Water operator knowledge search that answers from your SOPs and permit, and checks every dose against the permit before you set it
Procedures and permit conditions, answered on shift.
See one case, screen by screen ↓At night one operator runs the works. When a storm turns the river from 6 to 45 NTU in a few hours, the answer to “what dose now?” is spread across a 46-page coagulation SOP on a shared drive, the permit in the compliance team's library, the last jar test in a lab spreadsheet, the dosing pump's product strength in an O&M manual — and a laminated storm card taped by the dosing panel. Under-dose and floc carries over to the filters. Over-dose in soft water and the coagulation pH crashes, aluminium stays dissolved and passes the filters.
The person who knows how the river behaves is the senior operator on call, and she retires in December. Her knowledge is mostly not written down, and the storm card no longer matches the water: it was written in 2016 for raw alkalinity of about 55 mg/L, and since the 2021 upland blend it has been 30–38. In February 2023 a night operator followed it, raised alum from 20 to 60 mg/L in one step, and treated-water aluminium stayed above the permit for 5 h 20 min.
Estimated split for one coagulant dose change on a night shift, in minutes, by hand and with the solution.
Six specialist agents read the SOPs, the permit, the jar tests and the live plant; the operator sets every dose, the supervisor decides anything outside the permit.
Hollow Vale Water Treatment Works, Lumina Utilities: a 38 ML/d works on the River Kestle. Tuesday 13 October 2026, night shift. Daniel Osei is on his own in the control room; Brenda Holloway is on call. Here is what happens, screen by screen, in the working solution.
Sam Kirby hands over at 19:00: dry weather, alum 18 mg/L, all 6 filters in service, storm forecast after 23:00. The shift log fills from the plant historian and the weather feed: intake turbidity 6 NTU at 22:00, 19 at 00:40, the settled water alarm at 2.5 NTU at 01:50, then 45 NTU at the intake at 02:30. Alum is still on the dry-weather setpoint, 18 mg/L and 44 L/h on dosing pump DP-1.
Handover: “Dry weather. Alum 18 mg/L, all 6 filters in service, filter 4 due a wash. Storm forecast after 23:00.”
Daniel asks in his own words. Above the answer, a live strip from the plant historian, read-only: intake 45 NTU, raw pH 7.3, raw alkalinity 34 mg/L, 9.2 °C, 38.0 ML/d, settled water 2.9 NTU. The answer: raise alum to 38 mg/L, 93 L/h on DP-1. Every line of the working is cited — jar test JT-2026-0288 gave an optimum of 35 mg/L at 40 NTU; SOP-HV-021 Table 2 adds 0.6 mg/L per NTU above it; 38 g/m³ × 1,583 m³/h ÷ 645 g/L gives 93.3 L/h.
“Raise alum to 38 mg/L · Liquid aluminium sulphate on dosing pump DP-1, for raw water at 45 NTU and 38.0 ML/d”
The run tape shows each step. The question reader took the works, the stage and the chemical from the question and the live readings. The procedure finder found SOP-HV-021 rev 9 §5.3, §5.4, §5.6 and Table 2, the O&M manual §4.2 — and flagged storm card SC-HV-03 as superseded. The dose calculator split the change into two steps under the ramp rule, and the permit checker checked 9 limits, all inside. Brenda's approved note sits beside the jar-test curve.
Brenda Holloway, approved note N-102: “Above 40 NTU, step the alum up in tens and watch the coagulation pH, not the settled turbidity.”
At 38 mg/L every meter is green: treated-water aluminium 0.060 mg/L against the permit's 0.200 and a works target of 0.100, filtered turbidity 0.12 NTU against 0.30, coagulation pH 6.7 in the SOP window of 6.4–6.9, 15 mg/L of alkalinity left. Daniel slides the dose to the storm card's 55. Aluminium goes to 0.230 mg/L, coagulation pH to 6.2 and alkalinity left to 6.5, below the minimum of 10.
“Breaks permit HV-07 4.2: aluminium 0.230 mg/L. Also outside the SOP: alkalinity 6.5 mg/L.”
If Daniel tries to log the red dose, the solution stops him and offers to send it to Brenda to decide, with the permit check attached. The rule is locked on in Settings. The permit checker never softens or hides a limit, and the warning goes on the shift log: “Daniel tried 55 mg/L (the storm card): warned”.
“A dose outside the permit cannot be logged without the on-call supervisor. This rule is locked in Settings.”
Beside the answer, the incident matcher shows INC-2023-0417. Raw water rose to 52 NTU; the night operator followed the storm card and raised alum from 20 to 60 mg/L in one step. Coagulation pH fell to 6.1 and treated-water aluminium reached 0.24 mg/L. The root cause: the card was written for alkalinity of about 55 mg/L, and no jar test had been run in the storm band that month.
“Notified to the regulator · 2.1 ML run to waste · service reservoir flushed”
Daniel sets DP-1 to 69 L/h, 28 mg/L — step 1 of 2, no more than 10 mg/L in one step under SOP-HV-021 §5.3 — and signs the shift-log entry the knowledge keeper drafted with the reason. The agents never write to the plant; Daniel sets the pump. Because the night's change will pass 15 mg/L, Brenda is notified and acknowledges at 02:45.
Brenda Holloway, on call: “Good. Watch the pH, not the settled.”
After the warning, Brenda phones in a note, and the knowledge keeper drafts it in her words with the evidence. It waits with two more of hers for Dr. Priya Raman: filter 4 needs twenty minutes to waste after a backwash, and the Kestle's second flush comes six to eight hours after the first peak. Brenda retires in 66 days; 24 of her 38 notes are already in answers.
N-121: “Withdraw storm card SC-HV-03. It was right for the old water. Tonight it would have taken us over on aluminium again.”
SOP-HV-021 §6 asks for a confirm jar test within 2 hours of any change over 10 mg/L. Six jars are set up from the answer, 30 to 50 mg/L in steps of 4. Daniel imports the bench readings from turbidimeter TB-07 and pH meter PH-03: the lowest settled turbidity is 0.7 NTU at 38 mg/L. He saves JT-2026-0291 as the new reference for the 30–60 NTU band.
“The answer holds. Lowest settled turbidity at 38 mg/L (0.7 NTU), the dose set at 02:54.”
Helen owns permit HV-07, version 4. Her view lists the nine conditions a dose can move — aluminium, turbidity at each filter, outlet pH, free chlorine, contact time, trihalomethanes, iron, manganese, chlorate — with the permit limit, the works target, the 30-day maximum and tonight's prediction. It shows one warning tonight, at 55 mg/L, and no exceedances in the last 12 months.
“Permit text is quoted, never rewritten · owner: Helen Marsh”
The same storm night, seen by the five people who carry it — what it looked like before, and what it looks like now.
Six specialist agents read the SOPs, the permit, the jar tests and the live plant; the operator sets every dose, the supervisor decides anything outside the permit.
Reads the operator's question in their own words and pulls the latest readings from the plant historian — raw turbidity, pH, alkalinity, temperature, flow and the current setpoint — to work out the works, the process stage, the chemical and the dosing pump.
Searches the SOPs, O&M manuals, permits, standards, lab methods, incident reports and approved expert notes, and returns each passage with its document, revision and clause, preferring the current revision.
Starts from the most recent jar test in the raw-water band, applies the SOP correction table, converts to the pump rate with the product strength and the current flow, works the alkalinity used and the coagulation pH, and splits the change into steps under the ramp rule.
Predicts aluminium, filtered turbidity, coagulation pH, alkalinity, chlorine, contact time and by-products for the proposed dose and compares each with the permit limit and the works target: green inside both, amber outside a target, red outside the permit, with the clause.
Finds past water-quality incidents with conditions like tonight's and shows each beside the answer, with what happened, why and the lesson it left.
Drafts the shift-log entry from each answer, drafts notes in the senior operator's words from recorded sessions and calls, gathers the plant evidence and proposes SOP changes.
Decides any dose outside the permit. The agents propose; a named person decides.
The assistant answers from the works' SOPs, permit, O&M manuals, jar tests, incident reports and approved expert notes, and cites each one.
Why not the storm card's 55 mg/L?
Because the water has changed. Storm card SC-HV-03 was written in 2016 for raw alkalinity of about 55 mg/L; since the 2021 upland blend it has been 30–38, and tonight it is 34. At 55 mg/L alum uses 27.5 of it, leaving 6.5, below the minimum of 10. Coagulation pH falls to about 6.2 and aluminium stays dissolved: predicted 0.23 mg/L at the outlet, above the 0.200 in permit HV-07 condition 4.2. That is what happened on 14 Feb 2023.
Set up tonight's confirm jar test
Due within 2 hours of the change over 10 mg/L, so by 04:41. Set six jars around 38 mg/L in steps of 4: 30, 34, 38, 42, 46, 50. Use raw water from the intake tap, not the inlet channel, and the lab method's 2 / 15 / 20 minute timing.
Which limits can tonight's dose move?
Nine conditions of permit HV-07 are checked on every dose answer. For alum the four that move are aluminium (4.2), filtered turbidity (4.3), and two SOP limits that protect them: coagulation pH 6.4–6.9 and alkalinity ≥ 10. At 38 mg/L all are inside. The safe band tonight is 30–42 mg/L.
What has Brenda said about storm nights?
Nine of her 38 notes are about coagulation in storms. The one in tonight's answer: “step the alum up in tens and watch the coagulation pH, not the settled turbidity”. Two more wait for Dr. Raman: the Kestle's second flush six to eight hours after the first peak, and withdrawing the 2016 storm card. Brenda retires in 66 days.
13 screens from the working solution, on its sample data. Pick one to see it large.
The limits each dose is checked against, what warns and what stops, and the order sources are trusted in are settings, saved as a new version.
| Setting | Default | Choose from |
|---|---|---|
| Permit limits and works targets | Permit limits read from HV-07 and locked; works targets set inside them | Amber from the works target, red at the permit limit · coagulation pH: amber outside 6.4–6.9, red below 6.2 or above 7.1 · alkalinity left: amber below 10, red below 7 |
| A dose outside the permit cannot be logged without the supervisor | On | Always on |
| Warn when a prediction passes the works target | On | On · Off |
| Ramp | At most 10 mg/L per step, 10 minutes apart | On · Off |
| Confirm jar test | Within 2 hours of a change over 10 mg/L | On · Off |
| Show superseded storm cards in answers, marked superseded | On | On · Off |
| Agents may write setpoints to the plant | Off | Always off — agents read the plant, they never write to it |
| Sources, in the order they are trusted | Permit, SOPs, O&M manuals, standards and lab methods, approved expert notes, incident reports, storm cards | Storm cards never outrank an SOP |
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 works data · “target” = the design goal, measured in the pilot · “estimated” = our estimate (the minutes split for one dose change) · “typical” = published figures (AWWA, 2025: 21% of water utility staff can retire within 5 years). Regulatory context: aluminium 0.2 mg/L is the EU Drinking Water Directive 2020/2184 indicator value and the top of the US EPA secondary standard range (0.05–0.2 mg/L); 0.3 NTU filtered turbidity is from the US EPA surface water treatment rules. Lumina Utilities, Hollow Vale works, the River Kestle, permit HV-07 and the people, suppliers, documents and numbers named on this page are fictional — characters and sample data in the working solution.
It is a search for treatment operators and network technicians that answers from the works' own SOPs, O&M manuals, permit, jar tests and incident reports, with every line cited. When the answer is a chemical dose, it comes back as a worked calculation with a permit check beside it; when it is a procedure, it comes back as numbered steps with the SOP and permit text quoted.
The dose calculator starts from the latest jar test in the raw-water turbidity band, applies the SOP's correction table, converts the dose to a pump rate with the product strength from the O&M manual and the current flow, and works the alkalinity used and the predicted coagulation pH. In the sample case: 35 mg/L at 40 NTU, plus 0.6 mg/L per NTU, gives 38 mg/L and 93 L/h on DP-1, ramped in two steps.
Yes. Before an answer is shown, the permit checker predicts each limit the dose can move — in the sample works, nine — and marks it green, amber or red against the works target and the permit limit, with the clause. The operator can slide to any other dose and see which limit it would break. A dose outside the permit cannot be logged without the on-call supervisor.
The knowledge keeper drafts notes from recorded sessions and calls with senior operators, in their words, pinned to the SOP clause each one explains and backed by plant evidence. The process scientist approves or rejects each note; approved notes appear in answers with the author's name. Superseded storm cards and quick guides are flagged and can be sent for withdrawal.
No. The agents read the plant historian read-only and never write to the control system; that setting is locked off. The operator sets every dose on the pump and signs the shift-log entry the knowledge keeper drafts.
Yes. A network technician can ask, for example, when a repaired main can go back into supply, and get the mains-repair SOP as numbered steps with the flush volume worked out and the field chlorine and turbidity readings checked against each limit.
Yes. The operator sets every dose and signs every entry, the on-call supervisor decides anything outside the permit and is told of large changes, the process scientist owns the SOPs and approves expert notes, and the compliance manager owns the permit limits and works targets. Every answer, warning and signature is kept on the record.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your SOPs, O&M manuals, permit, jar-test records and incident reports — and it goes live once every quality gate has passed. We will walk you through it on your own works and a past storm night first.
Leak crews sent where the loss is certain, not where a meter is in doubt.
1-min demoWater Quality Compliance MonitoringExceedances caught the hour they happen, and low chlorine seen two days ahead.
1-min demoLeak Detection AnalyticsSend crews to real leaks, not false alarms.
1-min demoWe'll run Water Operator Knowledge Search on your own SOPs, permit and a past storm night.