Hydrogen Corridor Analysis

Interactive Map

Corridor Overview

Energy-optimal route via I-880 (orange) vs. shortest-distance route via Niles Canyon (gray). Green circles are real NREL-reported H₂ stations. The two routes are nearly the same distance. The difference is entirely terrain. Dashed blue markers show where each route passes within range of a station.

Part 1: Energy-Optimal Routing

Dijkstra's on Energy Cost

The road network is fetched from OpenStreetMap via OSMnx, filtered to major roads (motorway, trunk, primary, secondary). Each node gets an SRTM elevation from the Open-Elevation API; edge grades are computed from adjacent node elevation differences.

Edge weights: weight = (dist_miles + elev_gain_m / 25) × H₂_kg_per_mile. NetworkX Dijkstra finds the minimum-energy path. The two routes cover nearly the same distance (43.8 mi vs. 43.7 mi). The Niles Canyon route climbs 302m vs. just 43m on I-880.

Result: the energy-optimal route consumes 18.8% less hydrogen despite being the same length. Flat terrain beats shorter distance when the grade penalty is steep enough.

Route comparison: shortest distance vs energy-optimal

Part 1b: Terrain Analysis

SRTM Elevation Profile

Elevation profiles for both routes overlaid. The orange line (I-880, energy-optimal) hugs sea level at 10–30m for the full 43.7 miles. The gray dashed line (Niles Canyon, shortest-distance) climbs through the East Bay Hills to a 330m peak before descending to Fremont, adding 302m of effective uphill work that directly translates to extra hydrogen consumption.

Part 2: Infrastructure Coverage

Coverage Gap Analysis

Operational H₂ stations are pulled from the NREL Alternative Fuels Station Locator API (real stations, not synthetic data). 7 stations appear in the Bay Area corridor (True Zero network + Shell). The route is sampled every 500m and the distance to the nearest station is computed in EPSG:3310 (California Albers).

At a 30-mile conservative threshold (appropriate for hydrogen buses in hilly terrain), the Bay Area portion of this corridor is fully covered. The True Zero network provides continuous coverage. This is actually a meaningful finding: the corridor is infrastructure-ready for hydrogen fleet deployment today.

Dotted green lines show where each station projects onto the route. The chart structure generalizes to any corridor where coverage gaps would emerge.

Part 3: Optimal Siting

Infrastructure Placement Under Spatial Constraints

Candidate new station sites are placed at midpoints of coverage gap segments. Candidates are filtered by two spatial constraints: road grade ≤ 5% (flat terrain lowers construction cost and improves heavy-vehicle access) and minimum 10-mile inter-site spacing. The Bay Area corridor at a 30-mile threshold shows no gaps; the siting algorithm is designed to trigger on longer or underserved corridors where gaps do exist.

Constraint 1

≤ 5%

Maximum road grade at candidate siting location. Steep terrain increases excavation, retaining structures, and equipment delivery cost for hydrogen infrastructure.

Constraint 2

10 mi

Minimum spacing between recommended new sites. Prevents clustering and ensures each site adds the maximum new coverage relative to existing stations.

Bay Area finding

Covered

At a 30-mile conservative threshold, the Bay Area I-880 corridor is fully served by existing stations. No new siting required. Corridor is infrastructure-ready.

Data source

Real

Station locations from NREL AFSL API. Road network from OpenStreetMap via OSMnx. Elevation from NASA SRTM via Open-Elevation. No synthetic data.

Methodology

Pipeline

End-to-end spatial optimization pipeline. Each step uses real external data sources; the analysis is fully reproducible from the Jupyter notebook.

OSM Network
OSMnx → SRTM Elevation
Open-Elevation → Energy Weights
NetworkX → Optimal Route
Dijkstra → NREL Stations
REST API → Coverage Gaps
GeoPandas → Siting
Constraints

01

Road Network: OpenStreetMap via OSMnx

Fetches the drivable road network within the Oakland–San Jose bounding box, filtered to major road types via ["highway"~"motorway|trunk|primary|secondary|..."]. Graph is simplified to merge parallel edges and remove intermediate degree-2 nodes.

02

Terrain Elevation: NASA SRTM via Open-Elevation API

Batch-queries SRTM 30m elevation data for every node in the graph. SRTM (Shuttle Radar Topography Mission, NASA 2000) provides near-global terrain coverage at 30m resolution, a real remote sensing product. Edge grades are computed from adjacent node elevation differences.

03

Energy-Weighted Routing: NetworkX Dijkstra

Each edge gets energy_cost = (dist_miles + elev_gain_m / 25) × (1/60). The 25m constant comes from empirical H₂ vehicle efficiency data: 25m elevation gain ≈ 1 extra effective mile of consumption. NetworkX finds the shortest path on this energy graph.

04

Station Coverage: NREL Alternative Fuels Station API

Pulls all operational (status=E) hydrogen (fuel_type=HY) stations in California. The route is sampled every 500m and for each sample point the distance to the nearest station is computed in projected EPSG:3310 (California Albers). Coverage threshold: 30 miles (conservative for heavy vehicles on varied terrain).

05

Optimal Siting: Spatial Constraints

Contiguous gap segments are identified and midpoints become candidate sites. Grade at each candidate is computed from incident edge grade_abs values from the OSM graph. Candidates failing grade > 5% or spacing < 10 mi are eliminated. Remaining candidates are ranked by distance from nearest existing station.

View Notebook on GitHub → ← Back to portfolio

Hydrogen Corridor Routing &Infrastructure Gap Analysis