5746-23856-1-PB

Embed Size (px)

Citation preview

  • 7/26/2019 5746-23856-1-PB

    1/14

    STATISTICALDISTRIBUTION

    OF

    HORIZONTALWA V EFORCESON

    VERTICALBREAKWATERS

    JaniceMcKENNA

    1

    an dWilliam ALLSOP

    2

    ABSTRACT

    This

    paper

    discusses

    th e

    statistical

    distribution

    of

    wave

    impact

    forces

    on

    vertical

    wall

    structures.

    tescribesewnalysis

    arried

    ut

    ased

    n

    heistribution

    ype

    o

    identify

    the

    parameter

    combinations

    that

    lead

    to

    wave

    impacts.

    Comprehensive2-dimensionalrandomwavemodeltestswerecarried

    out

    tomeasure

    wavepressures

    n

    a

    range

    of

    structure

    types.nitialnalysisof

    these

    testshowed

    thatth eWeibull

    distribution

    could

    beused

    todescribe

    non-breaking

    waveforces.

    Forsomestructuralconfigurationshowever

    th e

    wave

    forceswere

    found

    to

    give

    a

    poor

    fi t

    with

    he

    Weibull

    istribution.

    hese

    ata

    ad

    een

    xcluded

    ro m

    he

    nitial

    analysis.

    he

    newanalysisdescribedin

    thispaper

    has

    resulted

    in

    a

    revised

    parameter

    map

    to

    summarisethe

    risk

    of

    wave

    impact,

    derived

    from th e

    full

    data

    set

    and

    based

    on

    th e

    distributiontype.

    1 .

    INTRODUCTION

    Vertical

    wall

    structures

    werewidely

    used

    in

    the

    UK

    as

    seawalls

    andbreakwatersbefore

    rubble

    mound

    breakwaters

    nd

    rip-rap

    revetmentsgained

    popularity

    inth e900's.n

    Japannd

    taly,

    aissonsontinue

    o

    e

    avoured

    or

    reakwateronstruction.

    thorough

    understanding

    of

    the

    relationship

    etween

    wave

    forces

    n

    vertical

    walls

    nd

    overallstructuregeometry

    isnecessary

    to

    enable

    th e

    design

    of

    asuitable,

    ost-effective

    structure,

    with

    anacceptably

    low

    risk

    of

    failure.

    t

    isalso

    desirable

    to

    minimise

    the

    risk

    of

    severexposure

    to

    arge

    breaking

    wave

    mpactorces,o

    hatuture

    maintenance

    costs

    are

    not

    undulyhigh.

    Verticalwallsin

    the

    marineenvironmentaresubjected

    to

    highlyvariablewave

    loads,

    yet

    existing

    esign

    methods

    reeterministic.

    hemain

    esign

    method

    or

    aissons,

    developed

    by

    Goda(1974,985),

    provides

    a

    good

    estimate

    of

    non-breakingwaveforces

    (McKenna,

    997).

    For

    wave

    mpacts

    owever,

    he

    redicted

    orces

    sing

    Goda's

    Assistant

    Engineer,

    BabtieGroupLtd.,

    9 5Bothwell

    Street,

    Glasgow,

    G 27HX,UK

    VisitingProfessor,

    TheQueen'sUniversity

    of Belfast,UniversityRoad,Belfast,

    B T7

    INN,

    UK

    Manager

    Coastal

    Structures,H R

    Wallingford,

    HowberyPark,

    Wallingford,Oxon

    OX10

    8BA,

    UK

    2082

  • 7/26/2019 5746-23856-1-PB

    2/14

    COASTALENGINEERING1 9 9 8 083

    method

    areeffective'alues,amped

    by

    the

    response

    of

    the

    structureandfoundation,

    ratherthan

    actual

    impact

    loads.

    hese

    wave

    impact

    loads

    were

    previously

    thought

    to

    be

    unimportantforthestabilityof

    massivestructuressuchascaissons,ut

    O umeraci

    t

    al

    (1995)

    have

    shown

    that

    repeated

    wave

    impacts

    can

    cause

    incremental

    displacements.

    It

    s

    ecessary

    o

    ecognise

    he

    wideariationnorce

    ype

    ndmagnitude,

    ndo

    incorporateameasureoftheir

    probabilityofoccurrence,nordertodevelopimproved

    designmethods.

    Work

    withinM A S TI I - P R O V E R B Sasbeenirectedtowardsthis

    Identification

    of

    a

    uitable

    tatistical

    istribution

    or

    wave

    orces

    nerticalwall

    structures

    is

    particularlymportant

    fo r

    the

    evelopment

    of probabilistic

    design

    tools.

    Previousworknth e

    tatistics

    ofwave

    orces

    asoncentrated

    n

    stablishinghe

    extreme

    istribution

    f

    eries

    of

    (theoretically)

    egular

    waves,

    ee

    or

    nstance

    Kirkgoz

    (1995),

    but

    these

    results

    cannotbe

    applied

    to

    real

    (random)

    seas.

    The

    Weibullistribution

    canbe

    used

    toescribepulsating

    wave

    forcesmeasuredn

    model

    tests

    on

    caisson

    breakwaters

    using

    randomwaves.

    llsopt

    l(1996a)

    av e

    shown

    that

    the

    onset

    of

    wave

    mpacts

    can

    be

    defined

    as

    a

    change

    n

    gradient

    of

    the

    probability

    plot

    where

    wave

    forcesstart

    to

    increaserapidlyabovethose

    predicted

    by

    th e

    simple

    Weibull

    distribution.

    The

    design

    load

    fo r

    many

    structures

    will

    however

    be

    du e

    to

    wave

    impact

    forces

    rather

    than

    on-breaking

    aves.

    t

    s

    herefore

    mportant

    o

    e

    ble

    o

    stablish

    he

    statisticaldistributionofwaveimpactforces,

    nd

    to

    knowth erelativeproportionsof

    pulsatingan dimpactforces

    fo r

    agivenstructural

    configuration.

    2.ODELTESTS

    Comprehensive

    arametric

    model

    ests

    were

    onductedn

    andom

    wave

    lume

    t

    H R Wallingford

    during

    994.

    hese

    tests

    were

    designed

    to

    nvestigate

    th e

    influence

    of

    structure

    geometry

    on

    wave

    pressures

    and

    forces,

    using

    random

    waves.

    ver

    200

    tests

    ere

    arried

    ut

    n

    ll,

    o

    xplore

    he

    ffects

    f

    arying

    he

    ollowing

    parameters:

    a)

    ignificantoffshoreandinshorewave

    heights,

    H

    s0

    and

    H

    s

    ; ;

    b)

    aterdepth

    in

    front

    of

    structure,

    h

    s

    ;andcrestfreeboard

    R ^;

    c)

    ave

    steepness,s

    m0

    ,

    and

    peak

    wavelength

    atstructuretoe,

    L

    p

    j ;

    d)

    ater

    depth

    over

    moundinfront

    ofwall,

    d;

    an d

    bermheight,

    h i , ;

    e)

    erm width,

    Bt,;

    f)

    ront

    slope

    ofmound,

    a;

    g)

    epthofembedmentofcaissonintomound,lvh

    c

    ;

    Thecaisson

    was

    instrumented

    with

    8

    pressure

    transducers

    on

    th e

    front

    face

    and

    4

    n

    the

    undersides

    hownnFigure.ressure

    datawere

    acquired

    nll

    ransducers

    simultaneously

    at

    4 0 0 H z

    fo r

    500wavesin

    each

    test.

    hese

    testshave

    been

    described

    previously

    by

    Allsopet

    al(1996b).

  • 7/26/2019 5746-23856-1-PB

    3/14

    2084

    COASTAL E N G I N E E R I N G

    1998

    Figure

    1 InstrumentedModelCaisson

    3.

    ATA

    ANALYSIS

    A

    data

    analysisprogramwasdeveloped

    to

    carry

    outtheinitial

    analysis

    of

    th e

    pressure

    dataacquiredfromthesemodeltests.hepressure

    data

    were

    spatially

    integratedat

    eachtime-stepto

    give

    horizontal

    anduplift

    forcetime

    histories.

    n

    orderto

    studyth e

    statistics

    of

    th eataet,nformationegarding

    he

    orce

    maxima

    wasequiredor

    eachwave

    event.nvent

    was

    efinedrom

    thebeginning

    of

    each

    rapidressure

    riseon

    th e

    transducer

    at

    still

    water

    level,

    nd

    th e

    analysis

    program

    was

    configured

    to

    searchfo rth erequiredinformationwithineach'event'.

    As

    horizontalandupliftpressureandforcemaximadonotnecessarilyoccuratth e

    sametime,anumberofoutputfileswerewritten,sothateachphenomenoncouldbe

    studiedseparately.

    Thenext

    stage

    of

    th e

    data

    analysis

    was

    to

    rank

    thepressure

    and

    forcemaxima

    containedin

    th e

    output

    filesin

    orderofmagnitudetoenabletheir

    statistics

    to

    be

    explored.hispaperconcentrates

    on

    the

    results

    ofthefurtheranalysis

    of

    the

    horizontal

    forces.

    4.

    TATISTICSOFW A V E

    FORCES

    Wavempact

    orces

    n

    oastal

    tructures

    re

    xtremely

    ariable,

    nd

    re

    herefore

    better

    described

    by

    their

    statistics

    than

    by

    any

    single

    deterministic

    value.

    n

    selecting

    an

    appropriate

    tatistical

    istribution

    fo r

    wave

    orces,

    t

    s

    n

    dvantage

    to

    maintain

    anyassociationbetweenwaveheightsandwaveforces.f anarrowbandprocessis

    assumed,waveheightscan

    be

    approximatedbyth eR ayleighdistribution,whichisa

    special

    form

    of

    the

    more

    generic

    3-parameterWeibull

    distribution.

  • 7/26/2019 5746-23856-1-PB

    4/14

    COASTAL

    ENGINEERING

    1 9 9 8

    2085

    Analysis

    of

    th e

    pressure

    datafrom

    these

    model

    tests

    by

    Allsop

    etal

    (1996a,b)

    showed

    that

    th e

    Weibulldistribution

    could

    be

    used

    to

    give

    a

    good

    description

    ofwave

    forces

    from

    on-breakingwaves

    referred

    o

    ere

    sulsatingwaveorces),s

    hownn

    Figure

    .t

    was

    lso

    ound

    hat

    the

    nset

    ofwave

    mpacts

    ould

    e

    efined

    y

    change

    n

    he

    radient

    of

    th e

    Weibull

    lot,

    where

    he

    wave

    orces

    tart

    ncreasing

    morerapidly

    withnon-exceedance

    level,

    Figure

    3 .

    f

    Non-exceedance

    probability

    ( )

    30.8

    3 .2

    3 .4

    m

    H si

    0.08m,

    snii

    0.03

    Structure9

    sw l

    1.70m

    -1

    ln(-ln(l-P))

    2.72

    ^

    Figure

    2 Weibull

    distribution

    ofpu lsatingwaveforces

    Non-exceedanceprobability

    (% )

    12.7

    0.83.23.49.9

    K 0

    lib

    0.367m

    hb

    0.457m

    --

    7.40

    Hso0.25msmo

    0.04

    swl1.61m

    0

    ln -ln l-P))

    8.37

    Figure3 Transition

    from

    pulsatingforcesto

    impactforces

  • 7/26/2019 5746-23856-1-PB

    5/14

    2086

    COASTAL

    ENGINEERING1 9 9 8

    Thismethodofestimatingth e

    percentage

    ofimpacts

    was

    appliedtoeach

    test,and

    was

    found

    o

    ive

    ood

    greement

    withbservationsromhelume

    estingndwith

    analytical

    considerations

    of

    the

    physical

    processes

    involved.

    hisanalysis

    led

    to

    the

    developmentof

    a

    parameter

    map,

    Allsop

    et

    al(1996a,b)that

    could

    beusedto

    estimate

    th e

    risk

    of

    wave

    impacts

    on

    a

    particular

    structure.

    Furtheranalysisby

    McKenna(1997)revealedthatth eWeibull

    distribution

    couldalso

    be

    used

    to

    describe

    wave

    impact

    forces

    in

    some

    instances,

    Figure

    4.twasalsonoted

    that

    fo r

    other

    casesth epercentage

    ofimpactsould

    be

    difficultto

    determine,

    s

    th e

    change

    in

    the

    gradient

    of

    th e

    distribution

    was

    not

    distinct.

    hese

    cases

    were

    selected

    fo r

    urther

    nvestigation,

    ndt

    was

    foundthat

    th e

    ata

    oints

    lotted

    s radual

    curve

    n

    Weibullxes,atherthan traightinewith

    harp

    hangenradient,

    Figure

    5.

    The

    tructural

    onfigurations

    orresponding

    o

    hese

    urved

    istributions

    ere

    investigated

    nd

    acommon

    linkwasound

    n

    th e

    relationship

    between

    theffshore

    waveconditions

    an d

    th e

    localwater

    depth

    at th e

    structure.

    imiting

    valuesofH

    so

    /h

    s

    =

    0.425

    nd

    mo

    /h

    s

    17were

    dentified,

    eyondwhichwaveorcesoongerithe

    Weibull

    distribution,

    Figure

    6.

    he

    distortion

    ofth edistributioniscaused

    by

    shallow

    water ave

    ransformations.he

    mallestaves

    each

    he

    tructure

    elatively

    unchanged,

    but

    there

    isa

    gradual

    increase

    in

    th eamount

    of

    modification

    of

    the

    wave

    shape

    as

    th ewave

    heights

    and

    lengths

    increase.The

    incidentwaveheight

    distribution

    was

    found

    to

    be

    non-R ayleigh

    for

    these

    cases,

    Figure7.

    Non-exceedance

    probability

    (% )

    12.7

    0.8

    3 .2

    3 .4

    9.9

    f

    0

    a

    Hsi

    0.15m,

    sini

    0.04

    V

    1.72

    Structure3

    swl

    1.52m

    ln -ln l-P))

    Figure

    4

    Weibull

    Distribution

    of

    horizontal

    waveimpact

    forces

  • 7/26/2019 5746-23856-1-PB

    6/14

    C OASTAL E N G I N E E R I N G

    1998

    2087

    N on-exceedanceprobability( )

    0.02

    2.7

    3 .2

    2

    t e s t _ 1 0 0 2 < r ]

    -

    -2

    ln(-In(l-P))

    Figure

    5 Non-Weibull

    distribution

    z

    0.14

    Figure

    6 ParameterrangesforWeibulldistributedforces

  • 7/26/2019 5746-23856-1-PB

    7/14

    2088

    COASTAL

    E N G I N E E R I N G

    1998

    offshoreH nshoreH

    Spectrum

    H2 5S 6E D ,

    S W L

    1.34m

    1

    - ln 1-p)

    Figure7

    Modifiedinshore

    wave

    height

    distribution

    5.

    EVELOPMENT

    OF

    PARAMETERM AP

    Attempts

    to

    analyse

    th e

    full

    data

    se tby

    Allsop

    et

    al

    (1996b)

    had

    highlighted

    th e

    need

    to

    eparate

    he

    ata

    nto

    egions

    f

    imilar

    esponse

    haracteristics.

    he

    dimensionlessparameters

    that

    influenced

    waveforces

    were

    identified

    as:

    elative

    mound

    height,

    hb/h

    s

    ;

    elativewave

    height,

    H

    s

    ;/d;

    elative

    bermlength,B

    eq

    /L

    P

    i .

    The

    effectsof

    these

    parameterswere

    summarised

    inth eparametermappresented

    by

    Allsop

    t

    al

    (1996a,b).

    hederivation

    ofthat

    parameter

    map

    ad

    concentrated

    ona

    central

    ore

    of

    data,

    where

    he

    waterepth

    ver

    he

    ubblemound

    erm,

    , as

    greater

    than

    twice

    he

    ignificant

    nshore

    waveheight,

    H;.

    ases

    where

    he

    water

    level

    wascloseto

    or

    below

    thetop

    of

    th erubble

    moundwereexcluded.

    Potentialweaknesseshad

    been

    recognised

    in

    that

    derivation,

    both

    inexcluding

    of

    a

    se t

    of

    data

    fromth eanalysis,andinusingth edimensionlessparameterH

    s

    ;/d,whichtends

    toinfinityasth ewaterdepthoverth erubble

    mound

    tendsto

    zero.

    twasconcluded

    that,

    fpossible,

    his

    arameterhouldot

    e

    sedn

    urther

    work

    oxtend

    he

    parametermap

    to

    include

    th e

    fulldataset.

    The

    approach

    used

    n

    this

    urther

    work

    to

    include

    th e

    ull

    data

    et

    in

    the

    parameter

    map

    onsidered

    he

    ffects

    fhe

    on-Weibull

    istributed

    orces,

    nd

    ncluded

    additional

    parameters

    to

    describe

    th eeffectsoftransformationfrom

    the

    offshorewave

    climate

    to

    th e

    inshore

    wave

    climate.he

    dimensionless

    parameters

    used

    nth e

    new

    parametermap

    (Figure

    8)

    are:

    elative

    offshore

    wave

    height,H

    S 0

    /h

    s

    ;

    elativeoffshore

    wave

    length,

    L

    m o

    /h

    s

    ;

  • 7/26/2019 5746-23856-1-PB

    8/14

    s

    a

    o

    a

    s

    n

    J

  • 7/26/2019 5746-23856-1-PB

    9/14

    2090 O A S T A L E N G I N E E R I N G

    1998

    elativeinshore

    waveheight,H

    s

    j/h

    s

    ;

    elative

    berm

    height,hb/h

    s

    ;

    elativeberm

    length,

    B

    eq

    /L

    p

    j .

    The

    significance

    ofeach

    ofthese

    parameters,

    andtheircontribution

    to

    th e

    description

    of

    the

    overall

    physical

    processes

    aredescribed

    below.

    5.1

    egree

    of

    Wave

    Breaking H

    so

    /h

    s

    andL

    mo

    /h

    s

    )

    Inorder

    to

    represent

    th e

    effectsofstructural

    geometry

    properly,it

    is

    first

    necessaryto

    identify

    hoseests

    whereth eeabedauses

    wavebreakingnthe

    pproach

    o

    he

    structure.heseestsmust

    e

    dentified

    t

    he

    eginningofth enalysis,

    s

    hey

    would

    distort

    anyconclusionsaboutth e

    overall

    response

    to

    th e

    structure

    itself.

    hese

    tests

    were

    easily

    identified

    by

    th e

    curved

    distribution

    of

    data

    when

    plotted

    on

    Weibull

    axes,asdescribedpreviously.

    Further

    work

    is

    required

    to

    identify

    th e

    parameter

    influences

    n

    ases

    or

    whichthe

    data

    are

    not

    Weibulldistributed. suitable

    statistical

    distributionmust

    be

    established

    fo r

    thesetests

    andth eresponse

    to

    the

    structuregeometry

    may

    then

    be

    identified

    from

    this

    new

    analysis.

    Itis

    not

    possible

    at this

    stageto

    speculateon

    effectsof

    individual

    parametersin

    these

    cases,

    but

    it

    is

    certain

    that

    structures

    in

    shallow

    water

    are

    at

    risk

    of

    exposure

    to

    breakingwave

    forces.

    he

    overall

    levelofforces

    on

    thestructuremayhoweverbe

    low,as

    th e

    incident

    wavesmay

    be

    substantially

    broken

    (and

    aerated)

    before

    reaching

    th estructure.

    lackmore

    and

    H ewson

    (1984)haveshownthat

    wave

    forcesdueto

    highlyaerated

    waves

    are

    significantly

    lower

    than

    corresponding

    'deep

    water'

    waves.

    Configurationsthat

    fallinto

    this

    category

    should

    be treatedasthough

    they

    will

    be

    subjected to

    high

    impact

    forces,anddesignedaccordingly,until

    further

    work

    has

    been

    carried

    out

    in

    this

    region

    ofth e

    parameter

    map.his

    isparticularly

    important

    for

    structures

    in

    areas

    with

    high

    tidalranges.

    5 .2

    averegime

    at

    structure

    H

    S

    i/h

    s

    )

    The

    most

    ignificantarameterffecting

    he

    nset

    fwave

    mpacts

    or

    Weibull

    distributeddata

    wasfound

    to

    be

    th erelative

    incidentwave

    height,

    H i/hs.

    his

    parameter

    representsheikelihood

    fwave

    reaking

    t

    heoe

    f

    the

    tructure,efore

    ny

    significant

    interaction

    with

    the

    structurehastaken

    place.

    hecriticalvalueofHsi/h

    s

    to

    cause

    th e

    onsetofimpacts

    was

    investigatedbyplotting

    the

    percentage

    ofimpacts,P i% ,

    for

    each

    testagainst

    H

    S

    j/h

    s

    .

    Initial

    investigations

    usingWeibulldistributeddatafromstructures0,,and2

    suggested

    that

    if

    th e

    valueofH

    s

    j/h

    s

    was

    lessthan

    a

    criticalvalueof

    0.2,

    there

    would

    be

    nowave

    breaking,and

    th e

    resulting

    forces

    would

    therefore

    be

    pulsating,see

    Figure

    9.

    Extension

    of

    thisnalysis

    o

    nclude

    ll

    tructureshowedthatmpacts

    id

    ccur

    or

    some

    configurations

    where

    H

    i/h

    s

  • 7/26/2019 5746-23856-1-PB

    10/14

    COASTAL

    E N G I N E E R I N G

    1998

    2091

    verticalwall

    tructures

    and

    structures

    with

    lowrubble

    mounds.

    second

    limitofH si/h

    s

    =

    0.3

    wasidentifiedasthe

    point

    beyondwhichomeimpactswererecorded

    in

    all

    tests,

    alsoillustratedinFigure

    10 .

    Figure

    9 Variation

    in

    percentage

    of

    impactswithH

    si

    /h

    structures

    0,1and2)

    ,

    0

    0 . 1

    ^11H-S^

    0 .2

    . 3

    Hsi/hs

    Figure 0

    Variation

    in

    percentageofimpactswith

    H

    si

    /h

    s

    all

    structures)

  • 7/26/2019 5746-23856-1-PB

    11/14

    2092

    C O A S T A LE N G I N E E R I N G1998

    Inthe

    zonebetweenthese

    limits,ie

    0.2