I +National
Council CanadaResearch institute for Marine DynamicsCanad
Conseil national de recherches Canada Institut de dynamique marine SYMPOSIUM ON SELECTED TOPICS OF MARINE HYDRODYNAMICS St. John's, Newfoundland August 7, 1991EFFECT OF FORWARD SPEED ON ROLL DAMPING OF A DESTROY MODEL
M. R. Had4tra and D. Cu.Lng
Facu1y of En&Lna.ring d Appli.d Scienc., Meaorial. Un.Lv.rsLcy of .vfound1and Sc. John's N.wfoud1 CANADA A1.8 3X3
Natioia1 Research CouocLl, Institut. for Marine Dyrtaaics
I o
Nalionál Research Council Conseil ñationaj de recherchesCnadaCanada
Institut de Øynamiqu manne lflstitût6 fôr Marne DynarhicsSYMPOSIUM ON
SELECTED
TOPICS OF
MARINE
IiThRODyNtJIj
St. John's, Newfoundland
August 7, 1991
r4
Ç-.A3CT
An extensive experiment.,l program has been
carried out to study the eff.c
of forwardspeed on the roll damping of
a self-propelled,
9 meters long destroyer model. The experimental
set.up was deèmed te be as clos,
to reality as
practical in a towing tank
sstng. The roll
decay records vere analyzed usf.ng the Energy
approach, and a mcv flOndi.monsional.ization scheme
vas used. This scheme allowed
to identify
certain trends for the data,
meidto relate
daping of
the. model moving th non-zeroforward speed te that for à non a&vancthg model.
Thà damping vas found
to vary n0linearly with
forward speed.
I. rgraopUcyxov
One of che primary
sources of error in
existing ship motion prédiction Algorithms is
caused by inadequate methods
of estimating roll
damp ing. This
is
especially true whenconsidering large angles of roll.
Roll dampingh.ù recently received much attenion from the
research counity
asscientis
strive
toforujate improved
criteria to
increase themargin of safety against capsizing. Estimation
of roll damping
isstill larly based
onempirical formulatio derived fròm physical
mádel experiments. The purpose mf the present
investigation La to provide
a daca base of roll
da2ptng infornaej for a warship hull form and
to study the
characteristics 0f the damping
moment
at
largeroll
amplitude tobetter
w1derstd the physics
of roll damping.The damping
coefficient ealmates in this
paper were obtaine using the rgy approach
described in (lj
- The primary admncage of thismethod is the áiltty
to analyze 'enry short rolldecay recorda. The accuracy of che
method was
tested by
recenstrcting the roll decay
curves
using the estimated damping coeicienta
for a
larg. number of ca-ges (2-4 j.
It
.0 been shOwnthat the damping
coefficients obnsd from
theanalysis of a single
roll decay cycle producedmere accurate raU. decay Curves than
those obtained using the
classj
averaging tâchnique
-The preàemt paper describes the results of
an experimental program carried mut using
an
EFFECT OF FORWARD SPEED ON ROLL DAMPINC
OF A DESThOY MODEL
N. R. Haddara and D. Cumiiuing
Faculty of Engineering d AppU.d Science Mémorial University of Xevfoundland
S t. John' s, Nevfowd1a, CANADA All
3X5
National Research CouncU, Institute fOr Marine Dynamics
St. John's, Nevfoundlanci CANADA, All 3T5
unconstraixed selfpropeLL.d warshipmodel in the Clear Vater Towing Tank of the Institute
for Marin. Dynamics () in St
John's,Newfoundland. Roll decay di vere obtained for
a range of forward speeds
static stability
conditions for the
odeL fitted wich andwithout bilge keels. This rk compliments a
previous test carried out using
a laterally
constrained model described in (2J.An urow,antior
isen-sionlji0
scheme is introduced. This scheme helps in shoving the consistency of trénda in the data
and helps in de undersri.rw1fng of the behaviour
of damping as a functionof other variables.
IX. ERThENTAL PROCZg
Th. roll decay empenimentsvere carried
out on a 9 meter long. 1:1.3.5 scale warship modal in the IND 200 m x 12 e towing tank. A
body plan of the model Laprovided in Figure 1. The twin-screw model wns appended with
propeller shafts
/brackets, five bladepropellers a large emororline rudder.
Bilge
keel particulars are presented in TAble 1. Table 1. luge Keel Details
The bilge keel length is3.093 n and its span
is 0.045 e.
The odel wag self-propelledunder the towing tank carriage usingtwo 3 Kv electric propulsion motors. A singl,steel wire fitted from the bow of the model to a fixed point under the carriagewas used to tow the model.
during acceleration. Thea
a steady state
forward speed Vas attainedand the model was being propelled under its awn power, the tow cable vas permitted to go slack. Model rolling was then inkiced by the release of a moment
St&tioQ Distance.fo Disiance
ßaseljne (n) Centerline's) 12 11 10 9 8 7 6 0.465 0.483 0.495 0.500 0.497 0.482 0.45$ 0.138 0.120 0.115 0.117 0.121 0.124 0.132
Applied to a 2.5 m mast located at the model cântr. of floatation. s the modal -vas
uncon.serained during the collection of
roll
decay data. The model vu restrained during deceleration by applying. tension to a single steel vire fitted from the stern of the modal, to
a fixed point under the carriage.
For zeroforward spa ed runs, the bow and stern toe vires wir, disconnected fról the model and tb. modal
arranged across the tànk to avoid tank vail
reflec tien. The experimental set-up was deemed
te be as close -to the te.ality as prActical in a
tow tank setting. S difficulty vas
experienced contrólling mich
a large
modelwithin the restricted tas; area under th. tow
carriage afld
for several runs, only a few
quality decay cycles vere attained.
Roll amplitude vas measured using a 2-aXis
electro-mechanical gyro fitted on -the model
centerline. The analog signal from the gyro vas igitized at 20 Hz and recorded on the carriage licroVax It computer.
Data vere collected for five metacentric
height..., five forward speed.. (F5 - 0.0 to 0.25)
and initial heel angles from 5 to 27 degrees.
The five motaceneric heights are given in Table 2
where B i..
the model breadth and Ql is the
metacencric, height. Ballast weights were moved
laterally to preserve a constant roll moment of inertia. All five conditions-were tested for the
nodal fitted with bilge keels, however., lack of time permitted only four cinditions (Ql 61 of
basa vas omitted) to be tosted for the model without bilge keels fitted. A total of 259 runs
vere executed for the model fitted with and
without bilge keels.
III. DATA ANALYSIS
The free roll decay curves were analyZed using the Energy method. The method has been
discussed in several previous publications, see
for example (l.4J; and it vas shown te produce
fairly accurate predictions for the linear and
nonlinear damping coefficieflts.
It was also
shown, that the method is especially suited to
short roil decay recorde, for which none of the methode available in the iLterature are suited.
We vili give a ver)! brief description of
the method here, for the sake of completeness.Ve asume that the fr.. response of a ship
rolling in cala sea
can b. described by the
following differential equation:
28
e D(+) - 0 (1)
where is the roi-1 angle, if and D are the
damping and restoring moments per unit virtual
mass moment of inertia of the ship. Dots over
the variable indicate djff.rentiaeion with
respect to tise.
Equation (I) can be revriete.n in the
following form
(2)
where E(t) is the total energy of cha ship p.r
unit virtual moment of inertia. E(t) can be
expressed in th. following form
E(e) -0.5
+ G()where
G(+) -fac+de
Equation (2) is am expression of th. equality of the rate of total ship energy loss and the
rate of energy dissipation by the damping
moment.
Integrating equation (2), we get
E(e2) - 2(Ç)
-f
N(,)sdt (3)
CtEquation (3)
can then b. used to
identify the parters of the
dLng
moment. However, a form for the moment has to b. chosenfirst. The Energy method can deal with any of the well known forms. As a matter of fact, it
can deal with a general polynomial form where the damping moment is expressed as a function of. the velocity as well as the dLsplacemcnt To
perform a parametric study of the damping of the destroyer hull, one can use the simplest
form for the daing moment. That is the
equivalent linear damping. It bas becO found that this form is very useful in studying the
effect of varying- d different parters e.g.
velocity, fxequey .
.etc on th. damping.However, it w.st be bastzed that a monlinear
model should, always be used for response prediction especially when large motions are
considered.
-Considering a d.ing moment given by
N(+.) -N04
(4)Table 2. List of Metacentric Heights Tested.
Condition
Very High (VH) 0.12
High (H) 0.10 Medium (M) 0.08 Low (L) 0.06
i - Zero Forward Soee4
In this cale eh. nondjmenajo1 equivalent linear damping coefficient is given
by
t,4 - 8,4/p T' e
Th. mond1esional damping coefficient has
been plotted, for th.
case of zero forwardspeed, as a function of F1.
It should, b.mentioned her, that for thi,
case F1 Lsproportional to the mean amplitude .#. These plots are shown in Figures 2 and 3 for the cases of the model fitted vieh sOd withOut bilg, keels. For both cases, on. can identify the following trends:
For snail values of the parameter F1, the mondinaionai damping coefficient - is
inversely proportional to F1. Thismeans that
the dimensional damping
coefficient 54,
a constant independent of the amplitude
of oscillation.
For large values of F1, thecurves of 3 for eh. different values of CXconverge to almost
the san.
conleant. In other words, thenondtns tonal damp ing coefflc Lent becomes independent of F1. This indicatos that the
dimensional equivalent linear damping
ce.ffjcijat becomes a linear functionof the
1inud
of the roll motion.
Thus, ther.lationship between B and #. is e straight
line. The value of P at which this region starts decreasing as the metacentric height
decreases. This indicates that sa the
metacentric hait decreases, the
diflg tends
to be more viscous at smaller anglesof roll.
This can b. easily explained, since a decreasein the metacentric height, in
this case, is
caused by a rise in the vertical level of the
centr, of gravity which vili result in an
increase in the magnitude of the damping fórce
arm.
2 - Forvardsoeed Greater than -Zero
From the nondimensjonalizationscheme
aggzsted above, one can relate the dlmensianaj.
equivalent linear damping coefficient for the
model moving with forwaxd speed V, to that for
the model roiling at zero forward speed as
follow.:
$44v - B G (F,. ø,) Ii. (V/ca,0r)2J
The function 01 Ls a function of the
foriord velocity the natural frequency. The
analysis of the, damping data has shownthat the
function c1 can be written as
- G2(P,)x o4
E(e) Sty.. above,
fr.. roll decay curve, using the(3) Ls evaluated from an experimentally obtainedexpression for and substiD:1.ng in eh. L.LS. of equation (3),
least square algorithm. The &.H.S. in squacion
the .agnteud. of N can b. obtained using a
/
Tb. equ.ival.nt linear damping
cosfficj.n,
$. can be obtath.d from N0 by multiplying the
latter by eh. virtual moment of in.rtja
of theship. It is assuned that 3oo La
function ofeh. moan roll amplitud. #..
It vili b. shown
lacer thac this assumpUon is based on
.xp.rimectal evidence.
Iv. NOIOmALYZATION SCHENE
The follOwing nondiensional quantities
ar. used:
- B
rj, s/p u2 T
/(G2, $)where
02 - Va
T+)2
and p L the water density and
T is the ship
draft.
A pseudo Frouda number F1 is defined as
and the following expression is used for Fronde
F, - VI v'T
The velocity U, used in the aforementioned
non - dimensjona.ljzatjon scheu.,
is the resultant
of two velocities:
the forward speed of themodel, V. a velocity representative of the
transverse velocity caused by the rolling
motion, T
..
The ratio of the lattet to theforaer can b.
cortsjdered as a measure of theangl. of attack of
the hull when treatedas a
wing section for
the purpose of finding the
mognitud. of the lift
damping, The rational.behi.d ch. chOic, of these nondjmensional
parameters lie., in the fact that lift damping
beco..s signLfIc vieh forward speed.
V. RESUL AND DISCUSSION
The fre, roll decay curves obtained
experimentafly vere analyzed using the Energy
method and
th.
equivalent lineardamping
coefficient of the model vas evaluated for the
dLffere cases outlined above. The results
C2 is plotted in Figur. 4 for the modal fitted with and without bilge keels. It is shown that
C2 is independent of the natural frequency.
The .quival.nt linear 4aaping coefficjei
for the modal moving with forward speed can tham be obtained, from the damping coefficient of model rolling at zero speed by multiplying
-latter by the function G3 given b
G3 - G1[l.(V/o03]
obere G3 can be obtained from Figur. 4.
The n.c effect of forward speed am the
damping coefficient, 1 C the function C3) ha..
been plotted in Figure 5 for the eases with and
without bilge keels. It is clear that in both
cases,
the effect is noñlinear.
owev.r, thenonlinearity becomes more pronoied as the cacencric height decreases. This behaviour is
justified by the fact that G3
La a nonlinearfunction of the natural frequency.
For the -same step in forward speed, the chaitge in C3 for the case with mo bilge keels appears much larger than for the model with
bilge keels. This óccurs, even though the change in the damping coefficient should b. smaller f.m
the absence
of bilge
keels. This apparentcontradictión is caused by the fact that
thezaro speed damping coefficient far the del without bilge keels is only
a fraction of the
coefficient for the nodal with bilge keels.
Thus, if the-coefficients increase by thØ same
amount, the increase in the case of the model
without bilge keàls would be much
larger than
that in the case of the model fitted with bilge
kaels_
fl.
CONCLUSIONsIn this work, we have pré.sented the results of an experimental study of the damping moment of a self-propelled 4estroyer model. The
model vas allowed to have six degrees of freedom. The main objective
was to study the
effect of forward speed on the damping at large rolling angles. The Energy method proved to bea very valusbie tool for the analysis of the
free roll decay curve!, since most of the roll
decay curves obtained had only on. or two cycles that can be analyzed.
The accuracy of the
standard methods of analysis would be greatly
Çuestionable in this case.
The nondimeflsionslization schame used her. proved useful in shoving the consistent trends
in the results. It
showed that for large viluesof rolling amplitude the nondinensi1 damping
b.comes independent of both the roll amplitude
and the natural frequency. It also allowed us to
derive an expression relating damping
coefficient for the model moving with forward
speed from the Coefficient for the model rolling at zero speed.
30
One of the main conclusions of this work
is the -confirmation of the fact that the
additional ding caused by forward speed is a nonlinear function of the speed. A method of lift damping prediction which is based on a
nonlinear theory ii still lMking. At present, a study is being conducted to develop such £
method.
AGOVLEDGq5
The authors would like to express their
gratitude to the Defence Research E.tabl.Lsheent
Atlantic and ch. National Science and
Engineering Research Coil for supporting this research financially.
-REPE&CES
i.
Bass, D.V., Haddara, M.R.'Nonljnear Model, of Ship Roll Damping', Internacion4Thiobuilding Progxess, Vol. 35 No. 401 (1988). pp. .5--24.
Cising, D., Haddara, M.L, Crah, Ross 'Experimental Investigation of Roll Damping Characteristics of a Destroyer Módel',
Proceedings
of the
Fourth International Conference oñ Stability of Ships and Ocean Vehicles (StAB '90), Naples, ItalySeptémber 24-28, 1990, also 11W report No. LI-1990-3.Raddara, M.R., Bennett, P. 'A Study of
the Angle Dependence of Roll Daping Monent',
Ocean EneineeriUg, Vol. 16. No. 4. w. 411.
-427, 1989.
4.. Maddara, M.ft, Bass, D.V. 'On the Form of
Roll Damping Moment for Small Fishing
Vessels', PceanEi,eineerfn,g, Vol. 17. No. 6, pp.525-539, 1990.
.7igur. I. Mod.l body plan
andrdinats axes.
03DI)(s1óaIL ¡04-04 y £ MODEL 391 BODY PLAN 0.00P!ZUVO 7I! WUMO!! PÌ
Fig. 2 Zero
Forward Speed, Model with B.R.
01 5 04
rozuo i» ri
Fig. 3 Zero FOrwax.d
Speed, Model without s.jc. MCOWDITXO DCOJiDmoj - Ya ÇOWDITTO1I 0.06 4 31 0.01 maa pvtqcrioi - - VLcQKØmow.w,c Mcomoa.vpz V!coxomops.Tpg N COD1TIO. X5C
-.-.- Va
corDrnoN. M)E 71000! NV1E2 rs'Fig. 4 Forward Speed Effect,
Speed Function G2.
VI.corDmoy.Ya(
Va COWDI?IOj( 1F5
N IrDLTIO5,. ßE
Va Coi(DrnoN. M)!
78001E MUI!?p
Fig. 5 Forward Speed Effect,
Speed Function G3.