Sense
Measured at the node, summarised in the cloud
Ground vibration from digging, blasting and pile driving, measured by sensor nodes we build and processed on the Field Agent beside them. Sending raw high-rate acceleration to a cloud is a bandwidth bill, not an architecture.
Where the work happens
Three stages, and the interesting one is in the middle.
Measure, do not think
Tri-axial acceleration sampled continuously and streamed over wired Ethernet with power-over-Ethernet. The node’s job is to deliver every sample and keep accurate time. Nothing else.
Signal processing on site
The full waveform lands in the Field Agent’s local time-series store , about 15 days, heavily compressed, and the metrics below are computed there. Events get a pre-trigger and post-trigger snapshot written to object storage.
Summaries, on the asset
Throttled summaries arrive, auto-create the node and a resource per metric in the asset graph, and feed the same alarm rules and activity feed as everything else. Vibration is not a side app.
What the Field Agent computes
Data Explorer
On the asset, not beside it
Auto-provisioned into the graph
A node that starts reporting creates itself and its per-metric channels under the right parent. There is no separate inventory of vibration sensors to keep in step.
The same alarm machinery
Thresholds go through the full alarm lifecycle, raised, acknowledged, cleared, with flood protection, rather than a bespoke notifier that emails somebody.
Classification, stated as heuristic
Events are labelled blast, machinery, traffic or wind by a rule-based classifier using onset shape and spectral content. It is deterministic and inspectable. No model, no training set, no claim to be one.
Snapshots you can go back to
Around a trigger, the surrounding waveform is written out whole, so an argument three weeks later can be settled by looking at the event rather than at a summary of it.
The same node on a machine
A vibration sensor does not care whether the vibration came from a blast or a bearing. The same node bolted to a compressor, a pump or a gearbox measures the same thing, on the same wiring, into the same asset graph, and it will tell you that a machine’s signature has changed and when it changed. For a lot of rotating equipment that alone is worth having, because the alternative is a person with a handheld meter once a quarter.
Be clear about where that stops today. Telling you the signature changed is not the same as telling you which bearing race is spalling. The rotating-machinery analysis that does that, envelope demodulation for bearing defect frequencies, severity zoning toISO 10816, per-asset learned baselines and degradation trending, is specified and scheduled and is not built. If condition monitoring is your first requirement rather than your third, tell us and we will be straight with you about timing rather than selling you the roadmap.
What this is not
This is monitoring. It is not a legal record.
If you need defensible blast records for a regulator or a neighbour dispute, use an instrument certified for it. Three specific reasons, stated plainly:
- The compliance-grade event tier is designed but has nothing writing to it yet.
- Peak particle velocity is computed with a simple baseline removal and integration drift is not corrected, so the figure reads high. It is useful as a trend and as a trigger; it is not a certified measurement.
- The node has no local storage by design. If the network drops during an event, that event is gone. The Field Agent can only keep what reached it.
Standards for blast monitoring are what this is being built toward, not what it currently satisfies. Anyone telling you otherwise about a system at this stage is selling.
Useful today for trend and trigger
Continuous awareness of what your site is doing to the ground around it, and an automatic work order when something crosses a line you set.