« AnteriorContinuar »
is reduced to a very few per cent., possibly to nothing. When this point is reached, the test piece gives way completely, without any additional weights being added. Now this gradual increasing of stresses is reproduced while flanging a plate. The shape does not matter much, except, perhaps, as regards the intensity of the stresses which are set up. In forming a tube plate, first one side is flanged, except at the corners, then the next one in the same way, and then the third. When this is finished the corners are once more heated and flanged. Now it is clear that, if the operation commences at A (fig. 252), the heat will have little effect at first; but even while B is being flanged and A is cooling, strong tension stresses would be produced; these are very much increased by the time that C is finished. Each one of these parts has probably had a chance of cooling slowly from about 1,000° F., and if the plate were now put aside, and
measurements could be taken, it would be found that these lengths of the plate had not shortened as much as would be due to their change of temperature. If a steel bar, 1 in. in area, is rigidly secured at its ends, and cooled from 212° F. to 32° F., it would then exert a pull of about 13 tons, so that it is not unreasonable to imagine that the cooling of the flange from a red heat would produce a stress of at least 20 tons. But a tension stress at one edge of a plate must produce a compression stress at the centre and a tension at the circumference. In other words, the plate yields a little, and possibly, instead of finding a stress of 20 tons at only one edge, we find a fairly uniform stress of 10 tons all round the plate, and a compression stress in the centre. While the second side is being flanged new stresses are set up all round ; then, when the third side goes through this manipulation, the stresses will be once more increased. To give a correct idea of the distribution of the stresses would be impossible, but for the purpose
of illustration it may be assumed that the flange ABC is now subjected to 30 tons tension, the flange D to 20 tons, and E to 10 tons. The unflanged parts, F, would be subjected to 30 tons, and the centre of the plate is in compression. The next operation consists in flanging the corners, and there is no doubt that heating them relieves the other stresses at the circumference a little, but on cooling they will be more intense than before. Even now, if the cooling were carried out quickly, there might be no danger, because one or the other part would gently elongate ; but, as is usually the case where a failure has subsequently occurred, the plate has been put aside for the night, and next morning the flat part had cracked (fig. 253). Slowly the breaking stress was reached, and then, as in the case of test pieces, the material gave way completely. In a case mentioned by the late Dr. Kirk (fig. 253), ‘N. A.,' 1882, vol. xxiii. p. 131, the crack extended about 18 ins. into the plate, and measured } in. open at the edge. It is not at all certain that, by cracking, the plate was relieved of all its strains ; but even assuming this, and assuming also that the stresses were uniformly distributed over the total circumference of about 14 ft., the opening of the crack to } in. would show that in this particular case the circumferential stress amounted to 13,000
or nearly 40 tons per 168 square in., showing that the above estimate of 30 tons was not too high. Other cases are illustrated in figs. 250, 251, 253, 254, 255, 256. It has not yet been possible to obtain tensile tests of plates which have not led, but which might be expected to do so. They would certainly throw a strong light on the subject. Test pieces should be
Fig. 253 sawn (not sheared) out of the plate all along its circumference. An accurate measurement of the limit of elasticity would give conclusive information as to the intensity of the stress which that particular part had been resisting; but the greatest care would have to be taken not to bend the sample (p. 159).
Before leaving this subject it may be as well to explain why cracks of this sort extend so far beyond the overstrained part, The contraction of area at the edges of the crack at its inner end, and even beyond this point, is a sure sign of the ductility of the metal, at least in the centre of the plate. A careful study of the subject will show that at the instant when a fracture takes place the two separating surfaces are travelling away from each other at a velocity equal to that of sound. For iron and steel it is 17,500 feet per second. The rate at which the
crack mentioned above was extending towards the centre of the plate must have been = 1,260,000 feet, or 550 miles, per second, and it is
therefore not surprising that it had overshot its mark. Instances could be mentioned where the crack had extended right across the
plate, and in one case a plate actually broke in two, one piece knocking down a man. If the above estimate of 40 tons stress per square
inch is a correct one, then the amount of energy which was relieved by the plate cracking was equal to lifting it bodily 16 feet into the air.
The loudness of the report when these cracks occur, and the destructive energy which some steel armour plates have displayed when cracking spontaneously, are proof that these estimates are not excessive. The existence of such mischievous powers in the interior of boiler plates should not be tolerated, and there ought really to be no objection to the annealing of plates which have been flanged. When this cannot be done at once, the centre of the plate should be heated to redness immediately after flanging, and before the edges have lost their heat. The compression stresses in the centre of the plate are thereby partly removed. If the plates have been very much buckled during fanging, they should not be flattened except during the annealing process, for it is chiefly the very flat plates which crack.
Working Plates when Partly Cooled. The other danger to which allusion has been made is the working of steel and iron at a blue heat. Experiments on this subject will be found in the chapter on 'Strength of Materials.' Here it is only necessary to draw attention to workshop practices. The ease with which such failures can be attributed either
to redshortness, coldshortness, or generally unsuitable material, and the absence of any chemical or mechanical test, make it difficult to be sure, in any particular case, what has been the cause of the breakage; but about the following one there can be little doubt.
The flange E (fig. 257) of part of a furnace front plate was so much out of shape that the furnace, f, could not be drawn up sufficiently to make a good job, and it had to be knocked in a little ; but being rather thick, a substantial heater, H, was first applied, as shown. A few weeks later, while the boiler was being tested by hydraulic pressure, a loud report was heard, and though no leakage took place, an internal crack, c, was ultimately discovered, extending round one-quarter of the flange, yet penetrating only to within one-sixteenth of the outside surface. Too little is as yet known about this subject, though every boiler-maker should be made aware of it, which can easily be done by letting him make the following bends ::
Strips of mild steel 6 ins. long, 1} in, wide, and about 3 in. thick, should be treated as follows:
No. 1 should be placed under a steam hammer (fig. 258), allowing
about 3 ins. to project as far as a. This end should then be bent down to the angle b by striking it with a sledge hammer. It should then be reversed to the position c, and again bent down, and the operation continued till breakage takes place. If good, the material will stand 20 half-bends, either with sheared edges, annealed, or hardened.
No. 2 should be placed between two heaters, H, H (fig. 259), and kept there till its edge turns straw colour to violet. It should then be
taken to the steam hammer and
bent as before. Two instead of H
20 bends will now suffice to H
No. 3 should be treated like Fig. 259
No. 2, but the bending should
only be carried on till there is the first indication of a crack. The sample should then be put aside for a day to cool slowly, when it can readily be broken with a hand hammer or by throwing it on an anvil.
No. 4 should be heated like No. 2, and then drawn out under the steam hammer till its thickness is reduced by about it in. After waiting a day it will have become as brittle as glass.
Cases in which the influence of working flanges at a blue heat may at least be suspected are shown in figs. 250, 251.
Local Heating: -It may be possible that plates can be injured by the influence of a blue heat even while they are partly red-hot, and that is another reason why they should always be annealed after working them.
Fig. 260 shows a plate partly flanged, and also locally heated near the centre of one edge. Evidently this red-hot part is surrounded by
a zone of which one part is blue hot. There is also a strong probability that one part of this zone will retain its temperature for a long while, for although the whole plate is radiating heat, the colder parts do not lose it as quickly as the bot ones, and are also being warmed by them, so that at some point the gain and the loss must be equal during a considerable period. If the plate is being hammered while this point of permanent temperature is just blue hot, very serious injury may be done to the material. It is therefore not as improbable as it might otherwise seem to find that such a plate contains as many brittle zones as there have been heats applied to it. The danger of these local