John Gilmer wrote:

A short circuit will trip the breaker but this is an arc. Ideally the
breaker would trip "instantaneously", but that takes a lot of current.


A lot? With a 15 amp breaker, a 30 amp arc should cause a trip within a
few seconds.


Might help if you read
http://www.cpsc.gov/volstd/afci/AFCIFireTechnology.pdf
hereafter called "the paper".

The paper quotes the UL standard as only requiring a trip in 2 minutes
at 200% of rated current.

Looking at the time-current curve for SquareD breakers, at 200% a trip
can take 40 seconds.

Breakers work on a thermal element to trip and it takes longer the lower
the overload - hence "inverse-time element". They also have a magnetic
trip for "instantaneous".


The available fault current in an extension cord may not be high enough
to trip a breaker on "instantaneous".

Oh!

Now we are speaking the same language. Is the NEC worried that some of
that crappy lamp cord has enough resistance to limit current flow to less
than, say, 30 amps when subject to essentially a direct short?


From the paper: UL field tested 1,590 receptacles in 80 dwellings to
determine the available fault current at the receptacle. 16% of the
receptacles would likely not trip on "instantaneous" with a short at the
receptacle. 44% of the receptacles would not trip on "instantaneous"
with a short at the end of a 6' of #18 cord.

For a SquareD breaker, a 15A breaker may require up to 11 times the
rated current to trip "instantaneously". That is 165A for a 15A breaker.
More than 16% of the surveyed receptacles would not supply 165A.

Not tripping on "instantaneous" puts the trip on the inverse-time curve.
And this was for a short. An arc reduces the current further, slowing
the trip. "Crappy" #18 is widely used for extension cords and appliances.


If it is high enough, the arc is
not necessarily continuous and the breaker still may not trip on
"instantaneous". And an arc is not a short circuit - the current will be
less than the available fault current.


An arc is quite close to being a short circuit. To a good approximate, an
arc can be modeled as a "short" with a fixed voltage drop that stays about
the same regardless of current. That voltage drop is only a few volts but
even if it were, say, 20 volts, the line cord would have to soak at 100
volts and to get a 100 volt drop in a line cord would take enough amps to
trip a breaker.


An arc may involve material that has been carbonized - which can run on
a lower current. And "parallel arcing faults have erratic current flow"
which reduces the current value.


That often leaves the breaker in
its inverse-time mode. With a constant load of 30A on a 15A breaker, for
example, the breaker may not trip for 2 minutes.

That long?

Yes - see above.

OK, then why not require breakers that "magnetic trip" at a current closer
the the rating for bedroom circuits? Or require a shorter "heat" trip
time?

Lower "instantaneuos" ratings can cause nuisance trips for motors and
lights which draw about 6x full load when they start. This is covered in
the paper - it was considered.


The time delay likely
with an arc leaves opportunity for the arc to start a fire. AFCIs were
developed to detect an arc and provide a fast trip.

The NEC requirement for AFCIs was based on research done by UL for the
CPSC. When the research was done AFCIs didn't exist. The CPSC, UL and
the NFPA (author of the NEC) think parallel AFCIs can prevent a
significant percentage of fires with electrical causes, estimated at
40,000 per year.


Just WHAT is that "significant percentage?"

If it's only, say, 10% then I say the AFCI is a WASTE.


I would say 10%, 4,000 fires a year, would be well worth it. Consider
deaths and injuries. And the cost of medical for burns and cost of
building loss, which of course we all pay for.

There was a "cost-benefit" analysis done - so the requirement for AFCIs
in bedrooms had some grounding, if you'll excuse the expression.

I think the extension to all 15 & 20A circuits is a lot more
questionable. Particularly since the new series/parallel devices are
barely on the market (if at all) 10 months before they are required for
widespread use.

Incidentally - parallel arcs were considered more dangerous than series
because the current availble was much higher than a series arc.

--
bud--