University of Strathclyde
Department of Ship and Marine Technology Glasgow. July, 1982
77:127:23112 trznn:TET
LEtoretoriurn vizorSaheepshydromechanica
Archief Mekeftveg 2, 2828 CD De!fti$LOth-?8--F*: ois
-181X$ ON THE PROPEFiTIES OFVERTTCAL WATER OSCULATION IN A MOONPCOL
by
B. S. Lee, B.S.E., M.Sc
A thesis submitted for the degree of Doctor of Philosophy
IP
SUMMARY
This project is concerned with the water oscillation in a Moonpool which is employed for a range of offshore support taskt. The problem is am important one, as not enough attention has been given to
its
optitUt design despite its Wide popUlarity among the offshore
support system operators. The fundamental aspects of the
ahipmoonpool .system is investigated and the emphasis
is
placed on the practical applicationof
the results. The research is performedin two
parts.
In the
first part the moonpool Is idenlised to a two-dimensional system so that itbecomes a narrow duct formed betWeen two
infinitely
long Obstacles in
the wave field separated by a Small distance.. Initially the barriers are fixed in space with a train of simple harmonic waves incident upon them in the beaMwise direction. The problem is treated as a singular perturbation problem and is solved by the Method of matChed asymptotic expansions. The reaponte amplitude operator of the water column oscillation is cdtputed and the results are compared to the experimental results. The barriers are then allowed to, heave only and the fixed problem solution isextended ta incorporate the heave motion. The results
are also
compared to experimental data.
This
study enables a key parametergoverning the behaviour of water oscillation in a two-dimensional
moonpool to be identified.
The second
part deals
with a three-dimensional coupled system of
ship and moonpool. The ship is allowed to heave only and a train of simple harmonic waves is incident along the length of the Ship. The coupledproblem
is formulated by Using a.two degrees of. freedomsystem and the
hydrodynamic coefficients are obtained through experimental approach.A practical design calculation procedure is put forward and its
ii
ACKNOWLELGEMENTS
The project was ,initiated
and
supervised. by Professor C.KUO ofStrathclyde University and.' an indebted
to him for
his
patient
gUidance and constantencourage bent
during the research.The help and advice of Dr. J.
Martin.
Of 'Edinburgh Univeraity, Mr + D.Vas4a,lOs and
Dr. P. Sayer ofStrathclyde
UniVersity aid PrOfessor Ursellof
Mandhester University are gratefUlly adknioWledged.The computer program used for the solution of the various bounaarY
value
probTems discussedin
this thesis is a modified versionof a
program originallyWritten
by Mr. Vassalos.T would. alSo like
to
offer my thanks to W. Wett, G. Vittand T.
Chapter
SUMMARY
ACKNINLEIXEYETIS ii
PART A ABOUT THE PROJECT AND KEY FINDINGS
INTRODUCTION
AIMS OF THE PROJECT 5
LITERATURE REVIEW
PROJECT STRATEGY 10
KEY FINDINGS AND_ :DISCUSSIONS 12
AREAS OF FUTURE _RESEARCH 14
MAIN CONCLUSIONS 16
PART B ON THE PROPERTIES OF WATER OSCILLATTON
IN '1W0-DIMENSIONAL DUCTS -18
INTRODUCTION 19
BASIC APPROACH TO THE PROBLEM 21
WATER COLUMN OSCILLATION BETWEEN TWO FIXED
BARRIERS. WITH THICKNESS 24
3.1 Fbrmulation of
the
Problem 24I.
iv
3.3
Solution of the Inner Problem26
3.4
Solution by Matching 283.5
Estimation of Water Column Response29
3.6
Experimental Verification31
WATER COLUMN OSCILLATION BEWEEN TWO HEAVING .BARRIERS
WITH THICKNESS
34
4.1
Formulation of the Problem .34
4.2
The Solution of the Inner Regian-35
4.3 The
Solution of the Outer Problem36
4.4
Matching39
4.5
Heave Problem 404.6
Estimation of Water Column Response42
4.7
Experimental Verification44
APPLICATION OF THE FINDINGS
45
rucussIoNs
48
CONCLUSIONS
51
LIST OF SYMBOLS FOR PART B 75*
PART C ON THE THREE-DIMENSIONAL MCONPOOL PROBLEM
77
INTRODUCTION
78
ON THE PROPERTIES OF WATER OSCILLATION IN A FIXED TUBE
79
2.1 System Representation with a Mechanical
Oscillator Model
79
2.2
Description of Experiment 8385
3.1 Introduction 85
3.2 Formulation of the Problem 86
3.3 SOlution Of the
Coupled Equations
88 StrateV for Experimental Studies 913.5
Description of Experiments '923.6
Methods of Analysis95
'3.7
Experimental Results Obtained97
3.8
Interpretation of the Experimental .Results and Comnents99
PRACTICAL DESIGN CALCULATION PROCEDURE AND
EXAMPLE COMPUTATIONS 102
4.1 Introduction 102
4.2 Design
Computational
Procedure 102 4.3 Worked Examples and. Comparison with Experiments 104.DISCUSSIONS 107
CONCLUSIONS 110
LIST OF SYMBOLS FOR PART C
148
REFERENCES
150
APPENDIX I
THE SOLUTION OF FIXED INNER.PROBLEM.BY
SCHWARZ-CHRISTOFFELTRANSPORMATION 154
APPENDIX II
FORMULATION OF BOUNDARY VALUE PROBLEM FOR
Vi
APPENDIX III
DESCRIPTIONS OF EXPERIMENTAL FACILITIES USED DI THE STUDY
APPENDIX IV
ANALYSIS OF FEM OSCILLATION 9:Esir RESULTS
APPENDDC
RESULTS OF FORCED OSCILLATION TESTS ON
WATER COLUMN IN :VERTICAL TUBES
APPENDIX VI
CCMPUTATION OF HEAVE EXC ITATION PUCE AND WATER COLUMN EXCITATION FORCE
169
17
INTRODUCTTON
With the :rapid expansion of Man's exploitation of ocean resources
in
the last two decades, the
demand for
the 'effective meanS oflaunching and retrieving equipment, personnel
and
materialsinto/fram the
sea has
increased tremendously. Despite the huge inVestment in developing more COSt effective and reliable sUpportsystems, at leatt one hurdle remains unresolved in launch/retrieval:.
activities,' i.e. the hatard of air/sea interface. Although various ideas of alleviating this problem have been put forward and in spite of a few claims as to the contrary, the ability to pass through t*6
interface succesSfUily
is
still at the .mercy of weather donditiOirsliThis difficulty is
expected to become greater as
the offshore -activities advance into more hazardous environment.The major hazard Of the air/Sea
interface in launoh/retnieVal operations lies in the forces exerted by the environmental element8 and experienced by, the objects launched/recovered at they pass frOMone medium into another. These forces can
be broadly divided into
two dompOnents: vertical
and
horizontal. The vertical force it mainlydue
to the relative vertical motion between the objectsand
the water
and can lead to a dangeroub situation, particularly wneh cotpled with horizontal forces OfWhich the
likely causesare
current, wind and 'theorbital motion
or
the water particles induCedby waves.
With this
problem in mind
several Methods of launching Andretrieving subsea Units, especially manned and =conned Submersibles
and diVing bells, have been developed and Some have been put
to
practical use with Varying degrees of success. One of these SysteMs Which has Shown considerable merits and consequently won reasonable
popularity
is
the mompool system.A moonpool is essentially a vertical well of various cross sectional shapes usUally situated at the centre Of floatation
or
a ship
with
its lower end open to the sea. At presentonly
relatively small arid'compact units, such as diving bells, remote control vehicles (RCV's) and one man atmospheric diving suits, are launched and recovered
through the moonpool, but there is a sign that the system will be widely used to handle manned submersibles and heavy equipment as
well in the future.
The relative advantages of using the moonpool system for these
purposes are threefOlds:
it provides good protection from the
horizontal forceelements; .
by positioning the,moonpool at the centre of floatation of the ship at the normal working draught, the adverse effects of the ship's angular motions can be. minimised;_
the Characteristics of the water oscillation ,inside the
moonpool is such that the higher frequency components of the wave are 'filtered out.'
For the
summary of various
diving support systems and thediscussions of their relative advantages and drawbacks, see Kuo [1].
Unfortunately, however, launch/retrieval operations are often hampered by 'unpredictable' large amplitude oscillations
of the
water column inside the moonpool. One other
aspect of the
'unpredictability' of the moonpool behaviour can be illustrated by a-very puzzled diving support operator who observed the prevalent wave period of six seconds and the resultant
water oscillation in a
moonpool which appeared to be of .about.ten seconds period. This puzzlement is due to the
lack of understanding of the
basic properties of the moonpool water column oscillation. The consequence of this present ignorance is that moonpools are usually designed simply as vertical wells of a size suitable for the subsea units tobe launched/retrieved. Various damping devices are added to them almost as an afterthought in an attempt to decrease the water column
oscillation.
4
The vertical water oscillation in the moonpool is important only
in
the context- of the forces imparted by the oscillation on the. ObjectS
to be launched or retrieved. Once the characteristics of the
water-oaciilation are
known,
however, these forces am be :calculated by using e.g. Mbrrizonis equation (see Madsen Fbr.this reason the research Work describedin this
thesis- is devoted to -the Investigation of the Vertical water oscillation problem alone.The behaviour of amconpool.is of a very oamplicated. nature. Apart
from the size and geometrical shape of the
moonpoal,
there are thelmillmoonpoO1
interaction and the effects Of various damping devicesto consider.. It is well.-known, however, that
the more
parameterthere are, - the more complex the problem becomes. It is, therefore,
,,necessary to simplify' the probleft
and
.start with as few parametersas possible. The Choice of these parameters is a difficult one
due
to lack of reliable data and for this reason a theoretical approach to the problem using
a
simplified geometry was adapted as a startingpoint.
In this approach the main features of the
moonpool
are retained but two-dimensionalisedin order
to accommodate the linearizedtheoretical hydrodynamics more easily.
Damping
devices are ignbred and main emphasis is placed on determining the characteristia.behaviour of such a system. This part of the work is described
in
Part B of the thesis.
Useful though this exercise is; it only provides qualitative
information Which could not be readily used for design purposes. For
this reason the problem is tackled once more using an entirelly
different approach with particular emphasis on practical_
application. This
entails
formulating the problem in terms of4
two-degrees of freedam-imedhanical, oscillator system.
Again
theeffects of damping devices are ignored and stoothwalled circular
cylinders are taken to represent the moonpool. This ,apprbach
is
discussed in Part C.2. AS OF THE PROJECT
The main .aim of the project. is to develop a. mathematical model describing the shiloHmoonpool coupled system, the ktOwledge of Which wiil..provide the basic, understanding of the
behaviour
of the watercolumn .oscillation in a moonpoOl and
an effective meansof
predicting
the moonp001 properties atthe
design stage. Morespecifically the aims are;
(i) To examine the
method of
solution tothe problem using
theoretical hydrodynamics and to verify the theory by
experiments;
1. To investigate
the
effects of ship heave motion on
the
..-behaviour of vertical water oscillation in.
the moonp001;
To establish a practical
method
of
evaluating the3. LrEERATURE REVIEW
Although the mconpool system has been in practical applicatidn forra
number of years,
it is 'onlyduring
the last 5 years that tlyiecharacteriptics.
of moonpoOls
have been studied tseribusIV. Consequently there hat been Very little work attempting todescribe. I
the vertical motion. Of the water column
in
the moonpool, t4Pinformation of Which is 'essential
for assessing the fore
experienced by subsea units laundhea/recovered throUgh the moonpoOl air/Seaintetface..-.
The papers Which have been-published to date on the subject are reviewed below.FUkuda [2] investigated the effects of water column oscillation
in a
-:...moonpool on Ship Motions with particular emphasis on the increase
of
ahip's:fOrward motion resistance. It was shown that,' When a model with a moonpool was towed in calm water, the resistance increased by up to 7% compared to a model without one. The work has provided good
iiisi,gbt into the importance of the water oscillation in Ship motion
study, but the behaviour of the water column was
studied with a
number of uniform forward velocities of the Ship and not in the wave
field. Therefore', there it no direct application to the present
problem, except the fact that an empirical formula for the Water
column added Mass was derived through a series of transient tests.-This formula was found fairly consistent with the result of ttie
present work described in Part C.
Madsen [31carried out a research specifically directed at modelling the water column motion in the mconpool mathematically so that .the
forces experienced by the diving bells passing through the- moonpoOl.
air/tea interface
could be
predicted. The equation of motion was derived by using the principle of equilibrium of forces. The addedmast of the water column was taken from Fukudes [2]
suggestion While thedamping
coefficient was determined by assuming a flOWpattern which- is not fully justifiable. This work Was the first attempt to give a full description to the water motion including
the
I
effect of
a
diving bell blocking the flow path. However, there Arethe following weak features:
Froude-Kryloff force alone is considered as external excitation force and other factors,
such as
wave diffraction, are not taken into account.The numerical values
of the water column hydrodynamic
coefficients are selected in an incorrect
manner.
(iii) It is believed by
the present author
that the various 'damping devices'mentioned in this
paper will have asignificant effect not only on damping but on added mass as well, but this point is ignored in his formulation.
(iv) A significant weakness is that the cross-coupling effects
between
ship
and moonpool are not considered.Knott and MacKley [14] performed experimental studies an the water
oscillation' in half immersed vertical tubes fixed in _ space. Their
'main interest was to study the eddy' motion neat the lower exit of the tube die to the water oscillation. A series of. free-oscillation tests was performed and
the decaying water Oscillation was
formulated by using Bernoulli's equation. The equation thus derived
is very similar to that describing
the
motion of a single degree of freedan mechanical oscillator with non-linear damping. The decaying motion of the Water column was then simulated on a computer by this equation using various numerical Values of damping coefficient until a satisfactory matching between the prediction and experimentalobservation Was achieved. The empirical formula for damping
coefficient thus
derived
proved to be very close to the result ofpresent study. Unfortunately, however, only the isolated tubes were treated and the coupled
systen of ship-moonpool was not included
in
the study.Flower and Aljaff [15] Showed that KnOtt and MacKley's experimental data could be analysed by a method based an a first order averaging
method due
to Kryloff and HogpliUboff (see Minorsky [20]). This is a very Useful technique in analysingthe
free Oscillation test data of8
Although these
researchers
have contributed to the general. knowledgeof the moonpool
probleth
considerably, it is felt that the basic understanding of the physical phenomenon is still lacking.A good
starting point would be to investigate the problem using theoretical hydrodynamic methods
but
without invOlVing hydrodynamic coefficientsas such. It is .natUral; therefore,, to examine the theoretical
studies which are relevant to the problem of vertical water column oscillation in
a
moonpool.Newman [6] solved the problem of a closely spaced double barrier with infinitesimal thickness in waves. The flow field. was diVided into inner and Outer regions and examinedseparatay. The inner
'region covers the flow field near the lower exit of the Slit formed
1
.)between the two barriers and the channel-like flow near
the
free-surfacelmaide the slit. The flow
in
thiS regal is solved'using- Sdhwar2-Christoffel tranSformatiOn,
but the
solution itsincomplete since one constant is still undetermined. In the. outer
region,
which covers the flow field fartemoved from the slit, the two barriers are assumed to be collapsed into one. The flow into and but of the lower entrance is represented by a source sing0Iarity of an Undetermined strength. The velocity potential in this region is obtained by summing incident wave potential, diffraction potential and the potential due to the source. The solution of the two regions arethen matched to
each other in the overlap domain to give the expressions for two unknown constants. This completes tha solutionfor the singular perturbation problem. and the water column
oscillation in the
slit can be estimated from linearised free-surface condition.. NUmerical studies based on this solution have shown that the characteristic features of thewater
column oscillation are very simil.Arto those of the heave motion response of a free floating thin solid Column in waves.Another
importantconclusion of this work is that the most influential parameter
governing the behaviour of watwer oscillation is the ratio
of
barrier Separation distance to the depth of immersion.
-tUbe. In the absence of prOper. evaluation of damping. terms the result
is
removed from the reality, but it ismeaningful
In a sense that this approach gives the maximum rate of energy -extraction of this particular device, and this was shown to be 0.5.The basic geometrical assumptions of these authors
ate
that thebarrier thickness is.infinitesiTAT and that the gap between the two barriers is much Smaller than the depth of immersion. In terms of
shipmoonpoOl
system the barriers correspond to theship hull
and the gap between them to the Size ofmconpool,
e.g. its diameter. The latter assumption may be valid for a large number Of cases, but the .zero thiakness.assumptign is not, since the ratio of ship breadth to , moonpool diameter isusually
larger than 5. Their. work has provided'Valuable insight into this complex problem And it is now &Own in
this thesis how the Method can be Modified to Suit. the present
problem of double
barrier with a
ship-like section,which is a
similar buta
more generalised problem of the double barrier withzero thickenss. The results obtained by Newman are particularly
In View of the critical review, it
is
clear that the ship-moonpoOl probleM'willhave
to be tackled in two ways as follows:Theoretical hydrodynamic method can be used to inhance the fUndamental understanding of the system through studying a two-dimensionalised model. The technique developed by
Newman [6] is followed closely, but the barriers are allowed
to have an arbitrary Ship-like
section. The barriers are assumed to be fixed in space initially and the solution is extended to incorporate the heave motion of thebarriei's at a later stage. This approach is described in
Part EL
Since the assumptions inVolVed
in the theoretical
analysiswould be
too
restrictivefor the
results to
be 1284b1directly for design Calculation purposes, a more practical
approach is required. This is achieved by formulating
the
Shipmoonpool system as
a
coupled mechanical oscillatorwith the
help
of insight gained during the theoreticalstudy. The
numerical values of hydtodytamic coefficients Including coupling terms are then obtained by experimental methods of both free- and forced-oscillation tests. The atOof providing a guide for designers
is achieved bysuggesting a design caldblation procedure. This part of the
.Study is presented in Part. C.
Finally, the whole project strategy is summarised in Fig.A.1. The, items inside double rectangles are
the work carried out in
thepresent study. The investigation into the fundamental aspects of the Problem is
described on the
left, While the approach aimed at practical application is shown on themelt.
It is worth noting that the theoretical line of research concentrates On the closely Spaced double barrier problem and not the widely spaced One. This point is discussed fUrther in Chapter A.6. The insight gained through the10
work
on double. barrier with shiplike sectionis
then utilised in fotmUlating the problem as a coupled mechanical oscillator. After solving the coupled equations with the hydrodynamic coefficients obtained, by experimental methods, the practical approach results in proposing a design calculation procedure.12
.5, KETFINDINGS AND BRIEF DISCUSSIONS.
Following the Strategy shown in the previous Chapter, the properties Of Vertical Water oscillation in a moonpool were investigated in two ways. Here the main features and the key findings of the research
work described in Parts B and 0 are highlighted and a few salient
points are briefly discussed.
In the
theoretical study the work by Newman[6]
with two closelyspaced and infinitely thin barriers was generalised to
deal With
barriers with arbitrary thickness,
Although in this
stUdYtie
. barriers were assumed to be symmetrical About the Vertical axis
and
have
a
Ship-like section.The result from the fixed barrier case
showed a similar
trend to theI
-thin barrier prObleM, but distinctly diffetent.in details. The two problemS use different transformation and the flow lines indicate
that
they are dissimilar physical phenomena. The thickness effecti,therefore, cannot be ignored.
The numerical studies
based on the
theoretical solution have produced Aset of graphs with water
column response amplitude operator plotted against wave number. The response characteristicsare
seen to be similar in many ways to those ofa
damped Medhanicai oscillator. In conjunction with experiMental observation,this
can
be viewed as a theoretical
badk. Up to the current practiceof
treating the Vertical water oscillation in a tube as aspring pasS
system. Another contribution of the theoretical exercise is that it clearly shows the composition of the excitation forces on the watelrcolumn. This knowledge was made use of in the study presented
in
Part C.
However, the theory has its
restrictions mainly due to
the'assumptions usually Associated with linearized hydrodynamics and ttl,
geometrical assumption of small duCt width to depth of .iMtersia4,
because of its two-dimensionality, but it can be a .starting point to tackle the three-dimensional problet or it Can be Used to check the viability of other theoretical methods Which
May.
be developed tocope
with
the three-dimensionality.Following
the line of approach aiMed at providing a practical design tool, the ,problem of shipmoonpool was reformUlated.aS.a coupledMechanical oscillator in Part C.. Four cross coupling coefficients. were introduced to link the ship's heave motion to the water column oseillatidn And vice versa. The numerical values of these. cOUplihg
coefficients as wU as
the hydrodynaMic coefficientsof the
individual
single degree of
freedom
systems were obtaned byexperimental Methods._ _Eased on this formulation, a .design calculation procedure was put forward and a-set of worked examples
even.
Some of
the experimental results were considered to be unreliable due to inadequate experimental facilities, but the values obtainedwere found
to be fairly consistent With the.findings reported
byother researchers. It is encouraging, but more data are required tO
firmly
establish these results.The calculation procedure suggested.
in
this theSia is ampleand.
eaaly adaptable. Once the hydrodynamic coefficients are known, the Simultaneous equations can. be solved without
any
difficulties, and the while process can then be computer programed.The forMUlation
however,
mayrequire further
refinementdepending
on how well it agrees with further experiments. The .major
contributions of this part Of research work, therefore, cart be
Summarised as having
suggested
a simple but versatile approach to the compleX. engineering problem and having demonstrated theuse of
experimental techniques to obtainhydrodynamic coefficients of the ship-mconpool coupled system.AREAS OF FUTURE
RESEARCH-The areas of fututre research
in
this field of practical importance are discussed below.(a) Solution for Wide Duct Problem
The present theoretical analysis described in Part B oniy
deals with very narrow ducts, but there are Many cases Where
this assumption is not Valid. More importantly it is
-difficult at present to identify the region of width to depth
ratio Where it becomes unacceptable. It: is, therefore, -necessary to tackle the problem from
the. other
eXtremity,i.e. a wide duct problem.
The
result can be brought side by sidewith the
present method and compared to each other neat1
the
intermediate region. This work would complete thepicture,
So
to speak.-. () Influence Of MbOnpool On Horizontal ship motioh
It is known that the water in the tOonpoOl dOes oscillate or sloSh in all
SiX
directions, i.e. pp and down, bad k and forth and side to side. The initial work by FUkuda [2] and the practical experienceof diving
Support vessel operatorSI tUggeSt that the Water oscillation Mentioned above increases
the
resistance of the Ship's fotWard MOVement and has adverse:effects on coUrse keeping abiltiy. This is an important poin! to consider when designing a vessel with a toonp601,-but
the
Work in this
field is far from complete yet, as no attempt has been made to take into account the effects of surfacewaves..
(C)
COnfirmation of Hydrodynamid Coefficients14
The hydrodynaMic coefficients including the coUP11;4g
4ria not totally conclusive, mainly due to lack of data. More
eXperitental and theoretical
studies
are required todetermine these all iMpottant quantities.
Inclusion of Other Modes of Motion
Although heave motion of. the ship is believed to be. the most important mode influencing the vertical water oscillation
in
the mconpool, the effect of other mode8 of motion need be
investigated at least to confirm that
4.t is
so.Effects of Various Damping Devices and MoOnpool Geometry
Many
damping devices are being eMployectin
an :attempt_ to reducethe
water oscillation in the moonpool at present, e.g.injection of high pressure air under
the water surface, Installing various forms of baffles,- etc. Their effectiveness, however,is
not at
all
clear and the many contradicting claiMs confuse the issue considerably. The saMecan be
said abOut the geometrical arrangements of themconpool itself. Some claim the best configuration
tO
be a .trapezdidal. one, while others favour a rectangular,triangular or circular moOnpool.
As
far as
the vertical water oscillation is concerned, theeffectiveness of these devices and the
geometrical
configurations
can be
investigated by .the experimentalmethods
described In
Part
C ofthis thesis.
When the
hydrodynam3.c coefficients are obtained through systematic
16
7. MAIN CONCLUSIONS
The main conclusions of this research work are as follows:
To understand the
fundamentals
of the water column oscillation inside a moonpool theoretically, it is esential to idealise it as a two-dimensional system represented by two surface piercing parallel barriers in waves and the results of this treatment showed that the most important parameter is the ratio of thegap
between the two barriers to the depth of immersion..By idealising the moonpool as a fixed vertical. tube, it. is possible to employ the knowledge of the spring-mass system to
obtain approximate response characteristics of the vertical oscillation of the water column.
The coupled system of amoonpool on a vessel can be treated
as a two degrees of freedom system involving the water column oscillation and the vessel heave motion.
The hydrodynamic coefficients contained
in the
coupledequations have been obtained by experimental methods, but additional experimental data are required for design purposes.
Any effective procedure for designing moonpools an vessels
must incorporate factors concerning the Characteristics of water column oscillation in the moonpool and vessel motions
FUNDAMENTAL ASPECTS
CLOSELY SPACED DOUBLE BARRIER WITH ZERO THICKNESS
No.
MATCHED ASYMPTOTIC EXPANSIONS
CLOSELY SPACED FIXED BARRIERS WITH
SHIP-UKE SECTION
THEORY; EXPOIMENT
1
HEAVING BARRIERS WITH SHIP-UKE SECTION
-THEORY, EXPERIMENT
17
FIGA.1 SUMMARY OF APPROACH
EVERIMENT ESTIMATION OF HYDRODYNAMIC COUTICIENTS: MOONPOOL; SHIP; COUPLING TERMS
APPLICATION
SOWTION OF COUPLED EQUATIONS
PROPOSING
DESIGN CALCULATION PROCEDURE SINGLE DEGREE OF FREEDOM
FORMULATION
MOONPOOL : NON-LINEAR OSCILLATOR SHIP: UNEAR OSCILLATOR
TWO DEGREES OF FREEDOM
-FORMULATION
PART B ON 'ME PROPERTIES OF WRIER OSCILLATION
IN TWO-DIMENSIONAL DUCTS
1. INTRODUCTION
It is now known
that
the water Oscillationin
a moonpool has a peak response phenomenon Which suggests its semblance tOa siMple
mechanical oscillator system. Fat' such a system it is essentialto
know the
peak response frequency either to detune it from the operational environment asin
the case of the mbonpool, or to tune Into it asin the
case of a certain type of wave energy device. Therefore, it is highly desirable to predict thebehaviour of the
water colt= oscillation
Ina
moonpool at the design stage. This requirement for predictionhas
led a few authors (e.g. Madaen[3]
and AUghes [T]) to represent the systemin
workable fOrtulae. It isfelt, however, that the
approaches used by
theft are ratherover-intuitive and the fUndamental understanding of the systemis still lacking.
Accordingly the linear potential theory is utilised to study the
basic behaviour pattern
of
the ship-moOnpoOl system.In
1974Newman
[8]
solved the problem of water oscillation between two
closely
spaced barriersin waves using
the method of matchedAsymptotic expansions si*elar to Tuck [5], by treating the problem
as a singular perturbation problem. Evan
[8]
extended the solutionto
the
three-dimensional case, i.e. the problem of water oscillationin
a vertical tube using the 'pipe-end correction factorgiven
by Levine and Sdhwinger [9].References
[6]
and [8] use the basic assumptions that the thickness of the barriers is infinitesiraland
thegap
between them is emailcompared to the depth of lmmePsion. The latter is acceptable for a large number of cases
but the
former is not for the mcohpoolproblem. Accordingly
an
approach simi_lar to these authors is adoptedto investigate the water oscillation between two thick barriers of ship-like section. The barriers are assumed to be fixed in space initially (fixed case) and at
a later
stagethe
heave motion.restraint is released so that the body, formed by a rigid connection of the two barriers undergoes pure heavng motion (heaving case).
20
Expansion to three-dimensional model using this approadh
in
a,
similar manner as
[8], however, is not Validin
this case since there will be considerable croaa-flOwt at the lower exit of thetube.
The basic
approach and the construction of a tWo=dimensiOrial Modelis discussed in Chapter 2. After Saving the problem of the fixed barriers case
in
Chapter3,
the solution is developeOrfUrther to incorporate the heave motion of the body in Chapter 4. PVadtideilapplications of the theory thus developed are discussed
in chaptee.
5.
2. BASIC APPROACH TO THE PROBLEM
The actual moonpool problem can be illustrated by the diagram shown in Fig.B.1(a),
and as
it stands it is a daunting task to tackle theoretically. Some simplifications and idealisations, therefore, are necessary and the steps involved are shown in Fig.B.1.The. problema considered in this Part are represented by Fig.B.1(.d) and (c). As mentioned above, Newman
[6]
solved theproblem
of twobarriers shown in Fig.B.2. This
can
be either a two-ditensionalised representation ofa
vertidal tube or a model simplified one stepfurther
frdit Fig.8.1(d) by removingthe
thickness ofthe
barriers.We Can therefore regard the problem in Fig.B.2 as
a
special case ofNoting its similarity to -the present. problem, we adopt a
SlmilAr
approach to that of[6]
and [8] and the method of tackling the fiXed barriers case.(Fig.B.1(d)) is summarised below:(i) The flow field i8 divided into two regions, i.e. the inner region comprising
or
the local flow near the lower exit Of the moonpool and the Channel-like flow near the free surface within the moOnpool; and the outer region coveringthe wave
field exterior to the barrierseii) The outer problem i8 Staved With
the
assumption
that the two barriers areeSsentiAlly
.collapsed into oneand
consequentlythe outer
contour of the barriers represents a solid boundary Of the Ship Section. The MBSS flow into and outof the moonP061
i8
then represented by A sourcesingularity placed Where the exit of the MoOnpdol
S.
Then the velocity potentialin this
regionis
the linearaddition of the incident wave,
potential,
diffraction potentialand
the potential due to the SoUrce. The litret of this compound velocity potential near the exit of the MoonpOol is taken as the inner expansion of the outersolution.
(iii)
The inner problem is solved with the chwarz,JChriatoffel
transformation and the asymptotic expression of this
Implicit solution for the region near the lower exit of the duct is taken as the outer eXpansiOn of the
inner
solution..(iv) The inner expansion Of the outer solution is then matched - to the outer expansion of the inner solution.
(V) With the solution obtained in (iv) the
behaviour of the
water column oscillation
in
the ductis
evaluated in the form of. the response amplitude operators, which are defined at the ratios of the oscillation amplitude to the incident wave amplitude.1
The heaving barrier problem is shown in Fig.B.1(c), and essentially it is a similar probleth as the fixed barrier case,
but it
differsI
from the
latter in that the heaving Motion Of the ship is expected to be influenced. by the source singularity placed at the exit of heduct, and that the water oscillation is affected by the heave
potential.
Therefore, the approach employed for the fixed case ismodified to stilt the present problem as follows:
IntrodUce an
unknown
complex heave amplitude todeal with
the influence of the heavepotential.
The inner soldtion- will now be
the one
obtainedin the
fixed da8a plus another term to satisfy the body boundary cOndition dUe to the heave motion of the hull.
The
velocity potential of the outer solution is the linear summation of incident waVe potential, diffradtion potential,heave potential
And source potential.22
solutions In the
overlap domain and
the heave motion equation of the body.With these modifications the .behaviour of the water column oscillation relative to a fixed datum and the
heaving
barriers can be estimated. We begin, however, by tackling the fixed barrier24
WATER COLUMN OSCILLATION BETWEEN TWO FIXED BARRIERS WITH THICKNESS
3.1 FOrmulation of the Problem
The
geometry
of the two-dimensional moonpool model isidentical
t0 that of an infinitelylong
ductfOrmed by
two barriers Ofa
Ship-like tectiOn. We, therefore, represent the prOblem as a doub. 0 barrier problet.A cartesian coordinate system (x,y) is Chosen with the
origin
atthe
undisturbed free
surface On the
centreline ofthe
dUct With y JoOsitive verticallydoWnwarda.
Tha barriers are.hIed in spaceand
the width of the duct is 2b, While the depth of immersion is a
(See
AlUA3.3)..
The
depth
Of Water is infinite and plane progressive waves of frequency (P/21tare
normallyincident
fPam x = 140 upon the
two-dimensional Ship hull. The Usual astumptiOns of lineariaed ideal_ fluid and ittotational flow are made and the time,-dependence of tbe velocity potential is expressed by
(xiY it)
=- eec.4x,y)
eg'xt]
The spatial-.dependence 4'(x y) satisfies the following
set Of
conditions.:
v*=
0 for y> 0 external to the body-Winds*.K+
+a+/a7
=-0 on y = O.alb/CRI = 0 an the by boundary.
- 0 as y
where i =
K = ci/g
g = gravitational acceleration R = reflectidn coefficient T = transmission coefficient
normal vector positive outward from the body. A solution
or
this boundary Value problem issought
with the
assumptions
on the
geometry Of the problem, i.e.b/a = e << 1 and Ka = 0(1),
Which means
that the width of the duct is small relative to the draught of the ship a and wave length .= 27r/K.The flow field is now divided into two regions.
The
outer region for (Kx,Ky) = 0(1) and the inner region for x/b = 0(1), 0 < y "<d; a areconsidered separately.
3.2 Solution of
the outer
ProblemThe outer problem can be best illustrated by an analogy to a giant observer far away from the eXit of the duct (see
Tack.
[5]). Fbr himthe two barriers will look effectively collapsed into one and will then represent a Ship section with
a
solid boundary.. Theoutgoing
ftow
of the duct is represented by a sourcesingularity
placedWhere
theexit
ofthe
duct is Situated as shown in Fig.B.4. This treatment is somewhat similar to the approach adopted byTaylor [10].
The .body is fiXed
in
space, and the time-independent part of the Velocity potential can be expressed bycp(x,y) = t+430+ 2mG (5. 1)
where4k=
incident wave potential!there
26
cPt, = diffraction potential
m = sOurce
strength
to be determined later G = Green's.fUnction.Let
0, =
ctti+4s,, and r =Cia +
(y=a)43ra Green'sfunction
G(x,y;a) for deep Water is giVen by Wehausen and Laitone [111 inthe
form NA0HougsAgx , G(x,Y;a) =44t.'"°
X4+(1%:/;°: 1 1-e
, 4.2
r
Jtv-d<-k (y+°)
y+a)
-se. COSk
X ,enotes Cauchy PrinciPal value integral.
If r-r0, then X 0 and y a, and the asymptotic expression of as r becoires Ocx,y ; a 11%."0 m
-3--
"wa 0-K
KA.ie
0((r/04)
)
ak
For
any
givenwave
frequencyand body
contourregarded
as a constant and We letDo=
I'm
n-ipo 4).
The inner expansion of the outer solution,
then, is
11'11
cx y) =
D. +27c
(L)
ftze±-
A
ime2K4
0 ((
ria)11Z )via
.r-ve
This
completesthe solution
of the outer region. with the source strengthm Still. undetermined. .3.3
The Solution of the Inner Problem
The geometry of the inner region is shown in Fig.B.5(a) and, .since, the flow will be syMmetridal about the y-agiS, we need consider onlil
(32)
be
the
eight
half of the 2-plane with the coordinate system as shownin
Fig.B.5(b).In order to find the exact solution of
the flow
in this
region bounded by the solid walls of Fig.B.5(b), the technique ofSchwarz-Christoffel transformation is utilised. The flow is
described by the implicit solution of the form (see Appendix I).
z =
b[Z
cexp cow +p3 + Pf'e4 ( dm/
+fS)
-2
OcjitCexPCom+P)
+I j vai (E3. 3)where j.
rlf.
and does not interact with iCI and
p
are constants real relative tow. is the complex potential of the form (
4, +
Sufficiently far above the lower exit the asymptotic expression of
equation (B.3) is
x + gay)
i4r ci)By redefining
p
2 the term (2 2log2) can be absorbed byp
andseparating the real
and
imaginary parts,7-=
(Y-a)
C5-4)
On the free surface within the
duct
equation (B.4) should satisfy the free surface condition, i.e.+ a+,/ay
= 0 on y-=o.
Substituting (.B.4)
into (B.5)
givesp =
+
it
with d'0.
In the
overlap dathain b << r <<a,
where r =Ix + j(ay)1._,
asymptotic : expression of equation (B.3).
is
+ 2 2log2)
(B. S)
28
x
14) [exP( clw +p)+
nY2.
Hence
zvi; + 0 ((
b/r)2
This completes the solution for the inner region With cL still undetermined.
3.4 Solution 1:m Matching
Now the
inner expansion of the outer solution is matched to the outer expension of the innersolution in the overlap &main tO
obtain
the two
unknowns cl and m. The outer limit of the inner'solution is given by (3.7) and the inner limit of the outer solution by equation (B.2). The outer limit of the inner solution, then it
expressed: by
the
outer solution perturbation quantity r/a after Van Dyke [4] asa?
jaa
-ar
'"93with the order term omitted.
Equation (B.8) can now be readily matched to equation (B.2) to give
m
/
211c= 2 /d
or
= LC/
m
and CB-9.) otr, _2kcka
click
ly Dof
di?
(8. °)
From (B.6), (B.9) and (B.10) ;eaM =
[
-a
logaft
+ ;71h5 (a -1/K) +
otk
+ ie
-2"
CB. 10.With
p
from (B.6), Cl from (B.9) and M frOM (S.11) the solution isat
Using this solution, the response Characteristics of the water coll.= oscillation in the duct is examined in the next section.
3.5
Estimation of Water Column ResponseOur Main interest here is the amplitude of water column oscillation and accordingly the Phase angle part of the transfer function is Ignored. Since we are dealing with a linearised prOblem, this can best be approached by estimating the ratio of response amplitude to the incident wave amplitude at various wave, frequencies.
From the free surface condition
K
+ ativa
y = 0 at y 0Therefore
=
(x,o)Thus the amplitude of the wave is
A -/c-41;
14Vx,o)
The velocity potential near y = 0 inside the duct
is
given by
equation (B.4)
...yc -a)
bAs8Uting a flat water. surface in the duct, the amplitude of the
water column oscillation is Obtained in similar manner as
A =
14,(0,0)1
Therefore the response amplitude operator M is
4 (0,0)1
MIf the
amplitude of the incident wave potential at y = 0 is unity, i.e. cl)x,(x,0) e-iKx= 14)(0,0)
I= rett)
la%(5
We now consider eqUation (B.11) in order to evaluate
DO
213:10()+
(1I-TD)
e-Ky
e-iKxbut we are interested
in the
regionx
-0 And y -*a. Therefore.ri.oie-14c4
OD is evaluated numerically from the relationShip
_ 4 *111on
Zs
-an
- 04
where
Se denotes the body Silt:face.. As shown in Fig.B.6, this constitutes a boundary value probIet,
and
it is solved by the source distribution method closely following the work of Frank [12].A computer program
MS written
to evaluate the response amplitudeoperator of the water oscillation and
a
numerical Studywas
perforted
for
two typical section shape8 shOWn in Fig.B.7..lhe twO sections shown as Fig.B.7(b) and (c)represent the
same section shApe but of different draughts. Fbr the reasons discussed below this rectangular section withrounded bilge was
chosen for the expetitental study.30
I
irI
'I
= (8.12)
oticb 4)(b
the
form
of R.A.O. of the water oscillation plotted against Ka. For the purpose of comparison the Characteristics of the Water oscillation between two barriers of zero thickness aregiven in
Fig.B.1I.
These wereevaluated using the
solution given in
Newman [6].
Experimental Verification
The two-dimensional fixed mconpool problem has been theoretically
investigated and as -a result the vertical water .column oscillation
in Atwo-dimensional moonpool.can be predicted for a ship fixed in
space. The qualitatiVe agreement with Newman's resUlts, as can be
seen from
Figs
8.10 and B.11, gives some credibility to the present solution.It was decided, however, to perform a series of experiments aimed at achieving the following objectives:
validate the theoretical results;
find the limitations of the theoretical analysis.
(a) Description of the Model
The basic configuration of the model used in this experiment
is Shown in Fig.B.12. Due to the cross-sectional shape of the mpdel the diffraction of the incident waves was expected to be considerable. However, the
inner
region geometry of the
.theoretical ahalysis Was assumed to
have. a 110 bottom
extenoarig to a reasonable distance,
and this
is
the mainreason for Selecting such a 'body Shape. The experienceof experiments with fixed vertical tUbes has .shown that an accurately identifiable
peak
does not occur below the depth of immersionof
approximately 15 am. Therefore, it was decided that the draught of the 'Model should be at least 20 an forthis experiment.
towing
tank
/ma
themain
dimensions32
25 mL x 1.5 mW x 0.8 mD and the model is to be placed across the tank to maintain the wo-dimensionality. Thus the length of the model was made to be 1.3 m which leaves a 10 am gap each side between the tank wall and the model.
Since the result of the theoretical study suggested the ratio b/a to be the most important
parameter governing the peak
response frequency, the width of the duct was varied as shown In Table B.1. The effect of the
depth of immersion was
investigated by testing for two draughts of 20 an and 25 an.
The
immersed cross sectional shapes when normalised relative to the draughts, are identical to those given in Fig.B.7(b) and (c).(b) Instruments Used and the Experimental Procedure
Once the duct was set to a desired width and the model placed
and fixed
to two cross bars across the tank at the first draught (20 am), the simple harmonic longcrested waves of
frequency ranging from
0.4
to 1.2 HZ were generated by the wave maker.In less than one minute after
starting a wave of a given.
frequency standing waves were quickly established between the model and the wavemaker due to the large reflection of the incident wave.
For this
reason the input wave height wasmeasured as near to the hull as practicality would allow.
The
resulting water oscillation and
the
input wave profile were measured and recorded on a three channel pen recorder. The wave in the tank was allowed to die down and the water surface calm again before the next frequency wave was generated. The model was then adjusted to the second draught (25 am) and thewhole process repeated before the duct was adjusted to a new width.
For the description of towing
tank
and instrumentation system see Appendix III.(c) Results Obtained and Comparisons with the Theory
The results of the experiment are presented in Figs B.13 and
B.14. By comparing these to Figs B.9 and .B.10, the trend of
the
peak
response frequency obtained from the experiment isseen to agree with those from the theoretical analysis. The
peak
response magnification factor, however, dhows an opposite trend, i.e. the theory predicts it would decrease as b/a increases, but the experimental results show otherwise, albeit In a much smaller scale. The discrepancy is due to following factors:experimental error arising from severe distortion of
waves caused by a relatively large model Obstructing the
now field;
the theory considers
only wave
radiation losses as damping but thepeak
response magnification factor is extremely sensitive to damping value. It is worth noting that the theory predicts a virtually infinite responsefor small b/a ratios, but it is
hirely
unlikely inreality.
It is very difficult to determine in what manner these factors
would affect the result, and therefore the information concerning the magnitude of peak response should be treated with caution. Finally, we can conclude from these results that
the b/a ratio is amain parameter influencing the response Characteristics of the water column.
34
4. WATER COLUMN OSCILLATION BETWEEN TWO HEAVING BARRIERS WTEH THICKNESS
4.1 F6rmUlation of the Problem
Following the conventions Of the previous chapter, a cartesian
coordinate system is chosen with its origin at the undisturbed free surface on the centreline of the duct
with y positive vertically
downwards. The two barriers are rigidly connected to each other and are allowed to heave only. The width of the duct is 2b while the depth of immerSion is a (see Fig.B.15).
It it assumed that the
depth of water
is infinite and plane progressive waves of frequency (.0/27C are normally incident from= +co
upon the tWo-diMenSiOnal Ship hull. The usual assumptionsof
lineariSedideal
fluidand
irrotational flow are made and thetime-del6endence is assumed to be harmonic SO that, for
instance,
the Velocitypotential
is written. as) =
ReCCx,y)e'
3.
The spatial-dependence 1?(x,y) satisfies the following
set of
conditions:
174+= 0 for y > 0 externel to the body boandary.
10-1, a+/1Y =.0 on
(14) a+Ain =
-icoAdose on the body boundary.-.0 as y
(v) 40(;
y) e-KY (6-1Kx' + Rei" ), x 4.1ms.ct(x,y)
,
-ca.g = gravitational acceleration R = reflection coefficient T = transmission coefficient A = heave amplitude of the body
e
= the angle between the vertical axis and the normal outward vector of the body surface R, T and A can be complex to representphase relationships.
A solution of this
boundary value problem is
soughtunder the
further assumptions on the geometry of the problem that:b/a
= 6
<< 1 and Ka = 0(1),,which means that the width of the duct is small relative to the draught of the barriers a and the wave length ).= 2n/K.
The
flow field is now divided into two regions to be considered . separately: the outer region for (Kx,Ky) = 0(1) and the inner region for x/b = 0(1), 0 < y a.4.2 The Solution of the Inner Region
The geometry of the inner region is Shown in Fig.B.16(a) with the
cooranate system as shown in Fig03.16(b). We consider the right half of the z-plane, since the flow
is
symmetrical about y-axis.Fbr
the
problem with fixed barriers, the line y = a, x > b is part ofa
Streamline andthue-acOiy_=
0 an y = a, X > b. However, this line is no longera
streamlinein
the heaving case. Suppose the unknown complex heave amplitude of the barriers is A, then ..4) must satisfy= 4)1- iwAY,
+,satlzfies the Same
prOblem
as4, in
the fixed barrier caseThe
asymptotiC
ekpressions of 47, are found in the previous chapterand
in Appendix I:4)1 al
JE-(y-a) _
36
near y = 0 for
Ix( <
bwhere d and p are constants
r =, .
'Hence near y = 0,
for
[xl < b, we haveIC4)
34) /ay = 0 on y = O.
Substituting (B.13) into (B.14) gives
p =
- a) + icaccA/K
The outer Limit of the inner solution is
(1)
log tk- ,
-
iwAy.
4.3 The Solution of the Outer Problem
This mutt satisfy the free surface condition within the duct, i.e.
(8.15)
(5.16)
This cpmpletes the solution for the inner region with pt,
p
and
A, still Undetermined, although equation (B.15) gives one telationShip between them.ft
log for b<<r<<a, y a,In the outer region, as 6 .0, the two barriers are effectively collapsed into one and the barriers then represent a. ship section with a solid boundary. The flow into and out of the duct is then represented by a source, placed at (0,a), of strengthmt Which is to be determined. This is shown in Fig.B.17.
The Section is allowed to heave only, and the spatial dependence of the potential is expressed by
(x,Y) = +
A tit 4- m t's
(B. 17)
Where indident wave potential
46-
diffraction potentialA.= unknown heave .amplitude of the barriers $6= heave potential
for
Unit aMPlitude of heave= unknown source strength
4k= unit
source PotentialThe
incident waVe
potential is given in the formand the four potentials of the
ral.t.
of equation (B.17) Should satisfy the free Surface condition and Laplace't equation.cp., $4
and +5 should also satisfy the radiation condition. In addition to these conditionS the following bOdy bOtIndary conditions apply:(a) Diffraction Potential
(Ii)
The diffraction potential
is
the correction to the incident potential required if the ship were fixed in space and the source absent (i.e. the duct sealed). It must satisfy'341)/an =
-a4/an
on Se (8.19)where is the northal Vector positive outwards
from the
body.
Heave Potential
(+0)
The heave potential represents the flow generated by the unit amplitude heave motion of the barriers with the source absent
(i.e. the
duct
'Sealed). It Must satitfy-icoAdose on Ss
where 9 is as given in §4.1.
Source Potential
The sourdepotential represents the flow due to
a
unit sourceat'
(0,e) in thepresence Of the
ship fixed at itsmean
position. We
require
thatciog P, as
e
+
(y-a)Z 0 with y a.The potential in y> 0 of a source at (0,a) of unit strength satisfying the free strface condition throughout y- 0
Laplace'S equation and the radiation condition is
just the
well-knOwel Green's function (see Wehausen and Laitone [11]):
G(x,y;4) = log
x:4-- 1=11._2.ele"4"Y440,
'-X cy+a)
Z
Je A-K
obakx dk 38. = G +51-W7+4)
e
Kx
We writeCs. al)
and compute the correction +51
in order
to satisfy theaVan
= 0 onSs.
Hence fsomist satisfy the boundary condition
a40/..a n = 0 - a
G/a
n on Sp.c5.23)
The explicit form of (B.23) is given inAppendix II.
These
three boundary value problems, as suMmatised in Figs B.18.-20, can be solved by various numerical methods and here- we followFtank's close fit source distribution method [12].
Fbr the purpose of matching we require the leading asymptotic forms of the four potentials as r 0, and thus we obtain
(x f.-xiog( ) - dic
-+ Um (
cfxj_ + 4) + A cis, + m4,51)rwe
+. 0((r/0), as r 0
which is the inner expansion of the outer solution,.
4.4 Matching
We now Consider the .matching between the inner expansion of the
outer solution and the outer expansion of the inner solution in the overlap daMain in order to obtain two 'equations for the unknown
quantities.
In the overlap domaih b<<r<<a, the asymptotic expression for the
velocity potential in the inner region is given by equation (B.16):
log()
-Jo) Ay.
Cr
CB . 24)
dp 14-*
Introducing the outer perturbation quantity
r/a and noting
that40
r log e-Tx +
With the order term omitted.
Equation
(B.25) is now matched toequation (B.24). By matdhing
0(log r) terMs
_2 en
air
and by
matching 0(1) terms we
obtain2
az
log -6-a- -
icaAa.-ak4
= De
+14
A +ti curt 13 ia)Aa where D, ( 42z. p )We thus bye equations
(B.15),
(5.26)
and (B.27)for the
four
unknowns.
4.5
Heave ProblemOne further-eqUatiOn is required for a
closed
System and this comes':from the equation of motion for the heaving of the barriers:
(4811iMe+
AN.*
+ Fs + Fo + mF5 .
Solving this
equation
for A givesA = CI
+ Cm
where
C, = -(P1+Pp)/(46+
FM -
FiCa =
-Fs
AdMi+
FM 130)Ma = gess of the barriers
FH
,
etc. = force due toetc.
Re = heave restoration.(9. 25)
(6. 29)
(8.26)
-
ime-2"
34;110 451 (13.2-7)_
The forces
FH ,
etc.can be estimated
by using
the linearizedBernoulli's equation,
thus
F(4),t)
=r1;1(X,5t;t)ftdS6 Se aE(X/Y;y
i55s
at
_ Ase,p
where p = mass density of the fluid
5
= normal outward vector on the body surface P(x,y;t) = time-dependent pressure at (x-,Y) St = body contour.Hence the vertical component of the complex-force amplitude is
F(ca) =
f p(x
,y) cos dS= - 463 fi(x ,y)
wee dSe
.(B-31)
'where p(x,y) = spatialrdependent pressure at (x,y)
The velocity potentials are given by
Saving
the boundary value:problems of
the
previous section, and therefore equation (B.29) can be evaluated to give a relationship between A and m.We thus have four equations (B.15), (B. 26), (B.27) and (B.29) for
the four unknowns d, p, A and t. Solving
in
particular for m, wehave
(
S. 30)
m = CD0 + Di -
-a)]
x
EakiogV -(A
-
1A)c2)(iE-
a) - Da + J.J7S2±Ldk +. le-2" limr
7C A-14 0-i+0;!
0 where Di = C,lizi+pi Da = CalAmeism(a. 32-)
This completes
the
solution and We can proceed to calculate the response of the water column oscillation in the duct.42
1L6 Estimation of Water Column Response
With the complete solution
obtained
above .the behaviour ofthe.
Watercolumn oscillation,
in the form
of a response amplitude operator,iS
now inVettigated.
FrOt the viewpoint
of moonpo01 applications the vertical Water oacillatiOn relative tothe ship is of more interest
tO us, but it LS- Convenient first tO consider
the oscillation
relative to the fixed Coordinate system.
P. the
free surface conditionK
4): ÷ tiVay =
0 on y 0,We have
Ott/ay )yzio =
-K cf,L(x,0).Thus the
amplitude of the incidentwave is
Aw
=1A144x,0)1
.
The water column potential near y = 0
is given by
equation (B.13) as/c
Pdip
(Sra)
4-i4
Ay.
The amplitude of the
Watercolumn oscillation
is
Obtained
In
similar
manner asJo
,o
1 .Therefore, the response amplitude operator
for
the
absolute oscillation113(9,0.)1 1 Ick(xi0) .
MA = I(0,0)1
=
-= 17;A
ic"/Ki
(5.33)
With m and A given by (B..32) and (B.29) respectively.
Now we turn to the water oscillation relative to the
heaving
barriers. The heave amplitude A is now known and the time-dependence of the heave motion can be described in the formytt
= Ae-t
.,7he velocity potential of the water column near y p is given by
(B.13)
from which we derive the position of the free surfacein
the duct at time t to betnt
i+ A \
b ch,)
)
The Water oscillation relative to the heave motion of the barriers -is thUS
_im
yo
-e
.
The response amplitude operator of the relative water oscillation
14k, then is
m I / 41,1
A computer program was written to compute the R.A.O. of the water column oscillation based an
the
solutions Obtained above,With the
aid of the
program a numerical studywas
performed for the threecross sectional
Shapes of the
barriers as describedin h3.5 and
Shown In Fig.B.7.
The latter two sectionsare
the shape ofan
identical model used for the experimental study described in the next section with different draughts and normalised relative to the draughts... The resUlts of the numerical .study are given in Figa B.21'
44
23.
The
R.A.O.'s-of the Water oscillation relative to the fixed coordinate systel are plottedagainst
Ka and shown asgraphs (a)
andthe
same
relative to the heaving barriers are presented asgraphs (b).
FU.B.24 shows the influence of the water oscillation inthe duct on the heave motion of the barriers for the case of the
semi-circUlar cylinder.
4.7 Ekperimental Verification
(a) Description of.Experitent
The results of
the
theoretical analysis on the heaving moonpoolproblem
discussed above areseries
of experiments Wing the same model
.3.6. The mpdel, however, was allOWed to pure resulting motion was
measured With an
arrangement
of
the instrumentation systemAppendix III.
two-dimensional Checked
by
aas described
in
heave and
the
For
the
eftployed see
The model heave guide medhanism is shown in Fig.B.-25 and the
experimental setup
is
Similar to that shown in Photos B.1and
The rest of the instrumentation system is exactly,identical to that used for the experiments
with the fixed
barriers. The combinations of the parameters tested are Shown, In Table B.2.
(b) Results
Obtained and Comparisonwith
TheoryThe experimental results are summarised in Figs B.26 and B.2 .
It may be
noticed
that these graphs do not show a well-defined pattern. This was due to the severe distortion of the waves near the Model caused by theheaving
motion of a relativelylarge
modelin
a small
tank, thus introducing appreciableexperimental error.
those fram the fixed barrier experiment, we can conclude that bOth