Superstructures and Structural Discontinuities

A superstructure may be defined as any structure which is built above the main
continuous structure of the ship. Such structures are usually shorter than the
length of the ship and they therefore introduce discontinuities into the structure.
Typical superstructures are poop, bridge, forecastle and deckhouses. It is
convenient to distinguish two different types of superstructure.

In the poop, bridge or forecastle the sides consist of a continuation of the side
plating of the main hull, whereas in the deckhouse the sides are usually set in
from the ship’s side. This difference in superstructures has an important
influence on how they behave structurally and this will be dealt with in detail
later.

The superstructure raises two important problems as far as structural strength

is concerned.

• The first is the discontinuity problem created at the ends of the

superstructure.
• The second is the distribution of bending stress in the superstructure.

It has long been recognized that the distribution of stress in a superstructure

may be non-linear or, in other words, bending theory may not apply. A great deal
of research has gone into the study of this latter problem in order to assess the
influence of the superstructure on structural strength and in this chapter the
fundamental principles involved in this problem will be discussed. Although the
superstructure represents the major discontinuity in a ship’s structure, other
discontinuities exist, and these will also be mentioned under this general
heading.

STRUCTURAL DISCONTINUITIES

Wherever the cross-section of a structural member changes suddenly, a

structural discontinuity exists, and under these circumstances the stress in the
member in the neighbourhood of the discontinuity cannot be determined by
simple theory. In fact, theoretical solutions (often based on plane stress theory)
only exist for a few simple cases. A typical one is the stress distribution in a plate
with a hole subjected to a tensile force as in Fig. 1.
Fig-1: Stress in plate with hole

The solution to this problem shows that, while the average stress clear of the
W W
hole would be  = and in way of the hole  hole = , the stress round the
A A−a
edge of the hole can be very much greater than these, and so it will be seen that
material which is otherwise subjected to quite a low average stress may locally
be stressed very highly because of the discontinuity created by the hole. The
ratio of the stress at any point to that obtained by simple theory is called the
stress concentration or stress magnification factor. The value of this factor
depends very much on the abruptness of the discontinuity, and it follows that it
is desirable to design structures in the neighbourhood of a discontinuity so as to
keep this magnification factor as low as possible.

The classical mathematical theory of elasticity has produced certain results for
holes and notches in laminae. The stress concentration factors at A and B of
Fig.2 of an elliptical hole in an infinite plate under uniform tension in the
direction of the &-axis are given by
A B 2a
jA = = −1 , jB = = 1+
  b

Figure 2
For the particular case of a circular hole, a = b and jB = 3, i.e. the stress at the
sides of a circular hole is three times the general tensile stress level in the plate,
while at top and bottom there is a compressive stress equal to the general stress
level. If a crack is thought of, ideally, as a long thin ellipse, the equation above
gives some idea of the level of stress concentration at the ends; a crack twenty
times as long as its width, for example, lying across the direction of loading
would cause a stress, at the ends, forty-one times the general stress level and
yielding or propagation of the crack is likely for very modest values of .

Figure 3

A square hole with radiused corners might be represented for this examination
by two ellipses at right angles to each other and at 45 degrees to the direction of
load. For the dimensions given in Fig. 3 the maximum stress concentration factor
at the corners is given approximately by

1 b
 ( 2b + 2r ) 
j= 1 + 
2 r
 b− r 

Figure 4 shows the effect of the variation of corner radius r on length of side b for
a square hole with a side parallel to the direction of stress and for a square hole
at 45 degrees. The figure shows
(a) that there may be a penalty of up to 25 per cent in stress in failing to align a
square hole with rounded corners with the direction of stress;
(b) that there is not much advantage in giving a corner radius greater than about
one-sixth of the side;
(c) that the penalty of corner radii of less than about one-twentieth of the side, is
severe. Rim reinforcement to the hole can alleviate the situation.
These results are suitable for large hatches. With the dimensions as given in
Fig.5, the maximum stress concentration factor can be found with good accuracy
from the expression,

It is of importance that the maximum stress occurs always about 5-10 degrees
around the corner and the zero stress 50-70 degrees round.

Butts in plating should be made at this latter point. Figure 6.43 shows the
results for a hole with L = B. Note that the concentration factor is referred not to
the stress in the clear plate but to the stress at the reduced section.

Figure 4
Figure 5: Rectangular hole in finite plate

Figure 5: Stress concentration factor for a hole with L=B

The effect of the high local stress is twofold.

• The stress concentration may be so great as to cause direct local failure of

the material, it is easily seen that with an average stress of say 4 ton/in2
and a stress concentration factor of 5 the material would pass into the
plastic region locally and failure might result.
• Should the member be subjected to alternating loads, even if the local
stress were below the yield point for the material, fatigue failure might
occur which could result in general failure of the structure.

From both of these points of view it will be seen to be desirable to reduce stress
concentrations as much as possible.

The study of stress concentrations can be made in different ways.

• On full scale structures the stress distribution at positions of discontinuity

can be studied by means of strain gauges and the electrical resistance type
is very suitable for this purpose.
• Another method of studying the problem is to coat the member in question
with a brittle lacquer. Under strain these lacquers crack, the cracks being
more closely spaced the higher the level of strain. By examining the
dispositions of the cracks, it is possible to obtain some idea of the intensity
of stress in the structure.
• A third and most useful method for studying this problem is the photo
elastic method. This method has the disadvantage that it cannot generally
be used on the actual material of which the structural part would be
constructed. The material must be transparent, so that a model of the
structure made of such material has to be used.

Figure 6: Stress distribution in notched beam

Figure 6 shows one of the beams tested and the stress distribution round the
discontinuity. In this diagram the stress in a direction tangential to the edge of
the beam is drawn perpendicular to the tangent. The beam was subjected to end
couples so that the bending stress should be constant over the parts of constant
cross section. This can be seen to be so, but locally in way of the discontinuity the
stress becomes very much greater than that on either of the parts of uniform
section. More recent work of a similar nature to this has shown that as the
radius of the corner is reduced the stress concentration factor becomes greater,
and one broad general conclusion which can be drawn from work of this sort is
that the more gradual the change in section the lower will be the stress
concentration factor.

Various discontinuities in the ship’s structure

Superstructure ends

As stated earlier, the ends of large superstructures represent the major

discontinuities which exist in a ship’s structure. For a superstructure such as a
poop or forecastle, where the ends are situated in positions of low bending
moment, the problem is perhaps not so severe. At the ends of midship
superstructures, however, where there is high bending stress, the greatest care
is required to avoid stress concentrations. The measures adopted to avoid
breakdown due to the high stresses generated may be summarized as follows:

Fig. 7 Methods of reducing height of bridge side plating

(1) The thickness of the sheer strake at the ends of the superstructure is
increased by about 50% and the thickness of the deck stringer plate is
increased by about 25%. This applies to long bridges. At the ends of short
bridges smaller increases in thickness are usually made.
(2) Bridge sides ate increased in thickness at the ends of superstructure' and
the side plating is tapered down gradually to the level of the bulwark in
the wells. Traditionally this has usually been a curve of the type shown in
Fig. 7 (a), but probably from a structural point of view some shape such as
shown at (b) or (c) would be better. These shell ' sweeps” as they are often
called are stiffened on their upper edges and supported by web plates.
(3) The stringer plate, deck plating and sheer stake of the thickness as fitted
in the wells is extended to the bridge for a distance of about one third the
breadth of the ship. The arrangements described here arc shown in Fig.8.
They are essentially practical measurements which do something to
relieve the abrupt discontinuity occurring at the ends of superstructures.

Fig. 8. Strengthening at end of superstructure

Deck openings

Holes cut in decks and particularly the uppermost continuous deck represents
other major sources of discontinuity. Such holes are unavoidable since they are
required for hatchway and machinery openings. This problem is also treated in a
practical way.

• In the first place, it is usual to concentrate the material resisting

longitudinal bending abreast the line of the openings and to reduce the
plating thickness of the decks within the line of the openings to a
minimum. The effect of this is to reduce the change in the section modulus
of the structure in way of the openings as compared with the modulus
clear of the openings. In this way the change in bending stress will be
reduced in going from a section clear of an opening to a section in way of
an opening.
• The corners of all deck openings should be well rounded and this is now'
usually adopted. Parabolic corners are sometimes adopted instead of
circular arcs. In addition to these measures it is usual to increase the
thickness of plating round the corners of openings in at least the upper
and second decks. In riveted construction this was done by means of
doubling plates but welded doubling plates are undesirable and with
welded ships insert plates are usually fitted. These have been shown
experimentally to reduce the stress concentration but apparently bad
fitting in this region can reduce the benefit obtained by such devices.
• Another desirable feature with hatch openings would be to reduce
gradually the height of the side coaming beyond the hatch end, and this is
done where the side and end coamings form a square corner.

Figure 9: Strengthening at hatch opening

Typical arrangements in way of openings are shown in Fig. 9. Where there are
openings in decks it would be ideal from a structural point of view to eliminate
altogether the plating within the line of the openings. The discontinuity problem
would then be removed. As, however, it is necessary to have this plating for
supporting cargo, maintaining watertightness or providing a walking platform,
this ideal cannot be achieved. On the other hand, it would not be impossible to
reduce the stress concentrations at the corners of opening by introducing a low
elastic modulus material for the plating within line of openings. For example, if
an aluminium alloy were used for this purpose, the stress in this part would be
only one third that of steel, and this should result in a much less difficult
problem from the point of view of stress concentration. It would be possible to
reduce still further the stress level if glass reinforced plastics were used, as
materials of this type have moduli about one-sixth that of steel.
Ends of girders and other structural members

Little need be said about such members. Wherever girders, such as deck girders
or intercostals, are situated in a region of high stress it is obviously sound policy
to avoid finishing them abruptly. They should be gradually reduced in depth and
tapered into the remainder of the structure. A good example of a girder changing
section is where a double bottom changes height, say in a machinery space. The
level of the centre girder in the double bottom and any inter-costal girders which
there may be should be gradually changed over a few frame spaces, as shown in
Fig. 175. In this way the effect of the discontinuity is greatly reduced.

Figure 10: Reduction in height of center girder in double bottom

Determination of scantlings of superstructure decks on the basis of simple
bending theory

The problem of the determination of the thickness of superstructure decks on the

assumption that they would take the full bending theory stress was first dealt
with by Montgomerie. In Montgomerie’s work the superstructure was assumed
to be made of the same material as the main hull. Since, however, it has become
not uncommon to build superstructures of aluminium at the present time, it will
be assumed that the superstructure deck is constructed of a material having a
different modulus from that of the main hull. It will be seen that it is then an
easy matter to obtain the result for a steel superstructure.

It is assumed that a structure whose cross-sectional area is A0, moment of inertia

I0 and distance from neutral axis to uppermost deck is y0, the material of which
has a modulus Es, is adequate to resist some external bending moment M. The
stress in the deck is
My
=
I
It is required to find the cross-sectional area a of a superstructure deck
constructed of a material whose modulus is Ea, so that the safe stress for that
material, a is not exceeded. The distance above the main structure at which the
deck is fitted is It (Fig. 184), and it will be assumed in the first place that the
vertical material, i.e. the deckhouse or superstructure sides connecting the
superstructure deck to the main hull, can be neglected. It is also assumed that
bending theory applies and that in the original structure the neutral axis is
below the half-depth of the section.

Fig. 11 Addition of superstructure deck

Referring to Fig. 11, it will be seen that the effect of adding the superstructure
deck is to raise the neutral axis to some position N’A' and the distance from this
axis to the superstructure deck becomes ys. Using the principle for composite
construction,

Ea
Equivalent steel area as = a
Es

as ( y0 + h )
Shift of neutral axis =
A0 + as

as ( y0 + h ) A0 ( y0 + h )
ys = y0 + h − =
A0 + as A0 + as

Moment of inertia of equivalent steel section about modified neutral axis

as 2 ( y0 + h )
2

= I 0 + as ( y0 + h ) − ( A0 + as )
2

( A0 + as )
2

I ( A + as ) + A0 as ( y0 + h )
2

= 0 0 = Is
( A0 + as )
Section modulus of equivalent steel section to superstructure deck

I ( A + as ) + A0 as ( y0 + h )
2
I
= 0 0 = s
( A0 + as ) ys

Now it follows that the external bending moment

I0 E I
M = s  = a  s  s
y0 Ea ys

Hence

Es  I 0 ( A0 + as ) + A0 as ( y0 + h ) 
2
I0
s  = a    
y0 Ea  A0 ( A0 + as ) 

Solving this expression for ɑs it will be found that

  s Es 
  ( y0 + h ) − y0 
 Ea
as = 0  A0   a 
I
I 0 + A0 ( y0 + h )
2
y0
It follows that

 s 
 ( y0 + h ) − y0 

a = 0  A0   a 
I
I 0 + A0 ( y0 + h )
2
y0

Where the superstructure is steel the conditions are a =s and Ea = Es, so that

I0 A0 h
a= 
y0 I 0 + A0 ( y0 + h )2

This formula is that produced by Montgomerie in the paper already referred to

and gives the minimum cross-sectional area required for a steel superstructure
deck so that the stress in that deck shall not exceed the stress in the upper deck
before the superstructure was added. It is instructive to consider a numerical
example to ascertain the thickness required in an actual case.
Numerical Example

A0 = 6251 inch2

y0=35.0 ft

I0=4601750 in2 ft2

It is desired to lit a superstructure deck 9 ft above the top deck, the breadth of
the deck being 80 ft

H = 9 ft

B = 80 ft

I 0 ( = 4601750 ) A0 ( = 6252 ) h ( = 9 )
a=  = 442in2
y0 ( = 35) I 0 + A0 ( y0 + h )
2

a = 442
Thickness of deck plating = = 0.46in
B(= 80 12)

This thickness would be quite high for a steel deck above a deckhouse. If the
superstructure deck was, however, the deck of a long bridge, then the thickness
would be reasonable, since the bridge deck would now be the strength deck for
the structure. In fact, it would be desirable to increase the thickness of the
bridge deck beyond the figure determined and to reduce the scantlings of the
decks below. In this way more material would be placed at the greatest distance
from the neutral axis which would lead to a better strength/weight ratio. This is
the procedure which is usually adopted. In the ship in question the thickness of
the strength deck was 1.06 in and the cross-sectional area including longitudinal
was 1160 in2. The corresponding figure for the deck next below was 450 in2.