DRP DER Growth Scenarios Workshop DER Forecasts for Distribution Planning- Electric Vehicles May 3, 2017
Presentation Outline Each IOU: 1. System Level (Service Area) Forecast 2. Disaggregation Approach Joint IOUs: Challenges 1
Overview of SCE s System Level Assumptions 2
Comparison of SCE 2017 IEPR Submittal with 2017 ACR (2016 IEPR Update) Light Duty * The CPUC ACR document containing planning assumptions for 2017 is based on the 2016 IEPR Update. 2016 IEPR Update mid case scenario assumes the EV prices and consumer preference will change enough over the forecast period that the ARB s ZEV Mandate will be met SCE s 2017 IEPR submittal utilizes the Navigant s forecast which applied discrete choice model and consumer surveys to account for customers preferences and adoption behavior, manufacturers vehicle availability, battery costs, and infrastructure quality and costs etc. *2017 CPUC ACR containing Assumptions and Framework for use in the 2017-18 TPP 3
SCE s Planned approach for 2017/18 DSP Internal system level forecast will utilize Bass Diffusion, discrete choice, multivariate regression models and/or TCO (Total Cost of Ownership) analysis Estimate maximum potential for EV adoption by considering the key attributes affecting EV adoption such as income, education and travel time to work, etc. Utilize the SCE s latest historical EV adoption and public sources such as U.S. Census Data Account for latest policy changes such CARB s scoping plan Calibration with historical adoption and benchmark against vendor forecasts Update EV charging profiles 4
Comparison of SCE 2017 Forecast with 2017 ACR (2016 IEPR Update) - Non Light Duty Non Light-Duty* (GWh) ACR** SCE 2017 Forecast 664 764 564 465 367 68 43 137 103 64 86 269 172 108 125 142 160 178 199 218 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 The CPUC ACR document containing planning assumptions for 2017 is based on the 2016 IEPR Update. 2016 IEPR Update forecast developed by UC Davis Institute of Transportation and Aspen Environmental Group and the method used in this study is similar to TEA study SCE used Phase 1 of ICF International and Energy+ Environmental Economics s Transportation Electrification Assessment report (TEA Study) in-between forecast. *This Forecast includes forklifts, Truck Stop Electrification, Transport Refrigeration Units, Port Cargo Handling Equipment, Airport GSE, Medium and Heavy duty vehicles. ** 2017 CPUC ACR containing Assumptions and Framework for use in the 2017-18 TPP 5
Overview of SCE s 2017-18 Disaggregation Methodology 6
Disaggregation Method - Light Duty Step 1: Map existing Electric Vehicles to the circuit level Customers on EV specific rates (38%) Rely on SCE zip code level EV data provided by EPRI Step 2: Gather most recent data for key attributes affecting EV adoption Sources include U.S. Census Data and other public surveys Existing EV Adoption in SCE (# of Vehicles) 1-75 75-200 200-350 350-500 500-1250 Center for Sustainable Energy (2016). CVRP Infographic: What Drives California s Plug-in Electric Vehicle Owners? Retrieved 05/01/2017 from https://cleanvehiclerebate.org/eng/programreports 7
Disaggregation Method - Light Duty (Continued) Step 3: Clustering approach will be used to sort and score the circuits based on the existing EVs as well as factors indicating potential EV adoption Income Education # of each household vehicles Travel time to work # of existing EVs for each circuit # of existing DC Fast chargers for each circuit Step 4: Determine growth rate for each cluster relative to its potential Step 5: Allocate system level EV forecast to each circuit based on the growth rate determined for each cluster in Step 4 8
Disaggregation Method - Non Light Duty SCE will consider some major categories for non light-duty load growth Forklifts Truck Stop Electrification Transport Refrigeration Units Port Cargo Handling Equipment Airport Ground Support Equipment Medium and Heavy Duty Vehicles Non Light-Duty will be disaggregated based on: known project locations non-residential customer concentration for each circuit 9
Overview of SDG&E s System-Wide Forecast Assumptions 10
System-Wide Forecast Assumptions Align Distribution Forecast to 2017 IEPR Submittal (avail. before 2017-18 DPP cycle begins) IEPR submittal will be supplemented with Electric Vehicle Survey, EVTOU Rate adoption and RL Polk (DMV) datasets 11
Overview of SDG&E s Disaggregation Methodology 12
EV Disaggregation Methodology Behind The Meter Utilize EV-TOU Rate adoption and Electric Vehicle Customer Survey to provide historical adoption data Compare EV rate customers and survey results to Polk (DMV) total ZEV count in service territory to obtain estimate. Charging Stations Multi-family and employment center charging stations will be applied to specific feeders based on using like load profiles and expected in-service dates. Include C&I Charging Programs at major usage centers 13
# EVs in survey # EVs in Survey EV Disaggregation Methodology Proportional Feeder Forecast Based on Past Adoption 80 60 40 20 0 Sample Zip Code A Sample Zip Code B 30 20 10 0 2011 2012 2013 2014 2015 2016 2011 2012 2013 2014 2015 2016 Year Year Note: linear adoption rate observed for available history. 50 40 Zip code level data to be mapped to feeder level distribution model. Historical Adoption Data EV Adoption by Zip Code Distribution Model EV Adoption By Feeder 14
EV Disaggregation Methodology Proportional Feeder Forecast Based on Past Adoption Estimated adoption at the feeder level to be spread proportionally throughout system based on past adoption rates. Historical Feeder EV Forecast Feeder EV System Wide EV System Wide EV Forecast Linear forecast to be used at feeder level for 2017-18 DPP to align with observed zip code level adoption rates. S-Curve type forecast may be possible with more adoption history. EV Load curves will be applied using the LoadSEER software. 15
Overview of PG&E s System Level Assumptions 16
000s of EVs GWh PG&E s Service Area EV Forecast EV Adoption (000s of EVs) EV Sales (GWh) 700 2,500 600 2,000 500 400 1,500 300 1,000 200 100 500 0 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 0 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 Developed 4 scenarios for EV adoption in California based on varying degrees of policy stringency and existing adoption trends Applied a percentage to reduce the California adoption scenarios to represent EVs in PG&E s territory Converted EV adoption scenarios to GWh forecast Uncertainty analysis and Monte Carlo simulation produce final Expected forecast and probability distribution
EV Load (GWh) PG&E s EV Forecast vs. the CEC s IEPR 3000 2500 2000 1500 IEPR Forecast for PG&E (2016) PG&E Forecast (2017) 1000 500 0 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 Both forecasts exceed the statewide ZEV Mandate PG&E s 2017 forecast is lower than the IEPR forecast, mainly due to different methodologies: The IEPR mid-case forecast includes a vehicle choice model, with an EV preference parameter that increases over time PG&E s Expected forecast is the result of a Monte Carlo simulation of its original EV adoption scenarios 18
Overview of PG&E s Disaggregation Methodology 19
Geospatial Disaggregation Total system-level EV forecast is allocated to each county based on electric vehicle registration data from Polk via EPRI Each county is allocated a number of EVs for each year of the forecast based on its current share of system-level EVs County-level EV adoption estimates are allocated to the feeder level using economic and demographic variables that characterize likely EV adopters: Income Homeownership Education 20
Joint IOU Challenges 21
Challenges to Developing Disaggregated EV Forecasts EV registration data is aggregated on the county/zip code level. Individual registration data is difficult to obtain and has restrictions on the use cases. Historical geographic adoption may not be indicative of future adoption EV load is mobile and not always tied to the residence or the same feeder 22
Q&A 23