Rooftop solar system fires 1

Just noted this article about fires Amazon has been experiencing. They downplay the gravity of the problem, but have turned off the solar power at all their facilities in the US. Is the problem equipment or installation workmanship?

Haven't read all of the way through but we should probably start here.

Not just Amazon, Walmart is now suing Tesla over solar panel fires
Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

There was discussion here recently about a Walmart fire in Indiana. Wonder if that one is part of the lawsuit.

I would like to hope they give a reason why they went on fire.

Turning them off doesn't really help if its the panels that are going on fire. You would have to cover them otherwise they will just sit at max Voltage and be able to produce max current @12 amps when a fault occurs. Wiring issues they can still produce and feed the flames.

Seems it was issues with the MC4 connectors which because all Tesla installations use optimisers even if there is not shading issues means they have 4 times as many connections compared to clean strings.

The MC4 are crimped and not soldered apparently. And they are solid tube crimp. Not fold over crimp.

And we're fitted onsite by contracted installers.

Just checked my panels and they have german sprung loaded cable grips inside them over a s grip. Which I presume allows thermal moving of wire strands and no arc gaps being produced.

Did a search for similar fire events in Europe and there hasn't been any since the regulations about diodes came out.

Crimp connectors are supposed to work by mechanically welding the wires to the connector. It very much depends on proper installation. I believe there are even tools that won't release until you've applied enough compressive force.

In my experience, crimp connectors done properly don't fail.

Soldered connections can "fail" at the melt temperature of the solder. Proper crimp connections fail at the melt temperature of the metals.

Spring-loaded connectors can fail if the springs lose their temper. I've dealt with it a few times.

Hard to believe, I know; but some installers are incompetent.


spsalso

I would be suspicious of incorrect wire insulation selection before I went after the crimped connectors. Everything uses crimped connectors, they are tried and true. However, running electric wire in a high UV exposure environment such as a rooftop might not register as a special environment to many installers.

From what I've seen on top of houses, the installers put the wire in conduit. So the insulation is not exposed to UV. Perhaps they do things differently on the roofs of Amazon facilities out in wherever.

It looks like there's either a fabrication error of the product, or an installation error by the installers. Which it is, or both, doesn't seem to be showing up.

On the installer's side, it seems to me that if one of his connections failed, the failure is contained inside a junction box. And one or more panels shuts down. I don't see how a giant fire would ensue. That's why junction boxes were "invented".


spsalso

So how hot is it the roof? Maybe 150F plus resistive heating.
And now you put that into conduit so there is no air cooling.
And now you cycle this each day with a soak period.
Better be some pretty good wire insulation.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, consulting work welcomed

"So how hot is it the roof? Maybe 150F plus resistive heating.
And now you put that into conduit so there is no air cooling.
And now you cycle this each day with a soak period.
Better be some pretty good wire insulation."

Typical insulation is rated for 194F. There are higher temp available.

Temperatures of working environments must be included in calculations: various de-rating factors. For example, Table 310.15(B)(2)(a) of the NEC has corrections for ambient temperatures up to 437F.

There is also a correction factor for the number of conductors in the conduit: Table 310.15(B)(3)(a)

Should the insulation still fail, the conductor will be shorted to ground; and a breaker will trip. Any damage will be contained in the conduit.


spsalso

The initial wire runs from the panels to the (switchboard for us generator guys, not sure what solar calls it) are unprotected. Any faults within that intial run are usually catastrophic. If you have ground fault detection you may catch it when the first wire chaffs through.

The initial run of a typical solar panel handles about 300 watts. If there's a failure in that initial run, the panel will either be disconnect from the load by the failure, or it will deliver its full power (300 watts) to the failure point. I am not sure what a relatively small amount of power will cause, but I don't think I'd use the word catastrophic. I could see some metal-melt maybe happening. But consider that it's likely that some of the metal that is melting is the wire delivering the power. And thus the fault may be "self-clearing". I would think it most important not to have ignitable materials nearby.

But start adding panels up, and....

I do agree that the designers of the system should pay attention to "failure to a short". It DOES seem likely that a ground fault system and/or an arc fault system gets appropriate pretty quickly.



spsalso

When I say soldered I was meaning silver soldered which in my limited experience in the oil industry they did on power systems but they were crimp then solder. But this was for explosive gas environments. But realistically that's not going to be used in the field for solar.

USA there seems to be major work and discussions about arcing industry wide and yet another code is likely to be issued for arc suppression systems and detection.
Solar in Europe and Australia it is hardly ever mentioned and seems to be a none issue. I can't work out why its a big problem in one area and absolutely nothing else where. In fact both those areas run much higher voltage strings than the US which is capped at 600V domestically, We are capped at 1000V domestically and the farms run 1500V strings. But already there is political lobbying that we should also change the codes to include it. I presume its linked to single market requirement economics and if the rest of the markets don't need it then it sticks out like sore thumb and increases costs/reduces choice in that market. Plus makes local suppliers uneconomic outside that region.

When I say sprung the wire is fed across a conductor edge and then the sprung part is another edge gripping the wire so it S through between the edges. So when it heats up and the strands move they are kept in contact. I think the main idea is that it can thermal contract and expand and rearrange itself and the contact area doesn't reduce over time.
In this case it may not be the crimp but the quality of the male - female connection causing arcing as they seem to be saying it is the connectors from one supplier.

As for the tripping to ground. We don't have fused/breakered strings in Europe. The wire is protected by the max current able to be produced by the panels which is under 13 amps. You can connect a panel directly to ground and a 16 amp fuse/breaker will do absolutely nothing. German inverters have to have a grounding fault detection system. Which works by the grounding of the panels goes back to the inverter and it uses that to work out if all the electrons are going to where they should be. But all it can do is trigger dry switch which can be linked to a audio or visual warning and send you and email/ flag it in the monitoring system. You can't stop the panels producing power unless you block the sunlight.

I think they also reported that they found when they inspected the other installations they found quiet a few of them not grounded at all. So its more than likely revolves round installation errors across a broad range of topics. Tesla seem to have issues with the in roof domestic solar setups as well in the USA but I can't find any reports in Europe. And we don't have the conduit requirements. In fact my strings just go through flexible plastic tube rated at 1500v through the internal walls of my house down to the cellar. USA seems to require most installations to be piped down external walls.

Construction materials may also be a feature, With our consumer units being made out of plastic not metal along with most of our electrical boxes its pretty much normal to use none flammable plastics for components.

I might add this inability to kill power to solar is one of the political lobbying arguments that enphase are trying to use to restrict/ban DC string inverters and only use micro inverters. And of course they hold the bulk of the patents for micro inverter tech. Its a mess of codes and practises. But engineering wise the problems are varied and region specific. The ones that were global they have tackled eg the bypass diodes and panels going on fire.

I don't think Australia is on top of this issue, either. It is reported that about 10% of house fires in Australia are started by rooftop solar. But that could be because domestic installations are done by cowboys.

I couldn't find much for AUS on the subject.

If you have any links I would be interested.

The cowboy issue is rearing its head in the UK there is just so many people trying to get installs at the moment. And a colossal shortage of hardware. The popular brands are 6 months plus lead times. But the brands are restricted for grid tied due to them setting up a cartel qwango of installers/ suppliers which isn't required legally but they have managed to get all the grid DNO's to require it and also house insurance company's. If you have an approved cowboy installer who then walks after a rubbish install of course they will wash their hands of it and say its a civil matter. So 10% price hike for not a lot.

Solder, braze, weld all imply a temperature range. Whether it's lead solder or silver solder, they all melt within a low temp range that's included by the definition of soldering. There is also a silver brazing process often erroneously referred to as solder. This is most familiarly used for joining copper pipe in refrigeration systems. I have never seen electrical connections joined by brazing.

Alistair,

Media reports:




There are many more, but these are all domestic installations, not the big commercial fires reported in the US.

"Solder, braze, weld all imply a temperature range. Whether it's lead solder or silver solder, they all melt within a low temp range that's included by the definition of soldering. There is also a silver brazing process often erroneously referred to as solder. This is most familiarly used for joining copper pipe in refrigeration systems. I have never seen electrical connections joined by brazing."

I've heard of the term "brazing", but never used it or heard anyone else apply the term. Up until now, my understanding has been that it refers to a joining of two metals using a brass alloy. I've never seen it done.

I have done work called "soldering" by all concerned with the task.

There was what I would call "soft soldering", typically using a lead-tin alloy that can be melted by a hot iron. This would be in the range of 600F-700F. I have typically used it in electronics.

And there was "hard soldering". Also called "silver soldering", since it uses an alloy with a high silver content. This required a torch, typically acetylene/air. Temperature range is 1200F-1500F.

You get a much better bond with solver soldering, if that's what you need. I did an experiment of laying a 12 gauge copper wire across another and soldering the two together with soft solder and another with silver solder. I then hammered them flat. The soft soldered wires separated. The solver soldered wires did not.


I do know there is a silver bearing soft solder, and have used it. In my case, I found no advantage.


spsalso

Sounds more like brazing I am on about you need a blow touch with MAPP gas in it a propane torch it will just still there.

I am more than happy with the possibility that crimping a connector onto a solar wire is outside the skill set of causal labour trained in 5 mins how to use the tool.