Most girders are made of two plates, one on either side of the stays which they have to support (figs. 366, 367, 368, 369, 370). These are connected to each other by rivets and distance pieces, and under the nuts of the stays.

caps are fitted

Instead of riveting the plates together they are sometimes welded at their ends. Cast-iron feet at the ends are also used.

Sometimes they are forged out of the solid (fig. 371) or are made of cast steel (fig. 372).

In some works the girders are arranged alternately with an odd and an even number of stays, the pitch being measured diagonally. The saving in weight, however, is too slight to balance other inconveniences.

Angle Iron and Web Girders. Instead of girders with stays, angle irons (fig. 373), and even beams (fig. 374), are sometimes riveted

FIG. 373

FIG. 374

to the combustion chamber tops, but this is a bad practice, as the plates generally crack under the doubled parts, in the same way as they do under palm stays (see fig. 37, p. 24).

Another plan, and one which is coming into more general use, is to flange the combustion chamber plates, and rivet them to vertical

U 2

webs (figs. 375 and 376). The ends of these vertical plates have to be forged with projecting feet (fig. 384, p. 293), which tuck in under the

FIG. 375

оо

FIG. 376

FIG. 377

combustion chamber top plate flanges. It is not difficult either to rivet these seams or to caulk them from the fire side.

Some other shapes of girders are mentioned by D. S. Smart (C. E.,' 1884, vol. lxxx. p. 132).

Fig. 377 shows a corrugated combustion chamber top which is occasionally fitted; its strength can be calculated with the help of a formula on p. 205. Occasionally the tops are rounded, as in fig. 378.

Suspended Girders. Instead of allowing the girders to rest entirely on the corners of the combustion chambers, they can be suspended by stays to the boiler shells (figs. 379, 380), or their ends need not rest on the tube plate at all; but in either case the end stays should be kept far away from the flanges, in order that the expansion of the shell diameter, which amounts to about 1 in.,

FIG. 378

О

FIG. 379

should not open their seams. The top ends of the suspending rods are attached to double angle irons, riveted to the shell. The top ends are flattened out as in fig. 381, so that they may be secured by several

rivets, but generally they end in an eye and a bolt, with a split key passing through their projecting end (fig. 382). Sometimes these stays are tapped both into the combustion chamber and shell plate.

FIG. 380

FIG. 381

FIG. 382

FIG. 383

FIG. 384

Vertical Stays.-It also happens that girders are dispensed with entirely, and every one of the combustion chamber top stays is carried up to the shell. This arrangement is carried out as shown in fig. 382.

The reason for fitting these stays to the shell is, in most cases, to relieve the tube plate of its load. But this relief may be carried to excess, and may lead to the stretching of tube plate holes if the stays are fixed near the flanges. (See p. 34.)

FIG. 385

Plate Stays are shown in figs. 383, 385; they are carried to the shell plate (figs. 383, 385) and riveted to angle irons. Fig. 384 shows how the two ends of the vertical plate have to be stumped up so as to fill the roundings at the end.

FIG. 386

These plates are usually fitted crossways (fig. 383). Holes are provided, through which the steam-space stays have to pass. For very deep combustion chambers, the top has to be supported by

FIG. 387

several webs, but their number depends on the thickness of the plates; see also s.s. Iberia,' ' Enging.,' vol. lvi. p. 207.

The Steam-Space Stays are sometimes tapped into the plates (fig. 386), and sometimes fitted with nuts on either side of each plate

(fig. 387). Both arrangements have their disadvantages. In the one case the stays have to be swelled at one or both of their ends and threaded, then screwed into the plates, caulked both inside and out, and nutted. After being in use for some time the diameter at the front end will be seriously reduced by corrosion, as indicated by the dotted lines (fig. 386), and the stays must be renewed. This is not the case with those stays which have been threaded without swelling, for the exposed part of the stay may be very seriously reduced before making it weaker than the screwed part; in this case the thread at the front ends should not extend beyond the inside nut. The objection previously urged against fine threads does not apply to nutted stays (see p. 289).

The trouble with these stays is that they leak, for it is only grummets and washers that can be used to prevent this. The hollow space is sometimes filled with sheet iron and caulked, but better results seem to be obtained with asbestos packing. The outside washers for these stays are very often riveted to the plate, giving it a better support. In confined spaces it may be necessary to fit the stays in two lengths. This is shown in figs. 388 and 389.

Boiler Tubes.-One of the last things to be done to a boiler is to fit the tubes. This is a simple matter: the tubes arrive cut to the right lengths and probably also annealed at their ends; they are passed through the two tube plates, and their ends are expanded. Formerly this was done by means of conical drifts, and some people still advocate their use, but the common practice is to use expanders. These consist of several-usually three-small rollers, partly projecting out of the circumference of an iron case. A taper mandril is place in the centre and driven in while being turned round, thereby causing the rollers and their case to revolve, at the same time exerting a pressure on the tube, which expands and gets firmly bedded against the metal of the tube plate. Plain tubes can be placed in position and expanded at the rate of about six per hour.

The recent disasters with the tubes of Navy boilers, leading as they did to the appointment of a special committee of inquiry, are a sufficient proof that the mode of securing them is not a perfect one. Undoubtedly the severe conditions of Navy trials search out any defects which may exist, and when leakage has once commenced it seems impossible to stop it again until the tubes are re-expanded. By some it is affirmed that the trouble is caused by overheating the tube plate (see p. 33). Others believe that the chilling effect of inrushes of cold air, or that structural peculiarities, are to blame. Thus each tube has a slight twist, due to the expander having been worked in one direction only. On heating the tubes, they may untwist slightly. This motion will be particularly injurious if the tube plate holes are not perfectly circular. The taper shape of the