What we use. What we add.

A pipeline or a unit modelled as coupled first-order hyperbolic equations, discretised in time only, so the infinite-dimensional structure, the energy and the observability are kept. Constrained observers, Kalman filters, moving-horizon estimators and predictive control designed on that model. Almost all of it validated in simulation.

Moving-horizon leak detection, localisation and constrained size estimation, with a transform that decouples size from location; a discrete observer with a classifier as the lighter variant. Wavepin reimplements these as comparison methods and builds the hypothesis layer on top; Program uses them as research-backed test scenarios.

Simultaneous data reconciliation and gross-error detection with a heavy-tailed noise model; a robust dynamic method when the noise is unknown. Balance's method.

Plant-model mismatch detection and quantification for predictive controllers: the pattern for deciding whether a persistent residual is a wrong model or a wrong measurement. Balance's second line.

A transient hydraulic model plus a learned corrector for volume-balance drift; checked against industrial benchmark data by its authors, no published field result. Impact takes interface arrival times as an input. Commercial batch trackers already offer the feature; the claim is explained error, not the feature.

Port-Hamiltonian modelling of process networks and sensor-location selection under delayed measurements. Register's physical-consistency check and Wavepin's placement argument.

Koopman reduced-order models for hierarchical predictive control and generative modelling of irregular observations. The fast engine behind deterministic twins; Impact's span check stays reduced-order and labelled unverified.

Physics-informed reconstruction of pressure, flow and hydrogen fraction along a line from sparse sensors; a Hamiltonian-preserving monitoring model. Nothing on the site claims it.

Gearbox fault diagnosis under varying conditions by domain adaptation and federated domain generalisation, on laboratory and wind-turbine data. The method behind a parked pump-health line; revisited only as a joint dataset project with real pump data.

A plunger-load observer for sucker-rod pumps from surface data, discrete-time thermodynamic models of heat exchangers, reactors with recycle delay. Not used by the entrance products.

An alarm-management research line on predicting alarm floods early and ranking operator support. Alarm draws on it as neighbouring evidence; the harness adds consequence, planned work, the rehearsed response and receipts.

A European collaborative-intelligence network on whether operator behaviour predicts correct alarm response, with decision-support screens. Context for Alarm's one-approval screen beside the SCADA display.

A systematic review, model and case study of digital twins for leak detection and fault diagnostics in gas pipelines. Program's test scenarios are measured against it.

Observers for hydrogen-blended networks on a graph, and optimal control of blending from a national laboratory. Watch items for a later blend line.

Multiproduct network scheduling with indexed batch numbering, fair-open-access scheduling, rolling-horizon methods. Impact's what-if tab borrows from this literature without becoming a scheduler.

Drilling-automation centres and simulators; virtual flow metering with years of well data. The rig twin presents rehearsal scenarios, not automation, and is parked until a published field result exists.

Running ductile fracture and depressurisation data from about ten full-scale tests worldwide. No product; the nearest asset for dense-phase lines is a non-isothermal pipeline observer.

Flow-loop and terminal-scheduling work. The harness stays with lines, stations and the control room.

The estimation literature we build on does not cover these. The site claims nothing here. The oil and gas procedure library needs a procedures writer, not an estimator.

Every cited method carries one of four badges. Research-backed: published, simulation-validated. Simulated: run in our own simulator. Lab-realised: shown on a laboratory rig. Field: a published result on a real asset.

Leak localisation on a laboratory pipeline manifold with an extended Kalman filter, in a preprint that is not yet peer-reviewed. An industrial pipeline application has been mentioned in a seminar without published detail. Everything else in the methods list is research-backed or simulated.

Everything near-product in the estimation literature is evidence: what the pipe is doing, from sparse sensors. That is read-only toward the control system and matches the harness rule that deciding is a person's verb. Control appears only inside rehearsal twins.

One straight line, one product, no branches, no batches, no gas. The most advanced member of the family and still simulation. Leak and Explain are ready for a data pilot.

Parked lines use Wi-Fi channel-state sensing. No hardware has run, no measurement exists, and simulated sensing is not evidence. Presence is not identity or a headcount; never a life-safety or muster system. Range and behaviour in metal-rich cabins and stations are unknown until the bench protocol runs; hazardous-area certification is open.

Gas transmission under 49 CFR 192.631, hazardous liquids under 195.446: the alarm management plan, the monthly review of off-scan, inhibited and forced points, the shift-change record, controller workload. Register's review pack and Alarm's shift view are built to these.

• 49 CFR 192.631 control room management
• PHMSA notice of amendment, March 2026 (49 CFR 195.446): handover gap, alarm-plan review without KPIs, workload

After a 2022 refinery fire the CSB recommended per-flood performance targets (number of floods, duration, alarm count, peak rate) in ISA-18.2; status Open – Acceptable Response. Alarm's flood finder reports exactly these.

• CSB notation item 2025-41 (13 May 2025)

API RP 1175 for the leak-detection programme: KPIs, testing, culture. The 2022 valve rule: closure within 30 minutes of rupture identification, review within 90 days, on transmission and hazardous-liquid lines (vacated for gathering lines). 49 CFR 195.402 for the procedures manual. Program compiles these as contracts.

• API RP 1175, Pipeline Leak Detection Program Management (2nd ed., April 2022)
• PHMSA valve rule (87 FR 20940, 8 April 2022)
• 49 CFR 195.402 procedures manual
• Valve rule recordkeeping notice (Federal Register, 31 July 2025)

FERC Order No. 809 set the gas day and the nomination cycles. 18 CFR 284.12 incorporates the NAESB standards and requires critical notices on the pipeline website. Order No. 587-AB adds geographic information to critical notices from 1 January 2027. Impact's tables and draft notice follow these.

• FERC Order No. 809 (80 FR 23198, 24 April 2015)
• 18 CFR 284.12 NAESB standards; critical notices
• FERC Order No. 587-AB (91 FR 31651, 28 May 2026): geography in critical notices

49 CFR 195.452: high consequence areas and segment risk analysis on hazardous-liquid lines. Impact's consequence segments.

• 49 CFR 195.452 high consequence areas

The OPC UA companion specification for ISA-95 gives the physical asset, equipment and location models; CFIHOS the capital-project information handover. Register's loaders target these shapes.

• OPC 10030 v1.00, section 4.1: ISA-95 companion specification
• CFIHOS (IOGP JIP36; news item, April 2020)

The economic and ESG cost of lost-and-unaccounted-for gas errors. Balance's business case.

• Pipeline & Gas Journal, July 2022