Multi-year DUPV profiles

[gschivley]

I'm trying to compile the new multiple weather year data from ReEDS and am looking for some help to understand the distributed PV capacity and profiles. I've previously taken state-level capacity and generation profiles directly from the Cambium results, which is based on the 2012 weather year.

ReEDS now seems to have a file (dupv_ba.h5 located at /inputs/variability/multi_year with 7 weather years of capacity factors by BA and resource class. I've also found the files DUPV_supply_curves_capacity_2018.csv and DUPV_supply_curves_capacity_NARIS.csv, which seems to have available capacity by resource class within each BA. Finally, I see that the folder dGen_Model_Inputs has various distPVcap_* files, the newest of which is Standard Scenarios 2022.

I could use distributed PV capacity from the files in ReEDS or directly from the 2023 Standard Scenarios results, but both give total capacity per BA and the generation capacity factors are by resource class within BA. I haven't found anything in the ReEDS code that deals with this. Should I fill up resources class capacity starting with the lowest number and create a weighted average capacity factor profile? Or is there another set of data that I should look at?

[wesleyjcole]

ReEDS has two distributed PV capacity inputs: DUPV and distPV. DUPV is distributed utility PV in urban region, and distPV is behind-the-meter rooftop PV. We disabled DUPV a while ago (maybe about a year ago?) because we merged the DUPV with the UPV, so now all utility PV is captured in the UPV supply curves and generation profiles. The distPV hourly profiles come from dGen, which only has a single year of weather data. Therefore, to approximate distPV profiles in ReEDS, we use the UPV profiles, but scale them to match the annual capacity factors produced by dGen. See https://github.com/NREL/ReEDS-2.0/blob/main/input_processing/LDC_prep.py#L156 for where that happens in ReEDS.

[gschivley]

Wes, can you confirm that the distPVCF_hourly*.csv files are all in local time? Looks like I can use reeds_region_tz_map.csv to shift them to a reference timezone.

While I'm asking, are all time series data stored in local time? Looks like the new EER demand (EER_IRAmoderate_load_hourly.h5) are probably also local.

[wesleyjcole]

Yes, these are in local time. We convert to eastern time in the input processing scripts when a run gets kicked off. At some point we'll change the inputs to all be in the same time zone rather than local time, so just a heads up that things could change down the road.

[gschivley]

Are the UPV profiles that are used as the basis of distPV aggregated within BA or for individual sites?

[wesleyjcole]

We use the worst UPV profile within a BA for the distPV profiles. See

ReEDS-2.0/input_processing/LDC_prep.py

Line 156 in 9b159a1

def get_distpv_profiles(inputs_case, recf, rb2fips, agglevel):

.

[gschivley]

Thanks for the quick answer, and good to know about the DUPV/distPV difference. Additional data files have been added to ReEDS over time and it can be difficult to determine which are current vs legacy.

The code in get_distpv_profiles doesn't seem to account for differences in distPV profiles by year. Is that correct? I've noticed that Cambium 2022 has different capacity factors for each model year. They generally drop over time, which I assumed was due to installing panels on better roof orientations in early years. But I just looked at a selection of normalized profiles for Iowa and am not sure if that's what is happening. Is distributed PV being curtailed in the Cambium results?

[wesleyjcole]

Some pieces of this should be from degradation (e.g., the last plot you posted looks like it is from that). The other outputs from Cambium are likely from curtailment. We recently changed ReEDS to no longer allow rooftop PV to be curtailed, but I don't know if that has been implemented in PLEXOS for running the Cambium outputs.