Ragging Detection
PumpIQ · reading a pump that is choking on rags, before the station tells anyone
A partially blocked impeller is pulled two ways at once, which is why current-based clog detection disappoints in the field.
The duty point shifts back along the curve toward shutoff, which on a falling power characteristic reduces shaft power. At the same time the obstruction destroys hydraulic efficiency and adds loss, which raises it. Which effect wins depends on impeller geometry, on where the mass has lodged, and on how force main head splits between static lift and friction. The change is small, it sits inside normal cycle-to-cycle variation, and its direction follows the pump geometry as much as the fault.
And stall is the failure itself.
Shaft-speed vibration, vane-pass energy, ultrasound, and volume moved per minute of run time.
Link 04 of the failure chain is duty degradation: effective capacity falls, so the same inflow needs longer runs and more starts per hour. Those are real run hours — not phantom ones — and they count against the 10.0 hr ceiling exactly like sanitary load. But the station does not need capacity. It needs a two-hour de-rag.
Infiltration, recirculation, stuck float, held run signal
Remedy. Correct the measurement or the fault. Hours were never real.
Fiber mass on the impeller, lost hydraulic efficiency
Remedy. Field de-rag. Hours are real but fully recoverable.
Build-out and connected demand in the basin
Remedy. Capacity. This is the only category a capital project addresses.
Reading all three as one number is how a station ends up in moratorium with a Remedial Action Plan written against the wrong problem.
NAPOT ComplianceRead in order, these separate a station that needs a two-hour de-rag from one that needs a crane, a crew and a rebuilt pump — and they give the crew a reason to believe the difference.
Fiber wraps the leading edge of one vane. The rotating assembly is heavier on one side and the pressure field around it is asymmetric, so a steady radial thrust acts on the shaft.
1× shaft speed amplitude rises; phase angle shifts
The mass bridges the gap between vane tip and cutwater. Each vane passage now strikes that restriction as it sweeps past.
Vane-pass amplitude rises; ±1× sidebands appear
The pump is forced left of its best efficiency point. Suction and internal recirculation set up, and vapor bubbles collapse against the vane surfaces.
Broadband ultrasound rises; random high-frequency energy
Effective capacity falls, so the same inflow needs longer runs and more starts per hour. Windings, cable entries and starters cycle harder than they were sized for.
Run time per cycle up; starts per hour up; ΔT up
Sustained radial thrust deflects the shaft. The lower bearing carries a load meant for intermittent duty, now applied continuously, and the mechanical seal faces open and close against a wobbling shaft.
Bearing defect frequencies in the envelope bands; seal-chamber moisture rises
Vibration reports that something changed. Capacity reports how much the station lost, and capacity is the number operations and finance both budget against. The wet well supplies it for free, provided the level signal and the pump-run state are read together.
A — wet-well plan area from the as-built. h — level. dh/dt — slope over the straight portion of each segment, discarding the transient at start and stop. Baseline — the first 14–30 days after a confirmed clean condition, held separately for P1 and P2 and for each level band.
Constraint. The expression assumes a prismatic well. Benched or irregular wells need a stage-volume curve built from the as-built drawings before the index means anything.
Three things break this measurement — and the control for each
The expression takes inflow during the run to resemble inflow just before it, which holds in dry weather. Storms move it, and wet weather mobilizes settled debris.
Control. Compute the index on dry-weather cycles only, using fill segments on both sides of the run. Wet-weather cycles carry condition data with no capacity number attached.
Where stations share a force main, a neighbouring station starting raises the head ours works against, and capacity falls for reasons unrelated to condition.
Control. Take a pressure transducer on the discharge and normalize against it, or gate on concurrency where the platform watches both stations.
Working volume is sized against minimum motor cycle time, so a basin pump-down runs in minutes and the drawdown slope lives inside that window.
Control. Sample at seconds while the run signal is active, and report segment slopes plus a cycle summary so radio budget and battery life both survive.
The best control is the other pump. Two duty-alternating pumps in one well see the same inflow, geometry and discharge head within minutes of each other. An index that falls on one pump while the other holds is close to unarguable. Both falling together points at the force main — a finding worth having in its own right.
Every rule is evaluated per pump cycle, gated by the run signal and the level band, and compared against that station's own learned baseline. Every rule carries a confirmation window of four consecutive cycles of the same pump — the shortest window that survives a stray plastic bag passing through. The window is counted in cycles on purpose, so a busy basin station and a quiet one are held to the same evidential standard.
| Rule | Signal | Condition | Reads as | Response |
|---|---|---|---|---|
| RAG-01 | 1× shaft speed amplitude | Sustained rise above learned baseline in the same level band, four consecutive cycles | Mass on the vane, hydraulic imbalance | Watch |
| RAG-02 | Vane-pass amplitude and ±1× sidebands | Vane-pass energy rising with sideband growth | Obstruction in the flow path | Act |
| RAG-03 | Ultrasound / high-frequency RMS | Broadband rise with vane-pass unchanged | Recirculation or cavitation — check suction and valve position before blaming rag | Investigate |
| RAG-04 Discriminating | Capacity index vs motor power | Capacity falling while power is flat or falling | Hydraulic blockage confirmed | Act |
| RAG-05 | Run time per cycle, starts per hour | Both rising at equal inflow | Duty degradation — thermal and starter stress accruing, and NAPOT run hours climbing | Act |
| RAG-06 | kWh per thousand gallons | Sustained rise against baseline | The cost of the condition, in the units the utility budgets in | Report |
| RAG-07 | Demodulated envelope bands | Bearing defect frequencies emerging | Secondary damage — past the point a field clean will fix | Escalate |
| RAG-08 | Housing temperature vs ambient | ΔT rising across cycles, conditioned on the wet-well level band | Thermal stress on windings and cable entry | Escalate |
| RAG-09 | Post-intervention signature | Signature still above baseline two cycles after the de-rag | Damage retained — schedule the pull | Escalate |
Discrimination is what keeps call-outs productive: a crew that finds the fault it was sent for keeps using the system. The pattern across signals names the cause; the magnitude of any one of them sizes it.
| Cause | 1× | Vane pass | Ultrasound | Capacity | Power | Distinguishing tell |
|---|---|---|---|---|---|---|
| Ragging | up | up+ sidebands | up | down | flat | Onset in days, and it clears completely after a de-rag |
| Impeller and wear-ring wear | flat | flat | slight | down | slight | Months of drift, no step change, no recovery after cleaning |
| Bearing degradation | varies | flat | up | flat | flat | Defect frequencies in the envelope bands with capacity intact |
| Low-level snoring / vortexing | erratic | flat | up | downlow band only | flat | Tied strictly to the bottom level band, absent at higher levels |
| Closed valve or air lock | flat | flat | flat | ~zero | flatshutoff value | Well holds level while the pump reports running |
| Misalignment or looseness (dry pit) | up | flat | flat | flat | flat | Axial component and harmonics rise, hydraulics untouched |
Battery powered, joins over LoRaWAN, computes spectra and envelope bands inside the enclosure so the radio carries indicators and selected bands while the full waveform stays at the edge. Vibration, ultrasound and surface temperature from one enclosure — which is what lets a single device separate a hydraulic obstruction from a bearing fault at the same station.
Learns the normal vibration, temperature and acoustic behaviour of a machine and reports how far it has drifted. The drift number ranks the fleet and points the diagnostic device at the asset that moved — at a price per asset that brings a second and third pump, a blower and a screen under watch as well.
The sensor mounts on the dry side of the load path — discharge elbow above the water line, guide-rail top bracket, or nearest pipe support — and that path shapes what arrives. Vane-pass energy, hydraulic pulsation and cavitation travel in the water column and pipe wall and arrive strongly. Shaft imbalance and bearing defect frequencies are mechanical, travel through the base elbow and its gasketed joints, and arrive attenuated by an amount specific to each station's pipework.
Bearing diagnosis in the strict sense belongs to dry-pit units, where the sensor sits on the bearing housing, and to submersibles carrying manufacturer-embedded monitoring, whose output we read into the same station view.
The pump chamber is wet, hydrogen-sulfide bearing and in many stations a classified area. Device selection follows the North American hazardous-location listing — a separate approval from the European ATEX marking most datasheets lead with — and enclosure and antenna materials are chosen for sulfide exposure.
Signal needs a path past a cast-iron hatch, so the device sits in the valve vault or dry chamber, or its antenna is brought out. Decided per station at site survey and recorded on the commissioning sheet.
Single-vane and two-vane non-clog impellers are both in common wastewater service. On a single-vane impeller the vane-pass frequency IS the shaft frequency — the two sit on top of each other — so discrimination moves to the harmonic structure above the fundamental, to the ultrasound band, and above all to the capacity index. Vane count is recorded at commissioning.
Detection pays when the loop closes: the condition becomes a work order, the work order becomes an action, and the action is verified by the same instrument that raised the alarm.
The sensor processes the cycle on board and sends indicators and selected spectra over LoRaWAN. The station network, the SCADA system and the control panel stay exactly as they are.
PumpIQ holds the station: level bands, lead and lag identity, run history, capacity index and condition, per pump, on one screen.
The rules set a stage — watch, act, escalate — with the evidence attached, so the notification says what was seen and why it matters.
The stage becomes a work order in the system the department already runs on, carrying the expected intervention: field de-rag, or pull.
Two cycles after the work, the signature either returns to baseline or holds high. That answer is the proof of value, and the input to the next decision.
The escalation runs across three stages, so it takes three rows — each with its own priority and crew response.
| Rule | Sensed | Trigger | Priority | EAM action |
|---|---|---|---|---|
| Mechanical degradation — onset | Vane-pass vibration energy and ultrasound band, read against the pump's own baseline in the same wet-well level band | Vane-pass amplitude and sideband growth sustained across four consecutive cycles of the same pump | 3 — Planned | Planned. Add a field de-rag to the next scheduled station visitation. |
| Mechanical degradation — confirmed | Effective pumping capacity, derived from wet-well drawdown rate, held against motor power | Capacity falling while power stays flat or falls, on dry-weather cycles, across four consecutive cycles | 2 — Urgent | Urgent. Dispatch to de-rag before duty run time forces the lag pump into continuous service. |
| Mechanical degradation — retained | Post-intervention signature against pre-fault baseline | Signature still above baseline two cycles after a completed de-rag | 2 — Urgent | Urgent. Schedule pull and bench inspection — something beyond the rag is holding the signature up. |
Two corrections to the adjacent rows
Both change how a rule fires, so they are carried as rule configuration rather than as commentary.
Cavitation is broadband and high frequency, best seen in the ultrasound band — and that is what makes it separable from imbalance and looseness.
A low-frequency energy jump is mechanical: impact, looseness, a lost balance weight, a bearing gone hard. The row carried the right priority; the physics belonged in the ultrasound band.
The thermal threshold is conditioned on the wet-well level band.
A submersible motor is cooled by the liquid around it. Run the well down low and skin temperature climbs with the loss of that cooling; keep it high and the liquid carries part of the winding heat away. Gating on level is what keeps those urgent call-outs accurate.
Reports load. The change from a partial blockage is small, sits inside normal cycle-to-cycle variation, and its direction depends on impeller geometry and force main head — so current turns decisive at stall, which is the failure itself.
Reports basin behaviour with a lag of days. It attributes to the basin, so a capacity loss at one pump and a genuine change in inflow arrive as the same number.
Reports a full spectrum, monthly at best, with a station visit and often a confined-space permit. Ragging develops inside that interval.
Reports continuously and accurately. Cost per station — conduit, panel space, station electrical work and permits — keeps this class of monitoring on a handful of assets.
Reports what the historian holds, which is typically control state and flow. High-frequency mechanical signal enters that record once the instruments above are in place.
A detection method is worth what its evidence is worth, so the pilot is designed around the event rate. Ragging is episodic: a station that clogs three or four times a year gives a six-month window perhaps two events, and a defensible result needs more than that.
Station selection starts from the department's own work-order history — the stations carrying the highest count of clog-related visits over the last two or three seasons — because those are where a six-month window holds enough events, and where the operating saving shows up in the department's own records.
Ground truth comes from the work orders. Each intervention is time-stamped, the condition found is recorded by the crew, and the signature either side is held against that record.
The four numbers that come out
Between the first confirmed stage and the intervention. Every lead time traces back to a work order.
Share of alerts where the crew found what the system said they would find.
Clog events that occurred with no prior alert. This measures whether the method can be trusted, and it belongs in the report beside the successes.
Movement across the stations under watch.
This watches the machines — pumps, motors and the hydraulics they move — which is a separate measurement from level-based blockage detection in the collection system. Where the department already runs that, we read it alongside.
The rules are tuned to a small number of actionable notifications per station per month, which is what keeps a crew acting on them. Everything below that threshold stays visible on the station view and quiet.
Detection thresholds are commissioning parameters, set per station during the baseline period against that station's own clean-operation record.
Senzary LLC · WaterIQ and PumpIQ on the IoT-LogIQ platform · prepared for Miami-Dade Water and Sewer Department. Detection thresholds are commissioning parameters, tuned against each station's own clean-operation record. Sensor capabilities are as published by the respective manufacturers and confirmed against the datasheet for the specific variant. NAPOT ceiling per Miami-Dade Code Sec. 24-42.3.