stratum atlas
Model DocsWider Book
M14

BTM-vs-Grid Separation Model

Which promised MW depend on the grid and which sit behind the meter with onsite generation.

Assumptions

Everything the model believes, stated before the math. If you do not buy one of these, do not buy the output.

  1. Grid-dependent and behind-the-meter MW are different risk objects and are never blended into one number.
  2. Classification is deterministic from interconnection data and generation data joined to the campus project.
  3. Ambiguous MW are labeled, never forced into a path.
  4. Behind-the-meter is not an ask on the grid: BTM MW does not appear in the grid ask accounting.

Derivation

Promised MW per project is classified into two paths by joining interconnection records (spec Section 11) and generation records (Section 12) to the campus project:

The Delivery Model then runs separately per path, each with its own dependency set and base rates. Only path-level results are reported: a grid MW figure carries grid dependencies (transmission, interconnection agreement, transformers); a BTM MW figure carries onsite generation dependencies (equipment delivery, fuel, permits). There is no blended number to hide the mix in.

Identification strategy

Why the estimate supports the claimed meaning - and where the model refuses to claim more than the data carries.

  • The paths carry structurally different failure modes, so blending them would produce an average that describes no actual asset. The model's identification is the separation itself: each path's answer is valid for its own risk class.
  • The ambiguous bucket is kept visible rather than allocated: forcing ambiguous MW into a path would manufacture precision where the record carries none.

Inputs and outputs

Reads

Interconnection data
Spec Section 11: queue positions, agreements, awarded capacity
Generation data
Spec Section 12: onsite generation records
Campus-project join
The project record's linkage to its campus

Writes

  • Grid-dependent MW vs behind-the-meter MW per project
  • Separate delivery analysis per path
  • Ambiguous MW, labeled

Calibration and gate evidence

Validation design

  • Classification audited by hand on the pilot corpus; ambiguous MW are labeled, never forced into a path.

Evidence on record

  • The hand-audit results are not yet published in the sealed source.

Failure modes

The ways this model declines, degrades or goes UNKNOWN - and what it prints when it does.

  • Missing interconnection or generation records: the MW lands in ambiguous, labeled.
  • A project with both paths live reports both figures separately; nothing is netted.

Sensitivity

  • The split moves MW at risk materially: a campus whose MW are mostly BTM carries equipment risk, not queue risk, and the Delivery answer differs accordingly.

Iron rules

The two paths carry different risks and are never blended into one number.

Source and audit

Written from the sealed model source. Every run is recorded - model version, frozen inputs, input hash, timestamp. There is no unrecorded model execution.