This section is from the book "Modern Buildings, Their Planning, Construction And Equipment Vol3", by G. A. T. Middleton. Also available from Amazon: Modern Buildings.
As the single main carries both flow and return waters, and all there is of them, it follows that it must remain its full size from beginning to end, and cannot be graduated as is possible with the two-pipe system.
The horizontal main is carried around the basement space under the building, passing nearly beneath the points where radiators are to come above, but it should not come immediately beneath the radiators. Assuming that it is a square-shaped building, requiring radiators against the outer walls in rooms or offices, then the main would be suspended from the basement ceiling, about 4 to 6 feet distant from the walls, for a reason connected with the radiator branches, as will be explained. The main circulation is given a rise from the boiler of at least 1 inch in 10 feet, as with other systems, but as it is one continuous pipe there has to be a point chosen for the main to reach its greatest elevation, this being, as a rule, about half-way round the circuit. In Fig 47 this "highest point" is marked, and the flow is given a rise from the top of the boiler to this point, while the return descends from here back to the boiler.
As stated, the highest point in a one-pipe horizontal circuit is, as a rule, arranged to come about as far out from the boiler as possible, thus making the flow and return of the circuit of about equal lengths. It does not matter if the flow is a little longer than the return, but, theoretically, if there is any difference at all, the flow should be shorter and have the least radiation on it. It is as well to arrange it so when possible.
At the highest point the expansion pipe is usually placed to relieve the circuit of any air that may collect in it, which would naturally gather here. It is essential that there be a way for the air to escape out of the pipe, but if the "highest point" and the expansion pipe cannot be located together, then the expansion pipe can be placed elsewhere on the flow-pipe, provided a radiator is placed on the high point for air to collect in, and to be allowed to escape periodically by opening its air cock. As previously stated, the collection of air in these works is not great after the first heating up, and although it may be sufficient to have a bad effect at the high point in a pipe circuit, it is not enough to be troublesome if vented into a radiator. The latter will take ten times as much air as a pipe before its efficacy is impaired.
The customary method of connecting and running the radiator branches is shown in Fig. 48, and this illustration will show how it is that the main circuit does not need to be run immediately beneath the points which the radiators are to occupy. The reason for keeping the mains from directly under the radiators is to admit of making what are known as "Swing connections." Long straight runs of pipe, such as these mains may be, give a very distinct push and pull (lengthways), as they expand and contract with heat and cold, sufficient in any fair sized installation to seriously strain, if not break, connections made with straight branches which are more or less rigid. The method of connecting, illustrated by Fig. 48, is certainly more expensive than straight branches, either vertical or horizontal; but unless the job be one with short mains, or mains that are continually breaking away from a straight line, this method of running the branches must be followed.
It will be noticed that one branch, the flow or that nearest the flow connection of the boiler, is taken out of the top of the main by a tee looking upwards. This is to ensure the radiator being served with the hottest water from the top of the main. The return is connected horizontally with the idea of delivering the used and cooled water as low down in the main as possible. The idea is a correct one, but it must be admitted that many jobs are done with both tees looking upwards, and the result appears to leaving nothing to be desired.
A horizontal connection that can be recommended is that to the flow branches of the first fourth of the number of radiators on a circuit. Thus, if twelve radiators were being connected to one line of pipe, it is a good plan to let the first three out from the boiler have horizontal connections to both their branches. It will be quite understood that the first radiator has the hottest water delivered to it always and in any case, so that a horizontal connection delivers as good water as any other. With second and third radiators on a fair sized circuit the water in the main is practically as hot as when it passes the first radiator, so that these again cannot be served badly by horizontal connections. The advantage gained is with the later radiators, as it may be assumed that some of the hottest water travels along the top of the pipe past the first three pairs of horizontal tees, and so arrives at the fourth radiator as hot as it came from the boiler. By this means the fourth is served much like a first radiator ordinarily is. Some such plan of getting the hottest water quickly along to the distant radiators is worth following, as a little fault which this one-pipe system has is the considerable difference in temperature between the first and last radiators when lighting up or heating up at any time.
When all are hot, the radiators heat with agreeable uniformity (if all the conditions are correct), but, as stated, when heating up the first radiators take all the hot water and keep the end radiators cool for some time. A good plan for obtaining uniform heating amongst the radiators is to divide them up into two or more circuits. When the radiators all stand around the outer walls of a square building there is always a temptation to run a single circuit to take the lot, but when the work is more broken up the circuit can be broken up also with advantage. Fig. 49 will show what is meant. In this, one circuit takes five radiators, while another consists of one flow with two returns, these twin circuits taking nine radiators together. In this work, as with the two-pipe system, there is rarely much objection to branching a flow main, but every effort should be made to let the returns enter the boiler separately.