Kardashev Labs · ERCOT
Site Clearance

Published Rubric v1

Methodology

You draw a Footprint. In Generation Mode we return a Clearance: a Band of strong, mixed, or weak, plus a Verdict (headline and drivers). Only these inputs may move the Band.

This is not an ERCOT interconnection study. The Band is a public-data judgment of a Footprint, not an IA result and not a time-to-power number.

The short version

  1. Clip to Texas counties

    The Footprint has to stay inside Texas. We intersect it with Census county shapes. Anything that spills into Mexico or out of state gets rejected.

  2. Pull the county Generation Queue

    GIS filings name a county, not a lat/lon pin. Pending Interconnection Projects and MW come from Scored Counties (coverage at least 5% of the Footprint). Tiny edge clips stay on the map as slivers and drop out of the Band.

  3. Stack Full Process and Settlement Zone prices

    Peer Full Process medians for the dominant CDR Zone and fuel, with sample size shown (n below 20 is not treated as a measured Full Process). Trailing LMP stress for the mapped Settlement Zone. Wire Proxy and DC Screen are Attached Evidence. They do not move the Band. Published Curtailment, if shown, is the same: shown, not in the Band.

Geography (county resolution)

ERCOT's public generator queue is county-resolution. Midland County is not a point on a map; it is a polygon with every pending Interconnection Project that listed "Midland." If your Footprint covers 60% Midland and 40% Martin, both counties can feed the Band, and the map paints the overlap so you can see what we counted.

We do not invent project pins from bus names or substations in v1. When we can geocode POIs honestly, the map will get sharper. Until then, county is the honest unit.

What enters the Band

Generation Queue pressure

Pending (never Approved for Energization) GIS Interconnection Projects and MW in Scored Counties, from the latest GIS Report.

Peer Full Process

Measured medians from about 97 months of GIS Film. Full Process is Screening Start to Approved for Energization. Browse the same series on our timelines tool.

Settlement Zone stress

Trailing-year negative-price hours, RT volatility, RT minus DA spread for the mapped LZ_* Settlement Zone. CDR labels like COASTAL map to the nearest published series. That mapping is a translation, not identity.

Attached Evidence (shown, not in the Band)

  • Wire ProxyPublic HIFLD transmission-line km per county area vs the Texas median. Approximate geometries; not CEII; not a thermal rating.
  • DC ScreenMicrosoft GridSFM Texas model (OSM + EIA synthetic network). We add your MW as injection (Generation Mode) or withdrawal (Load Mode), solve a DC power flow, and read local branches near the county. Not ERCOT CEII; not N-1; slack absorbs imbalance.
  • AC-OPF / contingency packsNot built. DC Screen only.
  • Published CurtailmentISO/fuel grain the ISO already publishes, if shown. Resource-level reconstruction is not ingested.
  • Load ModeERCOT does not publish Large-Load pins. Load Mode returns Attached Evidence (west/north/other aggregates). It is not a Clearance until Large-Load MW can be attributed to the Footprint. See LLWG.

Sources

Everything in the Band comes from public filings. Links below go to the sources, not our paraphrase.