Has anyone heard of roof fires caused by PV installations?  The attached article describes this in pretty good detail.  One suggested fix is to unground the dc side of things.  This will take some effort to get the National Electrical Code to buy into this since they presently require it to be grounded.

No. But I have heard of fires from lightning.
Wasen't the installation of lightning rods something like put metal rods on a farm house, and connect them to ground (Sort of like you are doing with solar panels)?

Somewhat, but PV panel impedance to ground is generally quite high, even assuming you could get through the glass covering.

TTFN

FAQ731-376: Eng-Tips.com Forum Policies
Chinese prisoner wins Nobel Peace Prize

I'm not sure this lightning diversion is worth pursuing.  One of these fires occurred in Bakersfield, where lightning has a pretty low incidence.

Just though this group might be interested in a real-life case history.  Does anyone have any comments on the analysis offered by Mr. Brooks?

I have seen structures and electrical components severly damaged or destroyed even when lightening rods were present.  No guarantees as far as I am concerned.

Mike McCann
MMC Engineering

 

There was no mention of lightning in the report.

Something still seems to be missing.  There just shouldn't be enough flammable material just lying around in these installations.  In one spot, the author talks of, "the flammable PV module backsheet or any flammable roof materials."  Seriously?  We go out of our way to buy fire resistant roofing materials to meet the fire code, and the roofing has "flammable materials?" Most roofing materials are designed to resist a direct flame for HOURS.  

And the PV panel has a flammable backing?  Seriously?  And most wiring insulation just isn't that flammable, since that's got be one of the only two significant construction requirements for wiring; "insulate the voltages" and "don't burst into flames if the wires get too hot."  

And why are there multiple faults in this installation?  Seems like there's something else that went on.   

TTFN

FAQ731-376: Eng-Tips.com Forum Policies
Chinese prisoner wins Nobel Peace Prize

Quote:

Most roofing materials are designed to resist a direct flame for HOURS.

If we're talking residential, cedar shakes are pretty flammable and still common for roofing around here.  And plywood burns pretty well.





 

David Castor
www.cvoes.com

Not sure where you are, David, but around here, Southern California, only old houses have cedar shakes grandfathered in.  New houses and new roofs have to meet something like a 4-hr fire resistance test.  That applies to shingles, tile, steel, etc.  Commercial roofs are usually, tar & gravel, tile, membrane, or metallic.  The membrane ones are probably most flammable, but would still need to have tested to resist fire for some number of hours.

TTFN

FAQ731-376: Eng-Tips.com Forum Policies
Chinese prisoner wins Nobel Peace Prize

I don't know why I supplied the attachment.  Looks like most of the comments indicate that they didn't read it.

It's not a cedar shake residential roof.  383 kW.  Big box store.

Ground fault current was over 300 A on a #12 wire.  Any idea what that does to the insulation and the wire?

There were problems with the expansion joints.  See pictures in article.

The PV panels I've inspected don't look very flammable.

I should have posted this on the Electric Power T&D forum but I thought since it involved PVs, this would be the place where it belonged.  I was looking for a better alternative to ungrounding the dc side since the NEC doesn't allow that.

Are there any protection people out there?  Will a ground fault protection system work where you wrap a CT around the plus and negative leads going to the collectors?  I was looking for a solution that would offer better protection than simply waiting for the existing protection to do it's thing.

 

A magnetic core CT won't do much of anything on DC.  There are other sensor types that might, but not a CT.

Most PV panels are current devices, so they can be shorted. So a protective device for a PV system only needs to short the positive to the negative.

However if what you are talking about is to small wire, with melting insulation, that is a different story.

David is correct, CT's won't work on DC, but there are other current measuring devices that will (metered shunt).

Some electrical system installations dispense with fuses on the ground, and if fault currents are too high (there are cases where faults such as these ARE designed for by the engineers) then the equipment-grounding conductors are attached to resistors (low-ohm) before connection to the ground rod.  This allows a certain amount of fault current to flow, and when significant faults occur the resistor acts like a shunt to actuate various alarm/safety/shutdown devices.

In a system with a potential 300Amp fault current then a 0.1 ohm resistor (or shunt) would give the detector an obvious error signal, yet allow a smaller 3Amp fault current through, or perhaps sound a lower priority warning.

... at least it sounds like a starting point.  There may be demons in that design that I wouldn't think of (not an EE).

Steven Fahey, CET