On 5/4/2012 7:57 AM, TomR wrote:
chaniarts wrote:
On 5/3/2012 12:17 PM, Sonny wrote:
If you have to dig, I would dig around each faucet first. Maybe one
of the faucets got hit by something and the leak is at one of the
elbows coming up or otherwise associated with one of them... i.e.,
the most likely spots to be damged by some above ground accident.

I'm not sure there's a way to tell if there's a leak along a long run
of pipe in the ground, but if you have to dig, maybe dig only every
20 feet, or where there's likely to be a joint in the piping.... My
second guess at where a leak would be is a joint.

Is there a driveway, of some kind, along the pipe's path? Maybe
driving over that area caused the leak at that spot. A tree root
nearby, that may have caused a break in the line as it grew larger?

Sonny

plastic pipe?

binary search. dig somewhere in the middle. cut and install a valve.
turn off. does it still happen? if so, it's somewhere between the
valve and meter. if not, it's the other half.

repeat until you get down to a length where it's not such a burden to
dig up a longer length of pipe to find out exactly where.

Interesting idea.

Now my mind can't help doing the math and wondering if a 1/3 -- 2/3 split
instead of a 50-50 split would work better. I don't know, but I am
imagining that every time a 1/3 -- 2/3 split is done, there is a 50-50
chance that the leak will be on the 1/3 section, which would mean the leak
would be narrowed down to a smaller section half the time. I know I am just
daydreaming here, but your approach did make me start to think about that.



you need to go back to your probability class.

a 1/3 -- 2/3 split is done, there is a 50-50 chance that the leak
will be on the 1/3 section

this is not correct.

you can find a name in a thick phone book in 14-15 searches. your task
can probably be done in no more than 4.