You are on page 1of 6

(Invited lecture)

R. E. D. BISHOP*

A SURVEY OF STRENGTH CALCULATIONS FOR SHIP HULLS

S U M M A R Y : A n elementaor introduction is given to, and a even in a flat calm sea. Aside from propeller excitation, the
survey is made of, practical methods employed for estimating the overall dynamic distortions arise from wave loading and
strength of ships. The evolution of techniques is described, showing are essentially a matter of random excitation. It is to this
their developmentfrom the statics of a rigid body to the dynamic last problem that this survey principally relates. There is
anal,sis of an elastic body subjea to random loading. The need at least some justification for assuming that enough is
for basic research on the side of structural dynamics is pointed known about unit analysis and about propeller excitation
out. for the important practical problems to be identified (and
even solved, if only approximately).
The loading action of waves on a ship is normally a con-
tinuous random process. There is an important special
1. Introduction. case however which is usually associated with emergence
of the bows from the water; re-entry may then be impactive
Many types of loading are applied to ship hulls, and and the whole hull subjected to a transient vibration as a
Fig. 1 gives some idea of the range of possibilities. It is consequence. This behaviour is known as "slamming". It
therefore understandable that a practical strength calcula- can usually be alleviated by reduction of speed and its
tion for a ship can never be complete. Considerable emphasis study is a specialised branch of naval architecture [3].
has to be given to the process of idealisation prior to ana- The purpose of this lecture is to give a brief survey of
lysis and, for the sake of definiteness, we shall here restrict the practical techniques available for estimating the overall
our attention to ships that are under way at sea. (Docking, strength o f a hull in the absence of slamming. As we shall
launching, collision, grounding, ... , all raise special prob- see this is a matter of very great interest at present, since
lems of their own). very large modern ships (which are not normally subject
Some hull stresses are essentially localised. Those asso- to slamming) are beginning to raise serious questions as
ciated with the support o f a concentrated weight, with the to their ability to withstand passage through heavy seas.
thrust of a propeller, with the hydrodynamic forces ap-
plied to a rudder, ..., are cases in point. )t ship's hull raises
many structural problems of this sort and their elucidation 2. T h e statics o f a r i g i d ship.
may be referred to as "unit analysis". The high natural
frequencies of the units concerned (in relation to the driv- _A ship hull.is normally treated as a beam with a vertical
ing frequency of any forced excitation that may exist) mean plane of symmetry. Its strength is estimated for both sym-
that many problems of unit analysis can decently be dealt
metric and antisymmetric loading.
with by Statics [1]. Consider first the symmetric loading; it stresses the beam
There are certain problems, however, where dynamical
in bending. Fig. 2(a) shows the beam in still water. The
effects matter. They are mainly connected with motion
weight and buoyancy forces applied to it are shown in
of the propeller and so are usually associated with the after Fig. 3. The distributed net vertical force per unit length
end of the ship. That these are unit (i.e. localised) problems may be obtained by subtraction, and hence shear force
is of course determined by the frequency of excitation, and bending moment curves found. The appropriate
which is commonly the "blade frequency". "nominal" stresses can then be determined by one of the
Not all propeller induced vibration is localised, however.
procedures of beam analysis.
It has been found that gross (i.e. overall) distortion may It is not suggested that a technique like this would ever
arise from that particular source [2]. But this latter phe- be one and it is not surprising that early calculations of
nomenon is recognisable by the frequency of vibration, ship strength sought to "correct" the foregoing static ana-
and steps can be taken to avoid resonance - - at least in
lysis for dynamic effects.
principle they can. Suppose the vessel traverses trochoidal waves whose
Overall distortions of the hull can in fact be static or length is equal to its own. Figs. 2(b) and 2(c) show the ship
dynamic. The static distortion arises from differences of when a wave trough and a wave crest are amidships. The
load distribution and from built-in stresses, and it exists corresponding buoyancy curves are shown in Fig. 3. F o r
each of these two cases a curve of net vertical force per unit
length can be found and, hence, stresses deduced. (The
* Kennedy Professor of Mechanical Engineering in London areas under all the curves of Fig. 3 are equal because equi-
University; Fellow of University College London. librium in the vertical direction is assumed to be preserved).

SPECIAL ISSUE 1972 9


EXTERNAL ,,FORCES
I ....... i
I I I
BOTTOM REACTION
I i
INERTIA MISCELLANEOUS
GRAVITY PROPULSOR FLUID
TORQUE
WEIGHT OF: --& COLLISICN
THRUST
DARE HULL
t I I I MJNES
BARE PAYLOAD ICE PROPULSOR
HULL & UNDERWATER
& STORES
MACHINERY It EXpL~JSlONS
MACHINERY
EQUIPMENT I
LAUNCHING
I
DOCKING
I
GROUNDING
I
SEA-BED I FIRING
PAYLOAD SUPPORTS I I I WEAPONS
FIXED MOVABLE SLOSHING
HUMANS ~, THEIR TETHERING
IMPEDIMENTA
I I TOWING
tEE LACK ,S)F GYROSCOPIC
I BALANCE JACKING
TRAPPED WATER HYDROSTATIC HYDRODYNAMIC WIN.__O (E ACV GN
G~qOUND)
CONSUMABLE ITEMS : WATER I
CIJSHION LANDING ,~
FUEL (ACV) LAUNC~;~,G
WATER OF ~/C
STORES I I I t FENDE~NG
STEADY UNS"(E ADY IM,PA(:TIVE W~VES PERIODIC

HULL HULL SLAMIv~ING PROPELLER

HYDROFOILS hYDROFOILS POUNDING

CONTROL CONTROL CAVITATION


SURFACES SURFACES

STABILISERS

Fig. 1 - Sources of loading of a ship's hull.

In this way an attempt is made to find "worst cases" of manner. They provide a reasonable way round a problem
bending [1] [3]. The nominal stresses so found give some of great inherent difficulty.
idea of the worst that can arise in the structure, though a This approach to the "longitudinal strength calculation"
strict assessment of their real importance would again is certainly adequate for most ships and it has been very
raise questions of stress concentrations, of fabrication stres- widely adopted. It was in use by Lloyd's Register of Ship-
ses, of metallurgical behaviour and so forth. In practice the ping, for example, in 1874 [4] and (with many modifica-
worst nominal stresses allowed are usually of the order tions) remained so until at least 1958 [5].
of 9 tons/in2 for mild steel with a yield point of about Hull loading by waves on the sea introduces the problem
21 tons/in s. The nominal stresses are used as a general of antisymmetric loading. Lateral bending is caused by
guide to the designer, who is able to compare them with oblique waves and, the properties of shear centres being
the corresponding results for previous satisfactory (and what they are, this is accompanied by twisting. As with
unsatisfactory) designs of similar ships. longitudinal strength, suitable quasi-static calculations have
This is the basis of what might be termed the "classical" been devised for the antisymmetric effects, based on em-
approach to the calculation of ship strength. In practice pirical assumptions [1] [3].
the calculations may be refined considerably, both as regards The literature on ship structures is colossal. It is care-
the postulated loading and the type of beam analysis [1] [3]. fully and systematically reviewed by committees of the
But the object of the calculations remains merely to tell International Ship Structures Congresses, the proceedings
the designer "where he is" in a useful, but rough and ready, of which are really essential reading in this field [6]. Con-

(b)
, I

Fig. 2. - Vessel in (a) still water; (b) sagging; (c) hogging; Fig. 3. - Distribution of vertical forces corresponding to con-
conditions. ditions of Fig. 2.

10 MECCANICA
temporary ship structural analysis is to some extent still
based on the techniques that have been outlined here and
some idea of its complexity may be had from reference [6].

3. "Rigid ship" analysis.

Even with a correction for dynamic effects, the statics of a


rigid ship is clearly open to several serious objections. It is
to be expected that a real ship will respond to waves in
modes, that each mode separately is capable of being
brought to resonance, that coincidence of the encounter
frequency of the waves with a natural frequency of the
ship will usually produce large motions. It is of course to
be expected that resonant motion will be limited by damp-
ing forces that may arise in the structure and from the Fig. 4. - Ship moving with constant velocity in long crested
action of the water. None of these factors are countenanced sinusoidal waves of unit amplitude.
in the foregoing static theory.
Since about 1960 the static analysis of a rigid ship has
gradually been supplanted by dynamic analysis of a rigid
sided mean square spectral density of M is then given
ship. This change of emphasis is of considerable theore-
by [8]
tical interest. To explain it, it is necessary to digress.
The last 15 to 20 years have seen the emergence of the a,,~,(~o,) = I ~,,,~(,o,) I~ a,~(,o,)
study of " seakeeping" [7]. Seakeeping analysis is concer-
ned with the motions of a ship, assumed rigid, in a random for the given ~:, where @(toe) is the one-dimensional en-
sea and results obtained in that field are now used in cal- counter spectrum for a long crested sea. In effect •(00e)
culations of hull strength in a manner that will next be embodies the oceanographer's contribution to ship strength
explained. calculations and, as one might expect, such quantities are
So far as linear theory is concerned, small parasitic the subject of much careful thought; see Committee t
motions of a rigid hull about a steady uniform motion are report in Ref. [6], for example. If the bending moment
described kinematically in terms of surge, drift, heave, is a narrow band random process (because the motion is
roll, pitch and yaw. Appropriate generalised coordinates lightly damped and capable of being brought to resonance),
for these six motions may be arranged in the form of a then the bending moment in question will possess a
column matrix q governed by six simultaneous equations Rayleigh probability density function [8] and we may
that are in practice coupled in two groups of three as a calculate the probability of M exceeding some specified
result of symmetry and antisymmetry and are of the form value. The convention has grown up of determining the
minimum plate thickness of a hull so as to be capable
Aq+B¢ I+Cq=Q (t). of resisting such a value of the bending moment as is likely
to be applied once in the lifetime of the ship, i.e. a bending
Here A, B, C are 6 × 6 matrices of "inertia", "damping" moment the probability of exceeding which is about 10 -8.
and "stiffness", all consisting of structural and fluid con- The beginnings of this type of analysis can be traced back
tributions (the latter being frequency dependent if q varies to the 1950s. But the approach has been in everyday use
sinusoidatly). The column vector of generalised forces for a relatively short time. (An interesting example of the
Q(t) represents the excitation by the sea. Notice that, as the technique, in action so to speak, appears in a recent paper
six degrees of freedom, those six principal modes of the on a new generation of container ships [9]). Murray [10]
dry hull are used which have zero natural frequency. Mo- described the technique, as the basis of the prevailing
reover, typically, it is the fluid forces which couple the Lloyd's Register thinking, in 1965 and at that time the
corresponding equations of motion. transfer functions, such as HM~(toe), were mainly deter-
Consider a two dimensional wave train of unit amplitude, mined by model testing. Since then seakeeping theory
frequency of encounter co, and relative angle ~ (see Fig. 4). (whose object is the calculation of the motions q as random
For it, we may specify Q(t) and hence calculate q and hence processes) has been very greatly advanced, largely by Ger-
q and q. It follows that we may write down the gravity, ritsma and his co-workers in Delft, to the point where
buoyancy and inertia forces for any transverse slice of the theory and model tests now agree fairly well [3].
hull. This makes it possible to calculate vertical and ho- Just as the static analysis of a rigid ship is fundamentally
rizontal bending moment and shear force and also twisting a clever approach to a very difficult problen-, so too is the
moment for any section of the hull, all "per unit wave dynamics of a rigid ship. Philosophically the advance
amplitude" and all "for the given frequency toe and wave from the former to the latter was huge, but it has only been
angle ~". made at the expense of much greater complication. The
Let us take vertical bending moment 2]/I at any section added complexity relates mainly to the representation of the
as an example. The foregoing analysis gives the transfer fluid forces and this aspect is apparently becoming do-
function /-/Mg(c0e) for the given wave direction ~e; that is rninated by strip theory [3] [7]. But the assumptions that
to say it gives the (complex) bending moment at any sec- have to be made (particularly those demanded by strip
tion due to an incident wave of unit amplitude. The one theory) are sweeping indeed.

SPECIAL ISSUE 1972 11


4. W a v e - e x c i t e d h u l l vibration. ¢~:(+Je)
(it2 +,ec)
?
By the mid 1960s, several instances had been reported of
significant wave-excited hull vibrations [11] [12]. The
6
motion is sometimes distinguished by a special name
500,000 D~/T
(viz. "springing" or "whipping") and is apparently to be 5
observed only in large hulls. There is also evidence to sug- 200,000 Di~
gest that wave-excited vibration usually occurs principally 50,000 D~£

in the 2-node mode of vibration in the vertical plane of 3


hull symmetry.
Fig. 5 is adapted from a paper by Goodman 113] and it 2
illustrates xvhy wave induced vibration has become a
1
problem and why that problem can be expected to become
more serious as hull size is increased. The curve shows
°O..5 ' I I -r
a standard encounter spectrum #~(co,,) for a ship making 1.0 2.0 3.0 ~ .0 5.0

16 knots in a head sea(l). Typical frequencies o f the 2-node


symmetric mode are marked for a number of tanker dead-
weight tonnages. We see that as the ship deadweight in- Fig. 5. - A standard sea encounter spectrum showing the na-
tural frequencies of a 2-node symmetric vibration of various
creases so the appropriate resonance frequencies are moved sizes of tanker (showing appropriate deadweight tonnages,)
further into the significant region of the sea spectrum. after Goodman [13].
Wave excited vibration cannot be treated as a vibration
with a mere nuisance value. It throws the question of hull
strength under conditions of fluctuating stress in sharper they treat the system as if it is a passive one in the familiar
perspective because of the possibility of dynamic magni- sense described in books on mechanical vibration [15].
fication. It is mentioned by T o d d [2] that two sister ships The position may thus be summed up as follows: Starting
has resonant magnification factors (i.e. Q factors) in the at co = 0 in a spectrum like that of Fig. 5 and attempting
2 node vertical symmetric mode of: to estimate responses (and hence bending moments, shear
forces and torsion), one would
68 (riveted hull) and
a) investigate surge, drift, heave, roll, pitch and yaw
132 (welded hull). under the assumption of no distortion,
b) change to a new form of approximation to the modes
N o doubt one should not take such figures literally - - un- to proceed to higher frequencies (and so to get away from
fortunately damping forces are like that - - but the message the hump in the sea spectrum).
is clear. It is that resonance is potentially serious.
Unfortunately this vibration lies outside the scope of This theoretical approach may very well suffice provided
the rigid ship theory. Methods of coping with this new the dominant frequencies of the six "rigid" motions are
situation have therefore been proposed by Goodman [13] sufficiently removed from the lowest dominant frequency
and van Gunsteren [14]. These writers regard the vibration of distortion. But it seems questionable whether such a
problem as essentially something to be "added" to the rigid level o f empiricism will survive for very long, if only
ship analysis. Both consider the excitation of the modes of because it seems to leave many obscurities.
a ship, treated as an elastic beam, by a random sea. These
two papers are particularly relevant to the present discus-
sion and both are significant contributions to ship structural 5. T h e o r e t i c a l considerations.
dynamics. In general approach they have much in common
and it is worthwhile to dwell on them briefly. The contemporary position as outlined above is one of
As already mentioned, these papers are concerned only some interest from a purely theoretical point of view.
with responses in distortion modes as distinct from motions While the writer may well have missed one, the fact is
that are commonly thought of as "rigid", but are not so that he has never seen any suggestion that the dynamics of
in fact (i.e. heave and pitch). In dealing with the distortions, a flexible ship in a seaway could be tackled in any other
the authors adopt the conventional theory of beam vi- fashion. It has become second nature to the naval architect
bration and allow for the effects of "added mass" and to regard ships as rigid in seakeeping analysis. This un-
damping, using strip theory. But in extracting the modes doubtedly simplifies that anlilysis very considerably. But
it is not obvious that the arbitrary modes (i.e. the rigid body
motions) that are most suitable for seakeeping are the best
(z) For the reader who has a specialist interest in statistical for other purposes.
theory, curves like this were adopted by Committee 1 of the To digress briefly, the writer's interest in this situation is
Ship Structures Congress [6] and details of the one shown
heightened by experience with the dynamics of rotating
are:
shafts. It is usual for an engineer to think in terms of
significant wave height, h~/z = 4.O/too = 10 ft "static" and "dynamic" balancing of a shaft in a balancing
average period at zero crossings, T = 2~/(mo/m2) = 6 sec machine, as if that shaft were rigid. I f in fact the shaft is
where me, m2 are the area and second moment of the spectral flexible and is going to run through (or near) one o f its
density curve. critical speeds, then strictly speaking static and dynamic

12 MECCANICA
balancing are irrelevant and unnecessary and a technique matters which seem to demand special attention at this
of "modal balancing" is then called for. time and it is perhaps of interest to mention one or two.
The modes attributable to a rigid rotor are not possessed As we have seen, the stress system at any point consists
by a flexible rotor. In other words, if static and dynamic of a steady and a fluctuating portion. Even assuming that
balancing is practised on a flexible rotor, then the proce- the steady stress can be estimated adequately closely and
dure o f balancing the various higher (distortion) modes is that the fluctuating stresses can be analysed by a modal ap-
complicated by the necessity of not upsetting the previously proach, we are still left with an awkward problem. H o w
acquired static and dynamic balancing and yet not posses- does the combination of modal contributions vary in space
sing a simple orthogonality condition to make this pos- throughout the structure and in time?
sible. This has in fact formed the basis o f some controversy A second major source of misgivings relates to the pro-
in the literature [16]. One wonders if there is scope for perties of materials. Fatigue damage is not considered in
controversy in the present context of ship strength - - for present day calculations of hull strength, so far as the
a flexible ship does not possess the rigid modes of the dry writer knows. N o w fatigue is known to be associated with
hull. While it is quite permissible to use a mixture of ar- cyclic plastic deformation and this undoubtedly takes place.
bitrary modes (i.e. rigid modes) and natural modes, the In large welded structures there are inevitably sharp crack-
advisability of proceeding in this way seems open to like defects present in the welded joints prior to service.
question. In service these small defects can be expected to increase
A t present our understanding of hull dynamics appears and crack growth will occur. In a welded structure that
to be fundamentally deficient. Let us examine some of the operates in a corrosive environment such as the sea the
questions that cannot be answered at present: situation is more serious, as the rate at which fatigue cracks
grow will then be increased.
1) Would a knowledge o f the heave mode be helpful
The potential severity of this situation may be seen from
in focusing attention on a particular location within a hull
the statement [17] that, when corrosion acts together with
where extra care is needed over stress concentration?
fatigue, there is no endurance limit. No-one really knows
2) What are the spacings between resonance frequencies how important this point is in practice. It may merely mean
(of heave, pitch and the two-node symmetric modes, for that welded joints are likely to grow defects that are of
instance) and what does closeness imply? geographically limited extent and which can reasonably
3) H o w large can resonant distortion stresses be in be repaired when necessary during overhauls. The present
heaving and pitching? What part does cargo damping situation is essentially one of ignorance and it is worth
play? noting that if this attitude to damage of welds is indeed
justified, then particular attention will have to be paid to
4) How close to the maximum of the sea encounter
design for higher strength steels. In general fatigue crack
spectrum do heaving and pitching take us (see Fig. 5)?
growth rate goes up, and the crack size that is critical from
5) What modes other than the two-node symmetric the aspect of brittle fracture goes down, as yield strength
mode are a potential hazard? is increased.
6) H o w reliable are estimates that are made on the basis It is indeed quite possible that so far the day has been
o f nonlinear theory (for the effects on bending moment in saved by the accommodating properties o f mild steel. It
the vertical plane of bow flare, for example) by dealing seems most unlikely that considerations of cumulative
with one mode at a time? damage can be avoided much longer where the strength
of ship hulls is concerned.
All these questions, and others like them, seem to be
sensible and worthwhile. If the trend that is illustrated in
Fig. 5 continues (as it will), then satisfactory and scholarly
answers will presumably have to be found one day. If they
are not sensible and worthwhile questions then surely the 7. Conclusions.
reasons why they are not should be exposed.
It is now freely predicted that in ten years' time tankers,
bulk carriers, liquid methane carriers and container ships
6. P r a c t i c a l s t r e n g t h analysis. will be nmch larger than they are at present. T o build them
in our present state of ignorance would, the writer suggests,
It is at once clear that the sort of calculations that we have be to invite t r o u b l e . . A s dj,nam#s, present analysis o f hull
been discussing would be very difficult to make with any strength is both crude and circuitous. In particular, the
accuracy. What sea spectra should be used? What sort of semi-empiricism of rigid ship dynamics (which is admit-
life will a given hull lead and, in particular, how much tedly a great advance on what went before it) places re-
consideration will the master have for his ship ? H o w much sponses in the seakeeping modes on a different footing
alleviation of wave induced vibration can be counted on from the responses in the natural modes. Yet even if those
from cargo damping? These and other imponderables will responses could be adequately elucidated and steady com-
inevitably remain and will presumably have to be kept ponents found, the metallurgical problems that would still
always under close scrutiny. There are however one or two remain are great indeed.

SPECIAL ISSUE 1972 13


REFERENCES Trans. N. E. Coast Insm. of Engrs and Shipbrs., Vol. 81,
pp. 261-280, 1965.
[11 ] S. T. M^THEWS,Main HnH Girder Loads on a Great Lakes
[1 ] K. J. RAWSONand E. G. TUVVER,Basic Ship Theory, Long- Bulk Carrier, S.N.A.M.E., Spring Meeting, Montreal,
mans, 1968. 1967.
[2] F. H. TODD, Ship HulI Vibration, Edward Arnold, 1961. [12] A. O. BELLand K. V. TAYLOR,IVave-Excited Hull VibraZion
[3] J. P. COMS'rOCK(Editor), The Principles of Naval Architec- Stresses. Measurements on a 47,000-ton Deadweight Tanker,
ture, S.N.A.M.E., 1967. BSRA Report NS. 115, Nay. Arch. Rep. 41, 1966.
[4] W. JOHN, On the Strength of Iron Ships, Trans. I.N.A., [13] R. A. GOODMAN,IWave-Excited Maiu Hnll Vibration in Large
Vol. 15, pp. 74-93, 1874. Tankers and Bulk Carriers, The Naval Architect (Jour.
[5] J. M. MURRAY, Longitudinal Strength of Tankers, Trans. RINA), pp. 167-184, Apr. 197I.
N. E. Coast Insm. Engrs. and Shipbuilders, Vol. 75, [14] F. F. VAN GUNSTEREN,Springing. IVave-lndnced Ship Vibra-
pp. 39-56, 1958. tion, Internat. Shipbdg. Progress, Vol. 17, pp. 333-347,
[6] Proe. 4th Internat. Ship Structures Congr., Oslo, 1967. 1970.
[7] KORVlN- B. V. KROUKOVSKY, Theory of Seakeeping, [15] R. E. D. BIsHoP, G. M. L. GLAOWELLand S. MmHAELSON,
S.N.A.M.E., 1961. The ATatrix Analysis of Vibration, Cambridge University
[8] J. D. ROBSON, A n Introdnction to Random Vihration, Edin- Press, sect. 2.1, 1965.
burgh Univ. Press, sect. 4.3, 1963. [16] R. E. D. BISHOPand A. G. PARKINSON,On the Use of Ba-
{9] M. MEEK, R. ADAMS, J. C. CHAr'MAN, H. REIREL and lancing Machinesfor Flexible Rotors, ASME Vibrations Con-
P. WIESKE, The Strttclurat Design of the OCL Container ference, Toronto, 1971 (to be published).
Ships, R.I.N.A. Spring Meeting, 1971. [17] D. BIRCHON, The Use and Abuse of ASaterials in Ocean En-
[10] J. M. MURRAY, Fnrther Notes on the Strength of Tankers, gineering, Proc. I. Mech. E., Vol. 185, 1971.

] 4- MECCANICA

You might also like