Berkeley, February 2024
This repository contains all the codes, figures, and some raw data tables used in the study 'Impact of Grid Infrastructure Limits on Electric Vehicle Charging Access for California Households'. ICA data tables could not be uploaded due to their large volume. The files are organized into folders numbered 1 through 7, each corresponding to a different stage of the analysis.
A diagram illustrating the architecture of the repository is displayed at the end of the README file.
All the R and python files in this repository are briefly described below.
01_demodata.Rmd
- Gathers demographic information for census block groups in California.
- Census data, downloaded through the Census Bureau, are from the "American Community Survey, ACS 5-Year Estimates"
- CalEnviroScreen data are from CalEnviroScreeen 4.0 results
- Output: bgCAcensusCES.csv
02_cpolys.Rmd
- Pulls in circuit polygon data (outputs from ArcGIS) and analyzes outcomes of spatial processing.
- SCE inputs: SCE_ICAall_cspoly.csv, SCE_ICAall_cpolybg.csv, SCE_ICAall_ctotpoly.csv
- PG&E inputs: PGE_ICA23_cspoly.csv, PGE_ICA23_cpolybg.csv, PGE_ICA23_cpolybg_ICAavail, PGE_ICA23_ctotpoly.csv, PGE_subs.csv
- Demographic data inputs: bgCAcensusCES.csv
- Outputs: PGE_all.csv, PGE_ICAall.csv, PGE_ICAallreal.csv, SCE_ICAall.csv, SCE_ICAallreal.csv
03_features.Rmd
- Assembles and cleans features to be used in regression runs.
- Inputs: PGE_all.csv, PGE_ICAall.csv, SCE_ICAall.csv, bgCAcensusCES.csv
- Outputs: PGE_ICAalldemo.csv, PGE_ICAalldemoreal.csv, SCE_ICAalldemo.csv, SCE_ICAalldemoreal.csv
04_allocation.Rmd
- Assigns hosting capacity to households.
- Inputs: PGE_ICAalldemo.csv, PGE_ICAalldemoreal.csv, SCE_ICAalldemo.csv, SCE_ICAalldemoreal.csv
- PG&E outputs: PGE_ICAalldemoalloc.csv, PGE_ICAalldemorealalloc.csv, PGE_ICAalldemotrees.csv, PGE_ICAalldemotrees_real.csv
- SCE outputs: SCE_ICAalldemoalloc.csv, SCE_ICAalldemorealalloc.csv, SCE_ICAalldemotrees.csv, SCE_ICAalldemotrees_real.csv
05_access.Rmd
- Calculates overall household access and evaluates household access based on the sizes of residential buildings.
- There is three other variants of this code, focusing on: access by all county, access by most populated counties, vehicles access.
- Inputs: PGE_ICAalldemoalloc.csv, SCE_ICAalldemoalloc.csv
- Outputs: Fig. 3.2, Fig. 3.3, Fig. 4.2, Fig. A.2, Tab. 3.1, Tab. A.6
05_access_scenarios.Rmd
- Projects scenarios of charging demand and analyses household access for each scenario and considering the residential building sizes
- Outputs: Fig. 5.1, Fig. A.8, Tab. 5.1, Tab. A.7, Tab. A.8
06_analysis.Rmd
- Analyzes the relationships among various service, infrastructure, and demographic features in relation to the hosting capacity.
- three other variants of this code, focusing on: (1) same relationships but at a county level and (2) relationships between proportion of multi-unit dwellings and demographic features.
- Outputs: Fig. 3.5, Fig. 4.1, Fig. A.6, Fig. A.9, Fig. A.10
07_regression.ipynb
- Calculates accuracy of linear and random forest (RF) regression models for hosting capacity prediction
- Assesses all variable importance as predictors
07_regression_plot.Rmd
- Generates plots of the results produced by the Python notebook '07_regression.ipynb'.
- Calculates the interactions between all variables using iterative random forest (iRF) models. More information on iRF in Basu et al., 2018.
- Outputs: Fig. 3.4, Fig. A.3, Tab. A.5
