resource_data = {
        'tz': -7, # timezone
        'elev': 1407.0, # elevation
        'lat': 40.3864, # latitude
        'lon': -104.5512, # longitude
        'year': tuple(weather_data_prize_2019.index.year), # year
        'month': tuple(weather_data_prize_2019.index.month), # month
        'day': tuple(weather_data_prize_2019.index.day), # day
        'hour': tuple(weather_data_prize_2019.index.hour), # hour
        'minute': tuple(weather_data_prize_2019.index.minute), # minute
        'poa': tuple(weather_data_prize_2019['pyranometer_(class_a)_pad_2_poa_irradiance_(w/m2)_o_149725']),
        'wspd': tuple(amb_weather_data_2019['wind_sensor_max_wind_speed_(m/s)_o_149732']), # windspeed
        'tdry': tuple(amb_weather_data_2019['weather_station_ambient_temperature_(c)_o_149727']) # dry bulb temperature
        }
#pv_model.SolarResource.assign({'solar_resource_data': solar_resource_data})
#Assign weather data from NSRDB
pv_model.SolarResource.assign({'solar_resource_file': '40.3864_-104.5512_psm3-5min_5_2021.csv'})

Run model

• pv_model.execute() to run SAM module

• Outputs in pv_model.Outputs, can export to dictionary with .export()

• Time series, annual summation metrics

• “gen”: time series kW output

pv_model.execute(0)
out = pv_model.Outputs.export()
predictions = pd.DataFrame({"AC": out["gen"]})
predictions["AC"].plot()
print(pv_model.Outputs.annual_energy, " kWh")

9580759.524304047  kWh

Compare to Measured Data

-Use measured data to compare against simulated results

-Can compare timestep to timestep (non-annual), or for entire year

measured_data = pd.read_csv("9068_ac_power_data.csv")
measured_data["measured_on"] = pd.DatetimeIndex(pd.to_datetime(measured_data["measured_on"], utc=False))
measured_data = measured_data.set_index("measured_on")
measured_data_ex = measured_data.loc["2022"]
predictions = predictions.set_index(measured_data_ex.index.copy())
measured_data_ex
measured_ac_kw = measured_data_ex["inverter_1_ac_power_(kw)_inv_150143"] + measured_data_ex["inverter_2_ac_power_(kw)_inv_150144"]

predictions['AC'].plot()
measured_ac_kw.plot()
plt.legend(["Predicted power", "Actual power"])
plt.ylabel("Power (kW)")
measured_ac_kwh = measured_ac_kw * 5.0 / 60.0 #5-minute data

Financial Model

• SAM has detailed financial models with discounted cash flows

• Front of meter, behind the meter perspectives

• Also able to run simple LCOE calculator

lcoe_model = lcoe.new()
lcoe_model.SimpleLCOE.annual_energy = pv_model.Outputs.annual_energy
lcoe_model.SimpleLCOE.capital_cost = 10000000 #capex in $, not based on actual cost data
lcoe_model.SimpleLCOE.fixed_charge_rate = 0.1 #Can be calculated
lcoe_model.SimpleLCOE.fixed_operating_cost = 50000 #annual opex in $, not based on actual cost data
lcoe_model.SimpleLCOE.variable_operating_cost = 0.1 #variable operating cost in $/kWh, not based on actual cost data
lcoe_model.execute()
print("Levelized Cost of Energy:",lcoe_model.Outputs.lcoe_fcr, " $/kWh")

Levelized Cost of Energy: 0.2095946513777333  $/kWh

Non-annual Simulations

• Run subset of year or single time step

• Monitor measured vs modeled based on incoming data streams

• Focused analysis on segments of year, individual days

• Faster parametric sweeps, iteration over input options and ranges

• Use solar_resource_data instead of solar_resource_file

pv_model = pv.new()
#Model single inverter instead of both inverters
with open( 'System_9068_Inv_pvsamv1.json', 'r') as f:
        pv_inputs = json.load( f )

# iterate through the input key-value pairs and set the module inputs
for k, v in pv_inputs.items():
    if k != 'number_inputs':
        pv_model.value(k, v)

#Unassign Weather file to use dictionary of weather values instead
pv_model.unassign("solar_resource_file")
#Unassign other annual values
pv_model.unassign("subarray1_shading_timestep")

#Pull weather data from site for specific day with trackers not moving all day (9/16/23)
weather_data_prize_example = weather_data_prize.loc[datetime.date(2023, 9, 16):datetime.date(2023,9,17)]
amb_weather_data_example = amb_weather_data.loc[datetime.date(2023, 9, 16):datetime.date(2023,9,17)]
#weather_data_prize_example = weather_data_prize["2023-9-16"]

#Assign solar_resource_data with poa and solar data from site data
solar_resource_data = {
        'tz': -7, # timezone
        'elev': 1407.0, # elevation
        'lat': 40.3864, # latitude
        'lon': -104.5512, # longitude
        'year': tuple(weather_data_prize_example.index.year), # year
        'month': tuple(weather_data_prize_example.index.month), # month
        'day': tuple(weather_data_prize_example.index.day), # day
        'hour': tuple(weather_data_prize_example.index.hour), # hour
        'minute': tuple(weather_data_prize_example.index.minute), # minute
        'poa': tuple(weather_data_prize_example['pyranometer_(class_a)_pad_2_poa_irradiance_temp_compensated_(w/m2)_o_149726']),
        'wspd': tuple(amb_weather_data_example['wind_sensor_max_wind_speed_(m/s)_o_149732']), # windspeed
        'tdry': tuple(amb_weather_data_example['weather_station_ambient_temperature_(c)_o_149727']) # dry bulb temperature
}

#Assign solar_resource_data to model, set irradiance calculation mode to POA from pyranometer (determined from site metadata)
pv_model.SolarResource.assign({"solar_resource_data": solar_resource_data})
pv_model.SolarResource.assign({"irrad_mode":4}) #4 - POA irradiance from pyranometer

#Run 1 day analysis
pv_model.execute(1)

#Load tracker data
tracker_data = pd.read_csv("9068_tracker_data.csv")
tracker_data["measured_on"] = pd.DatetimeIndex(pd.to_datetime(tracker_data["measured_on"], utc=False))
tracker_data = tracker_data.set_index("measured_on")

Custom Tracker Rotation Angles

• User-defined tracker rotation angles rather than default SAM sun-following algorithm

• SAM covers sun-following tracker algorithm with backtracking, slope-aware backtracking (single fixed slope)

• Useful for analyzing misaligned tracker, stuck trackers, custom tracker algorithms

• Currently SDK only, looking at desktop tool integration in future

#Pull tracker rotation angles from site data
stuck_tracker_data = tracker_data.loc[datetime.date(2023, 9, 16):datetime.date(2023,9,17), "tracker_1.1-101_tracker_angle_trkr_149827"]

#Turn off custom_rotation_angles
pv_model.SystemDesign.subarray1_use_custom_rot_angles = 0
pv_model.execute(1)
out = pv_model.Outputs.export()
tracker_predictions = pd.DataFrame({"sun_angles": out['subarray1_axisrot']})

#Turn on custom rotation angles, set array to tracker angles from array
pv_model.SystemDesign.subarray1_use_custom_rot_angles = 1
pv_model.SystemDesign.subarray1_custom_rot_angles_array = stuck_tracker_data
pv_model.execute(1)
out = pv_model.Outputs.export()
tracker_predictions["Stuck_tracking"] = out["gen"]
tracker_predictions["custom_angles"] = out["subarray1_axisrot"]
tracker_predictions["calculated_temps"] = out["subarray1_celltemp"]

#Look at results
measured_data_stuck = measured_data.loc[datetime.date(2023, 9, 16):datetime.date(2023,9,17)]
measured_data_kW = measured_data_stuck["inverter_1_ac_power_(kw)_inv_150143"] #+ measured_data_stuck["inverter_2_ac_power_(kw)_inv_150144"]
#measured_data_kW = measured_data_stuck["inverter_2_ac_power_(kw)_inv_150144"]
tracker_predictions = tracker_predictions.set_index(measured_data_stuck.index.copy())
tracker_predictions["Stuck_tracking"].plot()
measured_data_kW.plot()
plt.legend(["Non-moving tracker", "Measured"])
plt.ylabel("Power (kW)")

Text(0, 0.5, 'Power (kW)')

tracker_predictions["sun_angles"].plot()
tracker_predictions["custom_angles"].plot()
stuck_tracker_data.plot()
plt.legend(["Sun position angles", "Custom angles", "Measured tracker data"])
plt.ylabel("Tracker rotation angle (degrees)")

Text(0, 0.5, 'Tracker rotation angle (degrees)')

Custom module temperatures

• Replace cell temperature in calculation with user-defined cell temperatures

• Useful for using cell temperature models not in SAM, using measured temperatures

• Currently SDK only, evaluating for future integration into desktop tool

pv_model.SystemDesign.subarray1_use_custom_cell_temp = 1 #Turn on custom cell temperatures
pv_model.SystemDesign.subarray1_custom_cell_temp_array = amb_weather_data_example["thermocouple_pad_1_back-of-module_temperature_1_(c)_o_149728"]
pv_model.execute()
out = pv_model.Outputs.export()
tracker_predictions["custom_temps"] = out["subarray1_celltemp"]
tracker_predictions["calculated_temps"].plot()
tracker_predictions["custom_temps"].plot()
plt.legend(["Temperature Model", "Custom Temperature"])
plt.ylabel("Degrees Celsius")

Text(0, 0.5, 'Degrees Celsius')