A clogged filter forces the compressor to work harder. The compressor looks broken; the filter is the culprit. Getting it right instead of wrong, in this example documented in the ARRASTRE patent, was the difference between $375 and $12,000.
32:1 is the extreme example documented in the ARRASTRE patent — not a deployment case measured in industry · the typical range is 2:1 to 15:1
Heat, vibration, pressure or current propagate wear from one part to its neighbor. Most current systems treat each component as if it existed alone in the world.
Top-1 root-cause accuracy under noise, and up to 10× lower catastrophe rate, versus classic threshold/CBM. Internal benchmark on a synthetic corpus (31 cases × 7 conditions × 30 Monte Carlo runs) — illustrative, not an independent trial.
of industrial maintenance interventions today target the wrong component.
is the cost of every hour of unplanned downtime, depending on the asset.
They tell you something is wrong, not the cause. Every part is treated in isolation.
A single number points to the guilty component, with the contagion discounted.
Sometimes it's right, but you can't audit why. Inadmissible under IEC 61508 or 62443.
Deterministic and traceable. Every decision is explained step by step and can be certified.
Governs the machine as if it were still new. Chases an optimum that no longer exists.
Resets the target to the equipment's real state. Diagnoses and controls in the same loop.
Want to see it against specific tools — CMMS, vibration routes, predictive maintenance AI, digital twin? See the full comparison →
Only the physics of contagion changes. The structure of the method is identical across every sector.
Avoids shutting down a pumping station over a wrong diagnosis.
Keeps solar inverter performance up as it ages.
Prioritizes which network node needs attention first.
Anticipates fleet failure before service stops.
Tells apart a genuine fault from an induced one on the production line.
Avoids interrupting a production batch over a wrong diagnosis in a clean room.
Detects structural wear before catastrophic failure.
Anticipates failure in ore grinding before the process line stops.
Avoids replacing the compressor when the filter is the culprit.
Anticipates end-of-life for critical flight components.
Protects the availability of essential medical equipment.
Diagnoses axle and bearing before an in-service failure.
Reduces unplanned downtime on the cutting line.
Finds the true bottleneck in a full process.
Validates the method against NASA run-to-failure test benches (bearings, turbofan, battery).
Tests the diagnosis against reference academic benchmarks from MIT.
Cases marked REAL run on actual physical measurement, not simulation alone.
On a real operational incident, it isolates the CTRL-MODE root cause (a stuck setpoint) with the highest dominance in the corpus — matching the engineering record, and the second fault surfaces on its own.
Recovers the faults that are localizable and observable — including the condenser fault masked by the control loop — and declares, plainly, where it fails on the distributed ones.
The naive detector flags all 4 bearings on the shaft as suspect. Ln2's decoupling isolates B1, the documented root cause.
The root cause identified by the method matches the test bench's real tool-wear label.
With 4 well-instrumented components, the method ties with naive diagnosis (90.5% vs. 92.9%). We don't claim an advantage the data doesn't support.
That our hits and our misses fall exactly where our own premises predict doesn't weaken the comparison — it confirms our map of limits is real.
We apply the method to one of your critical assets. You bring the case; we bring the technology and the analysis.
Our patent portfolio, available for integration into your products or process lines.
For large industrial or defense projects. We work as an R&D partner, not a vendor.
Real-time monitoring, with a closed control loop, on the customer's own asset. Today it isn't a purchasable product: there's no onboarding backend, no ingestion from plant systems, and no pricing set. If you're interested, let's talk about how it fits your roadmap.
Tell us about it and we'll give you a straight answer.
Contact us