Raptrix Power
Raptrix Power
Request a pilot
Raptrix Sentinel

Planning-grade physics
from the live operating state.

Exhaustive N-1 on the live case. Simultaneous N-2 from the defined common-mode list. Sequential N-1-1 on by default. Recovery-validated against LTE / Normal windows. Sentinel ingests existing EMS exports at N-200+ scale without requiring you to maintain contingency lists. It surfaces only the scenarios that truly matter, each with an auditable certificate.

Request a pilotSee the workflow

Your control room is running yesterday’s assumptions on tonight’s grid. The State Estimator solves a high-fidelity picture every few minutes — and then your tools approximate everything they do with it.

Sentinel changes the equation. It takes the SE case you already have, screens every credible contingency at planning-grade physics, validates whether the system can actually recover within NERC limits, and hands you back a short, defensible list of what actually matters — before the next interval ticks.

And it does it without rip-and-replacing your EMS. No new SCADA. No new contingency lists to maintain. No fidelity traded for speed.

How Sentinel works

Six stages. One physics core. No shortcuts.

Every solved State Estimator case becomes an exhaustively screened N-1 plus defined N-2 study — recovery-validated, certifiable, and written to SAL.

  1. 01

    Ingest

    Solved SE case · default 5 min

    RAW, PSLF, or native State Estimator exports — no rip-and-replace.

  2. 02

    Screen

    Exhaustive N-1 + defined N-2

    N-1-1 default after intervening recovery. Simultaneous N-2 is the defined common-mode list.

  3. 03

    Solve

    Full-AC Newton-Raphson

    Strict Q-limit enforcement and PV-PQ switching. No FDNR shortcuts.

  4. 04

    Recover

    LTE / Normal-window recovery

    Q-limits enforced. Greedy reserve dispatch + switched-shunt recovery (on by default). Only unrecoverable cases are surfaced.

  5. 05

    Certify

    Auditable per case

    Every filtering decision is grounded in defensible physics.

  6. 06

    Hand off

    Canonical .rpf out

    One file carries the operating state forward into planning. Every cycle also writes SAL.

What makes Sentinel different

What operators get that static lists and EMS screens cannot deliver.

01

No manual contingency lists.

Exhaustive N-1 on the live case. Simultaneous N-2 from the defined common-mode list. Sequential N-1-1 on by default. Recovery-validated against LTE / Normal windows.

02

Tens of thousands of "what-ifs" → a short list of unrecoverable scenarios.

From the actual real-time operating point, Sentinel identifies the small number of contingencies that have no path to recovery within NERC limits using available reserves and controls — and tells you the auditable reasons why.

03

Operators get visibility they never had before.

Real-time, planning-grade analysis on the exact current operating state — exhaustive N-1 + defined N-2, N-1-1 default, recovery-validated against LTE / Normal windows — at a speed and fidelity no existing EMS or real-time tool delivers.

04

Planning gets a library of real operating states they never had before.

Every cycle writes the empirical library — Sentinel Archive Library (SAL) — plus a canonical solved .rpf and replayable events. Planning teams finally get high-fidelity cases from actual operating conditions — not stylized study cases.

05

True low-friction bolt-on.

Works with existing RAW, PSLF, or native State Estimator exports. No rip-and-replace of EMS or deep SCADA integration required.

06

Native modeling of large, dynamic computational loads.

Arms VRT / campus transfer from profiles on the live SE case. Purpose-built for AI data centers and other large, voltage-sensitive, rapidly varying loads under NERC Level 3.

Auditable by design

Every flagged contingency ships with a certificate.

Not a black box. Every filtering decision, every Q-limit enforcement, every reserve redispatch is grounded in defensible physics — with the receipts attached.

Recovery Certificate
sentinel://case/2026-05-22T13:05Z/ctg/L7842-out
Unrecoverable
contingencyN-2 · line + xfmr
case_modesolved · live SE
q_limitsenforced
pv→pq switches3
recovery_window15 min LTE
reserves_availinsufficient
// reason
Voltage collapse at bus 4421 under greedy reserve redispatch;
no admissible control sequence within NERC LTE envelope.
metadata.fingerprint 7c4a9e…b201schema v0.14.2 · arrow ipc
The physics, in detail

Engineering choices that hold up under review.

  • Full-AC Newton-Raphson

    with strict Q-limit enforcement. A converged solution that violates generator Q limits is marked infeasible — non-negotiable physics.

  • Intelligent PV-PQ switching

    prevents naïve "infinite Q" hacks while still recovering solvable cases.

  • Solver cloning

    for massive parallelism — hundreds of full-AC N-1 and defined N-2 cases in near real-time.

  • Hybrid contingency selection

    graph-theory plus operator-defined guardrails. Betweenness centrality, electrical distance, critical cut-sets, and historical patterns augment — never replace — operator judgment.

  • IBR-tagged screening

    IBR-tagged screening. FACTS/PST carried in .rpf; NR device control is not claimed. PTDF / LODF ranking is awareness, not FACTS-aware NR voltage control.

  • NERC recovery validation

    Q-limits enforced. Greedy reserve dispatch + switched-shunt recovery (on by default). Only unrecoverable cases are surfaced. RAS/SPS is enrichment, not closed-loop execution.

  • First-swing stability screening

    when dynamics data is available — on the same solved operating point. Large-load rejection scenarios today; inverter-centric models expanding.

NERC Level 3 · Large Loads

Built for the grid that’s arriving — not the one we modeled in 2010.

AI data centers and other large, electronically coupled loads behave nothing like traditional demand. Sentinel treats them as first-class citizens — so operators and planners get a defensible, physics-based view of system risk without waiting on EMS vendor roadmaps.

Sentinel for large loadsSite selection studies
Voltage sensitivity

Modeled natively at the bus, not assumed away with a generic load.

Up / down ramps

Rapid ramp and load-reduction schemes carried with the operating point in .rpf.

Contingency impact

First-class in N-1, defined N-2, and profile-armed VRT.

IBR interaction

IBR-tagged screening. FACTS/PST carried in .rpf; NR device control is not claimed.

VRT / large-load arming

VRT / large-load arming from load profiles.

Every operating state Sentinel evaluates ships with the computational-load profile attached — seasonal envelopes, buildout ramps, voltage guardrails, and reduction schemes.

carried in .rpf
Roadmap · Q3

Sentinel Phase 2 — look-ahead advisory.

Phase 2 extends the same high-fidelity core with an integrated LP. Using expected topology, commitment, and large-load forecasts, it will recommend proactive battery and reserve positioning — turning real-time visibility into forward-looking operational decisions.

Low-friction bolt-on to any existing EMS. One canonical solved record. Auditable certificates on every filtering decision. Planning-grade physics in operations timeframes. Built for the grid that’s coming.

No-cost pilot for qualified teams

Pilot Sentinel on a recent SE case.

Send us a representative State Estimator export. We will run exhaustive N-1 + defined N-2 with LTE / Normal-window recovery and walk through the results with your team.

Request a pilotEmail founder direct

info@raptrixpower.com · Founder direct: matthew.musto@raptrixpower.com · github.com/RaptrixPowerFlow