Tailrace, the full simulation.
A solar maintenance firm whose crews chase hundreds of alarms daily in record heat. The case followed on this page, a shutdown inverter in a ground-mounted solar plant near Buckeye, on a September Monday when temperatures reach 41 °C.
A solar maintenance firm whose crews chase hundreds of alarms daily in record heat.
Every alarm sorted before crews depart, the most costly first.
- Tailrace operates and maintains around sixty solar plants in Arizona and Nevada on behalf of asset owners, from warehouse rooftops to utility-scale ground-mounted facilities. The Phoenix-based company employs one hundred and ten people, including seventy technicians organised into regional crews.
- Every morning, the control centre receives hundreds of alarms from inverters, combiner boxes, trackers, and meters. Supervisors triage them manually, and a crew sometimes sets off to fix a minor fault while an entire inverter sits idle elsewhere.
- Across plants analysed in 2024 by a global industry report, inverters accounted for 37% of lost power. Extreme heat compounds every issue, and Phoenix recorded 70 days at 110 °F (43 °C) or above in 2024, an all-time record.
- An agent triages overnight alarms from six in the morning. It clusters those stemming from the same root fault and estimates the energy loss for each. It then suggests a dispatch schedule per crew, complete with the spare parts required.
- For every dispatch, it prepares the technician's job sheet with equipment service history and the manufacturer's procedure. The supervisor approves the schedule, and the technician files the field report on site.
What remains with the teams
The supervisor decides on every dispatch and its priority. On site, the technician alone assesses whether work can proceed safely, whether in extreme heat or on live equipment.
- Energy lost prior to intervention
- Time elapsed between alarm and intervention
- Wasted callouts
- First-time fix rate
Monthly reports for asset owners
Each owner receives a report on production, breakdowns, interventions and replaced parts. The agent prepares it from monitoring data and technician reports.
The account manager reviews and sends each report.
Manufacturer warranties
When an inverter or panel breaks down under warranty, the agent prepares the claim file with readings and photos.
The technical manager decides on each claim.
Drone inspections
Thermography flights produce thousands of images. The agent identifies panels showing abnormal heating and maps them onto the plant layout.
A technician confirms each fault on site before any replacement.
Spare parts inventory
The agent tracks parts used and anticipated breakdowns, then recommends replenishment for each regional depot.
The logistics manager places each order.
a shutdown inverter in a ground-mounted solar plant near Buckeye, on a September Monday when temperatures reach 41 °C
A shutdown inverter on a high-heat Monday, before and after.
- Overnight alarms
- At seven in the morning, the supervisor discovers hundreds of alarms with no prioritisation.
- By six in the morning, the agent has already grouped alarms by incident.
- First workstream
- Priority
- The team sets off for the nearest plant, even if the largest fault is elsewhere.
- The agent ranks faults by lost energy, and the shutdown inverter comes first.
- First workstream
- Spare parts
- On site, the technician discovers an inverter board is missing.
- The job sheet lists likely spare parts, loaded into the van before departure.
- First workstream
- Heat
- Work begins in the early afternoon, at the hottest point of the day.
- The agent suggests an early morning departure, which the supervisor approves.
- First workstream
- Owner reporting
- The monthly report is compiled manually at month end.
- The agent drafts the report using monitoring data and field reports.
- Workstream 2
- Warranty
- The inverter replaced under warranty is sent without a full file, and the claim drags on.
- The agent compiles the warranty claim file with logs and photographs.
- Workstream 3
- Receive
- It receives alarms from each plant's monitoring system.
- Group
- It groups alarms originating from the same fault.
- Estimate
- It estimates the energy lost for each fault, based on output from neighbouring equipment.
- Prioritise
- It suggests an intervention order for each team, with the most costly fault first.
- The supervisor approves the schedule.
- Prepare
- It prepares the job sheet for each intervention, including history and manufacturer procedures.
- List parts
- It lists likely spare parts and checks they are available in the depot.
- Track
- It tracks each intervention until generation resumes.
- The technician writes their report.
- Report
- It prepares a weekly review of recurring faults and lost energy.
- Plant monitoring
- Alarms and power generation for each piece of equipment.
- Read-only
- Maintenance software
- Interventions and history for each piece of equipment.
- Creation of job sheets, schedule approved by the supervisor
- Manufacturer procedures
- Manuals for inverters and trackers.
- Read-only
- Depot inventory
- Parts available in each regional depot.
- Read-only
- Team messaging
- Messages sent to team leaders.
- Dispatch of approved schedules
- Operations Director
- Sets the agent rules alongside us and selects the next workstream.
- Supervisor
- Approves each schedule and decides on each intervention.
- Team leaders
- Organise field interventions.
- Technicians
- Assess safety on site for each intervention and write their reports.
- The FDE
- Leads the mission and writes the agent rules alongside the team. Accountable for every agent.
- Triage agent
- Groups alarms and estimates lost energy.
- Scheduling agent
- Proposes intervention order and prepares job sheets.
- Reporting agent
- Prepares monthly reports for asset owners.
- Audit agent
- Compares proposed priorities with actual faults each week and alerts the FDE to any recurring errors.
- A critical fault is ranked too low.
- The supervisor sees all alarms, and any safety alarm moves straight to the top without triage.
- A technician is dispatched to dangerous equipment.
- The technician is the sole judge of safety on site, and lockout-tagout rules remain unchanged.
- The lost energy estimate is inaccurate.
- It relies on neighbouring equipment, displays its methodology, and the supervisor can adjust it.
- An asset owner's data is leaked.
- Each owner only sees their own plants, and all access is logged.
- The team works during the hottest hours.
- The agent factors in heat forecasts, and company heat policies apply without exception.
- We measure the baseline over one month, including energy lost before intervention, response times, wasted trips and first-time fix rates. Supervisors write the priority rules alongside us.
- The agent triages alarms in parallel with supervisors for two weeks, and we compare its priorities with theirs.
- Supervisors base their mornings on the agent's schedule, first for one region, then across all of them.
- We compare against the baseline and choose the next workstream with the Operations Director.
- Energy lost before intervention
- Unproduced energy between the alarm and production resuming, compared to the baseline month.
- Time between alarm and intervention
- Time gap recorded in the maintenance software.
- Wasted trips
- Interventions where no fault was found or the required part was missing.
- Faults resolved on the first visit
- Share of faults resolved without a return trip.