This section is from the book "American Plumbing Practice", by The Engineering Record. Also available from Amazon: Plumbing: A working manual of American plumbing practice.
A Plumber, Liverpool, O., writes:
"I should feel obliged if you can explain the following occurrence. The accompanying sketch is a diagram of the water supply (hot) in a house. There are two lines of return circulation taken off the expansion pipe. The returning ends of these lines join together, and, as one pipe, are branched into the flow pipe to water-back (from cylinder).
"When the stop-cock is open one can only draw cool water from the cylinder, although the upper part of it may be quite hot; so that it appears that the water from the bottom of the cylinder is drawn at the taps instead of from the top of the cylinder. Upon closing the stop-cock hot water only is drawn at the same taps. What I wish to know is, by virtue of what law does the flow of the water reverse itself when the stop-cock is opened and shut ?
"The water level in tank will be about 35 feet above the top of water-back.
"The point A on expansion pipe will be about 3 feet 6 inches below water level in tank. Cylinder holds 50 English gallons."
[The fact that water will flow most readily in the direction of least resistance seems to be the solution of this question. The water flows from the tank downwards through the pipe a, and enters the boiler at the bottom. Should the stop-cock be closed, the water to leave the boiler (cylinder) and go to either faucet must go by the pipes b and bl, and hence only hot water is drawn. Should the cock, however, be opened, the circuit by the pipe a through the bottom of the cylinder, and thence to the pipe c, cl, and c2, are presumably much the shorter than by the pipe b and b2; hence the flow of cold water at the faucets.
If your pipes throughout were of large diameter, excepting the return pipes, this probably would not occur, but they would have to be of sufficient size to supply the warm water by circulation in the proper direction. Several possible contributory causes may be given in addition to those caused by friction: (1) There may be partial obstruction caused by solder at a joint. (2) If the loss of head within the pipes when drawing freely at a faucet is sufficient to lower the water below the point A, then no hot water can be expected at the faucet d. though it might still run at the faucet e. If it does not run warm at the faucet e when drawing slowly, then our opinion is a partial stoppage will be found between the cylinder and the branch bl If the stoppage exists, remove it and then "choke down" the stop-cock until it is only sufficiently open to maintain circulation of the water at the temperature required. This will hold back on the flow of the water by the return pipe c, but if this is not sufficient, to put a swinging or easily worked check valve in near the stop-valve.]

 
Continue to: