Water main break prediction that ranks pipe renewals by the breaks and interruptions each dirham avoids
Three times the breaks avoided for the same renewal budget, ranked by risk per dirham.
See one case, screen by screen ↓Renewal budgets are far below need. Lumina, the utility in the sample, renews about one per cent of its mains a year, and like most it picks them by age, at best by break count. But old cast iron in benign ground can last, while younger pipe in corrosive soil, on a main hit by pump-start pressure surges, fails again and again. An age list spends the renewal budget on pipes that were not going to break, and the burst that floods six shop fronts happens on a main that was never on it.
Likelihood and consequence are rarely scored together. A main feeding a hospital and a cul-de-sac main with the same break rate get the same rank. The break record drifts, because repairs are logged against the wrong pipe or the wrong material, so any model trained on it learns from errors. And the programme is not checked against road resurfacing or cable works, so streets are dug twice and reinstated twice.
Estimated split, in working days, for a capital planner building one year's water main renewal programme by hand from a GIS export and last year's list, and with the solution.
Six specialist agents match breaks, score every main and draft projects; the asset manager signs the programme.
Lumina Utilities runs 3,912 km of water mains. On Monday 12 October 2026 Leila Mansour, capital planner for water distribution, is building the 2027 renewal programme for the Capital Review Board on 4 November, with a budget of AED 64M. The 2026 programme was ranked by age. On Friday a cast-iron main burst on Old Souk Street. Here is what happens next, screen by screen, in the working solution.
The Break matcher reads 3 new repair work orders. The Likelihood modeller refits the 10 cohort curves — 1970s cast iron now has a median life to first break of 41 years, down from 42 — and scores every main for 2027 from age, soil, pressure swings, traffic and each pipe's own break history. 14,712 mains are rescored, and the Project drafter re-ranks 24 candidate projects.
“Back-test on 2025 breaks · Top 10% of mains held 58% of breaks · threshold 45% · passed, scores published”
Yusuf Rahman's crew report on WO-26-31877 reads: circumferential crack on DN200 grey iron, wall heavily graphitised, water through six shop fronts; valves V-HQ-112 and V-HQ-118 shut at 03:05, supply back at 10:00. 1,840 customers were without water for 7 h 20 min. The Break matcher places the break on Old Souk Street, picks WM-HQ-0417 over a DN100 service main because the crew measured 218 mm outside diameter, and gives the match 0.94.
“Conflict: the next pipe east, WM-HQ-0421, is recorded as ductile iron, but the crew's photo of the trench end shows the same cast iron” — sent to Daniel Okoro to decide
Leila opens the network map. Every main is coloured by its chance of a break in 2027, from under 2% to over 50%, with last month's breaks marked and the General Hospital, Dialysis Centre and Secondary School on the map. The Harbour Quarter cast iron is the darkest red. The list beside it reads WM-HQ-0417 on Old Souk Street at 82%, WM-HQ-0419 at 76% and WM-HQ-0433 on Pearl Divers Lane at 75%.
“The Old Souk Street cast iron burst on Friday. Draw around it to see what renewing it buys.”
Beside the map are the survival curves for each pipe cohort, material by decade laid: the share of pipe length never broken, against age. Leila clicks Cast iron, 1970s and its pipes light up on the map. 186 km of it is in the network, it breaks 38 times per 100 km a year, and only 28% of its length has never broken at 52 years.
“Cast iron, 1970s: 186 km · median life to first break 41 years · 38 breaks per 100 km a year”
WM-HQ-0417 is cast iron, DN200, 268 m, laid 1972. Its cohort baseline is multiplied by sabkha soil (×1.70), 14 pressure swings over 3 bar a day from the Harbour pumping station (×1.60), 5 breaks in 36 months (×4.21) buses and lorries on an arterial road (×1.25) and its age past the cohort median (×1.18): 1.72 expected breaks in 2027. If it breaks, 1,840 customers lose water between valves V-HQ-112 and V-HQ-118 and the old souk's shops and basements flood.
Consequence 4 of 5 · about AED 196,800 a break · risk AED 338,000 a year
With Trace pipes, Leila clicks junctions along Old Souk Street, Pearl Divers Lane and Customs House Road, and the route follows the mains. The panel counts up as she draws: 2.4 km, AED 3.6M, 11 breaks avoided a year and 68,600 customer-hours without water avoided. It also feeds Harbour Quarter General Hospital, weighted ×3.
3.08 breaks avoided per AED 1M — rank #1 of 31, against 1.16 for the 2027 programme and 0.34 for an oldest-first plan
Saved as RP-27-014, the Harbour Quarter cast-iron renewal costs AED 3,580,480 at the 2027 open-cut unit rates, for 9 pipes. The Coordination checker reads the City Roads Department resurfacing plan and Lumina's electricity capital programme. Customs House Road is resurfaced in May 2027, then closed to digging for five years: renewing first saves about AED 410,000 of reinstatement. On Pearl Divers Lane, the 11 kV cable replacement E-27-118 opens the street in March–April 2027, so the trench can be shared.
“Drafted a joint-trench request for the electricity team — not sent: waiting for Leila”
The renewal projects are ranked by breaks avoided per AED 1M, with a line where the AED 64M budget runs out. 31 projects fit: 44.1 km of mains, 74 breaks avoided a year and 377,000 customer-hours avoided. Planners drew 7 of the candidates; the Project drafter proposed 24, each a contiguous run that can be built valve to valve under one street-works permit.
Projects below the AED 64M line wait for 2028.
Oldest first, as in 2026, renews 42.6 km and avoids 22 breaks a year. Most breaks first avoids 46. Risk removed per dirham avoids 74, and 377,000 customer-hours without water against 86,000. Over ten years at the same budget, network breaks in 2036 fall to 117 against 181 for oldest first, and 357 with no renewal.
“3.4× the breaks avoided for the same AED 64M.”
The Programme writer drafts paper CRB-26-41 for the Capital Review Board — recommendation, why the ranking changed, the largest project, the top six and the risks — with 6 citations to the scenario run, the back-test and the work order, and checks every total against the project list. Mei Tan confirms the unit rates and phasing at 11:40; Leila sends it at 11:52. Karim signs with the meaning “approved for the Capital Review Board”.
“Agents draft and check. Only Karim can approve the programme, and only the board can release the budget.”
The same renewal programme, seen by the five people who build, check, repair and sign it — what their year looked like, and what it looks like now.
Six specialist agents match breaks, score every main and draft projects; the asset manager signs the programme.
Reads each repair work order, crew note and photo, matches the break to one main in the network record with a confidence, and flags material, size and location conflicts.
Refits survival curves for each pipe cohort every night from the break record, scores every main's chance of breaking next year from age, soil, pressure swings, traffic and its own history, and back-tests against last year's breaks.
Traces the valve-to-valve shut-off area for each main and counts the customers and critical customers — hospitals, dialysis, schools — then adds road class, flooding exposure and the all-in cost of a break.
Groups high-risk mains into contiguous projects that can be built valve to valve under one street-works permit, costs them from the unit rates and ranks them by breaks avoided per AED 1M.
Overlays each project route on the road resurfacing plan, Lumina's electricity capital programme and street-works moratoria, and flags clashes, savings and shared-trench chances.
Drafts the Capital Review Board paper from the scenario run and the project list — recommendation, why the ranking changed, largest projects, risks — and checks every total against the project list.
Signs; network engineer fixes records. The agents propose; a named person decides.
Ask in plain words, or say what to change. Answers point to the curve, the ground investigation, the logger or the work order behind them.
Why is Old Souk Street at the top of the map?
WM-HQ-0417 has an 82% chance of breaking again in 2027. It is 1972 cast iron, a cohort whose median life to first break is 41 years, laid in sabkha with 2.1% chloride and hit by 14 pressure swings a day from the Harbour pumping station. It has broken 5 times in 36 months, most recently on Friday: 1,840 customers without water for 7 h 20 min.
What does AED 64M buy if we rank by age, like 2026?
Ranked oldest first, AED 64M renews 42.6 km and avoids about 22 breaks a year and 86,000 customer-hours without water. Ranked by risk removed per dirham, the same money renews 44.1 km and avoids 74 breaks and 377,000 customer-hours. In the 2025 back-test the oldest 10% of mains held 21% of breaks.
Which mains feed Harbour Quarter General Hospital?
Two: WM-HQ-0452 and WM-HQ-0453 on Customs House Road, both 1971 cast iron, both in project RP-27-014. The hospital's storage covers 9 hours; on 3 October it ran on its tank for 5 h 30 min.
Weight hospitals and dialysis centres ×4 instead of ×3
Done in Settings: critical customer weight ×4, pending your save. On the next refresh 23 mains move up; the 2027 programme keeps its top 12 and swaps one Saltmarsh project for a Citadel Hill main that feeds the Dialysis Centre. The change is recorded with your name.
13 screens from the working solution, on its sample data. Pick one to see it large.
How consequence is weighed, what renewal costs and who signs are settings, decided by the utility.
| Setting | Default | Choose from |
|---|---|---|
| Critical customer weight | × 3.0 | hospitals, dialysis, schools — set by the asset manager; a change applies on save and is recorded with the name |
| Other consequence weights | Customers × 1.0 · arterial road × 2.0 · flooding × 1.5 · trunk main × 4.0 | customers per 1,000 shut off · arterial road closure and diversion · shops and basements below road level · trunk mains DN400 and above |
| Unit rates 2027, open cut (AED per metre) | DN150 1,280 · DN200 1,450 · DN300 1,760 | including local-road reinstatement · pipe bursting DN150–DN200 1,120 · + 170 on a new surface inside a moratorium |
| Model refresh | Every night, 05:30 | curves refitted and every main rescored from the break record |
| Back-test threshold | 45% | share of last year's breaks in the top 10% of mains — below it, new scores are not published |
| Planning year and forecast horizon | 2027 · 10 years | renewed pipe expected breaks 0.02 per pipe a year |
| Renewal budget and finance sign-off | AED 64,000,000 | scenarios at AED 40M · 64M · 90M · finance signs projects above AED 5,000,000 |
| Programme approver and record corrections | Karim Aziz · Daniel Okoro | Head of Water Asset Management signs; the network engineer decides record corrections |
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 network · “target” = the design goal, measured in the live solution · “estimated” = our arithmetic, inputs shown (AED 4.3M = 52 more breaks avoided a year — 74 against 22 on the AED 64M plan — × AED 82,000 all-in per break, the demo utility's figure for repair, reinstatement, damage claims and lost water; the working-days split is our estimate) · “typical” = published figures (Utah State University, Water Main Break Rates in the USA and Canada, 2023: 11.1 breaks per 100 miles a year · ASCE Civil Engineering, 2024: a $452B problem · city repair figures: $5K–$10K a break). Lumina Utilities and the people, streets, districts, sites and records named on this page are fictional — characters and sample data in the working solution.
Software that predicts which water mains will break next and ranks pipe renewals by the risk they remove. Water Main Break Prediction scores every main's likelihood and consequence of failure, lets the capital planner draw renewal projects on the network map, and ranks the programme by breaks avoided per dirham. The Head of Water Asset Management signs it.
Each night the Likelihood modeller refits survival curves for each pipe cohort — material by decade laid — from the break record, then adjusts each main for its soil, pressure swings, traffic and own break history. The Consequence scorer traces the valve-to-valve shut-off area and counts customers, critical customers, road class and flooding. Likelihood times consequence, per dirham of renewal, sets the rank.
Yes. In the sample 2025 back-test, 144 of 248 breaks were on the top 10% of mains by predicted risk, and 31 of those 144 were on mains with no break before — found from cohort, soil and pressure alone. An age or break-count list would have missed them.
It ranks by risk removed per dirham rather than by age. On the sample AED 64M budget, oldest first avoids 22 breaks a year and 86,000 customer-hours without water; risk removed per dirham avoids 74 breaks and 377,000 customer-hours. The scenario screen compares both, and most breaks first, at three budgets.
The Break matcher reads every repair work order, crew note and photo and matches the break to one main with a confidence. Where the crew's evidence contradicts the network record — cast iron recorded as ductile iron, for example — it proposes a correction; the network engineer accepts or rejects it.
Yes. The Coordination checker overlays each project on the road resurfacing plan, the electricity capital programme and street-works moratoria. In the sample case it found a resurfacing in May 2027 — renewing first saves about AED 410,000 of reinstatement — and an 11 kV cable trench to share. It drafts the request; a person sends it.
Yes. Agents read, score, draft and check; they never commit budget, sign, edit the network record or contact an outside authority. Consequence weights belong to the asset manager, record corrections to the network engineer, and the programme needs the Head of Water Asset Management's signature, with finance on projects over AED 5M.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your network record, repair history, unit rates and approval rules — and it goes live once every quality gate has passed. We will walk you through it on your own mains 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 Main Break Prediction on your own network record and break history.