CMMS for Water and Wastewater Utilities

Most maintenance software is designed around a building or a plant: a bounded site, a hierarchy of rooms and equipment, technicians who walk between jobs. A water utility looks nothing like that. A handful of treatment works, dozens or hundreds of pumping stations, reservoirs and towers, valves and chambers scattered through the street network, and then thousands of kilometres of mains and sewers that are not equipment at all but continuous linear features nobody has ever fully surveyed. Choose a CMMS on a demo that showed you an HVAC hierarchy and you will meet the mismatch about four months into implementation, at the point where somebody asks how to raise a work order against a sewer.

The message up front: a water utility has two asset populations with genuinely different needs. Point assets (pumps, blowers, motors, valves, panels, instruments) are served well by a conventional CMMS. Linear assets (mains, sewers, rising mains, channels) are served badly by almost every CMMS on the market, and properly only by an EAM with real linear referencing and a GIS integration. Decide early which population dominates your work, and treat GIS as a co-equal system of record rather than a map layer bolted on at the end.

1. The dispersed estate and what it does to scheduling

A facilities team plans a day around a building. A water utility plans a day around a driving route. If a technician has six planned tasks across four pumping stations spread over sixty kilometres, travel is not an overhead on the work, it is most of the work. Any planning approach that ignores that produces schedules which look efficient in the system and are impossible on the road. Several consequences follow, and they should shape both your software selection and your planning practice:

• Geography has to be a first-class scheduling dimension. Zone, district metered area, treatment catchment: the planner needs to schedule along your operational geography. If the CMMS only sorts by priority and due date, planners rebuild the geography in a spreadsheet and you lose the audit trail.
• Align preventive intervals deliberately. Harmonising monthly, quarterly and annual routines so they land together on one station visit removes journeys outright. Pure planning discipline, and one of the highest-return changes available.
• Site visits should carry opportunistic work. The station's low-priority corrective backlog should surface on the mobile device while the technician is there. Treat the visit, not the task, as the unit of work.
• Offline mobile is non-negotiable. Chambers, remote reservoirs and rural stations have no reliable signal, and an app that needs connectivity to open a work order is replaced by paper within a month. Test offline in your actual worst location.
• Lone working changes the job, not just the paperwork. Check-in, check-out and escalation have to interact with the work order rather than sit beside it.

I would push hard on route-aware planning in any water utility selection, because it is the difference that shows up in cost per work order and is almost never demonstrated in a standard demo. Ask the vendor to plan a week for a two-technician crew across twelve dispersed stations using your own asset list, and watch what the planner has to do by hand.

2. Point assets and linear assets: two different problems

The single most useful distinction before you evaluate any water utility maintenance software is between point assets and linear assets. Point assets have a location, a nameplate, a serial number and a history belonging to that individual item. Linear assets are continuous: a 4.2 kilometre trunk main is one register record, but faults, repairs, surveys and renewals happen at positions along it rather than to the whole of it. That difference propagates into every part of the system.

Read that as a selection tool. If your workload is dominated by the Point rows, a good mid-market CMMS will serve you. If the Linear rows carry the money, the renewals programme and the regulatory exposure, you are in EAM territory whether you like the price or not. Most utilities are both, which is why the pragmatic answer is usually a CMMS or EAM for point assets tightly integrated with GIS for the network, rather than one product forced to be excellent at both.

3. Why most CMMS products model linear assets badly

This is where I would advise the most scepticism in a selection, because almost every vendor will answer "yes, we support linear assets" and mean something much weaker than what you need. The weak implementation is simple: the pipe is a record in the asset table with a length field, and work orders attach to the whole record. That is a point asset with a length attribute. It cannot tell you that there have been eleven bursts in one 300 metre stretch of a 4 kilometre main, because every burst was recorded against the same asset. The exact analysis you need to prioritise mains renewal is the analysis this model destroys.

A real linear implementation has four capabilities at minimum:

• Linear referencing. A location along the asset expressed as a measure: start at 0, end at 4,200, this repair at 1,860, either as points or as extents.
• Dynamic segmentation. Attributes that vary along the asset without splitting it: diameter changing at 2,100, material changing from cast iron to PE at 3,400, a railway crossing between 1,100 and 1,140. Overlaid ranges, not a fresh asset every time one attribute changes.
• Work and condition at a position. Work orders, CCTV defect grades, leak detections and repairs stored against a measure, so a heat map of failures along the asset falls out of the data naturally.
• Stable identity under change. When a 200 metre section is replaced, the parent route, network connectivity and history at that location must survive. Systems that renumber everything on a split lose the failure history that justified the renewal.

IBM Maximo has a dedicated linear assets capability; Hexagon EAM and Infor EAM both address linear and network assets; SAP PM can be made to work with linear referencing but that is configuration effort rather than out-of-the-box. At the lighter end, MaintainX, Limble, Fiix, UpKeep and eMaint are sensible products for point assets and were not built for linear referencing. That is a scope statement rather than a criticism, and pretending otherwise during selection is how utilities end up with a good CMMS and a network still managed in spreadsheets.

For the broader argument about when a CMMS stops being enough, see CMMS vs EAM: when you outgrow a CMMS and the enterprise asset management explainer. Linear assets are one of the clearest single reasons a utility crosses that line.

4. GIS as a co-equal system of record

In a water utility, GIS is not a reporting layer. It is the authoritative record of where the network is, what it is made of and how it connects, and operations, planning, capital delivery, leakage, hydraulic modelling and street works all run off it. Any CMMS or EAM arrives into an organisation where GIS already holds the network truth, and the implementation succeeds or fails on how honestly you draw the boundary between them.

The mistake I see most often is treating the integration as a data copy: nightly export from GIS, bulk load into the CMMS asset table, and within a year the two disagree about several thousand pipes and nobody can say which is right. The alternative is to decide, attribute by attribute, which system owns what, and integrate on identity rather than duplication.

The identity row decides whether the integration lives or dies. Asset identity has to be issued once, by one system, and treated as immutable by both: the GIS feature identifier travels into the maintenance system as a reference field that no interface, no data cleanse and no migration is allowed to regenerate. It sounds trivial and it is the single most common cause of a GIS and CMMS pair drifting apart, because a network is constantly split, merged, re-laid and re-surveyed, and every one of those operations is a chance for an identifier to change on one side only.

Three things belong in the integration from day one: a reconciliation report listing features present in one system and absent in the other, run weekly and read by a named person; an agreed rule for what happens when a pipe is split or merged, settled before go-live rather than discovered afterwards; and a rule that network assets are born in GIS and flow one way, never created directly in the CMMS. The wiring detail is in GIS and CMMS integration, and the register structure underneath it in asset hierarchy design.

5. SCADA and telemetry: alarms, runtime and the noise problem

A water utility is already instrumented. Pumping stations report to telemetry, treatment works run on SCADA, and there is usually a historian holding years of process data. That is a large head start on condition-based maintenance and an equally large opportunity to flood the maintenance system with rubbish. Three integrations are worth building, in this order of return:

• Runtime and starts counters into meter readings. The highest-value, lowest-risk integration available to a water utility, and frequently skipped. A duty pump that ran 6,000 hours and a standby that ran 400 should not receive the same annual service. Pull the counters from SCADA into the CMMS meter, let meter-based PM generate the work, and you stop over-servicing standby plant and under-servicing duty plant. Guard it with plausibility checks: counters reset when a panel is replaced, and an unguarded interface reads that as negative runtime or a jump of 50,000 hours.
• Selected alarms to corrective work orders. Not all alarms. A small, deliberately chosen set that genuinely always requires maintenance intervention: pump failed to start, high-high wet well level with pump unavailable, dosing pump fault, standby generator fault on test. Everything else stays in SCADA where the control room handles it.
• Process trends as condition data. Discharge pressure falling at constant speed, power draw rising for the same flow, blower amps climbing: condition signals you already collect. Reading them from the historian is cheaper than buying sensors and often gives earlier warning. Covered in the predictive maintenance practitioner's guide, with the plumbing in SCADA and CMMS integration and SCADA historian integration.

One architectural rule: keep control one-way. Maintenance systems read from SCADA. They do not write to it and they do not command plant. Blurring that boundary creates a safety and security problem far larger than any convenience it buys.

6. Regulatory and environmental compliance reporting

This is where water and wastewater maintenance stops resembling other industries. In a commercial building, a missed filter change is a comfort complaint. In a wastewater works, a failed pump at a station with no standby can become an unconsented discharge to a watercourse: a reportable environmental incident with statutory, financial and reputational consequences. The maintenance system is therefore part of the compliance evidence chain, not just an operational tool.

Regulatory regimes differ substantially by country and the details matter far too much to generalise. You may be answering to an economic regulator, an environmental regulator, a drinking water quality regulator, a public health authority, or several at once. As examples only: in England and Wales, discharge permits sit with the Environment Agency, drinking water quality with the Drinking Water Inspectorate and economic regulation with Ofwat; in the United States, Clean Water Act discharge permitting and the Safe Drinking Water Act run through the US Environmental Protection Agency and delegated state agencies; in the Gulf, obligations typically flow from a national environment authority plus a sector regulator. Treat all of that as illustrative and work to what your own regulator actually imposes on you, in writing, with your compliance team in the room.

What generalises is the shape of the requirement. Whatever the regime, a maintenance system that supports compliance has to do this:

• Flag compliance-critical assets and tasks explicitly. Not a note in a description field: a structured attribute marking the asset or routine as underpinning a permit condition, a quality obligation or a statutory test, so compliance work can be reported on, escalated differently, and never quietly deferred in a backlog cleanup.
• Record completion with evidence, to a defensible standard. Who, when, which asset, what result, readings and attachments held immutably. The test is whether the record stands up if a regulator asks two years later, after the technician has left.
• Make deferral an exception with a name on it. Compliance tasks should not defer through the same one-click route as a low-priority corrective. Deferral needs a reason, an approver and an expiry.
• Hold sampling and calibration traceability. Water quality monitoring depends on instruments being in calibration, so the record, the certificate and the traceability chain belong in the maintenance system against the instrument.
• Support incident linkage. An investigation will ask what maintenance history exists on the assets involved and whether anything was overdue. That query should take minutes, not a week of archaeology.

The uncomfortable corollary: once the maintenance system is part of the compliance evidence chain, its data quality becomes a regulatory exposure rather than an operational annoyance. An overdue statutory task visible in the system for eight months is worse, in an investigation, than one that was never scheduled, because the record shows the organisation knew. That is an argument for backlog discipline, not for recording less.

7. Confined space, chemicals and lone working: permit-controlled work

A far larger share of water and wastewater maintenance is legally permit-controlled than in most industries. Wet wells, dry wells, sewers, chambers, digesters, reservoirs and covered tanks are confined spaces with atmospheric hazards including oxygen deficiency, hydrogen sulphide and methane. Chlorine, sodium hypochlorite, ferric salts, polymers, lime and acids are handled daily. Work happens alone, at night, at roadside, in and around water. The practical implication is that the permit cannot be paper beside the work order. It has to be part of the work order lifecycle:

• Hazard flags on the asset, inherited by the work order. If the asset is a confined space, every work order raised against it should arrive knowing that, whether raised by a planner, an alarm interface or a technician in the field.
• Permit as a gate, not a document. The task should not be executable in the mobile app until the required permit is issued. That is the control that survives time pressure.
• Isolation and lock-out tied to the plant item. Electrical, mechanical and valve isolation, and the register of who holds which lock, linked to the asset rather than a book at the station.
• Gas testing captured in the task. Entry readings, retest intervals, and the detector's own calibration status, which is itself a maintained instrument in your register.
• Lone working check-in and escalation. Expected duration, a check-in, and automatic escalation if the check-out does not happen. A workflow requirement, not a policy document.
• Competency enforcement, live. The scheduler should not be able to assign confined space entry, chemical handling or electrical isolation to someone whose certification has lapsed. An annual audit is not the same control.

The integration pattern is in permit to work integration with a CMMS. The US Occupational Safety and Health Administration and the UK Health and Safety Executive both publish confined space and lone working guidance, but again, work to whichever regime has jurisdiction over your sites.

8. Critical spares for pumps, blowers and dosing plant

Water utilities have an unusual spares profile: much of the plant is large, slow-moving, long-lead and effectively irreplaceable at short notice. A submersible sewage pump for a specific wet well, a positive displacement blower, a large VSD, an actuator for a trunk main valve: weeks or months to procure, and the consequence of not having one is a flooding or discharge event rather than a production delay. The practices I would insist on:

• Stock the criticality, not the consumption. Conventional inventory maths, based on usage frequency, will tell you not to hold a spare for an item that has failed twice in fifteen years. For a pump whose failure means an unconsented discharge, that maths answers the wrong question. Say explicitly in the inventory policy that consequence of stockout drives critical-plant stocking, otherwise a well-meaning working capital review will delete exactly the spares you needed.
• Link spares to assets properly. A technician at 2am should see what fits and where it is. A bill of materials that exists only in a manual PDF in a cabinet is not a spares list.
• Manage rotables as rotables. Pumps and motors go out for refurbishment and come back. If each return is treated as a new part receipt you lose the unit's history and cannot tell which of four identical pumps has been rewound three times. Duty and standby rotation makes serialised tracking matter more here than in most sectors.
• Know the standby strategy per station. A standby pump that has not run under load for three years is not a standby.
• Watch the long-lead register actively. A short, reviewed list of items whose lead time exceeds the tolerable outage, with a named decision on each: hold stock, hold a framework contract, or accept the risk with the consequence documented.

The general discipline is in spare parts and MRO inventory in a CMMS, and the routines themselves in generator, pump and motor preventive maintenance.

9. Energy: usually the largest controllable operating cost

Pumping is the dominant energy load in most water utilities, and aeration the dominant load within wastewater treatment. That makes energy one of the largest controllable operating costs in the business, and pump and blower efficiency a maintenance concern rather than purely an engineering one.

The maintenance-relevant version of energy management is narrow and actionable:

• Track specific energy, not consumption. Kilowatt-hours per cubic metre pumped, or per kilogram of load removed, is the number that reveals degradation. Raw consumption moves with flow and tells you little.
• Trend pump efficiency and act on the drift. A worn impeller, a partially blocked suction or a failing bearing shows up as rising energy cost per unit pumped, often long before anything alarms, and a runtime-based service interval will not catch it.
• Treat aeration control as maintained equipment. Dissolved oxygen probes that have drifted out of calibration cause blowers to over-aerate continuously. The energy penalty from a neglected DO probe can dwarf the cost of calibrating it monthly.
• Check that duty rotation is actually rotating. Stations frequently end up running one pump almost continuously because a control setting was changed during a fault and never changed back. SCADA runtime meters make that visible immediately if anybody looks.
• Put the energy case in the renewal business case. The efficiency difference over the asset life is often a larger number than the maintenance saving, and it is the argument that gets capital approved.

A specific-energy trend per station, derived from data you already hold in SCADA and the CMMS, is a more persuasive artefact in a budget meeting than any backlog chart.

10. Capital maintenance planning across regulatory periods

Many water utilities plan and fund in multi-year regulatory periods, and the maintenance system has a role in that cycle which is easy to under-invest in. The renewals programme for mains, sewers, pumps and treatment plant is built from condition and failure evidence, and a large share of that evidence lives in the maintenance system. What that implies practically:

• Failure history has to be positioned and coded well enough to support renewal prioritisation. Burst counts per kilometre by material and laid decade, blockage frequency by sewer segment, repeat failures on one asset: all of that comes out of maintenance records, and only if a failure code and a location were captured at the time. The value of good failure coding is realised five years later in a capital submission.
• Condition survey results need a lifecycle. CCTV grades, structural inspections and leakage surveys have a validity period. A grade from 2014 is not evidence about 2026, so the system should know how old each grade is and survey effort should target the assets whose evidence has gone stale.
• Criticality scoring should be maintained, not set once. A sewer that ran under farmland when it was scored may now run under housing. Re-scoring on a cycle, using GIS-derived proximity and customer-impact inputs, keeps renewal priorities honest. The method is in asset criticality classification.
• Operating and capital expenditure need to be visible together. The argument for renewal is usually that reactive and consequential cost has exceeded what renewal would cost, and that comparison requires labour and materials booked to the right asset rather than a generic cost centre.
• Plan the maintenance consequences of the capital programme. New plant arrives with new PM requirements, spares, competencies and permits. The cheapest moment to get asset data right is while the contractor is still being paid: if your capital contracts do not specify the asset register deliverable, in your format and identifier scheme, verified before final payment, you will be paying your own staff to survey the plant you just built.

11. The honest part: data quality on a network nobody has fully surveyed

Everything above assumes you know what assets you have and where they are. In water and wastewater that assumption is weaker than in any other sector I have worked in, and any plan that does not confront it will fail quietly.

The reality of a mature network: parts were laid in the nineteenth century by companies that no longer exist, records were paper then microfilm then digitised with varying care, pipes have been abandoned in place, diverted, partially replaced and cross-connected, recorded depths and alignments can be out by metres, material and diameter are sometimes inferred rather than known, some sewers have never been surveyed along their whole length because doing so needed confined space entry or flow diversion nobody funded, and private or third-party sections sit inside networks with unclear ownership. Every one of those uncertainties propagates into the maintenance system as a record that looks exactly as authoritative on screen as the ones that are correct.

The approach I would recommend instead of a doomed completeness programme:

• Record provenance and confidence as attributes. Source (survey, as-built, historic drawing, inferred) and a confidence grade on every network asset. It costs almost nothing at load time and is impossible to retrofit.
• Improve data as a by-product of work, not as a project. Every excavation, CCTV survey and repair is a chance to confirm or correct material, diameter, depth and alignment. Make that a mandatory closure field on excavation work. It is the only data improvement programme that pays for itself, because the survey was already funded as part of the job.
• Target survey investment by criticality and confidence together. High-consequence assets with low-confidence data are the survey priority list. High-consequence with good data needs monitoring. Low-consequence with bad data can stay bad.
• Accept a differently-shaped register for the network than for plant, and say so in the business case. Bounded, visible treatment works plant can reach a high standard of completeness. The buried network cannot. A business case that assumes clean network data will overrun; one that budgets data improvement as ongoing operational activity survives.

There is a second data-quality problem entirely within your control and therefore less forgivable: failure coding on work already done. Blockage and burst records closed with a free-text comment are records you paid a van journey to collect and cannot analyse. A short, enforced failure code structure is the cheapest analytical investment available to a utility.

12. What to check during selection

Generic evaluation criteria will not separate the candidates. These will. Take the list into the demo and insist on seeing each item in the product, with your data where possible.

• Linear assets: burst history plotted by position along one main, two attributes dynamically segmented, then split the asset and prove the history survives.
• GIS: a live integration to your GIS platform, a named immutable identity field, and the reconciliation report.
• Meter-driven PM: runtime arriving from SCADA or a historian, with plausibility validation, generating a work order at a threshold.
• Alarm filtering: debouncing and de-duplication in configuration, not custom code.
• Offline mobile: airplane mode, a confined space task completed with readings and photographs, then sync.
• Permits as gates: try to start a permit-controlled task without an issued permit, and to assign it to lapsed competency. Both blocked.
• Compliance reporting: overdue compliance-critical tasks by site and permit condition, with the deferral trail.
• Route planning and rotables: a week planned for a dispersed crew by geography; a serialised pump out for refurbishment and back with its history intact.
• Data provenance: where source and confidence live on a network asset, and how a planner sees them.

A vendor who does eight of those convincingly is a serious candidate. One who does only the point-asset half may still be the right choice, provided you go in knowing the network will be managed in GIS and a separate planning tool and you have budgeted for it. What you must not do is discover the gap after signature. If you are earlier in the journey, start from the complete buyer's introduction to CMMS.

The idea to walk away with

A water utility is not a difficult CMMS implementation, it is two implementations sharing a work order. One is conventional and well served by the market: pumps, blowers, panels, instruments and structures at bounded sites, driven by runtime meters and SCADA condition data, controlled by permits, supported by critical spares. The other is poorly served: a buried linear network of uncertain extent, whose identity and geometry live in GIS, whose failure history only becomes useful when positioned along the asset, and whose condition evidence funds the capital programme.

Get the boundary right, decide which system owns which fact, and never let the identity key drift. Force the network into a plant-shaped hierarchy instead and you get a register that looks complete, analysis nobody can trust, and a renewals programme argued on instinct.

Final thoughts

The temptation in a water utility selection is to solve the visible problem, usually a plant maintenance backlog at the treatment works, and defer the network question. I would resist that. The network is where the asset value, the regulatory exposure and the capital spend sit, so network requirements should shape the architecture even if the first implementation phase is the treatment works. Choosing a product that can only ever do plant maintenance well, because plant maintenance was urgent, is a decision you live with for a decade.

And whatever you choose, budget for the honest thing: you do not fully know your network, you will not fully know it at the end of the project, and your systems should improve that knowledge steadily as a by-product of work you were already doing. Regulatory regimes, permit conditions and reporting formats vary by jurisdiction, so treat every example above as illustrative and work to what your own regulator actually requires of you.

Related reading: CMMS vs EAM: when you outgrow a CMMS, GIS and CMMS integration, SCADA and CMMS integration, Asset criticality classification, Permit to work integration, Spare parts and MRO inventory.