by J E Gordon
In this way a tiny unseen crack may start from any hole or notch or irregularity in a stressed metal and may spread across the material, which is not, as a whole, changed in any obvious way. Sooner or later, such a ‘fatigue crack’ will reach the critical length for an ordinary common or garden crack. When this happens, the crack will immediately speed up and run right across the material, often with very serious consequences. It is usually quite easy to diagnose a fatigue crack after failure because of its characteristic striped or banded appearance. Before rupture, however, an incipient fatigue failure may be practically impossible to spot.
Naturally metallurgists and others do a great deal of experimental fatigue testing on their materials, and a great many different types of testing machine are now available for the purpose. It is common to consider the fatigue properties of a metal in terms of a reversed stress (±s) – that is to say, the sort of stress which would occur in a rotating cantilever, such as the axle of a vehicle. (There are ways of converting these results to other conditions of fluctuating stress.) This reversed stress (±s) is usually plotted on a graph against the logarithm of the number (n) of times the stress has to be applied to a specimen to cause failure. This is sometimes called an ‘s-n diagram’.
The s-n diagram for a typical steel would look like Figure 3. It will be seen that the ‘curve’ is a dog-legged affair which flattens off after about a million reversals – which might be equivalent to about 3,000 miles of service for the axle of a car or train, or about ten hours of running for an ordinary car engine, which, of course, goes round much faster than the wheels. The existence of a definite ‘fatigue limit’ of this nature for materials like iron and steel constitutes a great comfort to the engineer. If his engine or his vehicle will run for 106 or 107 revolutions – which may only take a few hours – then there is some hope of its being safe indefinitely. But fatigue is a danger which always needs to be considered.
Figure 3. Typical fatigue curve for iron or steel.
Aluminium alloys do not have a definite fatigue limit but tend to tail off, something after the fashion of Figure 4. This makes them more dangerous to use and accounts for some apparently old-fashioned prejudices in favour of steel for use in machinery and other structures.
The Comet accidents, which occurred in 1953 and 1954, naturally caused consternation and well-justified alarm. The investigation of these accidents by Sir Arnold Hall and a large team of experts was a classical feat, not only of engineering detection, but also of deep-sea salvage. The broken parts of one of the aircraft, which had fallen into the Mediterranean, had to be dredged up from a depth of over 300 feet or 50 fathoms. The salvage people managed to recover practically the whole of the aeroplane and the innumerable fragments covered the floor of a large hangar at Farnborough. As far as I remember, no piece was more than two or three feet across.
Figure 4. Non-ferrous alloys such as brass and aluminium frequently do not show any definite fatigue limit.
The Comet was one of the earliest airliners to have a pressurized fuselage. The main purpose of this was, of course, to spare the passengers from the discomfort and danger of the atmospheric pressure changes associated with change of altitude. In the old days, when flying over the Rocky Mountains, one used to have to eat one’s lunch while wearing an oxygen mask: this now ranks as one of those lost skills. In a pressurized aircraft the fuselage becomes, in effect, a cylindrical pressure vessel, not unlike a very thin-walled boiler, which is pressurized and relaxed every time the aircraft climbs and descends.
The lethal mistake in the design of the Comet lay in not realizing sufficiently the danger of ‘fatigue’ occurring at stress concentrations in the metal of the fuselage under these circumstances. The Comet was built from aluminium alloys, and most of de Havilland’s previous experience had been gained with wooden aeroplanes, such as the triumphantly successful Mosquito. I am not suggesting for a moment that de Havilland’s very able design staff did not know a lot about fatigue; but it is possible that the danger of fatigue in aluminium alloys may not have burnt itself sufficiently deeply into their collective consciousness. Wood is much less susceptible to this danger than metals – which is one of its great advantages.
In each of these accidents cracks,seem to have started from the same small hole in the fuselage and spread, slowly and undetected, until they reached the ‘critical Griffith length’. Whereupon the skin tore catastrophically and the fuselage exploded like a blown-up balloon. By repeatedly pressurizing a Comet fuselage in a large tank of water at Farnborough, Sir Arnold Hall was able to reproduce the effect so that it could be observed, as it were, in slow motion.
Part of the trouble about the Comet accidents was that the fatigue cracks which must have existed were never spotted by an inspector, perhaps because he was not expecting to find them, but more probably because they were too short to be seen easily. Nowadays aircraft fuselages are designed to contain with safety cracks up to about two feet long, and one would think that so long a crack could hardly fail to be seen in good time. There is, however, the story about the two cleaners at London Airport. These ladies finished sweeping out the cabins of an empty airliner late one night. They shut the door and went down the steps on to the tarmac.
‘You’ve forgotten to switch off the light in the toilet, Mary.’
‘ How do you know?’
‘Can’t you see it shining through the crack in the fuselage?’
Accidents to wooden ships
Before the days of railways nearly all the heavy traffic went by water. Besides the deep sea trade and the Continental trade and the inland trade by river and canal, there was an even larger coastal trade. Many thousands of little wooden brigs and schooners, of the kind caricatured by W. W. Jacobs, transported anything and everything, not only into the coastwise creeks and harbours, but to almost every possible or impossible beach. A ship would be grounded on the beach at high water and, when the tide fell, would unload her coal or bricks or lime or household furniture into carts which were driven alongside. When the tide rose again, she would slip away to sea and then go and do it all over again somewhere else.
Naturally this was rather a risky business, but during the eighteenth century most of the smaller vessels could afford to lay up and refit during the worst of the winter – when the crews could see something of their families and of the local pubs. This slightly idyllic and not exceptionally dangerous state of affairs was upset by the more competitive conditions of the nineteenth century. Under commercial pressures vessels had to trade throughout the winter and could not afford, as a rule, to wait for weather. Indeed the regularity of some of these little sailing ships would put a lot of modern goods trains to shame.
But, of course, a price had to be paid. During the middle 1830s an average of 567 shipwrecks occurred round the coasts of this country each year; as a result, a yearly average of 894 lives were lost. Whether these figures are better or worse, per ton-mile of goods delivered, than modern lorries I do not know. At any rate the public conscience was disturbed at the time and Parliament appointed a Select Committee to investigate the ‘Causes of Shipwrecks’. After hearing a great deal of evidence, the committee reported that, apart from minor causes, shipwrecks in this country could principally be attributed to the following conditions in ships:
Defective construction.
Inadequacy of equipment.
Imperfect state of repair.
They pronounced’ That the defective construction of ships appears to have been greatly encouraged by the system of classification [i.e. the rules governing construction and repair of insured ships] which, from the year 1798 up to 1834, was followed by Lloyds.’
The committee went on to add that the system by which the government measured ships for tonnage dues encouraged thoroughly unseaworthy shapes of hull. The bureaucratic mind does not seem to change very greatly through the centuries.
To be fair, the problem of framing regulations for the strength and safety of ships, or any other kind of structu
re, is an extraordinarily difficult one. No doubt a certain amount of progress has been made in the matter since the 1830s. At the same time, and in a different sense, a great deal of technical progress has been prevented – especially by the various building regulations. As Pugsley points out in The Safety of Structures, it is inherently impossible to make regulations about the strength of structures which are proof against both fools and knaves without preventing, or at best handicapping, development and innovation. Regulations for structural safety are presumably necessary, but some of them are not only stultifying; they can be the actual cause of accidents.
To return to wooden ships: not only the clippers but the little brigs and brigantines and topsail schooners and barges – which were so beautiful and so satisfying – have all gone, and the yards that used to build them are now turning out yachts. The structural problem of a wooden yacht is both more and also less severe than that of larger vessels. Yachts’ hulls are not bumped on shingle beaches while carrying cargoes of stone or coal, but they have a more difficult problem with regard to local impacts which their thin skins are not well fitted to resist.
Now that long voyages in small yachts have become so fashionable this question of the impact strength of the hull has become important. Yachts voyaging in deep waters have repeatedly been attacked and sunk by killer whales. These animals weigh about six tons and swim at around thirty knots. They seem to have a special hatred for small yachts, which they ram and hole below the waterline. This has now happened so often that the possibility cannot any longer be classed as an ‘act of God’ (Poseidon presumably) but is a serious hazard which must be guarded against.
It is probably impracticable to make the sides of a small yacht thick enough and strong enough to resist such an attack. The best thing to do would seem to be to provide some sort of inflatable floatation gear to keep the yacht afloat – and preferably sailable – after she has been holed. So far, those who have survived these attacks have done so by taking to the dinghy, which, naturally, gave most of them a very unpleasant time before they were picked up by a steamer after many days or weeks.
More about boilers and pressure vessels – and something with boiling oil in it
For a considerable number of years before the railway system was completed much of the passenger and express freight traffic was carried by steamship. During the first half of the nineteenth century, not only were there far more steamers running to more Continental ports than is the case today, but there were also very numerous services between towns in Great Britain. Considerably the cheapest – and often the quickest and most comfortable -route from London to such places as Newcastle, Edinburgh or Aberdeen was by steamboat.
Accidents were fewer in steamships than in sailing vessels only because there were many fewer steamships. Nevertheless, between 1817 and 1839, there were ninety-two major accidents to steamships in British waters. Of these, twenty-three were due to boiler explosions. This is nothing like as bad a record as that of the American river steamers a few years later; but it is quite bad enough.
Some of the early boilers were made from unsuitable materials, such as cast iron. At least one cast-iron boiler, that of the S.S. Norwich, duly burst and killed several people. Even when boilers were more or less properly constructed from wrought iron, they were very commonly neglected and allowed to rust through until they burst. This was the cause of the loss of the For farshire on the Fame Islands in 1838. Five people were rescued by Grace Darling’s superb feat of seamanship.*
Again a Parliamentary Committee was appointed, which reported in 1839 and produced an extensive, thorough, factual and almost incredible document. During the boom years of the expansion of the steam engine, sober, let alone competent, responsible or intelligent engine-room staff were almost unobtainable, even at very high wages. These people treated their engines and boilers with a degree of ignorance and carelessness which almost passes belief. For instance:
A steamer, on her passage from Ireland to Scotland, was perceived by her commander during the night, and in a smooth sea, to be going with much greater than ordinary velocity through the water. The engineer was not at his post; the Captain inquired of the fireman how it was that the engines were going so fast: the man said ‘He could not tell, for he had very little steam and had been firing hard nevertheless’. The Captain began to look about him and, approaching the chimney where the exposed safety valves were fixed, he perceived a passenger fast asleep with the greater part of the weight of his body resting on the flat, cheese shaped, weights of the safety valve. This man had contrived, with some luggage, to make his bed there for warmth. On arousing and turning him off, the valve rose, the steam escaped with a roar which denoted its having attained a very elevated pressure.
There was no mercurial gauge to indicate the pressure of the steam to the fireman who was accustomed to keep it as near as he could to the blowing-off point: and not having heard it escape, he ‘fired-up* believing his steam to be low; and he was too ignorant to ascertain the fact, though the increased speed of the engines should have informed him that something unusual had occurred.
It is mentioned by several of our correspondents that engine men, firemen and even masters have frequently been caught sitting, or even standing, on the safety valves, or hanging weights and resting their bodies on the levers in order to raise the pressure of the steam at the moment of starting.
The report goes on to say that it was also the practice to stow surplus bunker coal on top of the safety valve. The steamship Hercules blew up from this cause. Altogether, it is rather remarkable that only seventy-seven lives were lost from boiler explosions in British steamers during the period under review.
The record of the railways was about as bad as that of the steamships and for much the same causes. There was a succession of very serious accidents extending over a period of seventy or eighty years. About the last of these occurred in 1909. A locomotive boiler blew up although the pressure gauge appeared to be showing zero pressure. It turned out that a workman had assembled the safety valve the wrong way round, so that it was incapable of blowing off at all. The gauge appeared to show no pressure for the simple reason that the needle had gone right round its full travel and was pressing against the wrong side of the stop pin. Three people were killed and three more badly injured.
In these latter days the number of boiler explosions has greatly diminished. This is partly because the manufacture and maintenance of steam boilers is now closely controlled by law and by the insurance companies, but perhaps more because the number of steam engines in service is now quite small and those that do exist are nearly all large plants, such as power stations, which are presumably run by competent people.
But – when is a boiler not a boiler? This is quite an interesting legal question. There exist in industry a large number of pressure vessels of one kind or another which are used in various manufacturing processes. Many of these vessels are of more complicated and less conventional design than traditional boilers and they may be less obviously dangerous. In general the control over their manufacture and use is less strict than for ordinary boilers. However, many of these vessels are heated by process steam or by hot oil under pressure, so that the consequences of fracture may be nearly as bad. It is well to bear in mind that the fatigue limit for the weld metal in mild steel structures exposed to wet steam may be as low as ±2,000 p.s.L
In one instance in which I was concerned, two large rotating drums, used for making plastic-coated paper, had been converted from low-pressure oil heating to steam heating – using process steam at a higher pressure. To make certain, the insurance company’s inspector had insisted that the drums be ‘strengthened’ internally by connecting the flat end-plates to the cylindrical surface by means of a number of large triangular gussets, or brackets, cut from mild steel plate and welded in place.
Both drums burst in service after being used for a short time with steam heating. From the drawings I calculated that, in the two drums, there were forty-eig
ht individual places at which failure should have taken place. In fact this was a pessimistic estimate; failure actually occurred at only forty-seven places. By the grace of God, nobody was killed or seriously injured: but it was naughty of the insurance company’s inspector, who, I expect, was a diligent and well-meaning little man.
Another case was more tragic. A firm of chemical engineering contractors had bought in from elsewhere a mixing vessel which they installed as part of a plant they were constructing for a customer. Since this mixing vessel was intended to be heated by oil under pressure, the pressurized heating jacket had been subjected to a ‘proof test’ with cold water. It had withstood a pressure of 65 p.s.i. without obvious damage before it was installed. However, when the plant was commissioned and the jacket was filled with very hot oil at only about 23 p.s.i., the jacket burst after a few hours of service, spraying a man with oil at 280° C, from the effects of which he died a few days afterwards.
According to the report of the official inspector, the accident could only have happened as a consequence of gross mismanagement by my clients, the firm of chemical engineers. As a result these people had become involved in very elaborate and expensive litigation in the High Court.
In fact the official report of the accident was based on faulty observation of the broken remains and was quite misleading. The vessel had burst, not because it was mishandled by my clients, but because it was of incompetent design and manufacture. Although the technical cause of the accident was, in reality, of a slightly subtle nature, both my clients and also the people who actually made the vessel had assumed the design of such a thing to be a trivial problem. In fact the vessel was never really ‘designed’ at all in any sophisticated sense but was simply put together ‘ by eye’ in a back-street welding shop.