Kent, William, Nick, Dav e,
et al,
Wish I had time to respond
sooner than this, but haven't. Where to start? Oh,
yeah. Dave P's comment about getting opaque tarps for his job trailer
is right on. Good advice for everybody.
I get the sense some
may think I am trying to reduce the maximum allowable OCP rating. I am
not. Go with the listed Series Fuse rating if you like.
My position is simply to
ALLOW a designer/installer to choose OCP rated LESS than 1.56 ISC. By
doing this, certain safety hazards can be reduced by up to 40%, or
more, compared to many system configurations being
installed today. WITHOUT reducing reliability in any way. At NO
ADDITIONAL COST. And there's no need to invent, patent, alpha-test,
beta-test, early-market-hype... None of that.
All you gotta do is ALLOW
OCP that is something between ISC & 1.56 ISC.
There are numerous examples
of burned-up equipment and roofs. Hell, there's probably something out
there cooking right now.... Many cases where selecting a safer, lower
amperage fuse would have reduced the damage. By opening up the currents
backfeeding into a fault sooner. At a lower amperage. With less heat
and lower incident energy.
Our power &
current-limited energy sources will continue to feed into a fault until
the sun goes down or until someone clears the fault. Whichever comes
first. On conductors that can safely handle it until somebody gets
there to fix it. That's gonna just have to remain the case until we
have RELIABLE, cost-effective, DC arc-fault circuit interrupters at the
source.* Our job, as an industry, is to reduce the
potential hazards and damage, when faults happen, to their lowest
responsible levels. This means opening up the backfeed from other
circuits as quickly as possible when something goes wrong.
Pick a module. Pick an Ipmax
and Isc. Let's do the strings and feeders and evaluate safety,
efficiency, cost... All the stuff that needs to be counted. And you
tell me.
How about a 7.99 Ipmax and
8.51 Isc with a 15A Series Fuse rating? I just made it up. Let's use
that. I'll say use a 9A midget in the string combiner. You're gonna say
15A midget in the string combiner. We're both gonna say the same size
wire. My situation is 6A safer. 40%. You're gonna say "Enhanced
Irradiance is a Normal Operating Condition!". And I say fine!
9A Fuse = 112% Ipmax. Hmmm?
Is that enough? Maybe not. Or maybe so. Especially when we know they
aren't "precision" circuit protection.... Using your own numbers, we
can figure on NOT tripping below 105% of the rating, which comes out
to.... 9.45. Or 118% Ipmax. Maybe worth a shot. If it was mine, I
would. And would be happy to report the conditions any time a fuse blew
over the course of its life.
But it ain't mine. And I'm
way good with using a 10A fuse here. 125% Ipmax. Which is still 33%
safer than your 15A fuse. And will NEVER blow under normal operating
conditions. NEVER! And it cost the same. And it took the same amount of
time to order and install. And, when the tree fell across the conduit
during a windstorm one night... Crunching a pipe on the roof...
Shorting out 1 of 4 strings... It didn't cook with all the other
strings coming from the combiner past 10:00 AM the next morning. By
10:00 AM the sun was up good enough that my 1.18 ISC rated fuse blew. Just like I hope it would. Because
it only needed 400 W/m2 to blow.
What if I had put in 9A? 350
W/m2.
But your 15A fuse is gonna go longer. Yours
is gonna need to be right close to 600 W/m2, and that much sun don't
always come around every day. Yours is going to cook with all the heat of
the injured string PLUS 176% ISC until... Well, maybe not until a
couple days later. Or a week? Will it cook until it faults other
conductors? Or starts a fire? Where's the National Fire Protection in
that? Will it still
be cooking when the service tech without a DC current clamp meter and
the knowledge that he needs one..... When THAT GUY gets there? Is yours
still gonna be cooking when THAT GUY gets there?
Could be a tree. Could be a
nicked wire. The cause doesn't matter much at this point. What matters
is that the current sources to a fault are removed as soon as
practicably possible. One way to do this is to use lower-ampere rated
fuses.
What's really interesting is
how this same exercise works out on a String Combiner to Master
Combiner feeder. Use that same module and build a 100kW or 500kW
system.
Figure 7 strings per string
combiner going into a 100A Fuse on a master combiner. Figure a dozen of
those. So, 84 strings total. (Really just making up numbers here...
Use your own if you like.)
Each feeder is 8.51 Isc * 7
strings = 59A Isc. On a 100A fuse. Figure a #1AWG feeder in an 1-1/2"
pipe. Figure that conductor is derated to 110A ampacity with
temperature, etc.
Using my method, I can put a
70A fuse in there and shave 30A of destruction off any fault. Right off
the top. 30 flippin' amps might totally mean the difference between a
fire and a not fire. And that 70A fuse is still 125% Ipmax.
I might also be able to do
something else. That 30% in amperage saving just might equate to less
combiners and feeders. Let's run that scenario. Start with the 100A
fuse. Using my same 125% Ipmax ratio, I can put 10 strings in a
combiner. 10 strings * 8.51A Isc * 1.56 = 133A so that's too many for
my conductor. How many strings can I get on my conductor? 110A / 8.51 /
1.56 = 8 strings. 125% Ipmax = 79.9, which bumps me to an 80A fuse.
Which is still 20A less destruction in a fault. On the same feeder. Now
work that out... 8 strings per combiner means I now need only 11
combiners and feeders instead of 12. 10 of them would be the
8-string/80A fuse sort and the last would be a 4-string/70A (lowest in
that physical size). Which results in at least 8% savings in combiners
& feeders and a 20A safer feeder. We might be able to up the
conductor one size and use even less combiners. Any way you go it's
money and safety savings.
If you really fear that
cloud-edge increases the actual, operational amperage in these circuits
for long durations, use your own numbers. I happen to think that 125%
Ipmax is reasonable in most cases. If you are up at 10,000 elevation,
you might have a different thought. I say use what you like up to the
Series Fuse rating of the module for OCP and 1.56*ISC minimum
conductor.
I just want the right to
protect my customers and the folks who work on this stuff as much as I
can.
Sunshine Wishes!
* Even once there is such a thing as a RELIABLE,
cost-effective DC arc-fault circuit interrupting gadget AT THE SOURCE,
you will still have a fault. Being fed by parallel strings. Until that
combiner fuse blows. At which time the fault will still be a fault. But
it will be a DEAD FAULT. If this new-fangled RELIABLE,
cost-effective whatchamadohickie on the normal load side (inverter,
charge-controller, margarita-mixer, etc.) of wherever a fault occurs,
you will still have a fault when the whatchama... That thing, opens.
And this fault will NOT be a DEAD FAULT. Nosiree! It will be the SAME
EXACT FAULT YOU WOULD HAVE IN 2010. Only this fault will now have cost
millions of dollars in grief, standards-making, R&D,
early-market-hype, paychecks to marketing... Fights with AHJs and
engineers and bosses and workers... Yes, this fault will have cost a
lot and it will be EXACTLY THE SAME as you have now. (Note to anybody
who thinks otherwise: The DC AFCI MUST go as near the source as
possible!)
** Use of the term "never" is... Well...
Rare. As in, I pretty much regard the word as sacred. Please understand
that my use of the term "NEVER" indicates a strong belief on my part
that this will not happen... Under normal operating conditions.
Matt,
There is a word on the KLKD time delay chart you referenced that is
hiding a very important fact. Those are "average" curves. To
illustrate what that means take a look at the time delay curve for the CBI
breaker commonly used for battery charging systems. There is a
gray area that extends from 105% to 130% of rating. Over that range,
the breaker may or may not trip. While not shown on the KLKD fuse time
delay chart that you referenced, a very similar gray zone exists.
These fuses and breakers are simply not precise enough devices to
distinguish between Imp and Isc at STC in the best of conditions.
if you did manage to select a fuse that would blow at Isc but not at
Imp, it still wouldn't blow during a short if the irradiance was 800
W/sq m. An arcing fault at 800 W/sq m can still burn a big hole. It
simply isn't possible to provide good fault protection for PV systems
with fuses. What's needed is a ground fault protector/arc fault
protector that is located in the combiner box. Until that exists, the
feeder between the combiner and inverter needs to be considered an
unfused conductor.
Kent Osterberg
Blue Mountain Solar
Matt Lafferty wrote:
Hi Kent,
Thanks for the 98% vote. Now I'm
gonna try to get the other 2% out of you.... You're a smart guy, so it
shouldn't be too difficult ;)
These aren't the days where we
were lucky to have a customer with barely enough money to afford a 300W
system. We are commonly dealing with >2500W strings of 200+ watt
modules. It's a new paradigm and the risks associated with faults
continue to grow. You are 100% right... OCP in a current-limited DC
application ain't simple.
To be clear, I am completely in
favor of 1.56 ISC as a minimum for conductors. I want that inherent
protection and any case I describe hereafter assumes this condition.
Also to be clear, I am not proposing that our current-limited power
sources should be able to trip the OCP from the source based on
amperage alone.
To
your contention that "1.56 ISC isn't in any way responsible for the
danger"....
Compared to a lower fuse
rating.... Something like ISC for instance.... 1.56 ISC as a MINIMUM
OCP rating does indeed increase the hazards. To both persons and
property. By at least 40% in terms of raw amperage. By more than that
in terms of kCal/cm2. The biggest difference comes in terms of time to
blow and the amount of damage or injury caused during that period. They
call it Incident Energy.
For those who have not studied
Time:Current curves of commonly used fuses, you should. KLKD
is a typical fuse I hope we are all familiar with. http://www.littelfuse.com/data/en/Data_Sheets/KLKD.pdf
I'm using KLKD as an example because Littelfuse put a little table
right on the front of the datasheet to make it simple. Other commonly
used fuses have similar characteristics.
Note that this class of fuse
will take 135% of it's rating for up to 1 hour. 200% of its rating for
up to 2 minutes. Be sure to check out the fuse curves. These things
will take ~125% of their ratings for pretty much indefinitely. How much
damage is caused waiting for it to blow? What about when irradiance is
<800 W/M2? How long then? Will
your conductor hold up during this time?
If you fault on the array side
of a combiner fuse in a 3-string system, you might NEVER BLOW that
fuse. (Example: 8.4A ISC module with 15A fuse). Especially if there is
an arcing DC fault. Temperatures of arcs are much higher than 90C. Is
your conductor up to that? I have been called to troubleshoot a lot of
low-performing and broke-down systems. One resi rooftop had a spot
where an arcing fault burned through the side of a NEMA 3R steel
pull-can on the roof and never blew a CODE COMPLIANT combiner fuse.
This system had 3 strings and was down about 1/3 on power output after
the original installer replaced the GFP fuse following a
"ground-fault". He had checked the combiner fuses and they were good so
he called me to troubleshoot it. He was convinced that there must be a
bad module and wanted a third party to verify it for warranty claim.
Hated to show him what he missed and that it was his fault. (The fault
had burned clear so the GFP didn't blow again.)
Another situation that I am
sadly familiar with has been burning holes in a steel roof since 2007 (I
told them not to do it over and over)... At least one spot is
about the size of a softball. The others vary in size. Their common
characteristic is that you can see the sky thru them from inside the
building. The first time it happened was
during an ice storm. This POS peel-n-stick system typically burns thru
the roof during low-irradiance periods. Most of the time it will
EVENTUALLY blow a combiner fuse, reducing the current feeding the
fault, and either weld to a short or burn clear (open). This
is ~500kW on central inverters with marginal GFP protection. I hear the
Tefcel front sheet is a nice insulator.... You want to walk out on that
thing in the daylight? I don't. This case happens to be a classic
installer-was-either-drunk-or-on-drugs-because-nobody-can-be-that-stupid
situation. Rolling UniSolar right over loose screws on the pan? WTF!!!!
I
can list a lot of similar examples where damage has been caused and
fuses have either not blown or taken longer than they should have to
blow. The key here is that the wiring
and OCP in ALL OF THESE CASES ARE CODE COMPLIANT!
I bring these cases up because,
if these things had fuses with smaller current ratings, the fuses would
blow before this much damage is done. It takes a lot of heat to burn
thru steel. You wouldn't think that a circuit with a 10 or 15A fuse
could do this much damage, but they do! Amperage = Heat. The more Amps,
the more Heat. When you use an Arc or Wire-feed welder, you adjust Amps
to get the heat you want. I want to minimize the amount of potential
Amps back-flowing into a fault to a lower level. A level that allows
safe and reliable NORMAL OPERATION, yet limits the catastrophic effects
in an ABNORMAL CONDITION.
My contention is that the NEC is
flat out wrong requiring 1.56 ISC as a MIMIMUM OCP rating. It creates
undue hazard. It is in direct conflict with the spirit, intent, and
other long-standing precedents in the Code.... With the exception of
Emergency Fire Pumps and other mission-critical equipment that is
specifically intended to stay alive until it completely burns to the
ground, all other minimum OCP ratings
are based on 125% CONTINUOUS OPERATING CURRENT of the equipment. In our
case, that equates to 125% Ipmax, as opposed to ISC. An example of an
AC equivalent to this asinine "minimum 1.56 ISC" OCP requirement would
be requiring motor circuits to be OCP at 156% of the Locked
Rotor Amps. I'm sure you can imagine what this would do to the size and
cost of starters, etc.
Since the last time my favorite
6kW system put out 7.5kW was... NEVER.... I'm gonna have to guess that
a 1.25 Ipmax fuse would hold just fine. This would be 100% consistent
with the rest of the Code. It's also not gonna blow with cloud-edge
effect or other irradiance enhancing events. Especially when you
consider that it's gonna automatically withstand an extra 20-25%% for
an indefinite, possibly forever, period due to the Time:Current curve
relationship.
My
NON-CODE-COMPLIANT-SELF prefers to size OCP at "ISC or next larger
standard fuse rating not larger than the LISTED Series Fuse Rating of
the module". You've got a 6-7% headstart between Ipmax and ISC
plus the ~25% indefinite Time:Current characteristic. Ain't no way in
hell that a NORMALLY OPERATING SYSTEM will blow fuses rated this way.
And it's ~2/3 the amp rating (or less) that is now required as a
minimum. Which is exactly what we want.
Reliability during normal operation and safety.
The only time we want an OCP
device to trip is during abnormal conditions such as a short circuit.
When we have this type of fault, we want that circuit to open up as
quickly as possible in order to minimize damage. At least I do. In the
case of 1.56 ISC, the NEC is GUARANTEEING GREATER DAMAGE AND INCREASED
HAZARDS compared to a lower-amp fuse. I have gone thru this logic with
numerous building inspectors over the years. Every single one agrees
with it. Some, but not all, have agreed to allow lower-amperage fuses.
The only, and I mean ONLY reason given by inspectors that have not
allowed lower-amperage fuses is because... "The Code requires 1.56
ISC so I have to require it."
String inverters only bother me
so much in this regard. Central inverter systems are where the real bad
ju-ju starts to happen. For the sake of definitions, I
consider a string inverter to be one that has one or less combiners and
the modules are configured in series strings. A central inverter is one
that has string-level combiners and one or more re-combiners.
I am seeing more and more faults
in the DC feeders between string combiners and re-combiners in these
central systems. The power levels you are dealing with here are pretty
significant. 100% of the recent faults I've been seeing are due to
shi##y workmanship, particularly in conduit installation and
wire-pulling. Some of these faults would certainly have been avoided if
they had selected a tougher insulation such as XHHW-2, instead of
Quik-Nick THWN-2/THHN. (THWN stands for "This Heiffer Will Nick".
THWN-2 stands for "This Heiffer Will Nick 2wice")
I have zero tolerance for crappy
workmanship and even less sympathy for the people who do it. Just
got off ANOTHER call this morning where the installer has re-pulled one
DC feeder four times and still can't pass megger testing. They have
re-pulled every feeder at least once. The spec only calls for 250 Mohms
even though wire and cable engineering formulas say the minimum should
be 688 Mohms for that size, length, and insulation type of wire. And
they can't even get it to 250. Every
set of wires that has been pulled out has obvious physical damage. The
sub is crying, wanting more money and to have the work accepted (NOT!).
Come to find out, one of the field guys working for the developer has
witnessed these guys beating on the 1/0 AWG with a mallet to get it
into the LB.
Says right here in my NECA
1-2006 Standard for Good Workmanship in Electrical Construction...
Section 9 Wire and Cable.... "c) Wires and cables shall be installed so
as not to damage the insulation or cable sheath." Sounds like this
electrician sub wannabe is in violation of his contract.... You know
that clause... "Workmanlike manner". (Sub, if you are reading
this... I am NOT your friend in this case. You WILL re-pull these
feeders correctly, at your own expense. I will repeat the advice
already provided: Use pulling condulets. I will add some advice: Fire
your electricians.)
The scary thing is, this
practice goes on every day. A LOT! The
sad fact is that many (most?) of these systems are not having thorough,
comprehensive Insulation Resistance Testing performed. And IRT will
only catch SOME of the future faults! I have been involved with
post-mortem in two cases where 500kW AC feeders have been properly
IRT'd and blew up later. Not good! (Side Note: Each of these cases
involved big feeders in standard LB's. Make a note of it.)
It's
only a matter of time before these things go Pop Sizzle Smoke! These
failures WILL occur. A lot of them already have and the number is
growing. I hope and trust that most of us on this list practice Good
Workmanship on every project. That being said, none of us are perfect.
What about cases where we miss something or even cases like a tree
falling across one of our conduits?
My contention is that we should
do whatever we RESPONSIBLY can to minimize the damage when this happens
and the hazards when it's being troubleshot and repaired. It's a simple
principle.
One RESPONSIBLE way we can
minimize the damage is to reduce the fuse size by ~40%. (i.e. ISC
or next higher standard fuse rating) This method will provide
adequate operational reliability. It will also ensure
that there is a better chance of the fuse blowing sooner when there is
a fault, thereby minimizing the damage caused. It will minimize the
hazard to personnel performing troubleshooting and repair because the
incident energy at the fault will be reduced in all cases. By ~40%.
Whether or not the fuse blew.
Happy to discuss this issue with
all who care and are not on NFPA 70 CMP. (Just kidding. You CMP guys
are welcome to discuss it, too... Just be ready to issue a
memorandum/addendum to the 2011 NEC allowing OCP with lower than 1.56
ISC...)
Extra Credit for BOS Mfrs: Make
a Combination Device that has DC Arc-Fault Interruption and OCP that
fits in a standard fuse configuration. Start with midget-class so we
can simply drop it into our string combiner fuseholders.
Pray for Sun!
Matt Lafferty
Matt,
I agree with you on about 98% of this. You are 200% correct that a
faulted high-voltage or high-current PV array is a serious and
dangerous situation and that the person looking for the trouble in a
faulted PV array needs the proper tools and knowledge of how all the
components work. But the 156% rule for fuse sizing per NEC 690 is not
in any way responsible for the danger. The danger is a result of the
nature of the PV module: a power source with the current nearly
proportional to the illumination and a short circuit current that is
only 10% greater than the normal operating current. If one were to
select a fuse that could blow when the array was shorted, occasional
edge of cloud irradiance enhancement would cause nuisance trips and it
still wouldn't clear a fault when the irradiance is 900 watts per
square meter. There will never be a simple fuse that can provide the
protection that is needed. The existing ground fault protection in the
inverters is inadequate and current plans for arc fault protection may
not be a satisfactory either. These have been slow incremental
improvements; much more is needed.
--
Kent Osterberg
Blue Mountain Solar
541-568-4882
www.bluemountainsolar.com
Wrenches all,
I 100% second Bill B's comment Correct that... I 200% second it. It should
be the law.... "Don't begin to troubleshoot a faulted PV circuit without a
reliable DC clamp meter."
The MOST DANGEROUS PV system is a wounded PV system. This includes danger to
persons and property. Safely and efficiently troubleshooting a faulted PV
circuit requires a voltmeter AND an ammeter. And PPE. And adequate knowledge
and understanding of operational and non-operational characteristics of PV
systems.
The simple reason for this is that, when one or more circuit conductors are
faulted to a short condition, the voltage between the faulted elements is
zero. Relying on just a voltage reading to determine whether or not to open
a circuit under this condition will result in an arc. The amount of energy
in that arc will depend on the amount of available sunlight and the amount
of PV that is feeding into it. The amount of potential hazard will
correspond to these factors as well.
Using a clamp style ammeter will allow you to understand where and how much
current is flowing in a circuit before you decide to open it. It is one
thing to know you have a 45 amp load in a circuit with a potential of ~450V
because you clamp it before you break it. With this knowledge you can assess
the situation. You can do something to mitigate or remove the potential
hazards... Cover the array, open a disconnect somewhere, put your PPE on and
go for it, select a different location to open the circuit, use insulated
cable cutters, wait 'til dark.... You have choices.
It is quite another to be surprised by the resulting arc in tight quarters
because you measured the voltage and figured it was a dead circuit! When you
react to the startlement (word?) by dropping your screwdriver and yanking
your hand back... Assuming you don't receive a shock, flash injury, or fall
off the roof in the process, of course.... The result just may be that the
now-dislodged conductor is arcing and zapping and spitting. Now you're gonna
have to stick something back into that box to deal with it. In the meantime,
a number of possible things can happen, most of which are not favorable....
Melting insulation and conductor material are the most common. The degree
(not just a pun) of damage and remaining hazard will be determined by the
amount of sunshine and amount of PV feeding into the arc.
The MOST DANGEROUS single point on the DC side of a PV system is ANYPLACE on
the Inverter side of a fuse(s). This is a simple function of the assinine
"1.56 ISC minimum fuse" rule in the NEC. The source cannot create enough
current to blow the fuse(s). If you have a fault between a combiner and the
inverter, you WILL have current flowing into the fault as long as the sun is
up! If you are relying on just a voltmeter in a central-inverter plant, you
could very well be in for a 15-20kW surprise, or greater!
The combination of shi##y wire, sloppy conduit installation, and crappy
wire-pulling methods have resulted in too many DC feeder faults to count. It
boggles my mind every time I hear of yet another guy nearly joining the dead
because he touched or opened up a connection somewhere in a faulted circuit
without de-energizing it. Time and time again I hear that they tested it for
voltage and it was "dead". Sometimes they even opened up the service
disconnect at the string combiner, "just to make sure". Time and again it's
a "journeyman electrician". I like it best when it's the same card-carrying
jackass who "built" the thing.
I consider THWN-2 to be on the list of shi##y wire types for DC, by the way.
I'm an XHHW-2 guy, personally. Why would anybody select an insulation that
is easy to nick/slice/tear when you can have a super-tough insulation for a
couple pennies more? Why would anybody select an insulation that only has
about 5% of the dielectric resistance of one that is a couple pennies more?
Why? Oh, I know... It's that race to the bottom on BOS costs...
Which leads to the next step in stupidity... Designing and building LARGE PV
plants without sufficient DC SERVICE disconnects... This is what's going on
out there.... PV plants with 500kW Central-inverters being installed without
string-combiner disconnects. Without any DC service disconnects.
The NEC considers the fuseholder in the combiner &/or the connector on the
module to be a "disconnect" and does not require a "service disconnect" in
the circuit. So these smart-ass engineers and project developers are out
there building this shi#. Some of these projects are being built by PV
module manufacturers masquerading as developers. "Vertically integrated..."
Others are being designed & built by formerly respected integrators who have
either sold out or lost their conscience altogether. The trend is to build
them to sell to PPA companies who ostensibly own and "operate" them. These
solar timebombs are being built on both sides of the fence. Frosty ain't the
only one with a solar flamethrower!
All in the race to the bottom of the $/Watt pile that they are now calling
LCOE. Har Dee Har Har!
I hate to say this, but I hope somebody gets really hurt out there, and
soon. I hope it's the same smart-ass engineer (or his boss) who thought it
was alright to design this way after some field technician walks away from
it because it's dangerous. And then I hope his family sues the crap out of
the company and companies involved with designing, supplying, building, and
owning it so they STOP DOING THIS SHI#! And then I hope he takes his cooked
carcass on the road doing safety awareness training so others don't repeat
these stupid, avoidable catastrophes! And then I hope these cheap-ass
developers go out to every site that doesn't have sufficient disconnects and
re-fits the systems with them to avoid further injuries and $$$$
settlements. What is the levelized cost of energy for that system now, Mr.
CFO?
Unfortunately it isn't likely to be that smart-ass engineer. Or his boss. It
is far more likely to be a Wrench. A Wrench without a DC clamp and the
knowledge that he needs one. A Wrench without the proper PPE because he
"tested it and it was dead" so, even if he had his gear on to "test it", he
took his gloves and face-shield off to work on it. A Wrench who doesn't
fully understand the operation of GFP circuits. A Wrench who doesn't
understand that not all faults are ground faults and the characteristics of
a fault change in terms of potential and magnitude with varying
environmental conditions. A Wrench that doesn't fully understand that power
can be coming from both directions. A Wrench who figures he doesn't have the
time to completely isolate a section of a circuit because there AIN'T NO
REAL DISCONNECTS. I hope it's not your Wrench.
As the size of the inverter grows, so does the hazard. To a point. The
idiotic 1.56 ISC rule only increases the potential hazards. Central-inverter
plants should not be serviced by anybody who doesn't have an extremely
comprehensive understanding of these systems, and the tools and PPE to
safely work on it. For systems with inverter-integral re-combiners, the most
dangerous spot in these systems are the feeders between string combiners and
re-combiners. Anything between the output of a string combiner and the input
of a re-combiner. For systems with standalone re-combiners, a fault between
the re-combiner output and the line side of the next disconnect is the most
dangerous point, but certainly not the only dangerous point. If either of
these systems are built without load-break disconnects at the
string-combiner level, the cost to service goes thru the roof. It either
goes thru the roof to do it safely or it goes thru the roof in terms of risk
to do it not safely. Pick one.
There is an interesting dynamic between the potential hazard on a faulted DC
homerun feeder and the kW of the inverter. The less re-combiner inputs you
use, the greater the potential hazard on faulted input feeders. Again, this
is because of the UNSAFE AND STUPID 1.56 ISC rule. In systems with a
relatively low number of re-combiner inputs, there are large portions of
time when there isn't enough combined amperage in the non-faulted feeders to
blow the re-combiner fuse of the faulted feeder. If your system only has 4
or 5 re-combiner inputs and it's winter-time, it is quite likely that a
faulted feeder is being fed from both ends. (Commonly 100A fuses in the
re-combiner with ~60A ISC feeding a string-combiner) That feeder can be fed
from the re-combiner end, by anything up to about 105% of the fuse rating,
for pretty much ever without blowing the fuse. The more parrallel inputs
there are, the more likely there will be sufficient current generated by the
other feeders to blow the fuse. Since the vast majority of systems out there
don't have load-break disconnects at the re-combiner inputs, the technician
needs to be able to open disconnects at each string combiner in order to
isolate this feeder. But what about systems without DC service disconnects?
Repair at night?
My hope is that anybody on this list will refuse... Say it with me now...
R-E-F-U-S-E to install PV systems without adequate disconnect provisions to
isolate faulted feeders. And only allow technicians with proper knowledge
and equipment to work on a busted PV system. "Journeyman electrician" does
NOT automatically mean that person has the proper knowledge to do it safely.
Safely working on a faulted PV DC circuit requires ALWAYS clamping the thing
for starters. It might also mean "not working" on it until the sun goes
down. A technician with the proper knowledge and equipment should be able to
determine the proper course of repair.
In the case of the faulted lightning arrestor, it was "only" a small
circuit, but it got the guy's attention and apparently nobody got hurt. The
bigger these systems get, the bigger the potential hazard.
To answer Tom's question about jumping around a fault: Maybe, maybe not,
depending on the nature of the fault (+/-, +/G, -/G) and the location of the
jumper relative the fault and the power source. Even if jumping to ground
eliminates the arcing when you are working with the terminal, you will still
have arcing when you land/un-land the jumper &/or remove the fault. If the
sun is shining and you have a DC fault, you will have arcing at some point
when you make/break the circuit. Hopefully it's safely contained and
localized to the contacts of a service disconnect!
Pray for Sun!
Matt Lafferty
