Reconstruct the evidence
Ingest the project documents and plant model. Every extracted fact keeps its document, page, method and confidence.
Deterministic extraction · SHA-256 provenanceAssure3D · Pilot programme
Fire and gas coverage,
verified by
physics.
Your mapping study proved a design on the day it was issued. Barrier Intelligence shows whether it still detects the releases your plant can produce, across real geometry, credible scenarios and every wind.
Functional prototype entering pilot deployment · Offline deterministic core · Engineer sign-off retained
of hydrocarbon releases in the cited UK HSE dataset went undetected by the as-installed F&G system.
Source: UK HSE RR67201 · The verification gap
Installed is not the same as effective.
A fixed coverage drawing cannot see the wind change, congestion grow, a detector move, or a process condition leave the assumed case. The plant keeps changing. The claim does not.
01
A handful of cases. Manual studies miss the combinations where blind zones survive.
02
A frozen basis. The design is rarely re-tested after MOCs, detector moves or geometry changes.
02 · From documents to a defensible verdict
The same engineering basis.
A much stronger answer.
Barrier Intelligence does not replace your HAZOP, LOPA, philosophy or 3D model. It carries their logic forward and tests whether the physical detection response earns the credit.
Run the plant, not a sphere
Enumerate credible release nodes, winds and process cases. Score detector response against the true 3D dispersion envelope.
Surrogate ranks · Solver decidesBind evidence to credit
Trace each HAZOP scenario through physical detection votes and issue a barrier verdict an engineer can inspect and sign.
The register signs · Physics owns the verdict03 · The F&G Barrier Credit Register
One statement of what the barrier can honestly claim.
Every scenario is tied to its hazard basis, detector votes, coverage, timing, evidence quality and smallest physical fix. Confidence may demote a verdict. It can never manufacture credit.
- Designed versus installed, side by side
- Dead zones and dynamic minimum detectable leak size
- Source-ready evidence for review and ALARP sign-off
05 · Proof, not projection
Better coverage does not have to mean more hardware.
Detectors in a published Kenexis optimisation example
The direction is clear: physics and optimisation can reduce device count while keeping the coverage objective in view.
Kenexis · published exampleDevices, with coverage increased
A useful signal for operators and EPCs: assurance and capital discipline can point in the same direction.
IChemE Hazards 28Target full-assessment processing
Compress weeks of manual cross-referencing and sampled cases into an engineer-reviewed, repeatable workflow.
Barrier Intelligence prototype targetSupported by The Utopia Studio.
Pilot with Barrier Intelligence
Proof, not paper.
Let us prove it on your plant.
Bring one mapping study, the supporting documents and the plant model. We will show where the design basis holds, where it does not, and what the smallest defensible fix looks like.
Assure3D is the first product on this engine. AssureLive, the planned live operational layer, follows.