EV Charging → Question 4

What Happens When the Grid Says No?

49 stations. 4,151 substations. An estimated 30 face grid constraints.

Model B's 49 Stations vs Real HIFLD Substation Capacity
Feasible (9 stations) Downsized (30 stations) Need Upgrades (10 stations)
Note: Substation capacity is estimated from voltage class (345kV ≈ 500 MW, 138kV ≈ 150 MW, 69kV ≈ 50 MW) and loading from county growth data — not utility SCADA measurements. The specific stations flagged may differ from reality, but the pattern (30–40% of high-demand sites face grid constraints) is consistent with NREL findings.

Model B optimizes for demand — 49 stations, 948 ports, 88.5% coverage. But when we check each station against its nearest real ERCOT substation using voltage-class capacity estimates and county-growth loading proxies, we estimate 30 of 49 stations would face grid constraints. The AIRLINE substation in Harris County, estimated at 71.4% loaded, could handle 13 ports, not 20. TENTH STREET in Bexar County (estimated 77.8% loaded) would drop from 20 ports to 8. The specific numbers depend on utility data we don't have — but the pattern is confirmed by NREL: grid constraints are the binding constraint for DCFC deployment.

The Grid Right-Sizes Model B

9
Feasible
Substation has capacity
30
Downsized
Ports reduced by grid
10
Need Upgrades
$5.5M total
294
Stranded Ports
948 planned → 654 deliverable

Top 5 Constrained Stations

AIRLINE — Harris County
Substation at 71.4% loading. Can deliver power for 13 ports, not the 20 Model B planned.
20 → 13
7 ports stranded
TAP305257 — Dallas County
Substation at 73.9% loading. Remaining capacity supports 11 ports.
20 → 11
9 ports stranded
TENTH STREET — Bexar County
Substation at 77.8% loading. Only enough headroom for 8 of 20 planned ports.
20 → 8
12 ports stranded
UNKNOWN307367 — Tarrant County
Substation at 70.1% loading. Can support 14 of 20 planned ports.
20 → 14
6 ports stranded
WELLS BRANCH — Travis County
Substation at 65.7% loading. Nearly full capacity — 18 of 20 ports viable.
20 → 18
2 ports stranded
Ports Planned vs Ports Deliverable — Top Constrained Stations
Finding
Based on voltage-class capacity estimates, we project 30 of 49 stations would face grid constraints. Model B puts 948 ports on paper; the grid can likely deliver ~654. Model C accounts for this upfront — it right-sizes from day one, avoiding an estimated $68M in wasted NPV.
~430 lines of code, 4,151 real substations. That is the entire difference between Model B and Model C. Without it, 294 ports sit dark and you are looking at $50M+ in grid upgrades nobody budgeted for. With it, you save 28% of 10-year NPV.

Model C: Grid-aware optimization — nearest-substation capacity check using 4,151 HIFLD substations with voltage-based capacity estimates.

Methodology note: Substation locations come from the HIFLD/DHS infrastructure database (real). Capacity is estimated from voltage class (345kV ≈ 500 MW, 138kV ≈ 150 MW, 69kV ≈ 50 MW). Loading is estimated from county growth classification, not utility SCADA data. Which specific substations get flagged will shift with better data -- but NREL consistently finds that 20 fast chargers can double or triple a building's electrical load, and interconnection takes 6–24 months. The pattern holds even if the pin locations move.

What a real deployment would need: Utility feeder data (from Oncor, CenterPoint, AEP Texas), transformer MVA ratings, real SCADA loading data, and protection coordination studies. We are identifying where the hard questions are, not answering them.

← Prev Q3: Station Sizing Study Back to EV Charging Next → Q5: True Cost