Correlated uncertainty propagation enables multi-impact decision support for electrical system decarbonization

Amir M. Gazar^1,2, Chloe Jackson³, Georgia Mavrommati³, Rich B. Howarth⁴, Ryan S.D. Calder^1,2,5,*

¹Department of Population Health Sciences, Virginia Tech, Blacksburg, VA, 24061, USA
²Global Change Center, Virginia Tech, Blacksburg, VA, 24061, USA
³School for the Environment, University of Massachusetts Boston, Boston, MA, 02125, USA
⁴Environmental Program, Dartmouth College, Hanover, NH, 03755, USA
⁵Department of Civil and Environmental Engineering, Duke University, Durham, NC, 27708, USA
* Contact: rsdc@vt.edu

Abstract

Decarbonization planning requires comparing diverse pathways across economic, ecological, and health dimensions under uncertainty. Capacity expansion models generally treat pathway uncertainties as independent, overestimating uncertainty around inter-scenario differences, which drive decisions. U.S.–Canada trade tensions and abrupt federal termination of offshore wind permits threaten key planks of regional decarbonization plans and illustrate the need for models spanning a wider pathway space. We present PHASED (Probabilistic Hourly Assessment of Scenarios for Electrical Decarbonization), propagating correlated uncertainties across prescribed pathways through hourly dispatch over a 26-year horizon and generating joint posterior distributions across modeled outcomes. Applied to eight New England pathways, correlated uncertainty tracking yields >90% confidence in pairwise cost differences despite overlapping absolute cost intervals. Pathways with similar monetized impacts (roughly $470–477 billion by 2050) diverge on land use, avian mortality, and air quality. Rural areas receive greater relative air quality benefits than urban areas, cutting against assumptions that shape siting politics.

Keywords: integrated assessment model; discount rate; capacity expansion model; decarbonization; renewable energy; energy policy; cost-benefit analysis

Reproduction Information Document

A comprehensive step-by-step guide to reproduce every analysis in this repository:

• Section 1: Configuration & setup, hardware (macOS 14.5 Sonoma, ARC cluster), R 4.4.2 & RStudio 2024.09.1, package versions, install & run times.
• Section 2: Conceptual overview of the modeling framework and code availability.
• Section 3: Decarbonization pathways data processing (Excel → R, metadata tagging, year‐range extraction).
• Section 4: Generation expansion model scripts, hourly wind/solar CFs, SMR specs, fossil facility & emissions processing, new fossil additions, imports, demand processing, randomization, dispatch‐curve generation.
• Section 5: Dispatch‐curve results processing.
• Section 6: Total cost modules; CAPEX, FOM, VOM for fossil, non-fossil & imports; fuel & import cost adjustments; GHG & air pollutant emissions cost interpolation & NPV; unmet demand penalties; hydropower cost assumptions & capacity modeling; consolidation of all costs.
• Section 7: Ecological impact metrics; land use, water withdrawals, avian mortality, viewshed.

View the full PDF here

Fossil‐Fuel Power Plants Data Portal

An interactive web portal offering probabilistic hourly generation and emissions data (incl. historical hourly data for the past 20 years) for all of United States fossil‐fuel power plants. Including:

• State Power Plants Data
  Download probabilistic hourly generation and emissions data for all fossil‐fuel plants in a selected state.
• Individual Power Plant Report and Data
  Access pre-generated PDF reports and CSV files showing hourly outputs and emissions for a single facility.
• Templates for New Power Plants
  Get probablistic templates (from similar facilities) to plug in new plant specifications and run them through the probabilistic models.
• Historical Generation and Emissions (U.S. EPA CAMPD)
  Retrieve historical U.S. EPA CAMPD data stored in a Harvard dataverse.
• API Bulk Download
  Retrieve all data available in this portal using Harvard Dataverse API token.
• Citation
  Instructions on how to cite this portal and its underlying datasets in your publications.
• Contact Us
  Support contact details for questions, feedback, or technical issues.

🔗 Or visit this link: https://amirgazar.github.io/powerplants/

Copyrights and Citation

Gazar, A. M., Jackson, C., Mavrommati, G., Howarth, R. B., & Calder, R. (2025). Correlated uncertainty propagation enables multi-impact decision support for electrical system decarbonization, Engineering Archive. https://doi.org/10.31224/4684

© 2025 Amir M. Gazar et al. All rights reserved. This work is licensed under a Creative Commons Attribution 4.0 International License .