"jack" wrote in message
Sealing the top doesn't seem right. The one time I had broken pipes, they
were drained, but there was some residue left in low spots. They had
plenty of room to expand into.

I thought a better test would be to put a right angle bend on the pipe;
then fill it and let it freeze. The right angle would reduce the ice's
ability to expand, and would be pretty much like real plumbing.

4 hours at zero and it didn't do anything but freeze. It didn't even
swell.

Apparently my simulation is deficient. Maybe the bottom of the L has to
be much longer?


If you don't seal the water in some manner it will move to the outside and
not exert pressure on the tubing walls. The water in low spots of your pipe
is a different scenario of a test pipe in the freezer. At some point the
water at the ends freezes and makes a plug. Later the center freezes and
exerts pressure in all direction but since the shallow water at the ends
froze first, it can burst the tubing.

You may find this interesting too. Hot water pipes can freeze before cold
water pipes.
http://www.gi.alaska.edu/ScienceForum/ASF4/440.html