4. Title and Subtitle
Development of a Time Domain Simulation for Ship Capsizing in Following Waves
5. Report Date
October 1973
6. Performing Organization Code 8. Performing Organization Report Na. 7. Author(s)
David M. Bovet
9. Performing Organization Name and Address Commandant (C-DST-2/62)
U. S. Coast Guard Headquarters
Washington, D.C. 20590
10. Work Unit No. (TRAIS)
11. Contract or Grant No.
13. Type of Report and Period Covered
Final Report
12. Sponsorin Agency Name and Address Commandant (G-DsT-2/62)
U.S. Coast Guard Headquarters
Washington, D.C. 20590 14. Sponsoring Agency Code
15. Supplementary Notes
16. Abstract
This paper describes the development of a time domain computer
simulation for ship capsizing in following waves. A survey of the
recent literature in this field is presented. The formulation of the
present approach is discussed, along with computer program limitations
and assumptions. The program developed is used to study the phenomenon
of low cycle roll resonance as demonstrated by a two-dimensional section
in forced heave motion and by a fast cargo liner in following waves, both
regular and irregular. The effect on capsizing tendency of variations
in ship characteristics, initial conditions, and wave conditions is
investigated. Qualitative comparison of results with open water experiments
is presented. Finally, the potential application of this program to the
determination of merchant vessel stability criteria is discussed, and recommendations are made for further work.
17. Key Words
Following Waves,
Ship Capsizing, Time Domain Simulation, Capsizing Literature Survey,
Computer Program
18. Distribution Statement
Document is available to the public
through the National Technical Informatic:
Service, Springfield, Va. 22151
19. Security Classif. (of this report)
Unclassified
20. Security Classif. (of this page)
Unclassified
21. No. of Pages 158
22. Price
The work reported herein was accomplished for the IJ. S. Coast Guard's Office of Research and Development, Marine Safety Technology Division, as part of its program in Vessel Safety Technology.
The contents of this report reflect the views of LTJG David M. BOVET, USCGR, who is responsible for the facts and the accuracy of the
data presented herein. The contents do not necessarily reflect the
official views or policy of the Coast Guard. This report does not
constitute a standard, specification, or regulation.
Reviewed by: . L. FOLSOM, LCDR, USCG
Project Officer
Submitted by:
/4%d '.
.ONES CDR USCG
h f, Marine Safety Projects Branch
(ThPg,
Released by:
C.L.'iAPT,
USCGchief, Marine Safety Technology Division Office of Research and Development
IT. S. Coast Guard Headquarters
BY
LTJG DAVID M. BOVET, USCGR
United States Coast Guard Headquarters Marine Safety Technology Division Marine Safety Projects Branch
This paper describes the development of a time domain computer simulation for ship capsizing in following waves. A survey of the recent literature in this field is presented. The formulation of the present approach is discussed, along with computer program limitations and assumptions. The
program developed is used. to study the phenomenon of low cycle roll resonance as demonstrated by a two-dimensional section In forced heave motion and by a fast cargo liner in following waves, both regular and irregular. The effect
on capsizing tendency of variations In ship characteristics, Initial conditions, and wave conditions is investigated.
Qualitative comparison of results with open water experiments Is presented. Finally, the potential application of this program to the determination of merchant vessel stability criteria is discussed, and recommendations are made for
further work.
This effort was undertaken in order to further understand a marine safety phenomenon of particular interest, that of
ship capsizings in following seas. The method of approach
was suggested by Professor J. H. Paulling, Jr., of the University of California, through his work on ship survivability. The
concept for the present effort along similar but simplified lines was greatly encouraged by Commander E. L. Jones, Jr., U.S. Coast Guard Headquarters.
My formulation of the problem was greatly aided by Dr. Nile Salvesen of the Naval Ship Research and Development Center, Carderock, who served as my thesis adviser. Discussions with him provided much impetus all during this project.
Assistance in computer programming was generously provided by Lieutenants H. K. Jenner and Frank Mlttricker, U.S. Coast Guard Headquarters. The graphics throughout this report are due to the drafting skill of Mr. Nonty Wing. Finally,
I wish to thank my wife for her moral support during this lengthy task. The author gratefully acknowledges the help extended by all these friends.
ABSTRACT
ii
PREFACE
iii.
LIST OF FIGURES
vLIST OF TABLES
vii.
NOMENCLATURE viii.
CHAPTER
I. INTRODUCTION 1
Description of Problem 1
Past Research 7
II.
METHOD OF APPROACH 20General Considerations
20
Detailed Formulation 2k
III. DISCUSSION OF RESULTS
38
A. Two-Dimensional Section Results
39
8. Results for Cargo Ship in Following Waves 50
IV.
CONCLUSIONS AND RECOMMENDATIONS 76REFERENCES CITED 78
APPENDICES
DERIVATION OF EQUATIONS .F OR HULL
IMMERSION RELATIVE TO WAVE PROFILE 80
DERIVATION OF COORDINATE
TRANSFORMATION EQUATIONS 85
ESTIMATION OF HYTEODYNAMIC TERMS
FOR THE AMERICAN CHALLENGER 89
C OMPUTER PROGRAM RMOMT:
HYIOSTATICS PROGRAM
93
COMPUTER PROGRAM SHIPMO:
SHIP CAPSIZING SIMULATION
105
Figure Page Wave
Encounter Period
and. Frequency asFunotion of Wc.velengthnd Ship :oa.ing
for a Given Ship Speed 55
Wave Enc.ounter Period as
a Function ofWavelength, in Following Waves, for a
Given Ship Speed
57
Time History of Vessel Roll Motion, Roll Restoring Moment, and Wave Height for a
Capsize Situation 59
Vessel Roll Notion for Three Different Cases 62
Unstable Roll Region for the
AMICAN
CHALLENG in Regular Following Waves 63
Time to Capsize as a Function of
Wavelength 65
Time to Capsize as a Function of
Wave Amplitude 66
21.. Linear Roll T).mping Effect on Time to
Capsize 68
Increased Stability Results for the
AMERICAN CHALLENGER in Regular
Following Waves 70
Time to Capsize as a Function of
V
Fiui
Page'1
Vessel With W.Ve Crest Amidships
2Vessel With Wave
Trough Amidships 2Time History of StabIlity and Roll Angle 2
Stable and Unstable Regions for the
Mathieu Equatioi
10
Approximate Regions of Stability for the
Mathieu Equation With Small Values of6
10
Experimental Results Compared
With Theory;
Paulling and Rosenberg
(1959)
12Coordinate System
25
Section Areas for the AMICAN CHALLENG,
Station 6.
at Various Heel
Angles andDepths of Immersion
33
Transverse Righting Moments About Keel
for the AMICAN CliALLENG, Station 6, at
Various Heel Angles and Depths of
ImmersIon..34
Flow Chart of Computation Procedure37
Unstable Roll Region for a Two-DImensional
Section ln Forced Vertical 0scillatio
43
Time to Capsize
as a Function of VerticalOscillation Frequency
46
Time to Capsize as a Function of Vertical
OscIllatIon
Amplitude
47
Time to Capsize as a Function of
Initial Roll VelocIty
48
TIme to Capsize as a Function of
Initial OscIllatIon Phase Angle
49
Table Page
1. Characteristics of Two-Dimensional Section LO 2, Stability Variation for Two-Dimensional
Section 41
Vessel Characteristics for the
AMERICAN CHMTENGER 53
. Irregular Wave Rims 73
A section area
Az heave added mass coefficient
sectional heave added mass coefficient
A0 pitch added mass moment of inertia coefficient A9 roll added mass moment of inertia coefficient
a9 sectional roll added inertia coefficient
B ship beam
Bz heave damping coefficient
bz sectional heave damping coefficient B0 pitch damping coefficient
BlO linear roll damping coefficient
b19 sectional linear roll damping coefficient B29 nonlinear roll damping coefficient
BN d.istance from center of buoyancy to metacenter
C(t)
time-varying heave restoring force coefficientC0(t) time-varying pitch restoring moment coefficient
C0(t) time-varying roll restoring moment coefficient wave encounter frequency
f(t) time-varying vertical wave excitation force g gravitational acceleration
GM transverse metacentric height
I moment of inertia
pitch mass moment of inertia viii
roll mass moment of inertia
k ship gyradius
height of center of gravity above keel height of metacenter above keel
1 ship length ship mass
S station spacing
Te wave encounter period T0 ship natural roll period
t time variable u wave displacement
V ship speed
X axis of coordinate system
X(t) general time-varying displacement x displacement in X direction
Y axis of coordinate system
Z axis of coordinate system
z heave displacement
wave amplitude
constant In Mathieu equation time step
constant in Nathieu equation
'1
wave elevation
e
roll displacement
NOMENCLATURE
'D.)
mass density of water
pitch displacement
heading angle
radial frequency
Description of Problem
The problem of severe ship roll motion in following seas has long been recognized but not adequately studied. It is a problem well-known to seafarers from experience, although its underlying causes are not generally understood. The fact that following and quartering seas pose a severe
safety hazard to many vessel types under certain extreme conditions is generally accepted. One of the paradoxes of this hazard is that its occurrence can be so sudden that a vessel may capsize before her crew is aware of the danger.
A brief description of the physical phenomenon involved can be presented with the aid of Figures 1 and 2. These
figures show a ship proceeding in the same direction as a two-dimensional regular wave system, with the vessel center-line perpendicular to the wave crests. With this orientation there is no external roll exciting moment present. In the first figure, a wave crest Is amidships, while in the second, a trough lies amidships. For a normal ship hull form with flared sections at both ends and wall-sided sections amIdships, It can be seen that the transverse inertia of the waterp].ane Is different In the two cases. With the wave crest amidships, inertia Is lost at the two ends without any corresponding
Figure 1. Vessel with Wave Crest Amidships.
Figure 2.
Vessel with
WaveTrough
Amidships.N
ROLL 4NGL'
N/
//
//
/
/
TIMe. O'
QOLL/
,/
I'"WI",
I/
/ NN //
Ro
TA.tLtY AXI5
- TIMEat the ends, the value of the transverse inertia of the water-plane is increased. Thus the position of the transverse
metacenter changes, and hence the stability varies, with a period equal to the wave encounter period.
This periodic variation in the vessel's transverse stability may lead to serious consequences under certain conditions. Suppose that the stability variation, caused by the vessel's encounters with regular following waves, has a
frequency twice that of the natural roll frequency. Starting with an initial roll angle and a wave trough amidships, as shown In Figure
3,
the vessel has a high stability value and returns rapidly to the upright position. Now as the vessel rolls to the other side, a decreased level of stability In the wave crest allows a heavy roll to occur. At the peak of this roll, Increased stability Is again caused by a wave trough moving Into the amidships position. This causes the vesselto right herself quickly and roll heavily to the other side against a decreased stability level caused by a wave crest amidships. In a regular wave situation, this process can continue over several wave encounter cycles, increasing the roll angle until capsizing ultimately occurs. In simplified
form, this is the capsizing phenomenon that will be investi-gated In this paper.
A recent case which illustrates the practical dangers of following seas concerns two 79-foot herring selners that
were brand new and were
tions in moderate seas. The designs had apparently been approved for government subsidy as meeting the 1)1CC stability criterion for fishing vessels. The conditions of the losses were both in following seas, with significant wave heights of about four feet, and speeds of about nine knots. The vessels
cap-sized suddenly causing the deaths of most of the crewmen.
Model tests confirmed the possibility of capsizing under these conditions, and pointed to the stability loss in following
seas as the likely capsizing mechanism.
The loss of a large West Coast tuna clipper in the
1950's
was analyzed in depth by Paulling(1960).
This vessel hadan overall length of 99 feet and a loaded displacement of Lj75
tons, with an initial metacentric height of 2.06 feet. In
still water, the righting arm curve peaked at around 15 degrees, declining then to yield a range of 60 degrees. Paulling
calculated the righting arm curve for two following sea condi-tions, one with the wave trough amidships, another with the crest amidships. A hydrostatic pressure distribution was assumed for the wave, and the vessel was posed in static
equilibrium on the wave. Wave length was taken equal to ship length, and height equal to length of wave over twenty. This calculation showed the vessel to have vanishingly small or negative stability at all angles for the case with wave crest amidships. It was theorized that the vessel capsized
in a wave crest, thus causing her loss during the maiden
voyage.
Bole and Kastner (l96) reported on a thorough
investi-gation carried out to determine the cause of capsizing of the German motorship LOHENGRIN In January 1963. This vessel was a 1,500 dwt cargo ship of 66.70 meters overall length,
built in 1958 for coastal service. At the time of the casualty, the LOHENGRIN was standing in to the Ki.el Fjord with some ice accumulation aboard as well as a full load of wood pulp.
Following seas with a significant height of about 2.0 meters were present when the vessel suddenly developed a k5 degree
list to starboard. This caused flooding of an internal com-partment through a deck vent which eventually led to the
sinking of the vessel. A major question for study was the cause of the Initial heavy list.
In a detailed study of this casualty, a 2.0 meter model of the vessel was built and tested under natural conditions on Pl5ner Lake. Wave conditions were obtained that
corres-ponded to actual conditions during the full-scale casualty. The model was self-propelled at various headings and with different stability levels. Capsizing of the model occurred In following seas with stability levels in the range of the estImated 0.13 to 0.22 meter metacentric height present with the actual ship. These stability values took into account the ice buildup, which alone was not sufficient to cause the
The tests demonstrated that it was through the effect of the following seas, chiefly related to the reduction of the righting arm in a wave crest, that the model reached an equilibrium
position at about a 5 degree heel.
The LOHENGRIN actually remained in this position for some two hours before down-flooding eventually caused the
vessel to sink. The existence of a second equilibrium position at a heel of LI5 degrees resulted from the particular hull form
and low initial stability of the vessel. However, this angle could only have been reached through the sudden large roll caused by reduced stability while the vessel was being over-taken by a wave crest. Thus, loss of stability in a following sea was a major cause of this casualty.
In fact, capsizing of small vessels in heavy stern or quartering seas is a significant safety problem. U.S. Coast Guard statistics (l9o8-71) indicate the loss of some 292 U.S.
fishing vessels and 25L1. tugboats due to foundering, capsizing, and flooding over a four-year period. How many of these casual-ties were directly related to stability loss in following
seas is not known. These small vessels are the most suscep-tible to such a hazard by the nature of their small size and low freeboard. A number of stability-related small ship
casualties were reported and analyzed by Rahola (1939) In his pioneering work which first advanced the concept of a
stabil-ity criterion based upon the area under the righting arm curve. Recently, Storch (1972) presented an excellent overall analysis
of stability-related problems of tugboats and a discussion of applicable safety criteria. However, the problem of deter-mining effective stability standards to withstand the hazards of following seas is not presently resolved.
This project has been undertaken with the purpose of developing a computer program for use as a research tool in the study of the phenomenon of ship capsizing in a following
sea. It is believed that a computer simulation will prove
to be the most efficient means of testing a number of hull forms under various following sea conditions. The simulation developed here is necessarily simplified, but it is believed that it can be used effectively to evaluate stability situations In pure following seas. The knowledge gained by extensive
simulation tests can be used to set safety levels guarding against this capsizing phenomenon. Once such criteria have been determined, they can be incorporated with other stability requirements f or wind heel, icing, towrope pull, etc. It is essential, though, that the safety hazard of following seas should be considered In the determination of overall vessel stability standards.
B. Past Research
Investigations into the problem of ship stability in waves extend back to Sir William Froude (1861). PaullIng
(1961)
briefly discusses the history of such research. Itstudy of ship rolling in following waves have been accomplished within the past twenty years. This body of work may be grouped under four headings: (1) work based on the study of the
Mathieu equation in the 1950's; (2) experimental and theore-tical work carried out up to the present time by Wendel,
Arndt, Roden, Kastner, and others in Germany;
(3)
experImental and theoretical work begun four years ago by a team underthe direction of Professor Paulling at the University of
California, Berkeley; and () other recent research. These
major research efforts will be briefly described here.
1. The Mathieu Eauation
An important realization in the study of ship roll motion in following seas was that of the applicability of the Nathieu equation to this problem. This equation describes the pheno-menon of parametric self-excitation, in which subharmonic oscillations of a system are caused by the periodic variation
of a time-dependent parameter of the system. This equation
governs a wide variety of problems in mathematical physics, and consists of a linear, single degree of freedom, system with a harmonically-varying restoring term. In the ship roll application the harmonic variation is caused by periodic
changes In the vessel's transverse roll restoring moment. These stability changes may occur due to changes in the
trans-verse moment of inertia of the waterplane or due to changes in the underwater volume, both of which may be caused by
relative vertical ship-water surface motion. In turn, this motion can be induced through forced heaving in still water, or through the slow variation of the relative wave profile in a following sea as illustrated in Figures 1 and 2,
The Mathieu equation can be written as:
+ (+eco3t)6 = o.
This equation does not consider damping, and assumes a pure harmonic variation in the restoring moment. However, the study of this equation does yield significant insights Into the general physical phenomenon.
Despite its limitations, the Mathieu equation can be used to determine regions of stable and of unstable solutions. Minoraky (1947) states that, in the linear form of the equation given above, a boundary between these regions can be found
based on the values of the constants,
b
and e
. In ourship roll application, 6 is the square of the ratio of natural roll frequency to the frequency of stability variation; E
is directly proportional to the amplitude of the stability variation.
Mapping the
£-6
plane results in stability chartsshowing unstable regions expanding up and outwards from points along the
b-axis as
follows:3 .
)
(=
o,%)iZ,±3)....)
Figure l. Stable and. Unstable Regions for the Nathieu Equation.
Figure
5.
Approximate Regions of Stability for the Mathieu Equation with Small Values of E.approximate representation of the main regions of interest, also due to Stoker, is shown in Figure
5.
These charts Imply the existence of unstable solutions In regions where the period of stability fluctuation has a ratio to the natural roll period of 1/2, 1, 3/2, 2, 5/2, etc. Within the unstable region, roll motion must increase inf in-itely, while outside the region any initial roll angle will be suppressed. At the borders ofthe region, periodic rolling
to some arbitrary maximum angle Is possible.
Paulling and Rosenberg
(1959)
were able to accomplish the experimental verification of the Mathieu4
diagram for the case of a )larIner ship model. The inclusion of nonlinearsecond-order couplings between heave and roll motions was
shown to cause the possibility of instability in roll, through excitation from the heave mode, as predicted by the Mathieu equation. This predicted Instability was verified by exper-Iments with the ship model where unstable roll motion was induced through forced heaving at certain frequencies and amplitudes. The comparison between experiment and theory
Is shown in FIgure 6.
Qualitative agreement between these model tests and the theory was good, although quantitative agreement was only fair. Differences observed between experimental results and theory
were attributed to the exclusion of damping and of certain
second-order derivatives in the theory, and to experimental problems of wave reflection from tank walls. It was stated
that inclusion of damping terms in the theory would tend to raise the amplitude required to produce instability at a given frequency. This can be understood as the absorption of excitation (input) energy by the damping (output) mechanism. Hence, values of the boundary line are critically dependent upon damping values, particularly in the vicinity of boun-dary minima along the 6-axis.
GO
LOCW.0.70 M0LEL
GM: O.0215X eAM, ZE0
IZOLL MOTIOP'J
H'JPUCEP bY 1OCEP VERTICAL
0cILLzmOPJ, \/ITh IlJITIAL 5'
0LL A)4GLE
o.Ia 0.10
0.08
0.06
0.040.02
0
0.% 0.15 0.20 0.25 0.50 0.55Figure 6. ExperImental Results Compared with Theory; Paulling and Rosenberg
(1959).
Kerwin
(1955),
durIng his stay at the Delft University of Technology, was one of the first to show that the single degree of freedom differential equation for roll motion In following waves can be written in the form of a Mathleuequa-tion. The main thrust of Kerwin's work was to calculate the change in stability due to following waves and to compute maximum roll angles for a given stability fluctuation. Cal-culations were based on the Mathieu equation using linear
and nonlinear damping, and some experiments were conducted with a fishing vessel model In following waves to confirm
the calculations. However, the stability of the model was such that maximum roll angles of only about 20 degrees were obtained, and took several hundred oscillations to build up. As Kerwin's computations were carried out by hand, the pro-cedure was rather laborious.
Following his work with Rosenberg, Paulling (1961)
carried out further Investigation of the transverse stability of ships In following waves, both theoretically and
experimen-tally. The appreciable decrease in stability (due to changed transverse moment of Inertia of the water plane) with the wave crest amidships and Increase with trough amidships, was
shown. Models were towed at fixed heel angles, free to pitch and heave, and righting moments measured for various following wave conditions. The variation of stability in waves was
shown to result from the altered characteristics of the im-mersed hull in waves. Further, the influence of model speed
stability variation was shown to be small also. Finally. Paulling pointed out that the effect of a following seaway on a vessel's ability to withstand capsizing should be con-sideréd when judging the vessel's ultimate performance at sea.
2. German Research
Since the mid-l950's, a team consisting of Wendel, Arndt, Roden, and Kastner has been studying ship stability in
follow-ing seas at the Universities of Hamburg and of Haimover. The body of work accomplished by this group represents the
only major concerted effort in the field outside of the Uni-versity of California. A major difficulty, however, is the
lack of English translations for most of the resulting
publi-cations. Brief discussion of two translated articles here
will have to suffice.
Kastner and Roden (1965) presented a paper to INCO describing the experimental approach to ship stability in following seas then underway in Germany. This consisted of model tests under different stability and sea conditions, conducted on a natural lake, to determine the minimum sta-bility required to withstand capsizing. As a result of these tests, required righting arms in a wave crest were determined
by statistical evaluation of the results. The authors sug-gest that vessel stability should be assessed by examining
the righting arm curve with a wave crest amidships, rather than by incorporating an arbitrary margin into the still water stability value.
Kastner
(1965)
carried out further analytical work based on the numerous ship capsizing model tests conducted. He utilized an analog computer to solve the equations ofmotion, and represented the stability variation as a normalized righting arm spectrum for all heel angles. The treatment
considered an irregular pure following seaway In a probabil-istic fashion. Results were obtained in terms of the sta-tistical length of time until capsizing. Limitations included accuracy problems of the analog computer, and the assumed spec-tra for righting arm curves. It is understood that Kastner is continuing his research in Hannover.
3.
Research at the University of CaliforniaBeginning in 1969, an intensive investigation was begun
into ship stability in a seaway at the University of California, Berkeley. This work was directed by Professor J. Randolph
Paulling, Jr., under U. S. Coast Guard sponsorship. Paulling approached the problem both theoretically and experimentally. As part of this effort, two frequency-based studies were
initially completed. Haddara (1971) studied the coupled,
three degrees of freedom, nonlinear motions of a ship in
regular oblique waves using the frequency domain. By consi-dering just the single degree of freedom roll equation,
Haddara (1972) was able to extend this analysis to the con-sideration of irregular wave spectra. However, these frequen-cy-based analyses were limited in their ability to represent the roll restoring moment variation caused by following seas. A ma1or thrust was carried out in the experimental direction to obtain improved understanding of the phenomenon of ship capsizing in following seas. An 18-foot model of the fast cargo liner AMERICAN CHALLENGER was equipped with extensive motion recording devices and was tested under remote control in San Francisco Bay. During these tests, variations to ship heading, speed, stability, and freeboard were carried out.
Results of these tests were presented in basic form by Paulling,
et. a].. (1972a), and analyzed by Kastner (1973). A good
summary of the overall project results was presented by Paulling,
et. a].. (l972b).
The most important contribution of these extensive model tests was the identification of three primary modes of capsizing. These may be described briefly as follows.
Mode 1, Low Cycle Resonance: This refers to a roll buildup during a series of particularly large following or quartering waves. As the crest of a wave is amidships, the vessel's stability is greatly reduced, causing a heavy roll. Next a wave trough moves into the amidships position, causing
increased stability and rapidly returning the vessel to the upright position. Now it a crest appears amidships, the vessel will roll slowly down on the other side against a weak
restor-Ing moment. If a wave trough now appears, the vessel will snap upright again. This process continues until either the model capsizes or it moves out of the wave group and
the motion dies down. This is the process which was described earlier in the paper.
This mode of capsize was characterized by a very reg-ular rolling motion which, in a group of three or four waves, grew rapidly to a large amplitude. The rolling motion was observed to occur at a frequency corresponding to the first
1'lathleu resonant frequency. Rolling sometimes occurred at the same frequency as that of the stability variation, hence corresponding to the second Mathleu resonant frequency. In
either case, the roll motion built up rapidly and. capsizing occurred as the model rolled into a wave crest.
Node 2, Pure Loss of Stability: It was sometimes ob-served that, with the model operating at high speed in very high waves, sudden capsizing might occur with little prelim-inary rolling when a crest amidships caused dramatic stability loss and capsizing. A high model speed was necessary to
induce this capsize mode, to produce a relatively stationary situation with the vessel caught in the wave crest. A Froude number of 0.4 is required for vessel speed to equal wave
speed, for a wavelength equal to ship length.
Node
3,
Broaching: This is the most dynamic mode, caused when the model is struck from astern by several steep breaking seas in succession. The waves cause the model toyaw off course to such an extent that the steering system cannot bring the model back on course between waves. The breaking seas striking the model combine with the dynamic heeling moment from the turn to capsize the model, again with the crest of a wave amidships.
The identification of these three modes of capsizing
and many other valuable insights were gained from the model tests conducted in San Francisco Bay. However, it was deter-mined that the number of test runs required to statistically
define the stability limit for safety from capsizing with adequate confidence was prohibitive. Therefore, Paulling's
current effort concerns the development of a six degree of
freedom numerical capsize simulation program, using as input fully directional wave spectra. Such a program would
vir-tually allow free-running model tests to be run on the computer. This represents a major digital computer programming effort, and results are not available at this writing.
M.. Other Recent Research
Several other investigators have recently attacked
certain aspects of the ship capsizing problem in following
waves. The Dutch government, concerned with fishing vessel safety problems, sponsored model tests in following waves
at the Delft University of Technology. The results, presented by Beukelman and Versluis (1971), show a fair comparison
waves for a series of four beam trawlers of varying prismatic coefficients. De Jong (1970), also at Delft University of Technology, carried out a probabilistic analysis of ship roll motion in beam seas and in following seas. His analy-tical treatment is based on the application of the Duff ing
equation to the beam sea problem, and on the use of the Mathieu equation for following waves. Recently in this country, a
group at the Virginia Polytechnic Institute and State Uni-versity has studied the phenomenon of nonlinear coupling between different modes of ship motion. Nayfeh, Mook, and Marshall (1973) reported analytical results indicating the existence of unstable roll motion Induced through nonlinear coupling with the pitch mode in the first and second Mathieu resonance regions.
Although there are undoubtedly other researchers working in this field, it Is believed that the efforts described
in this chapter represent the major advances to date. It
Is evident that a substantial body of excellent research has been carried out by investigators both in this country and abroad. However, we are still lacking the practical tools to permit detailed examination of ship capsizing situations to be made. It is in this direction that the present effort hopes to make some progress.
A. General Considerations
The objective of this effort is to develop a computer program for use as a research tool to study the phenomenon of ship capsizing in a following seaway. In particular, it is hoped to simulate the low cycle resonance mode of
cap-sizing. The method of approach chosen for this program
con-sists of a computerized time domain analysis of a ship's roll motion in pure following waves. This method utilizes the numerical integration of the equations of motion. The typical frequency-based equations, thoroughly discussed by Salvesen, Tuck, and Faltinsen (1970), are not adequate for this problem. Here the accurate time-varying representation
of the roll restoring moment is critical, and it can only be calculated by knowing the exact wave profile alongside the
vessel. This wave profile can easily be traced in a time domain solution.
The programming effort was carried out on an IBM 1130 computer until the computational procedure appeared to be completely functional. The program was then switched to a CDC 3300 computer using the remote job entry mode. The
pro-gram had to be slightly altered, but core capacity and run time were greatly improved. Fortran IV was used throughout
the effort.
The salient features and assumptions inherent in the approach used may be grouped as follows:
Three degrees of freedom.
Dominance of hydrostatic forces. Pure following waves.
Large roll amplitudes.
Numerical Integration techniques. These points will be briefly discussed here.
1. Three degrees of freedom
The analysis is limited to the consideration of three degrees of freedom. Heave and pitch are Included as they are considered essential In the determination of the Immersed
hull geometry, which Is critical to the transverse stability analysis. These modes are solved using linear,
frequency-Independent coefficients, except for the hydrostatic restoring force and moment, which are computed for the instantaneous hull-water surface position. Roll Is computed similarly, but with provision for a constant nonlinear damping coefficient. The instantaneous roll restoring moment is computed for each
ship section, thus incorporating the exact wave profile along
the ship's length, and Integrated lengthwise.
The vessel is assumed to advance at a steady speed on a constant course. Yaw and sway are Ignored, since the whole analysis Is limited to perfect following seas. Any
consider-ation of oblique quartering seas would require inclusion of these two modes, as well as inclusion of a wave exciting term in the roll equation. Finally, surge is ignored In order to simplify the computation of relative ship-wave
position. It Is hard to say what effect this assumption has on the stability analysis. Although it is believed that surge may affect the results, it is very difficult to obtain the necessary coefficients. In short, this assumption is made to allow a reasonably simple statement of the problem.
2, Dominance of hydrostatic forces
The treatment of static and dynamic forces adopted is intended to simulate specifically the condition of low encoun-ter frequency present In the following wave situation. Thus, the main effort is placed on a very accurate computation of the static forces acting at each instant of time. Conversely, the dynamic terms in the equations of motion are estimated using frequency-independent coefficients. It is anticipated that the dynamic terms are relatively unimportant compared to the dominant effect of the static forces under the low en-counter frequency condition.
3.
Pure following wavesThe wave environment considered consists of waves whose orestlines are perpendicular to the vessel's heading, arid which are advancIng In the same direction as the vessel.
The term "foflowing" seas is used to mean waves proceeding in the same direction as the vessel, regardless of whether the wave speed is greater or less than the vessel speed. Regular waves are represented by single sine waves.
Irregu-lar wave conditions are generated by a combination of sine waves of varying wave lengths, amplitudes, and initial phase angles. Vessel motion response is based upon this overall
irregular wave surface input, rather than upon the individual sine wave components utilizing the superposition principle of linear, frequency-based analyses.
Large roll amplitude
No restrictions are placed on the amplitude of roll, as would be the case in a linear, frequency-based analysis. This is a major advantage of the time domain approach. A roll angle of 90 degrees from the vertical is defined arbi-trarily as a "capsize" situation In an attempt to standardize the occurrence of this event. Although no special restric-tions are placed on pitch and heave amplitudes or on wave steepness, certain second-order inaccuracies are Introduced due to geometrical calculation methods if vessel motions in the longitudinal vertical plane are too violent. See Appendix A for details of the calculation procedure involved.
Numerical integration techniaues
mathematical techniques employed in this time domain approach. The equations of motion used to represent the physical situa-tion must be integrated time-wise. Nany techniques are avail-able for accomplishing this, including Runge-Kutta and pre-dictor-corrector methods. Here a Runge-Kutta method of the
second, order has been employed, using a straight-forward extrapolation of the motions based on constant accelerations across each time step. This is accurate so long as the tl.me
step used is sufficiently small. Areas and volumes used in the hydrostatic calculations are approximated using the trapezoidal rule. Thus, certain inaccuracies are inherent
in the results due to the mathematical approximations employed.
B. Detailed Formulation,
The specific formulation of the approach is discussed in this section, beginning with the coordinate system and the equations of motion. The calculation procedure employed for the time domain solution is then discussed in some detail.
1. Coordinate System
The coordinate system adopted for this study is shown in Figure
7.
Let (X,Y,Z) be a right-handed coordinate system fixed with respect to the mean position of the ship with Z vertically upward through the center of gravity of the ship, X in the direction of forward motion, and the origin in theFigure 7. Coordinate System.
plane of the undisturbed free surface. Let the translatory
displacement in the Z direction with respect to the origin
be z, the heave displacement. Furthermore, let the angular displacement of the rotational motion about the X and. Y axes
be
0
and0,
respectively, so that0
is the roll angle and0
is the pitch angle. Free surface elevatIon Isp.
As shown in Figure 7, the crestlines of the waves are perpendicular to the vessel's centerplane and are advancing in the same direction as the vessel for the pure following wave situation. Since the ship is advancing with a steady forward speed V, the expression f or the wave profile in this
coordinate system must account for the difference between
ship speed and wave speed. A derivation of the necessary equations Is given in Appendix A. Since the Z axis is per-pendicular to the undisturbed free surface, points on the
ship hull surface in the transverse plane must be transformed from their symmetrical coordinate system to that shown here, accounting for the roll and heave displacements. The equations
for this transformation are derived in Appendix B.
2. Eauations of Motion
The equations of motion adopted for this study are as follows:
Heave
(M+A),4e
+ ;£)z.
I f(x,t) ijx)
c&Pitch
(I##A.
+e =Roll
(t0+P18)Ô61é 466(81
+c (t)e = 0.
Alternatively, the accelerations may be expressed as:
=
-(e,,3. #c(e)
.-J'F
,v1x)ax) / (M#A)
(6
C#(e,,
)czjj*f(Z,t)It(Z)x
4.i
)/(104-A0')
0
-
(è
te!éI +C9( t.,it;1)Q
)/(t,+A9)
The basic approach here involves the estimation of linear,
time-independent coefficients for the added inertia and damping terms, and the computation of nonlinear, time-varying values for the restoring and wave exciting terms. Heave-pitch cou-pling in the added inertia and damping terms is ignored, in keeping with the simplified treatment of dynamic terms in
the low encounter frequency situation.
storing forces and moments are nonlinear and fully coupled, since they are computed at each step in time according to the actual elevation of the free surface, including the effects of heave, pitch, and roll along with the exact wave elevation,
For the wave exciting force in the heave and pitch equa-tions,. two assumptions are made. First, the wave diffraction effects are ignored, leaving only the free-wave profile (or Frou.de-Kriloff force) and discounting any effect of the vessel
on the wave structure. Second, the Froude-Kriloff force is
approximated by using the long wave approximation, thus ignoring the so-called Smith effect. This is equivalent to setting the term e in the Froude-Kriloff force equal to 1.0, which
leads to an error of less than twenty percent for a wavelength
equal to the ship1ength. This approximation for the exciting
force is necessary because of the difficulties involved in the time domain approach, where a specific wave profile is considered. possible improvement would be the use of con-volution integrals and Fourier transforms to compute the ex-citing force (and other dynamic terms as well) in the fre-quency domain, and then transform the results to the time domain. In practice, the hydrostatic restoring force and the exciting force, shown here as separate terms, are calculated as a single difference quantity by the hydrostatics subroutine. Finally, it should be pointed out that the inaccuracy in the
wave exciting force does not directly affect the roll motion
pure following seaway. Thus, the effect of this approximation on the capsizing results is not expected to be significant.
3.
Calculation ProcedureThe calculation procedure adopted f or determining the vessel trajectory in terms of heave, pitch, and roll can be outlined as follows:
a. Calculation of new position in time.
I. Vessel velocities and displacements are found by
integrating accelerations over the time step.
Ii. Wave position is found by Integrating wave velocity
over the time step.
III.. Relative immersed hull-wave geometry is determined
from items (i) and (ii).
b. Calculation of forces acting at a given point in time.
Hydrostatic forces are computed in a subroutine
util-izing item (a.Il.i).
L*mpirig forces are a constant function of velocities computed in item (a.1).
Added inertia terms are incorporated as constants within the vessel inertia terms.
c. Calculation of accelerations present at a given point
in time.
1. Accelerations are found with Newton's second law,
using the forces from item (b).
d.. Advano ement of time.
i. Time is advanced by a given step.
ii.. Entire procedure beginning with item (a) is repeated.
This is the basic process used to trace the vessel's motion in the time domain. Several simplifications are introduced
to keep calculation time at a reasonable level. Since the main emphasis of this simulation is on accurate representation
of the vessel's time-varying transverse stability, the largest calculation effort is spent on computation of instantaneous hydrostatic properties. The four Individual sections of this procedure will now be discussed In further detail.
a. Calculation of position
I. Vessel displacements and velocities
To start the program, vessel displacements and velocities are specified as initial conditions, along with the initial wave profile. Vessel accelerations are initially set equal
to zero. At the begInning of each time loop, vessel
displace-ments and velocities are calculated using a simple Integration procedure. According to Dorn and McCracken (1972), this
procedure can be considered a Runge-Kutta method of the second order, since it relies on evaluation of derivatives at only one point in time, and since the method used agrees with the Taylor series through terms in the second order in time step. The previous acceleration and velocity values are used to determine current displacement and velocity. Displacements and velocities In each of the three degrees of freedom are obtained as follows:
Displacements
X(t
X(t4
#1
+
if
or
x
(4:.)
+(h-)
(M.'2Velocities
(4k)
+I
(q)
at
or
x(-L''
(
th-IS)
+
Here, X(+.1)represents the new heave, pitch, or roll displace-ment. The time step Is
At,
which Is equal to tn-tn_i.ii. Wave position
The wave position at any point In time is found by
taking the initial wave position and adding to It the integral of the wave velocity up to that point In time. The first-order, small-amplitude linear wave theory is used, which
implies that the regular wave profile is sinusoidal. A possi-ble improvement would be the use of third-order Stokes wave
theory, as discussed by Salvesen (1969). Wave velocity Is computed according to the deep water assumptions. For irreg-ular waves represented by a sum of sine waves, the velocity and position of each sinusoidal wave component is computed and summed before the wave surface is applied to the vessel. This yields simple basic wave forms, but avoids any assumption
of linear superposition for the responses. Equations for the position of the wave crest in the X direction at a given point in time, with respect to the origin at initial time to,
are as follows:
u(t) = u(t0)
+f
(t) dt oru(t) = u(t0) +
i t where i =and u(t0) = initial position of wave crest.
iii. Relative hull-wave geometry.
Using the knowledge of the present positions of both the vessel and the water surface, their Intersect ion can be determined. An accurate estimate of this intersection Is needed In order to calculate accurately the hydrostatic restoring and exciting coefficients, c(t) and f(t), In the equations of motion. The detailed derivation of the equations
used to determine sectional immersion, or station "drafts,
is given in Appendix A.
b. Calculation of forces
I. Hydrostatic forces
Given the current relative position of the vessel in terms of roll angle and station drafts, it Is a simple matter to determine the required hydrostatic forces. The approach used is to pre-compute an array of data for a given hull form; this is accomplished by Program File RMONT, described in
Appendix D. The dimensions of the two arrays -- one for
section area, one for sectional transverse righting moment
- are heel angle, station draft, and station number. For
the cargo ship studied, ten heel angles, ten drafts, and eleven stations are used. Graphical representation of the arrays for section area and for righting moment are shown
in Figures 8 and 9, for one station of a cargo ship.
In order to find the hydrostatic values for the precise heel angle and draft present, a linear interpolation scheme
is employed. At this point in the computer program, the
values of heel angle and draft are fed to a subroutine (RSIM1) which carries out linear interpolation within the two arrays and returns sectional values to the mainline program.
A
trapezoidal integration is then performed along the vessel'slength to obtain the exact buoyancy (heave) force and the exact roll restoring moment. The pitch restoring moment is obtained from the longitudinal distribution of the buoyancy
forces.
ii. tamping forces
These are easily obtained at each point in time, since the coefficients are time-independent constants, and since the velocities are known for the given instant of time. The derivation of the damping coefficients for the cargo
ship studied is given in Appendix C. The only point of inter-est is the introduction of a nonlinear coefficient, B29
3500 3000 2500 w
2000
/500 /000 5000
)JOTE I7o5ITVE PAFT !6
0FJ.9 A'
SI1oil&i11EL
t'EGE
Pojró
6ttoir'J
dPO
A&e'
OFAVA1LAL TO
6Up MoTloj
?OGL1Ak1,?ornV 2AFT
Figure 8. Section Areas for the ANERICAN CHMTENGER, Station 6,
at Various Heel Angles and Depths of Immersion.
pgF% 58'
PZ4FTs Q' I
\
70,000
60,000
5000O
O)00O30,000
20,000
10,0001
PZAFT°PAFT
QRArT:II
38
28'
,;:;;1t:c
_v- .-&:
-U
-/0°
20030
40°50°
60°
70°80
HL ALE, P-3
POIP.iT6 Sr(O\/M S2E FOI't AE' O VILUE3
AVAJLASLE 10 5111? t4OTIOJf
PfQIM..
Figure 9. Transverse Righting Moments about the Keel for the
AMERICAN CHALLENGER, Station
6,
at Various Heelto account for some of the viscous effects and other
nonlin-cantles introduced in roll damping at large amplitudes of
motion. A value for this coefficient would have to come from an analysis of an experimental roll decrement curve from a model or full-scale vessel. In the case of the cargo ship studied, the value of this coefficient has been provisionally set equal to zero.
iii. Added inertia terms
The added mass and added mass moment of inertia coeff
i-dents, A, A# and A&, are taken as time-independent constants
for the hull form used. Thus the hydrodynamic inertia terms are incorporated with the vessel inertia terms. The derivation of the constant added inertia terms used for the cargo ship example is presented in Appendix C.
Calculation of' accelerations
The initial accelerations are set equal to zero. Then
accelerations at each instant in time are easily obtained from the forces calculated above, using Newton's second law. Expressions given for the accelerations in the earlier section on equations of motion are used.
Advancement of time
The last step of the time loop used for the numerical integration process is simply to increase the value of time
by some constant amount. The size of this time step is crit-ical to the stability of the numercrit-ical solution. Although more sophisticated integration procedures would permit larger time steps to be used, the approach adopted here is simply to select a sufficiently small time step such that further change in the time step would not produce any appreciable
change in the results. Following the addition of an incre-mental time step, the entire procedure beginning with item
(a) above is repeated.
This is the overall procedure utilized to obtain the vessel motions in the time domain. Further details are con-tained in the appendices, particularly In the two sections describing the computer programs developed. These are Appen-dix D, describing the hydrostatics subroutine, and AppenAppen-dix E, which presents the ship capsizing program file. A simple flow chart summarizing the overall computation procedure
T1M
The computer simulation developed was applied to the study of two particular cases. For the first example, a case was selected which was computationally simple and which clearly demonstrated the low cycle resonance mode of roll instability. This example consisted of a two-dimensional
section of rectangular shape which was harmonically oscillated in still water In the heave mode. The section was given
an initial roll velocity, and was allowed freedom In the roll mode. The heaving motion caused a periodic variation of the roll restoring moment, since the transverse stability was affected by the change In immersed area of the section. With the proper vertical oscillation frequency and amplitude, unstable roll motion was excited In a fashion similar to that
of the model experiments of Paufling and Rosenberg
(1959)
which were described earlier.The second example studied was the cargo ship AMERICAN CHALLENGER. This hull form was chosen because of the exten-sive open water tests conducted with a model of this vessel by Paulling In San Francisco Bay. Computer simulations were conducted for both regular and irregular following waves. The irregular wave case was selected to allow direct conipar-ison with Paulling's recent experimental results. The
investigations carried out with the A1ERICAN CHALLENGER were quite extensive, and represent a practical example of the type of vessel stability and capsizing analysis made possible by the time domain computer program.
Results obtained from the computer analyses of these two cases are presented in the following sections. The two-dimensional section results are described first, followed by the more extensive AMERICAN CHALLENGER results.
A. Two-Dimensional Section Results
1. General Observations
The rectangular section studied had a beam of twice the draft and other characteristics as presented in Table 1. This section was harmonically oscillated in the vertical direction, while being allowed freedom to roll. The poten-tial for roll instability lies in the periodic variation of transverse stability caused by the forced heaving motion. The transverse stability varies due to changes in the posi-tion of the center of buoyancy and of the metacenter for different degrees of section immersion. The change in meta-centric height (GM), which governs the roll restoring moment for small angles of heel, is computed for several depths
of immersion in Table 2. Inclusion of the section roll angle in this computation would yield the same general result,
Length 1.0 foot Beam 10.0 feet Depth 10.0 feet Draft 5.0 feet 4.07 feet 0.0967 feet
Roll gyradius in air 4.0 feet
Roll gyradius in water 4.58 feet
Pitch gyradius 100,000.0 feet
Heave damping term 0.56 x mass
Pitch damping term 0.00 x mass
Roll damping term 3.80 x mass
Heave added mass term 1000.00 x mass
Roll added mass moment of inertia term . . 5.00 x mass
B
A
C---10.0'
Two-Dimensional Section Section Immersed to Waterline A.
= 4.000 feet 3M = I/A = 1.041 feet KM = 5.041 feet = 4.000 feet
GM 1.041 feet,
Section Immersed to Waterline B.
= 2.500 feet 3M = I/A = 1.667 feet KM = 4.167 feet KG = 4.000 feet
= 0.167 feet.
Section Immersed to Waterline C.
= 1.000 feet 3M = I/A = 4.167 feet KM = 5.167 feet = 4.000 feet
= 1.167 feet.
where Height of center of buoyancy above keel
3M Distance from center of buoyancy to metacenter KG = Height of center of gravity above keel
GM = Transverse metaoentrio height
I = Transverse moment of inertia of waterline
same frequency as the vertical oscillation and an amplitude variation determined by the section shape and the vertical
oscillation amplitude.
This oscillation procedure simulated the case of an experiment conducted in calm water by force-heaving a body which' is free in the roll mode to respond according to the equation of motion for roll given earlier. In other words, the example studied was similar to the model experiments carried out with a MARINER hull form by Paulling and Rosen-berg (1959). The only difference lies in the simplified section shape used for the computer simulation. The object of these tests was to demonstrate the same type of roll in-stability that occurred in the experiments.
The basic results of the two-dimensional section runs, excluding preliminary tests, are given in tabular form in Appendix E.6 and are shown graphically in Figure 11. The figure clearly shows a pattern of results similar to that shown in Figure 6 for the experimental results of Paulling and Rosenberg (1959). Although the amplitude-frequency plane has not been rigorously divided into stable and unstable
regions, it is clear that the same upward and outward extend-tug unstable region is present here. The region shown cor-responds to the first roll resonance region, since the natural roll period was about six seconds; a stability variation
corres-0
T\f0- PMEU6I0KJAL
5.CT10PI, ZERO 6PP.
0.0967 FEET.
VA.I0U
IIJrflAL
IZOLL VLOCT5 AMP
IJITIAL PHM. A)JGLE3.
Figure 11. Unstable Roll
Region
for a Two-DimensionalSection
In Forced VerticalOscillation.
U&ITASLE poim POWJT
0 STABLE.
\
\
o \A
AA AA
AAc
,
c0
I.bo 1.90 2.00 2.10 2.20220
2.O
2. 0
2. Sensitivity Analyses
most runs were directed toward defining the first resonance region, capsizing also occurred in the second resonance region, with the period of stability variation equal to the natural roll period.
A typical computer print-out for the two-dimensional
section runs is given in AppendixE.5. In general, the results
fell into one of three categories: (1) the initial roll imposed on the section decayed with time; (2) the initial roll built up to some steady-state value which was maintained indefinitely; or
(3)
the initial roll built up over a period of from three to twelve cycles to the point where a capsize(90 degree heel) occurred. The result obtained was dependent upon the oscillation amplitude and frequency present, causing
either a stable or an unstable roll condition. The
synchron-ism of roll with the heave oscillations was obvious, with the section rolling heavily when the stability was low and.
snapping back upright when the restoring moment was high. In some cases, the initial roll disturbance actually subsided for a time until the roll motion reached the proper phasing with the heave oscillation, at which time the roll built
up again to finally capsize.
by the time required until capsizing occurred, was analyzed as a function of oscillation frequency and amplitude
separ-ately. In Figure 12, for a constant oscillation amplitude of 3.0 feet, the section clearly has the greatest amount of instability in the vicinity of an ostillat ion frequency of 2.24 radians per second. For frequencies outside the range of 1.92 to 2.37 radians per second, capsizing did not occur with a
3.0
foot amplitude. Of course, as one moves to lower frequencies, the second resonance region will eventually be reached. But each resonance region itself is clearly defined.Choosing a frequency of 2.24 radians per second, Figure
13
shows time to capsize as a function of the oscillation amplitude. As would be expected, a certain threshold ampli-tude is required to cause capsizing, above which the capsizing tendency increases with amplitude.For a given oscillation amplitude and frequency, the sensitivity of capsizing to initial conditions was
investi-gated. First the initial roll velocity was varied, with the results shown in Figure 14. These results indicate that, with regular heave oscillations, capsizing will occur
even-tually with even the smallest initial roll velocity, provided that the proper amplitude and frequency are present. The same general result was obtained by varying the initial phase angle of the heave oscillation. The results, shown In Figure
100
90
80
0 VIi
*60
50
'10F
20 /0 0VE11CAL 06CILL4T0PJ AM?UTU
3.0 FEET,IMITIAL PHA6E
2
AQA16.
J I I I I I I I I I I I I I I I I I I I I I $ I I I I I I -I I I I I I I I I I I
4
"ii
'4 I IUi
0
I4
I
-I I0
12
I I I I I I II.
S I I I I I I I I I I I I I I I I I $ I I I I I I I1.80 190 a.00 2.10 2.20
2.O
2.qo 2.50OCILLAT0PJ
QU3UC',
t6/6CO1u0
- Figure 12. Time to Capsize as a Function of Vertical
50
10
30
20
Fio
0
CLLTI0
2.2'4 2AVA/6ECOkJ2
1MrnAL PH4&$ MM3LE
2 7 A/il
At'L&U6.
I
0
.0
2.0
3.0
40
5.0
VERTICAL 0CILLATI0U AMF'LITUPe, FEE.T
FIgure 13.
Time to Capsize as a Function
of Vertical
l00
90
80
7o
60
'Iic3O
20
I0
0
R0 6PW.
0.0967
FteT. O3CILtAT1Ot4
AMPurUP3.0FEET,
2.25 RAPJ5/6ECOUP, IAJrrt&L. PIW6e AJGLE 90.
0
10 15ILLITIAL
0LL vL.ocrr'i', s/.cow
Figure ll.
Time to Capsize
as a Function of Initial
Roll Velocity.
60
40
30
I0
0
TIO- PM&IUoIOUAL 5ECT0U
,
R0 PaEP.
0.0967 FaT, RJtTIL 0LL V
Crr'= i6.7 PE62EE./6EcoWP.
VERTICAL 06CILL4TI0IJ FQUJ'JCY: 2.25
4MPL(TUP. = 3.0 FET
I I I I I I
90'
18dZ70'
360
IPJITlL O5CILLATIOU PljME AMGL, V4U
Figure 15.
Time to Capsize as a
Function of' Initial
initial conditions than with others, but that the final result is the same. The zero degree phase angle refers to an initial
immersed heave position; as the section moves upward from this position, stability decreases as indicated in Table 2. This critical initial phase angle causes the section to lose roll
restoring moment at the same time as it is heeling over from
the upright position due to the initial roll velocity.
The main purpose of the test runs with the two-dimensional
section was to demonstrate the capabilities of the computer
program. The results presented in this section indicate the successful simulation of unstable roll motion caused by forced vertical oscillation. The runs carried out represent the
type of experiment carried out by Paulling and Rosenberg
(1959),
and led to similar results. The buildup of rollmotion through periodic changes in the roll restoring moment can easily be understood by this simulation in still water. Thus the fundamentals of the phenomenon under study have
been successfully demonstrated using the time domain computer
program.
B. Results for Cargo Ship in Following Waves
An extensive Investigation of the AMERICAN CHALLENGER hull form was carried out in both regular and irregular following waves. Digitized hull offsets for the AMERICAN CHALLENGER, connected with straight lines as input to the computer program, are shown in Figure 16 for the eleven
FIgure 16.
stations used. The vessel characteristics and hydrodynamic coefficients for this 521-foot freighter are given in Table
3;
estimated values for the hydrodynamic terms are presented in Appendix C. Although not particularly critical from the stability viewpoint, the AMERICAN CHAT.LENGER was used since it was the subject of intensive model tests in San FranciscoBay. It was hoped to relate computer results to this exper-imental data.
Altogether, well over 150 runs were made with the AMERICAN CHALLENGER, first on the IBM 1130 computer and then on the
CDC 3300 computer. Very close agreement was found between results using the two different computers, although slight differences were present due to truncation errors. A number
of capsizings were obtalned;for example, 58 out of the 128
cases investigated on the CDC 3300 computer capsized. The results were exactly repeatible using the same computer and
input data, and appeared entirely consistent.
1. Determination of Critical Roll Resonance Condition
For a vessel proceeding with a forward speed in following
waves, the mechanism of stability variation is somewhat more complex than for the forced vertical oscillation of a section in still water. For the case in waves, the frequency of
stability variation 'is equal to the wave encounter frequency since this is the frequency of alternation of wave crests and troughs along the vessel length. The encounter period
TABLE 3.
VESSEL CHARACTERISTICS FOR THE AMERICAN CHALLENGER,.
Length between perperndicuiars 521.00 feet
Maximum breadth, molded 75 00 feet
Design draft, molded 29 25 feet
Displacement at 29.25 foot draft 18,871.21 tons S.W.
at 29.25 foot draft
31 64
feetPitch gyradius 143.00 feet
Roll gyradius In air 25 50 feet
Roll gyradius in water
27 50
feetShip sieed
16 50
knotsHeave added mass term ...1.00
x massHeave damping term 0
45
x massRoll added mass moment of inertia term. . .
105.60
x massRoll linear damping term
44 40
x massand frequency for a ship in deep water may be expressed as functions of wavelength, ship heading, and ship speed by the following equations, derived from Lewis (1967):
L/(J
cosp)
v
je -
cos9)
where Te = wave encounter period, seconds = wave encounter frequency, Hz = wavelength, feet
V = ship speed, feet per second
and = ship relative heading angle; zero degrees for following seas, 180 degrees for head seas. Wave encounter periods and frequencies for a vessel at a given speed with various wavelengths and vessel headings, adapted from Kastner (1973), are shown in Figure 17 for the case of a vessel speed of 16.5 knots. This graph clearly
shows the condensation of the encounter spectrum for the quar-tering and following sea cases. It can be seen that waves
of almost any length will all encounter a vessel at the same frequency in a quartering sea situation. The seriousness
of this situation can be imagined, if this encounter frequency happens to bear a critical relationship to the vessel's
natural roll frequency.
Since we are only considering the pure following sea