Condition Assessment System
Acoustic-emission based severity assessment for reciprocating gas compressor valves, so maintenance can be scheduled against condition rather than calendar.
The problem
Reciprocating compressor valves fail progressively, and the failure is largely invisible from outside the machine. Maintenance is therefore scheduled on a calendar — replacing components that are still serviceable, while occasionally missing one that is degrading quickly.
System architecture
Capturing emission from the valve at a sampling rate that preserves the event.
Filtering and transformation to isolate the components carrying condition information.
Deriving the measurable descriptors that correlate with degradation.
Classifying severity against a labelled reference set, with confidence reported.
A prediction only has value once it reaches the maintenance plan.
Capabilities
Technical overview
This is an active research and engineering programme rather than a published product. The approach is deliberately narrow — one machine type, one failure mode, one signal — because that is the scope in which industrial condition models are reliable. No public demonstration is available yet.
Active engineering programme. No public demonstration is published yet.
Have a complex process, digital product or industrial challenge? Let’s discuss the system behind it.
