density must have exceeded 51 ounces per gallon. This boiler was not worked hard.

The density as detailed in the above table is not a perfectly correct guide, for boiler water also contains magnesium salts, which add to the density, but which will not crystallise out at almost any temperature; therefore the percentages given in the table, which refer only to pure salt water, should be multiplied by 44:3: 36·01·23 (see table, p. 30), so that in modern high-pressure boilers the water must, at least locally, contain about 70 ounces of various salts before deposits occur. These remarks are not intended to encourage the use of highdensity water, but to enable engineers to fully understand such cases Collapses have certainly occurred before the density reached even, but this may possibly have been the result of purely local deposits of salt, due to hard firing and bad circulation. On standing idle after the accident the salt would dissolve away and leave no evidence except the harm it has done.

Floury Deposit consists chiefly of magnesia, and is formed in boilers if their waters are treated with caustic soda, generally for the purpose of preventing corrosion. This chemical certainly deposits the chloride of magnesia, and unless it is carefully introduced, and only in small quantities, the danger of a collapse is great. If there is grease in the boiler, the magnesia and the grease and water combine to form a substance of the same density as water, which can and does adhere to any part of the boiler and may cause overheating. See p. 128.

Grease in Boilers.-After the surface condenser came into use, engineers were troubled with furnace collapses, which are now explained as being due to deposits of grease, especially as it is known that grease is a ten times worse conductor of heat than scale; but even now the subject is a mysterious one, for whenever a furnace does collapse, due to grease in the boiler, the collapsed part never shows any grease at all, and even the other parts of the boiler are generally only covered with sufficient grease to dirty one's clothes, its thickness being probably less than 1 inch. Seeing that this film disappears from the collapsed parts, the only explanation which at present suggests itself is that at the temperature which exists in a boiler this grease changes its nature and becomes capable of forming tough bubbles wherever the heat is greatest, while either hydro-carbon vapour or super-heated steam is formed between the boiler plate and the skin of the grease bubbles, thus preventing the water from coming in contact with the already red-hot plate and cooling it. One objection to this view is that grease acts in a very positive manner, it being found that if one surface in a boiler has been overheated due to grease, so have the others. In one case nine furnaces in three boilers came down together; it also seems that the growth of collapses is a slow one, extending sometimes over days and even months. When furnaces collapse due to scale they do this very suddenly, and generally, too, it is only one furnace in a boiler which comes down at a time.

Grease and Scale.-It has been noticed that grease alone is far more dangerous than grease and scale together; probably the grease adheres so firmly to the scale that it cannot form bubbles. This may

account for the experience that the furnaces of new boilers have now and then come down, sometimes even on trial trips. In all such cases the cause was grease without scale. On the other hand steamers are running without mishap whose boilers are coated with scale, and in addition with a filthy covering of grease.

Collapsed Furnaces.-One of the greatest mishaps to a boiler which can occur at sea is undoubtedly the collapsing of its furnaces. Generally, but not always, this is due to ignorance or carelessness of the engineer in charge. For instance, it may happen that things have been left in the boiler which, by means of the water circulation, are landed on some plate exposed to the direct action of the flame, and may cause a local bulge, a ludicrous instance being where the head of a dead horse had been put into a Cornish boiler to prevent corrosion, and had settled on the furnace crown. With steamers it has happened that while ashore, the supplementary feed which had to be put on carried with it mud, sand, or infusorial earth, and this settled on the furnace saddles, or near the upper part of the flanges of Adamson's rings. Under specially favourable conditions the bulges which have then been produced were shaped similarly to the section shown in fig. 48.

FIG. 48

A very serious case of collapsed furnaces occurred in a ship with four elliptical boilers. One of them had given trouble, had just been repaired at sea, and when put into use again, all the furnaces in the three other boilers came down together. No doubt the indicated pressure of the repaired boiler was fictitious (see p. 2), and due to air in the steam space, although the water was still fairly cold. On connecting the boilers it disappeared, and was perhaps reduced to a partial vacuum, into which steam suddenly rushed from the other boilers. The consequent excessive ebullition, aggravated as it was in these boilers by very restricted water spaces, and very little clearance below the tubes, must have raised the water away from the furnace crowns, and these collapsed.

Very mysterious collapses are sometimes caused by emptying and refilling a boiler quickly in port, without removing the manhole doors or cleaning its inside. It would appear as if, under these conditions, the scale on the tubes falls off, lodges on the furnace crowns, and, as it is not removed, causes overheating. It also happens that on blowing out the boiler, the scum and oily matter which floated at the water level adhere to the heating surfaces, even when the new water is admitted.

By far the greatest number of collapses are due to scale, salt, or greasy matter; but then, instead of the furnace crowns coming down, it is the sides which come in; for, although the deposit may be uniformly distributed, the heat of the fire is greatest on either side just

over the burning fuel. Shortness of water and local deposits bring down the crowns, not the sides.

The collapse of furnaces may on occasion be due to very heavy firing combined with one or the other above-mentioned causes. Mr. Yarrow's experiments on tube plates (p. 126) show that, on account of the fire side of a plate being hotter than the water side, it curls towards the water. With the same fierce heat which he was using on flat plates, a furnace top of 20 ins. radius would tend to acquire a curvature of 203 ins. radius, and this would tend to increase the horizontal diameter by 1 ins. and reduce the vertical diameter by the same amount, making the furnace 3 ins. oval. The same influence is at work longitudinally. If free to move, the apex of the furnace crown, heated as above, would acquire a longitudinal curvature, of which the depression would be about 1 in. in a 6 ft. length. Of course, the end plates and the stiffness of the cool furnace bottom are opposed to these movements, but with the least sediment, grease or dense water, the metal is weakened and this support is much reduced and the furnace collapses. As these curvatures and depressions are independent of the thickness, it is well to make unstrengthened furnaces very thick if exposed to fierce heat, so as to give them extra strength.

In Mr. L. E. Fletcher's experiments on red-hot furnace crowns, there is a case of a furnace rising in. while raising steam, although the fires were very light.

Bulging of Flat Plates. Another part of the boiler where scale deposits produce visible deformations is the flat plates of the combustion chambers. With narrow water spaces the presence of much steam assists in causing the plates to get hot and. bulge, by keeping the scale partially dry. If this treatment did not tend to make the plates brittle, and cause them to crack between the stays, little harm would be done, because in their bulged shape plates are stronger than when flat. Possibly, too, the stretching of the plates enlarges the stay holes, and causes these to leak. When the bulging is serious the favourable conditions for further deformation are increased, because it is now more difficult than before to remove the scale.

In boilers with two or four furnaces, leading into one large combustion chamber, serious bulging sometimes takes place at the top of the saddle between the two central furnaces, but only if they are exposed to the heat of the fire, and if several angle irons have been fitted underneath, as is customary for strengthening these parts; the enclosed spaces form steam pockets, and the saddle plate above and between them gets overheated.

Tube-Plate Troubles in modern warships have recently attracted much attention, but it does not appear that any satisfactory explanations have been put forward. Doubtless here, too, it is a case of overheating caused by a high temperature in the combustion chamber, and an excess of steam bubbles near the plate on the water side. The phenomenon (see p. 35) that iron contracts permanently when heated and cooled again might account for the tubes shrinking and the tube plate drawing away from them, but it is difficult to suggest a remedy which has not been tried and failed.

Experiments on the subject have been made by A. F. Yarrow

D

(N. A.,' 1881, vol. xxxii. p. 106), who showed that tube plates tend to bulge towards the fire side, and thus to draw away from the tubes; but the strong curvatures which he was able to produce are not met with in practice. An idea of the temperature of a tube plate will be obtained by examining the following case. (Compare p. 127.) Let the thickness of a tube plate (fig. 49) be

1 in., if perfectly clean its temperature would be 40° F. higher on the fire side than near the water, while transmitting sufficient heat to evaporate 20 lbs. of water per square foot per hour. The tube, being only in. thick, will be 4° F. hotter on the one side than on the other, and therefore its average will be only 18° F. less than that of the tube plate, which corresponds to a relative shrinkage of about 3000 in. This can be neglected in comparison with the compression existing in the tube, due to the expanding. If covered with scale, as shown in fig. 49, or if in contact with a film of steam, so that the water can only reach the plate with difficulty, then the conditions are changed.

FIG. 49

Suppose that the scale is in. thick, and that only half of the above-mentioned quantity of heat is being transmitted, then the metal of the tube and tube plate would be about 1,000° F. hotter than the boiler water. But it is difficult to believe that the heat transmitted through the tubes is equal to that passing through the plate, and certainly, within a very few inches from the end, the difference will exceed 50 %, so that it is not unreasonable to assume that the tubes are only 500° F. hotter than the water. With 3-in. diameters this corresponds to a relative difference of 16 in. between the tube and hole. The

FIG. 50

1000

heightened temperature will reduce the elastic limit to, say, 13 tons, so that the pressure, due to the expanding, cannot dilate them more than 3 in., thus leaving an opening of half the difference, viz. in., or about in. all round the tube. This is sufficient to cause serious leakage. Salt from the boiler water is supposed to choke this opening as long as the heat is fierce, but dissolves out and renews the leakage when cold. (A. J. Durston, N. A.,' 1893, vol. xxxiv.)

In most boilers where this has occurred another force was at work intensifying the leakage.

It is well known that in order to reduce the crushing stress on tube plates of double-ended boilers, not only the sides but also the tops and bottoms of the combustion chambers are stayed to the shell (fig. 50).

When pressure is raised in the boiler its diameter increases, and tension stresses are produced in the tube plates, which they, with their small effective sections (only about 25% of the solid plate), are unable to resist, except by relieving the tubes of a little of the pressure which holds them in place. At the same time the tubes contract their diameters a little, due to the external steam pressure.

The chief difficulty in accepting these views is that the leakages are not diminished when the fires are drawn, and the phenomenon noticed by Wehrenfennig (see p. 24), that iron and steel contract permanently when heated, even if only up to boiling point, may help to explain the matter. The author's experiments on iron and steel bars which had been drawn out cold under a hammer, on cold rolled bars, on wire (drawn), and on stretched test pieces, show that they all contract permanently on being heated. The distortion which takes place during the annealing of flanged furnace front plates points to the same conclusion, viz. that leaky tubes are caused by a partial annealing of the expanded tube metal, due to excessive heat. The remedy which at once suggested itself is to anneal and re-expand the tube ends in place. According to this view, tubes which have leaked once (and have therefore been annealed) ought not to leak again after reexpanding, but although they do, the trouble is lessened. It is curious. that tubes which have leaked on account of excessive scale grow tight when it has been removed, even if the ends are not re-expanded.

The only effective remedy for these troubles, if they were not caused by scale, appears to be the use of copper tube plates. Possibly a wider pitching of the tubes, or reducing their diameters at the back end, may do good. Or the tube ends might be subjected to a preliminary compression. This, however, would require an amount of care in boring the tube plate holes which is not often bestowed on them. Differences of in. in the diameters of various holes in one plate are not uncommon, and most of them are distinctly oval, sometimes as much as in. Taking more care to make the holes circular and of equal size may reduce the trouble.

In a recent experiment (discussion on A. J. Durston's paper, 'N.A.,' 1893, vol. xxxiv. p. 150) a compressed tube was compared with another which had been expanded in the usual way. Both were fixed in a tube plate which was placed over a smith's fire and heated to a temperature at which about 100 lbs. of water were evaporated per square foot per hour, but the plate was only occasionally covered with water, which then remained in a spheroidal condition. After a few hours' exposure the plate was allowed to grow dull red hot, when both tubes grew slack. Careful measurements showed that the expanded tube had contracted its diameter in., while the compressed one had not altered, but its hole in the tube plate had changed its taper from 1: 12 to 1: 10. This compressed tube could not be withdrawn out of the hole, for its projecting end (4 in.) had expanded during annealing. This goes far to prove that leaky tube ends are caused by raising their temperature to redness.

Attempts are being made to electrically weld tubes into the plates, but the risk of burning an internal part of a boiler which is nearly finished must deter all except the most venturesome from trying this plan.