DET NORSKE VERITAS
P. 0. BOX 82, OSLO
TANK SIZE AND DYNAMIC LOADS
ON BULKHEADS IN TANKERS
By
E. ABRAHAMSEN
REPRINTED FROM
TANK SIZE AND DYNAMIC LOADS
Introduction
The general tendency to build oil refineries in the oil consumption areas has resulted in a great
demand for very large ships which for the greater
part of their life carry crude oil to the refineries.
Improved storage facilities for refined oil products and vastly increased production and consumption rates have made it possible and usual to distribute refinery products as one grade cargoes in what is usually called «handy size» tankers, but which are quite often tankers of 32,000 t.dw.
From the point of view of tankers being built to accommodate many grades of oil products
simultaneously, it is obvious that only some
smal-ler tankers need have the fine subdivision
com-monly required to-day. For the majority of the
larger tankers other considerations than the
multi-grade cargoes must be deciding for the subdivision
requirements for tankers. Among such
considera-tions may be mentioned standing waves and
agitation in slack tanks, stability, flooding due to
collision or grounding, and structural strength.
These problems are
discussed here in greater
detail, and it is shown that from rational
considera-tions it should be possible to increase the tank size
in proportion to the increment in ship size. This means that substantial savings of steel weight,
piping and valves may be made for larger size
tankers in comparison with present practice. The necessity to consider in detail structural problems
resulting from the increased tank dimensions is emphasized.
Based on this discussion some proposed recom-mendations for the arrangement of cargo oil tanks
and the maximum distances between transverse
and longitudinal oil-tight bulkheads and wash
bulkheads are given in relation to ship dimensions.
1) Head of Research Department, Det norske Ventas, Orlo, Norway.
ON BULKHEADS IN TANKERS
By E. Abrahamsen')
Flooding and Stability in Damaged Condition
The fine subdivision of conventional tankers
provides a very good safety margin for the vessels when they are subjected to collision or grounding causing flooding of compartments forward of the
engine room bulkhead. Flooding of the engine
room of conventional tankers may, however, prove
disastrous when the ship is in a fully loaded
con-dition.
When considering an increase in tank length,
the problem of loss of buoyancy due to bottom or side damage should be considered rationally. To
provide a background for such discussions, flooding
calculations vere carried out for a tanker with the
following characteristic dimensional ratios:
B/L 0.1440
D/L = 0.0724
d/L 0.0537
CB 0.800
L length between perpendiculars
B breadth mouded
d draught
and
CB block coefficient.
For the flooding calculation a computer
pro-gramme worked out by Ström-Tejsen [1] was used
and results obtained for the permeability factors
u. 0.2, 0.4, 0.6, 0.8 and 1.0. For 0.2 the
floodable lengths proved to be greater than the
ship length. Curves for floodable lengths with the other permeability factors are shown for the fully loaded ship with zero trim in Fig. 1.
The floodable lengths shown apply in principle
to flooding across the breadth of the ship. The curves may, however, also be interpreted as
flood-able lengths of an empty centre tank which has a
breadth ratio b'/B' corresponding to the
permea-bility factor p., or as simultaneous flooding of two
wing tanks with a breadth ration 2b'/B'. B' is
Fig. 2. Floodable lengths for ballasted tanker with flooding across the full breadth of the ship.
under consideration; b' is the mean breadth of
the centre tank; b' is the mean breadth of the
wing tank.
Fig. 2 shows the floodable lengths for the same ship in ballast condition with mean draught equal to 51 per cent of full draught, with zero trim and a trim aft corresponding to 0.0092 L respectively. The curves have been drawn for a permeability factor of 1.0 only. It is seen that the floodable
lengths for tankers in such a light ballast condition
constitute no great problem, even if penetration of the hull in the region of the forward engine-room bulkhead might lead to loss of the ship under very
unfavourable circumstances. For our considera-tions penetration of a fully loaded ship will be governing.
If a tanker in a fully loaded condition has empty wing tanks, the loss of buoyancy due to flooding
of such tanks will cause an angle of heel as well as sinkage.
Fig. S shows the angle of heel
in dependence of the length of the wing tank withthe breadth ratio of the wing tank b'/B' as a
parameter.
Fig. 1. Floodable lengths for loaded tanker, with width of centre tank as parameter.
ULL 4 30 'J) w w 20 a D 10 20 30 40
LENGTH 0F DAMAGED REGION OF SIDE
TANKS IN PERCENT OF
Fig. 3. Listing angle of a tanker in relation to the damaged length and the breadth of wing tanks.
If the wing tanks are full of cargo oil, the
per-meability factor for such tanks will
not be far
from zero so that flooding of wing tanks in suchcases will affect draught, trim and heel to a small
extent only. This means that from the point of
view of flooding, centre tanks only should be kept
empty when the full capacity of all cargo tanks
is not needed. Such loading may cause rather high
tension loads on the cross tie, web and stiffener connections in the wing tanks outside the empty
centre tanks, but these structural parts are designed to take such loads in general.
If a tanker is to sail with empty wing tanks, the
heeling caused by possible flooding on one side would probably be counteracted by flooding the
intact wing tank on the opposite side. The reason
for this is that the loss of freeboard due to the heeling caused by unsymmetrical buoyancy, might.
as may be deduced from Figs. i and 3, be nearly
as great as the sinkage caused by symmetrical flooding. If two empty wing tanks are required on each side when the tanker is fully loaded, it
would be wise to keep one full wing tank between
the empty ones.
The general remarks about the loading of wing tanks are also applicable in the case of ballast, but
the great reserve buoyancy may make special
precautions of this type less necessary unless the
tank lengths are increased beyond normal practice.
To maintain the inherent safety of the tanker, which is a result
of the fine subdivision and
excellent means of closing deck openings in such
vessels, a requirement should be that any two
tanks within the cargo tank area can be punctured when the permeability factor is 1.0 without
caus-ing the ship to sink. If the punctured tanks are
bio i
0.20-/
-0.300.25)b
0.40-DEC 0.57 AT WATER K E DGE LEVETa
a
-.loo n z 's 75 'f) u, Ui 550 o F-(n u .5 25 Ui C-) (n û-2 L,J u .5 .5 1.5 ° z o (n o 1.0 ° S u- >- F-0,5 n o Ui > Ui
Fig. 4. Relative distribution of collision damage along the length of ships, according to Comstock [2].
wing tanks it should also be possible to flood the corresponding wing tanks on the opposite side to compensate for undue list without causing loss of
the ship. These requirements may seem rather strict but, as can be seen from Fig. 1, they are
easily fulfilled with a much smaller set of
subdivid-ing bulkheads than is normal practice today.
When considering the question of floodable length,
the type of damage which a ship can
sustain is of considerable importance. Firstly, it is
interesting to note that for the engine-room area
of a tanker which from the flooding point of view
is most vulnerable, there seems to be less likelihood
of damage due to collision and grounding than for other parts of the ship. This is clearly reflected in
Fig. 4 which shows the distribution of damage
over the ship's length according to Comstock and Robertson [2]. The same authors have also made
a statistical investigation
of damage length and
damage penetration, the results of which are given in Fig. 5.
Considering that the data have been collected
from about 60 collisions between all types of larger
ships, Fig. 5 gives interesting information about
the probable damage due to collisions. It is seen that the median damage length in the material
investigated is 8.0 metres, and the probability that
the damage length will exceed 16.0 metres is about
10 per cent.
The median penetration is 80 per cent of the ship's breadth, with 10 per cent probability that the penetration depth
will be half the
ship'sbreadth. Even if penetration depth and damage
length are to a certain extent connected, the
pro-bability that there will be large values for both is much less than the separate probabilities. It
might be added that the investigations described
100 .5 r 60 C-u, 70 (I) 60 50 o I-40 30 .5 20 Ui lo LU o-o
(A) LENGTH OF COLLISION DAMAGE
lO 20 30 40 50M
in [2] mainly cover collisions in American waters and that the findings are not necessarily
represen-tative for the rest of the world.
Owing to the real possibility of collision damage
to a loaded tanker in general leading to flooding
of compartments with a permeability factor close
to zero, it seems that the subdivision of a tanker
such that any two compartments except the engine room may be flooded in the loaded condition will make the tanker an inherently safer ship than other
types of cargo ships.
In the ballast condition
tankers are even safer than in the loaded condition,
especially when the side tanks or every other side tank are used for ballast.
It will be seen from Fig. 1 that a subdivision
factor of 0.5 may be maintained for conventional tankers with tank lengths i given approximately by
the following formulas: For centre tank:
B'
= (0.1 L - 0.2 x)
b'For wing tanks:
B'
i
(0.1 L - 0.2 x)
where:x the distance between the half length of
the ship and the centre of gravity of the tank considered. x should not be taken larger than 0.25 L.
-Ship motions
As mentioned in the introduction, the agitation period in tanks should be considered in relation to the period and the amplitude of the ship motions.
A.P 90 80 70 60 50 40 30 20 10 FP
DISTANCE FROM FP PERCENT OF L
(B) PENETRATION DEPTH/ SHIP BREADTH
Fig. 5. Damage length and penetration depth in collisions, according to Comstock [2].
Of the six possible ship motion components we limit ourselves to considering three or four, viz.
rolling, pitching, surging and possibly heaving. With k as the radius of gyration of mass of the
ship plus entrained water about a longitudinal axis through G, the centre of gravity, and GMT
as the metacentric height, the rolling period T r of the ship in still water is:
27rk
Tr =
(1)g GMT
In this expression g is the gravity constant. If the ship is subjected to an atwarthships wave train of the same period, synchronous rolling will
occur. The rolling amplitude may in such cases be large, dependent on the amount of damping afforded by bilge keels and appendages, wave generation and friction,
and especially on the
magnitude of the metacentric height. The response of the ship to the various wave components in an
irregular
sea may be described by frequency
response functions.For a range of loaded tankers which we have
investigated, we found that k varied between 0.34 B and 0.38 B, GMT between 0.10 B and 0.17 B, where B is the ship's moulded breadth. If we use
mean values for k and GMT. formula
(1) forloaded tankers with B in meters reduces to:
Tr 1.91/B seconds. (2)
For light ballast conditions corresponding to a mean draught of 50 per cent of full draught GMT may be in the range 0.2 B to 0.3 B, whilst k may
have a much wider range than indicated above,
dependent on whether ballast is carried in centre-tanks, in wing tanks or both. If we again use mean values for ballasted tankers, we obtain:
Tr 1.41/B seconds. (3)
Rolling periods much less than given by (3) may
be encountered under unfavourable ballasting and
wave conditions.
Pitching motions of the ship are usually cross-coupled with heaving motions due to the lack of symmetry about the centre of flotation. The
corn-hined motions may give the ship an apparent
pitching axis which is a considerable distance from
amidships, usually well aft of the midship point.
The natural pitching period of a ship may be
expressed by:
T = 2
GML
1/SL' +w)
(4) where JL is the mass moment of inertia of the
ship about a transverse axis, GML is the
longitu-o
dinal metacentric height of the ship, w is
the lumped entrained water coefficient, and ¿ is thedisplacement of the ship.
For ships of the tanker type which vary little
in arrangement and form with size, we may write
approximately for the longitudinal metacentric
height CML:
L3B CJL L2
GML - CJL = d (5)
where L is the ship length, d is the draught. CB is the block coefficient and CJL is the coefficient of waterplane moment of inertia. For tankers CJL
0.059.
Further we have approximately according to
Vedeler [3]:
B
(1+w)
(0.6+0.36)
and
JL kL2L2L\ (6)
where kL is the radius of gyration of mass about a transverse axis, through the gravity centre of the
ship with entrained water. For tankers kL varies
between 0.20 and 0.26 dependent on the load
condition. If we assume CB 0.78, B = 0.143 L,
loaded draugth d 0.0525 L and mean ballast
draught d = 0.03 L we have approximately: Loaded condition: T
2.lkL\/L
Ballast condition: T 1.95kLVLFor a ballasted tanker kL will usually be some-what larger than for a loaded ship. A reasonable average for our purpose would be to put T O.45VL.
Iii regular waves the ship will pitch in the
ap-parent period of the waves. In irregular waves the
ship will be more strongly excited by the wave
components having a period at or near the natural
pitching periods of the ship than by the other
components. provided that a reasonable amount of the spectrum energy is concentrated near the
natural pitching period of the ship. The response of the ship to the various wave components may be described by frequency response functions.
Investigations by Cartwright et al [10] and
others, for example, show the frequency response
functions for pitching to be rather broad, such that a broad frequency spectrum for the pitching
motion is also obtained in irregular seas. In rolling the frequency response function is comparatively
narrow such that a narrow rolling spectrum may
be expected for a ship in irregular waves.
The angular amplitudes in rolling and in
-J o >-oz w D o U)
zq
- *! cc z1 -Jo -J z 0< 400. U) o w 300 o, w 200 o D -J Q. 100 B o o 4 2 03 0,6 FREQUENCY 0F ENCOUNTER, A o 0 20 CUMULATIVE 40 60 PERCENTAGEwith the three-dimensional wave spectrum
experi-enced by the ship at any time. To get an idea of the relative magnitudes of the angular motion in
roll and pitch one may compare the pitch
fre-quency-amplitude spectrum for a ship in a fully
developed two-dimensional Neumann-spectrum
with a direction 1800 to the ship's course with the
roll frequency-amp]itude spectrum for the same
ship in the same sea, but with a course 90° to the direction of the wave spectrum. Results from such
an investigation are shown in Fig. 6 for two
dif-ferent tankers in a fully developed Neumann
wave spectrum corrcsponding to 40 knots wind velocity. The relevant data for this investigation
are given in Fig. 6 and are believed to be average figures for modern tankers.
The pitching periods corresponds to natural frequencies of 0.90 for ship A and 0.75 for ship B. In rolling the natural frequencies are
0.61 for Ship A ando, 0.51 for ship B. Even if the rolling amplitude may change con-siderably within the limits of variation of the
me-tacentric height and the damping created by
dif-ferent ships, the general tendency will be as indi-cated in Fig. 6.
The heaving motions of a ship are of less
im-20 o w U) o 15 "L 10 u, o D F- Q--e = O
F-
L-I"
1
SEAWAY: FULLY DVELOPED NEUMANN SPECTRUM, 40 KNOTS SPEED
ROLLING IN BEAM SEAS. PITCI-IING IN HEAD SEAS
portance in connection with surging effects in
tanks, even if they help to augment vertical acce-leration forces in the tanks, especially in the for-ward part of the ship.
Not much information is
obtainable on the
surging motion. It seems that this type of motion
is mainly created by the orbital motions in the
passing waves and by the repeated transformation
of part of the kinetic energy to potential energy as the ship moves from wave troughs to wave
crests. Also the surging is coupled with the heave
and pitch motions. The surging period is gene-rally given more or less by the period of wave encounter. The longitudinal accelerations are, however, generally rather small and of such an
irregular nature that they will usually not build up and maintain a strong standing wave system
in liquid tanks.
Measured values for the surging of a T-2 tanker
model in regular waves indicate a maximum
surg-ing acceleration of 0.05 g in a wave height H
L/30, wave lengths À ranging from 0.75 L to
7.25 L and with speeds varying from O to 16 knots
for the full scale ship. This result agrees well with values given by Denis & Weinblum [9].
SHIP SI-lIP A TP TR V (PITCHING) 217,9 M 66000 TONS 8,35 SEC 12,25 SEC 9,5 KNOTS 153,3 M 22000 TONS 7,00 SEC 10,30 SEC 8,0 KNOTS
Fig. 6. Roll and pitch spectra for two different ships in a fully developed sea corresponding to 40 knots wind speed.
O 03 0,6 0,9 1,2
FREQUENCY OF ENCOUNTER, WE 1/SEC
1,2
09
WE 1/SEC
100 BO
IEASURING CELL
Fig. 7. Arrangement of test set-up.
Fluid Motions in Tanks
If a tank with a free liquid surface is subjected to periodical motions the acceleration of the fluid
particles will create a wave system on the free
liquid surface. Standing waves with considerable amplitudes will build up when the natural periods of oscillation of the fluid happen to coincide with
the period of one or more of the harmonic com-ponents of the tank motions. The standing wave
frequencies will depend on the shape of the tank
and of the free surface and on the distance
between the free surface and the bottom of the tank. The amplitude of the standing waves will
depend on the magnitude of the disturbing force
(the tank motions) and on the fluid damping. At greater amplitudes the free surface will break
down such that non-linear effects are introduced. The natural period of the fluid oscillations in a
rectangular tank may, according to the linear
theory (see Lamb [4]), be expressed by:
1l
1T-27T 'n7rg
nrh
(7) tanki
where,
t the period of oscillations in seconds
n an integer 1, 2, 3
. .the gravity constant (m/sek2) the length of the tank (m)
h = the depth of the liquid (m)
Now if h is of the same order of magnitude as
/, the period t may be approximately expressed by:
T= 1.13
If the rectangular tank is part of a ship which rolls
or pitches freely such that the periods of rolling Tr or pitching T
are the same as
the natural period of oscillation T, resonant fluid motionswill usually be generated in the tank. According 8
to Yoshiki et al. [5] the fluid motions at the
reso-nance point are
supressed when the
verticaldistance e between the liquid
surface and the
centre of rolling is:c
g T21/4r
(8)Further resonance may occur when Tr
2 T,
provided that n is an even number. Our investiga-tions show that it is difficult
to build up heavy
resonant motions when Tr 2T11.
When the ship is heaving and pitching,
thetanks, especially those positioned away from the apparent pitching axis, will experience both ver-tical and horizontal translations and simultaneous
angular oscillations. In such cases violent fluid
oscillations will result when the pitching period
Tp approaches the standing wave frequency T in
the rectangular tank. Resonant conditions may also
occur when T = 2 T11 provided that n is an odd number, but again the build-up is not so serious as at the fundamental resonance point with T =
T1
In order to verify the theoretical findings and
also to establish figures for the pressure build-up
at bulkheads and tank top. Det norske Ventas
initiated in 1958 a set of experiments on an oscil-lating rectangular tank with dimensions 800 mm
>< 600 mm X 700 mm. The experimental set-up is
shown in Fig. 7. Mainly due to the limited length
of the connecting rod drive the tank oscillations were not strictly sinusoidal, but the deviations
were according to our acelerometer readings not significant.
Fluid pressures were recorded with membrane pressure cells positioned on the tank walls as
indi-cated in Fig. 7. The membranes of the pressure cells which operated with bonded strain gauge
transducers had a natural frequency of more than
4000 sec' such that
the dynamical calibrationfactor of the cells was practically 1.0.
As indicated in Fig. 7 the tank was tested with
three different axes of rotation, and the liquid
height h as well as the angular deflections a of the
tank were varied systematically.
Fig. 8 indicates the good correlation found
between the observed standing wave periods T1
and the calculated values. The small systematic
discrepancies may be due to the afore-meritioned
deviations from the true sinusoidal motions as well as to the non-linear effects encountered under resonant conditions.
In Fig. 9 the pressure build up SP/y h at the
position of the bottom pressure cell (No. 4) is shown in dependence of the liquid height - tank length ratio h/i and the period of oscillation of
C 10 .9 .8 7 s 4 .30 .20 .10 00 10 20 0
Fig. 10. Maximum dynamical pressures on the pressure cells in dependence of the liquid depth.
4 £ 2i-68B j1/ I
..O
3í.0.250 3.7.0125..
POINTS MEASURED- Yr2/anh(Á)
íìI
Irr
'-I
Ah
hF
2.0 4. 80 10.0 12.0Fig. 9. Pressure increment at bottom of tank due to
reso-nant agitation in tank when the angular motions is
± 5.
reduction in dynamical pressure obtained when
the tank is nearly full or when the angular ampli-tude of oscillation is so great that wave resonance
build-up is prevented by the tank top. With
arolling or pitching angle of 7,50 which is
reason-able and often experienced at sea, tanks which are
between 70 per cent and 90 per cent full will suf-fer from great resonance forces on the bulkhead
tops. Deck girders, transverses and stiffeners will, however, help to alleviate the resonance build-up
when the liquid level is approaching 90 per cent of the tank height.
0,5 1.0 1,5
Fig. 8. Standing wave periods according to theory and
experiments.
the tank expressed by the
dimensionless ratioT/Vl/g. In the above expressions
P is the
ad-ditional pressure due to dynamical effects, and y
is the specific weight of the water. We notice that the dynamic pressure build-up in this case seems
to be more or less independent of the liquid depth,
provided that the wave surface is free. We see
that the dynamic pressure peaks at the bottom
are about three times the static amplitude in these
cases when the angular deflection a
is ± 5.0
degrees. The resonance peaks found theoretically
for T = 2T2
are only slightly visible for the smaller liquid depths.The arrows above the resonance peaks indicate the positions of the theoretical resonance period for the three liquid depths.
The high dynamical pressures obtained in
oscil-lating liquid tanks are mainly restricted to the
regions near the liquid surface. This is clearly demonstrated in Fig. 10 which shows the
maxi-mum pressures measured on the cells in
depend-ence of the liquid depth. This point is amplified
by Fig. 11 which shows measured values for the
dynamical pressure .P/y. h as a function of the
vertical position in the tank and of the rolling or
pitching angle a.
Attention should be drawn to the considerable
1. 3.0 2.0 1.0 6,0 5,0 4,0 3,0 2.0 1,0 o
ay
Fig. 11. Maximum dynamical pressure on the tank wall
for the liquid depth indicated and for 4 different angular oscillation amplitudes.
As is evident from Fig. 8 the standing wave
period is radically increased when the liquid depth becomes small compared with the tank length. It is thus obvious that the liquid in very slack tanks
may come in resonance with the ship motions. Since the lower parts of the tank boundaries are designed to stand up to very high static pressure which is not present when the tanks are very
slack, damage is not likely
to occur unless the
same tanks are repeatedly subjected to this type
of dynamic loading.
The damping of the motions in liquids is
strongly dependent on the viscosity of the liquids.
The dynamic pressure generated in a tank with crude oil will be less than that generated with
water ballast under the same conditions, provided
that also the specific gravity is the same in both cases.
The influence of wash bulkheads was
investi-gated with cut-out areas corresponding to 3.5, 15
and 32.5 per cent of the bulkhead area and
inserted mid-way along the tank. Even with the
largest cut-out area of 32.5 per cent the wash
bulkhead proved to be an efficient reflection
bar-lo
rier for the standing waves such that the resonant
period was reduced corresponding to the reduc-tion in i which may be taken as the distance be-tween the end bulkhead and the wash bulkhead
in these cases.
Partial bulkheads with heights 150 mm and 250 mm protruding from the tank top half way along the tank proved efficient in destroying
re-sonance build up in all cases when the liquid level
was higher than the lower edge of
the partialbulkhead. Even with small immersions, the partial bulkhead acted as an effective wave reflection
barrier.
In many ships other shapes than the rectangular tanks are used to carry liquids, for instance stand-ing and horizontal cylinders, spheres etc. When
tanks with such shapes become large enough,
resonance between the fluid motions and the ship motions may create large sloshing effects,
depen-dent on the specific gravity and the viscosity of
the fluid carried.
In rocketry the dynamic effects of liquid fuel
motions may cause grave stabilizing problems, for
which reason fluid motions in many types of oddly
shaped tanks have been investigated. Reference
is made to the works of Bauer [6], Abramson [7]
and McCarthy [8]. The curves in Fig. 12 give
mean values of experimental results for the two
lowest frequencies of the fluid motions in the type
of tank indicated on the figure. Resonances of
large tanks of such shapes should be checked
against the periods of ship motions in all cases
when free surface effects are expected for short or prolonged periods when the ship is at sea.
The standing wave period is not dependent on the specific gravity of the liquid. The dynamic
pressure resulting from the standing waves is, however, proportional to the specific gravity. If
standing wave resonance is created, the dynamic
pressure may reach high and unwanted levels,
even if the specific gravity is 0.42 as for liquefied
methane. It is therefore recommended that tanks
for liquefied hydrocarbons should not be built so
that standing wave resonance with the ship
mo-tion may occur.
Since the natural pitching period is proportional
to the square root of the ship length L and the
standing wave period in rectangular tanks is pro-portional to the square root of the free tank length i, it follows that the permissible free tank length i
may increase in
direct proportion to the
shiplength L, without dangerous resonant conditions
occurring.
When estimating how close to the pitching H
:!'
H H H H 1.00 .90 .80 70 60 50 .40 30 .20 .100.7 o 0. o 0.4 ori 0.3 0.2 0.1 o h/R
STANDING CYLINDRICAL TANKS
02 0.4 0.6 08 1.0
'2R
HORIZONTAL CYLINDRICAL TANKS WITH LONGITUDINAL WAVES
Fig. 12. Standing wave periods of the first and second order for different tank shapes.
period T we may allow the fundamental standing
wave period in a fluid tank to be, we may make use of many hundreds ship-years of successfull experience with tankers of about 150 metres in
length and free tank lengths of about 11.0 to 12.0 metres. These figures give T(fl1)/Tp 0.68 0.71, and it seems safe to allow free rectangular
tank lengths of 1 0.075 L to 0.08 L when the
tank is reasonably deep, say h _ 0.5 i.
As we have seen above, the fundamental
stand-ing wave period is drastically reduced when the liquid depth is small, or when the tanks have a
spherical or a cylindrical shape. In such cases one
on 0.6 1.4 1.2 1.0 0.8 0.4 0.2 0 o - In
2Tr6n\/'
n=2 02 04 0.6 08 1.0HORIZONTAL CYLINDRICAL TANKS WITH TRANSVERSE WAVES
02 04 06 08 10
h/2R
SPHERICAL TANKS
should calculate T and T on basis of the mf
or-mation available and keep the ratio T/T
0.70.
According to Fig. 6 the average amplitude in
pitching with a period which is 0.7 of the resonant period under the conditions shown, i. e. for WE = 1.29 for A and c0E = 1.07 for B, will be quite negligible. Even if the longitudinal motions of a
ship in pitching, heaving and surging aggravate conditions leading to longitudinal agitation in
tanks, the breadth of tanks should be kept so small
that rolling angles at the natural period of the
atwarthships fluid agitation in the tanks do not
0.2 0.4 0.6 08 1.0
1r2_
Tn2TTón,J tanhh,.Errn
3.81 FOR n1 -7.02 FOR rir2 -26nftarhflh
0.5 0.4 0.3 02 01 0SECTION 4-A SEC ION B-5 Fig. 13. Proposed arrangement of wash bulkhead designed
to carry large vertical and horizontal reaction forces.
materially exceed the small pitching angles ob-tained at the frequencies above.
It seems, however, that a ratio between the
atwarthships sloshing period and the natural
rolling period Tn/Tr = 0.6 would be reasonable.
This corresponds to 04E 1.01 for ship A,ÛE =
0.85 for ship B. The distribution of roll amplitudes
to be expected in a 40 knots fully developed sea with frequencies in the range oJE = 0.96 - 1.06
for ship A and oSE 0.80 - 0.90 for ship B are
given in Fig. 6. It is seen that a mean rolling angle
of about ± 1.3° is obtained for both ships within the frequency range indicated. Since the ballast condition must be taken into account, Tn/Tr
0.6 means according to (3),
that the
free tankbreadth should not be more than about 60 per
cent of the ship breadth, provided that free
sur-face effects in tanks cannot be avoided.
It should be added that the above treatment is
an approximation. Actually data of the type given in Fig. 9 should be treated as frequency response
spectra together with the input data which are
the pitch and roll spectra for the ship in question.
Fig. 9
indicates, however, that the frequencyresponse function for reasonable liquid depths will
be comparatively narrow, and that the agitation
energy spectrum with the ratios T/T and
T/
Tr chosen will be moderate. Further work onthis
point has been initiated.
Structural considerations
From the point of view of flooding and of
dyna-mic effects from liquids, tank lengths may be
increased beyond present practice in larger
tankers, especially when efficient wash bulkheads are fitted. It is a question whether such deviations from current practice may be made so that a
say-12
ing in weight can be effected without undue risk
of fractures occurring.
Firstly,
the design of the primary structural
members in the bottom of such large tanks should
be mentioned. Instead of building up the bottom girder system with a high central girder as the
backbone, it will be more efficient to have several longitudinal girders of almost the same scantlings and a set of transverses attached to the
longitudi-nal girders to form an efficient grid system. The
distribution of stiffness in both directions should
be made from optimum weight considerations
within the limits set by the permissible stresses in this part of the ship structure.
A structurally efficient grid system should have
a length and breadth which are nearly equal. If
full advantage is taken of the upper limit of com-municating tank lengths, it will pay to design the transverse wash bulkheads as effective boundary supports for the bottom and deck girder systems. Such wash bulkheads will have to sustain rather large dynamical forces of a repetitive nature. The high stress low cycle fatigue aspect therefore has to be considered in the design of wash bulkheads as well as in the design of ordinary bulkheads.
Since the static head on wash bulkheads is small, the strength of a wash bulkhead structure
need not increase towards the bottom of the ship
as has been the practice up to the present time.
On the contrary, it is the free surface effects which have to be considered, and these have a maximum near the top of the tanks as shown by our experiments. From Fig. 11 we find that for the fluid level giving the highest dynamic pres-. sures in the tank, an angular roll or pitch of 7.5°
in synchronism with the fluid motions will produce
dynamic pressures on the upper part of the bulk-head between 5 and 15 times the change in static
pressure on the bulkhead due to the same angle
of inclination. This means a dynamic pressure head
corresponding to from one third to the full length of the tank at resonance.
Since free surface resonances will usually be
avoided, a design pressure head for the upper third
of the bulkhead corresponding to one half of the free tank length i seems to be reasonable. The
previously applied static head for wash bulkheads corresponding to half the head for oil-tight
bulk-heads could reasonably be substituted by a con-stant static head corresponding to one half the
total static head on the bulkhead. Further
investi-gations are needed, however, in order to clarify
this matter more fully.
In any case, it is essential to take the dynamic
000± COOCO
IÛ 0 j?0 O
SS4
s
type of loading on the wash bulkheads into account
and design all details very carefully. A general increase in tank lengths will also lead to a higher rate of occasional heavy liquid blows on both oil-tight bulkheads and wash bulkheads and thus
in-crease the tendency for cracks to occur at hard
Strøm-Tejsen, J.: «Damage Stability Calculation
by the Computer Dask». Danish Shipbuilders Computing Office, June 1960.
Comstock, J. and Robertson, J. B.: «Survival of
Collision Damage Versus the 1960 Convention on Safety of Life at Sea'>. Trans. Soc. of Nay. Arch. & Mar. Eng., New York, 1961, advance copy. Vedeler, G.: «Sea going Qualities of Ships». Sixth
International Conference of Ship Tank
Superin-tenderts, Washington, 1951, p. 162.
Lamb, H.: «Hydrodynamics». Dover Publications, New York, Sixth Edition, p. 363.
Yoshiki, M. and Hagiwara, K.: «Experiments on Dynamical Pressure in Cargo Oil Tanks due to Ship Motions». Trans. Soc. Nay. Arch. of Japan,
1961, advanced copy.
REFERENCES:
spots and unsuitable details.
In Fig. 13 a wash bulkhead transversing the
centre tank is outlined. This type of bulkhead will
provide efficient support for the longitudinal girders and simultaneously make a reasonable flow
of water through the bulkhead possible.
Bauer, H. F.: «Fluid Oscillations in a Circular
Cylindrical Tank». Report, No. DA-TR-1-58, Dcv.
Operations Div., Army Ballistic Missile Agency
(Redstone Arsenal, Ala.), April 1958.
Abramson, H. N., and Ransleben, G. E.: «Simulation of Fuel Sloshing Characteristics in Missile Tanks by Use of Small Models», Tech.
Rep. No. 3, (Contract DA-23-072-ORD-1251),
Southwest Res. Inst., March, 1959.
[81 McCarthy, J. L.: «Investigation of the Natural Frequencies of Fluids in Spherical and
Cylindri-cal Tanks», NASA TN D-252, May, 1960. Weinbium, G. and St. Denis, M.: «On the Motions
of Ships at Sea'>. Trans. Soc. of Nay. Arch. &
Mar. Eng., 1950, Vol. 58, p. 184.
Cartwright, D. E. and Rydill, L. J.: «The Rolling and Pitching of a Ship at Sea>'. Trans. R.I.N.A.,