When producing power for the grid under a fixed price and limited duration of contract, it is in the interest of the producer to ensure that the array is operating at full capacity given light conditions at all times. So how do we know when a component fails to generate its share of the potential power?
I have no experience with monolithic string inverters, where I understand that failure is an all or nothing deal, that is either you are producing full power or nothing at all. My own system has Enphase micro inverters, each of which reports back to a central box ("Envoy") which in turn reports to the central Enphase database via Internet. Ideally, the Envoy gives a complete picture of the state of the array at any given moment or recent period of history, and the main Enphase database, via its Enlighten query system, gives the full history of production since installation based on information received from the Envoy.
I used the word "ideally" deliberately. The Envoy receives information about each inverter and the panel it controls. The inverters are fully autonomous, and are able to stop power production or maintain production according to the conditions on the panel and in the grid. The Envoy receives information across the power lines themselves, so any non-standard circuit such as an active power back-up system, while it may allow power to be generated and sent into the grid, may not allow information to be carried to the Envoy. Some systems do not have an Envoy box at all for other reasons, or may have an Envoy but the inverters, while they are producing power correctly may not be communicating properly with the Envoy. Under ideal conditions, with a fully functional Envoy receiving data from all inverters, a producer knows from the Envoy when a component is not behaving within normal parameters.
In addition, each inverter has a LED light indicating its current status. On the D380 dual inverters this is a small light which flashes green when all okay, orange when producing power but can't communicate with the Envoy and red when it it not producing power. So if a system does not have an Envoy, a daily trip to examine the LEDs gives a good idea of the overall condition of the array. All flashing green means all okay. All flashing orange means maybe all okay if there is no Envoy, but anything flashing red means a problem to be taken care of at the earliest opportunity since time is money.
The LEDs on the D380s are tiny, and the flashes very difficult to see in bright light, so the best time to monitor is just after sunrise when the inverters have detected power from the panels but before the sun rises high enough to drown out the light from the LEDs. This can happen quite quickly. The green is hardest to see in bright light, and the location of the LED on the side of the inverter makes the signal even harder to detect. Fortunately Enphase has learned that an easily visible LED is important; the LED on the new M190 is much brighter and located on the bottom of the inverter. The green is visible even in the brightest sunlight. Well done, Enphase!
In this image, note the two kinds of inverters. On the left are the two single inverters and on the right the single dual inverter attached to a pair of panels. The red arrows point to the location of the LED lights on each type.
The location of the LEDs is an important factor. I would imagine that located on the side is good for a roof mounted array, where it is difficult to see under the inverter. But for my rack mounted array, beneath the inverter is perfect. There are some stories of roof mounted systems without the Envoy reporter running into production deficits involving thousands of dollars simply due to the fact that it is hard to monitor production.
Envoy and LEDs aside, there really there is no substitute for monitoring the meter. A system may report that it is generating a certain number of kilowatts, but it will take some energy to push that into the grid, so the meter readings will be valuable in that they tell you what you will be paid for. Keeping a record of meter readings in addition to weather statistics gives concrete data which will be useful for the life of the system or the contract. With data generated on the kind of production for the time of year, time of day and given weather conditions it should be possible to set up a system which can judge whether the system is running on all cylinders or not.
One very useful system is to build a historical record of generation by month, then take the average of the month's readings. This will give the expected generation for the present month; the average of your historical monthly readings is statistically a best guess at what the current month's production will be. Tracking the actual cumulative generation and comparing with where you should be given the day of the month will give a good idea whether the array is behaving within normal parameters. There will be some variation around the expected generation, but it will only rarely be more than 10% from the expected value, and your intuitive observations of light conditions should explain that away, or in fact indicate a problem.
A more detailed approach would be to build a mathematical model based on weather and light conditions. A regression model such as:
kW = f( month, dayOfMonth, ambientTemperature, humidity, cloudCover, windSpeed)
should give a pretty accurate assessment of the power the array should be producing at any moment in time given the prevailing conditions.
Of course, the cumulative method or the regression prediction will indicate an overall possibility of an issue with the array. If the prediction raises a flag, it is time to check the Envoy or the LEDs. And the sooner this is done the more quickly the potential problem can be addressed and if necessary corrected to bring the array back to full production.