Meter event management that turns 18,214 AMI alarms into 23 actions, with every safety alarm in front of a person inside 5 minutes
18,214 alarms become 23 actions, and every safety alarm reaches a person in 5 minutes.
See one case, screen by screen ↓Every smart meter reports events: last gasp when the power fails, power restored, voltage sag, cover open, magnetic field, reverse energy, socket temperature high, register error, low battery, leak, backflow, comms failure. Each meter model sends its own codes for the same event. On an ordinary Tuesday Lumina Utilities' head-end raises 18,214 alarms before 14:20. Most need no action: a last gasp with a restore two seconds later, the same cover-open reported six times, 2,380 alarms from a planned outage the control room switched itself.
The few that matter sit in the same queue as the noise. A hot socket shows as three temperature alarms among eighteen thousand. Voltage on the house side of a meter disconnected for arrears means a bypass, or a generator back-feeding a line a crew will touch. The analyst sorts by code, exports to a spreadsheet and joins meter, premise, network and work-order data by hand, while the head-end, the outage system and the theft analytics each see one slice. Every visit sent to a meter in planned work, already restored or flagged by a firmware fault is a truck roll paid for nothing.
Estimated split of one metering-centre analyst's working week, in days, by hand and with the solution.
Six specialist agents read, filter, fold and route every meter alarm; a named person on each team confirms the action.
Lumina Utilities reads 1.24 million electricity meters and 680,000 water meters over its AMI network. On Tuesday 13 October, Leila Marsh runs the alarm queue in the metering centre. Here is her day, screen by screen, in the working solution.
Every alarm since midnight is one tick, in a row per family: power fail, power restored, voltage, tamper and cover, temperature, register and firmware, battery, comms, water flow. Under the stream, the fold: 4,860 duplicates, 2,408 blips under 3 seconds, 2,512 planned work and 1,522 information only taken out; 6,912 kept, in 23 incidents. Below that, the incidents sit in seven team lanes, with safety alarms at the top of each lane.
“18,214 alarms → 23 need action”
At 06:12:41 the meters on feeder F12 start sending last gasp; the last arrives at 06:13:19. Leila replays the window, and 1,860 last gasps fold into one incident: Feeder F12 outage, Kestrel Road primary, 1,860 meters. She sees why the incident exists before she opens it.
“1 incident · Feeder F12 outage · 1,860 meters”
The stream reader attached each meter's feeder, spur and transformer. The noise filter checked its 5 rules: no restore inside 3 seconds, and the F07 planned outage is a different feeder, so nothing was taken out. The correlator placed 1,860 meters on one feeder, three spurs and 12 transformers. The router proposed outage ticket OUT-26-10418 for the control room, P1. Graham Achebe confirmed it at 06:16.
Router: “1 min 40 s after the first last gasp; first customer call 06:19”
Switch S3-114 closes at 06:57 to restore spurs S1 and S2, and 1,412 meters come back. The 448 that fail the restore ping all hang beyond fuse FU-3317 on spur F12-S3, on transformers TX-F12-31, -32 and -33. Likely cause: fuse FU-3317 blown, 91 %; the alternative, recloser R-12 locked out, 7 %. Three customers on TX-F12-32 use life-support equipment at home.
Proposed: “Send a crew to fuse FU-3317” — Graham Achebe confirms · “Call the 3 life-support customers” — customer care
Meter E-40218837 at 22 Fennick Lane reports socket temperature 78 °C at 13:31, 84 °C at 13:52 and 92 °C at 14:11, against a limit of 75 °C, with voltage imbalance at the socket rising alongside. The electric oven and heat pump share the circuit; load 41 A. Likely cause: a loose jaw in the meter socket overheating under load, 88 %. The router raises it to field service, P1, on a 5-minute clock; Carla confirms the safety visit at 14:13.
EV-301 socket temperature high is a safety code: never taken out, never folded away.
Until the socket is replaced, a remote disconnect is on the table. The proposal goes to the field service supervisor with the alarm history attached, and the solution has already checked the account: no life-support or medical flag, no weather protection in force today. The supervisor decides.
“Agents never disconnect, reconnect or dispatch.”
At 118 Harlow Street, meter E-30551287 reports a cover open at 02:14, then six magnetic-field alarms to 02:40. The scorer adds what makes it a lead: use fell 61 % from 2 Sep, the night of an earlier cover-open, while the neighbours at 116 and 120 are flat, and there is no open work order — the last visit was the meter test in April 2024. Score 0.91 against the 0.70 case threshold. Sunita took it at 08:12.
“A tamper score is not proof. This solution never contacts the customer; Sunita Varga decides whether it becomes a case.”
11,302 of the day's alarms needed no action, each taken out by one of five rules with its approver and date — including Graham's switching order SO-2610-044 for the F07 planned outage, added at 08:40. The rule tuner proposes R-06 for reverse-energy blips on solar homes: 4,118 alarms in 90 days, none led to an action, and it would not hide 7 Linden Court, which has no registered solar. It goes to Amaal Bishara or nowhere.
Never taken out: EV-206 load-side voltage while disconnected · EV-301 socket temperature high · EV-704 backflow
Control room, field service, revenue protection, meter shop, water network, AMI network and customer care each get their incidents, their clock and the person on shift. The hand-off log shows who confirmed what: Carla Mendes confirmed the safety visit to 22 Fennick Lane; Ibrahim Sato sent an inspector to Orchard Mews Care Home for a backflow; Carla booked a two-person visit to 9 Pellow Close.
“Today, safety: 3 of 3 — hot socket 2 min · live load side 3 min · backflow 3 min”
577,227 alarms became 687 incidents in 30 days. 62 % of today's alarms were taken out by approved rules. 76 safety alarms reached a person in a median 2 min 40 s against a 5-minute clock, none over it. Outages were confirmed from the meters in a median 1 min 52 s, and 204 field visits were avoided.
“Safety alarms to a person · 2 min 40 s · median, 30 days · clock 5 min · none missed”
The same day's alarms, seen by the five people who act on them — what their queue looked like, and what it looks like now.
Six specialist agents read, filter, fold and route every meter alarm; a named person on each team confirms the action.
Reads the head-end alarm stream as it arrives and maps each manufacturer code to Lumina's event codes, with the meter's premise, feeder, spur, transformer, phase, water zone, collector, batch and firmware.
Takes out duplicates, momentary blips, information-only codes and alarms explained by planned switching or open meter work, and records the rule on every alarm it takes out.
Folds the alarms that are left into incidents — by feeder, fuse, transformer and phase; by premise; by batch and firmware; by collector — and names a likely cause with a confidence, the evidence for it and the best alternative.
Scores tamper and bypass incidents with 12 months of use, the neighbours, work history and past visits; at 0.70 or above it assembles the case evidence for revenue protection.
Sets the team, priority and clock for each incident and drafts the first action: the outage ticket, work order, safety visit or letter.
Each week, looks back 90 days for alarm patterns that never led to an action and proposes a rule with its exact condition, how many alarms it would take out and which open incidents it would hide.
Confirms; head of metering approves rules. The agents propose; a named person decides.
Ask in plain words, or say what to change. Every answer points to the alarm log, the network model or the meter's history behind it.
Why is part of F12 still off?
Fuse FU-3317 on spur F12-S3 has most likely blown (91 %). All 1,860 meters on F12 sent last gasp in 38 seconds. 1,412 came back when S3-114 closed at 06:57, and the 448 that failed the restore ping are all beyond FU-3317. Three of them use life-support equipment. The crew is waiting for Graham Achebe.
Which safety alarms are open?
Three, and each reached a person inside 5 minutes: the hot socket at 22 Fennick Lane, 92 °C at 14:11, visit ETA 14:46 with Carla Mendes; the live load side at 9 Pellow Close, 231 V on a disconnected meter, two-person visit 15:10; and the backflow at Orchard Mews Care Home, 0.41 m³ back toward the main, where Ibrahim Sato sent an inspector.
Is batch K19-B4 a recall?
Not on this evidence. Only the 38 meters on firmware 4.2.1 report checksum errors; none of the 4,162 on 4.3.0 do. That points at the firmware, not the hardware. The proposal is a sample test of 20 meters and moving the 38 to 4.3.0. Owen Pryce decides.
Stop raising reverse-energy blips on solar homes
In the last 90 days, 4,118 reverse-energy alarms under 0.02 kWh came from homes with registered solar, and none led to an action. The rule tuner has drafted R-06 for exactly that. It would not hide 7 Linden Court, which has no solar. Rules need the head of metering, so it is ready to send to Amaal Bishara.
13 screens from the working solution, on its sample data. Pick one to see it large.
How alarms are folded, what counts as noise, each team's clock and who decides are settings, chosen by the utility. A change to a rule goes to the head of metering first.
| Setting | Default | Choose from |
|---|---|---|
| Outage window | 120 seconds | Last gasps this close together count as one event |
| Outage threshold | 5 / 20 meters | Meters on one transformer / one feeder |
| Tamper window | 30 minutes | Cover, magnet and removal alarms grouped |
| Momentary blip | 3 seconds | Power restored within this time of the last gasp is taken out |
| Duplicate | 15 minutes | Same meter, same code, within this window |
| Never taken out | Locked | EV-206, EV-301, EV-704 — cannot be changed |
| Clocks to a person | Safety 5 minutes | Outage 10 minutes · water quality 15 minutes · revenue protection 2 working days |
| Tamper score to raise a case | 0.70 | Below it, the incident is monitored only; Sunita Varga decides when a tamper becomes a case |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample utility data · “target” = the design goal, measured in the live solution · “estimated” = our arithmetic, inputs shown: $0.6M = 2,400 avoided visits × $250, of about 3,000 alarm-driven visits a year at the demo utility's scale; the week's split is our estimate · “typical” = published figures: S&C Electric via Ventavid (truck roll $250–$500); DOE Smart Grid Investment Grant (AMI outage detection; tampering ≈ 1 % of revenue); Landis+Gyr (hot sockets ≈ 500–1,000 meter box fires a year in the US, vendor figure); US EPA, Using AMI for water quality surveillance (backflow and tamper alerts). Lumina Utilities and the people, streets, substations, meters and batches named on this page are fictional — characters and sample data in the working solution.
Meter event management is the work of reading the events and alarms smart meters send — last gasp, tamper, temperature, register errors, leaks, backflow — and turning the few that matter into actions. Meter Event Management does it with six agents: they read every head-end alarm, take out the noise by approved rules, fold the rest into incidents and route each to the team that acts. A named person confirms every action.
The stream reader maps each alarm to one code set and attaches the meter's network position, batch and firmware. The noise filter takes out duplicates, blips under 3 seconds, planned work and information-only codes. The event correlator folds what is left by feeder, fuse, transformer, premise, batch or collector and gives a likely cause; the router sets the team, priority and clock and drafts the first action.
It places every meter that sends last gasp on the network model and folds them to the smallest common device. In the sample day, 1,860 last gasps on feeder F12 became one incident, and an outage ticket was proposed 1 min 40 s after the first, before the first customer call. After switching, the 448 meters that failed the restore ping all sat beyond one fuse.
No. Socket temperature high, load-side voltage on a disconnected meter and backflow are locked: never taken out, never folded away, and each opens its own incident. Safety alarms run on a 5-minute clock to a person, 15 minutes for water quality; in the sample day all three reached a person inside 5 minutes.
The revenue protection scorer adds 12 months of use, the neighbours, work history and past visits to each tamper or bypass alarm and scores it; at 0.70 or above it assembles the case evidence. It never contacts or accuses a customer and never back-bills; the revenue protection investigator decides whether it becomes a case.
Yes. Each meter model's own codes are mapped to one dictionary with family, severity and default action; the sample utility runs two electricity models and one water model, plus codes raised by the head-end itself. A code it does not know goes to the analyst on shift.
Yes. The agents never dispatch, disconnect or reconnect. A named person on each team confirms every outage ticket, safety visit, case and letter, a remote disconnect needs the field service supervisor after the medical and weather protection checks, and every rule that takes alarms out needs the head of metering.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your event code lists, routing rules, team clocks and a sample of head-end alarms — and it goes live once every quality gate has passed. We will walk you through it on a day of your own alarms first.
Every held bill released inside its cycle, each estimate with its method and source.
1-min demoEnergy Theft DetectionAn estimated $8.4M a year recovered, with 3 in 5 site visits finding theft.
1-min demoSmart Water Meter Leak AlertsThe customer hears about a leak the next morning, with its cost a day, not on the bill.
1-min demoWe'll run Meter Event Management on a day of your own head-end alarms.