How does predictive maintenance help a municipal fleet stay on budget?
Predictive maintenance turns unplanned repairs into scheduled ones. It reads the telematics data a public fleet already collects, detects a failing component two to three weeks before it stops a vehicle, diagnoses the cause, and gives the service shop time to plan the repair. Costs that used to land mid-year as emergencies become work the fleet saw coming.
Why It Gets Missed
A municipal fleet runs on an appropriation set months in advance. That number does not move when a refuse packer’s aftertreatment fails in March or a plow derates in the first snow. The repair gets paid for anyway, out of next quarter’s plan.
The pressure shows in the largest city fleet in the country. New York City’s Fiscal 2026 Mayor’s Management Report shows average fleet age at a record 100 months, repair spending up from $327.0 million to $358.7 million in one year, and the daily in-service target met on 52 percent of days, down from 64 percent. The report ties the drop to older vehicles and fewer funded mechanic positions.
With older vehicles and flat headcount, preventive maintenance schedules carry most of the load, and they work well for oil, filters, inspections, and wear items on a set interval. They were not built to show which ten-year-old packer on a stop-and-go route is wearing out its DPF faster than the rest.
Experienced technicians often know a vehicle is not right before any screen does. What they are rarely handed is a month of trend data for every vehicle in the yard, ranked by risk. The telematics dashboard shows a code after the problem has escalated. It does not show the cause, how fast it is developing, or what to do next. That gap belongs to the data, not to the people reading it.
How AI Reads It
AI models learn what normal looks like for each vehicle from the time-series telematics data the fleet already streams: its own route, load, idle time, and climate. A refuse vehicle idling through hundreds of daily stops has a different normal than a pickup on highway runs.
When a component starts drifting from its own normal, the model tracks how fast. DPF soot load building between regens, EGR flow degrading, a NOx sensor reading off, cranking voltage sagging on a plow that sat all summer. The system raises a breakdown alert when the failure is close enough to act on with confidence, but still weeks before it would take the vehicle off a route.
The alert carries a recommendation, the likely cause and the repair to make, and lands as a work order in the system the service shop already uses. The technician confirms the diagnosis and does the work. The judgment stays with the people who know the equipment; the AI makes sure they see the right vehicle at the right time.
What Changes
- A repair in the service shop costs less than one on route. A breakdown alert brings the vehicle into the bay at planned labor rates, with the part on the shelf and a spare covering the route. The same failure on route adds a tow, emergency labor, overtime to finish the collection, and often a larger repair, because a failing component that keeps running damages the parts around it. Moving that work into the service shop is the most direct way to cut the downtime costs a fleet absorbs.
- Repairs get fewer and smaller. Fixing the cause early keeps a sensor replacement from turning into a rebuild, which helps extend vehicle life on units still waiting for a replacement budget. Troiano Waste Services, a refuse fleet of 55 vehicles in Portland, Maine, cut exhaust-related repairs by 41% in its first 10 weeks and saves $87,642 a year across those 55 vehicles. “The financial results speak for themselves,” said TJ Troiano, COO.
- Routes stay covered. In a ten-week pilot on an Iowa city refuse fleet of 50 vehicles, early alerts avoided $21,346 in repairs and downtime and two route disruptions per week, with zero vehicle comebacks.
- The budget conversation changes. A fleet manager who can show which repairs were caught early and which units are trending toward failure walks into a budget hearing with evidence. Unplanned work becomes a forecast, and a forecast can be funded.
Questions fleets ask
How can a city buy predictive maintenance software without an RFP?
Tensor Planet sells to public agencies through the RTA contract number 102924-RTA, a Sourcewell cooperative contract, so cities and counties get pre-approved pricing with no separate RFP process. It connects to the telematics already in place, such as Geotab, Samsara, and Verizon Connect, and needs no new hardware.
How do city solid waste fleets avoid missed collection routes?
Tensor Planet helps city solid waste fleets detect, diagnose, and prevent engine and aftertreatment failures two to three weeks before a refuse vehicle goes down on route, so repairs fit between collection days. In a ten-week pilot, an Iowa city refuse fleet of 50 vehicles avoided two route disruptions per week and $21,346 in repairs and downtime.