cussed, but, as there is no more difficult piece to deal with than a large furnace front, the subject has been introduced here.

The heating furnace is of the ordinary reverberatory type (fig. 272). It is usual to employ coal for fuel, but gas is also used, and it has been affirmed that it causes plates to grow brittle. The flame travels right across the furnace, and gets drawn down at the front. If the furnace is very large, the firing is done at one side, and the downcast is placed at the other. Doors should be fitted so that the plate can be watched,

and care should be taken to keep the temperature comparatively low, partly in order not to burn the iron or steel, but particularly not to heat one part of the plate before another; otherwise serious distortions will occur. Large plates are taken out, turned end for end, and replaced before finally allowing them to cool. J. H. Brinell, 'I. and S. I.,' 1886, vol. i. p. 365, has shown how best to anneal. The plates should be heated quickly to a good heat, cooled rapidly at first and then slowly. Long heating at a low temperature is bad.

Deformations Produced by Annealing. Even with the most uniform heating it will be found that the strains which have been produced while flanging will make themselves felt in the most annoying manner. As already mentioned, the tilting up of the outside flange shortens it, so that it is in compression, and the adjoining part of the plate will be in tension. In a furnace front plate this is counteracted by the stretching of the furnace-hole flanges, but only partially; for, on heating such a plate, it draws as indicated in fig. 273. The metal

at a, b, and c elongates, and the outer flange comes in, sometimes as much as

in.

The metal shortens at d and e, so that in one direction the diameters increase and in the other they decrease, their difference occasionally amounting to as much as 1 in. Partial relief is given by annealing the furnace front plate after the circumference has been flanged, and once more when the holes are finished.

To flange the furnace holes oval, which is sometimes done, does not give good results, partly because the changes of form cannot be previously estimated, but more particularly because the outside flange gets drawn in during annealing, and then, although the furnace holes may be circular, the outside flange is locally flattened.

FIG. 274

All these distortions depend not only on the shapes and sizes of the flanges, but also on the nature and number of the heat at which the various parts were bent. By making the outer flange and that of the furnace hole perfectly circular, then reheating both locally, and forcing them out at this point by means of a liner shown in black in fig. 274, and then annealing the plate, the flanges return from the position by the black line to their original shape, as shown by a dotted line. If they are still oval these parts are set by hand during the process

of annealing, the plate being drawn out of the furnace for this purpose and partly flattened, and then replaced and reheated, and if necessary reset. Where great accuracy

FIG. 275

FIG. 276

is aimed at, solid segments of circles (fig. 275) are wedged into the furnace holes before annealing; they have to be made rather strong, but for all that waste away if used too often.

A careful examination of machine-flanged holes, after they have been annealed, will show that they have closed in a little at their

edge (fig. 276). This, like the other deformations, is due to the stresses set up by the flanging operations. It is needless to say that all other flanged plates are distorted more or less by annealing. End plates cannot, therefore, be fitted together before this is done.

The plates are flattened when they leave the annealing furnace; this is very necessary, especially along the edges, where the riveted seams come. The operation is not difficult, but care must be taken

FIG. 277

that the necessary hammering of the flanges does not spoil them. is necessary to test the circumference by templet while the plates are still hot, and to gauge the furnace holes. Occasionally the circumferential flange will be found set up as in fig. 277. This happens particularly near the furnace holes. In such cases heavy doublehanded hammers have to be used to knock it down again. Portable steam hammers are also used for this and other purposes.

Plates in which the flanges of the edges and of furnace holes are turned towards opposite sides are troublesome objects. Not only is it difficult to get them level, but at the points where the two flanges are nearest each other a disagreeable tendency exists for them to tilt over (fig. 278), and also to bend down bodily, as shown in fig. 279. Considerable experience is necessary to get these parts into shape.

Plates should remain in an annealing furnace for at least two hours, but it takes a few more hours before they have cooled after being laid on the shop floor. A lesson in annealing may be learnt from the plan adopted for optical glass, which leads to perfect results. It is heated to the required temperature, and the furnace is then cooled quickly, but only through a small range of temperature, and it is only when the whole of the glass has adapted itself to the reduced heat that it is lowered once more. This is repeated till ordinary temperatures are reached. If the temperature is reduced steadily the surfaces would necessarily be slightly colder than the inside, and straining could not be prevented.

Final Flangings. Having been thoroughly annealed the various parts of the end plates are now clamped together, care being taken that the circumference is a true circle, and of the right diameter, so

as to fit the shell. A few holes are drilled through the various crossseams, and these are then screwed together by well-fitting bolts. Girders, or, for small-sized plates, angle irons, are also bolted to the plates, to keep them flat while suspended, and then one corner joint after another is heated and brought to its correct shape by hand hammering. This is rather a troublesome operation, and generally requires two heats, but if the corners are carefully prepared much time is saved. When, finally, the two plates are in close contact at the corners templets are applied, to test whether the curvature is correct. These parts should project slightly, as they get drawn in on cooling.

The shaping of these corners will be best understood by referring to fig. 280, which shows the irregular shape of the corner plates as they

have left the press. The corner is now heated, and the outer plate struck, as shown in fig. 281, starting at the bottom, so that the humps (which are shown in fig. 280) of both the outside and inside plate are driven in. Gradually the upper parts of the flanges are reached by the hammer. In the meantime, especially for light plates, heavy hammers or weights have been held inside the flange, in order to make the blows more effective and to bring the plates well together. At a later stage the inside is struck to drive that flange out and to stretch the outer one a little. All this hammering is performed with the end plates, suspended by a chain, so that the heated corner is quite accessible. The final operation is performed on a slab of iron (fig. 282). If there are two horizontal seams in the back end plates, it is usual to rivet up one of them, either before or immediately after the above operation. In shops where these corners are welded (fig. 283) none of the seams may be riveted up first; the shaping of the corners is done during

FIG. 282

and immediately after welding. After this operation it is well, and even necessary, to reheat the neighbourhood of these parts.

This has to do service instead of annealing. These welded corners contract about in., and it is necessary to make the proper allowances. It would be well to leave these corners full and trim them cold. The

FIG. 283

stay holes, unless they have been punched before annealing, or if it is intended to bore them in place, are now drilled, and the plates can now be fitted into the shells, drilled, and riveted up, as already explained. In some works the straight edges of all flat plates are bent so as to make bevelled joints; but this is unnecessary, costly, and dangerous, for heaters have often to be used, and the plates may thus be made brittle.

Riveting End Plate Seams. The cross seams are often riveted by machinery before fitting into the boiler, and as, particularly in the front plate, some rivet holes are countersunk, the plates do not expand equally during the riveting process, and all the trouble taken to ensure perfect fits at the flanges is in vain, in fact the difference of expansion is sometimes so great as to crack the welded corners. It is therefore advisable to bolt these seams well together and to start riveting from both ends (see p. 287). Even now, however, the corners are not likely to fit close to the shell, so that in some works after the greater part of the flange to shell riveting has been done, these corners are heated to redness by small blast furnaces, then hammered to close them up to the shell and riveted. It is not an ideal proceeding, but being sanctioned by Lloyd's Register it is not likely to be given up. The subsequent cracking of one of these heated corners indicates that if other means for making these corners tight can be shown to be satisfactory they should be insisted on.

Flanging Furnace Saddles.-Undoubtedly, one of the most difficult flanging operations is the shaping of furnace saddles. Recent attempts to do the work by Tweddell's flanging press are said to be highly satisfactory.

Two objects have to be kept in view, viz. not to burn or otherwise injure the material, and not to reduce its thickness. Suppose that a

FIG. 284

FIG. 285

FIG. 286

FIG. 287

cylindrical shell (fig. 284) is marked with a number of equidistant longitudinal lines, and that the end of this cylinder is flanged like a furnace saddle, as shown in dotted lines, the straight ones can be