PGE Distributed Energy Resource and Flexible Load Potential Study
Portland General Electric (PGE) sought an open, transparent tool that would quantify the impacts of electrification and distributed energy resource adoption across its service territory. This tool would address two specific needs for PGE: distributed energy resource (DER) supply curves for their integrated resource plan and a locational load forecast for their distribution system planning activities.
The Cadeo team (project partners: Brattle Group, Lighthouse Consulting, and Resilient Edge) developed a tool that allowed PGE to meet these objectives. We estimated the amount of technical, economic, and achievable potential for more than 50 DER technologies and programmatic measures in PGE’s service territory. These measures included behind-the-meter solar photovoltaics, behind-the-meter energy storage, demand response, direct load control, transportation electrification, building electrification, and more.
Phase 1: DER and Flexible Load Potential
First, we conducted a comprehensive measure characterization task that defined eligibility criteria, adoption curves, costs, lifetimes, and hourly load impacts for each measure. In order to collect this information, we employed a wide variety of tactics, including:
- Analyzing PGE customer data
- Reviewing evaluation reports and program filings from PGE and other utilities
- Running building simulation models
- Acquiring data from public APIs
- Employing web-scraping techniques
To estimate the technical, economic, and achievable potential over the study’s 30-year horizon (2021 through 2050), we developed and implemented our AdopDER model. AdopDER simulates the adoption and load impacts of each measure at each site (agent) to estimate DER potential for the service territory. We calibrated AdopDER’s results against system-level data. This ensured that the potentials were internally consistent with PGE load forecasts and program planning estimates. We also conducted a scenario analysis that estimated adoption and load impacts under various scenarios (reference, high, and low cases).
Phase 2: Locational Load Forecasting
Building on Phase 1, we added locational load forecasting capabilities to AdopDER in Phase 2. We used machine learning techniques to estimate site-level adoption probabilities for specific measures (electric vehicles and solar photovoltaics). We then used CalTRACK methods to create load profiles for each feeder and large customer. Finally, we used location-specific assumptions in AdopDER to generate a feeder-level load forecast, with and without projected DER adoption.
Cadeo’s feeder-level forecast has become a critical component of PGE’s distribution system planning processes. It characterizes load growth at specific areas on PGE’s grid, informs non-wires solution investments, and provides hosting capacity analysis. Thus, AdopDER gives PGE a powerful tool for exploring varying scenarios of DER adoption and how they propagate over time at each level of the distribution system.