Loading...

Arizona Utilities Turn to Sensors and Analytics for Water Leak Detection: Lessons From Mesa

Smart Utilities

Published by IOT Arizona Research & Editorial Team

Arizona Utilities Turn to Sensors and Analytics for Water Leak Detection: Lessons From Mesa

When Mesa's water resources planning adviser described the city's approach to water data in a March 2026 interview, two details stood out. First, Mesa reports citywide water loss under 6 percent, compared with about 10 percent across Arizona and about 16 percent nationally. Second, the city deliberately does not use artificial intelligence in its SCADA systems, because reliability is critical for treatment and service. Its control systems are kept internal rather than cloud-based, and at least two staff members monitor operations at all times.

That combination, strong results with a cautious approach to technology, offers a useful lesson as Arizona utilities face Colorado River cuts starting January 1, 2027. Leak detection is often sold as a story of advanced sensors and AI. In practice, the utilities that lose the least water tend to combine sensors, analytics and disciplined field work, and they are careful about where they trust automation. This article looks at how that works.

Why Losses Matter More Now

Every gallon a utility treats and pumps but never bills carries a cost: the water itself, the energy to move it and the chemicals to treat it. Under tighter supplies, there is a second cost. Lost water must be replaced from groundwater, stored water or expensive new sources. If Arizona's utilities as a group moved from around 10 percent loss toward Mesa's level, the savings would amount to a meaningful new supply without drilling a well or building a plant.

The Detection Toolkit

ToolWhat It FindsStrengthsLimitations
Water auditsTotal real and apparent losses by system or zoneStandard method; guides where to lookAnnual snapshot; needs good metering
District metered areasWhich zones lose water, often through night flow analysisNarrows the search areaRequires valving and flow meters at zone boundaries
Pressure monitoringPressure spikes and transients that stress pipesPrevents breaks; supports pressure managementDoes not locate existing leaks directly
Acoustic sensors and correlatorsSound of leaks on pipesPinpoints leaks for repair crewsWorks best on metal pipe; plastic pipe is harder
Smart customer metersCustomer-side leaks and unbilled useCatches leaks past the meterDoes not see leaks in mains
Satellite and aerial analysisAreas with unusual soil moistureCovers large areas quicklyNeeds field confirmation
Risk-based asset analyticsPipes most likely to failPrioritizes inspection and replacementDepends on good asset records

No single tool finds everything. Effective programs combine several, starting with an audit to know where to look and ending with field crews who confirm and repair.

Mesa's Approach, Piece by Piece

  • SCADA with control logic and advanced sensors across treatment plants and distribution, adjusted in real time by staff.
  • Smart customer meters that generate continuous-use notifications, more than 300 a month during the transition, with the city working with property owners to fix leaks.
  • Risk-based pipeline assessment using InfoAsset Planner, applying engineer-defined criteria to prioritize inspection and maintenance. The city does not classify it as AI.
  • Camera-based sewer inspection that produces 3D imaging and standardized condition ratings.
  • Fiber extension to water facilities, some of which still rely on radio communications.

The common thread is clarity about purpose. Each tool supports a specific decision, and humans remain in charge of operations.

Where AI Fits, and Where It Does Not

Mesa's choice not to use AI in SCADA reflects a reasonable distinction. Control systems that run treatment and pumping must behave predictably, and failures can affect public health. Analytics that support planning, such as ranking pipes for replacement, flagging unusual night flows or classifying acoustic signals, can use statistical or machine learning methods with less risk, because people review the results before acting.

For Arizona utilities, a practical rule is to use advanced analytics to decide where to look and what to fix first, while keeping real-time control simple, transparent and supervised. Our AIoT guide discusses how AI models can struggle with seasonal shifts, a real concern in a desert where summer demand and pipe stresses differ sharply from winter.

Arizona-Specific Challenges

  • Heat and soil movement. Expansive soils and temperature swings stress pipes, and summer demand peaks raise pressures in some zones.
  • Dry soils hide leaks. Water can drain into dry ground or caliche without surfacing, so leaks may run longer before anyone notices.
  • Mixed pipe materials. Older metal pipes and newer plastic pipes behave differently for acoustic detection.
  • Rapid growth. New subdivisions add pipe faster than records sometimes keep up.
  • Cyber risk. Connected monitoring must be protected, as our PLC security checklist explains.

Turning Alerts Into Repairs

Detection creates value only when crews fix what is found. Utilities that perform well tend to have:

  1. Clear triage rules that rank leaks by size, risk and location.
  2. Integrated work orders so alerts become scheduled repairs automatically.
  3. Verification after repair, confirming that night flows or acoustic signals actually dropped.
  4. Customer outreach for leaks past the meter, including help for those who cannot afford repairs.
  5. Feedback into planning, so repeated failures inform pipe replacement priorities.

An Illustrative Night Flow Investigation

Consider a utility that divides its system into district metered areas. At 3 a.m., when customer demand is lowest, flow into one district stays much higher than its neighbors with similar homes. Smart meter data from that district shows low customer use, so the extra water is likely leaking from mains or service lines. A crew deploys acoustic loggers on hydrants and valves overnight. The next morning, correlation software points to a section near an intersection where pipe records show older metal main. Excavation finds a cracked joint. After the repair, night flow drops back in line with neighboring districts, confirming the fix. The scenario is illustrative, but it reflects how combining zone metering, customer data and acoustic tools narrows a search from an entire system to a single block.

Pressure Management: Preventing Leaks Before They Start

Excess pressure increases both the rate of leakage and the frequency of breaks. Pressure-reducing valves and real-time pressure monitoring let utilities lower pressure in zones where it is higher than needed, especially overnight, while maintaining service and fire flow. Pressure sensors also catch transients, sudden surges caused by pump starts or valve operations, that can crack pipes. In growing Arizona cities with varied elevations, pressure management can deliver savings comparable to leak repair at lower cost.

Smaller Utilities

Arizona has many small water systems, from rural towns to private water companies, without large engineering staffs. They can still make progress: complete a basic water audit, meter wells and major mains, use portable acoustic equipment through a contractor, and join regional programs or shared platforms. Long-range wireless networks such as LoRaWAN can connect sensors at low cost, as our LoRaWAN guide explains.

Apparent Losses Matter Too

Not all lost water leaks. Apparent losses include meters that under-register, data handling errors and unauthorized use. Mesa's adviser noted that older residential meters were less accurate at low flow rates, a common reason utilities replace them. Smart meters reduce apparent losses by recording use more completely, while billing system checks catch accounts with zero use that should not be zero. Reducing apparent losses does not save physical water, but it improves revenue and the accuracy of every other water balance calculation, which in turn makes real loss estimates more reliable.

Building the Business Case

Utilities can estimate the value of loss reduction by multiplying recoverable losses by the marginal cost of replacement water, including pumping energy and treatment chemicals. Under tighter Colorado River supplies, that marginal cost increasingly reflects expensive alternatives such as recovered stored water or purified wastewater. Phoenix's Cave Creek recycling facility, for example, is expected to cost more than $359 million to build. Against those figures, sensors and crews that recover even a small percentage of system losses often pay for themselves quickly.

Measuring Progress

Useful metrics include losses as a percentage of water produced, losses per mile of main or per connection, leaks found per month, average time from detection to repair, break rates and customer-side leaks resolved. Reporting these publicly supports trust as rates rise, especially when utilities ask customers to conserve.

Publishing these figures alongside rate changes helps customers see that the utility is reducing its own losses before asking households to cut back.

Looking Ahead

As cuts begin in January 2027 and the long-term rules remain in litigation, reducing losses is one of the few supply strategies fully within a utility's control. Sensors and analytics make it possible, but the results come from combining them with careful asset records, disciplined field work and sensible limits on automation. Mesa's experience suggests that the best leak detection program is not the most sophisticated one, but the one that consistently turns data into repairs.

Our article on smart meters and tighter supplies covers the customer side, and our earlier leak detection systems overview introduces the basics. Find leak detection partners in the IoT Arizona company directory, including smart utility providers in Mesa. Browse more of our Smart Utilities coverage.

Frequently asked questions

Mesa reports citywide loss under 6 percent, compared with about 10 percent across Arizona and about 16 percent nationally, according to its water resources planning adviser.

Water audits, district metered areas, pressure monitoring, acoustic sensors and correlators, smart customer meters, satellite analysis and risk-based asset analytics.

Mesa intentionally does not use AI in its SCADA control systems because reliability is critical, and it keeps those systems internal with staff monitoring at all times.

In planning and analysis that people review, such as ranking pipes for replacement, flagging unusual night flows or classifying acoustic signals, rather than real-time control.

Dry soils and caliche can absorb leaking water without surfacing, soil movement and heat stress pipes, and mixed pipe materials affect acoustic detection.

Complete a basic water audit, meter wells and major mains, use contracted acoustic surveys, join regional programs and use low-cost wireless sensors such as LoRaWAN.

Related articles

Top