[PSH] Update PSH Supply Curves for Site-Level Processing

[jvcarag]

Summary

This PR updates the pumped-storage hydropower (PSH) supply curve input files with the latest (5th-generation) PSH resource and cost data as described at https://www.nlr.gov/gis/psh-supply-curves.

PSH input files are now stored at site-level spatial resolution, and input processing procedures are modified to aggregate the site-level data up to the spatial resolution of each corresponding ReEDS run. In effect, PSH supply curves are now processed using the same procedures as those used for other RSC technologies such as upv, wind-ons, and wind-ofs. The old PSH supply curves (pre-binned at the old 134-zone model resolution) are removed due to their incompatibility with the new input processing procedures which require the data at site-level resolution.

The new PSH supply curve input files also reflect new site exclusion scenarios labeled "Reference", "Open", and "Limited" that are more consistent with wind and solar resource availability scenarios. "Open" is most similar to earlier PSH scenarios including existing reservoir sites (wExist) and sites overlapping with ephemeral streams (wEph). "Limited" is a new case with more stringent PSH siting restrictions from the "Reference" case. Below is a comparison of the new binned supply curves for 8-hr storage duration PSH vs the old binned supply curves - the old default 12hr_ref scenario is also included for comparison between old and new defaults:

Technical details

Site-Level PSH Supply Curve Input Files:
PSH supply curve capacity and capital cost input files are now stored in a single, site-level spatial resolution input file (previously separated and at z132-resolution), using the latitude/longitude coordinates of the lower reservoir as the location of the powerhouse and its corresponding generating capacity potential. The coordinates of each site's paired upper reservoir are also provided in the input files, available only for the purpose of post-ReEDS mapping. Interconnection and transmission costs are appended to the PSH supply curve using each site's lower reservoir latitude/longitude coordinates to map to the nearest (Euclidean distance) reV site in inputs/supply_curve/interconnection_land.h5. Mapping with latitude/longitude coordinates instead of sc_point_gid mapping ensures PSH supply curves stay compatible with any future updates to the reV sites in the model.

PSH Supply Curve Scenarios:
To align more closely with siting scenarios for other resources (solar and wind), PSH site exclusion scenarios are now consolidated into 3 distinct scenarios: limited, reference, and open. Below are the exclusion layers included in each of the 3 scenarios:

• open: urbanized/developed areas, existing waterbodies, 100yr flood plains, various federal, state, and local protected areas and conservations, glaciers, snow/ice, and wetland ecosystems, and airports/airpads/helipads
• reference: all "open" exclusions [plus] ephemeral streams, critical habitats for endangered species, railways/pipelines/roads (incl. 30m buffer), and national wetlands
• limited: all "reference" exclusions [plus] agricultural lands, mature forests on federal lands, building footprints, and tribal areas (reservation and off reservation trust land)
  This change also simplifies the combinations of PSH supply curves and reduces file count in inputs/supply_curve,

Below are the new unbinned open, reference, and limited PSH supply curves for 8-, 10-, and 12-hr PSH storage durations:

Implementation notes

Input Files:

• Add new site-level PSH supply curve files and remove old vintages of BA-level PSH supply curve files
• Update inputs/storage/PSH_supply_curves_durations.csv with entries for new PSH supply curve files
• cases.csv: update options for pshsupplycurve switch, add numbins_psh switch for controlling PSH supply curve bin number, and set 8-hr Reference case as default for pshsupplycurve
• Update inputs/supply_curve/dollaryear.csv with entry for new PSH supply curve files
  Input Processing
• runfiles.csv: remove old psh_supply_curves_{capacity|cost}.csv entries and add new supplycurve_psh.csv entry
• writesupplycurves.py: update procedures in agg_supplycurve() function for assembling/aggregating supply curves for use with site-level PSH supply curves.
  • New procedure clips pre-binned supply curve values at $4000/kW [2004$] to improve binning resolution at the low-cost end
• io.py: add nearest-point mapping of interconnection costs to PSH supply curve to assemble_supplycurve() function

Additional changes

• b_inputs.gms: remove unit adjustment for pumped-hydro supply curve costs, as new input files are already stored at $/MW
• recf.csv: remove deprecated copy argument from pd.concat() calls
• reeds2.py: add new ReEDS result and associated preset function to calculate and display cumulative capacity additions by technology and model year

Switches added/removed/changed

• pshsupplycurve: updated Choices to reflect new exclusion scenarios
• numbins_psh: added new switch to control number of PSH supply curve bins
• numbins_{upv|windons}: updated Description to reflect current default setting

Issues resolved

Issue #174

Known incompatibilities

Note that while available generating capacity for PSH in ReEDS does account for frictional head losses in the conveyance system, the round-trip efficiency for PSH still uses a generic value that does not account for these losses, which are site-specific. Future work will explore ways to better account for efficiency reductions from frictional losses, either by incorporating site-level efficiencies into supply curve data or calculating a reasonable standard efficiency adjustment.

Relevant sources or documentation

Validation, testing, and comparison report(s)

Old vs New Default PSH Supply Curve Comparison

A comparison of the USA_Defaults case using the old 12-hr Reference PSH supply curve on the main branch vs this PR branch which uses the new default 8-hr Reference PSH supply curve was conducted. Below are the capacity, generation, and system cost results comparison:

comparisons_default_mainvsbranch.zip

The difference between these runs stem from a 68 MW reduction of PSH deployment by 2050 (23.041 -> 22.972 GW) resulting from the change to the less resource-available, lower-duration 8-hr Reference PSH supply curve.

High PSH Deployment Scenario

The new reference case only deploys an additional 3.5 GW PSH from 2025-2050 for a national total of ~23GW PSH storage capacity by 2050. To better understand the effects of the new supply curve and updated exclusion layer scenarios on PSH deployment, provided below are the PSH capacity deployment results [GW] of a "high PSH deployment" ReEDS scenario:
New PSH Capacity deployed from 2025-2050 [GW]:


New PSH Capacity deployed from 2025-2050 [GW]:

This scenario assumes favorable cost and performance metrics for solar and unfavorable metrics for li-ion battery technologies, as well as a linear reduction of PSH capital cost to 50% by 2050 to simulate a case that encourages high deployment of PSH capacity across CONUS. The 8-hr reference case deploys up to 83 GW PSH by 2050, with 47 GW deploying in CA, 10 GW in AZ, and 7 GW in NV.

Checklist for author

Details to double-check

• Charge code provided to reviewers
• Included comparison reports for appropriate test cases
• Documentation updated if necessary
• If input data added/modified:
  • Dollar year recorded and converted to 2004$ for GAMS
  • Units are specified
  • Preprocessing steps have been documented and committed in ReEDS_Input_Processing PR21
• Code formatting standardized
• Reusable functions used where possible instead of copy/pasted code

General information to guide review

• Zero impact on results of default case
• No large data file(s) added/modified
• No substantive impact on runtime for full-US reference case
• No substantive impact on folder size for full-US reference case
• No change to process flow (runreeds.py, reeds/core/solve/solve.py)
• No change to code organization
• No change to package requirements (environment.yml or Project.toml)

Tag points of contact here if you would like additional review of the relevant parts of the model

[jvcarag]

@mmowers I mentioned before that I may have inadvertently added procedure to hourlize/resource.py that looked to be missing (based on this comment mentioning "region-specific" class bins). Let me know if you think this is a welcome adjustment to the script, or if you think this change is not necessary!

[stuartcohen8]

It'd be good to have someone else take a look at the input processing scripts in case there are any code improvements there. Otherwise just some minor comments to help improve clarity.

[stuartcohen8]

Per our conversation, it could also be good to include a before/after results comparison for a scenario setup with at least a few GW of new PSH deployment.