Re: Emergency Water Turn-Off?

["Robert Green" <ROBERT_GREEN1963@xxxxxxxxx>]

"Matt" <mattmorgan64@xxxxxxx> wrote in message

> Yeah, I gotta agree, Robert.
>
> But, I think it would be fun to play around with!

No doubt about that!  I've got a bunch of fun projects.  I want to be able
to tell when FedEx has pulled up to the curb.  If I can sense their arrival
when I am in the basement, I can get to the door before the "We tried to
deliver" tag gets looped on the handle.  I set up cameras, shot gun mikes,
all sorts of stuff.  Eventually I settled on putting a microphone and a
camera on a lamppost by the front walk.  When I thought I had it working
perfectly, the FedEx guy delivered a package to my neighbor and then walked
across the lawn between our houses to ring my bell.  That's when I gave up!
You can't imagine how PO'ed I was to see the truck driving off, knowing my
system had failed me.

> I once worked for an industrial automation company - that
> company sold some major clients on some cool stuff that
> worked well (automated tracking sides of beef from the kill
> floor to the freezer) - to stuff that was an absolute multimillion
> dollar failure for the poor companies that bought them (using a
> machine vision system to determine if a keg of beer was leaking).

Sounds fascinating.  I think anyone who's been involved in the automation
field has similar stories.  The real world will apparently do what it
pleases, no matter what engineers or automators expect of it.

> I dunno, I think if you could get a fairly accurate flow meter,
> combine it with reliable indicators of whether or not people
> are in the rooms that should use water, and combine that with
> a schedule... I think the problem could be solved to a reasonable
> degree of accuracy.

That's the problem.  Accurate flow measurement.  Perhaps the mike on the
pipe will turn out to be just as good for that because water does make noise
going through a pipe.  I suppose I would start by strapping a mike to the
main valve's feed pipe and running that into a PC recording program like
CoolEdit that has an elapsed time indicator.  Then I would run around the
house with a pad and a clock, open various faucets just enough to fill a
gallon bucket about an hour and log various flushing, bathing and kitchen
water use events.  Then I would see what I could match on the CoolEdit log
to my own test log.

> Maybe that's part of the problem for the folks like watercop - it's
> designed to be used on it's own; it doesn't have the benefit of getting
> the data that a mid to upper class HA system can provide.

I think you're right.  In order to have mass-appeal, and, I expect to
increase reliability and predictability, they've gone with the water leak
detectors in critical areas.

> Oh well, I dunno either way. I do think I can safely say an acoustic
> solution can be ruled out.

I wouldn't rule it out until I performed the above test.  I just happened to
have moved a old PC with CoolEdit to the basement.  It would be simple
enough to try.  I also have an outside spigot with a leak that's shutoff
from the inside valve.  That would make a good "real world" sort of test of
an actual leak.

> I was thinking about that on the way
> home, and given the number of different materials that can be
> used (copper, pex, galvanized) for water supply, and the fact that
> their accoustic properties _must_ change over their lifetime, and
> probably over the seasons as well, just make it seem infeasible.

Hmm, hadn't thought about that complication!  It's just a reminder that
there's very little that's really trivial in the HA field.

> But if the OP gets something going, and wants a house to beta
> on - I'm game.

OK.  We'll use your house, Matt, to make sure it can detect someone smashing
the outside water spigot to simulate a car crashing into it.  "Can you hand
me that 15 pound sledge hammer?"  <grin>

> I just think it would be great fun to putz around with.

Yeah, this thread might inspire to do a little testing with the acoustic
method.  It may turn out that a truly sophisticated system will model human
leak detection:  Looks wets, sounds wet, feels wet.  There's a leak!  By
that I mean having moisture sensors, acoustic sensors and a flowmeter all
working together to determine if a leak has occurred:

IF FLOW > 10 minutes  AND floorwet=.t. THEN CLOSEVALVE!

> Then again, I'm putting in an old 1A2 multiline telephone
> system in my house next week..... with only one actual phone
> line - just so I have a phone(s) with blinking and flashing lucite
> buttons.

I've still got many of those floating around along with a dozen or so punch
down blocks and loads of the 50 conductor phone cable from when an office I
worked in got remodeled.  Bell didn't want it back so I took it.  The wire
makes great ties for plastic bags but at twice the diameter of RG6, it's not
useful for much else.   It's hard to believe how much phone wiring has
changed.

--
Bobby G.