12

Naval Architecture

UNTIL RECENTLY British warships were designed within the Admiralty or Ministry of Defence. Ship Department and its predecessors, DNC Department, E-in-C, etc, operated with a small design section, which would draw on the expertise of a considerable number of specialist sections. Initially, these dealt with the traditional naval architecture subjects of Stability, Speed, Seaworthiness and Strength while the E-in-C and DEE had their own specialist groups, such as Gearing, or Batteries. Following the merger of the old RCNC with the RNES, some were merged – Materials (Steels to paints), Stealth, Fire, Battleworthiness and others.

In earlier years these specialists would respond to queries from design sections and reply from their own experience or, where necessary, task the relevant research establishment – AEW, NCRE, AEL, AML etc. This interface was valuable, as it was not uncommon for a busy design section to ask the wrong question, and even more common for them to fail to understand or apply the reply. Increasingly, the role of the specialist sections changed to the setting of standards, first in the ‘General Hull Specification’ and later in ‘Naval Engineering Standards’. Once these standards were soundly based, the specialist sections were given authority to monitor compliance, though there was always a procedure to grant exemptions or make changes where necessary.

These standards were derived from hard and often painful experience, and modified in the light of new technology and changes in the availability of materials. Such standards must change from time to time, and it is as proper for the designer to challenge them as it is foolish for him to ignore the accumulated wisdom contained in them.¹

Curve of righting levers (GZ). (W J Jurens)

Stability

Intact Ship

The principles of stability for an undamaged ship were understood by about 1870, and by the end of the century mechanical aids were available which made the calculations merely tedious.² Stability after damage was much more difficult and, though the principles were well understood by the early twentieth century, only simple comparative calculations were possible before modern computers became available. Only these basic principles will be outlined in this book.

There were two major developments in the post-war years: firstly the criteria of acceptable stability were more clearly defined; secondly, later the introduction of powerful computers enabled much more realistic calculations to be carried out.

Even the word ‘Stability’ is used with several different meanings. The proper meaning to a naval architect is that stability is a measure of the force needed to heel a ship through a small angle – a ship is stable if a fairly large force is needed to heel it. A ship may also be said to be stable if it can be heeled to a large angle without capsizing. This second meaning is an extension of the first meaning but it is not the same. Finally, a ship is sometimes said to be stable if it rolls gently. This meaning is almost the opposite of the first two usages and should be avoided.³

For any floating body the buoyancy associated with the underwater volume will always be equal to the weight – displacement (Archimedes’ Principle). Both weight and buoyancy forces are vertical and when the ship is upright they are in the same line, balancing exactly (upper small sketch in diagram opposite).

As the ship heels the shape of the underwater volume changes and the position of the centre of buoyancy through which the force acts will move. Initially, the buoyancy force will move towards the more deeply immersed side and the pair of forces (Weight and Buoyancy) will act as a couple tending to bring the ship upright. As the deck edge goes under on one side and the bilge comes out on the other, the outward movement of the buoyancy force will slow down, stop and then reverse.

The distance between the forces of weight and buoyancy is known as the ‘righting lever’, represented by ‘GZ’. Values of GZ are plotted against angle of heel In the curves of righting levers, usually abbreviated to ‘GZ curve’. The key parameters are the value of the maximum GZ and the angle at which it occurs. Initially, the range of stability up to the angle at which GZ becomes zero was thought to be important but later it was recognised that this was unrealistic. The angle at which large openings such as boiler room inlets went under, the ‘down flooding angle’, was a more realistic limit.

The GZ curve and associated calculations can only be an approximation to reality because, for one thing, it assumes the water surface is flat and horizontal – no waves. However, over a century’s experience has shown that it is a good approximation.⁴

Stability after Damage

This case is much more complicated than that for an intact ship. For a start, there are numerous types of damage: small holes (splinters, cannon shell) over a considerable length, one big hole (contact torpedo – 30ft × 15ft), or splits from a non-contact explosion. In any of these cases flooding could spread through damaged bulkheads or systems such as ventilation that were not fully tight. Early welded structures with unsuitable materials and poor procedures were more likely to serious failure under explosive loading than riveted structures, but by the end of the Second World War these problems had been overcome and welded structures were seen as far stronger.

The intact portion of a damaged ship would be unsym-metrical and the ship would have large angles of both heel and trim. This lack of symmetry meant that many of the short cuts that made calculations feasible for the intact ship could not be used, and the complexity of direct calculation for a damaged ship made it almost impossible.⁵ It was customary to consider only flooding amidships so that trim could be neglected, and usually only two compartments flooded. This ignored the loss of stability when trim brought the quarterdeck under water, a contributory cause of numerous losses. A few examples of severe flooding were included in the Ship’s Book.⁶

There were few cases of extensive flooding to study. If the ship sank, it would not usually be possible to record the true extent of flooding. Even for ships that survived, repair was the first consideration and detailed records were rare.⁷ However, by the end of the Second World War there were a considerable number of well founded guidelines of ‘Do’s and Don’ts’.

Post-war Criteria

In the early post-war years there were no formal stability criteria, but the subject was taken very seriously, each case being considered on its merits. Senior constructor officers had experience of war, many had served at sea in uniform, and they knew all too well that damage in war is frequent. They also recognised that sinking from loss of stability following damage in ships of destroyer or frigate size was fairly rare, particularly among the more modern ships, showing that pre-war judgement was not far out. Almost alone amongst major navies the RN had lost no undamaged warships from poor stability and bad weather.

Some cases were re-examined. The author re-worked both stability and strength calculations for Kelly after damage. Probability theory was badly taught (and it still is) but began to influence thinking, particularly on the extent of damage and the spacing of bulkheads. The Type 82 design team (led by Eric Tupper) saw the need for more rigorous standards and adopted the US ‘Sarchin and Goldberg’ criteria. The Type 21 frigate was designed by contract and it was essential to have formal standards in the contract.

Stability Standards

It was decided to introduce the USN standards⁸ that derived from the investigations into the loss of three USN destroyers in the typhoon of December 1944.⁹ Technically, this tragedy went far to justifying the conventional GZ curve. The three ships which were lost had the poorest characteristics; the next poorest were in grave danger, whilst those which had good GZ curves were not in serious danger, though even in some of these the motions caused numerous injuries.

The USN inquiry placed the blame primarily on the admiral for requiring a course and speed inappropriate for small ships in such a storm. The direct cause of loss was a combination of wind pressure that forced the destroyers over to a large angle together with rolling in waves through large angles either side of the steady wind heel. This combination brought the boiler room intakes into the water and flooding began. To make matters worse, the main switchboard was below the intakes and shorted, causing loss of steering.

The new rules required a calculation of the wind force on exposed topsides to be equated to the righting force represented by the GZ curve with enough margin to allow for rolling. This wind-loading criterion was most demanding and many older ships in service failed to meet it. There were other rules concerning movement of weights and personnel.

An extract from a Ship’s Book showing the effect of flooding the engine room and boiler room of a Whitby. This limited damage reduces stability by about a third.

(PRO ADM 239/431)

The wartime destroyer Kelly in dock after being torpedoed. Recalculations of her strength and stability after surviving such massive damage were fed into the post-war ‘Tribal’ class design. (D K Brown collection)

It is essential to realise that stability standards are minimum values below which a ship in service must not fall. The standards are not just design aims, and must still be met after many years in service. A growth margin is included in the design for this reason.¹⁰

After the Falklands War Ship Department was asked to devise a simpler presentation of stability data for ships. They had a Stability Statement giving the results of the latest inclining, curves of stability to work out new conditions and a number of worked examples showing the effects of damage. Advice was taken from inside and outside the Ministry but it was concluded that there would only be a minor change – stability, particularly after damage, is not simple. The only change was to add an expiry date to the Stability Statement after which a new inclining was needed (see Appendix 4).

Power and Speed

For many years after the Second World War, the traditional Froude methods of estimating power and speed and selecting a hull form continued to give good service.¹¹ A provisional form would be selected from the Iso-K books, which summarised the results of all model tests ever carried out at AEW since 1872.¹² Almost always some changes were needed to suit the new design, such as a general increase in beam, lengthening the cut-up, or local changes to accommodate the weapon stowages. It is important to realise that there is no one ideal form (see photograph on page 17). The form suits the role. A model would then be made in wax and tested at Haslar, which would then be modified to reduce its resistance. Such modifications depended on the powers of observation, and experience of the experimenter in charge, who would look for waves or turbulence generated by the model.¹³ Typically, one would expect a reduction of about 5 per cent as a result with the saving in fuel, more than paying for the cost of testing. The most useful data for early nuclear submarines came from a Royal Aircraft Establishment report of 1923 on airship form, (see Chapter 9).

The diameter, pitch and rotational speed of the propeller would be selected from the tabulated data of tests carried out in the late 1930s with large, 20in diameter, models. Even when advanced design methods were introduced for quietness, this data proved very useful in providing a starting point. Other model tests measured the interaction between hull and propeller – the flow round the hull upset the inflow to the propeller, whilst the suction of the propeller altered the flow round the aft end of the hull.

Tabulated and plotted data were incorporated in computer databases. In general, only the more recent forms, suitable for modern roles, were fed into the database, but the old data was available if needed and it was easy to create a trend curve appropriate to a particular vessel.

Ship tanks are very versatile and there were many tests and trials of which only a few proved useful. One of the most valuable was that of transom flaps, based on the success in the ‘Gay’ class but neglected for many years, since on resistance alone the slight benefit at top speed was offset by a penalty at lower speeds. Eventually it was shown that there was an improvement in propeller performance at all speeds, leading to an increase in speed of 1.5kts when tried on a Type 21 (Avenger). It was shown by model tests and a full-scale trial that the introduction of long chain molecules (polyethylene oxide) at about 30 parts per million would reduce viscous resistance by about 30 per cent. Air lubrication was tried, but without success. (Propeller development is covered under Stealth in Chapter 13.)

For most of the period AEW had a dedicated trials section who would fit the required instrumentation,¹⁴ carry out the trial and pass the data back for analysis. Model testing is not exact and a correction based on previous ship-model correlation remains essential.

Seakeeping

‘There are three things too wonderful for me … The Way of a Ship in the midst of the Sea’ – Proverbs

By the end of the war it was recognised that the next generation of escorts – 1945 sloop merging into the Whitby – would need be able to maintain speed in rough weather in pursuit of fast submarines. The final configuration of the Whitby class was proposed by Neville G Holt, drawing on his experience as a yachtsman and sea-time as Constructor Captain during the war, and developed by R W L Gawn at AEW.¹⁵ It featured fine waterlines forward, a high freeboard forward (carried aft), and a reduced moment of inertia by concentrating weights (armament, engines) close to amidships. This design soon won a high reputation throughout NATO for seakeeping and even with more recent theoretical understanding of seakeeping has proved hard to beat.¹⁶

Coventry off the Falklands: hit by three bombs flooding five main compartments, she could not survive. The calculation of the effects of massive, asymmetric flooding could not be handled in design without a capable computer. (D K Brown collection)

Modern understanding of the motions of a ship is based on ‘strip theory’ in which the forces acting on transverse slices of the ship in a random sea are integrated along the length.¹⁷ The general conclusions have been verified by trials on ships, notably one in which a Leander (Hermione) and a ‘Tribal’ (Ghurka) were driven at high speed (22kts) through very severe seas.¹⁸ It was found that theory estimated pitch and heave quite accurately; occurrence of green seas over the deck was less well forecast, as the interaction between ship and wave was not properly represented (this has since been improved but is still a problem). It took many more years before rolling behaviour could be computed accurately.¹⁹

The well-established theory has made it possible to show the effect of varying the dimensions on ship motions. Pitch and heave are dominated by length, but longer ships are more highly stressed and there is a practical limit to length. Freeboard is roughly dependent on length – at the end of the war freeboard was taken as 1.1 √L ft. In later ships 1.3√L has been used as a guideline. Slamming is a complex problem but the starting point is adequate draught.

Perhaps the biggest problem in seakeeping is to decide what levels of motion are desirable and how much it is worth paying for improvement: neither is easy. Helicopter operation is of major importance in ASW and for frigates is usually limited by the vertical velocity at the landing spot; but in the smaller Leeds Castle it was found that landing was limited by roll angle. This illustrates a general point, that limits are almost always multiple – eliminate one problem and the next is not far behind. If vertical velocity is eliminated (a rock such as Rockall!), helicopter operation would be limited by wind speed, a point often neglected by over-enthusiasts for SWATH. Helicopter operation from frigates can be improved by making the ships longer or by moving the helicopter deck closer to amidships where motions are less.

Reduced operational capability in bad weather is largely due to human failings. Seasickness (and probably poor decision-making) is mainly associated with vertical acceleration exceeding 0.8m/sec² in the frequency range 0.15 – 0.3 Hz, though other aspects, even smell, can contribute.²⁰This was used in the design of the ‘Castle’ class, where the vertical acceleration was kept to an acceptable level in sea states averaged over the year.²¹ Manual tasks are governed by transverse acceleration levels (Roll acceleration), and these can be controlled by large bilge keels or active fin systems.²² More costly solutions lie in the SWATH or hydrofoil, the ‘cost’ often lying in limitations in other aspects, such as training, as well as financial.

No 1 ship tank at Haslar, built in 1887 but with a new carriage installed in 1960. All British surface warship hull forms for both world wars and up to the early 1990s were developed and tested here. It was drained in November 1993 and is now an office. (D K Brown collection)

A simple approach to the value of improved seakeeping has been published.²³ Annual Navy Estimates show that the cost of keeping one frigate at sea for a day is about £100,000 and all or part of this sum is wasted if capability is reduced by motions. This can be used to demonstrate the value of considerably longer frigates. Though this approach is crude, no one has come up with a better one.

Secondary factors in hull form individually have only a slight effect on motions, but the cumulative effect of getting them all right is significant. Flare and knuckles can affect wetness but their precise value is a matter for heated argument.²⁴ The effect of motions on the crew of fast attack craft is very severe.²⁵ Hydrofoils should be much more successful in the role.

The dimensions and form of a ship have to satisfy the requirements of both powering and seakeeping. Luckily it is very nearly true to say that seakeeping dominates the fore end and powering the aft end form.

Strength²⁶

The concept of the structure of a ship acting as a hollow girder resisting the uneven forces of weight and buoyancy goes back at least as far as the beginning of the nineteenth century. Traditional, frame-built wooden ships were lacking in rigidity and liable to flex, leading to rapid decay from rot. Sir Robert Seppings (1811) introduced a system of diagonal framing that greatly improved rigidity and made it possible to build much bigger ships. Lang and Edye (Admiralty naval architects) adapted Seppings’ work for the structure of Brunei’s Great Western.

Overall, weight and buoyancy must be equal in any floating body but this conceals large differences at points along the ship. Local differences in weight and buoyancy lead to vertical forces in the structure (shearing force) and to bending along the length (bending moment). This approach was set out in principle by Rankine in 1866,²⁷ and made into a usable design method by Reed and his assistant White.²⁸ Further refinements were introduced following Biles’ experiments with Wolf after the loss of Cobra.²⁹

Seakeeping – what it is all about. A Mark III Lynx on the flight deck of a destroyer in severe weather. Manhandling weapons in these conditions is very difficult and hazardous. (Westland)

The ‘Tribal’ Gurkha during a comparative trial with the Leander class Hermione in 1978. This trial verified most of the computer estimates of the time as regards motions. However, it did show that the effect of slamming meant that stress quite well forward of amidships was higher than expected from earlier theories. (D K Brown collection)

In the years leading up to the Second World War, and for some time after, strength calculations were based on two situations. The ship was considered at rest, head-on to waves of the same length (L) as the ship and with a height of L/20. In the first case the ship was supported with a wave crest at each end reducing support amidships (Sagging). The other case was with a crest amidships and the ends drooping (Hogging). When the ship was sagging the bottom was in tension and the upper deck was compressed. In hogging the opposite was the case and the deck was in tension and the bottom compressed.

It was then possible to design structure to resist these loads. Tension loads were easy to deal with since all that was required was to keep the stress below a figure found safe from experience for that material.³⁰ Compressive loads were more difficult: there was no way of calculating the buckling strength of a panel of plating with stiffeners. An approximation was to consider a single stiffener with an appropriate width of plating as a strut. With experience, this worked quite well though the solutions were probably conservative (that is, heavier than necessary).

It was recognised that this whole approach was a crude representation of reality and, in particular, very long waves with a height of L/20 were rare. For this reason, nominal stresses in long ships were accepted at much higher values than in smaller vessels (for example, 9.8 t/in² deck in Hood compared with 5–6 t/in² for a frigate). Since the overall approach was comparative, with previous successful ships as the basis, there was a proper reluctance to make significant changes. This conservative approach was largely justified by wartime experience when, despite the use of high speed in bad weather, there were no serious structural failures in undamaged ships, though minor leaks were too frequent. Many ships broke in half after underwater damage.

A major objective of the post-war Ship Target Trials programme was to understand the behaviour of structures under explosive loading, and hence to design ships better able to resist such attack. Preliminary analysis by E W Gardner indicated that thin plating with numerous, closely-spaced longitudinals was the way to go.³¹ This also led to a lighter structure, and memories of pre-war limitation treaties still saw virtue in light weight per se while it was still believed that cost was directly proportional to weight rather than to complexity of systems.

In service, this type of structure proved expensive to build, prone to minor damage, and easily corroded. The Leanders, and even more the ‘Tribals’, walked back a little, with slightly fewer longitudinals and thicker plating – enthusiastic young men are often right, but overdo it. More thought was being given to fatigue strength where a metal exposed repeatedly to alternating stresses well below the maximum permissible level will eventually fail. Fatigue cracks in riveted structures would usually (not always) stop at the first seam. K J Rawson in his design of the ‘Tribal’ structure took the line that cracking of ordinary steel, assembled under shipyard conditions, was inevitable but, by keeping the stresses low, such cracks would not spread. He was proved right: examination of a ‘Tribal’ near the end of a long life showed numerous small cracks, none of which had spread.

The hull forms of early post-war frigates were based on W J Holt’s experience as a yachtsman, developed by model tests by Gawn at AEW, Haslar. They proved hard to beat, even with modern theory. This pair of model and ship photos (Leander herself above) shows that model performance does accurately represent that of the full-size vessel – only the size of the spray droplets is different. (D K Brown collection)

Ken Rawson also introduced the use of an American method, credited to Schade, of calculating the overall strength of a ‘grillage’, a panel of plate with stiffeners running in both directions. Schade’s method was only approximate and soon superseded by more accurate calculations, but it was a big advance on anything that had gone before – junior constructors were made to analyse old failures using the new method to prove its value.

Another advance in the ‘Tribals’ was their flush upper deck. Stresses in a hull increase considerably at a discontinuity – to break a strong stick, it helps to cut a notch. Several wartime classes, like the Southamptons and the previous ‘Tribals’, had suffered cracking at the break of forecastle, and the Daring had a massive reinforcement in this area, as did the Blackwoods. Rawson was able to show that the weight of such reinforcement was greater than that of extending the forecastle deck to the stern.

Several leading academics were pointing out that the behaviour of a ship in waves was dynamic and that the old static approach should be dropped. This seemed possible, since the development of seakeeping theory combined with the use of big computers had led to the ‘Strip Theory’. The hydrodynamic forces on each longitudinal section (strip) of the ship could be integrated as it passed through representative sea states. This was a valid approach and was used successfully for moderate sea states, but it was hard to apply and mistakes were easy to make. It was possible to consider the flexibility of the hull, important in slamming loads. However, it was not possible fully to represent the above water part of the ship, and the calculation failed as the deck went under.

Another number-crunching computer program was ‘Finite Element Analysis’. This split the structure into a very large number of elements and calculated the interactions between them when the structure is loaded. This tool is essential for heavily loaded or complicated structures, but input of data is a lengthy task, not to be undertaken unless it is needed.³²

As mentioned above, comparative trials had been carried out in severe sea states between a Leander and a ‘Tribal’ class frigate, to assess the effect of hull shape on wave loading.³³ The structural aspects of the trial were later reported by Clarke³⁴ and in particular the effect of slamming. The ‘Tribal’ class were designed with closely-spaced transverse frames forward which it was hoped would resist slamming loads. Unfortunately, it was not then known that the peak slamming pressures occurred further aft in the hull form adopted, and some further stiffening was needed. The maximum stress amidships was in fair agreement with theory but did not decrease as rapidly towards the bow as expected. A number of ships have been fitted with recorders to measure how often different levels of strain were exceeded during years at sea. This enabled a reasonable prediction to be made of the maximum stress likely to be reached in a 20–25 year life.

This data provided a sound basis for a simple design method not very different from the old Rankine method. This time the wave height to be used for all ships is 8m regardless of the size of the ship.³⁵ It has been calculated that there is a 1.5 per cent chance of exceeding this loading in 10⁷wave encounters – roughly 22 years service with 30 per cent of the time in ‘average’ North Atlantic weather.

Structural design had come to be seen as ‘easy’ and new management systems encouraged progress charts to be ticked as complete under ‘structure’ without real checking. There were a number of problems, some potentially serious: the Type 21 and Batch I and III Type 42s needed stiffening, clearly visible in photos. However, this was not confined to the RN. The French Tourville, the Soviet Sovremennyy³⁶ and the US FFG-7 classes all show signs of stiffening in service. An Information Exchange Project (IEP) was set up between the UK, USA, Australia and Canada on structural design. The first meeting was embarrassing as we each confessed our structural sins. More careful quality control has, it is hoped, largely eliminated such problems in design, though fabrication remains a problem.

Most of these problems arose from discontinuities or sharp corners, which multiply the stress locally.³⁷ Problems at the break of forecastle have already been mentioned, but there are similar problems at the end of superstructure blocks. Such blocks are often made short in the hope that they will not carry load, a hope which is often misplaced. A rigid superstructure will not flex with the hull and may tear away from the deck at the ends. The best solution is to arrange the superstructure ends on main transverse bulkheads that will accept the load and distribute it, but this is not always possible or is forgotten.

As a means of eliminating the superstructure-end stresses an approach which was actively pursued by the IEP team was the use of GRP superstructures. Because this material is much less stiff than steel it cannot impose strains on the deck. Tests have gone well and most problems overcome. The problems of GRP design had been overcome in the design of the ‘Hunt’ class (see Chapter 10).

This account has been confined to the overall strength of the ship, but there are many other problems which, though of detail, are important and difficult. Perhaps the most important of these is bulkhead design to resist underwater explosions without leaking round the boundary. It is probably too much to say that this problem has been solved but it has been greatly reduced. There are also problems of wave impact, of vibration and many others.

Structural design is not easy, but common sense guided by theory, together with a good eye for discontinuities and sharp corners, will work wonders and should give warships 25 years of relatively trouble-free life.

The Canadian frigate Margaree in heavy seas. This class was designed by (Sir) Rowland Baker, who believed that it was inevitable that green seas would come on board and fitted a turtleback forecastle to get rid of water quickly. No evidence was found to support his view and turtlebacks were not repeated.

(D K Brown collection)

Above: Edinburgh, a Batch III Type 42, showing the strengthening along the deck edge. (Mike Lennon)

The large test frame at NCRE, Rosyth, with a prototype GRP minesweeper hull inside. Note the size of the men. (D K Brown collection)

Submarine Pressure Hull Design

Submarine hulls have to resist water pressure at the maximum depth to which the submarine may descend, with a very small factor of safety, usually 1.5.³⁸ This factor has to take into account inaccuracies in the theory, imperfections in building, etc, as well as allowing for accidental excursions below the intended depth. Normal engineering practice with a low factor of safety would involve a proof test to about 1.2–1.3 times design pressure, but the risk to a submarine is thought too great.³⁹

Prior to the war, the only guide to strength for British designers was the ‘boiler formula’ – Stress = Pressure × Radius/Plating thickness. This was used in comparison with figures for submarines that had inadvertently dived well past their operational depth. This simple formula is surprisingly accurate for plating strength (yield between frames) and is insensitive to errors in shape, but assumes the frames are strong enough. Since there was no way of calculating frame strength, these were selected in very conservative – heavy – fashion.

The diagram opposite shows the possible modes of failure for a pressure hull. Overall collapse due to instability involving a whole compartment is more dangerous, harder to estimate and very sensitive to errors in shape – out of circularity – and is usually associated with inadequate frames. Early theoretical work was used in the Porpoise class, which used a few deep frames at intervals with smaller frames between. The Oberons had an improved design, with special T-bars of uniform size. The aim was to ensure that frame yield would not occur until twice the design pressure, even with the maximum shape error. This led to study of fabrication methods to ensure minimum practical out-of-circularity.

There is also a failure mode in which the plating between frames buckles in a large number of nodes. This is unlikely under static loading but can occur under explosive attack.

Few submarines are truly cylindrical, and even these have end bulkheads that pose a problem both in themselves and in their intersection with the cylindrical hull, with these problems acerbated by penetrations for torpedo tubes, etc. Many SSN have conical sections increasing the difficulties of calculation.

Much of the development work involved tests on large structural models. These had to be very carefully made, simulating to scale the out of circularity of real submarines. It was not possible to represent weld characteristics exactly but it was believed that any errors were on the safe side.

Fatigue failures occur when a structure is loaded alternately in compression and tension. Since a submarine hull is always in compression it might be thought that fatigue could be forgotten; but this is not so – the contraction of welds as they cool leads to locked-in tensile stresses which, when compressed, can lead to fatigue failure, particularly at changes in hull diameter and at penetrations. All these problems have been explored and solutions found, largely by S B (Bill) Kendrick and his staff at NCRE (USN designers use his work too).

A submarine model under test at Haslar. This later method of measuring forces and moments used the Planar motion Mechanism, seen here on the back of the carriage of No 2 ship tank. The model could be oscillated ‘bow up – bow down’ while readings were taken which could be fed into computer simulations of manoeuvres. (MoD)

A wartime X-craft midget demonstrates local instability with dimples in the plating between frames, caused in this case by an underwater explosion. (MoD)

The diagram shows three ways in which a submarine pressure hull can collapse under pressure. Buckling between stiffeners is the easiest to calculate and the only one that could be designed for until after the war. Overall collapse was solved by work at NCRE leading to hulls which were both lighter and stronger. Local instability is unlikely under normal loading but common after explosive attack.

¹ D K Brown, ‘Defining a Warship’, Naval Engineers Journal, ASNE (March 1986). The last two sentences above were framed and hung on the wall of Preliminary Design in Washington – signed by the author.

² Intact stability has been explained in the author’s previous books, in great detail in Warrior to Dreadnought, p207, and also in The Grand Fleet, p199.

³ See Warrior to Dreadnought, p207.

⁴ The Great Pacific Typhoon of 1944 which caused the loss of three USN destroyers and the subsequent investigations showed the value of the traditional GZ curve whilst leading to better methods of using the information. See C R Calhoun, Typhoon – The Other Enemy (Annapolis 1981).

⁵ The author did carry out calculations with trim for the ‘Tribal’ class (Chapter 5). Even for this very simple ship it took 3 months, but did show the need for a extra bulkhead and more freeboard aft.

⁶ This was a compendium of important instructions carried on board and is not to be confused with the Ship’s Cover, held in the design section.

⁷ During the post-war target trials it was usually forbidden to sink the ship, as its scrap steel was needed quickly.

⁸ T H Sarchin and L L Goldberg, ‘Stability and Buoyancy Criteria for USN Surface Ships’, Trans SNAME (1962).

⁹ D K Brown, ‘The Great Pacific Typhoon’, The Naval Architect (Sept 1985); and C R Calhoun, Typhoon – The Other Enemy (Annapolis 1981).

¹⁰ Knowledge that there is a margin for growth may encourage its profligate use.

¹¹ Confirmed by towing trials with Penelope, discussed under Stealth in the next chapter.

¹² Had the Germans invaded in 1940, DNC gave the highest priority to getting these books to Canada.

¹³ The experimenter would usually be a senior draughtsman (PTO II), often with four years in the Dockyard Technical College behind him, regarded as pass degree standard.

¹⁴ Accurate thrust and torque meters (not easy), rev counters etc. The trials section also had a deep knowledge of cheap and comfortable hotels near trials sites which would serve meals at odd hours.

¹⁵ Known as ‘Dickie’ Gawn – but not to his face.

¹⁶ Naval architects distinguish between ‘seakeeping’, the motions of a vessel in rough seas, and ‘seaworthiness’, which brings in practical factors such as the watertightness of doors and hatches, the strength of structure exposed to the impact of green seas, safe access, etc. Both are important.

¹⁷ Much of this chapter is based on a state-of-the-art survey: D K Brown and E C Tupper, ‘The Naval Architecture of Surface Warships’, Trans RINA (1988). For those wishing to dig deeper into the technology, there is a lengthy discussion and 63 references to key papers. Only a few of the more easily understood of these references are repeated here. The authors received the Institution’s Silver Medal for this paper.

¹⁸ R N Andrew and A R J M Lloyd, ‘Full Scale Comparative Measurement of the Behaviour of Two Frigates in Severe Head Seas’, Trans RINA (1981).

¹⁹ William Froude established the basic theory in 1860 but it is still difficult to solve the equations for real cases. See D C Stredulinsky, N G Pegg and L E Gilroy, ‘Motion and Wave Prediction and Measurements on HMCS Nipigon’, Trans RINA (2000), p248 for an up-to-date comparison of estimates and measurements.

²⁰ The author at sea in Cygnet in rough weather felt fine until his stopwatch revealed motions at 0.18 Hz when he knew he should feel ill!

²¹ D K Brown and P D Marshall, ‘Small Warships in the RN and the Fishery Protection Task’, RINA Symposium (March 1978). See also: D K Brown, ‘Service Experience with the Castle Class’, Naval Architect (Sept 1983). Our ideas worked!

²² The ‘Castles’ had very deep bilge keels and fins. During the Falklands War both ships lost their fins and no one even noticed. See: K Monk,‘A Warship Roll Criterion’, Trans RINA (1987).

²³ D K Brown, ‘The value of reducing ship motions’, Naval Engineers Journal, ASNE Washington (March 1985).

²⁴ The author’s views are manifest in the photo of the ‘Castle’ (see page 178).

²⁵ D K Brown, ‘Fast Warships and their Crews’, RINA Small Craft (1984), pt 6. This generated much controversy among the Light Coastal Forces Association but Captain Peter Dickens gave me general support.

²⁶ This whole section is derived from: Dr D W Chalmers, Design of Ships’ Structures (London 1993). This was originally written as an internal manual and it was my job to approve it for use. It was hard going but worth it. Dr Chalmers also revised the draft of this passage.

²⁷ W J M Rankine, Shipbuilding – Theoretical and Practical (London 1866).

²⁸ E J Reed, ‘On the Unequal Distribution of Weight and Support in Ships and its effect in Still Water, Waves’, Phil Trans Royal Society (London 1866).

²⁹ D K Brown, Warrior to Dreadnought, p185

³⁰ Stress is load per unit area. It is not the same as strain, extension per unit length.

³¹ D K Brown, A Century of Naval Construction, p203.

³² There is a risk that it takes so long that the answer comes too late to be used.

³³ R N Andrew and A R J M Lloyd, ‘Full Scale Comparative Measurement of the Behaviour of Two Frigates In Severe Head Seas’, Trans RINA (1981).

³⁴ J D Clarke, ‘Measurement of Hull Stresses in Two Frigates during a Severe Weather Trial’, Trans RINA (1982), pp63–83.

³⁵ J D Clarke, ‘Wave Loading in Warships’, in C S Smith and J D Clarke (eds), Advances in Marine Structures (Dunfermline 1986).

³⁶ At the 1993 Merseyside Review, the Russian destroyer Gremyaschiy showed a newly-fitted, riveted doubler at the break of forecastle.

³⁷ Years ago, one Chief, inspecting structural drawings, would start flapping his arms saying ‘I’m a stress. How do I get from A to B?’ A modern structural designer commended this approach but would have said strain rather than stress. The RN College lecture notes drew attention to this problem from at least 1913.

³⁸ The external loading section of the British Pressure Vessel Standard (BS5500) is based on RN submarine design practice not the other way round.

³⁹ Many submarine COs would take their boat some 10 per cent below the nominal diving depth to give confidence. This practice was well known and allowed for in the design and in the recommended operational depth.