Multi-Objective Transmission Expansion: An Offshore Wind Power Integration Case Study (MOTEP-OSW)

This repository contains the source code used to implement the model described in our paper, "Multi-Objective Transmission Expansion: An Offshore Wind Power Integration Case Study." The paper has been accepted for publication in a future issue of the IEEE Transactions on Energy Markets, Policy and Regulation, and is available as an early access article on IEEE Xplore. A preprint version of the paper is also available on arXiv.

Citing

If you benefit from the theoretical frameworks, methodologies, and findings presented in this repository or the paper, please use the following citation in your work:

@article{khanal2024multi,
  title={Multi-Objective Transmission Expansion: An Offshore Wind Power Integration Case Study},
  author={Khanal, Saroj and Graf, Christoph and Liang, Zhirui and Dvorkin, Yury and {\"U}nel, Bur{\c{c}}in},
  journal={IEEE Transactions on Energy Markets, Policy and Regulation},
  year={2024},
  publisher={IEEE}
}

Test Systems

8-Zone Test System Based on ISO-NE Data ("Nodal", DC-OPF Model)

• Network topology based on Krishnamurthy et. al, 2015
• Can be utilized for any so-called nodal model
• Example configurations:
  • ISONE8busTS_N5X5/: Indicates 5 normal and 5 extreme days.
  • ISONE8busTS_osw1_17_N5X5/: Fixed POI (Point of Interconnection) specification with one degree of freedom (potential landfall nodes/zones of offshore power) and 17 candidate lines.
  • ISONE8busTS_osw6_51_N5X5/: Optimized POI specification with six degrees of freedom and 51 candidate lines.

The new operational scenarios (corresponding specifications denoted by _NS) better capture the effects of variable renewables and energy storage than those based on net load. The scenarios were generated using clustering methods from Scott et al. (2019), which used individual time series of wind, solar, and load, resulting in more temporally coupled samples.

9-Zone Test System Based on PJM Data ("Zonal", Pipes-and-Bubbles or Transportation Model):

• Network topology based on the U.S. EPA's Integrated Planning Model
• Can be utilized for any so-called zonal model
• Configuration adaptations include:
  • EPA_PJM9Zones_N5X5R/: With retirement data.
  • EPA_PJM9Zones_osw2_17_N5X5R/ and EPA_PJM9Zones_osw3_52_N5X5R/: Different POI specifications with specified numbers of candidate lines.
• _NS suffix denotes specifications incorporating the new scenarios.

Running the Simulations

• Begin by configuring the cases.csv, which sets the parameters and paths for input files. Note that the specification names used for the output directories are derived (starting) from the 5th column header in cases.csv. For serial execution of multiple model specifications, start them from the 5th column. For example, two headers, SO and MO in the 5th and 6th columns, can be used to run the two specifications serially in a single run. For each columns starting from 5th, update only parameter values that differs from defaults as defaults will apply from the Default Value column. See cases_isone.csv and cases_pjm.csv for references.
• Run TEP.jl for simulations.
• For PJM Case Study, update the model.jl and output.jl to model_pjm.jl and output_pjm.jl respectively in TEP.jl file. E.g., include(joinpath(@__DIR__, "src", "model.jl")) should look like include(joinpath(@__DIR__, "src", "model_pjm.jl")).

Inputs

Several input files from the inputs/ directory are required for the model to run successfully. Each file is described in detail below, focusing on the ISO-NE Test System. The PJM Test System data is also collected and curated using a methodology similar to that used for ISO-NE, as described in the paper. raw data/ and raw data epa/ contains the raw, intermediate and final files for ISO-NE and PJM Test Systems respectively.

Network Topology: Nodes/Zones, Lines, and Generators

• mappings_assumptions.csv

  • Details on per unit (p.u.) conversion bases and node/zone numbering.
  • Important for ensuring that nodal/zonal data are properly ordered as per the specifications in individual files.

• lines.csv

  • Provides transmission line data from Krishnamurthy et. al, 2015, complemented by line limits from https://github.com/jipkim/Trilevel

• generators_loc_emission_damages.csv

  • Enhanced generators.csv, which is the base generator data, with average marginal emissions values computed using InMAP.

Operations: Load and Renewable Generation Time Series

• weights_of_scenarios.csv

  • Weights of representative days (operational scenarios), used in conjunction with time series data.
  • Derived from clustering algorithms to prioritize certain days based on similarity and significance. See inputs/raw data*/Wind Scenarios*/ for details.

• load.csv

  • Hourly load data sourced from the ISO-NE website.
  • Source: ISO-NE Energy, Load, and Demand Reports.

• solar.csv and wind.csv

  • Actual hourly generation data for solar and wind respectively, for representative days (computed from clustering algorithms, see Wind Scenaarios for ISO-NE, as an example) of the year 2022.
  • Columns: Nodes/zones in order, consistent with nodes.csv.
  • Rows: Stacked by representative days, 24 hours at a time.
  • Source: Daily Generation by Fuel Type, ISO-NE.

• solar_normalized.csv and wind_normalized.csv

  • Timeseries data normalized by installed capacity for solar and wind respectively, to standardize the scale of generation across different zones and conditions.

Notes: To run the model with data sets other than those found in inputs/, ensure that all files are stored in the designated directory and are properly structured to avoid errors. To run the model with the pre-existing dataset, simply run TEP.jl and edit cases.csv as desired.

Outputs

The results of the simulation run will appear in the outputs/ directory, which is organized in line with the specification names listed in cases.csv. Each outputs/ subdirectory contains a number of files that are arranged to facilitate the extraction and analysis of the data. Below is a list of the primary output files along with their structures.

Configuration

• case*.csv
  • Contains the run configuration for each specification.

Investment Decisions

• new_build_gen_cap_*_[power/energy].csv

  • Details new buildout capacities in terms of power or energy (for batteries).
  • Rows: Nodes/zones in order.
  • Columns: Epochs in order, e.g., 1 through 4.

• new_line_decision.csv

  • New line investment decisions coded as binary (0 for no investment, 1 for investment).
  • Rows: Lines in order, consistent with lines.csv.
  • Columns: Categorized by cable types for each epoch in order.

• line_upgrade_decision.csv

  • Decisions on existing line upgrades.
  • Rows: Lines in order, matching lines.csv.
  • Columns: Epochs in order.

• line_upgrade_decision_size.csv

  • Size of line upgrade decisions (used only in pipes-and-bubbles model, in this study the PJM test system).

Investment Costs

• annual_cost_*_investment.csv
  • Discounted annual costs related to various categories such as line investments and externalities.
  • Rows: Sequential epochs, e.g., 1 through 4.

Operations

• *load*.csv and related files (*line_flow, theta.csv, wind_spillage.csv, *_out, [ch/dis]_node.csv)

  • Timeseries data for loads, line flows, bus voltage angles, renewable curtailment, power output, and charging/discharging at nodes.
  • Rows: Stacked by nodes/zones, then by representative days.
  • Columns: Stacked by 24-hour periods for each epoch in sequence.

• Emissions*.csv

  • Total annual emissions for various pollutants like CO2 and NOx.
  • Rows: Epochs in order, from 1 through 4.

Post-Processing After Run

The following post-processing scripts help to parse output files and prepare them for visualization and analysis of simulation results. Please see the complete instructions below for particular post-processing steps:

Processing Costs and Capacities

• Script: src/utils/process_costs_caps.csv
• This script processes output files to extract capacity additions and cost outcomes.
• Setup:
  • For the ISO-NE Test System, set n_lines_existing to 12, and for the PJM, set it to 13.
  • Set n_buses_existing to 8 for the ISO-NE and 9 for the PJM.
  • Specify the result directory path: result_dir = "/Users/skhanal/Globus_local/MOTEP-OSW/outputs/paper_draft/"
  • Use a base data directory for comparison across different runs: data_dir = "/Users/skhanal/Globus_local/MOTEP-OSW/inputs/ISONE8busTS_osw1_17_N5X5/"
  • Specify the base case (specification) directory name to compare different specifications, e.g., base_case_dirname = "SO"

Processing Topologies

• Script: process_topologies.py

• This script is used for processing and visualizing transmission buildouts.

• Setup:

  • Ensure you have the necessary Python packages installed. Refer to the pyproject.toml and poetry.lock files for setup guidance.
  • Update the result_dir and data_dir as required.
  • For PJM specific processing, rename scripts to use _pjm.py suffix (e.g., replace process_transmissions.py with process_transmissions_pjm.py and plots.py with plots_pjm.py).

  Notes: The IPM v6 Regions shapefiles can be downloaded from the U.S. EPA website and should be kept in the inputs/ directory. Despite being necessary for executing plots_pjm.py, this file is excluded because of file size restrictions.

Processing Generation Dispatch

• Script: process_gens.py
• This script processes generation dispatch.
• Setup:
  • Update the result_dir.
  • For PJM runs, set GSw_PJM = True in the process_generations.py script. You can use the same script the compute air quality damage details by simply changing csv_path = os.path.join(subdir, "Pg_reduced.csv") to csv_path = os.path.join(subdir, "Pg_aq.csv"), and the generations.csv then will have combined Pg_aq.csv from all specifications. You can also rename generations.csv to something like Pg_aq_all.csv avoid confusions in final_df.to_csv(os.path.join(result_dir, "generations.csv"), index=False)

Computing Differences

• Script: process_difference.py
• This script is used to compute the differences between identical files from two specifications.
• Setup:
  • Set the base directory: dir_base = "/Users/skhanal/Globus_local/MOTEP-OSW/outputs/ISONE/MO"
  • Set the comparison directory: dir_compare = "/Users/skhanal/Globus_local/MOTEP-OSW/outputs/ISONE/MO OPOI"
  • Results will be stored in a directory named MO OPOI-MO as CSV files with the differences, which can be used for further analysis such as plotting air quality impacts using the plot_airquality_damage_costs.ipynb Jupyter notebook.

Requirements

See Project.toml for the required packages.