As Mike says, this is not news, and is accounted for by any good simulation.

And most well-designed systems in the UK are oversized for the inverters to make best use of plant installed - this might be by a considerable margin for larger installations - so the actual output is often still limited by the inverter rather than the array theoretical capacity anyway.

My sisters friend has a large solar array and checked his output in current hot weather to be surprised that it was not working.

Called in an expert who found that the invertor had blown up and would cost around £1,000 to replace.

Not sure if it was damaged by lightning or if the household insurance will pay up? 

Also he needs to check the warranty - two reasons, first the inverter may be covered by it, that would be good, but 2nd even if it is not, any remaining guarantee on the panels  (could be 20 years..) may be invalidated if the inverter is changed by anyone other than the first installer, or for a non-equivalent type.

Or it may just  be part of the ongoing costs of of generating...  Note that inverters like any other electronics do very badly stuffed into poorly ventilated loft spaces and allowed to cook..
Mike

It may have died of its own accord.  The early ones were rubbish.

My first one came with a 5½ year warranty.  That seemed a rather strange warranty period, until it died after 6 years with a "relay fault".  The manufacturers had obviously found out how long they really lasted.

Modern ones should last a lot longer.  If it's not very old, check the warranty.

I have noticed much drop off in power over the last week or so. Our solar panels can, in theory, generate up to 5.5kW but we cannot export any more than 3.6kWh back onto the grid according to our DNO, which is the limit imposed by the inverter. The panels themselves didn't get too hot to touch.

Because of the solar panels, our energy bills these last 3 months to July have been -£15, -£5 and £3.47 so it they are saving quite a bit of money under the current climate.

The regulator failed? Did it hit an overvoltage condition or something? They're usually pretty reliable. 

Maybe use an SSR instead and control the output with a range? i.e. have a microcontroller look at the output from the panel and if it's within 10% of what you're looking for then switch the SSR, otherwise leave the output disconnected?