Key Takeaways
- Solar underperformance alerts catch a bad string, a tripped inverter, or a soiled array days or weeks before your monthly bill does.
- Good alerts compare today against a smart baseline (a nearby PV forecast or your own historical yield), not a fixed number.
- Set a sensible threshold and a grace period so you get real faults, not false alarms every cloudy afternoon.
- Most portals (Solis, Growatt, SolarEdge, Enphase) have built-in alerting; add a simple external check if yours does not.
- An alert is only useful if it reaches you fast and tells you what to check next.
A solar array almost never fails loudly. It just quietly makes less power, and you find out four weeks later when the bill lands. That is exactly the gap solar underperformance alerts close. Set up right, they ping you the day a string goes dark or the inverter starts clipping errors, instead of you eyeballing an app that most people stop opening after the first month. I have commissioned enough rooftop systems across India to know that the ones with working alerts get fixed in days, and the rest limp along for a season. In this guide you will learn how to pick a baseline, set a threshold that does not cry wolf, wire up alerts on common portals, and build a simple backup check if your inverter brand is stingy with notifications.
Why fixed-number alerts fail (and what to use instead)
The tempting first move is "alert me if today is under 20 kWh." That breaks the moment a monsoon week rolls in, and you either drown in false alarms or you mute the whole thing. A 5 kW system that makes 25 kWh in April may honestly make 9 kWh on a heavy-cloud July day, and nothing is wrong.
The fix is a relative baseline. You compare today's output against what a healthy system should have made under today's weather, not against a calendar number. Two baselines work well for home setups, and you can combine them.
| Baseline type | How it works | Best for | Catch |
|---|---|---|---|
| Historical self-baseline | Compare today to your own same-week average from prior years or months | Systems with 6+ months of data | Blind in your first season; drifts as panels age |
| Irradiance / forecast ratio | Compare actual kWh to a PV forecast for your GPS location | New systems, any location | Needs a weather/PV data source |
| Fixed floor (backup only) | Alert if output is near zero for a full sunny day | Catching total outages | Misses partial faults |
For real fault detection you want the ratio approach as your primary trigger, with a near-zero fixed floor as a dumb-but-reliable safety net for total inverter death.
Setting up low production alerts: the step-by-step
Here is the workflow I use when I hand a system over to a homeowner. It applies whether you use a built-in portal or an external tool. Do these in order and you will end up with low production alerts that are actually trustworthy.
- Establish your expected daily yield. A rough rule for India is 4 to 4.5 units per kW per day on a good clear day, less in monsoon. So a 5 kW system peaks around 22-25 kWh. Note this per season.
- Pick your baseline. New system, use a PV forecast ratio. Older system, use your own historical same-week average.
- Choose a performance ratio threshold. Alert when actual falls below roughly 70-75% of expected for the day. Below that is rarely just weather.
- Add a grace period. Require the shortfall to persist for 2 to 3 consecutive days before firing, so one cloudy day stays quiet.
- Add a near-zero floor alert. Separately, fire immediately if output is under 5-10% of a normal day, since that usually means a tripped or dead inverter.
- Choose a delivery channel. Email at minimum, push notification or WhatsApp if the portal supports it. A missed alert is a useless alert.
- Test it deliberately. If you can, cover a panel or switch off one string for a day and confirm the alert fires. Untested alerts are just hope.
Tip: Set your grace period to compare against the same day-of-week window, not a rolling 3 days. Weekends and weekdays can differ if you have self-consumption metering in the mix, and you want weather-only comparisons.
Performance alert setup on common inverter portals
Most brands already ship a performance alert setup screen. The names differ but the logic is the same. Here is where to look on the portals I see most in Indian homes, and what each one can and cannot do out of the box.
| Portal | Built-in alerts | Relative baseline? | Notes |
|---|---|---|---|
| SolarEdge (mySolarEdge) | Yes, per-panel with optimizers | Yes, panel-level comparison | Strongest for pinpointing a single bad module |
| Enphase (Enlighten) | Yes, micro-level | Yes, cross-panel | Flags individual microinverters that lag peers |
| Solis / Ginlong | Yes, fault + low-yield | Basic (mostly fault codes) | Reliable fault alerts; low-yield logic is coarse |
| Growatt (ShinePhone) | Yes, fault alarms | Limited | Good for outages; add external check for soft dips |
Panel-level systems like SolarEdge and Enphase have a real edge here. Because they see each module, they can flag one shaded or failing panel that a string inverter would blend into an average and miss entirely. If you are still choosing hardware, that difference is worth weighing.
Building solar fault alerts when your portal is basic
Plenty of budget inverters give you fault codes but no real low-yield logic. You can still build proper solar fault alerts on top of them, as long as the portal exposes your daily production somewhere you can read.
Option A: A simple daily digest and eyeball rule
The lowest-effort path is turning on the portal's daily production email, then setting one mental rule: if a clearly sunny day comes in under three-quarters of last week's best day, open the app and investigate. It is manual, but it beats no habit at all.
Option B: A small automated check
If your inverter has an API or a local Modbus/CT feed, a tiny script or a tool like Home Assistant can pull daily kWh, compare it to a PV forecast, and message you. Pairing this with proper CTs for consumption monitoring also lets you separate a generation drop from a consumption spike, which is a common source of confusion.
Safety warning: An alert tells you something is wrong, not that you should open the inverter or touch DC wiring. Rooftop DC can exceed 300-600 V even when the grid is off. Diagnose from the app, and call a qualified installer for anything past a resettable breaker or a visible soiled panel.
What I actually see trigger real alerts
When I commissioned systems and then watched them over their first year, the alerts that mattered were boring and repeatable. Soiling was the biggest one, a slow 10-15% slide over the dry pre-monsoon weeks that a good ratio alert catches long before you would notice by eye. String or fuse faults showed up as a sharp step-down on one MPPT input, obvious once you compare inputs.
Grid-side trips were the other frequent flyer, especially in areas with wobbly voltage, where the inverter safely disconnects and simply stops making power mid-afternoon. A near-zero floor alert catches that same day. The false alarms almost always traced back to a threshold set too tight or no grace period, which is why I keep hammering those two settings.
Make sure the alert actually reaches you
The best-configured alert in the world is worthless if it lands in a spam folder you never open. This is the step people skip, and it quietly undoes all the earlier effort. Decide up front where the alert should land and make sure that channel is one you check daily.
Email is the universal fallback, but it is easy to ignore. Push notifications from the brand app are better because they interrupt you. If the portal or your automation supports WhatsApp or SMS, that is best of all for a genuine fault, since almost everyone reads those within minutes. Whatever you choose, send a test alert to yourself and confirm it arrives before you trust the system.
One more habit worth building: when an alert fires, it should tell you what to check first, not just that something is off. A good alert reads like "generation 60% of expected for 3 days, check for soiling or a tripped string," which turns a vague worry into a two-minute rooftop or app check.
Frequently Asked Questions
What is a good underperformance threshold to start with?
Start by alerting when a day's output falls below about 70-75% of what a healthy system should make under that day's weather, and require it to persist for two to three days. That range flags real faults while ignoring ordinary cloud cover. Tighten it later once you know your site.
Will alerts work in my first monsoon without historical data?
Yes, if you use a forecast-ratio baseline instead of your own history. It compares actual output to a PV forecast for your location, so it works from day one. Add a near-zero floor alert too, which needs no baseline and catches total outages immediately.
Can a string inverter alert me about one bad panel?
Not precisely. A string inverter only sees the whole string's combined output, so one weak panel gets averaged in and often stays below the alert threshold. Panel-level systems using optimizers or microinverters can flag a single module, which is their main monitoring advantage.
How do I stop getting false alarms every cloudy week?
Two settings fix almost all false alarms. First, use a relative baseline that already accounts for weather rather than a fixed kWh number. Second, add a grace period so the shortfall must persist for several days before the alert fires. Together they filter out normal cloudy stretches.
Do I need extra hardware for underperformance alerts?
Usually not. Most inverter portals already log daily production and offer built-in alerting, so software configuration is enough. You only need extra hardware, like CTs or a data logger, if you want deeper insight such as separating generation from consumption or running your own automated checks.
Conclusion
Underperformance alerts are the cheapest insurance your solar system will ever get: a few minutes of setup that turns a silent, slow-bleeding fault into a same-week fix. Pick a relative baseline, set a 70-75% threshold with a short grace period, add a near-zero floor, and test that it actually fires. Then let it fade into the background and do its job. If you want the bigger picture of how alerting fits alongside dashboards and metering, read our complete guide to solar monitoring, or browse the Solar Monitoring FAQ for quick answers. You can also learn more about who is behind these guides. For neutral background on PV performance, the NREL photovoltaic research pages are a solid, non-commercial reference.
