Product · Hold
Minutes held. With a bound.
Per segment: how many minutes the estimate stays inside tolerance after the data stops, with a 90 % bound.
1
Minutes each segment's estimate holds, with a bound
2
Tolerance: twice the commissioning noise of the nearest gauge
3
Says “blind: cannot say” when it cannot
Hands on
Three steps.
2
Watch the minutes
Now showingA radio site Its gauges go quiet together. Line changed about 1.2× an hour before the outage.
Simulated pipeline
Holdover says
km 0–20: thin
About 0 minutes left at 90 %, given how often this line gets changed.
Smallest leak still seen at this point: 1 % of flow.
How it turned out
Predicted to hold at 90 %: 0 min (worst segment).
See the estimate along the line
Green: the estimate and its 90 % band while the line stays steady. Pink shading: blind segments, where an unseen change would not show. Show what really happened to compare.
3
One clear plan line
Now
Minute 0: km 0–20 thin. Hold flow; manual operation at minute 0 unless data returns.
Nothing runs on its ownA person decidesPlan is a table in this version
What it needs
- Historian export with dropout and freeze flags
- Wavepin's line model for the segment
- One month of history
What it is not
- Not a backup SCADA
- No regulator has accepted a computed blind time
Proven, in simulation
- Leaks starting inside an outage caught 91 % of the time at 6 % false alarms; a learned model 85 %, today's display 71 %, Wavepin's engine with validity masks 67 % (simulation)
- When it says the estimate still holds, it is wrong 3.7 % of the time; hold-last-value 22.5 %
- The estimate held 71 minutes (median) through an outage; the frozen display 63
- The 90 % bound on minutes held covered 85 % of outcomes
- Correlated loss shortens holding to 0.91× random loss at the same rate (interval 0.83–0.98), beyond the interval, narrowly
Not proven
- When an estimate does fail, the minutes forecast is off by 42 minutes (median); a five-minute open-loop rule is closer there
- The correlated-loss effect is small (9 %); another line or map may erase it
- No real outage log used
- No regulator has accepted a computed blind time
- Satellite latency modelled, not measured
- Two sites failing at once may read as one
Works with
See Hold on your asset.
Tournament run checked in. Simulation only. Not yet on a real asset. A conversation first; a data pilot when the engine has earned it.