document or support modules_per_string strings_per_inverter with pvwatts in modelchain

[Jawairia123]

Hi,

I am trying to run Modelchain with pvwatt model but it seems that the modules_per_string and strings_per inverter doesn't have any affect on the total output.

I am not sure why is it so. May be ModelChain isn't supporting so. If that's the case how can I achieve the desired result?

Here is my code:

Thanks in advance

# built-in python modules
import os
import inspect

# scientific python add-ons
import numpy as np
import pandas as pd

# plotting stuff
# first line makes the plots appear in the notebook
%matplotlib inline 
import matplotlib.pyplot as plt
import matplotlib as mpl
# seaborn makes your plots look better
try:
    import seaborn as sns
    sns.set(rc={"figure.figsize": (12, 6)})
    sns.set_color_codes()
except ImportError:
    print('We suggest you install seaborn using conda or pip and rerun this cell')

# finally, we import the pvlib library
import pvlib

tmy = pd.read_csv("http://re.jrc.ec.europa.eu/pvgis5/tmy.php?lat=29.74&lon=40.10")

tmy.Date = pd.to_datetime(tmy.Date, format='%Y-%d-%m %H:%M:%S')

tmy.rename(columns={' Ghor':'ghi','Dhor':'dhi','DNI':'dni','Tair':'temp_air',
                        'Ws':'wind_speed'},inplace=True)                    

tmy.set_index(tmy['Date'],inplace=True)
    #Drop unnecessary column

tmy = tmy.drop('Date', 1)
tmy = tmy.drop('RH', 1)
tmy = tmy.drop('IR', 1)
tmy = tmy.drop(' Wd', 1)
tmy = tmy.drop('Pres', 1)

#module =Jinko_Solar_JKM320P_72_V
#inverter = ABB__PVS980_58_2000kVA_K__N_A_V__CEC_2018_

lat = 29.74
lon = 40.10
altitude = 676
tz = 'Etc/GMT+3'  

loc = pvlib.location.Location(latitude=lat,longitude= lon,tz=tz)

#model = pvwatts 
pvwatts_system = pvlib.pvsystem.PVSystem(module_parameters={'pdc0': 320, 'gamma_pdc': -0.0041},inverter_parameters={'pdc' : 3200000, 'pdc0' : 2024292, 'eta_inv_nom':0.988, 'eta_inv_ref':0.986},surface_tilt = 20, surface_azimuth=0,
                    modules_per_string=30,strings_per_inverter=267, albedo = 0.2)

mc = pvlib.modelchain.ModelChain(pvwatts_system, loc, transposition_model ="perez",aoi_model = 'ashrae',spectral_model='no_loss')
print(mc)
mc.run_model(times=tmy.index,weather=tmy)
a = mc.ac
a = pd.Series.to_frame(a)
a = a * 530  # 530 = number of inverters in the system 

a['month'] = a.index
a.month = a.month.dt.month
monthly = a.groupby('month').sum()

[wholmgren]

https://stackoverflow.com/questions/49550656/run-pvlib-modelchain-with-pvwatts-model/50165303#50165303

The work around is to scale your module_parameters ['pdc0']. Pull requests welcome for improving the functionality and/or documentation.

[markcampanelli]

It seems that the system scaling provided by PVSystem.scale_voltage_current_power() is a system-level entity that should be included in PVSystem.sapm() and PVSystem.singlediode() computations, in addition to adding this to PVSystem.pvwatts_dc(). Currently, a higher level ModelChain function does this (except for pvwatts, as discussed above). If folks agree to this, then a question arises as to if the corresponding wrapped functions in pvsystem.py should still only calculate singlediode() for a single module/device instead of the whole system. (ATM, I think that they should.)

[wholmgren]

@cwhanse we need this for SPI. Do you have any concern with adding this

        self.results.dc = self.system.scale_voltage_current_power(
            self.results.dc,
            unwrap=False
        )

to

https://github.com/pvlib/pvlib-python/blob/56971c614e7faea3c24013445f1bf6ffe9943305/pvlib/modelchain.py#L732-L735

?

Or do you think we should go ahead with @markcampanelli's suggestion above? I think @markcampanelli's suggestion is better on the merits but it's a much bigger change and I don't know how to do it in a way that wouldn't cause user code to return unexpected answers.

[cwhanse]

I don't have a problem patching that into pvlib.modelchain.ModelChain.pvwatts_dc. I think it's an oversight that the scaling was left out, since it is included in the sapm and singlediode methods.

[wholmgren]

I think we left it out because it's arguably a departure from the pvwatts model in which you're typically specifying the pdc0 of the entire system. But I don't see a problem with the extension within our data model.

[cwhanse]

want me to open a PR? Or have you got it?

[wholmgren]

Would be great if you can do it.