Boundary Verification for Behaviorally-Coupled Computational Load

Two engineering specifications addressing a reliability gap in the interconnection of large computational loads to the bulk power system.

The problem

Reserve adequacy and frequency-response planning have historically been sized against an enumerated largest credible contingency — typically the loss of the single largest generating unit — treated as a physical, independent event.

Large computational loads introduce a contingency of a different kind. Geographically dispersed loads, operating under a shared orchestration or control plane, can reduce demand in near-simultaneity. This is a behaviorally-coupled block: its members are physically independent but behaviorally correlated.

Topology-based N-1 analysis does not see it. N-1 scores dispersed loads as independent low-probability events, so the coordinated-drop magnitude of the coupled block is never assembled into a single contingency and never tested against the system's frequency response.

The specifications

1. The Minimum-Inertia Condition

Defines the frequency-response absorption conditions a behaviorally-coupled load block must satisfy to be interconnected. Establishes the coupled-block magnitude ΔP_block, and mandates its evaluation against the swing-equation RoCoF limit and the primary-frequency-response adequacy bound — computed at a credible minimum-inertia condition H_min derived from operator dispatch data, not at average inertia.

2. The Coupling-Discrimination Specification

Defines the criteria by which a load is admitted into a coupled block. Separates orchestration-coupling — loads moved by a shared control plane, a common-mode contingency — from environmental correlation, where loads independently respond to a common public signal. Ensures the absorption test constrains genuine common-mode risk without penalizing independent demand-side flexibility.

What these specifications rest on

They rest on the swing equation, on inertia-duration-curve analysis of operator dispatch data, and on measured boundary telemetry. Every quantity in them derives from something an engineer can check against the physical record.

They are implementation-independent: they constrain measured boundary behavior and physical frequency response. They do not require, prescribe, or depend on any operator's software, market design, vendor, or certification service.

The evaluation criterion

These specifications propose a criterion that may be applied uniformly to any demand-side flexibility, storage, or orchestration proposal:

Does the proposal demonstrate operational response only — that a resource responded when called — or does it also provide for independent structural verification that repeated response preserves system invariants, does not create unaccounted-for coupling, and does not convert temporary flexibility into unverified dependency?

A demonstration that a resource responded correctly does not establish that the system's structure was preserved across repeated response. The two are complementary layers, not alternatives.