Explore EV charging demand with TEMPO¶
This tutorial walks you through loading the dsgrid TEMPO dataset — projected electric vehicle charging demand across the contiguous United States from 2024 to 2050 — and asking your first questions about it. Allow about five minutes.
What you'll load¶
The TEMPO dataset comes from NLR's TEMPO project (Transportation Energy & Mobility Pathway Options), part of the dsgrid framework. It models hourly county-level charging load for light-duty passenger EVs.
| Table | Description | Rows |
|---|---|---|
charging_profiles |
Hourly demand by census division, scenario, and model year | ~1.4M |
annual_state |
Annual demand by state, vehicle type, and scenario | ~16K |
annual_county |
Annual demand by county (FIPS), vehicle type, and scenario | ~448K |
Three EV adoption scenarios are modeled:
- reference — AEO Reference Case baseline
- efs_high_ldv — High electrification of light-duty vehicles
- ldv_sales_evs_2035 — All new LDV sales are electric by 2035
All energy values are in MWh. Vehicle types are split into BEV and PHEV variants across four body styles (compact, midsize, SUV, pickup).
1. Install datasight¶
Don't have uv yet? See Install datasight.
2. Download the dataset¶
This downloads ~19 MB from OEDI's public S3 bucket in about 10 seconds and creates:
tempo-project/
├── .env # Connection settings
├── dsgrid_tempo.duckdb # DuckDB database (~19 MB)
├── schema_description.md # TEMPO data documentation
└── queries.yaml # Example EV charging queries
3. Add an API key¶
Edit .env and add your Anthropic key:
Using GitHub Models or Ollama instead? See Set up a project for alternative configurations.
4. Launch the web UI¶
Open http://localhost:8084.
5. Ask your first questions¶
Try any of these prompts:
- Total projected EV charging demand by scenario and year
- Which states have the highest projected EV charging demand?
- Show the hourly charging profile for a summer day in the Pacific division
- Compare BEV vs PHEV charging demand over time
- How does the all-EV-by-2035 scenario compare to reference for California?
- Average daily charging pattern by season
Data source¶
The data comes from OEDI (Open Energy Data
Initiative) on AWS S3 at s3://nrel-pds-dsgrid/tempo/tempo-2022/v1.0.0.
Published by NLR based on the 2023 technical report by Yip, Hoehne, Jadun
et al. The data is publicly accessible with no credentials required.
What's next¶
- Build a dashboard — pin results, apply cross-card filters, and export as HTML.
- Ask questions in the web UI — follow-ups, clarifying prompts, and the schema sidebar.
- Configure semantic measures — override default aggregations for rates, ratios, and calculated measures.