[RE-wrenches] Hurricane damage to solar arrays

[Jason Szumlanski]

More early anecdotal data...

We are finding NO anchors pulled out of roofs, regardless of roof type or
attachment type. That is clearly not the failure point. We also see no
attachment to L-foot or L-foot to rail issues. And contrary to my
expectation, we see no t-bolt failures. What's happening? The panels
themselves are flexing (bowing) sufficiently to work their way right out of
the mid-clamps. This applies both to Unirac's older 1-inch space clamps and
newer 1/4 inch bonding mid-clamps.

This attached pictured system had panels on a north roof pitched very
slightly to the south. We had northeast hurricane winds in this area, and
the way the wind went under these panels was obviously what caused the
panels themselves to fail. This is an essentially flat roof commercial
application. It was bad luck to have the wind direction from the northeast
corner (the NW edge of the eye wall passed right over this area). But we
are seeing similar results on residential pitched hip and gable roofs in
terms of the failure mode.

What's interesting is that there is no rhyme or reason to where in the
array we see damage. I have seen absolutely no catastrophic damage on a
residential roof - just one or 2 modules mostly. And the missing module can
be on the lower edge, upper edge, or right in the middle of the array. More
often than not, the t-bolt and mid-clamp assembly is still sitting right
there in the channel of the Unirac Solarmount rail, but a module is
missing. It's quite freaky.

And much like tornado damage I have seen on TV, houses adjacent to each
other have very different fates. We have a new community (100+ homes slated
for solar) with about two dozen homes completed, and just one home had a
panel pop out in the middle of an array. It was gently deposited onto the
adjacent panel with absolutely no damage and the DC leads still connected
to the microinverter.

One issue we are facing is that when panels fly off, something has to give
with the DC leads to microinverters. No panel leads have been broken so
far. In most cases, the MC4 connectors simple un-snap somehow - no loss of
crimped connectors. We have a few cases of leads ripped out of the
microinverter case completely. The microinverter bracket is badly bent on
many microinverters, indicating that there was tremendous force until
something gave up. I am extremely hesitant to reuse these microinverters
because the force on the DC input leads must have been huge. I think we are
going to insist on microinverter replacement when replacing modules in
these cases.

Hopefully I will have more, but not too much more to come. It looks like we
did very well here (as an industry). There are spotty issues, but it's far
from the catastrophe that kept me up for multiple nights before and after
Irma. Then again, many people have not returned home yet.

​Jason Szumlanski
Florida Solar Design Group​



On Wed, Sep 13, 2017 at 6:18 PM, Jason Szumlanski <
jason at floridasolardesigngroup.com> wrote:

> [image: Inline image 1]
> I'm based in Fort Myers and we cover the hardest hit areas from Irma. We
> are in the "stuff" right now, so I'll make this brief until I have more
> time, which might be a while. We are seeing quite a bit of minor damage and
> some major damage. We're getting calls from all dealers' customers and a
> couple of our own clients. We have several homes with one or two panels
> dislodged. There is no rhyme or reason. Some are middle of arrays, some on
> edges. Panels are ripped right off rails, leads ripped from microinverters.
> Strangely, it looks like the panel j-box connection and MC4 connectors
> survived better than the microinverter end of the DC leads. Amazingly, we
> have several panels that were blown onto driveways, other roofs, and pool
> cages with NO DAMAGE except frame scrapes. Very weird. We haven't seen a
> shattered panel yet, but it's early.
>
> I'm heading to a self-storage facility tomorrow where there are three 25kw
> systems on different buildings. Two buildings are unscathed. One building
> lost ALL of the panels apparently. Tornado? Hard to say.
>
> So far (other than the 25kw I have not evaluated) we have not seen a
> single fastener pulled out. All of the failures are panel top and mid
> clamps at this time. Anchor and rails remain intact. Possibly installation
> errors? Possibly sheared off T-bolts? Hard to tell and we may never fully
> know.
>
> We generally require 48 inch spacing between anchors into trusses for
> engineered systems. The pullout values are pretty high. It looks like the
> attachment points into the roof are not going to be the failure point in
> the systems in Florida, but there is a LOT of work to be done still. It's
> going to be a very interesting few months ahead!
>
>
> ​Jason Szumlanski
> Florida Solar Design Group​
> ​
>
> On Tue, Sep 12, 2017 at 3:09 PM, James Rudolph <jamesrudolph99 at gmail.com>
> wrote:
>
>> Aloha Everyone,
>> I was just wondering how all the PV arrays did during these storms?
>> Does Florida have higher pull out values and wind designs for their
>> PV/H20 systems?
>> Is there any thing the rest of us could learn from all this?
>> Photos?
>>
>>
>> Mahalo Nui Loa,
>>
>>
>> *James B. Rudolph*
>>
>> *Hawaii Unified*
>>
>> *Director of Energy*
>>
>> *ES Electrician # 10816*
>>
>> *NABCEP Certified PV Installer # 091209-155*
>>
>> *80*
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.re-wrenches.org/pipermail/re-wrenches-re-wrenches.org/attachments/20170914/829a1f7b/attachment-0002.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.gif
Type: image/gif
Size: 42 bytes
Desc: not available
URL: <http://lists.re-wrenches.org/pipermail/re-wrenches-re-wrenches.org/attachments/20170914/829a1f7b/attachment-0004.gif>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 20170914_105345.jpg
Type: image/jpeg
Size: 230745 bytes
Desc: not available
URL: <http://lists.re-wrenches.org/pipermail/re-wrenches-re-wrenches.org/attachments/20170914/829a1f7b/attachment-0004.jpg>