Written by Dr. Robert Kowalski — Ph.D. in Agricultural and Biological Engineering from Penn State University, with 18 years of experience in on-farm electrical systems and energy data analysis. He has diagnosed hundreds of underperforming farm solar systems from their monitoring data across the Midwest and Northeast. He is a licensed Professional Engineer and a member of the American Society of Agricultural and Biological Engineers.
Your inverter app currently shows 47 alerts, and you understand none of them. That is fine — 42 of them do not matter. Here are the five numbers that do, the 10-minute weekly routine that catches 90% of production losses while they are still cheap, and the one comparison that tells weather from a real fault.
The Problem
Monitoring systems are built by engineers for engineers, then handed to farmers with no translation. The result on most farms is one of two failure modes.
Failure mode one: total ignoring. The app was opened twice in 2022. The array has been underperforming since a string fault in March. The loss: 15–30% of annual production — $1,500 to $6,000 on a mid-size farm system — discovered at the annual true-up.
Failure mode two: alert fatigue. Every firmware hiccup generates a notification, the farmer learns to dismiss them all, and the one alert that matters — string 2 offline — is dismissed with the rest.
Both failures have the same fix: stop reading the app like a dashboard and start reading it like a checklist. Five numbers, ten minutes, once a week.
The Five Numbers That Matter
1. Daily kWh — against the month, not the day. A single day means nothing (clouds exist). What matters is this month’s total versus the same month last year, or versus the installer’s projection. A month more than 10–15% below expectation, with weather accounted for, is a fault until proven otherwise.
2. Peak instantaneous power (kW). On a clear day around solar noon, a healthy array should hit 75–85% of nameplate. A 50 kW array that never breaks 33 kW on a perfect day has a string, shading, or inverter problem — no matter what the monthly totals say.
3. Specific yield (kWh per kW per day). This is the great equalizer: daily kWh divided by array kW. A spring day should deliver 4–6 kWh/kW; winter, 1.5–3. It lets you compare your 50 kW array against your neighbor’s 20 kW array, or against PVWatts for your county. If your specific yield trails the model by 15%+ for a month, investigate.
4. String or module-level status. If your system has string-level or optimizer-level monitoring, scan for the outlier: one string producing 30% less than its twins is a fault with an address. This turns “the array is down 12%” into “row 3, string 2” — a one-hour fix instead of a week of guessing.
5. Grid voltage events. Frequent over-voltage disconnects (common on long rural feeders) silently clip production every sunny afternoon. If the app logs grid faults more than a few times a month, that is a utility conversation worth real money — we covered the underlying issue in our net metering guide.
The 10-Minute Weekly Routine
Do it every Monday with coffee. Put it in the phone calendar or it will not happen.
- Open the app. Check this month vs. last year same month. (2 min) Within ±10%? Done — close the app, ignore the alerts.
- If below: check the weather alibi. (3 min) Pull your county’s actual sun/rain for the month. Bad weather explains bad months surprisingly often — and if it does not, you have a fault.
- If no weather alibi: check peak power on the last clear day. (2 min) Below ~70% of nameplate on a clear noon = hardware problem.
- Drill to string level. (2 min) Find the outlier string or module. Screenshot it.
- Log one line in a notebook or spreadsheet. (1 min) Date, month kWh, vs. last year, notes. Twelve lines a year, and you now own the most valuable document your system has: the production baseline — the same document that wins insurance claims and warranty disputes.
Weather or Fault? The One Comparison That Settles It
| Symptom | Weather | Fault |
|---|---|---|
| Low month | Below the model, but matching cloudy-day count | Below the model with normal weather |
| Daily curve shape | Normal bell, lower height | Flat-topped (clipping), stepped, or gap-toothed |
| Peak power | Normal on clear days | Never above ~70% of nameplate |
| String data | All strings equally weak | One string clearly worse |
| Trend | Random month-to-month | Steady decline over months |
The daily curve is the tell: weather lowers the whole bell curve evenly; faults distort its shape. A flat top means inverter clipping or derating. A midday notch means shade or a string dropping out under heat. Teeth mean intermittent disconnects. You do not need to diagnose the exact component — you need to know whether to call the installer or blame the sky.
What Monitoring Is Really For (Beyond Faults)
- The insurance baseline. As covered in our claims guide, exported monitoring history is the difference between a paid hail claim and a denied one.
- Warranty enforcement. “Feels low” loses arguments; 18 months of logged data wins them. Panel and inverter warranties are performance contracts — data is how you enforce them.
- Rate and expansion decisions. Your interval data is the only honest input for adding batteries, a second array, or an electric vehicle. Every sizing method on this site starts with it.
- Degradation tracking. Normal is 0.4–0.6% per year. Your annual log proves whether you are normal or have a warranty event developing — see our degradation guide for the full diagnostic.
Expert Tips
1. Export your data before switching anything. Changing installers, rep platforms, or monitoring hardware can orphan your history. Export the full CSV history annually — some platforms simply delete old data.
2. A monitoring gap is itself a red flag. If the app shows “no data” for days, the array may have been producing nothing — or producing fine with a dead gateway. Verify at the meter before assuming.
3. Cellular gateways die first. On rural systems, the monitoring gateway and its cell modem are the least reliable components in the entire system. Budget one replacement in years 5–8.
4. Alerts to actually act on: inverter offline over 24h, string at zero while others produce, ground fault, arc fault, repeated grid faults. Everything else can wait for Monday.
5. Compare against PVWatts, not the sales brochure. Enter your array specs into NREL’s free PVWatts with your county’s weather station. That model — not the installer’s projection — is your fair benchmark.
6. Teach one other person the routine. The system outlasts attention spans, vacations, and farm sales. A 10-minute routine only works if someone always knows it.
Conclusion
Solar monitoring is not a dashboard to admire — it is a checklist to run. Five numbers, ten minutes, every Monday: month versus last year, the weather alibi, clear-day peak power, string outliers, and one line in a logbook. That routine catches faults while they are cheap, builds the baseline that wins insurance claims, and turns your inverter app from a source of noise into the cheapest insurance your array has.
Frequently Asked Questions
Q: How do I know if my solar panels are producing enough?
A: Compare this month’s kWh against the same month last year (±10% is normal), then sanity-check against NREL’s free PVWatts model for your county. The fastest field test: on a clear day at solar noon, a healthy array should hit 75–85% of its nameplate kW. If your specific yield (kWh per kW of array) trails the model by more than 15% for a month with normal weather, you have a fault, not bad luck.
Q: What does it mean when one solar string produces less than the others?
A: A string producing 20–30% below its neighbors with identical orientation has a physical problem: a failed optimizer or bypass diode, a damaged panel, a loose connector, or new shading. The value of string-level monitoring is exactly this — it converts “the array seems low” into an address (row, string) that a technician can fix in one visit. Report it with screenshots; under warranty, this repair should cost you nothing.
Q: Do I need to check my solar monitoring app every day?
A: No — daily checking breeds alert fatigue and teaches you to ignore the app. A 10-minute weekly routine (monthly totals vs. last year, weather check, clear-day peak power, string scan, one log line) catches 90% of losses within weeks of onset. Configure push alerts only for the critical events: inverter offline, ground/arc faults, and strings at zero. Everything else waits for Monday.
Related Articles
- Solar Panel Degradation and Underperformance on Farms: How to Detect, Diagnose, and Recover Lost Production — The full diagnostic when your data says something is wrong.
- Farm Net Metering 2026: How Export Rate Cuts Change Your Solar ROI — Using your interval data to fight rate and grid-voltage problems.
© 2026 Farm Solar Guide. All data sourced from NREL PVWatts modeling, manufacturer monitoring specifications, and documented US farm operations. Last verified: November 5, 2026