t
r'. L'
L
r
For prèsentati'n at the 16th Meeting of the American Towing
Tank Conference, Sao Paulo, Brazil, 9 August 1971.
-Lab.
y. Scheepsbouwbnd
Hogeschool-.'DIft
t) e. r111 i J ROLL COMPUTER by Joseph E. RussNaval Ship Research and Development Center
29 MEl 1980
p
CONTENTS INTRODUCTION
PHASE A. FEASIBILITY STUDY OF TANK MOMENT MEASUREMENT USING PRESSURE GAUGES
PHASE B. HULL HYDRODYNAMIC CHARACTERIZATION PHASE C. EVALUATION OF ROLL-COMPUTING DEVICE FUTURE WORK
r)
ROLL COMPUTER
I. Introduction
The United States Navy is interested in developing a
ship-board device which computes the unstabilized roll of a ship,
given its tank-stabilized roll motion, hull hydrodynamic
char-acteristics, and the moment generated by the stabilizing passive
antiroll tank. The idea is to compare in some manner the
com-puted, unstabilized roll to the measured, stabilized roll in
order to evaluate the effectiveness of, and perhaps tune, the
passive tank. Cargo safety considerations and time* may
pre-clude making the comparison by simply running the tank dry and
then partially filled.
Some early developmental requirements imposed on the device were that it
1) be simple enough to be operated by members of the ship's
crew .
offer the user the option of determining under what
conditions the tank would effect minimum roll, i.e..
full, empty, or partially, filled. The device can be used to determine an optimum tank condition. for existing
environmental conditions, in addition to a simpler
eval-uation of the tank. Time and cargo safety considerations
may limit the device's use to evaluation of a relatively fixed tank condition.
provide a comparison of instantaneous values of
sta-bilized and estimated unstasta-bilized roll, and also a
comparison of various specifie time averages of these
quantities. Tentative displays include dual beata
oscilloscope traces of computed and stabilized roll,
and digital displys of one or .two weighted averages
of these quantities.
*The time to bring fluid to operating ievels in some tanks
i .acbfhaqvpn hciir
- -_'
y.--. . .predict unstabilized roll to within a maximum of ±20 per cent of its actual value.
5) be easily adaptable to many different operating
conditions of the same ship, on the basis of simple dockside measurements made of the metacentric height, natural roll period, displacement, and underway
measurement of speed.
IL
Phase A. Fèasibility Study ofTank Moment Measurement Using Pressure GaugesBesides hull hydrodynamic parameters and the stabilized
roll measuremént, the tank moment acting on the ship is an
additional qUantIty necessary to compute unstabilized roil
thotions, using an assumed linear model of ship dynamics. As
an inItial step in the development of the roll computer, a
feasibility tudy was made tö determine if the required tank
mothent could be well approxiniated on the basis of a practical
number of pressure measurements made at selected points in
the tank.
The shIp for which the device is to be initially
devel-oped is the ARIS-3 (Advanced Range Instrumentation Ship)
Maritime
Administration
Type C4-S-A3 troop ship. The ship isselected because a model of convenient size already exits.
The associated passive roll tank is shown in Figure 1.
The tank tnödel was mounted on a device equipped to
oscil-late the tank in a roll mode about a fixed axis. Three pres
sure gauge configuratidns of those judged most likely to enable accurate computation of tank moment were selected for test
purposes. These gauge configurations are shown s insets of Figure 2. The test objectives were o establish which, if any, of the arrangements (6 wäs arbitrarily chosen as the macinu
numbér of gauges toibe used in conjunction with the roll
corn-puter) provided pressure measurements enabling the total tank
I M - M1
t dt
Io
Tldt
i.e. a percentage error computed as the ratio of the average
value of the difference magnitude
- MT?I to the average
value of the true moment magnitude
hTI.
3
within IO per cent of its true value according to a specified
error criterion, over the frequency range of interest for this particular ship.
Additional objectives of the test were to determine the
necessity of gauges on the tank eiìd walls, and the relative importance of the varioùs gauge locations.
Tests were conducted for each of the three gauge config- L.
urations for sinusoidal ñotions of frequencies rànging from
.1 to .7 Hz, for roll amplitudes of 1, 3, and 5 degrees, the
maximum angle possible on the oscillator. For each run,
var-iables required for computation of
the
actual tank moment weremeasured and processed on-site by a small analog computer to
yield the time history of the actual tank moment. A similar
arrangement was used to cotfipute the approximate tank moment from pressure measurements made within the tank.
Boundaries of areas enclosing the gauges were chösen
such that each gauge lay at the centroid of its associated
area. The moment contribution of each gauge is then the. measured pressure multiplied by the area and an appropriate moment arm. The difference between the. tank moment NT
and
its approximation M was also corputed by the analog computer
for use in a performance index for each configuration. The
criterion chosen to evaluate the appröximate moment relative to the true moment was ....
Results. Results of the study ara presented in Figure 2. This fiure.indicates that, of the three arrangements tested, the one designated "B" best meets the imposed error criterion of 10 per cent, for the amplitudes and frequencies at which the tests were run.
III. Phase B. Hull Hydrodvnamic Characterization
In order to provide reliable input data for the feas
i-bility study, ARIS-3 hull hydrodynamic characteristics in the nature of a roll/tank moment transfer function were deter-mined by frequency response techniques, rather than the simpler and probably less accurate dockside technique envisioned for
use with the prototype roil computer. This roll/tank moment
transfer function is the heart of the computing device, and is essentially a calm-water characteristic independent, within linear assumptions, of the particular seaway in which the
ship is operating.
A roll axis passing through the CG and fixed in space,
was used for the tests. Frequency response tests were
con-ducted by oscillating the model, at speed, about this roll axis, using a push rod driven by an electric motor with
con-tinuously variable speed. The resulting roll records
and
the record of push rod force provide frequency response amp-litude and phase information relating the roll to the applied
momen t.
Figure 3 presents an experimentally derived roll/moment frequency response obtained for the ARIS-3 model at a speed of 2.27 knots. Superposed on the experimental data are mag-nitude and phase characteristics obtained by a digital
com-puter algorithm designed to fit the experimental data in a
mean square sense. The transfer function associated with
these frequency response characteristics is used in synthe-sizing the roll computer.
IV. PhaseC. Eva1uatIon of Roli-Coimuting Device
An expérimental study was next made to determine whether theroll computer met the specified performance requirements. A mathematical model of the dev'ie was programmed on an analog computer which was used iñ conjunction with model, tests of
the ARIS-3 in regular and irregular waves. RUns were first
made with the uns tabil.ized iodel operatIng' in bow and quarter-ing waves over the significant frequency range of its roll. This provided roll/wave slope transfer functions upon which were superposed ±20 per cent lines to define a band within
which estimates of the roll computer would be deemed acceptable. Runs were then made in beam and bow quartering seas with the stabilizing tank operating, and tank moment and stabilized roll were introduced as input to the ánalôg simulation of the roll computer. Its output, the unstabilized roll 'estimate,
was then sed to determine the unstabilized roll/wave slope
transfer function basedon the compute.d roll.
Figure 4 shows a plot òf ±20 per cent bounds of the "trué" unstabilized roll t.ransfe.r function and points of the estimated transfer function based on the computed roll in
a beam se.ao ' The data is preliminary. añd is a result of
cur-sory on-site analysis of strip chart records. The figure shows'
that the estimates are close to being within the ±2Q per cent specification for frequencies on each side of resonance., but that in the immediate neighborhood of resonance the estimated motion is decidedly smaller than the actual, motion and falls
within error bounds of ±30 per cent. It is hoped that a more.
carefulanalysis of the:data., including a study of the effect of hull damping factor in the sensitive resonant area, will improve the acòuracy of the computer over the entire frequency
range. Data for the. quartering sea case, run t determine coupling effects on thecomputerTs performance, is of similar
Future Work
Future work will first be directed toward a thorough analysis of the data obtained from the seaway experiment, and if necessary an improvement tqill, be attempted to resolve
the inaccuracies just mentionedo This will be followed by
a brief study of various averaging devices whose output is mt-ended to be displayed on the roll computer.
A prototype roll computer will be built, of integrated circuits and evaluated in full scale trials, in the latter part of 1971.
Preliminary Conclusions
The roll computing device, based on a linear model,
.g'ive surprisingly good estimates 'if uns tbilized roll,
except in the area of resonance. It is hoped this
discrep-ancy will be resolved by accurate data analysis.
For ARIS-3, the device appears valid for waveslopes up
to 0l8 at resonant frequency, corresponding to a full scale
sinusoidal wave amplitude of about 8 feet.
The assumption of a fixed roll axis appears to be jus-tified for this application.
J.E. Russ
PER CENT
RROR
-
Mrp1d
+ítMTIdt
GAGE CO?IGURATION "A"
GROE COiF'IGURATIOM "B" GAGE CcE'TGiJRATIOT "C"
(;U:; 2 -
:. r kt S Lìy
- T - -. .--iiH
-o
H
e-o
o
-o___=-.'J
- - --
I - b._o
II
.2
PHASE
MAGNI'IUDE
'ARIS-3 HULL MODEL
SPEED 2.27 KNOTS
O
MAGNITUDE (ExPErIMrìwL)O
PHASE (EXPERIMENTAL)COMPUTER FIT OF NAGNITUDE AD PHASE .\ \.;..
\t
qr't
.L LJL.!j
.5.7
1.0
2.0
5.0
7.0
10.0
FREC)uE?:cy (rad/sec)FIGURE 3 - Typical Rolì/Mc.:!cnt Frequency Response
of AlUS-3 Hull
ode11GJ
riL
O2L
A ARTS-3oor,
li
SFED 1.13 K'IS I BEAM SEA Ii'
t t 20 PERCENT OF TRUETS
FUNCTION I i I ¡ ¡ iII\'
I I I I,"
g7\
I C' 0 .3..2
.3 .11. .5 .6.f
.8
.9 FRF.CtJENCY (Hz) FIGURE 1i- Co:parion of "Truet' aud Estimated
O.
TRANSFER FUNCTION BASED CN ROLL COMPUTER ESTIMATE
r