SR329 Conveyance Meandering Channels HRWallingford

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 256

ConveyanceEstimationfor

Meandering
Ghannels

C S James
J B Wark

Report SR 329
December1992

HR Wallingford
RegisteredOffice: HR Wallingford Ltd. Howbery Park, Wallingford, Oxfordshire OXl0 8BA' UK
Telephone:0491 35381International+ 44 49135381 Telex: 848552HRSWAL G.
Facsimile: Cdlgl 32233lnternational + 44 491 32233 Registeredin England No. 1622174

s8329 1710493
Contract
This repoddescribesworkcaniedout as an exlensionof an NRA R&Dproject.
This report was furded by the Ministryof Agriculture, Fisheriesand Food
(MAFR underthe commissionon River flood protec{ion,nominatedproiect
officerB D Richardson.The HR iob numberwas QPS/OO13|J2.

This work was canied out by Prof C S James of The Universityof the
and Mr J B Warkof HR Wallingford.The proiectwas managed
Witwatersrand
by Dr NigelWalmsley.

This reportis publishedon behalfof the Ministryof Agriculture,Fisheriesand


Food, but any opinionsexpressedare not necessarilythose of the Ministry.

Preparedby
fu'0.0'l' x{tl 1,

Checkedby )*8.O*rk-
U

Approvedby *ell?3,

Date..Xt[..{.11

@ HR Wallingford
Limited1993

sR 329 2cil03/93
Title page
Contract
Summary
List of symfuls
Contents

Introduction 1.

Straightcompound chdnnel methods 2


2.1 Background 2
2.2 Straightchannelmethods 3
2.3 Field and laboratprydata on straightchannels 5
2.4 Applicationof stfaight channel methodsto straight
channeldata... 7
2.5 Application of straight channel methods to
meanderingcha4neldata. ...... I
2.5.1 The mdthods 9
2.5.2 The data sef . . 10
2.5.3 Resu/fs 11
2.6 Summary 12

Laboratory investigaticins into meandering overbank


flow . 12
3.1 Background...,. 12
3.2 Laboratorydata r 13
3.2.1 PhaseB SERCFCF . 13
3.2.2 Aberdeen ......15
3.2.3 US Arrfiy Vbksburg 16
3.2.4 Toebedand Sool<y 17
3 . 2 . 5 S m i t hi . . . . 18
3.2.6 JamesAndBrown 19
3.2.7 Kiely.i.... 19
3.2.8 SteinahdBouve ......20
3.3 GeneralOommedrts ......21
3.4 Bedfriction ... 22
3.5 Summary 23

Inbankmeandering
flort 24
4.1 Background...i. 24
4.2 Energyloss in cfiannelbends 26
4.2.1 MethodSof evaluatingnon-frictionlcsses 26
4.2.2 Data sdts 26
4.2.3 Resultd,and conclusions 27
4.3 Energyloss mecfnnisms 29
4.4 Stagedischargepredictionmethods . . . 32
4.5 Applicationof pr{dictionmethods 40
4.5.1 Data sdt 41
4.5.2 , FesultCand conclusions 41
4.6 Summary 43

sR 329 07/O7l93
Contents continued

Formulationof the procedurefor overbankmeandering


channels 44
5.1 lmportantmechanisms 44
5.2 Methodsavailablein the literature 44
5.3 Approachto conveyanceestimation . . . 47
5.4 Formulationzone 1 47
5.5. Formulationzone 2 56
5.5.1 Expansioncontmctionmodel 56
5.5.2 EmpiricalModel 61-
5.6 Formulationzones3 and 4 . . . . 64
5.7 Boundaryshearstresses 65
5.8 Summary 66

Verificationof the procedure 67


6.1 Background 67
6.2 Methods 67
6.3 Applicationto laboratorydata . 68
6.3.1 Data . 68
6.3.2 Totaldischargeand stage . 68
6.3.3 Di*harge distibutions 72
6.4 Applicationto straightlaboratorydata . 73
6.5 Applicationto fielddata . 73
6.6 Summaryand conclusions. 75

Futureresearchneeds .. . 76
7.1 Stage-discharge prediction for inbankflows . . . . 76
7.2 Laboratory studies . . 76
7 . 2 . 1 E x t e n s i o n oefx i s t i n g d a t a s e t s ...... 76
7.2.2 Laboratorysfudiesof loss mechanisms. . . . . 77
7.3 Fielddatacollection .....-78
7.3.1 Stntegy for field data ollectbn . . . . . 78
7.3.2 Suitablesites ...79
7.3.3 Hydraulbdata. ..79
7.4 Computational modelling. .. .. . . 80
7.4.1 Turbulencenndelling ...80
7.4.2 One dimensionalmodelling . . . 80

C o n c f u s i o n .s. . ...82

Acknowledgements

10 References ... 88

Tables
Table 1 Riverchanneland floodplainwidths
Table 2 Riverchanneldepthsand roughnessvalues
Table3 Meanenors for straightfield data
Table4 Meanenors for eachstraightfield site
Table5 Summaryof SERCPhaseB stagedischargetests

sR 329 07r'07/93
Contents continued
Table 6 Mean errods straight methods meandering data
Table 7 Summaryof SERCPhaseB tests
Table I Summaryof Aberdeenexperiments
Table 9 Summaryof Vicksburgexperiments2ft wide channel
Table 10 Geometricpfr:arneterslab studiesmeanderingchannels
Table 11 Nondimensipnal geometric parameters meandering
channels
Table 12 Bend losse* for 60o meander geometry, trapezoidal
crciss-sec{iorlr
Table 13 Non-frictionlosses for 60o meander geometry,natural'
cross-sectioil
Table 14 Non-friction lossesfor 110' meandergeometry,natural
cross-sectioir
Table 15 Summaryof averageenors in bend loss predictions
Table 16 Contraction lbsscoefficients (Rouse, 1950)
Table 17 MainchannQlintegrateddischarges
Table 't 8 Variablesfol definingmainchannelflow
Table 19 Adlustedvariablesfor definingrnainchannelflow
Table 20 Roughness dnd sinuosityadjustmenlto Q./
Table 21 Roughness 4nd sinuosityadjustmentto c
Table 22 Errors(%) in reproducingQ,/ for high valuesof y'
Table 23 Datasetsfoqinnerfloodplainanalysis
Table 24 Eqdationpailameters for y' greaterthan 0.2
Table 25 Geometricdhtaoverbanklaboratorystudies
Table 26 Geometricdpta Soolqy'slaboratorystudy
Table 27 Mainchanndlgeometricdata
Table 28 Mean% errQrsin dischargeFCF data
Table 29 Mean7o errdrsin dischargeAberdeen, Vicksburgand
Kielydata
Table 30 Mean% errqrsin dischargeSookydata
Table 31 Mean% errqr in dischargeall data
Table 32 Rankingof q\ethods
Table 33 Mean% errQrsin stage FCF data
Table 34 Mean% err{rs in stageAberdeen, Vicksburgand Kiely
data
Table 35 Mean% errdrsin stage Soolqydata
Table 36 Mean% errcirsin stage all data
Table 37 Sensitivitytq$s : effectof errors in wave length
Table 38 Sensitivitytlsts : effectof erors in channelside slope
Table 39 Measuredzqnaldischarges, Sookyand Kielydata
Table 40 Errors(%) ancalculatedtotalflows,Sootqyand Kielydata
Table 41 Medsuredartrdcalculatedflow distributions
Table 42 Reachavera{ged geometricparametersRodingstudy
Table 43 Errorsin prepictingoverbankdischarges
Table 44 Serlsitivityttsts on the effectof floodplainroughness

Figures
Figure1 Riverchanndlscross-sections
Figure2 Riv6rchannQlscross-sections
Figure3 Myerslabor{torychannelcross-sections
Figure4 Meanenorsfor straightchanneldata

sR 329 07/07/93
Contents continued

Figure5 Detailed plan geometry of FCF OOo meander


Figure6 Phn of flumeand naturalcross-section geometryfor 60"
meander
Figure7 Detailedplangeometryfor FCF 110omeander
Figure8 Planof flumeand naturalcross-section geometryfor 11Oo
meander
FigureI Plan geometriesof the Aberdeenflume with channel
sinuositiesof 1.40and 2.04 (afterWilletts1992)
Figure10 Plan geometryof the Aberdeenflume with channel
sinuosityot 1.21(afterWilletts1992)
Figure11 Plan and cross-sectionsfor Vicksburgflume (after US
Army1956)
Figure12 Plan and cross-sectionsfor Sooky's flume (after
Sooky1964)
Figure13 Plan and cross-sectionsfor Kiely's flume (after
Kiely1e9O)
Figure14 Flow processesin a meanderingc€mpoundchannel
(afterErvineand Jasem1991)
Figure15 Adjustmentto Manning'sn for bend losses:measured
and oredicted
Figure16 Predictedadjustmentsto n for bend losses:modified
Changmethod
Figure17 Cross-section subdivisionfor overbankflows Ervine
and Ellismethod(1987)
Figure18 Variationof main channeldischargealong a meander
duringoverbankflow (FCFPhaseB)
Figure19 Inbankstage-discharge relationshipfor 60otrapezoidal
channel
Figure20 Inbank stage-dischargerelationshipfor 60o natural
channel
Figure2l Inbank stage-discharge relationshipfor 110' natural
channel
Ftgure22 Variationof dimensionlessmain channeldischargewith
flow depth
Figure23 Variationof dimensionlessmain channeldischargewith
dimensionless flowdepth
Figure24 Variation of dimensionless nrainchanneldischargewith
dimensionless flow depthwith pointsadjustedfor friction
factorratio
Figure25 Additionaladjustmentto dischargefor relativeroughness
Figure26 Adjustmentto c for relativeroughness
Figure26b AdjustmenttactorYe
Figure27 Flowexpansionover a downwardstep
Figure28 Flow contraction over an upward step
Figure29 Expansionand contractionflow patterns
Figure30 Widthto depth ratioconectionfor expansionlosses
Figure31 Widthto depth ratioconectionfor contractionlosses
Figure32 Widthto depthratiocorrectionfor combinedexpansion
and contractionlosses
Figure33 Errorsfor SERCpredictions beforesinuositypredictions

sR 329 07107/93
Contents continued
Figure34 Erors for Aberdeen predictions before sinuosity
predictions
Figure35 Adjustment factor for inner flood plain discharges for
SERCPhaspB experiments
Figure36 Adjustmentfactor for inner flood plain dischargesfor
Aberdeenexperiments
Figure37 Adjustmentfactor for inner flood plain dischargesfor
y'> O.2, SERCPhaseB data
Figure38 Adjustmentfactorfor inner flood plain dischargesfor y'
> 0.2, Aberdeendata
Figure39 Variationof h with s
Figure40 Variationof b with B?A
Figure41 Compadsonof predictedzone 2 adjustmentfactor with
SERCdata
Figure42 Compadsonof predictedzone 2 adiustmentfractorwith
Aberdeendata
Figure43 Example of boundary shear stress distribution in a
meanderingpompoundchannel(afterLorena)
Figure44 Cross-sectioh subdivisionof overbankflows, Jamesand
Wark
Figure45 Enors in prepicteddischargeand depth BFO
Figure46 Errorsin predicteddischargeand depth JW
Figure47 Errorsin prepicteddischargeand depth JW2
Figure48 Errorsin predicteddischargeand depth EE
Figure49 Errorsin predicteddischargeand depth GH4
Figure50 Errorsin predicteddischargeand depth GH5
Figure51 Planof Rivei Rodingstudysite
Figure52 Samplecross-section BiverRoding
Figure53 PredictedstdgedischargesRiverRoding

Plates
Plate 1 FCF 60" cheinnelgeometry
Plate2 FCF 110"cliannelgeometry
Plate3 FCF naturalcrosssection
Plate4 FCF rod roughenedflood plain
Phte 5 FCF paniallyrod roughenedflood plain
Plate6 FCF breezeblockroughenedflood plain

Appendices

Appendix1 Methods for determining equivalent roughness of


channelswitfrcompositeroughness
Appendix2 The lateraldistributionmethod
Appendix3 The Ackers design method for straight compound
channels
Appendix4 Datafrom laboratorystudiesinto meanderingflow
Appendix5 Summaryof the AckersRod Roughnessmethod

sR 329 07/07/93
1 lntroduction
The estimationof channel conveyanceis probably the most comrnonly
enountered problem in river engineeringpractice. lt requiresquantitative
accountingfor the energydissipatedor'lost' by the flow. Energycan be lost
by a varietyof differentmechanisms,dependingon the physicalcharac{eristics
of the channeland the flow condition. In a straight,prismaticchannelthe
energy loss can be ascribedwhollyto friction. Bends in a channel induce
secordarycirculationin the flow,whicheffectivelyaddsto the energyloss by
reducingthe energyavailablefor the primaryflow. For overbankflows in a
straightchannel,further energy is lost throughthe interactionbetweenthe
rnain channel and flood plain flows. For overbankflow in a channel with
bends,the mechanismsare yet morecomplexandthe energylossstillgreater.

The Scienceand EngineeringResearchCouncil (SERC) Flood Channel


Facility (SERC FCF) was constructedat HR Wallingford(HR) in order to
providea nationalfacility in which large scale experimentson flow in flood
channels couH be carried out. Thus the effects of the complex flow
mechanismscould be studiedand quantified.

A researchprogrammeis beingcarriedout in three phasesas follows:

- PhaseA, straightand skewedchannels


- PhaseB, meandering channelsand dispersion
- PhaseC, sedimentexperiments (provisional)

The programmeis sponsoredprincipallyby the SERCand HR with financial


supportfromthe NationalRiversAuthority(NRA).The experimentsare carried
out by a numberof universityresearchgroups.

The PhaseA experiments werecompletedin June1989and a designmethod


for estimatingflood dischargesin straight compoundchannels has been
produced(Ackers,1991). This analysisstemmedfrom a needto makethe
results of the researchbeing caried out on the SERC FCF availableto
practisingengineersin a usefulform. Ackercwork was fundedby the former
RegionalWaterAuthoritiesof Englandand Wales, later the NationalRivers
Authorig(NRA), and HR Wallingford.

The PhaseB experimentswerecompletedin October1991 and in Septenrber


1991the NRAinvitedHRto tenderfor an B&D projectto producea procedure
basedon the resuttsof the PhaseB experiments.The objectiveof the project
was to preparean hydraulicmanualto allow the NRA to utilisethe resultsof
the PhaseB experimentscanied out in the SERC FCF.

It was understoodthat the intentionof the NRA in commissioningthis project


was that the end productshould be of immediateapplicabilityto the NRA
regionalotficesandtheirfieldstaff. Thiswas bornein mindduringthe course
of the proiectand significantlyaffectedthe forms of analyseschosen.

The projectinvolvedthe followingtasks:

1 Summarize the workdoneduringthe FCFprogramme straightcompound


channels,by the researcherswho took the measurementsand others
who analysedthe data.

sa 329 07/07/93
Applythe FloodChannelFacilityPhaseA Method,(FCFAM)to fielddata
and compareit with otherstraightchannelmethods.

Summarizelhe data availablefromthe SERC FCFand other laboratory


or field str.rdies
into meanderingcompoundchannels.

Developproceduresto describethe head loss in inbank meandering


flows.

Developproceduresto describethe head loss in overbankmeandering


flows.

Developguidelinesto assistin determining


boundaryshearstressesii
meanderingoverbankflow.

Deviseguidelinesto assistin applicationof the proceduresto naturat


channels.

The NRA requiredonly a shortsummaryof the resultsof each of thesetasks


in the final reportsubmittedto them. lt was felt to be importantthat the full
detailsof the developmentof the methodsand the conductof the research
should be recorded as well. Consequentlythe Ministry of Agriculture,
Fisheriesand Foodagreedto fund productionof this moredetailedreport.

The workcardedout on the Ackersmethodand straightcompoundchannels


is describedin Chapter2. Chapter3 gives a summaryof the available
information
on meandering compoundchannels.The analysisof inbankand
overbankmeanderingflowsare reportedin Chapters4 and 5 respectively.

The variousmethodsfor overbankmeandering channelsare appliedto the


availablelaboratoryand field data in Chapter6. Chapter7 gives some
recommendations for futureresearchand the mainconclusions
of the study
are sumarisedin Chapter8.

2 Straight compound channel methods


2.1 Background
The initial part of the NRA R&D project involvedpreparinga summaryof
previouswork canied out on the conveyancecapacityof straightcompgund
channels,Ackers (1991). The methodAckers producedis based on an
exhaustiveanalysisof the data collec{edon the SERC FCF duringPhaseA,
by teams of researchersfrom various universities. The NRA also
commissionedHR Wallingfordto evaluatethis methodagainstother straight
channelmethodsby applicationto a rangeof channeltypes and conditions;
but the NRA requestedthat the SERC FCF Phase A data should not be
analyzedfudher,since this has alreadybeen carriedout by Ackers (1991).
This workextendssomeearlierwork,(Warket al , 1991),caniedout at HR for
MAFFby includingthe newmethoddevelopedby Ackersin the analysis.This
methodwill be referredto as the FloodChannelFacilityPhaseA Method,
FCFAM.

In addition,some of lhese methodsfor calculatingdischargein straight


compoundchannelswere appliedto a selectionof the meanderingchannel
data collectedduringPhaseB of the SERCFCF work. The purposeof'this

SR 329 07197193
m
was to confirmthat straightchannelmethodsare inappropriatefor use when
analysingmeanderingchannelswith overbankflow.

2.2 Straight channel methods


The methodsdetailed below were ldentifiedas appropriatefor applbation to
measuredstagedischargesin straightcompoundchannels.

LDM Methodwith NEV = 0.16


LateralDistribution

DCM DividedChannel Method,using vertical division lines which are


includedin the wettedperimeterof the main channelbut omitted
from the wettedpedmeterof the flood plains

SCM Single Channel Method, applying the bankfull rnain channel


Manning'sn valueto lhe wholecompoundchannels

SCM2 Horton'sCompositeBoughnessMethod

SCM3 Lotter'sCompositeRoughnessMethod

SCM4 Einsteinand BanksCompositeRoughnessMethod

SCMS and ChristensenCompositeRoughnessMethod


Krishnamurthy

SSGM Sum of SegmentsMethod

DCM2 DividedChannelMethod,usingverticaldivisionlineswhichare not


includedin the wettedperimeterof eitherthe main channelor the
flood plains

FCFAM Method developedby Ackers, based on Flood Channel Facility


PhaseA data. Mainchannelslopeusedfor mainchannelflow.

These methodsfor calculatingdischargein straightcompoundchannelsfall


into four broadcategories:

- dividedchannelmethods
- methodof segments
- compositeroughnessmethods
- more complexphysicallybasedmethods

Two divkledchannelmethods(DCMand DCM2)wereconsidered.Theseare


based on verticaldivisionlines at the main channel-floodplain boundaries
therebydividingthe flow betweenthe main channeland the flood plains. In
a review of some simple methodsof estimatingdischargein compound
channels Ramsbottom,(1989)concludedthat the division lines should bq
includedin the wettedperimeterof the main channelbr,rtnot the flood plains.
This approachis adoptedin the DCM methodwhereasin DCM2the division
lines are not included in either the main channel or flood plain wetted
pedmeters.

The sum of segmentsmethodand the LDM allowvariableroughnessvalues


to be assignedacrossthe channelperimeter,whereasthe dividedchannel
methodsby definitionallowroughnessesto be designatedfor the wholeof the
main channeland the flood plains. The compositeroughness'methods

sF 329 07rc7l93
consideredherewere identifiedthroughearlierworkcarriedout at HR and are
listed in HR Report EX 1799. They are all basedon the assumptionthat a
representativeroughnessvalue for the whole compoundchannel can be
obtainedby takingsomeweightedaverageof the roughnessvaluesassociated
with differentregionsacrossthe channel.Appendix1 repeatsapperdix Fl of
Ex 1799and givesthe detailsof the variousmethods. lt is worth roting that
Lotte/s method(SCM3)and the sum of segmentsmethod(SSGM)are based
on ldenticalassunptionsand give identlcalresults. The sum of segments
methodis simpleto applyand is popularin numericalmodels.

The Lateral Distribution Method (LDM) is based on the work of


wark et al (1990). The LDM was developedfrom an anatysisof the results
obtained from Phase A of the SERC FCF work. lt is a fairly mmplex
mathematicalmodel of flow distributionsin straightcompoundchannelsand
is basedon approximations to the physicalprocessesratherthan an empirical
approach. The detailsof the methodare givenby Warket al (1990)and are
summarizedin Appendix2. The mainproblemwith the LDMties in the choice
of an appropriateNon-dimensional Eddy Viscosity(NEV)value. The work
repoiledherewas carriedout witha fixedvalueof 0.16. This vatuehas been
foundto giveacceptableresuhsin a fairf wide rangeof situations(seewark
et al, 1990and 1991and Chapter6 of Ackers,1991). The quotedvalueof
0.16gavereasonably goodresultsoverall whenappliedto the FCFphaseA
datra. However,when consideringindividualdata pointsthe optimumvalues
obtainedvariedwith flow depthand othergeometricvariablesover the range
0.1to 0.5.

The FCFAMproceduresare basedon a detailedanalysisof stage discharge


data collectedfrom the FGF during PhaseA. The FCF data is the most
comprehensive and accuratelaboratorydata set that exists. other laboratory
datasets werealsoconsidered by Ackers(1991)and alsoa selectionof fietd
data. The FCFAM procedures allow prediction of stage-discharge
relationships,division of total dischargeinto main channeland flood ptain
componentsand estimationof boundaryshear stresses. This methodfor
stagedischargepredictionis basedon the dividedchannelapproach(DCM2).

A basicdischargeis calculatedseparatelyfor eachzone usinga conventional


resistanceequation(suchas Manning's)and these are then addedtogether
to give the total basic discharge. This is then adjustedto accountfor the
effects of the interactionsbetweenthe zonal flows. Four regions of flow
behaviourwere identified,(Ackers 1991),within which the variationof the
interactioneffectwith flowdepthwasdifferent.A differentadjustmentfunction
is presentedfor each region. lt is not possibleto identifythe appropriate
regionand functionfor a pafiicularwaterlevelbeforehand,but a procedureis
given for selecting the correct result from those obtained using each
adjustmentfunction. An additionalcorection is also givento accountfor the
effect of rnoderateapgles of skewnessof the main channel. The total
dischargecan be dividedinto mainchannelandflood plaincomponentsusing
intermediateresultsfrom the primarycalculations.

The interactionbetweenmain channeland flood plain flows also affectsthe


magnitudeand distributionof boundaryshear stress in a compoundsection.
Proceduresare given for making provisional estimatesof the avenge
boundary shear stress in the main channel and lhe average and local
maximumvalueson the floodplains.

sF 329 07/07/93
The FCFAM method for stage-dischargeprediction,separationof rnain
channeland flood plain flow ard estimationof boundaryshear stress has
been summarisedas a step by step designprocedurein Appendix3. This
sumrary was used as lhe basis of an intemal NRA document which is
intendedfor distributionwithinthe NRA,R&D Note 44.

2.3 Field and laboratorydata on straight channels


Rivergaugingdata at siteswith compoundchannelswas suppliedto HR by
varbus NRA regionsas part of a previousMAFFfundedstudy into methods
of improvingflood dischargeassessment(Ramsbottom1989and Warftet al
1991). Additionallaboratorydata was suppliedby Dr WRC Myers of the
Universityof Ulster, who also supplieddetails of the River Main study in
Nodhemlreland. The MAFFstudyinvofuedthe evaluationof variousmethod6
of estirnatingdischargerincompoundchannels. The resuftsreportedin this
sectionextendthis earlierworkto includethe FCFAMmethod.

In total eightsetsof field gaugingdatawereobtained.The identifiedchannels


displayeda widevariationof geometricparameters,see Tables1 and 2. The
cross-sectionsare also shownin Figures1 and 2. As is to be expectedwith
data collectedunderfieldconditionsnoneof the informationis perfectand all
of the sites show some deviationfrom the ideal cases studied in the
laboratory. In particularnot allthe casesare truly straightprismaticchannels
with uniformflow. The varioussites and the data collec{edfrom them are
describedin detail by Ramsbottom(1989),Wark et al (1991)and Ackers
(1991). As can be obseruedfor someof the sitesit is difficultto definethe
bankfull level exactly in these cases each engineer must apply his own
judgement.Whenapplyingthe FCFAMproceduresit maybe dfficult to define
the geometricparametersused. lt is not ihe purposeof this reportto act as
a detaileddesignguideand if the proceduresgivenin Appendix3 are found
to be inadequatefurtherguidanceshould be sought from the report by
Ackers(1991).

Bed roughnessvalues

As with the geometricpiopertiesof the channels,the variationof roughness


valuesfrom the identifieddata sets is also notable. Table 2 gives the main
channeland flood plainroughnessvaluesused in this analysisalongwith the
bankfulland maximumstragevalues.Thevariousmethodswereappliedto the
data usingtwo basicestimatesof the rnainchanneland floodplainroughness
values.

The Authors'estimatesof main channelroughnessvalues vary from about


0.025to 0.046and the flood plainroughnesses from 0.025to about0.100.
The mainchannelroughnessvalueswere obtainedby backcalculationusing
the inbankmeasuredflows. The bankfullvaluewas usedat all higherstages.
The floodplainroughnessfor the Blackwater,Main,Sevemandthe Teeswere
estimated by comparingmeasuredand calculated distributionsof depth
averagedvelocityprofilesacrossthe flood plains. The LDM methodwith a
NEV valueof 0.16 was usedto simulatethe measuredfloodplainvelocities
and the valuesof Manning'sn whichgavethe bestfit betweenthe calculated
and measuredvelocityprofileswas adopted. In the case of the Tonidgean
estimatedvalueof n = 0.06was assumed.In generalthevaluesadoptedare
similarto those obtainedby Ramsbottom(1989)on the basis of the divided
channelmethod(DCM).

sR 329 07rp7l93
Ackers (1991)appliedthe FCFAMmethodto some of the sites listed in
Table1, MainSections6 and 14, Severn,Torridgeand the Trent. Table5.3
of Ackers(1991)repodsthe meandiscrepancies, definedas: (1.0- OJQ*J
obtainedfor eachsite witha rangeof roughnessvaluesfor both mainchannel
and floodplain.The secondset of roughnessvalueslistedin Table2 are the
valueswhich Ackersfoundto give the minimummeandiscrepancyfor each
site, in effectoptimumvaluesof flood plainManningn for the FCFAM.

The informationon the River Main used by Ramsbottom(1989),


Warket al (1991)and Ackers (1991)has been found to containan error
arisingfrom publishedinformation, Myers(1990)and Higginsonet al (1990).
In additionto providinga conectedcross-sectiongeometryfor Section14,
Myers(1992)also confirmedthat the channelgradientis variablealongthe
study reachand flow is non-uniform.As this reachof the Mainalso includes
berds, Ackers(1991)concludedthat it is not strictlyan appropriatetest case.
As a result a correction was issued to SR 281 by Ackers, effectively
withdrawing this site fromhis analysis.

However,the bendson this reachof the RiverMainare not particularlytight


and straightchanneltechniquesare regularlyappliedby engineersto rivers
withthistypeof planform. The stagedischargedatafromthe two RiverMain
sections is includedin the analysisdescribedhere since the Authors
consideredthat in practiceengineersregularlydealwithchannelsof thistype.
The applicationreportedhere was canied out using the correctedcross-
sectionand the averagehydraulicslope repoiledby Myers et al (1990).
Hencethe backcalculatedmainchannelManning'sn valuefor Section14 is
now 0.0278ratherthan0.0248. Flowin the studyreachon the RiverMainis
knownto be non-uniform, the hydraulicslopevariesalongthe reach. The
roughness valuesderivedusingthisaveragehydraulicslopeare notvalidfor
otherhydraulicslopes.

The othersites used are gaugingstationsoperatedby the NRA and water


surfaceslopesare not regularlymeasuredwhengaugingflowsat thesesites.
The implicitassumptionis made that the gaugingsites are situatedwhere
steady,uniformand normalflowconditionsoperate. No informationexiststo
confirmor refutethis assumptionbut it shouldbe notedthat naturalflowsare
rarely steady, uniform or normal. lndeed a close inspectionof the rating
sheetsfrom some of these sites showsthat duringthe two or three hours
requiredto measurethe dischargeby the pointvelocitymethodthe waterlevel
couldriseorfallby significantamounts.Othercausesof uncertaintyalsoexist
for someof thesesitesand the variousauthorsnotedabovehavediscounted
someof thesesitesas unsuitable.Forexample,the RiverBlackwatersection
has flood banksat the main channeledgeswhich may affectthe degreeof
interaction.The RiVerOuse sectionhas very smallfloodplainsand may act
morelikea simplechannel. Boththe RiversTeesand Tonidgesectionshave
irregularfloodplainswith flood banks. The analysisreportedhere has been
caniedout usingall,eightsitesand onlythe Sevem,Torridgeand Trentsites.

Myers'laboratorydata were obtainedfrom a scalemodelof part of the River


Mainand the two channelgeometries are shownin Figure3. The seriesA
channelhad a total width of about 1.6m with a main channeltop width of
0.67m. The series F case studiedthe effectof removingone flood plain to
give an asymmetricchannel. In this casethe totalchannelwas about1.3m
wideand the mainchannetwas0.78mwide. ln bothcasesthe flumewasset
to a longitudinal
slopeof 1.906x1O3. Themainchannelandfloodplainswere

sR 329 07/07/93
kept smoothduringthesetwo tests and the Manning'sn value was taken as
0.01.

2.4 Applicationof straight channelmethodsto straight


channeldata
In lotal, nine methods of estirnatingdischarge were used to conpare
measureddischargeagainstpredicteddischarge.The work repodedby Wark
et al (1991)was basedon observeddischargesderivedfrom ratingcurves
developed at each of the gauging sites, which were developed by
Ramsbottom (1989)and Myers,ratherthanthe individualobservations.This
procedurewassaidto averageout measurementerrorsandseasonalchanges
in vegetationand cross-sec{ion.Between5 and 12 points were selectedon
each stagedischargecurve. This procedurehas been criticisedand in the
work reportedhere the actual observedpairs of stage and dischargewere
used. ln the cases of the Sevem and the Tees these data points were
smoothed using running averages over three consecutive data points.
Betweenfive and thirtysix data pointswere obtainedfor each site. All of the
dischargeestimationmethodswereappliedwith identicalconditions,such as
the cross-section
geometryand roughness values.

The idealisedcross-sectionsshown in Figure 3 are approximateand are


includedonlyfor inforrnation.Theyare basedon the estimatedmainchannel
sideslopes.Themainchannelsideslopesand corresponding idealiseddepth
were estimatedfollowing the proceduredescribed in Section 6.4 of the
summarylo Ackers(1991). Thesetwo paameters were only used in the
FCFAMmethodand appearin the calculations for the correctionfactors. The
basicflows were calculatedwith the actualchanneland flood plain cross-
sectionsas recommended by Ackers(1991). Ackersalso recommendsthat
the factorQr."shouldhavea lowerlimitof 0.5 and this limitwas usedin this
work.

The meanpercentageerrorsfor the variousmethods,averagedover various


subsetsof the data, are shown in Table 3. These averagesand standard
deviationswereobtainedfrom the individualresultsand not by averagingthe
meanvaluesfor eachsite. As mentionedabovethere was somedebateover
the usefulnessof some of this data and the followingdiscussionis limitedto
the three sets of data collectedfrom the Severn,Torridgeand Trent. The
meanerorc averagedover each river site are shown in Table 4.

Resultswith Authors' n vatues

The roughnessvaluesusedfor lhe floodplainsin this workwere derived,in


the main, by adjustingthem to give good agreementbetweenthe measured
depthaveragedvelocityprofilesand the velocitydistributionpredictedby the
LDM. In generalthesevaluesagreedreasonablycloselywith thoseobtained
by Ramsbottom(1989),whointegratedthe velocityprofilesto obtainthe actual
flood plain dischargeand then estimatedthe flood plain roughnessusing a
dividedchannelmethod

The two dividedchannelmethodspedormwell over all the natural(Sevem,


Torridgeand Trent, case D) data with mean errors in the range -0.3% to
0.3%.The LDMgavethe next bestaccuracy with a meanerrorof O.7oh.The
FCFAMmethodfollowsin fourthplacewitha meanerrorof -2.2%. The single
channelmethods allshowmuchgreatermeanerrorsand standarddeviations
than the four best methodsand will not be consideredfudher. The mean

sB s29 07r'0293
enors over each individualriver site in Table 4 tend to confirm these
conclusions,althoughthe rankingof the variousmethodsvariesfrom site to
site. Thesedifferencesare probablydue, in part at least,to the wide range
of channelcross sectiontype, scaleand shape. lt is worth notingthat the
slandarddeviationsreportedare partlydue to randomerrorsin the measured
datraand partlydue to systematicdifferencesbetweenthe predictionsand the
measurements.For the four best methodsthe standarddeviationsare all in
the range4A% lo 6.7%.

Resultswith Ackers n values

Ackers(1991)reportsbroadlysimilarresultswiththe Authors'nvalues. He
alsoreportedresultsobtainedwithroughnessvaluesadjustedto reducethese
errorsfor the FCFAMmethod. However,Ackersdid not cornparehis method
with existingtechniques,as this lay outwiththe scope of his proiect. The
analysisreportedabovehas been repeatedusinglhese adjustedManning's
n values.The meanerrorsfor the LDMand thetwo dividedchannelmethods
have increasedfrom between-1to and +1% to between2Y" and 5o/". The
meaneror for the FCFAMmethodis actuallyslightlyworseat -3.1%. The
standarddeviationsare similarat approximately4o/oto 5%. Figure4 shows
these errorsand standarddeviations.The errorsreportedby Ackersfor his
methodwiththesedatasetsareslightlybetterandthisis probablydue,in part,
to different approachesin applyingthe method to non-symmetricnatural
channels.The Authorshavetreatedeachfloodplainseparatelyin orderto
calculatevelocitiesand basicdischarges.Thishasthe advantageof allowing
differentroughnessvaluesto be assignedto the two flood plains. Ackers,
1991,takesaveragedfloodplainwidthsand elevationsto definean average
floodplaingeometry,,which combinedwith an averagefloodplainroughness
valueis usedto deriveaveragefloodplain velocityandbasicdischargevalues.
No further attempthas been made to identifycausesof the differences
betweenAckers(1991)analysisand that reportedhere. The differenceslie
in the subjectiveinterpretation
of the geometryof naturalcross-sectionsand
the variationof order of a few percent indicatesthe likely computational
tolerancein practicethat may be obtainedby riverengineersin practice.

Gomparisonswith the FCFPhaseA data

Ahhougha comparisonwith this data set was specificallyexcludedfrom the


analysis at the request of the NRA it is worth mentioningthat such a
comparisonhas been carriedout by the secondauthor,Wark (1993). The
FCFdata is the best dataset availablefor the developmentand verificationof
methodsfor calculatingconveyancesin straightcompoundchannels. The
detrails
confirmthe workrepodedby Ackers(1991).TheFCFAMmethodgave
a mean error in predicteddischargefor the Phase A data of -2o/"with a
standarddeviationof 3.8"h,ahhoughover only the smoothflood plain cases
this improvedto -0.3% and 1.7o/o.These resuhsare very similarto those
repodedby Ackers(1991):0.001%,SD 0.8% and 0.08%and 1.467". The
slightdifferencesin these resultsare not significantand are probablydue to
smalldifferences in the computedbedfrictionfactors.

Discussion

The above resultsfrom field data showthat of the ten methodsinvestigated


four are worthyof furtherconsideration,
namely:The LDM,DCM,DCM2and
the FCFAMmethod.lt is difficultto makedefinitivestatementsas to whichis

sR 329 07/O7l9il
bes{on the basis of these results. The lack of independentcalibrationdata
meansthat the bed frictioncalibrationshave been based on the obtained
results. lt tras been shownthat the bed roughnessvalues for naturalrivers
and floodplainsare not welldefined,the choiceof valueis ofteninfluencedby
the methodusedto conputethe channelconveyance.Thereforetheseresults
are incorrclusive and do not confirm(in the strict scientificdefinition)that any
of the four methodsgivesmoreaccurateresultsthan the others. lt has been
demonstratedthat allfour methodscan be calibratedto matchfield data.

The case of straightcompoundchannelsis misleadingwhen considering


meanderingrivers. Thereare threemainprocessesoperating:

- bed frictionlocally,
- lateralshearbetweenthe co-flowingmain channeland flood plains,
- the exchangeof momentumthroughsecondarycurrents,Knightand
Shiono,(1991).

The relativestrengthof these processesis determinedby the distributionof


depthsand roughness andthe shearlayersat the edgesof the mainchannel
can extendoverthe wholemainchannel.

The mainchanneland floodplainflowsin the meandering caseare knownto


vary in directionboth throughthe depthof the flow and alongthe channel.
The lateralshear layerswhich have such an importanteffect in the straight
channelsdo notformand the effectof lateralshearon the flowstructureand
stage-discharge relationships
is minor. Thusthe impoilantftowmechanisms
in meandering channelsare likelyto be considerablymorecomplexthan in
straightc-ompound channels,this topicis coveredmorefully in Chapter5. lt
is thereforeunlikelythat methodsfor calculatingconveyancein straight
compoundchannelswill give accuratepredictionsif appliedto meandering
compoundchannels.Ackers,(1991),froma reviewof previousresearchalso
reachedthe conclusionthat straightchannelmethodsshouldnot be applied
to meanderingcompoundchannels. The work describedin the following
sectionsconfirmsthat noneof the straightchannelmethodsconsideredabove
is able to representthe meanderingchanneldata availablefrom PhaseB of
the FCF progmmme.

2.5 Applicationof straight channelmethodsto meandering


channel data

2.5.1 The methods


The methodsused in this work are listedbelow. Theseare simple methods
whichare practicaltoapplyby hand. The LateralDistributionMethodwhich
was foundto givethe best resultsfor straightchannelshas not been included
in this assessmentsinceit is a computational
modeland the NRA R&Dproiect
put a high priodty on hand calculationmethods. The various composite
roughnessmethodshave not been includedhere since their perforrnance
againststraightchanneldatawas poor.

DCM DividedChannelMethod,using verticaldivisionlines which are


includedin the wettedperimeterof the main channelbut omitted
fromthe wettedperimeterof the floodplains. Main channelslope
usedfor mainchannelflow.

SCM SingleChannelMethod,usingmainchannelslope.

sR 329 07/07/93
@
SSGM Sum of Segments Method, using main channel slope for rnain
channel segments.

DCM2 DividedChannelMethod, usingvefticaldivisionlineswhichare not


includedin the wettedperimeterof eitherthe main channelor the
flood plains.
Mainchannelslopeusedfor mainchannelflow.

FCFAM Method devetopedby Ackers, based on Flood Channel Facitity


PhaseA data. Mainchannelslopeusedfor mainchannelflow.

HOR1 DividedChannelMethod,usinga horizontaldivision lineat bankfutl


stage. Divisionline is includedin flood plain wettedperimeterbut
not in rnainchannelwettedperimeter. Floodplain slope used for
mainchannelandfloodplainflows

HOR2 DividedChannelMethod,usinga horizontaldivision lineat bankfull


stage. Divisionline is includedin flood plain and main channel
wettedperimeters.
Floodplainslopeusedfor mainchanneland floodplainflows.

HOR3 DividedChannelMethod, usinga horizontaldivisionlineat bankfull


stage. Divisionline is includedin flood plain wettedperimeterbut
notin rnainchannelwettedperimeter.Mainchannelslopeusedfor
mainchannelflow.

HOR4 DividedChannelMethod,usinga horizontaldivision lineat bankfull


stage. Divisionline is includedin flood plain and main channel
wettedperimeters.Mainchannelslopeusedfor mainchannelflow.

The two dividedchannelmethods(DCM and DCM2);the single channel


method(ScM); the sum of segmentsmethod(SSGM)and the FCFAM
methodwereall appliedas describedin Table4.1. Theseare methodswhich
could be appliedby handto calculateflows in compoundchannels. Ackers
(1991)specificallyadvisesthat the FCFAMmethodshouldnot be appliedto
meanderingcases. Howeverit has beenincludedin the work reportedbelow
since,as a handcalculation technique,it came withinthe scopeof the NRA
proiect. The varioushorizontaldivisionline methods(HORI, HOR2,HOR3,
HOR4) are simplificationsof the methodsproposedby Toebes and Soolqy
(1967)and Smith(19771.The mainchannelandfloodplainsare considered
to be splitby a horizontallineat bankfulllevel.The regionabovethe dividing
line is includedin the floodplainarea whencalculatingthe flood plainflow.
The sinuosityof a meandering channelis the ratioof the curvilineardistance
alongthe channeltothe straightdistancebetweenthe two points.

2.5.2 Thedata,set
The SERCFCF PhaseB stagedischarge test programmeis summarizedin
Table 5. Of this data series numbers832, 833, 846 and 848 have been
excludedfromthis analysisfor the followingreasons.

832, 846 Floodplain roughenedwith rowsof isolatedbreezeblocks,special


methodsmustbe usedto accountfor the headlossesdue to these
blocks.

10 sR 329 07/07/93
838 Floodplain only partiallyroughened. The roughnesszones were
limitedto lhe'meanderbelt',creatingtwo distinctroughnessregions
on the flood plains. The methodsdescribedaboveare suitablefor
flood plainswhichare homogeneously roughened.

848 Flood plains are totally blockedby breeze block walls which run
fromthe innerbendapicesto the outeredgeof lhe floodplain. This
simulatesthe casewere developmenthas occunedoverthe whole
flood plain. Again the simplemethodsused here are not suitable
for this geometry.

The series821, 826, 831, 884, 839, 843 andB.47from PhaseB of the FCF
wereall analyzedusingthe methodsdescribedabove. In total 107data points
were used in this analysis. The full detailsof the experimentsare given in
Chapter3. The bed fri-ctiontermsfor the varioustestswere calculatedusing
a npdified smoothlaw fgr the smoothcases and the Ackers rod roughness
method for the roughenedcases. The full details of these are given in
Chapter3.

2.5.3 Results
The mean errorsand standarddeviationsin the mean errorsfor the various
methodsare listed in Table 6. The resuftsdifferedconsiderablydepending
whetherthefloodplainswere roughenedor not and so meanerors are given
over the smoothdata,the rougheneddata and over all the smoothand rough
data.

Table 4.3 showsthat for the whole data set the mean errorsfor the various
methodsvaryfrom7.3"/"lo7O.1o/". All the methodsover-predictdischargeby
significantamounts. The correspondingstandarddeviationsvary between
16.8%and56.7%showingthatthe erors varyby verylargemarginsaboutthe
meanvalues.

tt is worth lookingmorecloselyat the resultsaveragedover the smoothand


roughdata sets. The meanerrorfor the fourthhorizontaldivisionline method
(HOH4)taken over all lhe data is 7.37" arld so this methodwouldappearto
give the best results. However,when the mean error is calculatedover the
smooth and rough data sets the mean errors are 19.5Y" and -19.8%
respectively.Thus the relativelylow meanenor achievedby consideringthe
whole data set is actuallythe resultof large positiveerrors for the smooth
casesand largenegativeenorsforthe roughcases. This wide bandof enors
is highlightedby the largevaluesof the standarddeviations.

The resuttsdiscussedabove show that the simple methodsdevelopedfor


straightcompoundchannelsare likelyto give rise to largeerrorsin estimated
dischargesif appliedto meanderingcompoundchannels.

The rangeof erors to be expectedwillvary with the followingparameters:

1 sinuosity
2 floodplainwidth/ mainchannelwidth
3 floodplainroughness/ mainchannelroughness
4 floodplaindepth/ mainchanneldepth

11 sF 329 07t07/93
For cases similar to the Phase B geometries considered the errors in
calculated discharges may be as large as 100%. Hence a different method is
required to calculate the discharge in meandering compound channels.

The four methods based on a simple two way division with a horizontal line at
bankfull stage appear to pedorm slightly better overall than the other methods.
This suggests that horizontal divisions are most appropriate for meandering
channels. In straight compound channels the best divisions are based on
verticaldivisionsat the edges of lhe rnain channel.

2.6 Summary
This chapter repods the results obtained by applying various methods
developedfor dischargeestirnation in straightmnpound channelsto fielddata
collected on straight reaches and to meanderinglaboratorydata. The
applicationto straightfielddatawasinconclusive, TheLDM,DCM,DCM2and
FCFAMmethodsgivingsimilarresufts,dependingon the roughnessvalues
used. Applicationsof thesemethodsto laboratorydata,repofiedelsewhere,
showedthatthe FCFAMmethodandthe LDMmethodscanpredictdischarges
in laboratorycompoundchannelsmoreaccuratelythan the other methods.In
practicaldesignexercises the FCFAMmethodshouldbe usedsinceit is based
on a carefulanalysisof the bestavailablelaboratory data. lt will alsogivea
slightlymoreconservative solutionthanthe otherfour methods.

Of the simple methodsappliedto the meanderinglaboratorydata the


hotizontaldivisionmethodsgave marginallymore accuratepredictions.In
generalstraightchannelmelhodsare not appropriatefor the analysisof
meandering mmpoundchannels.This confirmedthat the development of a
newprocedure for dischargeestimation
in meandering compoundchannelsis
wotthwhileand the nextchapterliststhe detailsof laboratory
datawhichwas
available.

3 Laboratory investigations into meandering


overbank flow
3.1 Background
In the previouschaptermethodsof calculatingdischargein straightcompound
channelswere appliedto some of the data collectedduringphase B of the
FCF work. The poor resultsobtaineddemonstratedthat straightchannel
methodsare notappropriatefor usewithmeanderingchannels.The nexttask,
therefore,was to developa new procedurefor the estimationof dischargein
meanderingchannels. In order to carry out this work informationon the
behaviourof both inbankand out of bank meanderingchannelswas required.
The following laboratoryand field experimentswere identified from the
literature.Boththe SERCFCFdata and the Aberdeendata were usedin the
developmentof the procedures,other sets of laboratorydata were used to
verifythe methods.The development and verificationof the procedureswere
carriedout separately.

The rnaincharacteristics
and findingsof eachof the laboratoryinvestigation
are detailedbelow.

12 sR 329 07/07/93
3.2 Laboratory data

3.2.1 Phase B SERC FCF


The FCFat HR Wallingfordprovidedthe primaryset of data usedin this work.
The detailsof the experipentsand the data collectedwere madeavailableto
the authorsby the variousuniversityresearchersinvolvedin the experiments.
This projectput a high priorityon stagedischargeand flow distributiondata
and rpst effod was put into obtaining details and data from these
expedments.

The PhaseB geometrieswereconstructedfromconcretewith a smoothmodar


finish.Thebasicsudaceroughness sizewasidenticalwithinthe mainchanne-l
and on the floodplains. In generalthefullfloodplainwidthof lom was used
althoughtests with a reducedflood plain width were also undertaken. Two
basic meandergeometrieswith differentsinuositieswere constructedfor the
Phase B experiments. The first geometryhad a sinuosity of 1.37 (60"
meander)and includedfour completemeanders.The secondgeometryhad
a sinuosityof 2.O4(110'meander)and includedfour and a half meander
wavelengths, see Figures5, 6, 7 and 8 and Plates1 and 2.

The designlongitudinal
slopeof the flood plainwas 1.0 x 1O3. The actual
longitudinal
slopeof thqfloodplainsurfacefor the firstchannelgeometrywas
0.996x 103and 1.021x10-3 for the second.

Two mainchannelcross-section geometrieswereusedwiththe 60'meander


geometry.Thefirstwasa simpletrapeziumwitha basewidthof 0.90m, side
slopesof 4Soanda depthof 0.15msee Figures5 and 6 and Plate1. The
secondwas an approximationto a typicalnaturalgeometry,reproducingthe
relativelydeep pools which form on the outsidesof the bends and the
symmetricgeometryat thecross-over sections.Onlythe naiuralmainchannel
cross-sectional
geometrywas used with the 11Oomeandergeometry. See
Figures7 and 8 and Plates2 and 3.

The fullfloodplainwidthof 10.0mwasusedfor mostexperiments, but stage-


dischargeswere also measuredwith the flood plain edgestangentialto the
meanderbends. This was carriedout for both the meandergeometrieswith
naturalmainchannelcrosssections.The flood plainsideswere slopedat 45o
(1:1)in thesecases.

In generaltheexperiments of the floodplains


includedno artificialroughening
or mainchannel,the sudacesbeingleft as trowelledmortar. However,a few
expedmentsdid includeartificialrougheningon the flood plainsby utilising
veilicaldowelrodsplacedin differinggeometricpatterns.The rod roughened
tests had the rods placed over the whole flood plain, while the padially
roughenedtests had only the innerflood plain or meanderbelt coveredwith
rods,see Plates4 and 5.

Additionaltests were also carriedout by introducinga blockageto the flood


plain flow. Concreteblocks aligned in the flow direction were used to
approximate the behaviourof bridgepierson the floodplain,Plate6. In the
case of the 110'geometrythese blockwere also used to constructwalls
acrossthe full widthsof the flood plainsto the inner bend apices. This
simulatedthe effectof completedevelopmenton the flood plain,whichwould
restrictthe flow to the rnainchannel. This data has not been used since no
reliablemethodof calculating the flowresistanceof the blocksor wallsexists.

13 sR 329 07/07/93
The srnoothand rod roughenedcasescorrespondto calibrationtestscanied
out during PhaseA of the FCF work into straightcompoundchannetsand
Ackers (1991) providesaccuratebed friction calibrationsfor these two
conditions.

The tesdprogrammeincludedinvestigationof in-bankand out of bankflows.


Measurementsof stage,velocity(magnitudeand direction),boundaryshear
stress,turbulenceand dispersionwereall includedin the programme.

The procedurefollowedwhen carryingout the stage dischargeexperiments


was as follows. The pumpswere set to give the requireddischarge,which
was measuredusing orificeplatesin the supplypipes. The tail gateswere
then adjustedto give a watersurfaceslope approxirnately equalto the flood
plainbed slope. water surfaceslopeswere measuredusingtappingpointsat
intervalsalong the centrelineof the flume. The water level at the tapping
points was measuredin stillingpots attachedto the side of the flume with
pointergauges. A least squaresprocedurewas then usedto determinethe
slopeof the measuredwatersudace. In orderio obtainstage-discharge data
for uniformflow readingswouldbe taken for severaltail gate settingsat the
sameflow. The stageand watersudaceslopevaluesfor eachof thesegate
settingswouldbe slightlydifferent.Linearinterpolationbetweentheseslope
and stage valueswas then carriedout to providestage values'corrected'to
the actualbed slope. considerablecarewastakenduringthe projectto check
the measuredstagedischargescollectedduringphase B. The interpolated
stagesfor eachcasewere recalculated and a few minorerrorsand anomalies
corrected.

For the other measurements the flow would be set and then the tail gate
adjusteduntil the water sudaceslopewas approximately equal to the bed
slope(within!2%). Pointvelocities weremeasuredusinga two stagemethod.
Miniaturepropellermeterswereusedto measurepointvelocities.ln orderto
give accurateresultsthese instrumentsneed to be alignedwith the local
velocitydirection. First the local flow directionwas measuredusing vanes
mountedon rotary potentiometers.When positionedin the flow the vane
alignedits self withthe localhorizontalvelocitydirection.The resultingvoltage
fromthe potentiometerwas relatedto the anglebetweenthe assumeddatum
and the velocityvector. These angle readingswere storedand used in a
subsequentexperimentto alignthe propellerwiththe flow direction. Thusfor
any flow conditiontwo experimentswere carriedout, one to measurethe
directionsof the velocitiesand a secondto measuretheir magnitudes.

Separatemeasurements of anglesand velocitieswere made in the rnain


channeland flood plain areas. For the main channelmeasurements were
takenon elevencrosssectionsalonga quafierof a wave lengthof the 6d
channeland on foudeencross sectionsover half a wave lengithof the 1ld
channel. Thesemain channelsectionsweretakenperpendicularto the main
channelcentreline
and extended3O0mmon to thefloodplainson eithersides.
Measurements were madeon a grid with horizontalspacingsof 150mmor
50mmand vedicalspacingof 15mm. On the floodplainmeasurements were
madeat 13 (60")or 11 (110o)traversescoveringhalfa wavelengrth.Readings
weretakenat spacingsof 0.5mlaterallyand 1Ommvertically.Onlythe rnain
channelvelocitydata wereusedin this project.

Boundaryshearstresseswere measuredin both geometriesusinga Preston


tube on the smoothsurfaces.The shearstressmeasurementswere madeat

14 sR 329 07/O7l93
the same sectionsin the main channeland on the flood plain as the point
velocitydata. This informationwas not useddirectlyduringthe projectbut the
resultsand analysisreportedby Knightet al (1992)were used.

A two componentLaserDopplerAnemomelerwas usedto measureturbulence


data across the flood plain and main channelat one bend apex. Detailed
watersurfacelevelswere measuredusinga churchill probe both in the main
channel and across the flood plain over one meanderwave length. Dye
dispersiontestswerecaniedout at inbankflow conditions.Theseexperiments
involvediniectingdye at variouspointsacrossthe channeland rnonitoringthe
concentrationsat selected positions downstream. Flow visualisation
experimentsinvolvedphotographing the npvementsof eitherinjecteddye or
floatingbodies. The visualisationexperimentsprovidequalitativeindications
of the complexityof the flow structurespresent. Becausethis projectput a
high priorityon the estimationof both the total dischargeand the distribution
of dischargenone of the turbulence, water surfacelevel, dispersionor flow
visualisationinformationwas used in this project.

A full listingof the experiments


undertakenin the PhaseB tests is givenin
Table 7 and Table 5 liststhe stage dischargeexperimentscarriedout. The
test numberingsystem was devised by researchersfrom the Universityof
Bristol. Tests one to nineteenwere carried out during phase A of the FCF
work into compoundchannels. To ease data handllngthe stage discharge
data availablefrom other sourceswere also assignedserial numbers. The
Abeideenstagedischarge datawereassignednumbersbetween100and 199.
The Vicksburgdata were assignednumbersbetween2OOand 299. Kiely's
one set of stagedischargedata was giventhe test number3O1and Soo(y's
data were assignednumbers400 to 499.

3.2.2 Aberdeen
The Aberdeenflume, Willettset al (1991)was constructed as a scalemodel
of the SERCFCF. Althoughdue to spacelimitationsa smallerverticalscale
was used leadingto the modelhavingdistortedchannelcross-sectionswhen
comparedto the equivalentgeometryon the SERC FCF. The meandering
channelswereformedin expandedpolystyreneand painted. In allcases the
flood plain width was 1.20 m and the flood plain and main channel
roughnesseswere identical. Experimentswere conductedwith four different
channefsinuosities, vi2..1.00,1.21, 1.4Oand 2.04. An identicaltrapezoidal
cross-sectionfor the main channelwas used for all sinuositieswith a base
widthof 0.139m,a depthof 0.050mand side slopesof 7O.T (0-35:1).

Two additionalexperiments wereconductedwith'natural'mainchannelcross


sections. Thesewere createdby infillingthe trapezoidalchannelsto a depth
of 20 mm with bakelitepowder. This moveablebed was subjectedto banldull
flow until a stable bed topographyhad evolved,which was then fixed and
painted. The water temperaturewas 14oC+1oC in all experiments.

The experimentalconditionsare summarizedin Table 8. Stage-discharge


measurements weretakenfor all conditionsexceptfor inbankflows in the 2.04
sinuositychannelwith the naturalcross-section.Figures9 and 10 showthe
variousplanand crosssectiongeomelries.

Willettset al (1991)reportsome preliminaryobservationsand conclusions.


The study includedmeasurementof stage discharges,point velocitieswithin

15 sR 329 07/07/93
the rnainchanneland watersudacelevelsovera wavelengthof the meander
pattern. The objectsof the investigation
were:

1) To identify the flow structuresassociatedwith channel flood plain


interaction.

2l To explorethe dependencyof thesestructureson channelcrosssection


shape.

3) To determinewhethermain channelsinuosityand cross-sectionshape


havesignificantinfluenceon stagedischargerelationships.

The stage dischargemeasurements showthat the dischargecapacityof th6


floodwayreducesas the sinuosityof the mainchannelincreases.Fora given
slnuositythe mainchannelcrosssectionhas a strongeffecton the capacity.
The 'natural'caseshad largertotaldischargesthan the equivalenttrapezoida
channelsat high sflages,even thoughthe trapezoidalchannelshad larger
cross sectionalareas. An explanationof this unexpectedphenomenonwas
indicatedby dye dispersiontests. The trapezoidalchannelsexhibitedfar
strongersecondarycurrentsand moreinteraction betweenmainchanneland
flood plainflowsthan the naluralcases. Thuschannelcross sectionhas a
strongeffecton the conveyancecapacityof a floodway.

A completeset of stagedischarge datahasbeensuppliedby Willetts(1991).


Onlythe datacollectedin thetrapezoidal
mainchannelandmeandering cases
havebeenusedherehowever.The inbankstagedischarges collectedin the
straightchannelwereusedto calibratea modifiedsmoothfrictionlaw.

3.2.3 US Army Vicksburg


A series of stagedischargeexperimentswere conductedon compound
meandering channelsby the UnitedStatesArmyCorpsof Engineers(1956)
at the WaterwaysExperimentStationin Vicksburg,Mississippi,USA. The
mainpurposeof these experimentswas to determinethe effectson floodway
capacityof: radiusof curuatureof bends; sinuosityof mainchannel; depthof
overbankflow; ratio of overbankarea to main channelarea and flood plain
roughness.Two basicsizesof mainchannelcrossseciionwere constructed.
The smallerchannelwas constructedwith a trapezoidalcrosssectionof base
widthone foot (0.305m)and was 0.5 feet (0.152m)deep. Stagedischarges
were measuredfor ten basicconditions.The resultsobtainedwith this one
foot wide channelwereinconclusive anda furtherset of testswerecarriedout
with a largermainchannel.

Themainchannelcross-section wastrapezoidal in allcases, withsideslopes


of 63.4". The bottomwidthwas two feet (0.610m).The testswereconducted
in a flume 30.5 m long and 9.2 m wide. Main channelswith three planform
geometriesweremouldedin sandand stabilizedwith a concreteveneer. The
flood plain width was variedby installingtemporarybrick walls. The basic
flood plain roughness was plain brushed concrete. Two additional
roughnesseswere obtainedby coveringthe sudace with expandedmetal
grating,laid with the openingsorientedparalleland normal to the flow
direction.

The three differentmeanderplanformswere constructedwith this 2 ft wide


channel,all witharcsof circlesconnectedtangentially,withno straightreaches
betweenthem. The meanderwavelengthwas heldconstantat twentyfour

16 sR 329 07/O7l93
feet (7.315m).Threeand a hafffull wave lengthswere constructed for the
three sinuosities. The valley slope was 1.0x10'3in all cases. For each
conditionthe dischargewas measuredat banHulland threeoverbankstages.
The main conclusionsbf the studywere:

a) Where the main channel is narow (and small) comparedto the flood
plain, the effect of channelsinuosityon the total dischargecapacityis
small.

b) The effect of increasedrnain channelsinuosityis to reducethe total


dischargecapacity.

c) Whenthe floodplainis morethanthreetimesthe widthof the meandet


belt the effect of channelsinuosityon the total dischargecapacityis
small.

d) The effect of incieasedflood plain roughnessis to reduce the total


dischargecapacity.

In all stage-discharges
weremeasuredfor elevenseparateconditionswith the
2 ft widechannelbut it wasfoundthatthe roughened floodplaincasescould
not be usedsincethe quotedManning's n valuesof 0.025and 0.035couldnot
be verified. The snpoth surfaceof the Vicksburgflume was similarto the
SERC FCF and both facilities were constructedat similar scales. The
Vicksburgflume had a quotedManning'sn of 0.012but this could not be
verifiedand it was decidedto modelbed frictionusingthe modifiedsmooth
turbulentlaw developedfor the SERCFCF. Thusonlythreeof the measured
stage-discharge curuescouldbe used.The experimental mnditionsforthe 2ft
widechannelare listedin Table9 and Figure11.

3.2.4 Toebesand Sooky


Toebesand Soolry(1907)and Soolqy(1964)carriedout a laboratorystudyof
overbankflow with a meanderingchannel,varyingonly the main channel
depth. Measurementsof stagedischargeas well as data on water sudace
and velocityvariationsacrossthe channeland flood plains were recorded.
The nrodelledgeometrycovered5.5 meanderwave lengths. The sinuous
main channelwas constrirctedwith a sinusoidalplan form and a rectangular
crosssectionin a flume7.3mlongand 1.18mwide,Figure12. Two separate
channel depths and seven longitudinalslopes were tested to give eleven
individualstagedischargecases. Calibrationtests were also canied out in
straightrectangularchannelsand this providedthe necessaryinformationto
calibratea modifiedsmoothlaw for Sooky'sflume.

Sookyanalyzedthesestagedischargesbasedon divisionof the crosssection


into two zonesby a horizontalline at bankfull. He assumedthat flows in both
regionsare controlledby the longitudinalvalleyslope. Applyingbasicfrictional
lossesand calculatingthe dischargesin these two regionsseparatelyover-
predicteddischargeand so did not accountfor all energylosses. In orderto
accountfor these extra energylossesSooky introducedan extra length of
wettedperimeter(T)to boththe mainchannelandfloodplaincalculations.He
usedhis laboratorydatato backcalculatethe valuesof T requiredto givezero

17 sR 329 07/07/93
@
enor in the predicted discharges. T was found to be a complicated function
of:

a) overbankflow depth
b) meanvelocitiesin the two zones
c) longitudinalslope

On the basisof this analysisit wasconcludedthat the additionalenergylosses


(otherthan bed friction)introduced in overbankflow in meandering channels
willdependon theseparameters.In additionthe followingconclusions were
alsodrawn:

1) The additionallossesincrease(from zero) with over bank stage up to


some maximum.As depthincreasesbeyondthis pointthe extralosses
then reduce.

2l The deeperand nanowerthe main channelthe smallerare the extra


energylosses.

3) For the purposesof calculatingdischargein meanderingcompound


channels,cross sectionsare best dividedby a horizontaldivisionat
bankfull.

The measuredvelocitiesalsoprovidedusefulinformationon the flow structure


withinthe channel. lt is well knownthat flow aroundchannelbendsinduces
spiralcurrentsand superelevation. Inbanksecondarycurrentsare knownto
rotatewith the sudacecurrentsdirectedtowardsthe outsideof the bend.
Duringoverbankflow Soolqyobseryedthatthe secondarycurrentsrotatein the
oppositesense, ie the surfacecurrentsare directedtowardsthe insideof the
bend.Thisobservation hasbeenconfirmed (SERCFCF,
by otherresearchers
Stein et al, 1988 and 1989). In additionthe velocitieswere integratedto
providedischargevaluesin the variousregionsof the channel.

3.2.5 Smith
Smith(1978)has publisheddetailsof a laboratoryinvestigationintooverbank
meanderingflow. He carriedout stagedischargeexperimentsfor threecases
includinga straightcompoundchannel,a meandering compoundchanneland
for the floodplainalone. The flumewas set at a tongitudinalslopeof 1x103
and in bothcasesthe rnainchannelwastrapezoidalwith a top widthof O.27m
and bankfulldepth0.076m. The modelchannelhad 7 meanderwavelengths
and all three caseswereconstructedof trowelledmortarin a flume 24m long
by 1.2mwide. The meanderingplanformwas constructedwith a sinuosilyof
1.172andfilledthe full widthof the flume. Smithcarriedout someanalyses
of the meanderingcase usingthe straightchanneldividedchannelmethod
(DCM2). He concludedthat straightchannelmethodsare inappropriatefor
calculatingthe dischargein meanderingcompoundchannels. He canied out
somedye injectionlests to investigateflow patternsand foundthat the flow in
the rnainchannelvariedalongthewavelength,spillingoutof the channelonto
the flood plainand back. The flow in the channelwas obseryedto be lowest
at the cross-overreach,halfway betweenmeanderbends. At deepovelbank
stagesthe valley flow was observedto pass over the main channel. A
separation zoneoccurredand a spiraleddyin the mainchannelwasinduced.

18 sR 329 o7l07/93
The mainconclusions
of Smith'sworkwere:

1) Straightchannelmethodsare inappropriatefor calculatingdischargein


meandering compoundchannels.

2') For meandedngoverbankdischargesthe main channeland flood plain


flows interact. This interaclionhas a strong effect on the discharge
capacityand variegstronglywith stage.

3) The flow in the mainchannelvaries


alonga meanderwavelengrth
and is
minimumat some pointbetweenbends.

Smith provided a bed friction calibrationfor Manning'sn of about 0.01-,


althoughthis did appearto vary for the variouscases. Duringlater work on
verifyingthe authorsmethodthis dataset was foundto behavedifferenttyfrom
the olherdataavailable.Thiswasattributedto the poorbed frictioncalibration
and it was decidedthat Smith'sdata is unreliableand so shouldnol be used
in any numericalwork.

3.2.6 Jamesand Brown


James and Brown (19771carried out measurementsto determinethe
geometricparameters whichinfluence floodplainflow in a tihingflume26.8m
long by 1.5m wide. Out of fourteentests they conductednine with straight
channels,three with singleskewedcrossovers,one with two cross-oversand
one withthreecross-overs.Onlythe lastcaseevenapproachesa meandering
geometrybut is ratherunrealisticwhencomparedto typicalnaturalplanforms.
The bendradiuswastoo smallandthe lengthof straightcross-overtoolarge.
The straightand meanderingchannelstage-discharge data was anatyzedin
termsof Manning'sn values. On the basisof this theyconcludedthat 'The
resistancefactor increasedas the crossoveror meanderlengfihdecreased'.
This is equivalentto sayingthat the @nveyancecapacitywas reducedas the
sinuosityincreased.

This data set was not usedin any comparisonsbecauseof the relativelypoor
meanderinggeometryand the lack of adequatebed frictioncalibrationdata.

3.2.7 Kiely
Kielyet al (1989and 1990)caniedout experimentalwo* intoflowsin straight
and meanderingcompound channels. Discharges,point velocities and
tutbufencemeasurements were rnadein a 14.4mlongby 1.2 m wideflume.
A straight, single meanderwave length and multiplemeander (4.5 wave
lengths)caseswere investigated, see Figure13. The flumewas hydraulically
smoothwith a test sectionconstructedof glass and perspexfor use with a
singlecomponentLaserDopplersystem. The mainchannelwasrectangular
in all three cases and the ftume was set at a valley slope of 1.Ox1O€.
McKeoghand Kiely(1989)providea modifiedsnroothlaw whichgivesthe bed
frictionin this flume. The lasersystemwas usedto investigatedetailedflow
structuresin both the main channeland the flood plain. Kiely identifiedthe
followingmechanismsin overbankmeanderingflow.

1) Secondarycurrentsin the main channelduring overbankflow were


observedto rotatein the oppositedirectionto thoseseenduringinbank
flow. A detailedexaminationof the secondarycurrentpatternssuggests
that the mechanismsproducingthesedifferentpatternsare bothpresent
during over bank flow but that the curvatureinducedcurrentsare less

19 sR 3an ozozs3
intense and become nullified. The energy losses due to secondary
currents during overbankflow are greaterthan the losses during inbank
flow.

2l Velocities within the rnain channel were generally observed to follow the
direction of the main channel side walls. The direction of velocities at
points over lhe main channel and above bankfull level were observed to
vary with level. Above bank level the direction of flow changes from
being paralleltothe main channelat bankfulltobeingalmost parallelto
the flood plain, close to the water level. This change in lhe direction of
local velocity lhrough the water column indicates the presence of a
hodzontalshear layer between lhe rnain channeland flood plain flows.
rl

3) At the crossover reachesihe water on the flood plain^isobserved to pass


into and across the main channel. Thus fluid from'the left hand flood
plain crosses the main channel and ends up on the right hand flood plain.
As the flow crosses into the channelthe depth increasesand as it passes
out onlo the flood plain the depth decreases. This expansion and
contraction of the flow area is known to induce energy losses in
analogoussituations.

4l Velocitieswere seen to vary stronglyacross the flood plain. Outwiththe


meander zone the velocities were approximatelyuniform. Within the
meanderzone an area of reducedvelocitywas observed. lt was felt that
this was caused by the interactionof the main channel and flood plain
flows, with relativelylow velocity fluid leaving the channel at the cross
over reachesand passing down the flood plain.

The multiple meander data has been used by the authorsto test and verify
various methods of calculating the conveyance capacity of meandering
compoundchannels.

3.2.8 Steinand Rouve


Steinand Rouve(1988,1989)haveinvestigated the detailedflow structures
present over one rneander wave length for overtank flow conditions.
Sophisticatedlaserdoppleranernometrywas usedto measureallthree point
velocitycomponentswithinthe flow for one water leveland discharge. The
meanderingchannelwas constructedin a flume 15.0mlong by 3.0m wide.
The rnainchannelwas rectangularwith a widthof 0.4mand a bankfulldepth
of 0.1m. The preliminary
resultspresentedallowedthe followingconclusions
to be drawn.

1) Secondarycurrentsin the mainchannelrotatein the oppositedirection


to thosefor inbankflow.

2) Fluid'wellingout of the mainchannelslowsthe dischargeon the flood


plain.

3) shearlayerexistsbetweenthe lowerand upperpartsof the


A horizontal
mainchannel.

20 sR 329 07/07/93
3.3 GeneralComments
Experimentalwork on flows in meanderingchannels during overbank
conditionshas been identifiedfrom the literature. The investigatorsand the
main characteristics
of their experimentsare summarizedbelow.

SERCFCF PhaseB Muttiple meander, two sinuosities,two cross


sections, lwo flood plain roughnesses, stage
discharges,velocity,water surfaceand bed shear
stresses.

Willettset al Multiple meander, three sinuosities,two cross


sections,stagedischarges,watersurfacelevelsand
velocities.

US Army Vicksburg Multiplemeander,threesinuosities,


threefloodplain
, roughnesses,stagedischarges.

Kiely Singleand multiplerneander,one sinuosity,stage


discharges,
velocity,water sudaceand turbulence
measurements.

Toebesand Sooky Multiplemeander,onesinuosity,two crosssections,


seven slopes, stage discharges,water surface
levelsand velocities.

Jamesand Brown Multiple meander,stage discharges and velocities.

Smith Multiple meander,stage discharges.

Steinand Rouve Single meander,water sudace levels,velocitiesand


turbulencemeasurements.

The key physicaldimensionsof the test channelsare given in Table 10.


Table11 showsthe relationships betweenthe key geometricparametersfor
the laboratoryflumetests. Variousauthorshavepublisheddetailsof empirical
equationsderived by regressionanalysescarried out on natural meander
patterns. The exact equationsvary from authorto authorbut in generalit is
possibleto say that in natural,fully developed,meanderbends: the wave
lengthis approximately ten times the channelwidth; the channelwidth is
approximately tentimesthechannelbankfulldepthand the radiusof curuature
of the bendsis betweentwo to threetimesthe channelwidth.

A studyof Table11 showsthat mostlaboratorystudieshavebeen caniedout


with mainchannelaspectratios(B/h)whichlie between3.5 and 5.0. Onlythe
SERC FCF geometrieshave channelcross sectionswhich approximateto
naturalcrosssectionswithan aspectratioof 8.0. Toebesand Soolqy(1964
and the workcaried out at Vicksburg(1956)demonstrated that mainchannel
cross sectionshape can have a strong effect on the dischargecapacityof
meandering channelsduringoverbankflow. Theseobseryationshavebeen
confirmedby Willettset al (1991).

Most of the investigations have used meanderswith wavelengthto channel


width ratioswhich are closeto the naturalratio of 10. Only the meander
investigated by Jamesand Brown(19771witha value of about 33 is totally
unrealisticin termsof this ratio. The final geometricratiobetweenthe bend

21 sR 329 07/O7l93
radiusand channelwidth generallyfalls within the naturalrange of 2 to 3.
soolry, James and, Brown and smith all constructedchannetswith low
sinuosityand this producedy'B ratiosof about 4.0. Howeverthis is not a
serbus deviationsincethe relationships
for naturalchannelswerederivedfor
fairlysinuouschannels.

It has been demonstratedthat all three of these geometricratios effect the


stage dischargecapacityduringoverbankflow. Since only the large scale
expedments carriedout in the SERCFCFsatisfiedallthreerelationships it is
likelythatthe flow patternsand stagedischargerelationshipsfor the FCF will
be closer to those observedin naturethan for the other experimentaldata
collectedin small scale models. Most investigatorsidentifiedwell-defined
structureswithin the flow including: secondarycurrents within the rnaih
channeland bulk exchangesof flow betweenmainchanneland flood plain.
Figure14 showsthe flow processestakingplace duringoverbankflow in
meandering channels.

3.4 Bed friction


ln later chapters stage discharge data from these laboratory studies is used
to develop and verify methods of estimating discharge in meandering
channels. In order to carry out this work it is necessaryto calculateenergy
losses due to bed friction. All of the laboratory models were constructed with
hydraulicallysmooth sudaces. One of the conclusionsarisingfrom the earlier
work by Ackers (1991) is that the bed frictionin hydraulicallysmoothconditions
should be obtained from a smooth turbulent expression. This expression
correctly predicts the effect of viscosity on friction factor. In any given case it
may be appropriateto derive a modifiedversion of the smooth turbulent law
which fits the data betterthan the generalversionquoted in the literature.This
approach to determiningbed friction has been followed and for each of the
various data sets a modifiedsmooth law has been obtained.

The original referenceseither gave the appropriatemodified smooth law or


stage-dischargemeasurementsin straightsimple channelswhich could be

used to calibratethe constantvalues in the smooth law. The generalform of


the modified smooth law is:

1lt1n = A log (Re fr21+ B (1)

Where Re is the Reyrnoldsnumber of the flow defined by:

Re= 4RV

in which

V is the flow velocity,


R is the hydraulicradius,and
v is the kinematicviscosityof the water.

The kinematicviscositycan be calculated by the


fromrecordedtemperatures
equation

v = ( 1 . 7 4 1- 0 . 0 4 9 9 T + 0 . 0 0 0 6 6 T
x t2O1 ' 6 (21

sB 329 07^r7l93
in which T is the temperature in oC.

The values of the constants A and B derived for each of the data sources are
listed below.

Data source A B Commenls

Vicksburg 2.O2-1.38 The SERCFCFsmoothlaw was used


Toebesand Sooky 0.68 2.42 Values calibrated to given stage-
discharges
Kiely 2.10 -1.56 Valuesprovidedby Kiely
Serc FCF 2.O2 -1.38 Values provided by Ackers
Aberdeen 2.48 -2.91 Values calibrated to given stage-
discharges

Someof the experiments carriedout on the SERCFCF involvedroughening


the flood plains with verticalrods extendingthroughthe full depth of water.
The patternof rodsconsistedof a triangulardistributionof angle6f. Thiswas
designed to have a density of 12 rods per square metre. Under these
conditionsthe resistanceto flow is madeup of drag of the rodsand the shear
force at the channelboundaries.Ackershas analyzedsomecalibrationlests
carriedout during phaseA of the SERCFCF work and developeda method
of obtainingthe totalfrictionfactordue to rod roughness.

He assumedthat the rod drag and bed friction can be treated separately,
accountingfor the blockageeffectof the rodson the meanvelocity. The drag
of the rods is relatedto the squareof the mean flow velocitypast the rods.
Ackerscalibratedan expression for the dragcoefficient
whichdependson the
ratioof rod diameterto flow depth. The expressionis quitecomplexand in
orderto obtainfrictionfactorvaluesfor a specificdepthiterationis required.
The equationsand datafor the methodare givenin Appendix5.

3.5 Summary
This chapterpresentsthe resultsof a literaturesearchin to overbankflow in
meanderingcompoundchannels. The main purposesof this reviewwere:

a) To identify laboratorydata to use in developingand verifyinga new


procedure for dischargeestimationin overbank flow in meandering
channels.

b) To sumrnarizethe current state of knowledgeon the detailed flow


structurespresentduringoverbankmeanderingflow and to gaugethe
effectthese mighthaveon the dischargecapacity.

Eightlaboratoryinvestigations
wereidentified,includingfhe SERC FCF. The
two most nrodemand extensivedata sets (SERCFGFand Aberdeen)were
consideredto representthe best qualitydata availableand it was decidedto
usethesetwo setsin developing a newprocedure.Threeotherinvestigations
(Vicksburg,Kiely and Sooky)were deemedappropriateto use in verification
of the new procedure.

23 sR 329 07/07/93
The internalstructureof currentsduring overbankflows has been found to be
highly complex see Figure 14. The most impodant observationsare:

1) The longitudinalvelocitiesbelow bankfulltend to followthe main channel


side walls while the floodplain velocities are generally in the valley
direction. Thus the floodplainflows pass over the main channel and
induce a horizontalshear layer.

2l The energy loss due to secondarycurrentsin the main channel is greater


than for an equivalent simple channel and the currents rotate in the
opposite sense compared to inbank flows.

3) Fluidpassesfrom the main channelonto the flood plainand back into the
main channel in the followingmeander bend. Hence the proportionof
discharge passed by the main channel and flood plain varies along a
meander wavelength. These bulk exchangesof fluid between dow and
fast moving regions of flow introduce extra flow resistance.

4l Flows on the flood plain outwiththe meanderbelt are usuallyfaster than


those within the meander belt. lt would appear that the extra flow
resistance induced by the meandering main channel has a relatively
small effect on the outer flood plain.

The followingchapter examinesthe topic of energy losses during inbank flow


in channel bends and meanders. Althoughthe main thrust of the projectwas
to deal with overbank flow it was felt that to be important to examine the
inbank case as well. The transitionfrom inbankto overbankflow is often the
critical aspect of practical problems and it would be impossibleto study it
properly with out some knowledgeof the characteristicsof inbank flow.

4 lnbank meandering flow


4.1 Background
It has been recognisedfor many yearsthat meanderingcan increasethe
effectiveresistanceof channelssignificantlyfor inbankflows. Laboratoryand
theoreticalinvestigations
in to thecharacteristicsof flowin channelbends
have
shownthat complicatedflow structuresform and that thesecan havea large
effecton the dischargecapacityof the channel.

Secondaryor spiral currents are induced by differencesin centripeta!


accelerations actingon a verticalcolumn. The longitudinalvelocitiesare
greaterfor particlescloseto the watersurface.This impliesthat the lateral
forceson the watercolumnare not in equilibriumand so lateralmovements
of particlesare induced. The cunentsmovetowardsthe outsideof the bend
at the water sudaceand towardsthe insideat the channeibed. These
secondarycurrentsalso affectthe water surfaceprofileacrossthe channel.
In straightchannelsthe watersuface is uniformbut in bendsthe surfaceis
displacedand slopesdown from the outsideto the insideof the bend to
balance the non-unifonnlateral pressuredistributionintroducedby the
secondarycurrents. These secondarycurrents affect the distributionof
longitudinalflow within the channelcross sectionby advectingthe faster
movingfluid towardsthe outer bank. The flow distributionand associated
longitudinalbedshearstressesbecomesnon-uniform acrossthe channel.The

24 sR 329 07/O7l93
secondarycurrentsalso inducea lateralcomponent of bed shearstresswhich
obviouslyincreasesthe total shearstressactingon the bed.

The strengthof these secondarycunentsis knownto vary along a bend. In


the case wherea singlebend has straightreachesboth up and down stream
then it has been observedthat thereare no secondarycunentsat the inlet to
the bend. The strengthof the cunenls increasesalong the bend until they
becomefully developedand are then uniformuntilthe bend exit is reached.
The secondarycurrentspersistin the straightreachdown slream, becoming
less and less intensewith distancefrom the bend exit. Where the stnaight
reachesbetweenbends are not long enoughto fully dissipatethe secondary
curents then the residualcunentsat the bend entrancecan have a strong
effecton the flow in the bend. The growthand decayof secondarycunents
has a s*ronginfluenceon the flow distributionwithina channelbend.

It is knownthat the bend tightness(radiusof curvature/width)has a strong


influenceon the secondarycurrentsdescribedabove. The tighter the bend
then the more pronouncedthe secondarycurrents. Tight bends also induce
zonesof flow separationparticularlyagainstthe innerbank. The effectof this
is to introducea 'dead zone' closeto the inner bank in which there is no
significantlongitudinalflow. A shearlayer is inducedand targe horizontal
vortices are induced within the zone. The effectivewidth of the bend is
reducedand the effectof secondarycunentsin displacingthe longitudinalflow
outwardsis enhanced.

Naturalchannelsare formedby the typicaldischargesthey pass. The size


and shapeof the channelvariesbothwithdischargeand plan geometry.lt is
generallyacceptedthat the importantchannelformingdischargefor natural
channelsis closeto the bankfullcapacityof the channel.In straightchannels
the flow inducessedimentmovement whichdeepensand widensthe channel
until some equilibriumstate is reached. The processeswhich inducethe
formationof rivermeandersare notwell understoodbut it is likelythat they are
relatedto efficiencyof the resultinggeometryin terms of both dischargeand
sedimenttransport. Thatis to say the resultinggeometryis the nps* efficient
shapefor passingthe bankfulldischarge and sedimentload.

Giventhata meanderhasdevetoped thenthe secondarycurrentswilldevelop


up to somemaximumstrengthand thendecayaway. Thesecunentsstrongly
affect the local bed shear stressesand form a channel cross section that
varies stronglyalong the bend. At the entranceto the bend where the flow
distributionand bed shear stressesare apprcximatelyuniform across the
channelthe cross sectionis approximatelyrectangularor trapezoidal. The
secondarycurrentstend to deepenthe channelon the outsideand transport
the materialtowardslhe insideof the bend wherethe lowervelocitiesallow it
to settle out. Thus many naturalbends exhibit deep pools at the outside
bankswith shallowregionsalongthe innerbanks. The poolsare deepestand
the shallowarea widest'ataboutthe apex of the bends. Althoughthe shape
of a channelvaries along a meanderit has been obseruedthat the cross
sectionalarea remainsapproximately constantthroughoutthe bend.

25 sR 329 07/O7Al3
4.2 Energyloss in channelbends
It is apparentthat the presenceof bendsin a channelwill affectthe discharge
capacity of the channel. In straight channels the only significantloss
mechanismis bed friclionbut in curvedchannelsotherloss mechanismsmay
also be inportant. lt was decidedto investigatethe relativeeffectof bendson
stagedischargerelationships.The inbankstagedischargedata frorn phase
B of the FCFwork was availableand was used in the followingwork.

4.2.1 Methodsof evaluatingnon-frictionlosses


The availabledata was collectedfor uniformflow conditions. The rate or
gradientof energydissipationalongthe channelswas constantand can be
assumedto be representedby the bed slope (S.).

The total energyloss is composedof frictionloss, bend lossesand all other


losses. The rate of energydissipationinducedby thesevariousmechanisms
are all assumedto be constantand the totalenergygradientis the sum of the
individualgradients.

Thefrictiongradient(S,)can be calculatedfromthe Darcy-Weisbach


equation,

s,= fv'?l(8gR) (s)

Subtractingthe friction loss from the total loss gives the sum of all other
losses. This can be representedby the differencebetweenthe total energy
and frictiongradients,So- 51,or as a bendlosscoefficient K., where

lq = hL| (\Fl2g) (4)

in whichh. is the headlossthroughthe bend. This can be evaluatedas

hL= (s"-sJl (5)

in whichI is the lengthof the channelthroughthe bend.

The lossesassociatedwith bends can also be accountedfor in terms of a


resistancecoefficient,most commonlyManning'sn. The ratio of the value
includingbendlosses(n) to the basicvalue (n) can be expressedin terms of
the energygradientsthroughManning'sequation,i.e.

(n?n)= (S/SJ'" (6)

4.2.2 Data sets


The effec,tof meanderingon flow resistancecan be inferredfrom the stage-
dischargedataobtainedfromthe inbankPhaseB experiments in the SERC
FloodChannelFacility. Threesets of data are available,one for eachof the
geometries tested,i.e.

the 60' meandergeometrywithtrapezoidalcross-section,


the 60omeandergeometrywith the naturalcross-section,and
the 110'meandergeometrywiththe naturalcross-section.

The measurements were all taken under uniformflow conditions(for the


naturalgeometriesthe bed undulatesconsiderablyand uniformityis assumed
to implyidenticalflowconditionsat the samepositionson successivebends).

26 sR 329 07/07/93
4.2.3 Fesu/tsand conclusions
Forthe 6d meandergeometrywiththetrapezoidalcross-section the difference
betweentotal energyloss and frictionloss can be whollyascribedto effects
associatedwiththe meanderplanform.This loss has beencalculatedfor each
measuredstage-discharge pair and expressedin each of the forms outlined
above. The bed slopeof the channelis givenby

So = Sr/s (7)

in whichS" is the slopeof the flume, and s is the channelsinuosity.

Forthe 6f meandergeometrythe slopeof the flumewas0.996x 10-3and the


sinuositywas 1.374,givinga channelslope of 0.7248x 1O3.The lengthof thd
channelthroughthe bend, L, was assumedto be measuredthrough half a
meanderwavelength, i.e.8.245m. Thisdistanceincludesthe straightsection
of channelat the cross-over.

Estirnationof bend losses for the 60o meandergeometrywith the natural


cross-sectionis complicatedby the variationof the cross-sectionshapealong
the channel. The flow distortionsassociatedwith this variationcan be
expectedto causeadditionalenergylosses, and so the differencebetween
the averagebed gradientand the frictiongradientcannotbe attributedto the
effectsof the meanderplanformalone. No experimentswere performedwith
a straightchannelwiththisnaturalgeometryandthe naturalcross-sec"tion was
about half the size of the trapezoidalone, so the lossesassociatedwith the
meanderplanformand the cross-section variationcannoteasilybe separated.

Becausethe hydraulicradiusvariesalongthe channelthefrictiongradient,as


calculatedusingEquations1 and 3, willalsovary. A valueof S,at eachof the
defined cross-sectionswas thereforecomputedand an average obtained,
weightedby the relativedistancesrepresentedby eachcross-section.For the
60omeandergeometrythe weightedaverageis givenby

Sr"" = ( 1.25S,, + 0.5745( S,o+ S,. + S,, + S,, + S,o))/ 4.1225 (8)

in whichS,, is the valuecalculatedfor the cross-sectionat the cross-overand


S,oto S,nare the values calculatedat the cross-sectionsdefined at equal
displacements throughthe bend.

The waterlevelwas meqsuredat the cross-oversectiononly. For calculating


flow areasand wettedperimetersat the other'sectionsit wasassumedthat the
water surfacewas flat, with a slope equal to the averagechannel slope of
0.7248x 10'3.Thisassurnptionis reasonabtebecausethe cross-sections were
designedwith a constantcross-sectional area. An energybalancebetween
the crossoverand apex sectionsfor one dischargeconfirmedthat the change
in water levelassociatedwith the cross-sectionvariationwas negligible.

Each natural cross-sectionwas compound,with a deep section and a


hodzontalberm. lt was assumedfor these calculationsthat the dischargein
the channelwas the sum of the dischargesin the deep and berm sections,
with any interactionbetweenthe two regionsunaccountedfor. The friction
gradientis thengivenby

S , = ( Q / ( A d ( 8 9 R , l t o l ' o+ A o ( 8 9 R ' / f o ) t o ) ) t (9)

27 sR 329 07n7l93
in whichQ is the totaldischarge,A is the flow area,and the subscriptsd and
b referto the deep and berm sectionsrespectively.

Thefrictionfactorswerecalculatedusingthe appropriatemodifiedsmoothlaw,
see Section3.4. Equation1 was modifiedslightlyby expressingthe velocity
in the Reynoldsnumberin termsof thefrictiongradientthroughequation3, i.e.

1 lfn = 2 . o 2 l o g( ( 4 ( 8 g ) t o l v ) R u s , 1 2 1- 1 . 3 8 (10)

Equations9 and 10 were solvediterativelyto obtainthe necessaryvaluesof


S, for each sectionand the averagevalue then calculatedby Equation8.

The non-frictionlossesfor the 60omeandergeometrywith the naturalcross:


sectionare presentedin Table 12. The lossesrepresentthe sum of those
associatedwith curvatureand the varyingcross-section.

The lossesforthe 110'meandergeometrywiththe naturalcross-section were


evaluatedin the sameway as for the 60" meandergeometrywith the natural
cross-section.In this casethe slopeof the flumewas 1.021x 1O-3and the
channelsinuositywas 2.043,and hencethe bed slopewas 0.49972x 1O3,
by Equation7. The length of the channelthrougheach bend, L, was
10.532m. Forthis geometrytherewas no straightcross-overreachand the
cross-sectionswere definedat equaldisplacements throughthe bend. The
averagefrictionslopewasthereforecalculateddirectly,withoutweighting.

The non-friction
lossesfor the 110omeandergeometrywiththe naturalcross-
section,whichagain includelossesassociatedwith channelcuryatureand
varyingcross-section,
are presentedin Table13.

The meanvaluesof S, - S,, K and n'ln for the threecasesare listedbelow.


The standarddeviationsare in brackets.

Channel So-St K n'/n


(x tos) F $'A')
60o trapezoidal 0.107 0.081 1.078
(0.032) (0.026) (o.o2e)
60" natural 0.236 0.415 1.217
(0.012) (0.113) (0.013)
110onatural 0.186 0.954 1.262
(0.005) (0.108) (0.010)

It is interestingto notethatthe standarddeviationson the energygradientsare


muchless in the caseof the naturalcross-sections.This impliesthat the rate
of energydissipationdueto bendeffectsis moreuniformwithstagefor natural
channelsthan for trapezoidalchannels. These sectionswere designedto
mimictypicalnaturalriversand so in real chdnnelsthe bend lossesmay not
vary as strongly with depth as in the case of trapezoidalor rectangular
channels. By comparingthe resultsfor the two 6d geometrieswe can see
that channelcross-sectionshape stronglyaffectsthe non-frictionlossesand

sR 329 07/07193
that these are approximalelytwiceas largefor the naturalchannelas for the
trapezoidal channel.Thedifferences betweenthe resuftsfor the 60'and 11O'
are less conclusivebut the non-frictionlossesappear to vary with channel
sinuosity.

In order to assessthe significanceof these extra lossesthe mean gradients


above have been normalisedby the total energy gradientsand are quoted
bebw. Theseresultsshowthat non-frictionlossescan be very significant. In
the casesexaminedthe non-frictionlossesformedbetween15%lo 4OY"of the
total energylosses. lt is impossibleto draw generalconclusionsfrom these
data but they do indicatethat further investigationof non-frictionlosses in
channelbends is required.

Channel (q-sJ/q
60'trapezoidal 0.15
6O'natural 0.32
11d natural 0.37

4.3 Energyloss mechanisms


The resuhs presentedabove confirmthat the presenceof bends in open
channelflows affect the energy loss comparedto straightchannels. The
questionwhichstillremainsto be answeredis: Howsignificantare thesenon-
frictionenergylossesand whatarethe impodantparameterswhichaffectthem
? The followingauthorshave tried to identifyand quantifythe mechanisms
whichinducethis extraflow resistance.

Shukry(1950)carriedouta set of experiments in rectangularchannelbends.


He constructed singlebendswhichturnedthroughangles(0)of 90o,135"and
18d. The experiments wereconductedfor depthto widthratios(y/B)of 0.6,
0.8, 1.0 and 1.2and alsofor bendradiusto widthratios(y'B) of 0.5, 1.0,2.0
and 3.0. Shukryanalyzedthe extra energyloss inducedby the presenceof
the bend usinga bendlosscoefficient(K.),definedas:

tr,_= l( (v2t2g) (11)

in which h. is the head loss due to bendsonly,and V is the overallmean


velocity. He showedthat the bend loss coefficientis a functionof:

1) The Reynoldsnumber
2') Depth ratio (y/B)
3) Radiusof curvature(r/B)
4l The anglesubtendedby the bend(0/180)

In additionhe foundthat the propoilionof these extra energylossesinduced


dudngdevelopmentof the secondarycurrentswereapproximatelyconstantat
4Oo/o.

Rozovskii(1957)publishedthe seminalanalyticalwork on flows in channel


bends. He examinesthe theoryof manyof the mechanismsdescribedabove
and comparespredictionswith both field and laboratorymeasurements.He
identifiedthe followingsourcesof energyloss:

1) The redistribution acrossthe channel


of longitudinalflow
2) Energylostin initiatingsecondarycunents

29 sR 329 07rc7l93
3) Increasedbed frictiondue to the secondarycurrents
4l Increasedintemalenergydissipationdue to internalfrictioncausedby the
secondarycunents
5) The redistribution
of longitudinalflow
in the vertical
6) Separationand the forrnationof eddy zonesin sharpbends.

Rozovskiianalyzedthe energydissipatedby each of these mechanismsin a


widerectangularchanneland concludedthat the impoilantmechanismswhich
significantlyincreaseenergydissipationin bendsare the increasedbed and
internalfrictiondue to the secondarycurrents.

He providedthe followingexpressionfor the extra energylosses:

h. = (24 g'otc + 60 g/c'a)(ytr,l2(Vy)(vztzgl (12)

whereh. is the total extraenergyloss in a bend of lengthI and radiusr". y is


the flow depth,g the accelerationdue to gravity,V is the averagetlow velocity
and C is the Chezy bed friction parameter. This equation was derived
assuminga logarithmic distribution
of the longitudinalvelocities
in the vertical
and that the secondarycurrentsare fullydeveloped.In generalthisanalysis
showsthatthe energylossesdue to a bendincreasewithchannelroughness
and the squaresof flowvelocityand depthto radiusratio. Hencethe tighter
a bendthe largerthe energydissipated.

Muchof Rozovskii'sanalysiswas approximate:he was forcedto makemany


assumptions and he concludedthatfurtherexperimental
andtheoreticalwork
is required.

Leopoldet al (1960) explainedthe resistancebehaviourof meandering


channelsby identifying
threemajortypesof resistance.

1) Skin resistanceis associated withthe surfaceroughnessof the channel


and varieswiththe squareof the flow velocity.
2) Internaldistortionresistanceresultsfrom energydissipationby eddies,
secondarycirculationand increasedshear rate,whereverany boundary
featuredeflectspartor all of the flow from its formerdirection. lt will also
vary with the squareof the flow velocity.
3) Spill resistanceis associatedwith localaccelerations followedby sudden
expansionsin the flow and can be relatedto Froudenumber.

Leopold et al conductedexperimentswith inbank flows and moderately


sinuous channels. They found that channel curvaturecould, by intemal
distottion,induceenergyloss of the same order as that due to skin friction,
and doublethat amountin tight curues. This type of loss could be relatedto
radiusof curuatureof the bendsand the ratio of channelwidth to radius.

Energyloss associatedwith spill resistanceappearsto be just as significant


but only comes into effect at a critical value of Froude number (which is
substantially
less than 1.0). lt appearsthat the Froudenumberat banldull
depth is generallyless than this criticalvalue in naturalchannels. This
mechanismmay be responsiblefor determiningchannelwidth by inducing
bank erosionat its onset. lt is unlikelyto be a nrajor loss mechanismfor
inbankflowsin naturalriversand will be neglected.

30 sR 3Z' 07rc7l93
Formresistanceassociatedwithflowaroundsmall-scaleallwial bedformscan
be consideredtogether with skin resistance by estimating a combined
resistancecoefficient.As notedby Onishiet al (1976),skin resistanceis not
independentof internaldistortionand may be enhancedby the non-uniformity
inducedby secondarycurrents.

Internal dis{ortion resistanceresults from energy dissipation by eddies,


secordarycirculationand increasedshearratewhereverany boundaryfeature
deflects pad or all of the flow from its former direction. The secondary
circulationinduced by meanderingis a major contributorto this type of
resistance.

Onishiet al (1976)investigatedinbankflowsin meandering, allwial channels.


They attributedhead loss to the followingfour categoriesof flow resistance.

1) Surfaceresistanceor boundarystress,which may be enhancedby the


nonuniformdistributioninducedby secondarycurents.
2) Form drag, resulting from the unsymmetricaldistribution of normal
pressurearound curves and deformationson the boundary. These
losses are due primarilyto separationbut are also influencedby
secondary currents.Theydependon the Froudenumber,channelwidth,
and the stream-wiseand transversenon-uniformityof the channel
geometry.
3) Superelevation, which causes additionalasymmetryof the normal
pressuredistributionon walls and large scale bed forms, resultingin
'waveresistance'.Theselossesdependon the channelgeometryand
the Froudenumber.
4) Bed forms,in alluvialchannels.

Onishiet al (1976)describedthe total loss due to bendsusinga bend loss


coefficient,
definedas:

hr-= K_ N2l2g) (13)

in whichh. is the head loss due to bendsonly, and V is the overallmean


velocig. Theyshowedthat this couldbe expressedas:

K-=L/4Rb(fb"-fJ (14)

in whichL is the lengthof the bend,Rois the bed hydraulicradius,f* is the


bed frictionfactorfor the meanderingchanneland f* is the bed fdctionfactor
for a similarbut straightchannel. The bed loss coefficientcould be relatedto
channeland flow characteristics by:

K=f (v/(gRJo'', R/D*, B/r") (15)

In whichD* is the mediansizeof the bed material,B is the channelwidthand


r" is lhe centre-lineradiusof curuature.

The resultsobtainedby Onishiet al showed|(. to be stronglydependenton


the Froudenumber. In some cases I was negative,implyingan energy
gradientlessthanfor correspondingstraightchannels.Thiswas attributedto
a relative decrease in bedform drag and possible decreases in wave
resislanceand boundaryshear.

31 sR 929 07r'07/93
Hayat(1965)obtainedvalueof K_(asdefinedabove)for meanderingchannels
with rectangularcross-sectionsand rigid beds. In contrastto the altwial
channel resuftsof Onishi et al (1976), K. was found to be approximately
constantwith Froudenumber.

The variationof cross-sectionalgeometryalong a channel has also been


as a sourceof energyloss(e.9.Chow,1959).Kazemipour
identified andApeft
(1979, 1983),however,have shown that such irregularitycontributesno
additionalenergylossprovidedlhat noflow separationor brokensurgesoccur.

Fromthe aboveit is apparentthat the mainsourcesof energyloss in channel


bendsare:

1) Bedfriction
2l Increasedbed frictiondue to secondarycurrents
3) Internalenergy dissipationdue to increasedturlculenceinducedby
secondarycurrents.

The energyloss in a bend has been found to dependon the following


parameters:

Bed roughness(f, C n etc)


Flowdepth (y)
Bendradius(r")
Lengthof bend(l) or Angleof bend(0, | = r" 0)
The cross-sectional
shapeof the channel.

Any generalmethodfor predictingflows in bendsshouldaccountfor these


three processesand be formulatedin terms of the five parametersabove.
Manymethodshave been identifiedin the literature.The majorityof them
havebeenderivedempirically fromlaboratorydataand maynot includeall of
the importantparameters.The followingsectiondescribesmethodswhich
havebeenidentifiedin the literatureand in additiontwo methodshavebeen
modifiedto improvethe predic,tions.
The morepromisingof thesemethodsare
thenappliedto the availablelaboratorydata.

predictionmethods
4.4 Stage-discharge
Althoughthereis nowbetterunderstandingof the mechanismsof energyloss,
mosthydraulicstextbooksstillrecommendaccountingfor theireffectstogether
by a simpleadjustmentto the valueof Manning'sn for a similarbut straight
channel. Such adjustmentshave been proposedby Cowan(1956)and the
Soil Conservation Seruice(1963). Thesemethodsare very similarand only
the laterone is coveredbelow.

The Soil ConservationService(SCS)(1963)Method

TheSoilConservation Service(1963)proposedaccounting for meanderlosses


by adjustingthe basicvalueof Manning'sn on the basisof sinuosity(s), as
follows.

n?n = 1.0 for s < 1.2

n'ln = 1.15 tor 1.2 > s < 1.5

n?n = 1.39 for s > 1.5 (16)

sR 329 07/O7l93
in whichn/ is the adlustedvalueand n is the basicvalue.

Becausen is proportionalto
f12,theadjustment
shouldbe squaredwhenusing
the Darcy-Weisbachequation.

The LinearizedSCS (LSCS)Method

The step natureof the SCS recommendationintroducesdiscontinuitiesat the


limits of the defined sinuosityErnges,with consequentambiguityand
uncertainty.To overcomethis problemthe relationship
has been linearized
and is expressedas:

n'ln = ((10112= 0.43 s + 0.57 for s < 1.7

n'ln = ((ff1t8 = 1.39 for s > 1.7 {17)

in which f is the adjusted Darcy-Weisbachfriction factor.

The Method of Scobey (1933)

On the basis of flume tests Scobey suggestedthat the value of Manning's n


should be increased by 0.001 for each 20 degrees of curvature in 100 ft of
channel. These recommendations are not expressed in terms of
dimensionlesschannel characteristicsand are unlikely to have consistent
accuracy at different scales.

The Method of Mockmore (1944)

Mockmore (1944) analyzed data from afiificial channels and rivers for bend
angles between 90oand 180' and proposedthe relationship:

hL = (2 blr.) y2 | 29 (18)

in whichh,_is the energylost througha bend,in excessof the frictionloss.


The frictionlossis obtainedfromnormalhydraulic eg the Darcy-
calculations,
Weisbachequation,i.e.

V = (8gRS,/f )r2 (1e)

in which g is gravitationalacceleration,R is the hydraulicradiusof the cross-


section,S, is the energygradient,and f is the frictionfactor. For uniformflow
S, can be equatedto So, the bed gradient. The energyloss due to friction
alonga lengthof channelI is givenby:

h= (f l/4R) v2t2s (20)

Combiningthesegivesthe total headloss:

hu+h= ((fll4R) +2Blr"l Yzl29 (21)

on re-arranging
this becomes:

h.+h= (f +8RB/lr")(l l4H)V2 l2g (221

by comparingwith Equation2Oit is apparentthat the extrabend head losses

33 sB 329 07/07/93
canbe considered
as an adiustment
to the straightchannelfrictionfactorwith:

{ = f +8RBllr" (2g)

This form of the method is easier to apply to stage dischargedata from


meanderingchannelswherea bedfrictioncalibrationfor an equivalentstraight
channelisavailable.

The Methodof Leopoldet al (1960)

From a set of laboratoryexperimentscanied out on meanderingchannels


formedin sand Leopoldet al presenteda graphicalrelationshipbetweenthe
ratioof the additionalboundary
shearinducedby channelcurvature (r) to the
boundaryshearassociatedwith friction(tJ and the ratioof flow width (B) tci
meanradiusof curyature(r"). This can be expressedas:

r/r" = 2.632(WrJ - 0.526 (24)

and appliesbelowa criticalvalueof Froudenumber(approximately 0.5). At


higherFroudenumberstheadditional shearwasa functionof Froudenumber.
By refatingboundaryshearstressesto velocity,Equation24 canbe interpreted
as an adjustmentto the frictionfactoras follows. The basicDarcyequalion
relatesshearstressto the squareof velocitywiththe coefficientf:

r = ptvzlzg (2s)

Assumingthat the total shearslress is composedof the two components


definedand that each componenthas a corresponding frictionfactorthen
Equation24 becomes:

tlt = 2.632(B/rJ - 0.526 (26)

the totalfrictionfactoris givenby


( =t +f, (27)

rearranging
and dividingby f gives

tilt=( lt-1 (28)

substituting
in Equation26 gives
( lt = 2.632(Btr")+ 0.474 (29)

The Toebesand Sooky (1967)Method

From experimentalresultsin a small laboratorychannelwith a sinuosityof


1.09,Toebesand Sooky(1967)proposedan adjustment to f. Belowa critical
valueof the Froudenumberthe adjustmentdependssolelyon the hydraulic
radius(in metres)accordingto:
(lt = 1.0 + 6.89R (30)

They confirmedthe conclusionsof Leopoldet al that abovea criticalFroude


numberthe increasein lossesdueto channelcuryatureis a functionof Froude
number. The criticalvalueof the Froudenumberwas foundto dependon
hydraulicradiusbut wasnotexceededin anyof the applicationsreportedhere.

34 sB 329 07/O7l93
The Methodof Agarwalet al (1984)

Agarwalet al (1984)performeda regressionanalysison previouslypublished


data from alluvialchannelsto definea conectionfor bend losses. The actual
flow velocityis determinedby dividingthe velocitycalculatedaccordinglo
RangaRaju's(1970)resistancelaw by ( where:

€ = 2.16 f.R{'o12 (31)

with

f.n= Re (0 / 18oo)'4'6s
(B / y)''tt (r"/ b)r's Ff'a (92)

in which0 is the bendangle,y is the flowdepth,Fr is the Froudenumber,in


termsof the hydraulicradius,i.e. V / (gR)'', and Re is the Reynoldsnumber,
(4RV/ v), wherev is the kinematicviscosity.

RangaRaju's(1970)resistancelaw is intendedfor use in alluvialchannels.


The adlustmentfor bendlossesis independentof the frictionlosscomputation
and it is assumedthat it appliesto rigidboundarychannelsas well,with any
appropriate resistancelaw. lt is unclearwhetherRe and Fr in Equation32 are
in termsof the actualvelocityor thatcalculatedfromfrictionlossesonly,and
the latterhas beenassumed.

(1983)
The Methodof Pacheco-Ceballos

Pacheco-Ceballos (1983)re-analyzed the resultscollectedby Shukry(1950).


He relatedthe headlossdue to the bendto the velocityat the bendentrance.
Otherauthorsexpressheadlossin termsof the averagevelocitythroughthe
bend. By assumingthat the lateraldistributionof velocitywithinthe bend
followsthe free vortexprofilehe producedthe followingequationfor l(':

K- = ( Vr - y. + Vr2l 2g - ( N y" V'l y^' 29 ) ) sg / V'? (33)

Wherey, and V, are the flow depthand velocityin an equivalentstraight


channel;y and V are theflowdepthand velocityat the bendentranceand y.
is the averagedepthat the positionalongthe bendwherethe secondaryflow
becomesfully developed. N is a parameterrelatedto free vortexflow:

N = [ ( In rol r, )t ror, / Bt ft (34)

whereroand r, are the radiiof the outerand innerchannelbanks. The term


yr - v. in Equation33 is approximatedby :

log (y, - yJ = 2.11v - ( Q+ o.7 rtB- 0.oGG"/B)'z+ y) (35)

Qis a parameterwhichvarieswiththe bendangle(0). For Shukry'sbendsof


45o,90oand 180oit hasvalues2.98,2.7Oand 2.64respectively.Intermediate
valuescan be obtainedby interpolation.

The Methodof Chang(1983)

Chang (1983) deriveda generalanalyticalmodel for the rate of energy


expenditureper unit channellengthassociatedwith transverseflow. This
modelis basedon the conceptualmodeldeveloped by Rozovskii(1957)and
assumesthatthe extraenergylossis due to increasedbed frictionand internal
turbulencerelatedto thesecondary currents.Chang(1983)assumeda power
law for the verticaldistribution velocity. This gave a different
of longitudinal
expressionfor the secondarycurrentcomparedto Rozovskii's.

sR 329 07/07193
For the case of a wide rectangularchannelwheresuperonewordelevation
and the lateralvariationof secondarycurrentsare smallhe approximatedthe
sec-ondarycunents with a linear distributionand produced a simplified
expressionfor the energyloss in a bend wherethe secondarycunents are
fully developed.

s, =(2ffiff ,2n tl l'rT Frz (36)


( o . s o s + f %J [ t . l
in whichS/ is the energygradientassociatedwith transverseflow (h,_/l).For
uniformflow,

S.-S,=9'

or

1-S,/So=S'/So (37)

Sincethe ratesof energylossare linearlydependenton the frictionfactors,

1-S/S"=1-t/( (38)

Rearranging it is possibleto seethat Chang's(1983)method


andsubstituting
can be interyretedas an adjustmentto the basic straightchannelfriction
factor.
(lt = 1/ (1 - s/// s.) (39)

Chang(1988)repodsa slightlydifferentformof Equation36:

()
sr/ = 2.O7t + 4.68
tr - 1 .83 f*l
-) lI I Fr2 (40)
0565 +
trl
Howeverboth forms were foundto give very similarresultsin preliminary
calculations
andthe simplerEquation36 has beenusedthroughout.

The ModifiedChangMethod

Chang's (1983) method was developedfor wide, rectangularchannels.


Becausemost riversand floodchannelshave largewidthto depth ratiosthis
was not considereda major limitation,but the effect of shape warrants
investigation
at a laterstageto confirmthe method'svalidity.

Chang's(1983) methodalso.assumesthat secondarycirculationis fully


distanceto develop
developed. In fact, the circulationtakes considerable
througha bendand.beginsto decayoncethe channelstraightens out. For
meandersthe circulationmust reversebetweensuccessivebends and the
associatedenergy gradient must drop to zero at two points over each
wavelength. The avercrgeenergy gradient associatedwith secondary
circulationalong the channelmust thereforebe substantiallyless than
predictedassumingfull development.Rozovskii(1957)studiedthis growth
anddecayof secondary currentsanalytically.He assumedthatthedistribution
remainsconstantduringthe processof decay. He showed
of the circulations
that the angleof bend requiredfor the secondarycurents to becomefully
developedis:

O r a= 2 . 3 C y / ( g r z r " ) (41)

JO sR 329 07/07/93
whereC is the Chezycoefficient.This can be writtenin terms of Darcyf:

O r c= 6 . 5 y 1 ( t 1 n r " ) (42)

The conespondinglengthof channelrequiredfor fully developedsecondary


cunents is

fo=6.5ylt'o (43)

Applyingthiscriterionto the SERCchannelgeometryshowedthat undersome


flow conditionsthe circulationwould never developfully. Significantly,the
degreeof development variedgreatlywith stage in the same channel. For
example,for the 60otrapezoidalchannel,secondarycirculationwouldbe fully
developedonly after152oof curyatureat bankfulland after54' for a flow depth
of 0.06 m, which is approximatelythe lowestdepthtested. The curvatureof
each bend in this geometryis 120', so secondarycirculationwould probabS
be fully developedover a considerableproportionof the channellengthat low
stages,but not at allat relativelyhighstages.

Ghang (1984)accountedfor the effectsof growth and decay of secondary


circulationby applyinghis full secondarycirculationloss modeltogetherwith
nonuniformflow calculationsto predictthe distributionof lossesand boundary
shear stresses,as well as water levels through bends. This requires
integrationof streamwiseand transversevelocitiesat each computational
sectionand wouldbe impracticaltouse. As an allernative,his approachwas
simplifiedto applyto uniformflow througha sequenceof identicallyrepeated
meanders.Becausethe energygradientvarieswith the growthand decay of
secondarycirculation, flowcan not actuallybe uniform.Alsothe bed slopeis
unlikelyto be constant;it will vary over a meanderwavelengtheven for
ldealizedlaboratorymeanders. The assumptionof uniformityis thereforea
butthe primaryvelocityand flow depthwill not vary greatly. For
simplification,
determiningthe effective resistancein meanderingchannels avenge
conditionsare sufficientand minordeparturesfrom uniformityare unlikelyto
influencethe conclusions.This approachenablesa correctionfactor to be
c.omputed whichcan be appliedto the energygradientpredictedby his wide-
rectangularequation(44),to accountfor growthand decayof circulat'ton.
Chang(1983)presentedan Equationfor the energygradientassociatedwith
fully developedtransversecirculation, 9',0.

Fr? (Ml
[iJ
in whichFr is the Froudenumber.

It is assumed(as by Chang,1984)that the patternof secondarycircutation


remainsconstantduring growthand decay. The strengthof the circulation,
and its variation,can then be representedby the transversevelocityat one
positionon the profile,and particularlyat the watersurfaceat the centreof the
channel. lf it is furtherassumedthat the local value of energy gradient
associatedwith secondarycirculation,S/, is proportionaltotransverseveloc'lty
(longitudinalvelocityis constantby the uniformityassumption), then 9/ can be
relatedto the fully developedvalue by

a" = S1o(v,"/ v,",0) (45)

in whichv," is the transversevelocityat the watersurfacein the centreof the


channel,andthesubscriptfd denotesthe fullydevelopedvalue. Similarly,the

37 sR 329 07/O7l93
average value of $/ through a meanderwavelengthis given by

9'.ro = 9'ro (Vr""r" lvr.u) (46)

In Equation 46 V,""". is the average of absolute values because the sense of


vrcreverses between successive bends.

The total gradient of energy losses is the sum of friction gradient (S) and the
secondary circulation loss gradient (it is assumed here that there are no other
sources of energy loss, or that these are accounted for in the basic friction
factor). Under uniform flow conditions the total gradient of losses is equal to
the bed gradient So.and S//can be represented by S/",". Therefore

Sr+S/.r"=5" (47')

S, can be estimatedusing the Darcy-WeisbachEquation,i.e.

S, = (f V'?y(8g R) (48)

in whichV is the meanflowvelocityand R is the hydraulicradius.

Substituting
Equation48 for S, and Equation46 tor S/"""in Equation47 and
rearranginggives

V = ((g qy(ff(8R)) + K))'o

with

K = ((2.86tla+2.o70/(0.565 +tto))(ytr"lz(BlA)
(V,""u/V."ro) (49)

in whichA is the cross-sectionalarea


and B is the surfacewidthof the flow.

Equation49 can also be expressedas

V = ((8gRS.)//)12

with

f =tl(1 -g/""d/s") (so)

For evaluationof Equation49 or 50, v.",0can be calculatedfrom the equation


for the distributionof tranwersevelocity(v) underfully developedconditions
givenby Kikkewaet al (1976),i.e.

vlY = F (ytrl (1/r) ( F,(zJy)- (1/r) gru)F"(ztyl)

with

F = ((yly)(rtr)l1n

Fr(z|y) = - 15( (zly)zln(zly)'112(zty)z+ 15124)

F"(zly) = 1512( (zty)" ln(zlyl - (zly)" ln(zlyl


+ 112(ztylz- 19154) (51)

in whichr is the von Karmanconstant,V. is the shearvelocity,y" is the flow


depth at the channelcentre,z is the verticaldirection,and r is the radial
direction.

38 sR 329 07/07/93
At the channelcentrey = v. and at the watersurface2 = t, dfid so F = 1. Fr
= 10/3,and F. = 10/9. Substituting
thesevaluesin Equation51 givesthe fully
developedtransversevelocityat the watersurfaceat the channelcentre,

v,.roff = $/r")(1/r)(10/3-(1/r)(VJVX10/9)) (52)

The von Kannanconstanthas a value of 0.4 and the shear velocitycan be


determinedby

V. = (g R S,)t" (53)

with Sr = Sofor uniformflow.

The averagesudacetransversevelocity,v,"""" is also requiredfor evaluatingi


Equation49 or 50. Chang(1984)presentedan equationfor computingthe
transversevelocityat the watersudacealongthe centreline througha bend.
Thevelocityis computedat discretecrosssectionsalongthechannel,andthe
value at any sectionis relatedto that at the precedingsectionby

. = ((qJi+ (t2l12(10/q: (1/rx5/9xf/2)r1


(v,")r,
(V/r)'exp((rc/yXf/2;ln exp0(rc/y)(f/2)12
as1as11 As (54)

in whichthe subscriptj is the cross sectionindex,and As is the distance


betweensectionsjand j + 1.

Equation44 includeslwo terms,one describingthe growthand the otherthe


decayof secondarycirculation.The full Equationappliesto flow througha
bend. Alonga straightreachaftera bendonly the decayterm appliesand

(vJi.r = (vJ,exp(-(r</yl(tl2)1P
ts (55)

Calculationof v,".u.requiressolutionof Equations44 and 45, withV givenby


Equation49 or 50 and v,",0is obtainedfrom Equation52. Becauseof the
implicitnatureof this set of equationsthe solutionis iterativeand is obtained
as follows.

1) A first estimateof the mean velocity is calculatedneglectinglosses


associated with secordary circulation, using the Darcy'Weisbach
Equationand an appropriateformulafor the frictionfactor.

2) This velocityand (v,"),= 0 are used in Equations54 and 55 to compute


an initialdistributionof v," throughone completemeanderwavelength.

3) The value of v," at the last section is substitutedfor (vJt and the
distributionis recomputediterativelyuntil the value of v," at the first and
lastsectionsare identical,withina specifiedtolerance.This conesponds
to uniformconditionsthrougha seriesof identicalmeanders.

4l The averagevalue of absolutev,"throughthe wavelengithis calculated


as

(v,").," = (E V,"Asy(t As). (56)

s) The meanflow velocityis recalculated,accountingfor lossesassociated


usingEquation49 or 50.
with secondarycirculation,

6) The recalculatedmeanvelocityis then used in Equations54 and 55 to


of v,"throughthe wavelength.
computea new distribution

39 sR 329 07/O7l93
m
7) This procedureis repeateduntil the recalculatedmean velocityis the
s€rmeas the previousone,withina specifiedtolerance.

This method is obviouslynot suitablefor direct applicationin practice,but


could be appliedto differenthypotheticalsituationsto develop relationships
between9/*. and geometricand hydraulicparameters.This would provide
a method for estimatinghead losses without the limitationsof the LSCS
method.

4.5 Applicationof predictionmethods


Various methodshave been identifiedfor accountingfor the additional
resistanceto flow inducedby channelcuruature.Theseare as proposedby:

Scobey(1933)
Cowan(1956)
Soil ConservationService(SCS)(1967)
Toebesand Soolq (1967)
Leopoldet al (1960)
Shukry(1e50)
Mockmore(1944)
Onishiet al (1976)
Agarwalet al (1984)
Rozovskii(1957)
Chang(1983)
Chang(1e88)
Pacheco-Ceballos (1983)

Someof thesewere not consideredfuflherfor variousreasons. Scobey's


methodgaveunrealistic predictions for the datasets used,probablybecause
it is not expressedin termsof dimensionless variablesand suffersfromscale
effects.Cowan'sapproachis similarto the SCS method,whichallowsbetter
quantitative descriptionof channelcharacteristics. Shukry'smethodcouldnot
be appliedto the datasetsavailablebecausehis curvesfor someparameters
did not extendto theirconditions.The methodproposedby Onishiet al was
intendedfor mobilebed channelsand requiresspecification of sedimentsize;
it is thereforenotappropriatefor the conditionsunderwhichthe availabledata
setswereobtained.Rozovskii'sEquationis very similarto Chang'sand it was
not thoughtworthwhile to considerboth. Chang'sequationproducedbetter
results in preliminaryapplicationsand is also extended in subsequent
publications; it was therefore selected in preference to Rozovskii's.
Chang's1983and 19BBequationsare virtuallyidenticalandthe 1988version
was rejectedas it pedormedslightlyworsein the preliminaryapplications.The
methodof Pacheco-Ceballos is difficultto applyand has not been considered
at this stage. Thus the followingmethodshave been considered:

Seruice(SCS)(1967)
Soil Conservation
Toebesand Sooky(1967)
Leopoldet al (1960)
Mockmore(1944)
Agarwalet al (1984)
Chang(1983)
ModifiedChang(1984)
Service(LSCS)
LinearizedSoilConseruation

To demonstratethe effectof meanderingon channelconveyanceand to


provide a basis for comparisonof the other methods,stage-discharge
relationships
werecalculatedignoringnon-friction
losses.

40 sR 329 07/O7l93
Friction loss only

For a givenstagethe dischargeis givenby

Q = AV (57)

in whichA is the cross-sectional


area and V is the flow velocity,given by the
Darcy-Weisbach equation,i.e.

V = (agRs,/flt" (S8)

in which g is gravitationalacceleration,R is the hydraulicradiusof the cross-


section,S, is the energy'gradient,and f is the frictionfactor. For uniformflow
Q can be equated to So,the bed gradient.

4.5.1 Data set


The selectedpredictionmethodswere appliedto the followingthree sets of
data, none of which were used in the developmentof any of the methods.

1. A full stage-discharge
relationshipfor a lrapezoidalchannelconstructed
in the SERCFloodFacilityat HR Wallingford, UK. Thischannelhad a base
widthof 0.90m, sideslopesof 45o,a depthof 150mm and a bed gradientof
0.00073.Thesinuositywas1.374andfourcompletemeanderswereinstalled.

2. Full stagedischargerelationshipsfor trapezoidalchannels at the


Universityof Aberdeen(Willetts,personalcommunication).These channels
aff had base widths of 139 mm, side slopes of 71", and depths of 50 mm.
Sinuositieswere 1.21, 1.41 and 2.O43with bed slopesof O.O0083, O.OOO71,
and 0.00030respectively.

3. Bankfulldischarges for trapezoidal


channelsmeasuredby the US Army
Corpsof Engineersat the WaterwaysExperimentStation,Vicksburg. The
channelshad sideslopesof 63o,depthsof 0.152m and basewidthsof either
0.305m or 0.610m. Forthe widechannelsinuosities of 1.20,1.40and 1.57
were tested. Forthe narrowchannelthe sinuositiestestedwere 1.17, 1.22,
1.33,1.49,1.50,1.75and2.54. ln allcasesthe valleyslopewas 0.001. Full
detailsof the channelsand experimentsare reportedby the US Army Corps
of Engineers(1956).

4.5.2 Resultsandconclusions
Eachof the methodsdescribedabovewas appliedto predictthe dischargefor
everyflow conditionin thesedatrasets. The frictionfactorfor the SERC and
Aberdeenchannelsvariedwith Beynoldsnumberand were calculatedby the
appropriatemodifiedsmooth law, see Section 3.4. This requiredthat the
equationsrepresenting thedifferentmethodsbe solvediteratively.Therewere
no data to establishvariationsof frictionfactor for the Vicksburgchannels,
and a constantvalue for eachchanneltypewas calculatedfrom the bankfull
flowsin the conesponding straightchannel.

The (per cent) error in eachpredictionwas calculatedas

Enor = 100 (Qe - QJ / Q.

The averageerrorand standarddeviationof errorsfor each data set and for


allthe datatogether,are listedfor eachmethodin Table14. Twovalueswere
computedfor some of the Vicksburgdata with the SCS Method. This was
becausethe sinuositiesfell on the thresholdsof the correctionfactor defined
by Equation16. Valueson either side of the thresholdswere used and
averagesincludingboth resultspresented.The first columngivesthe error

41 sR 329 07/O7l93
obtainedby ignoringbendlossesandthereforegivesan indicationof the effect
of meandedng on resistance.

ln terms of averageerrorand the standarddeviationof enors, the Modified


Changand SCS Methodsappearto pedormbest with meanerors withinthe
range €% to +5% and standarddeviationsof less than 1O%. lgnodngthe
energy loss inducedby meanderinggives unacceptablyhigh errors in the
predictionof dischargefor inbankflows. Of the methodsconsidered, those
of Agarwalet al (198a),Mockmore(1944)and Chang(1983)appearto be
unsatisfactory. The Chang methodsare the only methods with a sound
theoreticalbase, but the ModifiedChang Methodis not easy to apply in its
present form. All the other methodsare empiricaland based largely on
laboratorydata;their generalityis thereforenot assured

Theoverallperformance of theSCSmethodis surprisinglygood,andsuggests


that adjustingManning'sn by a factorrelatedsimplyto sinuosityis reasonable.
The relationshipbetweentheadjustmentfaclorandsinuosityas recommended
by SCS and as derivedfromthe data used hereis shownin Figure15. (Ihe
valuesderivedfromthe dataare approximate.They werecalculatedfrom the
dischargesas measuredand as calculatedassumingfrictionloss only. The
variation of friction coefficientwith Reynoldsnumber as bend losses are
introducedare thereforenot accountedfor.)

One undesirablefeatureof the SCS recommendation is that it is a step


function. The consequencesof this are apparentin the predictionof the
Vicksburgchanneldischarges.Forthe wide channelwith a sinuosityof 1.2
the error is 30.71%or 13.66%,dependingon which side of the step the
sinuosityis assumedto lie. Similarly,for lhe nanow channelwith sinuosity
of 1.50the errorcouldbe -8.1O% or -18.71o/o. lt wouldobviouslybe advisable
to replacethe SCS stepfunctionwith a smoothcurve. lt is difficuftto know
wherethis curve shouldlie becausethe data are fairlyspread out. One
reasonfor the data spreadis that bendlossesare not causedby sinuosityper
se, but ratherby the degreeof curvature.This is well demonstrated by the
Vicksburgnarrowchanneldatafor sinuosities of 1.49and 1.50. Althoughthe
sinuosities are almostidentical,the bendsin the 1.49sinuositychannelare
tighter,with longerstraightreachesbetweenbends. The tighterbendscause
greaterenergyloss and the adjustmentfactoris 1.23,comparedwith 1.06for
the other, nore gently curvingchannel. This effect is acmunted for by the
ModifiedChangMethod,as shownin Figure16 whereit was usedto compute
the adiustmentfactorfor eachdata point. The spreadof the predictedvalues
is stillconsiderable,confirmingthat it is associatedwith factorsnot accounted
for by sinuosityalone,ratherthan experimental scatter. The rangebarsfor the
SERC and Aberdeendata points in Figures 15 and 16 show lhat the
adjustmentfactor also varies considerablywith stage, and that this is
reproducedby the ModifiedChang Method. lt is therefore not entirely
satisfactoryto accountfor bend lossesin meanderingchannelsin terms of
sinuosityalone. A more reliableadjustmentfunction in terms of nadiusof
curvatureand bend angle could be determinedusing the ModifiedChang
Methodin hypotheticalapplications.

Usingthe SCS methodas it standswouldnot lead to majorerrors,however.


To makeit moresatisfactory,the stepsin the relationshipcould be eliminated
by using the curve shown in Figure16, ahhoughthe inherentlimitdtions
remain. This has beendoneand the resultingmethodis referredto as the
Linearized SoilConservation Servicemethod(LSCS).Predictionerrorsusing
this linearization
are listedin Table15 in the columnheadedLSCSMethod,
These reducederrorsshowthat the linearizedversionis superior.

42 sR 329 07/O7l93
The followinglimitationsremainwith a relationshipbetweenan adjustmentto
Manning'sn and sinuosity.

1) lt cannotaccountfor the variationof the adjustmentwithstageand radius


of curvature.

2l lt cannotaccountfor the effectsof cross-sectiona!


shape. This can be
significant,as shown by the SERCresults : the ratio of total to friction
lossesin termsof Manning'sn for the 1.37sinuositychannelwas 1.078
with a trapezoidalsectionand 1.22with a pseudo-naturalsection.

3) An adjustedn value is usefulfor riversand other channelswith fairly


uniformplanformswhich can be characterisedby sinuosity. In many
casesthe planformis irregularand it wouldbe preferableto accountfoi
lossesin individualbendsseparatelyin non-uniformprofilecomputations.

The relationshipbetweenbend radiusand sinuosityis probablynot highly


variablein naturalrivers, however,and this may not be cause for concern.
The samecommentappliesto artificialchannelsdesignedin accordancewith
regimerelationships.

Chang'stheorycouldbe appliedlo addresstheseissuesdirectly. The form


used in this study alreadyaccountsfor stage and bend radius effects. The
completeform (Chang,1984)wouldaccountfor cross-sectional shapeeffec{s
(but probablynot for varialionsof cross-sectionalong a reach). lt would not
be necessaryto simulatethe flow througheach bend using the full theory.
Rather,the theory could be used to generategeneralconectionsto n, or
preferablyto the Darcy-Weisbach f, to accountfor these effects. lt could also
be used to developa generalrelationshipfor the loss coefficientfor single
bends.

4.6 Summary
The effect of bends on flow resistancein open channels has been
investigated. Laboratorydata collected from meanderingchannels was
analyzedto showthat lhe meanderingplan form increasesthe resistanceto
flow comparedto equivalentstraightchannels.

A literaturesearchwas carriedout to identifythe importantprocesseswhich


inducethis extraflow resistance. The main sourcesof flow resistancein a
channel bend are: bed friction; increasedbed friction due to secondary
currentsand internalenergydissipationdue to increasedturbulenceinduced
by secondarycurrents. The flow resistancein a bend depends on bed
roughness(f, C, n etc); flow depth(y); bend radius(r.); lengfihof bend (l) or
angleof bend(0, | = r" 0) and the cross-sectional
shapeof the channel. Flow
resistancein a set of meanderbends is likely to differ from the resistance
inducedby a singlebendin an otherwisestraightchannel.This is due to the
interaction(growthand decay)betweenthe secondarycurrentsinducedin the
individualbends.

Variousmethodswhichaccountforthe extraflow resistancewere identifiedin


the literatureand a selectionof methods were applied to the available
laboratorydata. The methodswere evaluatedby comparingthe meanerors
in predicteddischarge.The SCS methodwas foundto giveacceptableresuhs
for most practicalpurposeseventhoughit does not accounlfor the impodant
mechanismsexplicitly. An improvedversion of the SCS method was
formulatedto removethe undesirablestepfunction(LSCS)and this linearized
version gave better predictions. Although these methods, which adjust
Manning'sn basedon the channelsinuosity,gaveacceptableresultsthey are
empiricaland their generalityis not assured. Chang'sapproachin explicitly

s8 329 07/07/93
modelling the resistancesdue to secondarycurrents combined with bachryater
cahulations along the channel is based on sound theoretical considerations.
This approach is applicable to both single bends and series of meanders.

5 Formulation of the procedure for overbank


meandering channels
The mainobjectiveof the studywasto developa procedurefor calculatingthe
@nveyance of a meandering channelduringoverbankflow. The mecfianisms
whichaffectthe @nveyancecapacityof meanderingchannelswereidentified
in Chapter3 and are summarizedbelow. A further literaturesearch was
carriedout to identifythe meansby which other authorshave acc"ounted foJ
these mechanisms. Armedwith this knowledgeof physicalprocessesand
modellingtechniquesit was then possible to decide on an appropriate
approachto be followedin developinga new procedure.

5.1 lmportantmechanisms
The internalstructureof currentsduringoverbankflowshasbeenfoundto be
highly complex. The available laboratorydata has been reviewed in
Chapter3. The mostimpodantobseruations are:

1) The longitudinal
velocitiesbelowbankfulltendto followthe mainchannel
side walls while the floodplainvelocitiesare generallyin the valley
direction. Thus the floodplainflows pass over the main channeland
inducea horizontalsheartayer.

2) The energylos5dueto secondary currentsin the mainchannelis greater


than for an equivalentsimplechanneland the currentsrotatein the
oppositesensecomparedto inbankflows.

3) Fluidpassesfromthe mainchannelontothefloodplainandbackintothe
mainchannelin the followingmeanderbend. Hencethe proportionof
dischargepassedby the mainchanneland flood plainvariesalong a
meanderwavelength.Thesebulkexchangesof fluidbetweenslowand
fast movingregionsof flow introduceextraflow resistance.

4) Flowson the floodplainoutwiththe meanderbelt are usuallyfasterthan


those within the meanderbelt. lt would appear that the extra flow
resistanceinducedby the meanderingmain channel has a relatively
smalleffecton,theouterfloodplain.

5.2 Methodsavailablein the literature


Toebesand Sookv(1967)accountforthe interactionlossesby separatingthe
mainchannelandfloodplainflowsby a horizontal planeat bankfulllevel.The
apparentshear on this plane is accountedfor by adding a solid boundary
equivalentto the wettedperimetersof both flow regions. The dischargesin
the two regionsare lhen calculatedseparatelyand added. Experimental data
wereobtainedfrom small-scalerectangularchannelsand thesewere usedto
evaluatethe solid boundaryaddition. The additionwas found to vary in a
rather complex way with overbankflow depth, main channel depth, and
channelgradient,butno general,practically
usablerelationshipwasproposed.

Jamesand Brown(19771proposed accountingfor the interactionlossesin


straightand meanderingcompoundchannelsby adjustingthe value of
Manning'sn. Fromlaboratorytestresultstheydeveloped an adjustment to the
bankfulln value,dependenton relativeflow depth and the ratio of floodplain
widthto main channelwidth. The adjustedn value is then appliedto the

44 sR 329 07/07/93
cross-sectionconsideredas a single channel. Their experimentswere
conducted mainly with straight channels, however, and the data for
meanderingchannelsare very limited.

Yen and Yen (198i1)also consideredthe compoundsection as a unit and


treatedthe main channelas a resistanceelement. They proposeda Darry-
Weisbach type resistancecoefficientwhich accounts for expansion and
contractionlossesinducedby the rnainchannel.This modeldoesnot accrount
for flow in the main channel,and dependson empiricalinformationobtained
for closedconduitswhich is unverifiedfor channels.

Ervineand Ellis (1987)also proposeddivisionof the cross-sectioninto three


zones, viz. the main channelbelow bankfull level, the floodplainwithin the
meanderwidth,and the'floodplainbeyondthe meanderbeh. They identified
the main sources of energy losses in each of these zones. In the main
channeltheseare :

1) frictionon the wettedperimeter,

2l boundary resistance due to transverse shear and intemal friction


associatedwith secondarycurrentsinducedby the meanderbends,

3) the turbulentshearstressgeneratedby the velocitydifferencebetween


the mainchanneland the co-linearcomponentof the floodplainflow at
the horizontalintedaceat bankfulllevel,and

4l form resistanceassociatedwiththe undulatingriffle-poolsequence.

Overthe floodplainwithinthe meanderbelt the main sourcesof energyloss


are:

1) frictionon the wettedperimeter,

2l expans'nnof the flow as it entersthe mainchannel,and

3) contractionof the flow as it re-entersthe floodplain.

The only lossoverthe floodplainbeyondthe meanderbelt is due to frictionon


the wettedperimeter.

Ervineand Ellisproposeda modelforpredictingstagedischargerelationships


by quantifyingthe more impodantof these loss mechanisms.Frictionlosses
are estimatedusingthe Darcy-Weisbach equationwith the frictionfactorgiven
by the Colebrook-White equation.Lossesassociatedwith secondarycurents
in the main channel are estimatedusing the method proposedby Chang
(1983) for fully developed circulation in wide, rectangularchannels.
Subsequentexperimentalobseruationshaveshownthe secondarycirculation
to be generallyoppositein sense for overbankflows comparedwith inbank
flows. This is becauseit is drivenby the horizontalshearat the bankfulllevel,
ratherthan by centripetalacceleration.Chang'smethodwas derivedfor the
inbankmechanism,and is thereforeinappropriate for overbankcases. Ervine
and Ellisaccountfor the growthand decayof secondarycurrentsby applying
only halfof the headlosspredictedby Chang's1983model.

Expansionlossesfor the floodplainflow are determinedby applicationof the


force-momentumprinciple,and contractionlosses by using loss coefficient
valuespresentedby Rouse(1950)and used by Yen and Yen (1983). The
losses in the main chartnelassociatedwith the shear across the horizontal
intedace and with pool-riffle undulationwere considered minor and not

SR 329 07r!7/93
accounted for. They applied the model to the experimental conditions of the
US Army Corps of Engineers, Watenvays Experiment Station (1956) and
Toebes and Sooky (1967) and producedfairly accurate predictions.

The method is summarized below.

Total discharge

Qr=Qr+Qr+Q.+Qo (5e)

The zonaldefinitions
are shownin Figure17.

Mainchannel

Qr = ArVr (60a)

v1
2g(S./s)Rl
(tl | 4) + ((2.86 tl + 2.a7 f,) / (5.s65 + ff))

WhereA, is the areaof the mainchannelV, is the meanvelocityin the rnain


(R,/rJ*
J*
channel,r" is the bendradiusof curvature,S, is the valleyslopeand s is the
sinuosity.

Inner flood plain

Q. = ArVt (61)

v2 =[
2g(S",*")
( t 2 1 4 ) * ( W . - B s ) / y z + s s i n ' z ( O J( ( 1 - y r i

Where$ is the widthof the innerflood plain, B is the width of the main
(yr+h))'z+K")
J"
channel,y. is the depthof flow on the floodplain,h is the bankfulldepth,0.
is the meananglebetweenthe mainchanneland the valleycentrelinesand
(" is the contractioncoefficient.Thevaluesof contractioncoefficientgivenby
Rouseare listedin Table16.

Outer flood plain

The flows (4, and Q.) are calculatedassumingonly bed friction with the
divisionlinesomittedfromthe definitionof the wettedperimetersand the flood
plainslopeSois usedin the calculation.

Greenhil!(1992) hastriedvariousdifferentmethodsof calcutating


discharges
for a selectionof the SERCFCFdata (tests26, 31 and 39). No attemptwas
madeto identifyor modelindividualloss mechanisms and the methodsare
based on dividingthe channel into the four zones and calculatingthe
dischargein each zone assumingonly bed friction. The two best methods
Greenhill4and Greenhill5)havebeenconsideredhere.

The mainchanneldischarge is calculated


assumingthatthe horizontaldivisio
is includedin the wettedperimeterof the main channeland the innerflood
plainzone. Greenhill'smethod4 appliesvedicaldivisionsat the meanderbelt
edgesand method5 is basedon divisionlinesinclinedoutwardat 45o. These
divisionlinesare includedin the wettedperimeterof the innerfloodplainzone
but not the outerzones. The mainchannelhydraulicslope(S"/s)is usedwhen
calculating the mainchannelandmeanderbeltdischarges whilethefloodplain
gradient(S.) is usedin calculating lhe outertloodplainflows.

sR 329 07/o7l93
5.3 Approachto conveyanceestimation
The possibleformsof analysiswere constrainedby the time availablefor the
study, the specificrequirementsof the NRAand the amountand type of data
available.

The NRA requiredthat the resultingmethodsshould be applied by hand


calculation.lt mustbe recognisedthatflow in compoundmeanderingchannels
is very complexand the developmentof methodsto analyzeit accuratelywill
probablyfollowdirectionsthat are highlycomputational.A hand calculation
methodis unlikelyto be conpatiblewith this developmentand shouldnot be
viewedas a contributionlo it. By disqualifyingthe mostpromisingavenuesfor
complete description of the processes involved, a hand method must
compromiseaccumcyand be limitedto first estimateapplications.The NRA
also requiredthat the methodsbe designoriented. They shouldthereforebe
developedin termsof physicalparameterswhichare meaningfulin a design
context. For example, the dependenceof channel capacity on design
variables,such as cross-sectionshapeand size, shouldbe fairly explicit.

The SERC Phase B tests were limited to just two different planform
geometries, withsinuositiesof 1.37and 2.04. The PhaseA tests,carriedout
in straightchannels,representthe limitingcase of sinuosity1.0. This wide
rangeof sinuositiesis suchthat it wouldbe unreasonable to expectto be able
to interpolateflow characteristicsbetween them. This makes a purely
empirical,descriptiveapproachunrealistic,as it could be appliedonly to new
situationswhichare very similario the experimentalones.

To ensuregeneratity of the designmethods,it was decidedto basethemon


conceptualmodelsof the physicalprocessesinvolvedin dissipatingenergy
and determining flow structure.The SERCdatawas usedto quantifythese
processes,in termsof geometricand fluid state parameters.This involved
theoreticaland empiricalformulations. The relative importanceof the
individualprocesseswere expectedto vary with the scale of the physical
system,and alsowiththe flow condition.Separation and individualtreatment
of the processesaccountedfor the effectsof thesevariationson the required
predictions(of stage-discharge for example)betterthan if these
relationships,
weremadein termsof the geometriesand fluidstateparametersdirectly. The
approachalso has the advantageof beingable to includedata from different
sourcesobtainedunderdifferentconditions,and allowingincorporationof new
resultsand analysesas they becomeavailable.

The divisionof the channelinto four zones as proposedby'Eruine and


Ellis (1987)was adoptedas the mostflexibleapproach.The stage-discharge
relationshipfor a compoundmeanderingchannelwill be predictedby dividing
the cross-sectioninto zonesand calculatingthe zonaldischargesseparately.
The divisionwill be by a horizontalline at bankfullleveland a veilical line on
eitherside of the meanderbelt. This approachalso recognisesthe limited
scope of the present investigationand allows for improvedmodels to be
substitutedfor the var'rouszonalcalculationsin the future.

5.4 Formulationzone 1
The flow mechanisms in this zoneare complex,and havebeendescribedby
Willetts(1992),for example.The majormechanisms for energy
responsible
dissipationare:

1) friction,

2l secondary drivenby theshearimposedby the floodplainftow,


circulation

47 SR 320 O7r!7/93
3) the apparentshearstresson the hodzontalintedaceassociatedwith the
gradientof collinearvelocitycomponentsacrossit,

4l the bulk exchangeof waterbetweenthe main channeland the flood


plain.

Losses associated with variations in cross-sectiongeometry and flow


separationhave been shown to be insignificantfor the conditionslikety to
occur(Kazemipour and Apeft,1979,1982and 1983).

It was originallyintendedto developphysically-based deterministicnpdels to


accountfor the effectsof the variousloss mechanismson stage-discharge
relationships.This hasprovednotto be possible,at leastfor the mainchannel
zone,owingto lhe currentlackof understanding of the mechanismsand their
effects. An empiricalapproachhasthereforebeen resodedto, basedon the
Phase B data and a rational selectionof dimensionlessvariables. The
procedureis to calculatethe bankfulldischarge(Qo,)using an appropriate
methodfor inbankflows,and then to adjustthis to accountfor the effeclsof
overbankflow. The bankfulldischargeincludesallowancefor the effectsof
bendlosses.Thiswas usedratherthanan equivalent straightchannelvalue
to separatethe inbankbendlossesfrom the ultimateadjustmentfactor. This
will allowfuturedevelopments in inbankflowassessment to be incorporated
Also, it is likely that in some designapplicationsinbankstage-discharg
measurements will be availablefor the specificsite,and thesecan then be
used to evaluateQ" directly.

Dischargesin this zone have been obtainedby integrationof the velocity


magnitudeand directionmeasurements taken in some of the Phase B
experiments.The relevantexperiments and integrateddischargevaluesare
listedin Table17. The dischargeswerefoundto varyalongthe channelin a
way consistentwith the descriptive
obseruations
repoiledby Willets(1992),
Figure18.

A studyof Figure18 showsthat for both the 60" and 110ogeometries the
dischargesvary alonga meander,beingmaximumat the bendapices(2 X /
L = 0.0, 1.0)and minimumat somepointin between.Figure18Ashowsthat
cross-sectionshape does not affect the distributionstrongly, with the
trapezoidaland naturalcases givingsimilarvariationsof discharge. Figures
188 and 18C showthat whilethe roughnessof the floodplain mayaffectthe
magnitudesof the mainchanneldischargesit doesnothavea significanteffect
on the flow distribution. The effect of channel sinuosityis apparentfrom
Figures188 and 1BC. The moresinuouschannelwasfoundto havea much
wider variationin mainchanneldischargeat simihr depthsconparedto the
lesssinuouschannel.For exampleat a flowdepthof 200mmlhe 110omain
channeldischarge variedbetweenabout0.4and 1.3of the meanwhileforthe
60' mainchannelthe variationis between0.8 and 1.2of the mean. The effect
of depth is more pronouncedfor the more sinuouschannel. The 60omain
channeldischargesvary betweenabout0.8 and 1.2 of the meanfor all three
depthswhilefor the 110'mainchannelthevariationwasbetween0.9and 1.1
at low depth(165mm)and 0.3 and 1.3 at highdepth(200mm).

Thesevariationsare ignoredin this analysisas neitherthey nor their effects


will be explicitlyaccountedfor in the stage-discharge
predictions;the values
listedin Table 17 are averagesof the integrations at all the measuremen
sections,weightedby the channellengthsrepresented by the sections.

The main channel bankfulldischargeswere not measuredduring the


expedments,and have been determinedindirectly. For the trapezoidal
channel,the modifiedversionof Chang's(1984)methodfor accountingfor

48 sR 329 07/O7l93
bend losses has previously been found to predict the stage-discharge
refationshipvery accurately(-1.767"averageerrorover all measuredvalues).
It was thereforeused to predictthe bankfulldischarge. The stage-discharge
relationshipis shown in Figure 19. No method has yet been found rvhich
predicts the stagedischarge relationshipssufficiently accurately for the
pseudo-natural channels.The bankfulldischargeswere thereforedetermined
for these cases by graphicallyextendingthe measured stagedischarye
relationships,as shown in Figures2O and 21. The bankfulldischargesfor
eachchanneltypeare also listedin Table17.

The ratiosof main channeldischargeto bankfulldischarge(OQ" = Qr/) are


ploftedagainstflood plainflow depth(yr)on Figure22. This showsthat as the
waterlevelrisesabovethe floodplain,the dischargeinitiallydecreasesbelow
the bankfullvalueand then graduallyrisesand rnayexceedthe banldullvalue
at high stages. The relationshipbetween main channel dischargeand
overbankflow depthis clearlyaffectedby the channelcross-sectiongeometry,
the channel sinuosity (s) and the flood plain roughness. lt is obviously
desirable to express these characteristicsin nondimensionalterms and
appropriatemeasureshave been selected. The flow depth is normalizedby
the hydraulicdepth of the main channelat bankfull,i.e. A/8, whereA is the
cross-sectionalarea and B is the sudacewidth. This has beenchosenrather
than a flow depth or hydraulicradiusbecauseit probablyvaries least along
naturalchannels and will requirethe leastfield surveyinformation.The data
have been replottedin Figure23 in terms of the dimensionlessflow depth,
yJWBI ( = y'). The cross-sectiongeometryis characterizedby the ratio of
surfacewidth to hydraulicdepth. This is a physicallymeaningfulparameter
becauseit representsthe ratioof the areaon whichthe apparentshearstress
on the hodzontalinterfaceis appliedto a measureof the volumeaffected.
Expressedas B2lAit is also a shape factor, describingthe deviationof the
channelcross-sectionfrom square. The flood plain roughnessis expressed
as the ratio of flood plain and main channelDarcy-Weisbach frictionfactors,
tltl (= f). Forthe mainchannel,boththe basicstraightchannelvalueandthe
effectivevalue accountingfor bend resistancewere considered,and the final
resultsfoundto be indistinguishable. Thebasicvaluewill be moremeaningful
to mostengineersand hasthereforebeen used.

Quantitativeinterpretation
of the relationshipsbetweenthe channeldischarge
and the variousphysicalcharacteristics is severelyconstrainedby the arnount
of dataavailable.In rnostcaseseffectsare presentedby only two data points.
The exclusiveuse of linear functionsto describethe relationshipsin this
analysisis a consequenceof the lack of data; it is unlikelythat the pr@esses
are actuallylinear.

Figure23 suggeststhat for any particularchannelthe relationshipbetween


Q'/ and y' can be representedby two straightlines. At low ovetbank$ages
the slope of the line is negativeand not appreciablyaffected by channel
geometry,sinuosityor flood plain roughness. At higher stages the slope is
positiveand both the slope and positionof the line are atfectedby these
characteristics.

The straightline describingthe variationof Q,/ with y' at low overbankstages


mustobviouslypassthroughthe point(1.0,O.O),whichdefinesthe constant
in the relationship.The slopeis definedby the four pointsat y' approximately
equalto 0.2. Thesepointsare very closetogetherahhoughthey represent
widely differentconditions,suggestingthat the variation is not appreciably
affectedby these conditions.The effectof sinuosityappearsto be similarto
that for highervalues of y', but there are insufficientdata to distinguishthe
effectreliablyand a commonslope has been assumed. This was calculated

49 sR 329 07/07/93
as the averageof the slopesfor all four data points. The variationfor low
slages is then definedby

Q,'= 1.0 - 1.69y' (ffi)

Foroverbankstageshigherthany' equaltoapproximately 0.2, the relationship


between Q'' and y' is more complex and is cleady affected by channel
geometry,sinuosityand flood plain roughness.Thesecharacteristicshad to
be quantifiedfor all data points. Bothchannelgeometry(as representedby
the ratioB?A)and sinuositywereconstantin all experimentsfor eachchannel
type, but the flood plainfrictionfactorvariedwith flow depthfor both smooth
and rod-roughened experiments.As describedbefore,flood plain roughness
was ac@untedfor in termsof the ratio(/) of the Darcy-Weisbach frictionfactor
for the flood plain (fr) and the equivalentstraightchannelvalue for the main
channel at bankfull stage (f"). The main channel and smooth flood plain
valueswere calculatedusingthe relationshipfor snpoth channelsderivedby
Ackers(1991)fromstraightchanneldata, Equation1. Forthe rod*oughened
flood plain, values were obtainedfrom the proceduredevelopedby Ackers
(1991)and summarizedin Appendix5. The variablevalues are listed in
Table18.

betweenO,', y', B?A, s and / for valuesof y' greaterthan


The relationship
about0.2wasdetermined in sevendifferentways. Theseare discussedin the
followingparagraphs.

Method1

As thereare no morethantwo data pointson the curvefor eachchanneltype


on Figure23, it is assumedthat all relationshipsare linear. The basic
relationship
is

Q,'= my' * (64)


"
in whichm is the slopeof the line and c definesits position.Both m and c
may be functionsof B2lA,s and f. lt is impossibleto determinethe effectof
sinuosityon m becauseonlytwo sinuosities wereused(1.37and 2.04),and
only one floodplainflow depthwas usedfor the 2.04 sinuositychannelin this
range of y'. The acceleratingeffect of the apparentshear stress on the
horizontalinterfacemust decreasewith sinuosityand the rate of increaseof
Q,'with y' will be less,implyinga smallervalueof m. However,assigninga
valueof m to a sinuosityof 2-O4wouldbe totallyspeculativewithoutadditional
data,and it was assumedthat it wouldbe the sameas for a sinuosityof 1.37.
It was assumed,therefore,that m dependson B2lAand f only, i.e.

tn = frl (B2lA,( | (65)

Becauseno experimentswere per{ormedwith the trapezoidalchannelwith


rod-roughened flood plains,there is no evidencethat the effectsof B?A and
f on m are not independent,and they are assumedto be so.

The positionsof the lines(and hencethe valuesof c) are ctearlydependent


on B?A, s and fl. The dependenceon B?A can be seen by comparingthe
pointsfor the 60otrapezoidaland pseudo-natural
channelswith smoothflood
plains. The dependenceon sinuositycan be seenby comparingthe pointsfor
the 60'and 110opseudo-natural channelswith smoothflood plains. The
dependence on f can be seenby comparingthe pointsfor the 60" pseudo-

50 sR 329 07/07/93
natural channels with smooth and rod-roughenedflood plains. lt was
assumed,therefore,that

c = c(B?A,s,/) (66)

As for m, there is no evidencethat the effects of B2lA and f are not


independent.No experimentswere done with the trapezoidalchannelwith
differentsinuosities,so there is also no evidencethat the effectsof B7A and
s are not independent. However,the effects of s and f are clearly not
independent, as can be seenby comparingthe pointsforthe srnoothand rod-
roughenedflood plainsfor the 6d and 110opseudo-natural channels: the
effectof rougheningthe floodplainis muchless if the sinuosityis greater. To
accountfor this dependence,f was initiallyomittedfrom the expressionfor c,
i.e. it was assumedthat

c = c(BzlA,s) (67)

Thecombinedeffectof s and/ was accountedfor by a subsequentadjustment


to Q,/.

It can be seen in Table 18 that f is substantially differentat differentflow


depths,evenwhenthefloodplainsarenotroughened andthe physicalsudace
roughness are identical.Forexample,for the 60opseudo-natural channelwith
smoothfloodplains,fl variesby a factorof two overthe rangeof y' tested. lf
the dependenceon f is beingsought,it is thereforenot corect to connectthe
pointsfor eachchannelgeometry,as done in Figure23, becausethey have
different/ values. This difficultywas addressedby adjustingthe positionsof
the pointsso that pointswith identicalfvaluescouldbe connectedto define
the relationships.Pointswiththe samey', s and B?A wereusedto definethe
gradientof Q,/ with / for that channeland y'. Assumingthis gradientto be
constantwith Q,', the positionof a pointcouldbe adjustedto representthe
same/ as anotherpointwiththe sames and B2lAbut differenty'. the line
throughthesepointswouldthenrepresentthe relationship betweenQ,/and y'
for a given channel with constant /. As no data were obtained for the
trapezoidalchannelwith roughened floodplains,the pointsforthe trapezoidal
channelwereadjustedusingthe gradientof Q,/withf as calculatedforthe 60o
pseudo-naturalchannel. Theadjustedpointsand resultinglinearrelationships
are shown in Figure24 andTable 19. This diagramforms the basis of the
dischargerelationship for y' greaterthanabout0.2.

It was assumedthat Equation65 has the form

m = 8r B?A + a"( +a" (68)

One point and the line throughit were selectedto representEquation68 for
eachof the 60'trapezoidal,60opseudo-natural with smoothflood plain,and
60opseudo-natural with roughened floodplainchannels.The averagevalue
of f for eachchannelwas used. Threeequationsfor m werethereforeset up
and these were solvedsimultaneously to determinevaluesfor a1,ar and a..
The resuftingequationfiorm is

m = 0.0147B2lA+ 0.0320f + 0.169 (69)

The sameapproachwasfollowedto evaluatec, with Equation67 assumedto


havethe form

c=b.rBzlA+brs+b, (70)

51 sR 32g 07/07/93
The same points and lines as used to define m were again used to set up
three equations for c, which were solved simultaneouslyto determine values
for b1,b., and br. The resultingequationfor c is

c = 0 - 0 1 3 2 B . 2 1 A0-. 3 0 2 s + 0 . 8 5 1 (711
The initialadjustment is thereforegivenby
to bankfulldischarge
l
Q/ = ( o . o 1 4 z B z t+A o . o 3 2 o f + 0 . 1 6 9 ) y '

+ 0.013?B?A - 0.302s + 0.851 (721

Equation72 does not accountfor lhe joint effectof s and f, on c. A further


adjustmentwasderivedby calculatingthe ratiosof measuredQ,/to the valuei
calculatedby equation(72),and relatingthemto s and f' The predicted(4,/o)
and measurd (4,1 valuesfor thoseexperimentsin the appropriaterange6f
y' are listedin Table20.

The adjustmentrequiredto the dischargeratio predictedby Equation72 is


pfottedin Figure25. The data for high valuesof ( aretoo sparseto infer a
variationwiths. lt wasassumedthatthe adjustment varieslinearlywithf and
linearregression was usedto obtainthe relationship

q?O,/o=K=1.o7- 0.0698( Qg)


factorfor y' greaterthan about0.2 is then givenby
The adjustment

Q,' = (my' + c)K

with m = O . O 1 4 7 B z l A + A . c F l 2+ (0 . 1 6 9

Q = 0.0132B2lA - 0.302 s + 0.851

K = 1.07 - 0.0698fl (74)

Method2

The second adjustmentin Method 1, representedby K, was intendedto


accountfor the interdependence of the effectsof s and f on Q,/. As it turned
out to be a functionof f only, whichis accountedfor in m, it needreallyonly
be appliedto c. The adjustmentto c (as predictedby Equation71) was
derived by calculatingthe ratios of the requiredvalues to these predicted
values,and relatingthemto s and f. The valuesof c requiredwerecalculated
using Equation64 with m given by Equation69. The predicted(c") and
required(c) valuesfor those experimentsin the appropriaterangeof.y' are
listedin Table21.

The adjustmentrequiredto c as predictedby Equation71 is plottedagainstf


in Figure26. Again,the data are too sparceto infera variationwiths, and the
followinglinearrelationship
withf was obtainedand is assumedto applyfor
all sinuosities.

cy'co= 1.14 - 0.136f (75)

The adjustmentfactor for y' greaterthan about 0.2 is then given by

Q'' = ty' + Kc

52 sR 329 07107/93
with m = O . O 1 4 7 B z l A + O . O 3 2 (+ 0 . 1 6 9

c = 0.0132B2lA - 0.302 s + 0.851

K =1.14-0.136/ (76)

Method3

Methods1 and 2 werederivedby simultaneous solutionof one equationfor


eachof the channeltypes.Someof the datawerethereforenot used,and a
nroreaccurateformulationmightbe obtainedfroma regressionanalysison all
the data. In Method3 a straightforwardmultiplelinearregressionanalysis
was performed,givingthe followingrelationship.

Q/ = o.+egy' + o.o392B1A- 0.0645/ - o.tgss +0.382 (77')

Method4

Figure24 presentsa set of sfraightlines for y' greaterthan about O.2,the


slopes and positionsof which appearto dependon B?A, / and s. In this
methodmultiplelinearregression analyseswereperformedseparatelyon the
slopesand interceptsof the lines.

Becausethereis onlyonedatapointfor the 110ocrossoveranglechannelfor


each roughnesscondition,no slope could be determined. Slopes were
thereforeknown for only one sinuosityand consequentlyno variationwith
sinuositycouldbe considered.lt wasthereforeassumedthat the slopesof the
linesdependonlyon B2lAand /. Usingthe dala for the channelswith a 60'
crossoverangle,the followingrelationship
for slopewas found.

m = 0.0183B'alA+ 0.0128/ + 0.159 (78)

The intercepts(c) of the lines were assumedto dependon B?A, / and s.


Valuesof c for eachlinewerecalculatedusingEquation64, withQ,/as given
in Table 18 and m measuredon Figure24. For the 110ocrossoverangle
channelsm was calculatedusingEquation78. The resultingrelationshipfor
c is:

c = 0.007688'z/A- 0.0708( + 0.cr.72s + 0.435 (7e)


The adjustmentfactorfor y' greaterthanabout0.2 is then givenby

o i = m y '+ c
with m = 0.01$B?A + 0.0128/ + 0.159

c = 0.0076882lA- O.O7o8(- O.OAZZs+ 0.435 (80)

Method5

The slopesof the linesfor y' greaterthan about0.2 on Figure24 do not vary
greatly,and a simplerequationfor Q,,/wouldresultif the slopewereassumed
constant. lt was assumedthat the averageslope(m = 0.433)appliesto all
linesand valuesof c werecalculatedusingEquation64 and this value. The
relationshipbetweenc and the channelvariableswas reanalysedand the
relationshipfor Q,/ is thengivenby

Q/ = 0.4my' + 0.00715B2lA- 0.0532/ + 0.0246s+ 0.45e (81)

53 sR 32{t 07/07/93
Method6

There are only two data pointsfor lhe 1100crossoverchanneltor t' greater
than 0.2. The pointfor the experiment with roughened floodplainssuggests
that sinuosityhas no effect on Q,/, while the point for the experimentwith
smoothflood plainssuggestsa significanteffect. Methods1 to 5 attempted
to reconcilethis information.

In Method6 it wasassumedthatQ,/is independent of sinuosity,as suggested


by the experimentswith roughenedflood plains. Afthough intuitively
unappealing, there is some justificationfor this assumption. Sinuosityis
accountedfor in the estimateof Q61, and the implicationis that the magnitude
of mainchannelenergyloss associatedwith meanderingis similarfor inbank
and overbank flows, although it is recognisedthat the mechanismsare!
radicallydifferent. The validityof this assumptionneedsto be investigated
usinga dataset witha widerrangeof sinuosities.

This methodis thereforesimilarto Method4, but the regressionanalysisfor


c excludedthe data for the 110'crossoverchannel,resultingin a different
formulation.The adjustment factoris givenby

Q'' = my' + c

with m = 0.01$ B'alA + 0.0128/ + 0.159

c = o . 0 o 8 8 8 9 2 l -A 0 . 0 7 2 9 { + 0 . 4 0 2 (82)

Method7

In this methodit was assumed(asfor Method5) that m is constant,and that


(asfor Method6) c is independent
of sinuosity.Valuesof c weredetermined
as in Method5 andthe regressionanalysisrevised.Theadjustmentfactoris
then givenby

Qr' = 0.433y' + 0.0182B2lA - 0.0614( + o.4o2 (83)

Evaluationof Methods

The erors in reproducingthe data by each of the methodsare listed in


Table22. The selectionof the most appropriatemethodor methodswas
based on the magnitude,nature and distributionof errors, with some
considerationof simplicity. Enors are most acceptableat high values of y'
becausethe mainchannelcontributionto total dischargebecomesrelatively
less significantas stageincreases.Errorsare consideredmoreacceptableat
highsinuosities as the latterare morecommon
thanat moderatesinuosities,
in design applications. Negative enors are preferableto positive errors
becausethey wouldintroduceconseryative underestimation of mainchannel
conveyancein designapplications.

On the basis of the above criteria, Method 2 was selected, its worst
peformanceis for highy' and highsinuositywith roughflood plainsand the
error for the latter conditionis negative. lt should be noted that the
expedmental floodplainroughnesswas e)dreme,and the errordecreasesfor
smootherfloodplains.

Summary

The procedureis to calculatethe bankfulldischarge (Q"), and thento adiust


thisto accountfor the effectsof overbankflow. Thebankfulldischargecan be

sR 329 07107/93
estimatedusinginbankflow methodsor obtainedby measurement, if possible.
The hydraulicslope which controlsthe flow in the main channelzone (S) is
relatedto the flood plainor valleyhydraulicslopeby the channelsinuosity,(ie
S = S. / s). lt shouldbe notedthat S" can eitherbe a groundslopeif uniform
flow is assumedor a water surfaceslope.

The adjustmentfactor was determinedfrom the SERC FCF Phase B data.


Actual dischargesin this zone were obtained by integratingthe velocity
magnitudeand directionmeasurements taken in some of the experiments.
The ratioof actualto bankfulldischargedefinesthe adjustmentfactor,Q,/.

Q'/ was foundto dependon:

1) the flood plain flow depthat the edgeof the main channel(yJ;
2l the channelsinuosity(s);
3) the cross-sectiongeometryand
4) floodplainroughness.

These characteristicsare representedby dimensionlessparameterswhich


werechosenas beingbothmeaningfuland easyto measure.Theflowdepth
is normalized
by the hydraulicdepthof the mainchannelatbankfull,Equation
84, whereA is the cross-sectional
areaand B the sudacewidthof the main
channelatbankfull.

y' = yrl (NB) (84)

The cross-sectiongeometry is characterizedby B2lA. The flood plain


roughnessis expressedas the ratioof floodplainand mainchannelDarcy-
Weisbachfrictionfactors,i.e.

f = tzltt (85)

The Darcy-Weisbachfriction factor can be calculated using the Colebrook-


White equation. lf Manning's n is used then f is relatedto n by

8g n2 (86)
f=
--1-

The ratiof can thereforealso be expressedin termsof Manning'sn

f = (n"/n,)21R,/R.1to (87)

The relationshipbetweenthe adjustmentfactorand these variablesis shown


schematically in Figure26b. Thisshowsihat the mainchanneldischargeis
initiallyreducedas stage rises above bankfull,and that this reductionis
independentof channelcharacteristics.At higher stages the discharge
increaseswith stage at a rate which depends strongly on B?A, s and f.
Variousexpressionsfor the relationshipat high stages were derived by
differentmethods. The relationshipderivedby method2 abovewas chosen
as the best.

Thus the variationin main channeldischargewith overbankstage can be


accountedfor by choosingthe adjustmentfactorto be the greaterof

Q , ' = 1 . 0 ' 1 . 6 9y ' (88)

oi = tny' + Kc

55 sR 329 o7l07/93
with m = O . O 1 4 7 B 2 | A +0 . 0 3 2 / + 0 . 1 6 9

c = 0.0132 B2lA - 0.302 s + 0.851

K =1.14-0.136f (8e)
and the correct flow in zone 1 is given by

Qr = Qu Qr' (90)

5.5 Formulationzone2
This sectiondescribestwo alternativemethodsfor predictingthe dischargein
the inner flood plain zone. The first methodattemptsto account for the
principalloss mechanismsusingphysically-based deterministicformulations.
The formulationsare basedon a very simpleconceptualmodelof the loss
mechanismsand requireempiricaladjustmentto accountfor the additional
complexities involved.

The secondmethodis purelyempiricaland followsan approachsimilartothat


used for the mainchannelzone. A basicdischargeis calculatedassuming
frictionlossesonly,and this is then adjustedto accountfor the effectsof flow
interactionwiththe mainchannel.The adjustmentis basedon dataobtained
from the SERC PhaseB experiments and data providedby ProfessorB B
Willetts(personalcommunication) obtainedfromexperiments conducted under
his supervisionat the University of Aberdeen.

5.5.1 Expansioncontraetion
model
The major energy loss mechanismsin the inner flood plain zone have been
identifiedpreviouslyby other researchers(for example, Ervineand Ellis 1987,
McKeoghand Kiely 1989)as

1) friction on the wetted perimeter,

2) expansionof the flow as it enters the main channel,and

3) contractionof the flow as it re-entersthe floodplain.

The energy loss due to friction (trf over one meander wavelength (L) can be
estimatedusing the Darcy-Weisbachequation,

t2Lv:
[ .' = (el)
89Rz

in which t2 is the Darcy-Weisbach frictionfactorfor the innerfloodplain,


g due to gravity,
is the acceleration
% is the flowvelocity,and
R2 is the hydraulicradius.

area to the
The hydraulicradiusis definedas the ratio of the cross-sectional
wettedperimeter.The innerfloodplainzoneis rectangular and the cross-
so
sectionalarea is the productof the innerflood plain width (WJ and the flow
depthon the flood plain(y.). The wettedperimeterincludesthe flood plain
sudaceonly,and not the horizontalplanedividingthe innerflood plainand
mainchannelzones. By considering the areasof the floodplainand division
planeover a wavelength,it can be shownthat the effectivewettedperimeter
is the widthlessthe productof the sinuosity(s) and the mainchanneltop

56 sR 329 o7l07/93
width(B). The hydraulicradiusfor frictionlosscalculations
is thereforegiven
by:

R,= g (e2)
" Wr- B s

Expansionand contraclionlossesdependon the pafternof flow acrossthe


main channel,which is complex. The flow expansionis accompaniedby
deviationof the primaryflow directionand entrainmentof some of the flood
plainflow intothe mainchannelalongthe cross-overreach(Jasem,1990).
Thereare alsobulk exchangesof waterbelweenthe mainchannelardflood
plain associatedwith the bend apex regions(Willetts,1992). The model
developedhere assumesstraight flow across a slot in which there is no
transverseflow, and is thereforea very simplifiedrepresentationof the real
situation.A complete,quantitativedescriptionof the interactionsbetweenflood
plain and channelflows would requiredetailedcomputationalmodelling.
Appropriatemodels do not exist at present,nor do the understandingand
quantitativeinforrnationnecessaryfor theirdevelopmentin the shortterm. For
presentpurposes,however,only the energy loss associatedwith the flood
plain - main channel inleractionis to be predictedand the model does not
needto be completeand accuratein all respects. lt is assumed,therefore,
that the magnitudeof energyloss and its dependence on the mainflow and
geometricpropertiesis similarfor the simplifiedand real situations.In fact,
insufficientinformationis availableat presentevento providea good,general
descriptionof energyloss for the simplecase and some broad assumptions
are necessary.

The expansionloss overa simpledownwardstep (as shownin Figure27) can


be estimatedby applicationof energyand momentumequationsbetween
sections1 and 2. This gives

,
n" t) * (1- (y"ry,)")
=lo ,l' ,(T,/t',, \v: (s3)
" l-
tYz+h*Yr )zg
in which h" is the energylostin expansionof the flow,
yr is the flowdepthin the mainchannel,
h is the step height,
V2 is the velocityat Section2, Ftgure27

lf it is assumedthat the watersur{aceis flat and unaffectedby the step,then


yr = ye+ h and Equation93 reducesto

h"=
[*l x
Equations93 and 94 haveboth beenappliedto somedataobtainedby Jasem
(e4)

(1990)andthe differences in theirpredictions


foundto be negligible.Equation
94 is thereforeacceptedas an adequatedescription. Notethat this result is
independentof 0, the inclinationof the downwardstepto the directionof flow.

The flow patternfor contractionover an upwardstep is shown in Figure28.


The step inducesa vena contractaa shortdistancedownstreamof the step,
beyondwhichthe flow expandsto the normalflow conditions.The loss of
energyassociatedwiththispatternis concentratedin the expansionregion,i.e.
betweenSections3 and 4. The contractionloss (h")could thereforebe

57 sF 329 07/0743
describedby an expansionloss equationsimilarto Equation94, i.e.

V.'
h" = (1 -yJy")" (e5)
Ts
Equation95 cannotbe used,however,withoutknowledgeof the contraction
coefficientnecessaryto definethe flow conditionsat Section3. This cannot
be determinedanalyticallyand hasnot beeninvestigatedexperimentally.yen
and Yen (1983)recommended accountingfor the contractionloss betweena
meandering channeland its floodplainwiththe relationship:

h"=K" (e6).
&
in whichK" is a contractionloss coefficientwhichvarieswiththe ratioof flood
plainto mainchannelflowdepths,as givenin Table16. Thesevaluesfor K"
were given by Streeterin Rouse(1950)and are apparentlybased on data
obtainedfromexperimentsin pipesconductedby Weisbachin 18SS.Jasem's
(1990)resuftsfor widerectangular slotsagreewellwiththesevaluesandthey
can thereforebe acceptedas reasonablyaccuratefor free surfaceflows as
well.

Both expansionand Contractionlossescan be expectedto dependon the


widthof the mainchannel(B). The expansiondevelopsover some distance
fromthe downwardstep. lf this developmentis incompletebeforethe upward
step is encountered,then clearly the associatedexpansionloss will be
propodionately
reduced.Incomplete development willalsomeanthatthe flow
contracliondoes not begin from the bed of the main channel,but some
distanceaboveit, andthe associated losswill be less. Flowpatternsfor wide
and narrowchannelsare illustratedin Figure29.

Jasem (1990)measuredexpansionand contractionlossesover slots with


width to depth ratios rangingtrom 2 to 20. His resultshave been used to
dedvecorrectionsto the expansionandcontractionlosscoefficients to account
for widthto depthratio. The ratiosof measuredexpansionloss coefficientto
(1 - y/yr)" are plottedagainstwidth to depth ratio (B/h) in Figure30. The
ratios of measuredcontractionloss coefficientto interpolatedvalues from
Table 16 are plottedagainstB/h in Figure31. As the loss coefficientsare
additivea conectioncould be appliedto both together,and the ratiosof the
sumsof the measuredvaluesto the sumsof (1 - y/yrl" andvaluesfromTable
16 are plottedagainstB/h in Figure32. ln each case linearregress'lonwas
usedto obtaina relationshipbetweenthe corection factorand widthto depth
ratio. These relationshipsare given on the figures. They are remarkably
similar,and it wouldthereforebe mostpracticaltoapplya singleconectionto
the two coefficientstogether,i.e.

= 0.02(B / h) + 0.69
Widthto DepthRatioCorrection (97)

For a channelangledacrossthe floodplainthe widthpresentedto the flow


would be greaterthan B and wouldvary with the angle. Attemptsto refine
Equation97 to accountfor this are not worth while at the currentstate of
knowledge of the processes.

Bothexpansionand contractionlossescan also be expectedlo vary with the


sideslopesof the mainchannel.The effecton the contraction lossshouldbe
particularlysignificantbecauseof the influencethe bank slopemust have on
the contractioncoefficient.Theseetfectscannotbe describedanalyticallyand
no directlyapplicableexperimentalresultscan be found. Chow (1959),

sR 329 07/07/93
however,has presentedresultsobtainedby Formica(1gss)for energykcsses
in hteral expansionsand contractionsin channels.Thesehave been used to
obtainfirst estinntesof the effectsof the transitiongeometries.

Formicameasuredenergylossesacrossan abruptconlractionin width and a


oontractionwith a straighttaper of 30". The energy bss varied considerably
with dischargebut on averagethe loss withthe taperedcontractionwas aDo-r.lt
0.3 times that with the abruptcontraction.lf the contractionloss is assumed
to decreaselinearlywith the cotangentof the side slope,a conectionfunction
can be writtenas

ContractionSide SlopeConection= 1 - (S"/ 2.5) (e8)


in which S" is the cotangentof the side slope. lt wouldbe realisticto set :i
minimumvalueabovezero,say 0.1,to this conection.

Formicaalso conductedexperimentswith an abrupt expansionin width and


expansionswith$raighttapersof 1:1, 1:2, 1:3 and 1:4. For the 1:4taperthe
energyloss was about0.3 timesthat for the abruptexpansion,and decreased
quite uniformlyoverthis range. Assuminga lineardecreaseof the expansion
loss coefficientwith side slope,the correctionwould be

ExpansionSideSlopeCorrection= 1 - (S"/ 5.7) (99)

Again,it wouldbe advisableto set a lowerlimit of, say, 0.1 to this correction.

Bothdownwardand upwardsteps betweenthe main channeland flood plain


extendover a widthof (W - B) over the innerflood plainbetweenconsecutive
bendapices. Overa meandeiwavelengththerewili be two downwardand two
upwardsleps. lf the lossesare assumedproportionalto the widthover which
expansionand contractiontake place, then the losses should be further
correctedby

S t e p L e n g t h C o n e c t=i o 2n ( W r - B l l W " (100)

The head loss over one wavelength associated with expansion and
contraction,\, is thereforeestimatedas

hr-= h" + h. (101)

i.e.

h.=K. v2
q
where

h. = C, C,,o( C"- ( - y)yr)"* C* K.l Y"2| 2 g (102)

in which C"l is the step lengthcorrection


= 2(tlz-B)/W.,

C*d is the widthto depth ratio conection


= 0.028/h + 0.69,

C"* is the expansionside slopecorrection


= 1-s15.7,

59 sR 3an 07/07/93
C"* is the contractionside slope correction
= 1-s12.5,

K" is the basic contractioncoefficient,as given in Table 16.

The total energy loss over one meander wavelength is the sum of the friction
and expansion and contraction losses, i.e. tr, + h.. Each of these rnajor loss
components can be expressed as a multiple of the velocity head, so that

= (f2Lt4R2 + tC) Vjtzg


h + rr,_ (103)

in which l(. is the total expansion-contraction


loss coefficient,as definedby
Equation(102). Underuniformflow conditionsthe total energygradientis
equaltothe floodplainbed gradient,So,so that

h + h = SoL l (104)

Equations(103)and (10a)can be combinedto give

V^=r
'
,rrJ I (105)
* K")
[tr'L)/(4Rr)
Consideringthe complexitiesof the flow mechanismsEquation105couldnot
be expectedto accountfor all the energy losses under all conditions. A
comparisonof the SERCPhaseB and Aberdeendatashowedthe non-friction
lossesto be stronglyinfluencedby the cross-sectional
geometryof the main
channel. The basic model, Equation105, was found to predict stage-
dischargesreasonablywell for the SERC 6f channelsbut to underpredict
dischargequitebadly(errors-20%l for the Aberdeenchannelwith a similar
sinuosity.

Apartfromthe scalethe onlysignificantdifferencebetweenthe two channels


is the crosssectionalshape.Thishasbeendescribedin the zone1 modelby
the factor (B'?/ A) and was assumedto be an appropriatemeasurehere as
well.

It was assumedthat the frictionpart of the model is adequateand that the


non-frictionterm shouldbe adjusted. Also it was assumedthat the error in
total dischargewould most likely arise in the zone 2 model because its
contributionis mostsignificantat higherstages.

The oonectionis basedon the SERC60'trapezoidalchannel(82I A = 9.14)


and the Aberdeen1.4 sinuositychannel(gzI A = 3.84). lt was foundby trial
that the shapeeffectobservedin thesetwo channelscouldbe accountedfor
by multiplyingihe non-frictionterm in equation105 (K")by a factordefinedby

Fr = 0.1 B?A for B?A < to

Fr = 1.0 forB?A > 10 (106)

in whichA is the cross-sectional


areaof the mainchannelbelowbankfull.
The SERC Phase B resuftssuggesta further effect associatedwith rnain
channelsinuosity(s). Application
of the basicmodel(Equation105)to this
data showedthat,for smoothfloodplains,the predictionswere reasonablefor
both the 60o trapezoidaland naturalchannels. The errors in calculated
dischargeoverthe wholerangeof stagesare shownin Figures33 and 34.

60 sR 329 07&7/93
The causeof the high positiveerrorsat low relativedepth (y') is not known.
beingclearlypresentfor the 6d trapezoidaland 11Oo
Theyare not c-onsistent,
naturalchannels,but not for the 60onaturalchannel.

A conectionfor sinuositywas madebasedon the differencein enors for the


differentsinuositiesat highervaluesof y'. The predictionfor sinuosity2.04
needsto be reducedby,about 10ol",whilefor sinuosity1.37 no adjustmentis
required.

As for the shape corection factor it was assumedthat the non-frictionterm


stpuld be adjustedand that the non-frictioncomponentof the zone 2 model.
It was foundby tdal thatthis can be accountedfor by muftiplyingK. by another
factor,definedby

Fz = s|1.4 (107)

The dischargein the innerflood plainzone is thereforegiven by

Q, = WY.V. (108)

with V, givenby

(^\t
v12-|ry,J
=l 2gs"L I (10e)

Preliminaryapplicationsof this npdel showed it to be quite insensitiveto


estimationof the step height,h. lt is recommended
that this be approximated
by the hydraulicdepth(= A/B).

5.5.2 EmpiricalModel
The physically-basedmodeldescribedaboveis unableto accountadequately
for the energylossesin the innerflood plain withoutempiricaladjustment
basedon the SERC PhaseB and Aberdeendata. Eventhen,the enors in
predictingdischargeover a range of overbankstages are inconsistentfor
some geometriesand there is a case for consideringfurther empirical
adjustment. Beeauseof the significantempiricalcontent that would be
required anyway, it would be practically expedient to apply empirical
conectionsto a basicdischargecalculatedin a sirnplerway than as required
by the previousmethod.,

The simplestempiricalapproachmightbe to disregardthe horizontaldivision


betweenthe main channeland innerflood plain zonesand considerthe two
together. This would be physically realistic consideringthe significant
interactionand exchangeof flows obseruedbetweenthe zones (e.9. Wllefts,
1992);the separationof the zones is acknowledgedto be rather adificial.
However, knowledge of the zonal distribution of the discharge, albeit
longitudinallyaveraged,would be most useful in design applications. The
analysis for the main channel zone exposed significant variations of
conveyancewith s{agewhich,ahhoughbasedon limiteddata,are significant.
This inforrnationwouldbe lost in a methodwhichdid not separatethe zones.
Subdivisionof the cross-sectioninto the previously defined zones has
thereforebeen retained.

Empiricalanalysisof the innerflood plainzonal dischargeis compromisedby


the lack of directlymeasureddischarges.Velocitymeasurements were taken
over the flood plains for only five conditions in the SERC Phase B
experiments:at two flow depthsfor each of the 6f trapezoidalandpseudo-
naturalchannels,and at one flow depthfor the 11f pseudo-natural channel.

61 sRSan 07/07/93
lllrtll I

There were no measurementsfor the rod-roughened flood plains. Where


velocities were measured,they could not easi$ be used to compute
dischargesoverthe definedzone. This was becausethe measuringsections
abovethe plain surfacesand abovethe main channeldid not fit togetherto
provide single, continuoussections across the whole zone. There were
thereforeno practicallyusablevelocitymeasurements for lhis zone. Inner
floodplaindischargeswereestimatedby subtractingcalculateddischargesfor
the main channel and outer flood plain zones from the total measured
discharge. The main channeldischargeswere calculatedusingthe method
developedeadier. The outerfloodplaindischargeswerecalculatedassuming
frictionlossesonly, usingthe Darcy-Weisbach equation. The frictionfactors
were estimatedusing the appropriatemodifiedsmoothlaw for the padicular
expedments, see Section3.4.

The use of calculated,ratherthan measured,dischargesfor analyzingthe


innerfloodplainflowshasobviousdisadvantages.The methodfor calculating
mainchanneldischarges is basedon verylimiteddatafromthe SERCPhase
B experimentsonly. As discussedin the sectionon the rnainchannelanalysis,
only a limited range of stages and channel characteristicswere included.
There is, as yet, no evidenceto confirmthat the mainchannelmodelapplies
to the Aberdeenchannels,as has been assumedin this analysis. On the
otherhand,obtainingdischarges by calculationenablesallthestage-discharge
data to be incorporatedin the analysis,ratherthan iust the few conditionsfor
which velocitieswere measured. This makestrends very much easier to
detectand providesadditionalinformalionfor interpretation.

The Aberdeendata were includedin this analysisbecausethey represent


some conditions(an extra sinuosityand differentmain channel geometry)
whichwere not coveredby the SERCPhaseB experiments,and whichwere
shownin the developmenlof the physically-basedmodelto be significant.

The basicdischargefor the innerfloodplainzone was calculatedassuming


frictionto be the onlylossmechanism.lt was furtherassumedthatthe plane
separatingthis zonefromthe mainchannelwouldofferthe sameresistance
as lhe flood plain suface. An analysiswas also done with the separating
plane subtractedfrom the wetted perimeter. This did not reducethe final
adjustmentrequiredand the simplercalculationwas thereforeadopted.

The flow in the innerfloodplainzonecan be expectedto be affectedby much


the same characteristicsas that in the rnainchannel. This is supportedby
inspectionof the variablesappearingin the physically-based model.Thesame
dimensionlessparamelerswere thereforeused. The flow depth was rnade
nondimensional (y') by dividingby the hydraulicdepthof the main channelat
bankfull,andthe cross-sectional shapeof the mainchannelwasaccountedfor
by the valueof B2lA.

The data used in the analysis were the overbank stage discharge
measurements for the SERC PhaseB standardgeometrieswith smoothand
rod-roughened floodplains,and the overbankstagedischargemeasurements
for the Aberdeentrapezoidalchannels. The geometricconditionsfor these
sets are listedin Table23.

For each measuredstagethe actual and basic innerflood plain discharges


were calculaledas describedabove. The ratio of thesevalues (Q/) defines
the adjustmentto be appliedto the basicdischarge. Q./ is plottedagainsty'
for the SERC Phase B data in Figure 35 and for the Aberdeendata in
Figure36. The numberedpointson Figure35 are calculateddirectlyfromthe
integratedmainchanneldischarges usedfor derivingthe predictionmodel;the

62 sB 329 07rc7l93
numbercindicatethe relevantexperiments.Thesepointsprovidecheckson
the accuracyof the rnainchannelmodelfor some conditions.

Both Figures35 and 36 showa clear pattern. For smallvaluesof y' there is
a rapid increaseof Q./ with y', in which no distinc{variationwith channel
characteristics can be discemed. For largervaluesof y', a! decreaseswith
y' nonlinearlyand at a ratedependenton B2lA,s and roughness.The ranges
of y' less than and greaterthan 0.2 were treatedseparately. lt would have
been difficuftto establishrelationshipsbetweenthe variablesfor lhe SERC
data abne becauseonly two sinuositiesare representedand the two values
of B?A are not greatly different. The inclusion of the Aberdeen data
contributedconsiderablesupplementary information.

The relationshipbetweenQr/ and y' tor y' less than about 0.2 is dfficult to
quantifybecauseof the limitedamountof data and their considerablescatter.
Muchof the scattercan be ascribedto the procedureusedto determinethe
zonaldischarge.At low overbankstagesthe innerflood plaincontributionto
total dischargeis very small. As it was calculatedas a small difference
betweenrelativelylarge quantities,enors can be expectedto be significant.
For the SERC Phase B data, shownon Figure35, it is only for the 6d
trapezoidal channelthatthereare sufficientvaluesto definea trend. Datafor
the otherchannelssuggesta decreasein Q./ for lowvaluesof y'' but there are
insufficientpointsto establishthe effec;tsof the channelcharacteristics
on the
trend. The data for the Aberdeenexperiments, shownon Figure36, also
showa distinctand similarlrend, but againthis is not sufficientlywelldefined
to quantifythe effects of channel characteristics. The data for the 1.215
sinuositychannelarepailicularlywidelyscatteredandsomeQr/valuesforvery
low y' are too highto appearon the graph. As it was not possibleto estabtish
multiplecorrelations, a singlestraightline was drawnthroughall the data,
passingthroughthe originto ensurepositiveadjustmentin all cases. This
gavethe relationship

Q.' = 6'o y' ( 11 o )

For valuesof y' greaterthan about 0.2 the relationshipbetweenQr/ and y' is
nonfinearand clearlydependenton B?Aand s. The dependenceon { (=t/t)
is questionable.The curvesfor the 6Cfcrossoveranglechannelswithsmooth
and rod-roughened floodplainscoincidefairlyclosely,althoughthe measured
points suggest that roughnesshas some influence which is opposite at
relativelylow and highvaluesof y'. The value of f variesover the rangeof y'
beingconsideredfrom about1.8 to 0.78for the smoothflood plaincases,and
from about3.1 to 11.9for the rod-roughened flood plain cases. Considering
thesevariationsand the closecoincldenceof the cutves,it wouldappearthat
flood plain roughnesshas negligibleeffect. However,the curvesfor the 11d
crossoveranglechannelswithsmoothand rod-roughened floodplainssuggest
that roughnesshas a very considerableeffect. lt would be extremelydifficult
to quantifythis effectbecauseno two pointshavethe samevalue of f and, if
roughnessis significant,each point actually lies on a different curve. A
comparisonof the SERCand Aberdeendata suggestedthat the curuefor the
rod-roughened caseis consistentand thatthe curveforthe smoothcaseis out
of character

The curvesin Figures35 and 36 can bestbe representedby an equationwith


the form

QJ = ay'o (111)

63 sR 329 07/O7l93
Valuesof a and b were determinedfor each case by plottingthe data on
logarithmicpaperand fittingstraightlinesthroughthem,as shownin Figures
37 and 38. The resultingvaluesare listedin Table24.

Visualassessmentof Figures37 and 38 suggeststhat a dependsstronglyon


s, only slightlyand not consistentlyon B?A,and on f for high sinuosities.As
discussedabove, the dependenceon fl is very difficuftto establish. The
parameterb dependsstronglyon B2lA. The combineddependenceof b on s
and f suggestedby the datafor the 11Oo channelwithsmoothfloodplainsis
again problematicand cannot be accountedfor withoutfudher information.
The slope of the line representingthis conditionwas assumed to be
inconsistent and disregarded.lt couldthenbe assumedthat a dependsonly
on s, and b dependsonly on B2lA.

The dependence of a on s was determined by calculating


the averageof the
valuesof a in Table 24 for each sinuosityrepresented.Theseare plottedin
Figure39 and the relationship can be describedby

a = 1.02 s'o'e15 (112)

The dependenceof b on B2lAwas determinedin the same way, ignoringthe


value for Run 839. The averagevalues of b are plottedagainst B?A in Figure
40. The relationshipfor b is given by

b = -0.91 (Bzlg1-o-ttt (113)

The adjustmentfactor,as predictedby Equations(110)to (113),is plotted


togetherwith the SERC data In Figure41 and with the Aberdeendata in
Figure42.

The dischargefor the innerfloodplainzoneshouldthereforebe obtainedby


first calculatinga basicdischargeusingan appropriateresistanceequation
(e.9. Darcy-Weisbach, Chezyor Manning). For this calculationthe wetted
pedmetershouldbe equalto the widthof the meanderbelt and the friction
factor should be appropriatefor the inner flood plain surface. The basic
dischargeshouldthen be adjustedby multiplying by the lesser of

Q.' = 6.0y' (114)

and

A : = ^ y'o (115)

with
a = 1.02s-o'els (116)

b = -0.81 (B?41'o'o" (117)

5.6 Formulationzones 3 and 4


The importantmechanismswhichaffectdischargein the outer zonesare

1) Friction
2) Shearon the intedaceswithzone2

Unfortunately therewas not enoughdatato evaluatethe relativeimpoilance


of each of thesemechanisms.Howeverotherauthorsworktendsto indicate
thatshearon thedivisionlineswillbe relativelyunimpodant.Henceflowin the
outer flood plain zones is assumedto be solely controlledby friction. The

sR 329 07/07/93
zonaldischargesare calculatedusingan appropriatefrictlonequationwith the
dMsion lines separatingthese zonesfrom Zone 2 excludedfrom the wetted
perimeter.

Q. = &V.

O. = &% (118)

where

Vo =
" t E' f3-11
3
) (11e)
/Y
v.= It nrt. t. l
tt1 )
5.7 Boundaryshear stresses
Boundaryshear stresseswere also measuredfor some conditionsduring
PhaseB of the SERC FCF work. Thesedata have beenanalyzedby Knight
et al (1992)and Lorena(personalcommunication)
and form the basisof the
provisionalrecommendationspresentedhere.

There is no simple,generalmethodfor predictingboundaryshearfor inbank


flows in meanderingchannels,but sweral simulationmodels have been
developedwhichcan be usedfor this purpose(forexample,by Bridge,1992,
and Nelsonand Smith,1989).

For overbankflows, Knight et al have shown that the sectional average


boundaryshearstressin the mainchannelis lessthanwouldoccuratbankfull
stageat all cross-sectionsthrougha meanderwavelength.Sectionalaverage
valuesare insufficientfor designingscourprotection,however,becausethe
distributionsof boundaryshearacrossthe sectionsare not uniformand vary
with flow condition.The measureddistributionssuggestthat duringoverbank
flows the shear stress on the main channel banks rnay be higher than for
inbankflowsat somelocationsthroughihe meander.The shearstresson the
bed, however,is less than for inbankflows. Designshearstressesfor scour
protectionshould thereforebe based on inbank flows for the bed and on
overbankflows for the banks.

Underoverbankflow conditionsthe bank shear stresson the upstreambank


does not exceed1.6 T y" Soin any of the measureddistributions,where y is
the unit weightof waterdefinedby pg (9.81x 103N/ms). On the downstream
banka high,localisedstressconcentration was obseryeddownstreamof each
bend apex, associatedwith the expulsionof waier from the main channelto
the flood plain (see Figure 14). This concentrationis shown in Figure 43,
whichpresentsLorena'splot of contoursof shearslressfor the 2.04 sinuosity
channelwith a flow depth on the flood plain of 50 mm. The concentrations
were centredat pointsbetween6d and 70odownstreamof the apex section
for allthe experimentalconditions.The maximumobseruedshearstressesin
the concentrations approached5 y yz So. The stressconcentrationsare very
localisedand decrease rapidly with distance but, because of the limited
experimentalc-onditions and consequentuncertaintyregardinglocations,they
shouldbe assumedto be more extensivewhen designingscour protection.
The enhancedshear stressesalso extendfor some distanceover the flood
plainon the downstreamside of the channel.

65 sR 329 07rc7l93
For the design of scour protection, it is recommended that boundary shear
stresses be detennined for the main channel bed aM banks for the full range
of inbank stages, using curently available methods. ln addition, the banks
should be able to resist stresses of

N = 1.67y.S. (120)

on the upstreamside, and

tr, = 5Ty.So (121)

on the downstream
side.

The observedshear stressdistributionssuggestthat the sedimenttransport


capacityin the main channelwill be lowerfor overbankflows than for inbank
flows. Net depositionof sedimenlmay thereforeoccur in the main channel
dudngprolongedflood events. The shearconcentrations on the downstream
banksduringoverbankflows suggestenhancementof meandermigrationin
the valley directionduring prolongedflood flows, and also corroboratethe
mechanism of meandercutoffby openingchutesacrosspointbars.

5.8 Summary
Experimentaldata from PhaseB of the SEBC FCF has been analyzedto
producetwo methodsfor the estimationof dischargesin compoundchannels.
The best approachwas foundto be basedon dividingthe cross-sectioninto
zonesand calculatingthe dischargein each zone independently.The four
zoneschosenare:

1) The mainchannelbelowbanKulllevel.
2l The floodplainwithinthe meanderbelt.
3) The floodplainbeyondthe meanderbelt on the leftbank.
4) The floodplainbeyondthe meanderbelt on the rightbank.

The zonesare illustrated


in Figure44

For a given stagethe total dischargewill be calculatedas the sum of the


componentdischarges, i.e.

Q=QI+Q+Q.+QI (22)

The zonal dischargeswill be calculatedindependently,accountingfor the


appropriateenergyloss mechanismsin each.

Zones 3 and 4

The dischargesin zones3 and 4 are assumedto be controlledby bed friction


onlyand are givenby Equations118and 119.

Zone 1

The lossmechanisms whichaffectthe mrainchanneldischarge are complex.


It was not possible to develop a physically based descriptionof these
mechanismsand an empiricalprocedure was developed.A correctionfactor
is appliedto the bankfulldischarge
to obtainthe variationin the mainchannel
dischargewith over bankstage. The form of the correctionfactor is given by
Equations88, 89 and 90.

66 sR 329 07/O7l93
Zone2

Two altemativemethodsfor predictingthe dischargein the innerflood plain


zonewere developed.The first methodattemptsto accountfor the principal
loss mechanismsusing physically-baseddeterministicformulations. The
formulationsare based on a very simple conceptualmodel of the loss
mechanismsand requiredempiricaladjustmentto accountfor the additional
omplexities involved. The model explicitlyaccountsfor bed friction and
expansion/ contraction
overthe mainchannelandis describedby Equations
109,108,107,106,and 102.

The secondmethodis purelyempiricalandfollowsan approachsimilarto that


used for the rnainchannelzone. A basic dischargeis calculatedassuming
frictionlossesonly, and this is then adjustedto accountfor the effectsof floW
interactionwith the main channel. The form of the conectionfactor is given
by Equations 114,115,116and 117.

Thesetwo approachesto computingthe innerfloodptaindischargesgive two


sepanate methods of calculating discharges in meandering compound
channels.Forconveniencethe modelincorporating the expansioncontraction
losseswill be refered to as the Jamesand Wark methodwhilethe empirical
procedurefor zone 2 will be referredto as the Jamesand Wark2 method.

Bed shear stresses

Bedshearstressdatawasmeasuredunderoverlcank conditions duringPhase


B of the FCF work. The analysis canied out by the investigatorsis
summarised above.Themainchannelbedshearstressesare reducedduring
overbankflow comparedto bank full conditions. The shear stresseson the
floodplainsadjacent to the main channel show peak values which are
associatedwiththe exchangeof flow betweenthe mainchanneland the flood
plain. Equations120 and 121 give a roughestimateof the likelypeak bed
shearstresses.

6 Verification of the procedure


6.1 Background
The previouschapter details the developmentof two new proceduresfor
estimating the conveyance of meandering compound channels. One
procedure(Jamesand Wark) includesa semi-empiricalmodel of the inner
flood plain flows, while the other (Jamesand Wark 2) is basedon a purely
empidcalapproachto lhe inner flood plain discharge. Other methodswere
also ftJentified
in the literature.The work repoiledin this chapterwas canied
out to comparethe new and existingmethods. A selectionof the laboratory
data availablefrom varioussourcesdescribedin chapter3 was obtainedand
the methodswere appliedto predictthe stage dischargevalues. In some
caseszonaldischargeswerealso measuredand theseprovidea checkon the
predicteddistributionof flows in additionto total discharges.

6.2 Methods
Of the methods listed in chapter 5 the following have been used in this
verification:

Bedfrictiononly (BFO)
Jamesand Wark (JW)
Jamesand Wark 2 (JW2')
Ervineand Ellis (EE)

67 sR 329 07/07/93
Greenhill4 (cH4)
Greenhill5 (cHs)

The James and Wark Ervineand Ellis and the Greenhillmethodsare


describedin detailin chapter5. The bed frictiononly methodis basedon the
James and Wark channel subdivision. The discharges are calculated
assumingonly bed friction is actingand the areas, wetted perimetersand
hydraulicslopes for each zone are as defined for the James and Wark
melhod.

6.3 Applicationto laboratorydata


6.3.1 Data
The availablelaboratorydata is reviewedin Chapter3. Five of the eight
availabledata sets are consideredto be of goodenoughqualityfor use in lhe
developmentand verificationprocesses. Two of these sets were used to
developthe new procedures(SERCand Aberdeen). The data sets which
havebeen used in this verificationworkare listedin Table25 and 26. These
are stagedischargedata collectedUnderoverbankflow conditions. The test
numbershavebeen assignedfor easeof data handling.Tables25 and 26
also list ihe valuesof parametersrequiredfor the variouscalculations,such
as the width of zones 2 and the wholefloodplain; the side slopesof the
floodplainedgesand radiusof curyatureof the channelcenterline.Theseare
shownon Figure17. Table27 liststhe valuesof parametersrequiredwhen
calculatingthe mainchanneldischarges.0- is the meananglebetweenthe
channelcenterlineand.the flood plain centreline, averagedover a wave
lengrth.This mean angle is requiredfor the Eruineand Ellis calculations
Becausenpst of the geometrieswere constructedusing a combinationof
straightreachesand circulararcs these parameterswere easily calculated.
In thecaseof Soolqy's geometrya numericalintegration
sinusoidal wascarried
out to determine0^.

6.3.2 Totaldischargeand stage


Eachof the abovemethodswereappliedto the availabledata as follows.

1) The predicteddischargeswerecalculatedfor the measuredstages. This


allowedthe enor in the predicteddischargesto be calculatedaccording
to:
o/oError
in predictedflow = 1Oo(Q""b- Q*"J / Q*

2) The calculated stage discharge curves were then used to obtain


calculated stage values by linear interpolationusing the measured
discharges.The errorin the predictedstage was calculatedin terms of
the depthof flow in the mainchannel(H) accordingto:
o/"Errorin predicteddepth = 100 (H*. - H*.") / H*."

The mean values of these errors were calculatedfor each condition. In


additionmeanswere calculatedover combinationsof the data as follows:

1) All SERCdata
2) SmoothfloodplainSERCdata
3) Rod roughenedSERCdata
4) Vicksburgdata
5) Aberdeendata
6) Sookydata
7) All data

68 sB 32{' 07/07/93
Errors in discharges

The mean errors in the predicteddischargesfor the SERC FCF data are
shownin Table 28. The BFO methodover predictsby considerablemargins
with meanerrorsof M.8o/o,12.3%and32.5o/" for the smooth,roughard all the
data. The JW methodgives errorsof less than 5% for all three subsets (-
3.3o/o,-5.3o/"and -4.0%)afthoughit is tendingto under predict. JW2 gave
good resultsfor the smooth dala, -5.2o/",but gave poor resuhsfor the rod
rougheneddala,-22.80/", whichshowsup in a largererrorof -11.8V"over all
of the data. EE also gavereasonableresultsover boththe smoothand rough
data with errors of 4.97", 8.2"/" and 6.1% respectively. GH4 tends to over
predictfor lhe smoothdata and underpredictfor the roughdata with enors of
-22.5%,4.2"/" and 13.9"/"over all the FCF data. GHSdoes reducethese
efforssfightlyto 12.7Yo,-7.5%and 5.1%overall.

Fourof these melhods(JW, JW2, GH4 and GHs) were developedbased on


the SERCFCFPhaseB data. lt is notsurprisingthatthesefour methodsgive
gpodaccuracywhenappliedto this data set. The rnainconclusionis that the
JW methodis moreaccuratethan the JW2 methodfor caseswith roughened
floodplains. Noneof the otherdatawascollectedwith roughenedflood plains
and so it has been impossibleto verifythis conclusionagainstindependent
information. Future experimentalwork into conveyanceof meandering
overbankflow shouldcoverconditionswith roughfloodplains.

Table29 givesthe mean errorsfor the various methodsover the Aberdeen,


Vicksburgand Kiely data sets. lt is clear from the Aberdeenand Vicksburg
results that the BFO method becomes less accurate for more sinuous
channefs. In generaltypicalerrorsof about 30% to 4OT"were obtainedwith
thesedata. Boththe JW and JW2 methodsgive very similarresultsfor the
Aberdeendata with averageerrorsof 0.6%and 0.8%respectively.This is not
surprisingsincethisdatawas usedto developbothof the modelsfor the inner
flood plainflows. Againthe EE methodgave reasonablepredictionswith a
meanerrorof -2.37". GH4and GHSoverpredicteddischargeby 16.9%and
12.7%respectively.Thereare too few datafor eachindividualconditionof the
Vicksburgand Kiely data sets to make any detailed conclusionsbut it is
possibleto say that the JW, JW2 and the EE methodsgave similaroverall
predictions.

The applicationto Sooky'sdata is summarizedin Table 30. Againthere are


too few data for each individualconditionto makeany meaningfulconclusions
but over all 63 data pointsthe BFO methodgave a mean eror in discharge
of 2O.8%.The JW and JW2 methodsgave errorsof -1.9%and 1.2Y"and the
EE methodan eror of 14.6o/".TheGH4and GHSperformedwellon this data
set with mean erors of -1.4Yoand 5.2o/"respectively.

Table31 summarises the meanerrorsover the varioussub sets and all the
data. The BFO methodover predicteddischargeby 34.1%on averageover
allthe laboratorydata available. The JW and JW2 methodsgenerallygave
similarresuftsfor the smoothdata and this is reflectedin the mean enors of -
2.1o/oand-4.3%respectively.The EE methodgavea meanerrorof 5.3% and
GH4 and GHS gave mean errors of 11.5o/o and 8.0% respectively. Given
these resultsTable 32 showsthe six methodsrankedin order of accuracyof
predicteddischargefor the varioussub sets of the data. lt is obviousthat the
bed frictiononly methodis the worstof all six methodsfollowedby Greenhill's
methods4 and 5 respectively.lt is more difficuftto distinguishbetweenthe
bestthreemethods.

The abovediscussionhas concentratedon the mean errorsand has ignored


the standarddeviations(SD) in lhese means. In generalthe JW and JW2

sB 329 07/07/93
methodshave SD's of between5o/oand lOTotor the varioussub seis of the
data. The Ervineand Ellismethodalthoughgivingroughlyequivalentmean
effors shows SD's between 15% and 2Oo/". These quite large standard
deviationsare not causedby randomscatteraboutthe meansbut are due to
systematictrendsin the enors withdepth,this is discussedin detailbelowbut
it is possibleto say that the JW and JW2methodsgaveslightlymoreaccurate
predictions.Over all the data the JW methodperformedbetterwith a mean
enor and standad deviationof -2.1o/"and9.7Y"comparedlo -4.3o/o and 13.21"
for the JW2 method. In additionit has been shownto be moreaccuratefor
the data with roughenedflood plains, mean errorand SD -4.O%and 8.4o/o
comparedto -11.8%and 14.4o/o for the JW2 method.

Whenconsidering onlythosedatanot usedin the development of any of the


methodsslightlydifferentresultsare obtained. The JW2 methodturnedor.il
bestwith meanerrorand standarddeviationof 1.8Toand 10.8olo, followedby
GH4(2.O%and 15.6%).The JW methodgaveslightlyworseresultsal -2.6o/o
and 11.7%respectively. ThedatafromVicksburg, KielyandSookyusedhere
did not cover roughenedflood plain conditionsand this shouldbe borne in
mindwhenconsidering the relativemeritsof the variousmethods.

Errors in stage

The resultsshownin Tables33 to 36 are the mean7oerrorsin calculated


depthfor the variousdatasets. ln generaltheseresultsfollowthe discussion
of the enors in dischargeswith two impoflantexceptions.

1) Wherea methodover predictsdischargethen it underpredictswater


level.

2l The values of enors in calculatedstage are much less than the


corresponding
errorsin discharge.

Thiscan be demonstratedby comparingthe valuesin Tables31 and 36. The


Bed FrictionOnlymethodoverpredicteddischargeby 34.1%on averagebut
under predictedthe channeldepth by 4.7"/". Similarcomparisons can be
madefor the othermeihods.

The variationof errorsin predicteddischargeand stagefor the six methods


with relativedepthare shownin Figures45 to 50. lgnoringall lossesexcept
bed fdction gave errors in the predicteddischargeswhichfall mainlyin the
range10%to 50%,withcorresponding errorsin depthbetween-1oo/oand-2o/o.
It is apparentfrom Figure45 that the enors dependon the geometryof the
channelwiththe variouscasesdisplayingdifferentdistributions of enorswith
the depth of over bank flow. In generalthe errorsshow strongtrends with
stage.

The JW methodgivesa muchsmallerrangeof errors,Figure46. Mostof the


data fafls in the range-10%to A% tor dischargeand-2Yoto 3% for the water
depth. At low over bank stages(H-h/H< 0.15)the methodtendsto over
predictdischargeswith the errorsreducingat higherstages.

Figure47 showsthe errordistributionsfor the JW2 method. The majorityof


the errors in predicteddischargefall in the range -2Oo/"
to 10% with the
correspondingerrorsin depth lying between2"/" and 15%. There are more
noticeabletrendsin the errorsfor this methodcomparedto the JW method.
The rod rougheneddata(33,34 and43) showa strongincreasein the under
predictionof dischargewithdepth.

70 sR 329 07/07/93
The Ervineand Ellismethodgave errorsin predicteddischargein the range -
3oo/olo 5O%with the correspondingerrorsin waterdepthlying in the range -
8t" lo 1O%,Figure48. This methodtends to over predictdischarge(and
underpredictwater level)at low relativedepthsand underpredictdischarge
at high depthswithan approxirnately lineargraduationbetween. This agrees
with the limited numberof resultsquoted by Eruineand Ellis (1987). lt is
interestingto notefromTable29 that for the datacollectedfrom the Vicksburg
flume the Ervineand Ellis methodis the npst accurateof all the methods.
Ervineand Ellisonly appliedtheir methodto the Vicksburgdata and reported
goodagreement.

Figures49 and 50 showthe variationof errorsfor Greenhill'smethods4 and


5 respectively. Both these methodsgive variationsof error for the various
caseswhichare similartothoseobtainedwith the bed frictiononly methodbtit
shifted towards the zero error line. Greenhill'smethod4 shows enors in
dischargewhichare shiftedby approximalely2A"/o- 22% while method5 is
gives errorsin dischargeshiftedby about 25 - 27%. Bothof these methods
displayquite wide rangesof errors.

The resuhsaboveshowthat the semi-empiricalexpansioncontractionmodel


developedby the authors(JW) is moreaccuratethan the other methodswith
a meanenor that is wellwithinexperimental tolerances.The tendencyof the
method to under predict dischargesis a conseryativefault. In a design
situationa channelis usuallysized to have a requireddischargecapacityat
a given water level. The aurthorsmethodgivesa slightlylargerchannelsize
than actuallyrequiredhencewaterlevelswill be slightlylowerthan predic'ted.
The Ervineand Ellismethod,whichis basedon a similarconceptualnndel,
gives a mean enor which is probablyacceptablein practicebut the larger
standarddeviationindicatesa wider and more systematicspread of errors
about lhe mean. The alternativeempiricalmodeldevelopedby the authors
(JW2)was foundto be less accuratewhen the flood plainsare rougherthan
the main channeland is not recommended.By ignoringthe effectsof loss
mechanismsother than bed frictionthe above resulls indicatethat enors in
totaldischarges in the orderof 35%maybe expected.Theempiricalattempts
by Greenhilltoreducetheseerrorsdo succeedto a limitedextentbut do not
significantlyreducethe spreadof the erors. The James and Wark method
has the additionaladvantageover the others that it is based on measured
velocitiesand dischargesfor zone 1, and shouldgive morereliablepredictions
of the zonal distributionof conveyance. There is limited independent
informationavailableon zonal distributionsof flow and this is consideredin
Section6.3.3.

Sensitivityanalysis

The Jamesand Wark methodrequiresvaluesof geometricparameterswhich


are well definedin laboratorychannelsbut usuallypoorlydefined in natural
channels. The valuesof meanderwavelengthand main channelside slopes
(requiredfor the zone2 model)in particularare dffiicuftto define exacttyfor
natural channels. The following sensitivityanalysis was caried out to
determinethe degreeof precisionrequiredwhen estimatingthese parameters
in practice.

The values of the wave length (L) and side slopes (SJ for the available
laboratorydataare knownexactly. Errorsin the predicteddischargesare not
due to uncedaintiesin L or S" but to othercauses. The effectsof uncertainties
in L and S" were investigated as follows.

The knownvaluesof L and S. for all of the availabledata were factoredup or


down by fixed amounts. The JW melhod was appliedusing these factored

71 sR 3An O707l93
values of L or S" in the calculation. The mean errors in predicted discharge
were calculated over all279 data points. Thus the variation in errors could be
related to the known errors in L or S".

The effect of uncertainties in wave length are summarized in Table 37. The
mean enor in the predicted discharges is reduced trom-2.1"/" to -10.3% by the
50% reduction in wave length and increases to 2.3% tor a 50o/oincrease in
wave length. Thus an eror in wave lengthof !5Oo/"resultsin a 110% change
in the mean error in predicted discharge. Similar resufts are shown in Table
38 for changes in side slope. The mean enor is reducedfrom -2.1o/oto-5.3Yo
by a 100% reduction in side slope and increases to 2.4% for a 100% increase
in side slope. Thus changes of +100% in side slope values results in a t5%
change in the mean error. These results, although not conclusive, indicate
that predicted discharges are relativelyinsensitiveto errors in wave lengrthand
main channel side slope and great accuracy in their estimation is not
necessary. However similar sensitivity test should be carried out in any
practicalapplicationto confirm these findings.

6.3.3 Dischargedistributions
The resultsabovedemonstratethe overallaccuracyof the variousmethods.
The methodsare basedon similarchannelsubdivisions.The dischargesin
the variousparts or zones of the channelare calculatedseparatelyand
summedtogetherto obtainthe totaldischarge.Hencethe methodsgivethe
distribution
of flow betweenthe zonesin additionlo the totaldischarge.

There is very little independentinformationavailableon lhe distributionof


dischargein meandering overbankflow. Sooky(1966)carriedout detailed
velocitymeasurements in shallow(403,depth0.0613m)anddeep(409,depth
0.080 m) meanderingchannelswhich were otherwiseidentical. These
experimentswere carried out in a channelwhich was built at a scale
approximately 8-9 times smallerthan the SERC FCF PhaseB geometries.
Sookyintegrated thesevelocitymeasurements to obtainthe proportion
of the
totaldischargewithineachzone. Kiely (1989)gives similarinformation
fortwo
depths(Test301,0.060mand 0.080m).The measureddischargesin all four
zonesfor thesefour casesare givenin Table39. Table40 givesthe enors
in the predicteddischargesfor these four cases. The BFO method over
predictsby up to 50% whilethe JW methodgave resultsaccurateto within
tl0oh, the JW2 was accurate lo t17o/o,EE 30% and Greenhill'stwo methods
to 30%. The JW methodgave very good overallaccuracyfor Kiely'sdata
while none of the methods were parlicularlyaccurate for Sooky's two
conditions.The mainreasonwas probablya poor definitionof bed frictionfor
Soolqy'sdata.

Table41 showsthe measuredand calculateddistributionof flowsbetweenthe


variouszonesas percentagesof the totaldischarge.The resultsobtainedfor
Sooky'sdatashowlittledifference betweenlhe variousmethods,theyallgive
similardistributionof flows. This may be a functionof the very low sinuosity
of Soolqy'schannel. Kiely's results show more differencesbetween the
methods.The JW methodgivesexcellentpredictionsof totaldischarge(within
4o/"and 1ol.).At the lowerdepththe JW methodgivesthe distributionof flows
alrnostexactlybut underpredictslhe zone 1 dischargefor the higherdepth,
this is coupledwith a generalover prediction for the otherzones. The JW2
methodgave similar resultsto the JW methodfor the flow distributions. lt
gaveexcellentresultsfor the smallerdepthcaseand underpredictedthe main
channeldischargefor the deepercase. In fact all the methodsunderpredict
thezone1 discharges for thedeepercaseandthis wouldindicatethatthe bed
frictionlaw is a betterfit to the data at low stages. The EE and the two GH
methodsover predictedthe zone 1 dischargeand underpredictedthe zone 2
discharges.

sR 329 07/O7l93
@l
These measureddistributionsof flow were derived from integratingpoint
velocitymeasurementsand the derivedzonalflowsare probablyaccurateto
about 5%. The comparisonsshow that in general both the JW and JW2
methodsgive flow distributionswhichagreewith the measureddistributions.
On the basis of this very limiteddata it can be concludedthat the author's
mettpd (JW) gives superiorpredictionsof both the total and zonal flows in
meanderingcompourdchannels.lt is hopedthatfutureexperimental workwill
concentrateon the collectionof datagivingthe zonaldistributionof discharges
to confirmthese conclusions.

6.4 Applicationto straight laboratorydata


The SERCFCFwork hasbeencarriedout in two phases. PhaseA dealtwith
straightcompoundchannelsand PhaseB dealt with meandedngcompound
channels. The PhaseA data has alreadybeen usedto developa methodof
calculatingconveyancein straightcompoundchannels,Ackers(1991). The
that thesestraightchannelmethods
work reportedin Chapter2 demonstrates
cannotbe usedto predictdischargesin meanderingcompoundchannels.

The James and Wark methodwas developedbased on the Phase B data.


The independentdata availablefor verificationincluded: the Vicksburgdata
with sinuositiesof 1.57, 1.4Oan 1.2; Kiely'sdata with sinuosity1.22 and
Sooky'sdata with sinuosityof 1.09. The authorsmethodgave reasonable
predictionsfor all of these data. Sincethe straightchannelmethodsare
known to give inaccuratepredictionsfor meanderingchannelsit is to be
expectedthat the meanderingchannelmethodswill give poor predictionsof
dischargein straightcompoundchannels.

The performanceof the Jamesand Warkmethodappliedto datafrom straight


compoundchannelshas been investigated.The Phase A data set was
availableand was usedin this evaluation alongwiththe straightchanneldata
availablefromtheAberdeenflume(100)eindKiely'sresults(300).Thedetails
of the various tests and results from Phase A are reported in full by
Ackers(1991).Of the PhaseA datatests1, 2, 3, 5, 7, 8, 9, 10 wereusedin
this evaluation,a total of 198 data pointswere available. The appropriate
modifiedsmoothlaw (section2.3) or the Ackersrod roughnessmethodwere
used to obtainthe bed frictionfactors.

Whenapplyingthe meanderingchannelmethodsto straightchanneldatathe


innerffood plain zone disappears. Zone2 has the same width as the main
channeland it was assumedthat the zones 1 and 2 could be consideredas
a single unit. The channeldivisionthereforehas reducedto the straight
channeldivision method(DCM2)describedin Chapter2.

The Jamesand Wark methodgave an averageerror in predicteddischarge


ot -27.3o/o This generalunderprediction
with a standarddeviationof 17.Oa/o.
of dischargesby up to 50olodemonstratesthat althoughthe method can
accuratelypredictdischargesin channelswith sinuositiesas low as 1.09 it
cannotbe appliedto straightcompoundchannels. Obviouslyfurtherwork is
requiredto investigatethe conveyanceof compoundchannelswith sinuosity
between1.0and 1.09.

6.5 Applicationto field data


The procedurepresentedabovewas developedand verifiedusinglaboratory
model data. There is very little field informationavailableregardingthe
performance of full scalemeanderingchannelswith flood plains. The only
detailedfield investigationknown at presentwas carried out on the River
Rodingin Essex,see Sellinand Giles(1989)or Sellinet al (1990).One other
site is alsocunentlybeinginvestigatedby Sellin.A physicalmodelofa 25Om
longsectionof the RiverBlackwater in Hampshire hasbeenconstructed in the

73 sR 329 07/07193
SERC FCF at a scale of 1:5. Field measurements are scheduledto
commencein early 1993and are to run for three years. The resultsof this
study were not availableat the time of writingbut shouldprovideirnprcved
validationdata.

The Roding study

The Rodingis a relativelysnrallriverwith a channelwidthand depth of about


7m and 1.5m respectively. Full details of the field and laboratory
measurements carriedout on lhis siteare availablein Sellinand Giles(19S8)
and Sellinet al (1990). The studyreachliesdownstream of Abridgeand as
part of a flood alleviationscheme a two stage channel was formed by
excavatingapproximately 30m widebermson eithersideof the mainchannel,
Figures51 and 52. Theoriginalchannel was untouchedand remainedin the
naturalstatewitha bankfullcapacity of approxinrately
3 cumecs.The resulting
flood channelhas a low flow channelwhich meanderswithinthe berm limits
witha sinuosityof 1.38and a wavelengthof approxinrately 96m. Hencethe
channeldoesnot possessouterfloodplainzones. The bermswereformedat
a level belowthe surroundingfloodplainand were intendedto provideextra
flood dischargecapacityand so relievefloodingon the existingfloodplainfor
flowswith a returnperiodof up to thirtyyears. Shortlyaftercompletionof the
schemeit becameclearthatthe actualcapacityof the channelwas lessthan
the designvalue. This was partlyassignedto the differencebetweenthe
assumedbermvegetalion(shortgrass)and the actualvegetationwhichwas
extremelydense. The designcase assumedthat the berm wouldbe grazed
by farm animalsbut in fact this did not happenand the NationalRivers
Authority(NRA)wereforcedto cut the growthmechanically at considerable
cost.

The field and laboratoryprojects investigatedthe effects of different


maintenancepolicieson the channel capacity. Most of the conditions
investigatedwere with the flood berms covered,totally or partially,with
extremelydensevegetation andverificationof calibrated
bedroughness values
was not possible. The roughnessvaluesvariedstronglyboth with stageand
duringthe growingseason. The data recordedaftera full cut on the berm
showedmuch less variationin berm roughnessvaluesand so were feh to
providethe best informationfor validationof the author'sprocedure. The
methodwas appliedto the stage-discharge datafromthe followingtwo cases.

P2 The bermgrowthwas cut immediatelyafterthe summergrowingseason


and so the bermswere coveredin shofi grass.

M2 The laboratorymodeldataconespondingto the smoothberm case (P2


on the prototype).

In orderto applythe procedureto these measurements the sevenavailable


surueyedsectionswereusedto providedreachaveragedareas,widthsetc for
both flow zones at stagesup to 1.0m above the berm level and these are
givenin Table42. The information providedby Sellinand Giles (1988)and
Sellinet al (1990)combinedwithwidelyacceptedguidelines, Chow(1959)and
Henderson (1966)allowedthe bermManning'sn valuesfor thetwo cases, P2
and M2 to be estimatedas 0.050and the main channelManning'sn was
estimatedas 0.044. The longitudinalslopeof the bermwas 1.405x10-3.

The meanerrorsin the predicteddischargesgivenby the BFO,JW and JW2


methodsare shownin Table43. lt is apparentthat the recommendedmethod
(JW) improvesthe overallaccuracyof the predicteddischargesto about -2%
and that by ignoringthe non-frictionhead losses dischargewill be over-
predictedby about10%on average. The empiricalJW2 methodgave very

74 sF 329 07/07/93
poor predictionsresultingin a mean error of approximately-30%. These
resultsare confirmedby Figure53, the JW and BFO methodsgive stage
dischargecurueswhichfollowthe generaltrendof the data. The JW method
tendsto underpredictdischargeat low flood plain depthsand over predictat
high flood plaindepth,whilethe BFO methodover predictsfor all stages.

Some sensitivitytesdswere carried out to investigatethe effect of berm


roughnesson the total channelcapacity. Table 44 shows the variationof
meaneffors in predicteddischarges,for the BFO and JW methods,with berm
roughnessfor case P2. Bothmethodsover-predictdischargeswith low berm
roughnessand under-predict with high bermroughness.The author'smethod
alwaysgivessmallerdischargesbecausethe non-frictionenergylossesin the
two zones are explicillyaccountedfor. The differencebetweenthe mean
effors for the two metlrcdsreducesfrom >100% at very low roughnesstd
about1O%at the calibratedroughness.At higherroughnessesthe difference
betweenthe two methodsremainsapproximatelyconstantat about 10%.

Theseresultsshowthat as the floodplainbecomessmootherthe two methods


divergemore. Thus the effectof increasedflood plain roughnessis to make
the non-frictionheadlossesless important.Bed frictionis likelyto be the most
impodantsinglesourceof energyloss in naturalriversand remainsa potential
sourceof significanterror in conveyanceprediclions. The estimationof bed
frictionfactorsis largelysubjectiveeven given the comprehensiveguidelines
presentedin standardtexts such as Chow (1959)and Henderson(1966).
Thus it is not possibleto give generalguidelineson the choiceof bed friction
valueas site specificaspectsare likelyto governthe relativeimportanceof the
variousloss mechanisms.Testsshouldbe caried out for eachapplicationto
gaugethe sensitivityof the solutionto variationsin roughnessvalues.

6.6 Summaryand conclusions


Thetwo methodsdeveloped by theauthorsandfourothermethodshavebeen
used to predictdischargesand stagesfor the availablelaboratorydata. The
authot's semi-empiricalmethod(JW) was found to give the most accurate
predictionsof total dischargeand acceptablepredictionsof the distributionof
discharges.

The available data used in this verificationcovered a limited range of


conditions. Further experimentalwork is required to look at both total
dischargesand the distributionof dischargesfor:

a) channelswith low sinuosities(<1.09).


Meandering

b) compoundchannelswith roughfloodplains.
Meandering

c) Low overbankdepths(y'. 0.2).

The sensitivityof the Jamesand Wark methodto variationsin the values of


both meander wave lengrthand main channel side slope has been
investigated.The resultsindicatethat in great precisionin estimatingthese
values is not required.

The procedurehasbeenappliedto the bestfielddataavaitableand has been


shown to give improved predictionscompared to current practice. The
sensitivityof the results to variationsin bed roughnessvalue has been
investigated.The non,frictionenergylossesare shownto be less important
as the floodplainis roughened. Bed frictionremainsthe most significant
sourceof energyloss in riverswith overbankflow.

75 sR 329 07lo7l93
7 Future research needs
The futureresearchwhichstill renrainsto be carriedout falls into three rnain
categories:

1) The collectionof independentinformationto use in verifyingthe work


presentedin this repod. This includes laboratorystudies and field
measurement programmes.

2) The development of two and threedimensional numericalmodelsand


their applicationto the availablelaboratorydata.

3) The use of the proceduresdevelopedin this repod within ond


dimensionalrivermodelshasnot beenfullyconsidered.The procedures
weredevelopedto mbdelsteadystatestagedischargesand the type of
workrequiredto confirmthatthe methodsare appropriate
for use in 1-D
modelsis describedbelow.

predictionfor inbankflows
7.1 Stage-discharge
The currentproject has put a low priorityon inbankflows. lt is clear,
Chapter4, that the effectof meandering
on inbankchannelconveyanceis
considerable,
and the importanceof main channelcapacityin a two-stage
channeldesignor analysisis obvious.

The SCS and LSCS methodsof adjustingthe frictionfactorto accountfor


meandereffectshasbeenshownto be reasonable.Theyhaveno theoretical
basis,however,and sufferfromthe main limitationof relatingbendenergy
lossesto only one parameter.In orderto circumventthese limitationsit is
recommended that Chang's(1984)approachbe furtherdevelopedto provide
simpleguidelines for estimating
lossesthataccountfor a widerangeof all the
relevantparameters,Chapter4. The guidelinesshouldallow lossesto be
evaluatedfor individualbendsas wellas for a meandertrain. The effectof
variationof cross-sectionalongthe channelshouldalsobe investigated, but
this wouldrequirea trore completedescriptionof flow in bends.

7.2 Laboratorystudies
7.2.1 Extensionof existingdata sets
Existing laboratorystudies cover a relativelynarrow range of conditions.
Furtherlaboratorywo* wouldbe requiredeitherto verifyor extendthe present
method for conditionsother than those covered by the existingdata. In
particularthe followinglist of expedmentswould fill gaps in the availabte
laboratorydata. lt shouldbe notedthat this list is not in any particularorder
or impodance.

1) Undertakeexperiments to measurestage-discharge,velocityand bed


shearstresses channelswithsinuosities
for meandering between1.0and
1.09. This is impodantbecausethere is a need to establishat what
sinuositya compoundchannelanalysistreatmentshouldswitchfrom
straightto meandering.

2) Undertakeexperiments to measurestage-discharge, velocityand bed


shearstressesfor lowovebankstages,ie (y/hl valuesbetween0.0and
0.1. Thereare few datapointsin this regionand it is probablythe most
commonrangeof overbankflow conditionswhichoccurin nature.

76 sR 329 07/07/93
3) Undertakeexperimentsto measurestage-discharge, velocityand bed
shearstressesfor flood plainswithtranwerseslopeawayfrom the main
channel. There are few laboratorydata for this conditionand natural
flood plains tend to slope laterallyin this manner. There is some
conjecturethat it may be morerealisticto analyzeoverbankflow in these
geometries using straight channel techniques,as the flow will be
constrained paralleltothe mainchannel.

4l Undertakeexperimentsto measurestage-discharge, velocityand bed


shearstressesfor 'sinuositiesbetween1.09and 1.20; 1.20and 1.40;
1.40and2.01andfor sinuosities greaterthan2.01. All knownlaboratory
experinentshavebeencaniedout at or very closeto sinuositiesof 1.09,
1.20,1.40 and 2.01 and this is obviouslyleavesgaps in the available
information.

5) Undedakeexperimentsto measurestage-discharge, velocityand bed


shear stressesin meandedngchannelsfor a rangeof channelto flood
plainwidthsand for caseswith asymmetricflood plainson eitherside of
the rnainchannel. All existingdata have been collectedfor a limited
range of channelto flod plain wldth ratios and with symmetdcflood
plains.

6) In ordertoconfirmthe SERCPhaseB datait wouldbe usefultoconduct


experimentsin smallscaleflumeswith geometrieswhichare exactscale
modelsof the phaseB tests. lf such experimentswere canied out and
provedto be positivethen the gapsin the PhaseB resultscouldbe filled
usingdata collectedin muchsmallerlaboratoryfacilities.

7) Undertakeexperimentsto measurestage-discharge, velocityand bed


shearstresses in meanderingchannels withroughenedfloodplains.The
onlyinformationcunentlyavailablewasobtainedfromthe SERCFCFfor
onlytwo channelsinuosities.The methodof usingverticaldowelrodsto
roughenthe floodplainalsoproducede)dremefloodplainroughnesses.
lndependent is requiredto confirmthe SERCFCFdata.
information

8) Undertakeexperimentsto measurestage-discharge, velocityand bed


shearstressesin meanderingchannelswithdifferentcross-seclions.The
SERCFCFphaseB investigation coveredtrapezoidalandpseudonatural
cross-sections. CIher studies have been conducted either with
rectangular or trapezoidal main channel cross-sections. Fuilher
informationon the effectsof varyingchannelside slopesin trapezoidal
channelsand the effects of changesin cross-sectionshape along a
meanderwouldbe useful.

Laboratoryworkintendedto extendexistinginformationshouldbe carriedout


in channelswith idealizedgeometriessimilarto thosefrom whichthe existing
laboratorydatrawereobtained.Forexamplethefloodplainsshouldbe uniform
in wldth along the lengthof the channeland the meanderingmain channel
plan geometryshouldbe a simplerepeatinggeometricshape.

7.2.2 Laboratorystudiesof tossmechanisms


The formulationof modelsof loss mechanisrnshas exposedsome surprising
gaps in experimentalresults. Someusefulinformationcouldbe obtainedfrom
relativelysimpleand inexpensivelaboratorystudies. The followingstudies
wouldcontributeto the descriptionsof lossesin the identifiedflow zones :

1) A quantificationof contractionloss over an upwardstep.

77 sR 329 07/o7t93
2l A study of the effect of slot alignmenton expansionand contraction
losses.

7.3 Field data collection


The lack of adequateand reliablefielddata has beena majorconstraintin the
verificationof analysismethodsfor meanderingcompoundchannels. The
methodpresentedis basedon resultsfrom laboratoryexperimentsand white
this is appropriatebecauseof the high degree of control of the relevant
variablesrequired,thecorrespondence betweenlaboratoryandfieldconditions
is not firmly established. The relativeimportanceof differentenergy loss
mechanismsmaychangewithscale. some informationwasavailablefromthe
River Roding study (Sellin and Giles, 1988) and providedgood initial
verificationof the findingsreportedabove. Howeverfurlherfield data should
be soughtto fully verify methodsof estimatingconveyancein meandering
channels.

The River Blackrrvater study which combines laboratoryand detailedfield


measurernents will providea usefuldata set to comparemanyof the details
of the method. lt is proposedto makemeasurements of stagedischargeand
pointvelocitydistributions,both in the laboratoryand the field. This research
progammeis plannedto take placeoverthreeto fouryearsand will provide
a gooddealof detailedinformation on flow distributionsbetweenthe various
zonesin pafiicularcases.

7.3.1 Strategyfor field data eollection


It is apparentthat the analysismethodhas not yet beenfully verifiedagainst
field data becausevery few relevantfield measurementshave been made.
Giventhat it is desirableto collectmore field data it is importantthat the
correcttypesof information are obtainedin orderto makethe mostetficient
useof resources.

In generalthereare two levelsof validationpossibleand thesedifferin the


amountof hydraulicinformationto be measuredat eachsite.

1) Collectonlystage-discharge
information
at eachsite.

2) Collect$age-discharges,point velocities,and water levelsboth along


and acrossthe studyreach.

Obviouslyit will be possibleto carryout measurements at a largernumberof


sites if only totaldischargesare to be measured. This wouldprovidea wide
range of data for the validationof the overall methodbut would not provide
informationto validatethe calculatedflow distributions.lf the more detailed
validationis requiredthen it is likelythat fewersiteswouldbe considereddue
to the increasedcosts.

It will eitherbe possibleto partiallyvalidatethe overallmethodon a relatively


largenumberofsites,orcarryoutmoredetailedvalidation on a limitednurber
of sites. The detailedvalidationwouldrequirethat at leastthreeor four other
projectssimilarto the Blackwaterprojectbe set up and the costsof running
these projectsover threeof four yearsare likelyto be considerable.

Partialvalidationof the methodusing stage-discharge data from a wider set


of sites wouldprobablybe sufficientin the dhoftterm combinedwith the long
term aim of collectingsufficientinformationto carry out full validationover a
numberof sites.

78 sR 329 o7l07/93
7.3.2 Suitablesites
Sincethis documentis concernedwith meandering compoundchannelsany
fielddatashouldalsorelateto meandering channels.The typeof reachto be
considered forfield datacollectionshouldconformto thefollowingguidelines.

1) Sitesshouldhave significantmeandersor bends. The meanderzone


shouldform a significantpad of the floodplainand the meandersshould
be distinctand welldeveloped.

2') Sites should preferablyhave a fairly regular meander pattern. The


meanderwavelengthand amplitudeshouldnot vary significantly within
each site.

3) Landusage,(vegetation
etc) on eachftoodplainshouldbe reasonably
uniform.

4) The presenceof buildingsor otherobstructions on the floodplainshould


not disqualifya site providedthatthe obstructionhasa minoreffectof the
flow patternthroughthe site.

In orderto carryoufany hydrauliccalculationsrelatingto a chosensitecedain


information is requireddetailingboththe plan and crosssectiongeometries.

5) Enoughsuryeydata shouldbe availablefrom mapsand channelcross


sectionsto estimateboth the main channeland floodplainlongitudinal
slopes. Wherethe local bed slopesat the site differfrom the overall
reachslopesbothshouldbe given.

7.3.3 Hydraulicdata
In orderto provideenoughvalidationdatafor eithera partialor a fullvalidation
thenthe followinghydraulicdata shouldbe measured.

1) Watersurfaceslopes.The importanthydraulicslopewhichcontrolsflow
in open channelsis the watersudaceslope. In uniformflow this slope
will be equal to the valleyor floodplainslope. Water sudaceslopes
shouldbe measuredover the reachesof interest. lt may be possibleto
do this relativelyeasilyand cheaplyusingmaximumwaterlevelrecorders
set at intervalsalongthe reach.

2) Pairsof measuredstageand correspondingdischarge.Theseshouldbe


providedat both inbankand out of bank stages. lt may be possibleto
irJentifysuitablesiteswhichare closeto existinginbankgaugingstations.
Maximum water level recorders would provide stage values with
dischargesbeing obtainedfrom the neaiby gaugingsites. This would
probablybe the mostefficientmethodof collectingstagedischargedata
in meanderingoverbankreaches. In suitablereachesnot close to
existinggaugingstationsspecialarangements would be requiredto
measuredischarge.

3) Velocityprofiles.Thesemay be eitherjust in the mainchannelregions


or across the whole channeland floodplain. This would require a
cablewayto be set up at selectedsites in the reach.

To provideinformationfor a partialvalidationitems 1 and 2 aboveshouldbe


measuredat as many sites as possible. lf a more completevalidationis
requiredthen item 3 aboveshouldalso be measuredat each site. In the
immediatefuture it is recommended that suitablesites shouldbe identified
and, if possible,a parliatvalidationcarriedout. ln the longerterm detailed

79 sR 329 07/07/93
measurements shouldbe soughtto add to the data set providedby the
project.
Blachruater

7.4 Computationalmodelling
7.4.1 Turbulencemodelling
Threedimensionalturbulencemodellingis the mostpromisingapproachfor
developing methodsto describethecomplexmechanics of flowin meandering
compoundchannels.lt is not envisagedthatturbulencemodelswill be used
directlyfor routinedesignapplications,but ratherthat they could be used in
parametricstudiesto generategeneralresuhsfor incorporalionin standard
designmethods. By followingsuch an approachthe resultsof experimental
work(suchas the SERCFCF PhaseA and B studies)andfieldstudiescould
be extendedand generalized.The procedurewouldbe to calibratethe model
on the existinglaboratorydataandthento usethe computational modelrather
than the laboratoryto generateinformationabouta widerrangeof conditions.
Turbulence modellingshouldbe usedto complement laboratory studiesrather
than replacethem.

In designapplicationsuseof a 3-Dflowandturbulence
modelis unlikelyto be
practicalfor the foreseeablefuture. Howeverusefulinformationmay be
obtainedfrom a two dimensional,depth integratedmodel. This type of
approachhas provedto be usefulin the simplerstraightchannelcase,for
examplethe LateralDistributionMethod(LDM),Appendix2. Development of
suitable2-D modelsshouldbe encouraged.

7.4.2 One dimensionalmodelling


Manyone dimensional modelsof riverflowsexistwhichare basedon the St
Venantequations of 1-Dflow. Theygenerally
usethecomputationaltechniqu
of finite differencesto solve the St Venantequationsand so providethe
variationof water leveland dischargealonga reachof channel. Typically
these modelsare basedon the use of pre-computed tablesof conveyance
whichare accessedduringthe calculations.

Existing methods used to calculateconveyance

One dimensionalmodelsrequirechannelcross-sectionsto be suppliedat


locationsalongthe river. Thesecross-sectionsand otherdata describingthe
bed roughnessof the channelare then usedto calculatethe mnveyanceof
eachcross-sectionwithinthe model. Conveyanceis a convenientmeasureof
a rive/s capacityto passdischarge.Typicallythe methodsusedto compute
conveyanceare basedon variantsof the dividedchannelor sum of segment
methods.Thesemethodsare appropriatefor straightcompoundchannelsbut
have beenfoundto give poor resultswhenappliedto meanderingcompound
channels(Chapter2). In the case of meanderingcompoundchannelsthe
existingconveyancecalculationcould be replacedwith the new procedure
reportedabove.

tnctusionof the new procedurein l-D modets

The new methodfor calculatingstagedischarges in meandering compound


channelshas a numberof implications regardingits use in 1-D rivermodels.
Primarilythese are changesin the data specificationfor the cross-sectional
data (ie additionaldata items)and changesto the conveyancecalculation
procedures.

80 sR 329 07/07/93
Data requirements

The data reguirementsfor the new methodare slightlygreaterthanthose that


wouldcunentlybe specifiedin existingpackages. Modifications to the cross-
sectionaldata inputswould be requiredto accountfor additionalitems such
as:

1) sinuosityof the channel


2l meanderwavelength
3) mainchannelside slope
4l pointersto indicatelhe limitsof the innerflood plain meanderbelt

Wherepossiblereach averagevalues,basedon sub-reachesof the model,


shouldbe used to specifythese additionaldata items. The sub-reachesare
likelyto covera numberof cross-sectionlocationsin the modelandstpuld be
selec{ed such that the geometric parameters(main channel side stope,
sinuosityand widthof meanderbelt)remainapproximately constantthroughout
the sub-reach. Thesedata itemsare readilyavailablefrom a combinationof
cross-sec{ionand plan surveysof the river reachand would not requireany
additionalresourceswhen undedakinga modelstudy.

In unsteadyfloodmodelling,storageon the floodplainscan playan impodant


rolein the attenuation of floodpeaks. In a highlymeandering riverspecifying
the flood plain lengthequivalentto the river lengthbetweenadjacentcross-
section locationsmay have a tendencyto over-estimatethe storage area
availableon the floodplains. This maythen leadto errorsin the attenuation
of a flood wave. lt is importantthereforeto specifythe river lengthand flood
plainlengrths separately.

lmplicationsfor 1-D river models

Thereare a numberof otherissuesto be consideredwhen usinga package


withthe new methodof calculating conveyances.The usualprocedurewhen
modellingcompoundchannelsis to calibratefirstlyfor the inbankroughness
and then proceedto calibratethe overbankroughness.Analysisof the Phase
B data has shownthat the inbankdischargefalls as the water level moves
frominbankto overbankconditions.In existingmethodsthis mayleadto large
erors in the flood plain roughnessas the calibrationprocedureinplicitly
assumes that the main channel dischargeremains constant at ovefuank
stages. This impliesthat the calculatedmainchannelflowsand velocitieswill
be too high at overbankstagesand thoseon the flood plain will be too low.

This results in incorrectvalues for the energy and momentumcoefficients,


which in tum leadsto errorsin:

1) afflux calculationsat structures,


2) shearstressand sedimenttransportpropertiesand
3) the effectiveflood wave speed.

A major factor to be considered,should the new hydraulic method for


meanderingcompoundchannels be incoporated in existing modelling
packages,is that the calibrationcoefficientsobtainedfrom earlier model
studiesmay no longerbe applicablein the revisedversionsof the modelling
software. The calibratedroughnesscoefficients(Manning'sn, Colebrook-
White k" or ChezyC) wouldbe effectivelycompoundroughnesscoefficients
whichtake accountof surfaceand formroughness,vegetation,and resistance
lossesdue to meandering.The latterof theseis includedexplicitlyin the new
hydraulicmethod and should therefore not be includedin the roughness
estimatesfor the channelor floodplainin any revisedmodel. Considerable

81 sR 329 07/07/93
effort may thereforebe requiredin re-calibrating
existingmodelsif further
studieswereto be undertaken usinga revisedmodellingpackage.

Recommendations

Due to the lack of field data for meanderingcompoundchannelsit has been


impossibleto verifyfullythe new hydraulicmethodand it is suggestedthat the
methodonlybe includedin 1-Dmodelling packagesfordevelopment purposes
at this pointin time.

The most appropriatedevelopmentpath to follow would bd to include the


methodin a single'trial'packageso thatan assessment and evaluationof the
methodcouldbe made. For easeof applicationand interpretation of results,
it would be desirablefor this to be a steady-statebackwaterpackage (oi
steady-statemodule of an unsteadymodellingpackage)with a switch to
enablethe methodof conveyancecalculationto be selectedusinga number
of alternativecalculationproceduresincludingthe newlyproposedhydraulic
method. Testscouldthen be carriedout to find the mostappropriatemethod
of specifyingthe data requiremenlsand lo makecomparisonswith measured
field data over river reacheswith knownor obseruedstage and discharge
information.

8 Conclusions
1) A need to disseminatethe resultsof recent high quality laboratory
researchinto straightand meanderingcompoundchannelshas been
identified
and HRWallingford wascommissioned to presentthisresearch
in a formaccessibleto practisingengineers.

2) Thisreportpresentsworkcarriedouton meandering compound


channels
and extendsearlierwork,Ackers1991,on straightcompoundchannels.

3) Variousstraightchannelmethodswereappliedto fielddatafromstraight
compoundchannels.TheAckersmethodwasverifiedas suitablefor use
in designingstraightcompoundchannels,Section2.4.

4) Variousstraightchannelmethodswereappliedto meandering overbank


data. The per{ormanceof these methodswas found to be poor with
typicalmean errorsin the range-50%to 50%. These straightchannel
methodsshouldnot be usedto estimateflow in meanderingcompound
channels. This confirmedthat the developmentof a new procedurefor
dischargeestir,nation
in meandering compoundchannelsis worthwhile,
Section2.5.

5) A literaturesearchwas carriedout to identifylaboratoryand field data


collectedin meanderingcompoundchannels,Chapter3. The following
sets haveprovedto be of sufficientqualityand quantityto be usefulin
this project.

Laboratorydata: SERC Phase B data; Aberdeendata; Vicksburg


data;Kielydata;Soo(y data.

Fielddata: Rodingstudy

This studywas carriedout by Prof Sellinof the Universityof Bristolfor


the NRA. Recentlyhe has carriedout a laboratorystudy of the River
Blacl$vaterin Hampshire,using the SERC FCF. A three year
programmeof field measurements is due to commencein early 1993.

sR 329 07rc7l93
Thesedatawere not availablefor the workreporledherebut will provide
good validationin the future.

6) Theavailablelaboratorydatacollectedfrominbankmeanderingchannels
were analyzedand,the non-frictionlosseswere foundto form between
15% to 4oo/o of the total energylosses. This confirmedthat bends en
significantlyaffectthe dischargecapacityof channels.

7l A literaturesearchwas caniedout to identifypossiblesourcesof energy


loss in inbankmeandering channels,Chapter4.

8) The mainsourcesof flow resistancein a channelbendare: bed friction;


increasedbed friction due to secondarycurrentsand intemal energy
dissipationdue to increasedturbulenceinducedby secondarycunents.
The flow resistancein a bend dependson bed roughness(f, C n etc);
flow depth (y); bend radius(r"); lengthof bend (l) or angleof bend (0,
| = r" 0) and the cross-sectional
shapeof the channel,Section4.2.

9) Flow resistancein,a set of meanderbendsis likelyto differfrom the


resistanceinducedby a single bend in an otherwisestraightchannel.
This is due to the interaction(growthand decay)betweenthe secondary
currentsinducedin the individualbends,Section4.3.

10) Various methods which account for the extra flow resislance were
identifiedin the lileratureand a selectionof methodswereappliedto the
availablelaboratorydata. The methodswereevaluatedby comparinglhe
meanerors in predicteddischarge,Sections4.4 and4.5.

11) The SCSmethodwasfoundto giveacceptableresultsfor mostpractical


purposeseventhoughit doesnotaccountforthe importantmechanisms
explicitly. An improvedversionof the SCS methodwas formulatedto
removethe undesirable stepfunction(LSCS)andthis linearized
version
gavebetterpredictions, Section4.5.

121 Althoughthesemethods,whichadjustManning'sn basedon the channel


sinuosity,gaveacceptableresultsthey are empiricaland their generality
is not assured. Chang'sapproachin explicitlymodellingthe resistances
due to secondarycunentscombinedwith backrruater calculationsalong
the channel is based on sound theoretical considerationsand is
applicableto both singlebendsand sedesof meanders.

13) A literaturesearch was carried out to summarizethe current state of


knowledgeon lhe detailed flow structurespresent dudng ovettank
meanderingflow and to gauge the effect these might have on the
dischargecapacity,;Section5.1.

14) The internalstructureof cunentsduringovettankflowshas beenfound


to be highlycomplex,Figure14. The most impoftantobseruationsare:

A) The longitudinalvelocitiesbelow bankfulltend to follow the rnain


channelsidewallswhilethefloodplain are generallyin the
velocities
valley direction. Thus the floodplainflows pass over the main
channelandinducea horizontalshearlayer.

B) The energylossdue to secondarycurrentsin the mainchannelis


greaterthan for an equivalentsimple channeland the currents
rotatein the oppositesensecomparedto inbankflows.

83 sFt 32{' 07/07/93


C) Fluidpassesfromthe rnainchannelonto the floodplainand back
intothe rnainchannelin the followingmeanderbend. Hencethe
proportionof dischargepassedby the mainchannelandfloodplain
variesalonga meanderwavelength.Thesebulkexchanges of fluid
betweenslow and fast movingregionsof flow introduceextra flow
resistance.

D) Flowson the floodplainoutwiththe meanderbeltare usuallyfaster


thanthosewithinthe meanderbelt. lt wouldappearthat the extra
flow resistanceinducedby the meanderingmain channelhas a
relativelysmalleffecton the outerfloodplain.

1 5 ) Various methods were identified in the literaturefor estirnating


conveyancein meandedngcompoundchannels. The most promising
ones were proposedby: Ervineand Ellis(1987)and Greenhill(1992).
Themostappropriate channelsubdivisionis basedon a horizontal
plane
at bankfulllevel,Section5.2, Figure17.

1 6 ) The detaileddataavailablefrom PhaseB of the SERCFCFwere used


to developprocedures
whichdescribethe flow resistancein eachof the
zones.

17) The dischargewithinthe mainchannelzonewas foundto varyalonga


meanderwave length. The maximumdischargeis foundat the bend
apices and reachesa minimumat some point in between. The
correspondingdischargein the inner flood plain zone will also vary,
beingminimumat bendapices. The patternof the variationvarieswith
plan geometryand floodplaindepth,Section5.4.

1 8 ) Thesevariations
in discharge
alongthe channelhavebeenignoredin the
analysisand the meandischargein the mainchannelwas used in all
subsequentanalysisand modelling,Section5.4.

1 e ) The dischargein the main channelwas foundto vary with flood plain
depthcomparedto the banKullflow. For low overbankdepths(y'.0.2)
the mainchanneldischargereduces,the rateof reductionappearedto
be independentof the channelgeometry. For higheroverbankdepths
(y'>0.2)the main channeldischargestartsto increase. The capacityof
zone 1 was foundto dependon sinuosity(s); channelshape(B'?lA)and
the relativeroughnessof the floodplains(f'), Section5.4.

20) An empiricaldescriptionof the main channeldischargecapacltywas


developed.Thedischargein the mainchannel(zone1) is calculatedby
adjustingthe bankfulldischargeas calculatedusingstandardmethods.
Theadjustment factoris the greaterof thevaluesgivenby Equations
(88)
and (89),Section5.4.

211 The dischargecapacityof the inner flood plain was consideredin two
ways. Oneprocedureis basedon a semi-empirical modelofexpansion
and contractionlosses. The other method is based on an empirical
analysisof the SERCphaseB and Aberdeendatasets,Section5.5.

221 Independentdata on expansion-contraction


lossesover slots (Jasem,
1990) were used to derive correctionsto the standardmodel, for
expansion-contraction
losses,to take accountof channelwidthto depth
ratioand sideslope,Section5.5.1.

23) Mainchannelcross-section
shapeand sinuositywerealsofoundto affect
the magnitudeof the expansion-contraction
lossesin the inner flood

sR 329 07/07/93
plains. The Aberdeendata was used to developconectionfactorsfor
theseparameters,Section5.5.1.

(Jamesand Wark) modelfor lhe innerflood


241 The expansion-contraction
plainflowis givenby Equations
(109,108,107,106,102),Section5.5.1.

26) Empiricalanalysisof the innerflood plain dischargesshowsthat at low


overbankstages(y'.0.2) lhe innerflood plain dischargeincreaseswith
depth. At higher overbank depths (y'>0.2) the discharge reduces,
Section5.5.2.

271 The empiricaldescription(Jamesand Wark 2) for the inner flood plain


ffowis givenby Equations
114,115,116and 117,Section5.5.2.

28) The dischargesin the outerfloodplainzones(3 and 4l arecontrolledby


bed frictiononly,Section5.6.

2e) The bed shear stress data collected on the SERC FCF has been
analyzedin orderto provldegeneralguidelines. Bed shearstressesin
the rnainchannelduringoverbankfloware lowerthanthosewhichoccur
at bankfullconditions.On the flood plain during overbankflow high
concentrationsof bed shear stresseshave been obserued. Equations
120 and 121 give the peak values on up and down stream banks,
Section5.7.

30) The two methodsdevelopedby the authors(JWand JW2)and four other


methods have been used to predict dischargesand stages for the
availablelaboratorydata. The authorssemi-empiricalmethod(JW)was
foundto givemarginallymoreaccuratepredictionsof totaldischargeand
acceptablepredictions Section6.3.
of discharges,
of the distribution

of theJamesand Warkmethodto variationsin the values


3 1 ) The sensitivity
of both meanderwavelength(L) and mainchannelsideslope(S") has
been investigated.The resuhsindicatethat greatprecisionin estirnating
thesevaluesis not required,Section6.3.3.

321 The two models(JW and JW2)have been appliedto the best field data
available,the river Rodingstudy. The JW methodwas fourd to give
better predictionsof total dischargeand is recommendedfor use in
practice,Section6.5.

33) The sensitivityof the resultsto variationsin bed roughnessvalue has


been investigated.The non-frictionenergylossesare shownto be less
importantas the floodplainis roughened.Bed frictionrernainsthe most
significantsourceof energyloss in riverswithoverbankflow,Section6.5.

34) The Jamesand Warkmethodwas appliedto laboratorydatacollectedin


straightcompoundchannels.The dischargewas underpredictedby27o/"
on average.Thisconfirmedthat the methodsdevelopedfor meandering
compound channels should not be applied to straight compound
channels,Section6.4.

35) It is recommended that the JW methodshouldbe used for compound


channelswithsinrrcsities lessthanor
greaterthan 1.02. For sinuosities
equalto 1.02it is recommended that a suitablestraightchannelmethod
shouldbe used with an appropriatecorrectionfor sinuosity,eg Ackers
(1991),Section6.6.

85 sR 329 l0l07/93
36) The availabledata usedto verifothe Jamesand Wark methodcovered
a limited rangeof conditions. Furtherexperimentalwork is requiredto
look at both total dischargesand the distributionof dischargesfor :

A) Meandering (<1.09).
channelswith low sinuosities

B) Meanderingcompoundchannelswith roughflood plains.

C) Low overbankdepths(y'. 0.2)

37) With the currentlyavailabledata, no furthersignificantimprovementsof


the new methodcouldbe achieved. New informationmust be obtained
beforeany substantialfurtherdevelopmentof the methodis undertaken.

38) The work uncovered some gaps in the existing knowledge and
recommendations have been given for furtherresearchto improvethe
currentunderstandingof the mechanicsof flow in meanderingchannels,
Chapter7. The futureresearchwhichstill remainsto be canied out falls
intothreemaincategories:

A) The collectionof independentinformationto use in verifyingthe


workpresentedin this repod. Thisincludeslaboratorystudiesand
field measurement programmes.

B) The development of two and threedimensional numericalmodels


and their applicationto the avaihble laboratorydata.

C) The use of the proceduresdevelopedin this report within one


dimensionalriver rnodelshas not been fully considered. The
proceduresweredevelopedto modelsteadystatestagedischarges
and the type of work requiredto confirm that the methods are
appropriatefor usein 1-Dmodelsis describedbelowin Section7.4.

3e) Computationalmodellingincluding3-D turbulenceand 2-D modelling


techniqueshave been identifiedas promisingmethodsto use in further
developmentof the understanding of the complexmechanicsof flow in
meanderingcompoundchannels.

40) Before includingthe new method in general 1-D river models it is


recommendedthat the method is incorporatedin a 'trial' rnodelling
packageso that a full assessmentand evaluationof its pedormancecan
be made. i

9 Acknowledgements
The workwas doneduringthe first autho/s sabbaticalvisitto HR Wallingford,
for which additionalfinancialsupport was providedby the Foundationfor
Research Developmentand the Munay and Roberts Charles Skeen
Fellowship.

The contributionsmadeby Dr PaulSamuelsand Dr NigelWalmdeyto the


conduct of the proiect are greatly appreciated. The assistanceof Mrs M
Johnstone, in collatingthe Phase B Data sets is gratefullyacknowledged.
The assistanceof The NRA membersof the SteeringGroup is gratefully
acknowledged.

The academics ard research assistants who canied out the Phase B
investigationon the SERCFCFalsogaveinvaluableassistancein collalingthe

sR 329 2007/93
availabledataand providedthe projectwithsomeresultsof theirown analysis.
The assistanceof the followingis gratefullyacknowledged:Prof B B Willetts
and Dr R Hardwick(The Universityof Aberdeen); Prof R H J Sellinand Dr R
Greenhill(The Universityof Bristol); Dr D A Ervineand Dr M L Lorena (Ihe
Universityof Glasgow)and Dr D W Knight(The Universityof Birmingham).

87 sR 329 07/07/93
10 References
AckersP (1991)The hydraulicdesignof straightcompoundchannels,Reporl
SR 281, HR Wallingford,
October.

AgarwalV C, GardeR J and RangaRajuK G (1984) Resistance to flowand


sedimenttransportin curyedalluvialchannels,FouilhCongress,Asianand
PacificDivision,IAHR,Thailand,11-13September,pp 207-218.

BridgeJ S (1992)A revisedmodelfor waterflow, sedimenttransport,bed


topographyand grainsize sodingin naturalriver bends,Water Resources
Research,Vol 28, No 4, pp 999-1013.

Corps of EngineersUS Army, WaterwaysExperimentStation, Vicksburg,


Mississippi(1956) Hydrauliccapacityof meanderingchannelsin straight
floodways.

CarlingP A (1992) Personalcommunication.

ChowV T (1959)Openchannelhydraulics,
McGrawHill,ISBN0-07-
Y85906-X.

ChangH H (1983) Energyexpenditurein curvedopenchannels,


Journalof
Hydraulic Vol 109,No 7, pp 1012-1022.
Engineering,

{r" Chang H H (1984) Variationof flow resistancethroughcuruedchannels,


JournalofHydraulic
Engineering,Vol 110,No 12,pp 1772-1782.

ChangH H (1988)Fluvialprocesses
in riverengineering,
JohnWiley& Sons,
lsBN 0-471-631396-6.

CowanW L (1956) Estimatinghydraulicroughness,


Agricultural
Engineering,
Vof 37, No 7, pp 474-475.

EinsteinH A and Banks R B, (1950). -Fluid resistanceof composite


roughness,
Transactions,
AmericanGeophysicalUnion,Vol 31, No 4,
pp 606-610.

ErvineD A and EllisJ (1987) Experimental and computationalaspectsof


overbankfloodplainflow, Transactionsof the Royal Societyof Edinburgh:
EarthSciences,Vol 78, pp 315-475.

ErvineD A and JasemH K (1991) P/ersonalcommunication

GreenhillR (1992)An investigation


intocompoundmeandering channelflow,
PhDthesis,Depadmentof Civil Engineering,Universityof Bristol.

Hayat S (1965) The variationof loss coefficientwith Froudenumber in an


open channelbend. Thesis presentedto the Universityof lowa in pailial
fulfilmentof the requirements
for the degreGof Masterof Science.

HendersonF M (1966)OpenChannelFlow,Macmillan.

HoldenA P and JamesC S (1988) Dischargecomputation for compound


channels,The CivilEngineerin SouthAfrica,Vol 30, No 8, pp 371-376.

HortonR E, (1933)."Separateroughness for channelbottomand


coefficients
sides.. Engineering
News-Record, Vol 111, No 22, pp 652-653.

sF 329 07/07/93
HR Wallingford (1988) Assessing the hydraulic performance of
environmentally
acceptablechannels,Repoil EX 1799, HR Wallingford,
September.

(1991) FhaseA & B FloodChannelFacilityManual,Inception


HR Wallingford
Report,HR Wallingford,ReportEX 2485(November)

HR Wallingford
(1992) PhaseA & B FloodChannelFacilityManual,Interim
ReportEX 2548(March)
Repofi,HR Wallingford,

James M and Brown B J (1977) Geometricparametersthat influence


floodplainflow,Corpsof EngineersUS Army,WaterwaysExperiment
Station,
Vicksburg,Mississippi.

JasemH K (1990)Flowirrtwo stagechannelswiththe mainchannelskewed


to the floodplaindirection,PhDthesis,Universityof Glasgow.

Kazemipour A K andApeltC J (1979) Shapeeffectson resistanceto unilorm


flow in openchannels,Journalof HydraulicResearch,Vol 17, No 2, pp 129-
148.

KazemipourA K and ApettC J (1983) Energylossesin irregularchannels,


JournalofHydraulic Vol 109,No 10,pp 1374-1379.
Engineering,

Kiely G, Javan M and McKeoghE J (1989) A comparisonof turbulence


measurements in straight,singlemeanderand multiplemeanderchannels,
Int'|.Gonfr.on ChannelFlowand CatchmentRunoff,CentennialofManning's
and Kuichling'sRationalFormula,Universityof Virginia,May.

Kiely G (1989) Expedmentalstudy of the mechanismsof flood flow in


meanderingchannels,Proc.of the 23rd IAHR Congress,Ottawa,Canada,
August.

Kiely G (1990) Overbankflow in meanderingchannelsthe important


mechanisms,Procof IntlConfon RiverFloodHydraulics, 17-20
Wallingford,
September.

KikkewaH, lkeda S and KitagawaA (1976) Flow and bed topographyin


curuedopenchannels,Journalof the HydraulicsDivision,ASCE,Vol 102, No
HY9, pp 1327-1342.

KnightD W and Sellin'RH J (1987)The SERC FloodChannelFacility,


Journalof the Institution
of Waterand Environmental Vol 1, No
management,
2, pp 198-204.

KnightD W, YuanY M and FaresY R (1992)Boundaryshearin meandering


channels,Proc.Int. Symp.on HydraulicResearchin Natureand Laboratory,
Wuhan,China,November, 1992.

M and ChristensenB A, (1972). 'Equivalentroughnessfor


Krisnamurthy
shallowchannels*.Journalof HydraulicsDiv, ASCE,Vol 98, HY12,
pp 2257-2262

LeopoldL B, BagnoldR A, WolmanM G and Brush L M (1960) Flow


resistancein sinuousor irregularchannels,USGSProfessional
Paper282-D,
Washington, DC,ppl 11-134.

LorenaM L (1991) Personal


Communication

89 sR 329 07/07/93
LotterG K, (1933). 'considerationson hydraulicdesignof channetswith
differentroughnessof walls', Transations,All-UnionScientificResearch
fnstitute,Leningrad,Vol 9, pp 238-241.

McKeoghE J and KielyG (1989)A comparisonof velocitymeasurements in


straight,single meanderand multiplemeanderchannels,Int'|. Confr. on
ChannelFlowandCatchmentRunoff,Centennial of Manning's
and Kuichling's
RationalFormula,Universityof Virginia,May.

MockmoreC A (1944) Flowaroundbendsin stablechannels,Transactions,


ASCE,Vol 109,pp 593-618.

MyersW R C (1991)Personal
communication

NationalRivers Authority(1991) Design Method for StraightCompound


Channels,R&DNote44. NRA.

NelsonJ M and SmithJ D (1989)Flowin meandering channelswith natural


lopography,in River Meandering,lkeda S and ParkerG (Eds),American
GeophysicalUnion,WaterResourcesMonograph12, pp 69-102.

OnishiY, JainS C and KennedyJ F (1976) Effectsof meandering


in alluvial
streams,Journalof the Hydraulics ASCE,Vol '102,No HY7,
Division,
pp 899-917.

PoseyC J (1976) Computation


of dischargeincludingover-bankflow, Civil
ASCE,pp 62-63.
Engineering,

PrinosP and TownsendR D (1983) Estimating


dischargein compoundopen
channels,in Proceedings
of the 6th CanadianHydrotechnicalConference,
Oftawa,Canada. R D Townsend(Ed),vol 1, pp 129-146.

RangaRaju,K. G. (1970).Resistance
relations
for alluvialstreams,
La Houille
Blanche,
No 1.

RamsbottomD M (1989) Flooddischargeassessment,


interimreport,Report
SR 195,HR Wallingford,
March.

RouseH (Ed)(1950)Engineering
Hydraulics,
JohnWileyandSons,NewYork.

RozovskiiI L (1957) Flowof waterin bendsof openchannels,TheAcademy


of Scienceof the UkraineSSR, translatedfrom the Russianby the lsrael
Programfor ScientificTranslations,
Jerusalem,lsrael,1961.

ScobeyF C (1983)Theflowof waterin flumes,US Department


of Agriculture,
TechBull No 393.

SellinR H J and GilesA (1988) Two stagechannelflow, Final Reportfor


ThamesWaterAuthority, of CivilEngineering,
Department of Bristol.
University

SellinR H J and GilesA (1989) Flow mechanisms in spillingmeander


channels,Proceedings,
23rdCongress,1AHR,Ottawa,August1989.

SellinB H J, GilesA and BeestenD P (1990) Post-implementation


appraisal
of a two stagechannelin the riverRoding,Essex,JrnlIWEM,1990,4, April.

SmithC D (1978) Effectof channelmeanderson floodstagein valley,Jml


HydrDiv,ASCE,Vol 104,HY1,pp 49-58.

90 sR 329 07107/93
SoilConseruationService(1963) Guidefor selectingroughness
coefficient'n'
valuesfor channels,USDepailmentof Agriculture,
SoilConservation
Services,
Washington.

SoohyA (19il) The flow througha meanderfloodplaingeometry,Ph D


Thesis,PurdueUniversity,(65-2&{7),Universitymicrofilmslnc, Ann Arbor,
Michigan.

SteinC J and RouveG (1988) 2D-LDV-Technique for measuringflow in a


mearderingchannelwith wettedflood plains - a new applicationand first
results,ProcIntl Confon fluvialhydr,Budapest,June 1988.

numericalpredictionsof
SteinC J and RouveG (1989) 2D depth-averaged
the flow in a meandering channel with compound cross seclion,
Hydrosoft,1989,Vol 2, no. 1.

Toebes G H and Sooky A (1967) Hydraulicsof meanderingrivers with


Jrnl\A/Wand Har Div,ASCE,Vol 93, WW2,pp 213-236.
floodplains,

Wark J B, SamuelsP G and ErvineD A (1990) A practicalmethodof


estimatingvelocityand dischargein compoundchannels,Procof IntlConfon
Wallingford,'17-20September.
RiverFloodHydraulics,

Wark J B, RamsbottomD M and Slade J E (1991) Flood discharge


assessment method,RepodS.R277,HR Wallingford.
bythe lateraldistribution

Wark J B (1993) Overbankflow in straightand meanderingcompound


channels,ph.dThesis,The Universityof Glasgow,Underpreparation.

WillettsB B, HardwickR I and Maclean A G (1990) Model studiesof


overbankflow from a meanderingchannel,Procof Intl Confon RiverFlood
Hydraulics, 17-20 September.
Wallingford,

WillettsB B and HardwickR I (1991) PersonalCommunication

WillettsB B (1992)The hydraulicsof overbankflow in meanderingrivers,


UK.
Seminaron MultipleStageChannels,HFIWallingfordLtd.,Wallingford,

WormleatonP R, AllenJ and HadjipanosP (1982) Dischargeassessmentin


compoundchannelflow,Joumalof the HydraulicsDivision,ASCE,Vol 108,
No HY9,pp975-994.

WormleatonP R and MerrettD J (1990) An improvedmethodof calculation


sections,Journal of
for steady uniformflow in prismaticmain channeUflood
HydraulicResearch, Vol28, No 2, pp157-174.

Yen B C and Yen C L (1983) Flood flow over meanderingchannels,


Rivers'83, RiverMeandering,
Proceedings, ASCE,pp 554-561.

sR 329 07/O7l93
91
Tables

sB 329 07/07193
Table 1 River channel and floodplain widths

Rivers Site BT Bmc Bfp Slope ob

Blackwater Blackford 72.O 6.0 66.0 1.60x10-3 8.50


Main6 Section6 27.5 1s.7 13.8 1.906x10-3 12.75
Main14 Section14 12.8 13.5 3
1.906x1O 16.06
26.3
Ouse Skelton 68.5 54.0 14.5 1.46x104 250.20
Severn Montford 125.0 35.0 90.0 1.95x10{ 183.30
Tees Low Moor 186.0 56.0 130.0 8.00x10{ 266.24
Tonidge Torringiton 120.0 30.0 90.0 1.45x10-3 190.00
Trent N. Muskham 180.0 72.O 36.0 3.20x10{ 389.60

Notes

1 All dimensionsin metres


2 B total widthof channeland floodplains
3 Bmc totalwidthof channelat bankfull
4 Bfp lotal width of floodplains(Bt - Bmc)
5 Ob bankfulldischarge

Table 2 River channel depths and roughness values

AuthorsEstimate Ackers Estimate


River BFSTGE h. Hmax nb nfl nb nfl nf

Blackwater 1.70 1.70 3.58 0.046 0.094 0.099


Main 6 0.90 0.90 2.20 0.032 0.040 0.040 0.030 0.050 0.050
Main14 0.92 0.92 2.00 0.0278 0.040 0.040 0.02 0.020
Ouse 4.30 8.85 9.94 0.0448 0.060
Severn 4.09 5.75 7.75 0.031 o.025 0.045 0.0307 0.0338 0.0338
Tees 8.50 4.36 6.67 0.056 0.100 0.100
Tonidge 17.20 2.78 5.29 0.027 0.060 - o.o24 o.oru o.oru
Trent 7.60 5.70 8.21 0.032 0.032 0.032 - 0.032

Notes

1 BFSTGE Bankfullstage
2 Channeldepthat bankfullstage
3 Hmax Channeldepthat stagecorresponding to the highestsurueyedlevelon the cross-section
4nb MainchannelbankfullManning'sn
5 nfl, nfr Left and riightfloodplain Manning'sn

sR 329 07/07/93
Table 3 Mean errors for straight field data

Data Set A B c D
NOP 127 118 77 68
MSD MSD MSD MSD

Authors n values
LDM 1.2 6.7 1.2 6.8 0.7 5.2 0.7 5.2
DCM -1.0 6.6 -1.0 6.7 -0.4 4.5 -0.3 4.4
scM -11.9 15.6 -12.6 15.9 -10.9 17.5 -12.1 18.2
scM2 -17.6 20.3 -18.7 20.6 -12.8 19.8 -14.1 20.6
scM3 8.1 11.9 8.0 12.2 6.1 9.6 5.8 9.9
scM4 -18.2 20.6 -19.3 20.9 -13.3 20.4 -14.7 21.2
scMs -12.9 16.2 -13.8 16.5 -10.8 17.5 -12.O 18.2
ssGM 8.1 11.s 8.0 12.2 6.1 9.6 5.8 9.9
DCM2 1.0 7.4 1.1 7.5 0.1 6.6 0.3 6.7
FCFAM -4.4 8.4 -4.2 8.0 -2.8 6.8 -2.2 5.5

Ackers'n values
LDM 5.0 6.s 4.5 6.6 5.5 4.9 4.6 3.9
DCM 2.2 7.4 1.5 7.O 3.5 5.0 2.5 3.8
scM -15.0 13.1 -16.9 11.4 -17.2 12.5 -20.9 7.4
scM2 -20.5 16.s -22.8 15.0 -19.4 13.9 -23.3 8.9
scM3 14.0 9.6 't3.4 5.4 4.3
9.5 14.4 13.3
scM4 -20.7 17.O -23.1 15.0 -19.4 13.9 -23.3 9.0
scMS -16.3 13.4 -18.4 11.4 -17.8 12.8 -21.5 7.8
ssGM 14.0 9.6 13.4 9.5 14.4 5.4 13.3 4.3
DCM2 5.0 6.8 4.6 6.6 5.3 4.9 4.6 4.1
FCFAM -2.8 7.6 -3.4 6.8 -2.O 6.3 -3.1 4.5

Notes:

1 NEV = 0.16 in LDM 2M - meanSD - Standarddeviationin mean


3 Error= loo*(Qc.b- Q*JQ'*""
4 The data for the Severn and Trent has been smoothed using runningaveragesof three
consecutivedata points
Meanstaken over followingsubsetsof availablestagedischargedata:

A Blachruater,
Main 6, Main 14, Ouse,Severn,Tees,Torridge,Trent
B As A with Ackersestimateof BankfullStagefor Toridge
c Severn,Torridgeand Trent only
D As C with Ackersestimateof BankfullStagefor Torridge

sR 329 07rp7i93
Table 3a Mean errors for Myers lab data

M SD
LDM o.2 7.2
DCM -1.O 6.8
SCM -13.2 6.3
scM2 -13.2 6.3
SCM3 5.7 7.2
scM4 -13.2 6.3
SCMs -13.2 6.3
SSGM 5.7 7.2
DCM2 1.7 7.7
FCFAM -5.2 5.2

Notes

1 NEV = O.16in LDM


2 M - mean SD - Standarddeviationin mean
3 Error 100'(Qc.b- Q,o"J/Q,,*u"
=
4 This Datawas obtainedfrom seriesA and F of Myerslab data nb =0.01,nf = 0.01,slope=
1.906x103,numberof observations= 20.

sR 329 07/lr7l03
o a q qc oqqgq c9.c - q . ! q q q g q o qq a
o (ltot-t\@t-Fcr(oo @(ot-i-@i-t\@@@
cct
E()
F< qq.q.ta?.c aa? 1q qc a a al.t aa? a ol (o
lO(4r(O(OF@(oF(tOl
' . F F l e * U)Ci)(o(oF@(oF(trN
o 1a
U'
o
() ol\olq.tGt\c?\q e .tc!ulalulolo!s?q
u)lo@o@oq)(0 lou) (O(oOrO)FOrO)F-@(('
'o
o)
1J
Flt (!Ol@O$tOOr 14ql\':--'::-t':
oO t ': : d. .( c\ tdt ; o i l ; r i . f rt(OtNu)Fu)U)\tT U)
tsa u)(\1

o o!\ -cYo|q o!qq . q o ?\ o q u ? q q q - g


() ( o ( o a t { ( o ) l o t (\o ( t o F-}-\tlo}-ro3F-t-o
FOI OIF

c!G?C-ol q -q ol(' ro o.q e Ah ol u) t o,l


o. i < ,cF jo ;l u t o i o. i d c j d ddoadaoddaj u;
NF
o
a
o \( a ?aaqqqolag \( oa. ?O -F qO r( o{ @
e Oq lq@o@l @
ag
o) O(OraO(o@Ol(o(oO
Ec>
F? qoqc!qcqagaol oloqc!qcqaqao?
t O s It | - f - @ r t t - O ( o $o$F (o
rC\l OIF
' T. F$- @ u ) F . Oq@, I T (9

o
o
r( vg)o( tql /q) u. qt oor l( q) u. )t (' t: rqt g , \\qq-u?q-qa?
otollr)@(r)@@(D('r(r)

RO
o
'6 e? ll)oo@t@9\trr)F
* o i c t d $ d -u i + u t t '-t
qq-qu?o?u?aa?-q-
(9FU)(')\t€D(OSU)t
qr$-cycI-
CO
ql qt ql c,.l
E
o .9
E Eo olaqq.tqqaoqoq olaqqaqqaoqoc .;(d
aa t\t\NN0oN1..(Ot-F. t-F-t-I-@t*f.-@NF.
q)
* o
*. 3<: qoltqcqalagola qotu?qcolaqqa E(!
€ d? (ooN@(o@(tlooo (ooot@u)@('rooo
.9 E
c(l)
(ug)
S
v, o
u)
\r +o ttcu@l qN e$ oq t\oor q! cq! iooq \eu?qqqqs?\q
@lO@(o(o@@(o(ot-
6S
'I6 or E
r3
€ EA.v
o : cqqolea\qc!q qqqqqu?qolu?-
s
oC
(It -(It
G
o
.EO
E?
tr)OOrO(DrF(oqrF
Ol-
t\ttO)rOFF-O@F
g trf
\ 6Ho
e P69
E;E
e
g
o
a
C qqaqaclgrol
-ortNNolOlOlNo)
qq\\otq\o!-':
O)@IOF()FOOF.o TJ (E'F
gPE
(O

.EO N OrtO 9{'!'cq:olN:r.(o


€6 e
o E? oouluisuio.+-o)
ut;?TNTThjsq 9toq'RPFePd{, F.
b. --( U0Q8
q 6JJ<
G €\1 sr
S tA
o
o OFCtOloo)or|t)rc?
ddr<tdt.dc,i.dqtdP
OFC,OlJ)O,O,lI)FO?
d@utddoi.addP E?E
nP
$c>
\t t?
6
9-o\ola?qga?o-t $-o\olo?q9c?oa
E*rsFE$$Ed;
q
9 o*8
z<E
a
o o E.n$Fsssstof c.
a o o
o
N 'F ! FIIJ
6= $asFgFFfrfifr$
€asHHHfrgFg$ o
ZrNC/)

sR329 07/07/9
Table5 Summary of SERCPhaseB stagedischarge fests

Run Angle FlowType X- FP FP Config No of


No Sn Data
Points
B.20 60 lnbank T S NI/A 17
B,21 60 Ovefuank TSSt16
825 60 lnbank NSSt10
826 60 Oveftank NSSr16
831 60 Overbank NSNa'14
B'32 60 Overbank N R.BB St 13
833 60 Oveftank N R-PD St 12
834 60 Overbank N R.D St 18

838 110 lnbank NI/A 11

839 110 Overbank NS st 14

843 110 Overbank N R-D St 15

846 110 Overbank N R.BB st 14

B.47 110 Overbank NS Na 14

848 110 Overbank W I

Notes

1 T = Trapezoidal
2 N = Natural
3 S = Smooth
4 R-D = Roughenddowel rods
5 R-PD = Parthlly rougheneddowelrods
6 R-BB = Roughenedwith breezeblocks
7 St = Standad
I Na = Nanow
I W = Walled
10 i.l/A = Not Applicable

sR 32{' l7l0993
@
Table 6 Mean errors straight methods meandering data

Method All Data Smooth Data Rough Data


Mean SD Mean SD Mean SD
Error o/" Error o/o Error o/o
DCM 38.5 17.8 44.8 16.2 24.4 12.5
scM 37.6 56.7 7.7 10.0 104.6 61.0
SSGM 70.1 30.6 69.4 29.9 71.6 32.5
DCM2 41.6 16.8 47.O 15.9 't1.7
29.3
FCFAM 24.8 26.0 39.9 14.2 -9.1 8.2
HORl 30.8 29.0 42.4 21.5 4.6 26.6
HOR2 13.8 23.5 25.5 16.8 -12.3 12.7
HOR3 20.6 23.3 32.9 14.2 -7.O 14.4
HOR4 7.3 23.2 19.5 16.6 -19.8 6.6

Notes

1 o/oError = 100*(Qcalc-Qmeas)/Qmeas
2 SD = $16n6ardDeviationin Meano/oError
3 Alf Data- 8.21,826,831, 834, 839, 843, 847
4 SmoothData- 821, 826, 831, 839, 847
5 RoughData- 834, 843

sR329 0707
o-
.9! t tt
o

..2


3 tl
F

o
c0
o.v,

Eg
€ie e
o tat t+tt

o-
L
(D S
z* * zz z Eg€ *E
f

c(U
EEE.E+
,.c
()
(D
f
Itt f,tttla+
;ffiFEE$;-
o r I SS<.. <<<s. i.
U' zzz zzzz zzz ltililllllllllllll

oc0
g o(Ld)
q
CD
c arirLrL6232 -
o
rh. o
o {--eefit

>aa@@@aaaa@z@@@@@u) Or(\lC){lO(o}-@
a (L
ll.
FFFTFTFFF

o
v, o
CL
G eQOaooo
F
to o-
|r
@ourcnaq@@@ao@9tEEEE
EE' c
o
EE
fr
(') x
c
a?F F F I - I - Z Z Z Z Z Z Z Z Z Z Z Z Z 83oo-
$$Frg
2
\
o
b
G
E
$EiEAt€:
EEsffifiEgfffif;gsggf
l.trtrtr

TEBISE'gRRr
s
u, -oo r o
c (0 o
(cr o o o o o o o o o o o o o o o o
Gt (o (c, (o (o (o (o (o (o (o (cl Gt (0 (o (0 (0
=
U,F,EE
55 LEE
8Eg
8E€g,gEf
t\ 11il ilnilil[llll

o o
a Z o - 6t o \t ro @ t- @r o, o r or o) $ lo @ t-
a (\t (\t (\t N N c! N (\t ol N cqt (v) (r) cl (f) (f) ({l (Il
= ao co dt d) dt dl dt co dt (D {D dl dl gl co co dl cl
ji ts IsFsE;=
N CE o
Z rN O.tlO(Ot-@O,

sR929 l7lo3r9e
o-
.9, r
o

.9, +tt

€ ata
J
F

@ r3t
CD

o r*+ cc

(L
LL 8e
@ zz e8
f 5E
,^30
c
(U
-c
o
€f H e6
ata E E5
(D
f EsE
E dE
*E
Eg
- q,
Il.
o Q+r
c Etr
I a I b = o H.s
=
io u)zzzzz
.h.

a
o
€t
c
FsFgEgi:
o ilil[illlllllll
(rl
G
6
(L O(LCD
oa0
d&d6272=
f tr

o o
.o-= - ^ o o o H =SP:P9FP
ff
to
FA@ct)cno===ry@u)
LErscc
ll.
\ E,'
o U'C
C
co
tat ulzzzzzzzzzzz
x gE .h a
E
E + = = g + = = + i -Y
tE*r#E
ggEE
5 -oE€918R€18R€€€
v,
:6AAA6AA666
Eu-a;;i(o;xooo

$
q g^ooooooooooo
o
{r
t\
YrFrr

ll lt ll ll ll !t ll ll ll ll

o o
tg SsESE*=,
a
zog)or(\r(Y)vrrr(gN@
ecqt(f)ssf!fr(tsvsr.rrsr o
o
g 5COco(DCrld)co(D(Dc0d)dt
[E oo
Z r(\l (')r+u)(Oi:@Or-

sF 329 l7r'm/93
Table 8 Summary of Aberdeen experiments

Sinuosity Cross-section ValleySlope Test No


1.00 Trapezoidal 0.00100 ABlOO
1.@ Trapezoidal 0.00071 ABlOOA
1.21 Trapezoidal o.oo100 AB1Ol
1.40 Trapezoidal 0.00100 AB102
1.40 Natural 0.00100 AB103
2.06 Trapezoidal o.00062 ABl04
2.06 Natural 0.00062 AB105

Table 9 Summaryof Vicksburgexperiments2ft wide channel

Test Floodway Sinuosity MeanderBelt Radiusof Assigned


No Width (m) Test No
widrh Curuature
(m) (m)
xtl 4.877 1.570 4.420 1.829 201 202203
xilt 4.877 1.400 3.761 1.865 204 205 206
XIV 4.877 1.200 2.822 2.137 207 208 209
XV 9.144 1.200 2.822 2.137 210
XVI 9.144 1.570 4.420 1.829 211

SR 3?g lTrW'&l
Table 10 Geometricparameters lab studies meandering channels

Source L B h rc s
Wave Channel Channel Radiusof Sinuosity
Length Width (m) Depth(m) Curuature(m)
(m)
SERCFCF 12.000 1.200 0.150 2.743 1.374
10.310 o.174 0.150 2.743 2.U3
Aberdeen 2.570 o.174 0.050 0.413 1.215
1.909 o.174 o.050 0.413 1.406
1.154 o.174 0.050 0.307 2.O43
Vicksburg 7.315 0.762 o.152 1.829 1.571
7.315 o.762 0.152 1.865 1.400
7.315 o.762 0.152 2.136 1.200
Kiely 1.803 0.200 0.050 0.400 1.224
Toebes& 1.280 0.209 0.038 1.392 1.090
Sooky
Smith 3.352 o.274 0.076 1.097 1.172
James& 9.144 o.279 0.051 1.143 1.068
Brown
Stein& Rouve 6.500 0.400 0.100 1.800 -1.200

sR 329 17lO3/9
Table 11 Nondimensional geometric parameters meandering
channels

Source UB B/h AJB


NaturalRivers 10.0 10.0 2-3.O
SERCFCF 10.0 8.0 2.3
8.6 8.0 2.3
Aberdeen 14.8 3.5 2.4
11.0 3.5 2.4
6.6 5.0 1.8
Vicksburg 9.6 5.0 2.4
9.6 5.0 2.5
9.6 5.0 2.8
Kiely 9.0 4.0 2.O
Toebes& Sooky 6.1 5.5 6.7
Smith 12.2 3.6 4.O
James & Brown 32.8 5.5 4.1
Stein& Rowe 16.2 4.O 4.5

sR 329 l7l0993
Table 12 Bend losses for 60"meander geometry, trapezoidal cross-
section

'Flow Discharge T So-S, K n'ln


Depth(m) (m%) fc) (x10-)
0.05932 0.01975 11.4 0.0634 0.0851 1.0477
0.06726 o.02512 11.0 o.0228 0.0312 1.0201
0.07198 0.02654 12.4 0.1039 0.1168 1.0805
o.07714 0.03056 10.8 0.0701 0.0690 1.0524
0.08263 o.03308 12.4 0.1176 0.1146 1.0920
0.08608 0.03630 11.6 0.0882 0.0780 1.0673
0.09170 0.04015 11.5 0.0947 0.0786 1.0279
0.09765 o.04425 11.6 0.1050 o.0824 1.0806
0.10192 0.04708 12.7 0.1203 0.0916 1.0943
0.10302 o,04.782 11.5 0.1185 0.0894 1.0930
0.10593 0.05015 11.7 o.11BB 0.0868 1.0931
0.10596 o.44974 10.9 o.1252 0.0930 1.0995
0.10680 0.04953 12.4 0.1489 0.1136 1.1215
0.11150 0.05467 13.7 0.1261 0.0869 1.0995
0.11394 o.o5702 12.7 0.1163 o.0772 1.0916
0.11900 0.06035 13.6 o.1377 0.0899 1.1110
0.13150 0.07073 12.9 o.1457 0.0867 1.0576

Average: 0.1073 0.0809 1.0782


StandardDeviation 0.0320 o.4262 0.0285
'At cross-over
section

sR s29 17w,93
Table 13 Non-friction losses for 60" meander geometry, natural
cross-section

*Fbw Discharge T K n?n


So-S,
Depth (m) (m%) fc) (x10')
0.09957 0.01019 16.1 0.2569 o.6762 1.2329
0.10359 o.01207 16.1 o.2284 o.5202 1.2084
0.10860 o.o1442 16.1 0.2130 o.4228 1.1900
o.11225 0.01612 16.0 o.2248 0.4209 1.20/,O
0.11&18 0.01806 16.1 0.2351 0.4146 1.2166
0.12316 o.02150 16.0 o.2407 0.3807 1.2237
0.12566 0.02288 16.0 o.2417 0.3669 1.2249
0.12923 0.02498 16.2 o.2413 0.3445 1.2244
0.13165 0.02646 16.0 0.2398 o.3287 1.2225
0.14235 0.03341 17.1 0.2398 0.2789 1.2227

Average: o.2362 o.4154 1.2170


StandardDeviation: 0.0118 0.1125 o.o127
'At cross-oversection

sF 329 1710993
Table '14 Non-friction losses for 1'10"meander geometry, natural
cross-section

*Flow Discharge T So'St K r{ln


Depth(m) (m%) fc) (xto'3)
0.11006 0.01135 10.4 0.1805 0.7819 1.2512
0.11516 o.4fi22 15.0 0.1920 0.7558 1.2743
0.12030 0.01533 10.5 0.1857 0.fl&+ 1.2615
o.12073 0.01560 14.4 0.1881 o.6228 1.2ffi4
o.12420 0.01699 14.1 0.1881 o.6228 1.2664
o.12791 0.01873 10.3 o.1774 0.5459 1.2452
o.13072 0.02006 15.9 0.1878 0.5502 1.2657
0.13566 o.02206 14.9 0.1950 o.5473 1.2806
0.13854 o.02342 10.3 4.1872 0.5057 1.26/,6
oi4a27 o.02432 10.5 o.1868 0.4908 1.2638
o.14672 o.02778 10.4 0.1844 0.4{}97 1.2589

Average: 0.1860 0.5935 1.2622


StandardDeviation: 0.0051 0.1077 0.0103
* At cross-over
section

sB 329 l7l@/93
g qqqeqEqqEq Ei Eq
9 \' c,i trt .d h di g c,i f ; ci = T ot

o
c
=6
I E T;h B; E S B R R e d
8 s ut R EF-
.E o { qi $ I c\i oi .), T
!E
E
o

o
.oE
^
-g'EBqbiRN:e= ls Eg
ci S ct N 9
E
Hg. q, $ F. e al ? d ? S
o

E
ES E E
r
8 E
Gi
q R E E q ts
to ..i
q s
ut
E g
g-g $ F e $ D E fi S i

(U
-? ro \r - Iq to
h
ol a R
q o h
q, t P
q ; o oq
?d 6 6i c) t d iv
P ai I d g
(', c.i
' oi oi
'- ul -t Ar t
Eg
6-
tt,
S
T e. qt $
.Es
' (5s>a 6

E €E
6^ pF
;A q $
;
x sct d $
oi
E E
;
R
;
E
oi
s 3
ut
E E
o;
(0.c
EP<- I I S
', S
-. 9 r. Po
5- 4.9
-9 r( ,t(E
D

.o6' g6
*9 I o, ,., @ g E E q S g q E> io
: E R R t
'E;
t ; c,i E
I ci ai i ? ; ; I gF
oo.
FE
a
=o
9-;
;e.e
g q q q q { i e q q hkh e dhk q Fpb
go
u(D
I $r @ I cl I or : o,i6i I a? nq,d o-c
g :Tg
o oE
o(D
g€
(!) =
c

I qqqEEqEEE6 EEF- rEot EE


ot oi I ot = ci R ci b ct I I .qE

.E
tJ- E.3
6E
>L

b9
gn;$;$j$!g€ 9=g

g E
Eg
d'F
=8 t
&-a
Jr

SR32{, 17r!3/&l
Table 16 Contraction loss coefficients (Rouse, 1950)

(y/(y. + h) 0.oo 0 . 1 0 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

K" 0.50 0.48 0.45 0.41 0.36 0.29 o.21 0.13 0.07 0.01 0.00

Table 17 Main channel integrated discharges

Run Channel Flow Depth on MainChannel Bankfull


Type Flood Plain Discharge Discharge
(mm) (m%) (m%)
823 60" Trap. 50.5 0.0580 0.0865
smooth
B.24 100.9 0.0710 0.0865

828 60o Nat 15.2 o.0270 0.0385


smooth
829 50.0 0.0334 0.0385
830 99.7 o.0437 0.0385

835 60" Nat 15.3 o.o271 0.0385


rough
836 50.1 o.0223 0.0385
837 101.0 0.0243 0.0385

841 11OoNat 15.0 0.0190 o.0297


smooth
B.42 50.8 0.0204 o.0297

B,44 110"Nat 15.6 0.0208 o.0297


rough
845 50.3 0.0173 o.0297

sR 3e9 zalogle3
Table 18 Variables for defining main channel flow

Run cy, / B2IA s t2 fr(


823 0.671 0.385 9.142 1.37 o.02221 0.01682 1.320
B,24 0.821 0.769 9.142 1.37 0.01708 0.01664 1.026

828 0.701 0.185 14.60 1.37 0.03691 0.01936 1.907


829 0.868 0.607 14.60 1.37 0.02158 0.01900 1.136
830 1.135 1.211 14.60 1.37 0.01670 0.01883 0.887

B3s 0.704 0.186 14.60 1.37 0.05150 0.01956 2.633


836 0.579 0.609 14.60 1.37 0.10850 0.01945 5.578
837 0.631 1.227 14.60 1.37 0.18050 0.01938 9.314

841 0.640 0.18dt 14.60 2.O4 0.03776 0.02050 1.842


B,42 0.687 0.620 14.60 2.O4 0.02906 0.02066 1.407

B.44 0.700 0.190 14.60 2.O4 0.05200 0.01996 2.605


B4s 0.583 0.614 14.60 2.O4 0.10850 0.01993 5.444

Table 19 Adjusted variables for defining main channel flow

Test c/t / B% s (
B'23', 0.690 0.385 9.142 1.37 1.026
B,24', 0.802 0.769 9.142 1.37 1.320

B2g' 0.884 0.607 14.60 1.37 0.887


830' 1.113 1.211 14.60 1.37 1.136

836' 0.336 0.609 14.60 1.37 9.314


B,37', 0.963 1.227 14.60 1.37 5.578

SR 3aO lTrG/ql
Table 20 Roughnessand sinuosity adjustment to d,

Tesl d,o cYr cyr/dre


B.23 0.691 0.671 0.971
B,24 0.816 0.821 1.006

829 o.885 0.868 0.981


830 1.129 1.136 1.005

836 0.693 0.687 0.991


837 o.770 1.583 0.757

Table2l Roughnessand sinuosityadjustment to c

Test c^ c clco
823 0.558 0.537 0.962
B.24 0.558 0.562 1.007

829 0.630 0.613 0.973


830 0.630 1.636 1.010

836 0.630 o.237 0.376


837 0.630 -0.205 -0.325

B,42 o.428 o.427 0.999


845 o.428 o.240 0.561

sR32e 17rc0/9
Method
Run 1 4 5 6 7
No
B,23 0.70 -0.40 -15.33 0.85 -2.47 -0.06 -2.57

B'24 -o.71 -o.52 €.55 0.49 1.87 -0.03 2.08

829 1.00 0.90 3.47 -1.62 -7.91 -0.04 -0.09

830 -3.03 -2.73 5.51 -4.22 -5.36 1.13 o.20

836 14.30 0.62 5.78 2.59 -2.58 0.05 1.63


837 -2.42 19.90 4.81 12.74 o.32 6.63 -o.71

B'42 -0.41 13.16 11.61 20.06 24.56 23.65

B,t5 -8.85 11.94 -15.51 -12.9'1 1.17 1.29 2.72

Ave 5.75 0.68 -0.58 1.69 0.64 4.19 3.27


SD 6.13 8.25 9.92 7.51 7.97 7.98 7.88

Table23 Data sefs for inner flood plain analysis

Run ChannelType B2IA s


B,21 SERC60" tnapezoidal 9.142 1.374
826 SERC60" natural,smooth 14.600 1.374
BU SERC60" natural,rough 14.600 1.374
839 SERC60" natural,smooth 14.600 2.M1
Brl.ai SERC110onatural,rough 14.600 2.441
AB101 Aberdeen,trapezoidal 3.88t7 1.215
AB102 Abetdeen,trapezoidal 3.837 1.406
AB104 Aberdeen,trapezoidal 3.837 2.O41

sR 34, 17100/9it
Table24 Equation parameterstor y' greater than 0.2

Run a b
B.21 0.675 -0.2846
826 o.792 -0.2051
834 0.760 -0.2051
839 0.660 -0.0356
843 0.490 -0.2468
AB101 0.910 -0.3912
AB102 0.710 -o.3741
AB104 0.510 -o.4743

sa 3ag i7lGv93
Table25 Geometricdata overbanklaboratory studies

Test 0L*rc B so L w. wr sc
f) (m) (m) (m) x10-3 (m) (m) (m)

SERC FGF PhaseB


21 60 2.500 2.743 1.200 0.996 12.@0 1.374 6.107 10.000 0.00
26 60 2.500 2.743 1.200 0.996 12.000 1.374 6.107 10.000 0.00
31 60 2.500 2.743 1.200 0.996 12.000 1.374 6.107 6.107 1.00
3€! 60 2.500 2.743 1.200 0.996 12.000 1.374 6.107 10.000 0.@
34 60 2.500 2.743 1.200 0.996 12.000 1.374 6.107 10.000 0.00
39 110 0.000 2.743 1.200 1.421 10.310 2.O+3 8.560 10.ooo 0.00
4{r 110 0.000 2.743 1.200 1.O21 10.310 2.U3 8.560 10.000 0.00
47 110 0.000 2.743 1.200 1.O21 10.310 2.M3 8.560 8.560 1.00

Aberdeen
101 40 0.984 0.413 o.174 1.000 2.570 1.215 1.OOO 1.200 0.00
102 60 o.477 0.413 o.174 1.000 1.909 1.406 1.000 1.200 0.00
104 110 0.000 o.307 o.174 0.621 1.154 2.O43 1.000 1.200 0.00

Vicksburg
201 90 0.000 1.829 o.762 1.000 7.315 1.571 4.420 4.877 0.00
204 78.7 0.000 1.865 o.762 1.000 7.315 1.400 3.761 4.877 0.00
207 58.8 0.000 2.136 o.762 1.000 7.315 1.200 2.822 4.877 0.00

Kiely
301 45 o.475 0.4000 0.200 1.000 1.803 1.224 0.770 1.200 0.00

SR329 l7r!0r03
Table 26 Geometric data Soolcy'slaboratory study

Test NOP rc B so L w2 W, S"{


(m) (m) xl0-3 (m) (m) (m)

Geometry4
401 5 1.392 0.209 0.675 1.280 1.090 0.462 1.'t84 0.00
402 6 1.392 0.209 8.700 1.280 1.090 0.462 1.184 0.OO
403 6 1.392 0.209 1.600 1.280 1.090 0.462 1.184 0.00
444 6 1.392 0.209 3.670 1.280 1.090 0.462 1.184 0.OO

Geometry5
405 5 1.392 0.209 0.300 1.280 1.090 0.462 1.184 0.00
406 7 1.392 0.209 0.675 1.280 1.090 0.462 1.184 0.OO
407 7 1.392 0.209 0.870 1.280 1.090 0.462 1.184 0.00
408 5 1.392 0.209 1.OOO 1.280 1.090 0.462 1.184 0.OO
409 6 1.392 0.209 1.600 1.280 1.090 0.462 1.184 0.00
410 5 1.392 0.209 3.000 1.280 1.090 0.462 1.184 0.OO
411 5 1.392 0.209 3.670 1.280 1.090 0.462 1.184 0.00

Note : 1 NOP = numberof data points

sR 329 l7l03/t3
Table 27 Main channel geometric data

Test Type of xs h Ai Pr Qb s" 0,o


(m) (m1 (m') (Ys) f)

21 Trapezoidal 0.150 0.1575 1.324 86.50 1.OO 39.10


26 Natural 0.150 0.0988 1.288 38.50 1.OO 39.10
31 Natural 0.150 0.0988 1.288 38.50 1.OO 39.10
38 Natural 0.150 0.0988 1.288 38.50 1.00 39.10
34 Natural 0.150 0.0988 1.288 38.50 1.OO 39.10
39 Natural 0.150 0.0983 1.281 29.70 1.00 55.O0
4!f Natural 0.150 0.0983 1.281 29.70 1.00 55.00
47 Natural 0.150 0.0983 1.281 29.70 1.00 55.OO

101 Trapezoidal 0.050 0.0078 o.245 1.76 0.35 32.61


102 Trapezoidal 0.050 0.0078 o.245 1.72 0.35 40.66
104 Trapezoidal 0.050 0.0078 o.245 0.94 0.35 55.00

201 Trapezoidal 0.152 0.1045 0.950 34.60 2.OO 45.O0


204 Trapezoidal o.152 0.1045 0.950 39.08 2.OA 39.35
207 Trapezoidal 0.152 0.1045 0.950 4€1.90 2.OO 29.41

301 Rectangular 0.050 0.0100 0.300 2.32 0.oo 32.19

401 Rectangular 0.038 0.0080 0.286 1.30 0.00 11.73


402 Rec'tangular 0.038 0.0080 0.286 1.50 0.00 11.73
403 Rectangular 0.038 0.m80 0.286 2.18 0.00 11.73
404 Rectangular 0.038 't1.73
0.0080 0.286 2.90 0.00
405 Rectanguhr 0.076 0.0160 0.362 3.55 0.00 11.73
406 Rectangular 0.076 0.0160 0.362 3.55 o.oo 11.73
407 Rectangular 0.076 0.0160 0.362 4.20 0.00 11.73
408 Rectangular 0.076 0.0160 0.362 4.65 0.00 11.73
409 Rectangular 0.076 0.0160 0.362 5.98 0.00 11.73
410 Rectangular 0.076 0.0160 0.362 7.62 0.oo 11.73
411 Rectangular 0.076 0.0160 0.362 7.99 0.oo 11.73

sR329 lT lxIrS
Table 28 Mean 7o errors in discharge FCF data

TEST NOP BFO JW JW2 EE GH4 GH5

21 16 38.7 3.9 7.1 12.2 16.3 15.4

8.4 6.1 1.3 15.1 3.3 3.5

26 16 29.2 -2.7 -1.3 4.2 9.8 8.7

5.4 2.9 2.6 14.1 6.0 5.5

31 14 37.0 -7.5 -5.5 0.8 10.0 9.6

5.4 5.3 22.8 19.1 5.2 4.9

33 12 8.9 -6.4 -13.7 6.0 -3.2 -6.0

12.5 4.9 3.9 11.0 5.1 5.9

34 18 9.3 -6.9 -20.2 5.8 -5.3 -7.8

15.1 3.7 8.5 13.8 9.1 9.7

39 14 59.1 -3.8 -11.7 3.2 39.0 15.1

6.9 12.5 8.6 24.3 6.4 9.8

43 15 18.5 -2.4 -33.3 12.8 8.3 -8.2

24.6 8.1 17,1 23.1 21.6 23.7

47 14 63.0 -7.3 -16.7 3.0 40.2 14.9

6.0 13.3 9.2 27.5 5.7 9.9

smooth 74 44.8 -3.3 -5.2 4.9 22.5 12.7

14.7 9.6 10.1 20.2 14.6 7.5

rough 45 12.3 -5.3 -22.8 8.2 -o.2 -7.5

18.4 6.0 13.7 16.9 15.0 15.0

Atl 119 32.5 -4.0 - 11 . 8 6.1 13.9 5.1

22.6 8.4 14.4 19.0 18.4 14.6

Note: The lower valuesare standarddeviationsin the means


NOP - Numberof data points
Smooth - 21 26 31 39 47
Rough - 33 34 43

sqgze zl@la
Table29 Mean lo errors in discharge Aberdeen, Vicksburg and
Kiely data

TEST NOP BFO JW JW2 EE GH4

Aberdeen

101 38 28.2 -6.1 -3.4 -4.O 0.5 2.5

11.8 5.5 7.4 16.7 7.4 7.6


102 30 41.6 1.5 3.4 -2.O 14.2 11.9

7.9 4.5 4.7 14.4 6.5 5.2


104 20 77.O 10.4 3.6 0.1 47.8 30.8

22.4 8.3 7.1 15.7 14.6 11.7

Atl 83 44.8 0.6 0.8 -2.3 16.9 12.7

23.7 8.7 7.2 15.5 20.7 13.6

Vicksburg

201 59.3 -2.5 6.0 1.1 34.6 26.0


7.6 6.1 3.8 13.4 6.3 4.2
204 3 45.3 -5.3 5.3 6.8 21.1 19.2
9.9 11.2 10.0 2.5 9.6 8.4
207 3 32.9 -8.0 1.3 1.8 8.1 12.6
5.8 11.3 12.6 4.9 7.O 7.8

All 45.8 -5.3 4.2 3.2 21.3 19.3


13.3 8.9 8.6 7.7 13.3 8.4

Kiely

301 5 39.2 -0.8 6.3 -1.9 11.1 14.9

8.0 3.1 8.2 11.1 5.6 6.0

Note: The lower valuesare standad deviationsin the means

sR 3ee l7llxv&l
Table 30 Mean 7" errors in discharge Soolcy data

TEST BFO JW JW2 EE GH4 GH5

401 22.8 3.2 2.7 15.8 0.1 8.6


5.0 4.3 4.7 4.8 4.2 4.7
402 23.5 3.7 3.8 16.6 0.8 9.4
6.3 5.2 6.4 6.5 5.3 5.9
403 24.8 5.0 7.3 18.6 2,O 10.9
7.7 6.1 7.5 8.1 6.2 6.8
404 40.2 12.8 17.9 34.4 14.4 24.2
17.7 12.9 14.5 17.8 14.2 15.2
405 -13.9 -18.4 -14.9 -18.7 -29.8 -25.5

4.9 7.9 6.0 6.0 4.0 3.7


406 7.5 -15.0 -7.9 1.1 -12.1 -6.5

10.8 7.0 7.2 11.4 8.7 8.7


407 12.O -10.8 -4.3 5.7 -8.5 -2.9

11.4 8.5 6.3 12.1 9.1 8.9


408 't7.9 -2.4 -o.7 -4.1 o.4
12.4
9.2 11.0 5.1 10.2 6.9 5.7
409 20.7 -4.3 1.4 14.6 -1.3 4.4
13.1 10.9 5.0 13.4 10.4 10.2
410 37.0 6.2 6.3 30.8 11.6 16.7
8.1 9.6 3.7 8.8 6.0 4.6
411 39.7 3.0 1.7 33.4 13.9 19.4
1 1. 1 10.4 5.4 11.3 8.6 8.O

Atl 20.8 -1.9 1.2 14.6 -1.4 5.2


17.5 12.4 10.6 17.6 14.2 15.1

Note: The lowervaluesare standarddeviationsin the means


NOP - Numberof data points

se3€9 zrcllB
Table 31 Mean lo errors in discharge all data

JW JW2 EE GH4 GH5

74 M.8 -3.3 -5.2 4.9 22.5 12.7


14.7 9.6 10.1 20.2 14.6 7.5
4s 12.3 -5.3 -22.8 8.2 -o.2 -7.5
18.4 6.0 13.7 16.9 15.0 15.0
119 32.5 -4.O - 11 . 8 6.1 13.9 5.1
22.6 8.4 14.4 19.0 18.4 14.6
83 44.8 0.6 0.8 -2.3 16.9 12.7
23.7 8.7 7.2 15.5 20.7 13.6
I 45.8 -5.3 4.2 3.2 21.3 19.3
13.3 8.9 8.6 7.7 13.3 8.4
63 20.8 -1.9 1.2 14.6 -1.4 5.2
17.5 12.4 10.6 17.6 14.2 15.1
279 34.1 -2.1 -4.3 5.3 11.5 8.O
23.2 9.7 13.2 18.3 19.3 14.7
77 24.9 -2.6 1.8 11.9 2.0 7.4
18.8 11.7 10.2 17.2 15.6 14.8

The lowervaluesare standarddeviationsin the means


1 . SERCPHASEB SMOOTH21 26 31 39 47
2 . SERCPHASEB RODROUGHEND33 34 43
3 . ALLSERC
4 .ALLABERDEEN 101 1O21O4
5 - VTCKSBURG201 2M 207
6 -ALLSOOKY 401 - 411
7 -ALL DATA
I - VICKSBURG, KIELYAND SOOKYdata only

Table 32 Ranking of methods

TEST 1st 2nd 3rd 4th 5rh 6th


1 JW EE JW2 GH5 GH4 BFO
2 GH4 GH5 JW EE BFO JW2
3 JW GH5 EE JW2 GH4 BFO
4 JW JW2 EE GH5 GH4 BFO
5 EE JW2 JW GH5 GH4 BFO
6 JW2 GH4 JW GH5 EE BFO
7 JW JW2 EE GH5 GH4 BFO

sR3?B 07lo7t1g
Table 33 Mean 7o errors in stage FCF data

NOP JW JW2 EE GH4 GH5

21 16 -5.3 -0.6 -1.2 -1.0 -2.5 -2.4

1.6 0.9 o.4 2.3 1.2 1.1


26 16 -4.O 0.8 0.5 0.5 -1.7 -1.5

2.O 1.3 1.1 2.4 1.5 1.3


31 14 -4.9 2.4 1.6 2.5 -1.7 -1.6

2.2 3.4 2.7 5.9 1.5 1.4


33 12 -0.8 0.9 3.2 -0.4 1.3 2.1
1.0 0.9 2.2 1.0 1.6 2.2
34 18 -0.6 2.2 8.5 o.2 3.1 4.1
2,2 1.4 5.8 2.4 3.5 4.O
39 14 -7.2 2.3 3.8 1.9 -5.3 -2.1

3.5 3.1 3.5 4.O 2.6 o.7


43 15 -1.3 1.5 19.7 -o.2 1.2 7.7
2.9 2.1 15.5 3.2 4.4 9.3
47 14 -7.1 3.3 5.1 2.7 -5.1 -2.O

3.5 4.3 4.9 5.5 2.5 o.7

-5.6 1.6 1.9 1.2 -3.2 -1.9

2.9 3.1 3.6 4.3 2.4 1.1

rough -0.9 1.6 10.8 -0.1 2.O 4.8


2.2 1.6 11.6 2.4 3.5 6.3

Atl 119 -3.8 1.6 5.2 o.7 -1.2 0.6


3.5 2.6 8.8 3.8 3.8 5.1

Note: The lowervaluesare standarddeviationsin the means


NOP - Numberof data points
Smooth - 21 26 31 39 47
Rough - 33 34 4{t

SRS?9 AIW)
Table 34 Mean lo errors in stage Aberdeen, Vicksburg and Kiely
data

TEST NOP BFO JW JW2 EE GH4 GH5

Aberdeen

101 -4.8 1.4 0.9 2.6 0.0 -0.3

2.2 1.2 1.5 4.7 1.3 1.3


102 -5.7 -0.4 -0.7 2.2 -2.5 -2.1

3.0 0.8 0.7 4.2 2.1 1.7


104 -8.7 -1.8 -0.7 1.7 -6.5 -4.7

5.3 1.6 1.1 3.5 4.1 2.7

All B3 -6.1 0.0 0.0 2.2 -2.5 -2.O

3.7 1.7 't.4 4.2 3.5 2.5

Vicksburg

201 -8.5 0.1 -1.5 4.9 -5.9 -4.8

3.9 1.2 1.4 10.o 3.0 2.3


204 -7.3 0.5 -1.6 -1.5 -4.2 -4.0

3.8 1.8 2.3 o.2 3.0 2.7


207 -5.9 1.0 -1.0 4.O -1.9 -2.8

2.9 1.7 2.4 7.5 2.O 2.4

Atl -7.2 0.5 -1.3 2.4 -4.O -3.9

3.3 1.4 1.8 6.9 2.9 2.3

Kiely

301 5 -6.6 2.3 -1.6 3.3 -2.3 -3.0

1.8 5.0 1.7 7.3 1.1 1.3

The lower valuesare standarddeviationsin the means


NOP - Numberof data points

sR 320 22103193
Tahle 35 Mean Voerrors in stage Sooky data

TEST NOP BFO JW JW2 EE GH4 GH5

401 5 €.1 -o.2 0.0 -4.5 0.8 -2.6

1.4 2.5 2.9 1.3 2.7 1.2


402 6 -5.9 -0.9 0.1 -4.4 0.9 -2.5

1.1 1.4 3.5 1.2 3.3 1.4


403 6 -5.7 - 1. 1 -1.7 -4.5 0.5 -2.6

0.8 1.3 1.5 1.1 2.8 1.1


404 6 -7.8 -2.5 -3.7 -7.O -3.0 -4.9

1.3 1.9 1.9 1.7 2.2 1.7


405 5.0 8.4 5.6 8.1 11.8 9.9
5.0 6.7 5.8 6.7 7.4 6.6
406 -1.2 4,7 2.5 0.6 4.3 2.5
2.8 3.5 3.1 3.8 4.1 3.5
447 -2.1 3.4 1.5 -0.5 3.1 1.4
2.6 3.3 2.5 3.4 3.6 3.1
408 -3.6 1.2 0.9 -1.5 2.2 1.1
1.6 5.3 3.3 4.2 4.3 3.7
409 -4.0 1.2 -0.1 -2.8 0.8 -0.6

2.O 3.1 1.3 2.6 2.9 2.4


410 -7.O -1.6 -1.6 €.1 -2.5 -3.3

0.8 3.3 1.2 1.0 1.1 0.6


411 -7.3 -0.5 -o.2 €.5 -2.9 -3.7

1.1 3.1 1.6 1.3 1.5 1.1

Atl 63 -4.1 1.2 0.3 -2.6 1.5 -0.5

4.0 4.4 3.5 4.9 5.1 4.7

Note: The lowervaluesare standarddeviationsin the means


NOP - Numberof data points

sR 329 l7lo3/93
Table 36 Mean /" errors in stage all data

NOP BFO JW JW2 EE GH4 GH5

74 -5.6 1.6 1.9 1.2 -3.2 -1.9

2.9 3.1 3.6 4.3 2.4 1.1


2 45 -0.9 1.6 10.8 -0.1 2.O 4.8
2.2 1.6 11.6 2.4 3.5 6.3
3 119 -3.8 1.6 5.2 0.7 -1.2 0.6
3.5 2.6 8.8 3.8 3.8 5.1
4 83 -6.1 o.o 0.0 2.2 -2.5 -2.O

3.7 1.7 1.4 4.2 3.5 2.5


5 I -7.2 0.5 -1.3 2.4 -4.O -3.9

3.3 1.4 1.8 6.9 2.9 2.3


6 63 -4.1 1.2 0.3 -2.6 1.5 -0.5

4.O 4.4 3.5 4.9 5.1 4.7


279 -4.7 1.0 2.2 0.5 - 1. 1 -0.6

3.8 3.0 6.6 4.7 4.3 4.4


8 77 -4.6 1.8 0.0 -1.4 0.6 -1.0

4.O 4.7 3.3 5.9 5.1 4.5

The lowervalues are standarddeviationsin the means


1 . SERC PHASEB SMOOTH 21 26 31 39 47
2 . SERC PHASEB ROD ROUGHEND33 34 43
3 .ALLSERC
4 . ALL ABERDEEN101 fi21}4
5 - VfCKSBURG201 204 207
6 -ALLSOOKY 401 - 411
T.ALLDATA
I - VICKSBURG,KIELYAND SOOKYdataonly

SR 329 O7rO7lB
Table 37 Sensitivity fests : effect of errors in wave length

Factor % Errorin L mean7" Errorin


discharge

0.50 -50.0 -10.3

0.75 -25.0 -5.5

1.00 0.0 -2.1

1.25 25.O o.4


1.50 50.0 2.3

Table 38 Sensitivity fests : effect of errors in channel side slope

Factor 7o Errorin S" mean7o Errorin


discharge

0.00 -100.0 -5.3

0.50 -50.0 -3.9

1.00 0.0 -2.1

1.50 50.0 -0.1

2.OA 100.0 2.4

Table39 Measuredzonal discharges,Sooky and Kiely data

Discharge(l/s)
Testsource Depth Total zone 1 zone2 zone3 zone4
(mm)

403 Sooky 61.3 7.89 2.09 2.58 1.54 1.68


409 Soolqy 99.4 12.62 4.92 3.33 2.11 2.36
301 Kiely 60.0 3.10 1.49 1.11 0.40 0.40
301 Kiely 80.0 11.10 s.10 5.48 1.26 1.26

sR 329 l7l03/91
Table 40 Errors (%) in calculated total flows, Soolry and Kiely data

Method
Test Depth BFO JW JW2 EE GH4 GHs

403 61.3 33.9 12.6 16.0 27.8 9.5 19.1


409 99.4 12.4 -10.3 2.2 5.8 -7.8 -1.8
301 60.0 52.9 3.8 6.8 14.8 20.6 24.6
301 80.0 36.0 -0.9 12.6 9.0 10.5 15.3

sR 3e9 t7l0993
Table 4'l Measured and calculated flow distributions

"/"(Zonalflow / Total flow)


Zone Measured JW JW2 EE GH4 GHs

Test Sookydepth= 61.3mm


1 26.5 2 1. 3 19.1 1 9 . 1 2 1. 9 19.6 18.0
2 32.7 38.2 32.9 32.9 35.7 31.0 38.8
3 19.5 20.2 24.O 24.0 21.2 24.7 21.6
4 21.3 20.2 24.O 24.A 21.1 24.O 21.6

Test SooW depth= 99.4mm


1 38.2 41.1 33.5 29.4 42.2 40.4 37.8
2 26.4 28.5 28.7 37.6 25.8 23.O 29.6
3 16.7 15.0 18.9 16.5 16.0 18.3 16.3
4 16.7 15.0 18.9 16.5 16.0 18.3 16.3

Test Kiely depth= 60.0mm


1 48.1 61.2 47.7 46.7 59.5 59.4 57.4
2 35.9 27.2 35.1 36.9 25.1 26.2 29.0
3 8.0 5.8 8.6 8.2 7.7 7.2 6.8
4 8.0 5.8 8.6 8.2 7.7 7.2 6.8

Test Kiely depth = 80.0mm


1 27.8 19.2 14.8 13.0 21.O 18.0 17.3
2 49.4 57.2 53.0 58.6 4i!.8 53.0 58.3
3 11.4 11.8 16.1 14.2 1 7 . 6 14.5 12.2
4 11.4 11.8 16.1 14.2 17.6 14.5 12.2

SR 329 t7r8/g
Table 42 Reach averaged geometric parameters Roding study

Yz(m) Area Width Wettedperimeter


(m') (m) (m)
Main channel at bankfull
0.0 5.3 7.1 7.7

Zone 2
0.1 2.1 20.5 10.8
o.2 4.4 25.3 15.7
0.3 7.1 27.9 18.4
o.4 10.0 29.0 19.5
0.5 12.9 29.9 20.4
0.6 15.9 30.4 20.9
o.7 19.0 30.9 21.4
0.8 22j 31.3 21.9
0.9 25.3 31.8 22.5
1.0 28.5 32.4 23.1

Table43 Errors in predicting overbankdischarges

Case P2 M2
Method Mean Standard Mean Standard
Error ("/d Deviation(%) Enor (%) Deviation(%)
Bed FrictionOnly 9.5 9.0 7.3 8.6
Jamesard Wark -2.O 1.7 -2.2 3.2
James and Wark 2 -27.1 10.0 -30.5 10.9

Note: o/oEnor = 100*(Q".r" - Q,r""")/Q,."""

sR329 22r0gl93
Table 44 Sensitivity fesfs on the effect of floodplain roughness

Zone 2 Mean ToErors Difference in Means


Manningn BFO JW (BFO - JW)

0.01 305 142 163


o.o2 122 78 44
0.03 61 40 21
0.04 31 16 15
0.05 10 -2 12
0.06 1 -10 11
0.08 -14 -25 11
0.10 -24 -33 I
0.18 -40 -47 7
0.30 -48 -58 10

Note: T"Error= 100'(Q*r"- Q*J/Q*."

SR 329 l7r03rg3
Figures
iql

-3 _ oN

-3
:
"s t -
sr 0.)b ": - -3
6
o EE
eF
'5
E (6l , !: gE
8CI
+ 32E _o
T

U ) cA -
e.-J
v
= R€
E T A
o
o
'ao
b!c
3 U,x, -cO
t E

aa) |
R 6g
bo : P g3
_ oN

i=g E o
-() i=9
o6 -o

t a
o
-o
o
E. Itrlll
ttlttttttttll
cr(,oroou)o|oorooroo o qqqqqq
dddr{ciaicic,i;;dci C' e(\.0(\l*@
l l l

_o - oo
@

-ro
l()

o
c -6 o
6c -oa €
b> c c
a
- E

(o6 o

6E
o
c - t t (t E
a a

CF : -R
E
68 I
o
_ o!
.96 a
a
I

5ce + -at ()

EE
@c
T

EE =, _lo9
t - 6
( ,9E Fo t
a
E
C E C
I (UO a
-R
E$
o o
I

o,
I
_a -ct

v
a
.9a I
o
o

EP
|q
o
C (r9 o
c
c

o
c
a

Eo
a
c G
o o o
I
a II
I
a
o
(, a
o
-o
o o
E E tl tll
| | t t-o
ro o q'Orl,O olo qqq
ci c,i oct (\| ?oo

Figure1 Riverchannelscross-sections
- oo -8
N

E -8
Ee
i
-E
E Eg *
b3 fI F E { E i - e.
sg - :, n f F
;6 SgEE
o
-(D 6 q

io E -o
H-9
(g6 - 8-e
8 tCI
E -89
EE s 2E NO
E

e,; + -R -E
LA
(l)a E$
9o
-8
E(j 'o a
E - oo
:
i -K E5 -3

-o
*t
I -o
o o
oi Foo@ts@6soN

E c
Q>
€E
ctr E,E
oo L

=E
cuc o
t
EE
LO -8
s 3 (O-
co :
-
!L
e oo
otj E
D
PE -3
-(l) .=o
di u) xu)
f

gs
.3n

(r()
o
o
o
E
, aa
b9
-e
o
c
.>()
E.
I
a
I
-R

tttttltt
qooooooo
dd+aic\i-ci;

Figure2 Riverchannelscross-sections
Myers Laboratory Flume SeriesA
cross-section geometrY
o.15

0.10

0.00
I I I ll I I
0.00 o.25 0.50 o.75 1.00 1.25 1.50 1.75
Metres

Myers Laboratory Flume SeriesF


cross-section geometry
0.15-

0.10 -
o
. L
o
o
=

0.05 -

I I I I I I
0.00 o.25 0.50 0.75 1.00 1.25
MEtres

Figure3 Myerslaboratorychannelcross-sections
I =
:a
I ()
a
= N N

9o I o
o
lL(l)
=
p:
ocg
6>
I o
a
OF
@

N 6=
=.8

12 g) I d' raO

8g() N
I
=N
ol\
o
.=< N o
=
o
F(I'
I 3tg
u
f)
tr(0
a-.
N gs
fio I t)€
_=
beg N
E FI
Fi=
o
N =
1 8
N =
o=
I J O

9
q ,eN o
NoOo o o o
a
U
z

N €tL Y
a I ? -to I o
3Oa
iTo
N N

6
:
6(6
rr< s N

o
=
X(U
€> T 6
E E
rtU)
6.q I
N (t
EE o
S-
J
o s/) I ni iiOa,
gs I
o=

8A
(o0 NS\:\ = N 6H = NDI
o€ I 6iN O= - Ei
.sa
a T g .c[
N
c)

8 F o 8 g UJ
b.s
t(o N\\:S g
,nl-L
s 99 Nl\\ Sae
uro I 8 E .FE - 8F=
sb FN\.\
I FI "_"€&
N\\
I 8l
FiE N
CJ o
o E .Eg
-9$ T
o
o

N N\\S
6 E =z
=3
9
aaooooo oono6oro o
NO ; .TT a
U U
z

Figure4 Meanerrors,straightchanneldata
F__3.054m

H 1.2)m
H
0.S78m
'-r
I
II
't'..'
t\\
II
i. I

II
h..
--{.1-l\ E
q
@

1 | l-t---
l-.--;i
I
l/
1/
y
\ \

-r \
I

i
I

.4
-l I
I r
.t1'
I
1 I,
v'

t/
iiil
t--,1
| 0-90m I

Figure 5 Detailedplan geometry of FCF 60" meander


W
10.0m

Fivetailgates
Endof mouldedsection

, 89.0 , 150.0
l-l
o
ot
@

I /

II
q
(\l
(\I
o /

) )
F--+t
t o
d
F-
19

o
d
{
't'
s

t
+
$
89.0 c.
45.0 .9
C)
o
L

E
3
€$
o
LL

tts
OE

fiE
s$
E$
$$E €s
Ei

Startof mouldedsection

Overflowweir

Figure6 Planof flumeand naturalcross-sectiongeometryfor


600meander
@l

no--[),

\-/-.------

tt
,..1*
,rr"w E
a?
o

+7.36m

vvO
l-r*'-[
l.-r.z'l

Figure 7 Detailedplan of geometryfor FCF 1'l0omeander


tql
10.0m

Fivetailgates
Endof mouldedsection

89.0
k---'l
o
ot
@

I
I
q
ol
(\l
o

I
l.e-rl
H
' 45.0
89.0'
aX

E$
gtr OE

fiE
r$
..EE
$$E ET

TE
Iro
lo
lc
r(\l
Startof mouldedsection

Overflowweir

Figure I Plan of flume and naturalcross-sectiongeometryfor


110o meander
Lql
Adjustabtef-l
tailsate
U

-r- T
l- I
39mm 174mm I
-.'-'}|

I
1200mm

i
I

1145mm

l.- 174mm

139mm---J

Figure9 Plangeometriesof theAberdeenflumewith channelsinuosities


of 1.40and 2.06(afterWilletts1992)
Lq

T
II
000mm
K---- I
I
II
I

Enlarged view of one bend


1

PIan view of whole section

F__1semmff
I
*.
\.
Cross-section through channel
Jglt,nU1242tgD

Figure10 Plan geometryof the Aberdeenflume with channelsinuosity


ot 1.21(afterWilletts1992)
Sinuosity: 1.00

8
4
o
6o 4
_c8
o
.s Sinuosity= 1.57
o
c
@
o
E8
o
o
(, 4 Ig
c
g0
o
64 '/ \v
b

I
I I

Sinuosity= 1.40

Length in feet
ll
100

Sinuosity= 1.20
16.0'

2.O'.
End view
1012345
re
Scalein feet

Figure11 Planand cross-sections


for Vicksburgflume
(afterUSArmy 1956)
Plan view

VN ^s.\=/'/1;

Water
Gravel supply
baffle

Cross-sections

Geometry3: Meanderingnanowchannel

b = 0.687ft

Geometry4: Compositechannel

B = 3.886ft
b = 0.687ft
Yoz= 1.5in

Geometry5: Gompositechannel

B = 3.886ft
b = 0.687ft
Yoz= 3'0 in

Figure12 Planand cross-sections


for Sooky'sflume (afterSooky1964)
Single meander compound channel

Testsection

I
B
2254

Multiple meandercompoundchannel

Figure13 Plan and cross-sectionsfor Kiely'sflume (afterKiely 1990)


3

6
c
c
(g
E
c)
.E
(U
s.sI
\ BE€
36'6
o()tr
z

EIE
r lE
.Elo
\
3
o
{
,x' Y
F
3
o
o i.t_
z

3
OE
'6
.Sb E
s-dE
PE€
6E
FI
El>^
tr9
o EIE
=to
.cl>
o, I
b co
(It =l (-
E
c
8 tr(U
E.C
U'E
Y -;p<>
o o9
a, Ig
3 o( ,(Eu
(l,
z 'l

EIE
.Elo I
lt

sl€ I
+
l" \
Y I
I
I

(l)
z

Jsi,Jr1311242rJo

Figure14 Flowprocessesin a meanderingcompoundchannel


(afterErvineand Jasem,1991)
1.5

1.4

1.3

1.2

1.1

t.0
1.6 1.8 2.0
Sinuosity

A Aberdeen + SERCTrapezoidal )F SERCNatural

o McksburgNarrow SCS Method

Figure15 Adjustmentto Manning'sn for bendlosses: measured


and predicted

1.5

1.4

1.3

1.2

1.1

1.0
1.8 2.0 2.6
Sinuosity

A Aberdeen + SERCTrapezoidal fl McksburgNarrow

SGSMethod LSCSMethod
JB/Y/I1r6/|2-e/3O

Figure16 Predictedadjustmentsto n for bendlosses: Modified


Changmethod
@l

Plan

-L

Q=Qr +Qr+Qa+Qo

subdivisionfor overbankflows,Evineand Ellis


Figure17 Cross-section
(1e87)
G'E'I
388383
F N d F N O
zzzzzz
888888
oooo@F
cNoooo
@@6doo

$rf{ryq

o
o
o
C
.c
o,
5
o
o
E
o
o
rf
F\
a?
Fc
\qt
'9^ eE
6o
Cc
6i'
c?
d)

Gfr
RRFR
,-FZZ
oo60
9RPR
zz2z
8888 oooc)
ortoo
NNNO
dtc)(D.o F N * O
ast=
to dI co tD
+: t x: l x o
ltx: rd: x

(r,
o (o o
CL o o
C' c
c .-c
tt, o)
f
.t)
x co 6
E
e s
(o6
-E o ct5
o o E
E
E 6
o g
o =
o *\
o :i c) dx
s
t\
6l
$
al o
FG oi
}E
'-^ E
.aE
t4E
6o 9o
=: .Sz
66 @i5
ON o ot\
ct ct o

Js'{|1U12.o2/siD

Figure18 Variationof mainchanneldischargealonga meanderduring


overbankflow (FCFPhaseB)
0.16

0.14

0.12

.E 0.10
.c
o-
(l)
E
3 0.08
o
tr

0.06

0.04
x Measured
Predicted,ModifiedChangmethod
o.02

0
0.07 0.08

relationshipfor 600trapezoidalchannel,
Figure19 Stage-discharge
inbankflows
0.15

0.14

0.13

a12
E
E
CL 0.11
o
!,
3
o
IL

0.10

0.09
Measured

0.08

0.07

channel,
relationshipfor 600pseudo-natural
Figure20 Stage-discharge
inbankflows
W

0.15

0.14

0.13

o.12
E

o- 0.11
o
E
3
o
tr
0.10

Measured
o.09

0.08

o.07

relationshipfor 'l'l0opseudo-naturalchanne
Figure21 Stage-discharge
inbankflows
@l

c;

o o
o

@
o
ct
E
.cEt
ff,) 5

F*ae-
qi6
oE5
€9
z;
I bP
6(or
(E@

sl1
qs
oE -
PEE
It E-yO
(,oo

It
It 3 rie o E-
ci E .-:

tl
lt
8Eg
tl XFzo
E'E 2
6 b
( ob( o9
I x+El
I ol
q
o

o
o

Figure22 Variationof dimensionlessmain channeldischargewith flow


depth on flood plain
x 60" Trapezoidal,smooth e 60" Natural,rough
+ 60oNatural,smooth @ 110"Natural,rough
E 110"Natural,smooth

JSNMn24?lw

Figure23 Variationof dimensionlessmain channeldischargewith


dimensionlessflood plain flow depth
E

\
o

(o
{

eq
EE
EF
t()
o

b3
(oF
ts
e@ o a
o
tt
o
E
o
at

=2e €
P E
(U a?
o
E E
b-
E
(U
E E U'

8Es
F,E 9
.E
o
c!
o
FZ; c
b b3
@(o
o
(L

ci
x+El

JgtNHtl2-92J9

Figure24 Variationof dimensionless mainchanneldischarge- yi!.h


dimensionless flow depthwith pointsadiustedfor friction
factor ratio
'-g
de_
dE
qE
F(g

2=
bP
(OF

(E@
.E
o
PeE
:E P
-v'-
S H
8EE
E"
bb9
E?
(o(o

x+tr
p //e

6,
t/A
//

\qq
oo
--lo-
o ld

Figure25 Additionaladjustment
to dischargefor relativeroughness
W]

eH
o)
3q
Eg
E?
b9
(OF

(E@
E

g€E
o

EEE
8EE
gE?
bbP
(O(OF

x+tr

.l,f

Jg|tilz$nz-e2JcD

Figure26 Adjustmentto c for relativeroughness


q s9 q
.E bP . o 9P P
F
c': 3 *e :
E
- 'hil
F F' tft

i+t+f

(o
F-
E -
.9
(E c
'6
f
u o-
IU E
.q8
OF
o
(f)
(0 s
o_
c o
!
.9 o
(g
f (!
ct E
uJ E

(t,p) 1 euozrol Joloelgueulsnfpy

Figure26b Adjustmentfactorfor Zone1 discharge


Plan

Figure27 Flowexpansionovera downwardstep

Jg!ilr27.28n2-92nD

Figure28 Flowcontractionoveran upwardstep


a- Wide channel b. Narrow channel

Jgi|,t2til1242f3o

Figure29 Expansionandcontractionflow patterns


t+
ol

$t
ol
()J
E
EE
5o ct
tJ) OI
FE
{n(D
co
q
qa
o @
+
-Gl x(E
b8
NO
-i (0
tl
tl
!
'
o \f

cF {c€o

o
F

(E xx

P'hc

Figure30 Widthto depthratiocorrectionfor expansionlosses


W

(l,
of
EE
5o
o,
(rl
E
Eb
@
o qa
\
o
1o x(E
5(f,
ai ol
@o
$n
d-
ll

o r€
Fd)

P,tlc

JB,flt31t12-92rfi

Figure 31 Width to depth ratio correctionfor contractionlosses


W]

!(r
OI

(v
(\l

o=
E
EE
9ru o
E(Il
o) (\l

Eb
ct)
q
ge
@
o
+ x(E
+o
dt ol
q|ct (0
-I
il

o
E
\f
F

ot€
Fqf

<)

prc

JgttlRwl2-92JW

Figure32 Widthto depthratiocorrectionfor combinedexpansionand


contractionlosses
tr
EI

-c
o
PEq-
c=o
3
gHE P
'0i6
ss5
F* g
FZ; E
b bP
(g(oF
E
x+EI
E

EI

E
EI
x.x
s
E cix
EI + x
.(v
x
xx+ d+

t(, o lf,
F

(7.) rcr3

Figure33 Errorsfor SERGpredictionsbeforesinuositycorrections


N

o
o x
o
-x
o

ol
o o

\
oc,
+
to(o(f,
FO.<f
otto +
;;oi
[ltl
o<to

xO+

*+ .aJ*

Figure34 Errorsfor Aberdeenpredictionsbeforesinuositycorrections


x 60" Trapezoidal,smooth o 60" Natural,rough
+ 60" Natural,smooth @ 110"Natural,rough
EI 110"Natural,smooth

B.29

F36e -o-
(}
B,24

841
Q'2

B,44 ""-F -"'-1


3--E --r@
__
@ __e _ _ @ _ _ @
___@

0.1 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.O

JBW35/r

Figure35 Adjustmentfactor for innerflood plain dischargesfor SERC


PhaseB experiments
*
A
A s = 1.215
;{e. s = 1.406
A,*
* ,tA,
A O s = 2.043

**^* A
A
lqa *
,r.
** x. ^^ a
a
A,
* A
xo A,
o ;F,
AA
*o :ft.
A

o
@q #
l(
le o )F
)F

o ;F1
o )r
*.p
@o o
o

o
o
o@
o o

Figure36 Adjustmentfactor for innerflood plain dischargestor


Aberdeenexperiments
smooth
x 60" Trapezoidal, e 60" Natural,rough
+ 60" Natural,smooth o 110"Natural,rough
E! 110oNatural,smooth

1.0
0.9
0.8
0.7
839
0.6 834
826
0.5
B,21
o.4

0.3

Figure37 Adjustmentfactor for innerflood plain dischargesfor y' , O.2


SERCPhaseB data
M

A, s = 1 . 2 1 5
:*. S = 1.406
o s = 2.049

;F
1.0
0.9
0.8 o
o.7
0.6

Q'2
0.5

0.4

0.3

JBilvr3an2-92F/O

Figure38 Adjustmentfactor for innerflood plain dischargesfor Y' > O.2,


Aberdeendata
1.0
0.9
0.8
o.7
0.6
0.5

o.4

0.3

a = 1.02s{'915
0.2

0.1

Figure39 Variationof a with s


tql

1.0
0.9
0.8
o.7 b = -9.9119214)4'4n
0.6
0.5

b 0.4

0.3

5 67
B2IA

Figure 40 Variationof b with B2lA


x 60" Trapezoidal,smooth e 60" Natural,rough
+ 60oNatural,smooth o 110"Natural,rough
El 110'Natural,
smooth
Equations114to 117

823-

P36e
60" Natural

841
Q'2 EI

60" Trapezoidal
B,44

0.1 0.2 1.8 2.O

J$Nt1l11242tfi

Figure41 Comparisonof predictedZone2 adjustmentfactor with


SERCdara
@l
^a
A s = 1.215

)F
;r s = 1.406
t\ o s = 2.043
A
* Equations114lo 117
*

s = 1.215

s = 1.406

o
o @

0.3

o.2

o.2 0.4 0.6 1.O 1.2 1.4 1.6 1.8

Figure 42 Comparisonof predictedZone 2 adjustmentfactor with


Aberdeendata
W
<-- Flow direction

,.0----/o'r--'o2

{o.s/ a-,_os
u\
o,s -
eo
o ol

w,
E
z
L
=l
o
c
o
o
v,
o

ffi
o
L

o
L
(U
I (D
s' E
o
,/ E
-"t' o
co
o3-

t\r,
rO
o

s.'

/\0
o.r_.--....-.-o

qu?qqqqqqeqq
tr) .+ .+ (f) cr) OI C{ F O_. O

Figure43 Exampleof boundaryshearstressdistributionin a


compouhdchannel(afterLorena,1991)
meantlering
Plan

Zone3
l-L---------------+

Zone4

Section
*Wrg
I Stage I
tvl
Zone2 : I zone4
Bankfullstage I

SE=(Ss+S"r)12
-
Pt =cd
_> -___>
Pe=bc+de-B(s-l) s = sinuosity
+
Pg=ab
Pl = ef-

Js'Nt4t1242r3o

Figure44 Cross-sectionsubdivisionof overbankflows,Jamesand Wark


0.s5
0.50 o o &.
o ,t". q a ,p@**s
0.45 @

*.i" "So
0.40 o+:.:gv "oo tr.t *-
EB

o ,E+
+trq,e
0.35 + o )*fS f e* x +
)c. o.so
I
5 0.2s
XA
eEl
t1s' ;i1-t ;' F+
+jE.
+g+
i(

0.20
X +
"Ti#gu.
..-.%;,f::&,F #
+fi
X GIX g"++
0 .r 5 X +
&t!
0 .r 0 +
# +* +
0.05 +

0.00 voo
-50 -40 -30 -20 -t0 0 t0 40 50 60 70 80
1(0catc - 0meas)/Omeas
A2l v26 o3l o55 o34 +39 +45 s47 ol0l ol02
+ f04 x20l a2O4 v2O7 tr501 o 401 o 402 + 403 i( 404 x 405
a 406 e 407 8408 +409 x 4t0 *41I
Bed Fr i ct ton 0n Ly : ( H - h )/ H v s Z ( O c aL c 0 m e a s )/ O m e a s
b
0.55
0.50
+
$o tg
0.45 ,o w\
ceB
0.40 tr e.,
o.35 l'#*^1
= - - )'.'[E+
E
0.30
I
r 0.25

0.20
0.r5
o.t0
0.05
0.00
-t4 -12 -10 -8 -6 -4 -2 l0 12 14
I (Hca[c - Hmeas)/Hneas
a21 e26 o5l o55 o54 +59 +45 847 o l0l o 102
+ 104 x20l a204 e2O7 tr301 o 401 s4O2 + 405 >..404 x 405
e 406 e 407 E 408 +409 x4l0 *4ll
Bed Fr i ct ton 0n Ly ( H- h)/H vs Z ( HcaLc H m e a s )/ H m e a s
JBffll15t12'0?IJID

Figure45 Errorsin predicteddischargeand depth: BFO


0.5s

d.#
0.50
0.45
0.40

$"ffi
0.35
=
)E o.so >:x
I
5 0.25
0.20
0 .r 5 +
+*
0 .r 0
0.05
0.00
-50 -40 -30 -20 -10 0 40 70
I (OcaLc - Qnreas)/Omeas
A2l v26 tr31 o33 o34 +39 +45 a47 ol0l e 102
+ 104 x20l a2O4 v2O7 tr501 ' o 4 0 1 o402 +403 x404 x 405
8406 e4O7 tr408 +409 x4l0 *4ll
James and flank Method r (H-h)/H v s Z ( O c aL c O m e a s/)O m e a s

0.55
0.50

$*o;
0.45
0.40
0.35
)-c. o.so

#F
I
= 0.25
0.20
0.r5
0 .t 0
0.05
0.00
-14 -r0 -8 -6 -4 -2 t0
l(HcaLc - Hneas)/Hmeas
621 v26 tr3t o33 o54 +39 +45 a47 o10l o!!?
+ 104 x20l a 204 v2O7 E 301 o 401 s402 +405 >a404 xqu)
e406 e407 tr408 +409 x4t0 *4ll
James and Uank Method r (H-h)/H vs Z ( HcaLc Hm eas)/Hm eas
JAilt{6t12{€,tlD

Figure46 Errorsin predicteddischargeand depth: JW


a
0.55 +
+ o€
0.50 + q l .-'e" %+

0.40
+
o.45 +
+ F. i "; f,'Y#*" *;-d
"
+
:s+-o$ "%"qli^*
0.35 E^* s + ''t,.,*
=
)-c, o.so
I
5 0.25
0.20
f

+
++. 's
OX m- *
" d x**.d,o,
* ,&"#t
3*:
g l r'o,
a_+\ S

0.r5 %fltr* +
0.t0
0.05
*'ff. ," rff€.l
voO
0.00
-50 -40 -30 -20 -10 0 t0 zv 60 80
I (0ca [c - Q m e a s )/ 0 m e a s
A21 v26 cl51 o33 o34 +39 +45 w 47 o l0l e 102
+ 104 X20l a2Q4 v2O7 E50l o 401 o402 + 403 i.404 x 405
al406 e407 s408 +409 x410 *4ll
James and Uank llethod 2 ! ( H - h )/ H v s Z ( O c aL c - 0 m e a s )/ 0 m e a s
b
0.55
+ oo
0.50 +d" to
+9q o"9
0.45
o.40
Jd %l'
i'i ooo.-*
c{x "-hn' 9 o + E E
t.,''tE :."8 .i.
o.35
=
)s. o.so
I
5 0.25 o"*
L1T**.*
0.20
0.r5 ffiu'
ffi*'
H""'
0.r0
0.05
v (}9
0.00
-14 -12 -t0 -8 -6 -4 -2 4 l0 l2 t4
l(Hcatc - Hmeas)/Hmeas
A21 v26 o3l o33 o54 +59 +45 s 47 ol0l o 102
+ 104 X20l a204 e207 tr 501 o 401 o 402 +403 x404 x 405
8406 e407 8408 +409 x4l0 *41|
James and Uank Method 2 ! (H-h)/H vs Z ( HcaLc - Hmeas)/Hmeas

Figure 47 Errors in predicteddischargeand depth: JW2


0.55 +
^(R 4 {
oovl
0.50
*ffi
0.45
f;fo + ..og
0.40
%Sq;
0.35
)c. o.so
"" E
a '+tr
c
eo.p.
I
X ?&I a+
5 0.25
$oso * *
0.20
''%&"r *#n.
0.15
0 .r 0
-ebsf?
0.05
E :qA
0.00
-40 -30 -20 -r0 0
I (0ca [c - 0meas)/0meas
L21 v26 tr31 o33 o34 +59 +43 a 47 el0l el02
+ 104 x20l L204 v207 tr 301 o40t o4OZ + 403 >:404 x405
4406 e407 s408 +409 x4l0 *41|
Env i ne and ELLts llethod : ( H- h)/H vs Z ( 0 c aL c Q m e a s/)0 m e a s
b

0.55
oo
0.50 0
o
q
0.45 o
o
+
o
+
a . "-tg*nt
0.40 xGi o

=
0.35
i(
i(
+
"E**.e
)E o.so
&*i".
ia
I
lt 0.2s
JF(
0.20 X
X p*+
t 4e
0 .t 5 _s
IF '*S+,$
4
0.10 6X
x
0.05
0.00
-r4 -12 -t0 -8 -6 -4 -2
I (Hca [c - Hmeas) /l-lmeas
A21 v26 tr51 o55 o54 +39 +43 a 47 ol0l e 102
+ 104 X20l 4204 v207 tr301 o 401 @4O2 + 403 i( 404 x 405
8406 e407 8408 +409 x4l0 *41I
Envtne and ELLrs Method ! ( H- h) /H vs Z ( HcaLc Hmeas)/Hm eas

Figure48 Errorsin predicteddischargeand depth: EE


0.55
0.50
9€ B
UI

lq "" li:- J'f * e{R


+E +
0.45
*'.
?lo* e o +E-
+E
+
0.40
0.35
: -* i(o {u;?"
x
.E+
a Je+x
:tr
)-c. o.so
3'o *q\ tu
):
x Eea
I
:5 0.25
es fe oF*
*rE

0.20
X + dp.
efr :B*"-."^
rPa
'os
+-
t!+

0 .t 5 X
OE El tr x&+
o ++ +
0 .t 0
X a
-b' !i * +' .+Y *e*
o
4A
0.05 ocF +

0.00 6ro

-50 -40 -50 -20 -10 0 t0 50


3(0calc - 0meas)/Omeas
L21 v26 n3l o53 o34 +59 +43 a47 o l0l e 102
+ f 04 x20l L2O4 v2O7 tr301 o 401 @402 + 403 i.404 x 405
tr406 e4O7 tr408 +409 x4l0 *41|
Gneenht LL llethod 4 ! ( H - h )/ H v s Z ( O c aL c - 0 m e a s ) / Q m e a s
b

0+55
@e^
0.50 .*, g
% r - @ *- e av o
0.45 4* o
+
$
_ T, 6.o
0.40 n(F
x ^*t*"-1g. n'
0.55
0.30
d..., ^,-fl
I
a-rF
t! a
I & Eeoea
x 0.25
1ilft
.:s#
?X
0.20
0.r5 X
X
0.r0
alalF ffi
0.05 +
0.00
-t4 -r0 -8 -6 -4 -2 t0 t2 t4
l(Hcalc - Hmeas)/Hmeas
A21 v26 tr31 o53 o34 +59 +43 847 o l0l C I1 0 2
+ 104 x20l a204 v207 tr501 o 401 o402 + 403 i:404 x405
a 406 e 407 4408 +409 x4l0 *41I
G n e e n h tLL M e t h o d 4 : ( H- h) /H vs Z ( HcaLc - H m e a s )/ H m e a s
JBiNI19|124.?JU

Figure49 Errors in predicteddischargeand depth: GH4


a
0.55 +
+
0.50
0.45
i** o3," l^-tq,***+
:;.. .*1:3H€
t*'*##T
+

"#&it-"
+
"
0.40 *:*tg"
=
0.55
)s. o.so
I
x
x *
"" ##*.
t4H.+a x " :;
ffia:"
5 0.25
x +o
0.20
o
X o
0 .t 5

:.fttr'%"-:
X +
0 .r 0
fs+
0
ooo' +
0.05
6rg
0.00
-50 -40 -30 -20 -10 0 r0
I (0ca Lc - 0meas) /Omeas
A21 v26 o35
oJl o34 +59 + 43 a 47 o10l @102
+ 104 x20l v207
A204 Er501 o 401 s4O2 +403 i:404 x405
a406 e4O7 +409
8408 x410 *41I
G n e e n h tL L Method 5! (H-h)/H vs Z ( 0caLc - Om eas)/Om eas
b
0.55
0.50
OO
0.45 o
o oB
0.40
ooo
oo
0.55 tato*+
)c. o.so ^OO a
I
5 0.25 il.*-* " " ",
0.20
0 .r 5
0.10
0.05
0.00
-t4 -12 -r0 -8 -6 -4 -2
Z(HcaLc - Hmeas)/Hmeas
A21 v26 tr31 o33 o34 +59 + 45 a 47 ol0l @102
+ 104 x20l a204 v2O7 tr301 o 401 o402 + 403 i(404 x405
4406 e4O7 4408 +409 x4l0 *41|
G n e e n h tt I M e t h o d 5 r (H-h)/H vs Z ( HcaLc - Hmeas)/Hmeas

Figure50 Errorsin predicteddischargeand depth: GHs


i--irlo.rao-----

, Main channel
7'nrrr.nr- Location of berm edges

200m
I

Figure51 Locationplanof studyareaon RiverRoding


(afterSellinet al, 1990)

23.0

22.5

22.4
o
o
at
21.5
(D
o
21.4

20.5

20.0

Figure52 Roding at Abridgesamplecross-section


(\l

*( !*( g
rs=
8E
f=o=at
H 6 E
.:/
ggEFE
v, ut o) o

xt( I
l:l
i :

o
(\l
.D
o
c)
E
f
t)
o
g,
rr)E
o
.9
o

ol q 0q q a o{ q
F o o o o o
(nr)uaq a^oqPqldsg

Jgwts3t'2-9P'm

Figure53 Rodingat Abridgemeasuredand calculatedstage-discharges


Plates

sR329 l7l0il93
Appendices

sR 329 t7lo993
Appendix1

Methodsfor determining
equivalent
roughnessof channelswith
compositeroughness

sB 329 t7l0993
Appendix 1 Methods for determining equivalent
roughness of channels with
composite roughness,

In all the following cases, the channe! section is divided into N parts. The
hydraulic radius, wetted perimeter and Manning's roughness coefficient of an
aribtrarySection i are R,, P,and n,, respectively.

Hodon, 1933

Hodon assumed that each part of the cross-section has the same mean
velocity, which at the same time is equal to the mean velocity of the whole
section. On the basis of this assumption, lhe equivalent mefficient of
roughness may be obtained by the following equation,

"=[{''t'*'f
(P,n,Q * .... + PNnN3/2)23
_

of thisequationmust
whichallowsthe derivation
Thevalidityof theassumption
be questioned. The velocity and thus the mean velocity are functionsof
roughnessand depth, and so the mean velocity of parts with different
roughnesses and depthsmustbe different.

Lotter,1933

By assumingthat the total dischargeis equalto the sum of the dischargesin


all the sub'sections,Lotterderivedthe followingequationfor the equivalent
roughnesscoefficient,

n=PRsg
N
E (P,R,*)
ln,

ie

PR5/3
(P,R,* * P.R.ot * * p"Rff)
t\ r,, nN

In derivingthis equationit is assumedthat the bottomshearstressis constant


alongthe wettedperimeter.lt is well knownthat the shearstressactingon the
slopingsidesof the channelis lessthanthe shearstressactingon the bed.

sR 329 l7l0g93
lf the Colebrook-White equation is used instead of Manning's equation the
equivalentroughnesslength is defined by:

q)D
rog,.(10';
K = 14.8nyro(FhLt$p,n,*

Einstein
and Banks,1950

By assumingthatthe totalforceresistingthe flow is equalto the sum of the


forcesresistingthe flow developedin the individualareas,a formulafor the
equivalentroughness can be derivedwhichis,
coefficient

N
(f,(e,n,t;;* _ ( P , n ,+t P " n ! + . . . . . + P n 2 ) E
I
n=.
ph pk

As with the methodproposedby Lotterit has been explicitlyassumedin


derivingthis formulathat, in the channel with the constant equivalen
roughnesscoefficient,the bottomshearstressis constantalongthe wetted
perimeter.One moreassumptionmadeis that the hydraulicradiusof each
sub-dividedseclionis equalto the hydraulicradiusof the wholesection;this
neednot be the case.

Einsteinand Bankstested the above theory by carryingout a series of


laboratory Theyuseda 17ftflume,12incheswideand 18 inches
experiments.
deep with sidesof paintedsheet metal. The bed of the flume compdsed
concreteblocksinto which pegs c-ouldbe inserted. A series of experiments
werecarriedout withthe concreteblocksveilicallyoffsetrelativeto each other
and withand withoutthe pegsinserted.By measuringthe watersufaceprofile
the total resistancewas computed. The resistancedue to each of the
componentsof the bed was also calculated. lt was found that the total
resistanceexeiledby combinedtypesof roughnessis equalto the sum of the
resistanceforcesexededby eachtype individually.

Krishnamurthy
and Christensen,
1972

andChristensenderiveda methodfor calculatingthe equivalent


Krishnamurthy
of a compositechannelby makingthe followingassumptions:
roughness

(a) the wholecross-sectionis assumedto be shallow.


(Thesectonis dividedinto smallervedicalsub-sections).

(b) the hydraulicradius,R,,of eachsub-sectioncan be approximatedby the


vedicaldepth, d,.

(c) the verticalvelocity followsa logarithmic


in eachsub-section
distribution
law.

sR 329 2ZO3l93
is,
and Christensen
The formuladevelopedby Krishnarmurthy

N
E P,d," In n,
Inn=
N
E PP'"'
I

This formulais not applicableto rectangularchannelsbecauseit does not take


accountof side wall effects. However,if the channel is wide and the influence
of the side walls if negligiblethe method of Krishnamurthyand Christensen
can be used. Under lhese conditionsthe above equationcan be modifiedto
give,

N
I P,lnn,
lnn= |
P

and Christensen
ln orderto verifytheirmethodKrishnamurthy useddatafrom
the LowerMississippi river. Theyshowed
that for this datatheirmethodgave
closeragreementwiththe measuredroughnesscoefficient thanthe methods
of Horton,Lotteror Einsteinand Banks.

SR329 2Z0Al93
Appendix2

The lateraldistribution
method

SF 329 t7lOA/93
Appendix 2 The lateral distribution method
In orderto nrodelall of the complexflow mechanismsthat are knownto occur
in onpound channelsa complexthreedimensionalflow andtutbulencemodel
is required. Such modelsare extremelycomplexand requiresophisticated
numericalschemesand powerfulcomputers. Inoder to obtain accurate
representations of the turbulenceandflowfieldsvery srnallnumericalgridsare
required. The cost of collecting such detailed suruey data and the
computational effortrequiredis notjustifiedin typicalengineeringapplications.
Howeverit is possibleto introducesomesimplifyingassumptionsintothe basic
mathematicsand so accountfor the effectsof the small scaleturbulenceon
the overallflow pattern.

This is the approachfollowedin developingthe lateral distributionmethod.


The basic equationsof turbulentflow are known as the Reynolds'equations
and are the mathernaticaldescriptionof all tutbulentflows. By making the
followingassumptionsit is possibleto simplifythese very generalequations:

1) Flowis steady.
2\ Flowis unidirectional.
3) Turbulentshearstressesare linearfunctionsof local velocitygradients.
This is the eddy viscosityconceptand in simple unidirectionalflow this
may be expressedas equation1. Wheret is the shearstress,p is the
fluid density,r, is the eddy viscosityand U is the velocity.

a='-H] (1)

Thenextsimplificationis introduced bydepthintegrating the resuhingequation


overihe watercolumndepth. it is necessaryto assume that the watersurface
is horizontalacrossthe channelwidth. This depth integrationresults in the
following equationfor the lateral distributionof depth integratedflow in a
channel.

sDs+HF.+[",#,]=o (21

Where B is a factor relatings*resson an inclined surface to stress on a


hodzontalplane,D is the localflow depth, f is the Darcyfrictionfactor, g is
q is the unitflow (ie the dischargeper unit width =
gravitationalacceleration,
UD), S is the sur{aceslopeand U is the depth averagedvelocity(refs 1,2).

The variableq mustbe continuousevenacrossa vedicalstep in depthwherd


as the depth averagedvelocity will display large discontinuitiesin these
situations. lt is obviouslypreferableto base calculationson a vadablewhich
is knownto vary smoothlyacrossthe domain.

Turbulencemodelfordepthintesratedflow : At this pointsomemodelmustbe


assumedfor the lateraleddyviscosityq. lt is possibleto use a sophis*icated
turbulencemodelbut a pricemustbe paid in terms of computationaleffott. lt
has beenfoundthat the simplemodel(3) can give acceptableresultsin many
situations.

0t= I,U.D (3)

sR 329 l7l0993
The problemin applyingthis modelis in choosingappropriatevaluesof 1,,the
NondimensionalEddy Viscosity(NEV) and U. is the local shear velocily.
Knight et al (ref 2) have reportedderivedvalues of l' which vary strongly
acrosschannelandfloodplainboth in laboratoryand naturalchannels.While
not disagreeingwith this conclusionthe authors experiencein applyinga
modelbasedon equations2 and 3 indicatesthat adequateprecisioncan be
achievedwith a single value of l, appliedto both channel and floodplain.
Howeverthis is likelyto be true only for the grossdistributionof flow across
the channel.The transponof pollutantsor suspendedsedimentsis far more
sensitiveto the localturbulentstructure,secondarycurrentsetc, whichaffects
the value of 1,. Hence if one is interestedin the distributionof transported
substances this simpleone valuemodelrnaybeinappropriate.

3 The numerical method


The sets of equations2 and 3 whichform the lateraldistributionmethodmay
be solved analyticallyonly in cedain simple situationsand in general a
numericalsolution must be sought. The authors use a finite difference
techniquewitha staggeredgridand Newtons'methodto linearizethe coupled
non-linearequations.lterationis requiredand the initialguessis providedby
settingor = 0. Convergence is usuallyattainedwithin5 iterations.Typically
over 100 pointsare usedfor the numericalintegration acrossa section.

4 Limitations
The LDM is based on the assumptionthat the flow is relativelyuniformly
distributedwith depththroughthe watercolumn. Wherestrongsecondary
currentsexistsuch as in tight bendsthen these simplemodelswill not give
goodpredictions.lt is possibleto modifythe basictheoryto accountfor mildly
curvedflow paths and differingslopes in the main channeland floodplains.
These empiricaladjustmentsare intendedto widen applicationof a model
which is theoreticallyonly applicableto straightchannels. The simple one
parameterturbulencemodel(eqn3) is attractivewhenconsidedngriverflows
since it relatesthe turbulentshear stressesto the channel bed fr'tction. In
riversbedfrictionis usuallythe dominantprocessbut in situationswhereother
effects becomeimportantthis model is less appropriate. One difficultyin
practiceis that calibratedvaluesof l, includethe effectsof secondarycurrents
on the lateraltransportof momentumand so it is difficultto give definitive
guidanceon appropriatevalues.

5 References
WARK J.8., SAMUELSP.G. and ERVINED.A. (1990)- "A Practical
Methodof EstimatingVelocityand Dischargein CompoundChannels",
Intl. Confr.on RiverFloodHydraulics,Wallingford,Oxfordshire,Sept.

KNIGHTD.W.,SHIONOK. and PIRTJ. (1989)- "Predictionof Depth


Int'|.
MeanVelocityand Dischargein NaturalRiverswith fuerbank Flovry'',
Gonf.of Coastal,Estuarineand RiverWaters,Bradford,England,19-21
Sept.

sR 329 17lO993
Appendix3

The FCFAMdesignmethodfor
straightcompoundchannels

sR 32(r '19107/93
Appendix 3 The FCFAM design method for
straight campound channels

7. Introduction
A 'conpound' channel consists of a main channel,which accommodates
normalflows,flankedon one or bothsidesby a floodplainwhichis inundated
duringhighflows.Figure1.1 illustratesa typicalcompoundcross-section
and
definesthe geometricvariables used in the procedures
to follow.

For water levelsabove the flood plain,the flow is stronglyinfluencedby the


interactionbetween the fast-flowingwater in the main channel and lhe
relativelyslow-flowing wateroverthe plains.Thissignificantly
complicates the
estimationof stage-discharge The extraturbulencegeneraledby
relationships.
the flow interactionintroducesenergylossoverand abovethat associatedwith
boundaryresistance.This is not accountedfor by the conventionalresistance
equations(suchas Ch6zy,Manningand Darcy'Weisbach), and their direct
applicationmay resultin considerable error. lf the channelcross-section is
treatedas a unit with one of these equations,the dischargefor any given
stagewill invariablybe underestimated.

The usualapproachpresentedin hydraulics text booksis to dividethe cross-


corresponding
sectionintodistinctsubsections to the mainchanneland flood
plainflows.The dischargefor each subsectionis then calculatedseparately
usingthe Manning(or othersirnilar)equation,andthe totaldischargeobtained
by adding these together.This approach invariablyoverestimatesthe
dischargefor any givenstage.

Variousattemptshavebeenmadeto improvethe latterapproach,usuallyby


includingsuMivisionintedacesin the wettedperimetersto accountfor the
apparent shear stresses induced by the interaction,or by locating the
subdivisionson planesof zeroshear.To date,all of thesemethodshavebeen
basedon the resuhsof smallscale laboratoryexperimentsand are unreliable
on naturalriverscales.

The procedurepresentedherewas developedby P Ackers(1991)and follows


the channelsubdivisionapproach.Subsectiondischargesare calculatedand
addedto obtaina 'basic' discharge,whichis then adjustedto accountfor the
etfects of the interactionbetween the subsectionflows. The adiustment
requireddependson the characteristics of the channeland also varieswith
stage.Fourregionsof flow behaviourare identified,as shownin Figure1.2.
This diagrampresentssometypicalexpedmentalresults,showingthe ratioof
actualtobasicdischarge(onthe horizontalaxis) for a rangeof dimensionless
flow depths (on the verticalaxis). The effect of flow interactionis complex,
altematelyincreasingand decreasingwith flow depth through the different
regions.Alsoshownon this diagramis the curveof channelcoherence.This
is the ratio of the conveyancecalculatedas a single cross-sectionto that
calculated by summing the conVeyancesof the separate flow zones. A
differentadjustment functionis definedfor eachregion,butas the limitsof the
regions vary with channel charactedsticsit is not possible to identify the
appropriateregionfor a padicularwaterlevel beforehand.A logicalprocedure
is given, however, for selecting the correct discharge value from those

sR 329 0707/93
steps are interspersedsa that the correct region is identifiedat the earliest
opportunhy,to avoid unn€cessry calculations.

Step 14 appliesthe additionalconectionto accountfor devbt'lonbetweenthe


mainchannelandfbod plainalignments.

Preliminaryinvestigationssuggestthat mostUK riverswithcompoundsect'ons


wiflflow in Regions1 or 2 for floodswith recurrenceintervalsup to about 20
year. Calculationsshouldbe carefullycheckedif higherregionsare indicated.
Artificialormodifiedchannelsmay operateover a widerrangeof regionsthan
naturalones.

Step 1. Determinethe longitudinal gradient of the channel reach, So,


from survey information.

Step 2. Determine the geometric variables required to defino the


adjustment functions. The basic dischargesfor the rnainchannel
andfloodplainzonescan be computedusingflow areasand wetted
pedmetersobtaineddirectlyfrom the appropriatesurueyedcross-
section.The dischargeadiustmentfunctions,however,includethe
geometricvariablesdefined in Figure 1.1, and their estinration
requiresrepresentationof the cross-sectionby a basictrapezoidal
geometry.This is done usingthe followingsteps.

as illustratedby Figure2.1, tor


2.1 Plot the surveyedcross-section,
example.

2.2 ldentifythe pointson the cross-sectionwhichmostrealisticallymark


the divisionsbetweenthe mainchannelandthe floodpbins on both
sides. Draw vert'tcallines throughthese pointsto define the bank
lines separatingthe main channel and flood plain zones. The
didance betweenthe bank lines is 2w". lf there is a flood plain on
one sideof the mainchannelonly,theniust one bankline is defined
and w" is half lhe main channelwidth at the level of the division
point.

2.3 Determinethe river bank elevation.This is defined by the bank


elevationsat the locationsof the bank lines - one value if there is
onlyone flood plain,and the averageof the two valuesfor two flood
plains.

2.4 By eye, fit a uniformslope to the main channelbankon each side'


lf the banksare inegularand the actualslopesvary,.fttthe straight
lines to the uppertwo thirds of the bank profiles.The average of
theseslopes,expressedas ratiosof horizontaltoverticaldis*ances,
definess".

2.5 Calculatethe cross-sectionalarea of the rnain channelbelow the


rlverbank elevation(as determinedin s{ep2.3 above)and between
the bank lines,ft"u*, from the surveyedcross-section.

2.6 Determinethe depth of the rnain cfiannel,h. This is the distance


belowthe river bank elevationof a horizontalchannelbed located
so that the area of the trapeziumdefinedby the bed, the top width

sR 329 l7l03/93
(2w")at the riverbank elevation,and the side slopes(s"), is the
sameas &",.. lt can be calculatedas

h = 2w" t ((2w")t - 4 s" A"*.)'t


2s"

It will be obviouswhich of the two solutionsof this equationis


correct.

2.7 Determinethe bottomwidthof the mainchannel,

2b = 2wc- 2hs"

2.8 ldentifythe positionsof the backsof the floodplains.The distance


betweenthesedefinesthemaximum totalcompound channelwidth,
28, for two flood plains.For one flood plainthe maximumvalueof
B is the distancefromthe backof the floodplainto the bank line,
plus w". Note that if the flood plains slope upwardsand are not
completely inundated, thetotalwidth(28) is lessthanthe rnaximum,
with the dry part ignored(see Figure2.1).The limitsof the water
surfacecan be determined fromthe surueyedcross-section.

Step 3. Estimateroughnesscoefficientsfor the mainchanneland flood


plains.Theresistance equationusedis a matterof personalchoice.
Manning'sequation(withcorresponding n values)is probablythe
mostwidelyacceptedandwillbe usedfor describing the procedure,
althoughthis does not necessarilyimply recommendation for its
generaluse. lf measuredstage-discharge dataare available,they
shouldbe used to estimateroughnesscoefficients.For the main
channel,the value(n") adoptedshouldcorrespondto nearbank-full
flows. lt is not possibleto infer the value for the flood plains(nr)
directlyfrom measureddata;a valuemustbe assumed,whichcan
be checked subsequentlyand refined. The slope used for
calculatingthe n values should be the hydraulicgradient,but if
reliable measurementsof this are not availablethe surveyed
channel gradient (S") can be used. lf no measureddata are
available,n" and n, shouldbe estimatedin the usualway.

Step 4. Specify a value for H, the flow depth measured above the
idealized bed of the main channel. The stepsthat follow leadto
an estimateof the dischargeforthis waterlevel.Thesestepsshould
be repeatedfor the requiredrangeof H valuesto definethe stage-
dischargerelationship.

Step 5. Gatculatethe basic dischargesin the main channetand flood


plain zones for the specified flow depth, using Manning's
equation.ln thesecalculations the bank linesbetweenthe zones
shouldbe excludedfrom the wettedperimeters.Areasand wetted
pedmetersshould be measuredfrom the surveyedcross'section,
not the idealizedtrapezoidalsection.

Step 6. Add thE zonal basic dischargestogether to obtain Q5g",the


basic discharge for the whole cross-section.This must now be
adjustedto accountfor flowinteractioneffects.Theadjustmentmust

sR 329 r7l0J93
be made usingthe adjustmentfunctionapplicablein each of four
possibleflow regions;thecorrectvalue will be selectedfrom these
as calculationsproceed.

Step 7. Adjust Qo-* assumingflow is in Region 1.

7.1 GalculateH., the ratio of flow depthson the flood plainsand in the
mainchannel,

H* = (H-h)
H

7-2 Calculatethe Darcy-Weisbach frictionfactorsfor the mainchannel,


f", and the floodplains,fr, usingthe relationship

f = SgRS
v2

in which g is the gravitationalacceleration= 9.81 m/s2,


R is the appropriatehydraulicradius(= AlP, excluding
the banklinesfrom P) (m),
S is the hydraulicgradient,equal to the channel
gradient(S.) for uniformflow, and
V is the appropriatebasicaverageflow veloc'tty(m/s).

V" and V, canbe calculated by dividingthe basiczonaldischarges


(step 5) by the appropriateareas. lf there are two flood plains a
singlevalueof f, shouldbe calculated by usingthe combinedareas,
wetted perimeters and basic discharges.

floodplaindischargedeficit,
7.3 Calculatethe dimensionless

f
Q.zr = - 1.0 H. ;r F

7.4 Calculatethe dimensionlessmain channeldischargedeficit,

Q.ec= - 1.240+ 0.395 =? + G H.


wc

for one floodplain,or

Q.2G = - 1.240+ 0.3953 + G H.


2w"

for two flood plains.

sR32{r 17103/93
In theseequations

c -- 10.42 * o.fiy for S" > 1.0


lc

G = 10.42. [ *c 0.171+ 0.34(1 - S") for S" < 1.0

The value of Q.." should not be less than 0.5. lf the calculatedvalue
is less than this, set it to 0.5 and set Q.rr to zero,

7.5 Calculatethe aspect ratio adjustmentfactor,

2b
ARF =
10h

ARF should not exceed 2.O. lt the calculatedvalue is greater lhan


this, set it to 2.0.

7.6 Calculatethe total discharge deficit, the differencebetween Qo."o


and the actual discharge,

DISDEF = (Qzc + N, Q...) (V" - VJ H h ARF

in which NF is the number of flood plains (1 or 2), and


V", V, are the zonal main channel and flood plain
average flow velocitiesrespectively.

7.7 Calculatethe Region 1 adjusted dischargefor the specified water


level,

Qnr=Q*"o'DISDEF

Step 8. Adjust Qo-,. assuming flow is in Region 2. The adiustrnent is


defined by the channel coherence at a flow depth greater than that
specified. (Channel coherence is the ratio of the conveyance
calculated as a single cross-section to that calculated by summing
the conveyances of the separate flow zones).

8.1 Calculatethe'shift'to be applied to the specifiedflow depth,

shift = 0.05 + 0.OSNF for s" 2 1.0

shift = - 0.01 + 0.05 Nr + 0.06 s" for s" < 1.0

8.2 Calculate the shifted flow depth,

H/=A(h -
shift H)

8.3 Calculatethe channelcoherencefor the shiftedftow depth, F/,

COH = (1 + A.) {((1 + AJ / (1 + f.P.))}u'


1 + A. (A. / (f.P.)))or

sR 329 17103193
i n w h i c hA " = & / & ,
A. is the totalfloodplainflow area (i.e.for both sides
if there are two flood plains),
fu is the mainchannelflowarea,
t. = trltc.
tF is the Darcy-Weisbach frictionfactor for the flood
plains,
fc is the Darcy-Weisbach frictionfactor for the main
channel,
P. = PclPc,
PF is the totalfloodplainwettedperimeter(i.e.for both
sides if there are two flood plains),excludingthe
bank lines,
Pc is the mainchannelwetted perimeter,excludingthe
bank lines.

The areasand wettedperimetersshouldconespondto the required


flow depth,i.e. F/ for this calculation.

The friction factors should also be recalculated,as in Step 7.2,


using F/. lf the shiftedflow depth is abovethe extremelateral
points of the surveyed cross-section,extend the cross-section
verticallyfromthesepointsto the requiredlevelto enableareasand
wettedperimetersto be calculated.

8.4 Definethe Region2 dischargeadjustmentfactor,

DISADF2= COH

8.5 Calculatethe Region2 adlusteddischargefor the specifiedwater


level,

Qne= Qu"","XD|SADF2

Step 9. Determineif Q", is the actual discharge,Q.

lf QBr I Qn then Q = Q",

lf Q = Q", the calculationsare completefor the specifiedwater level,unless


a skewoonection(step14) is required.lf QFr< Qo the actualdischargeis still
unknown;in this case proceedwith step 10.

Step 10. Adjust Qo-,"assumingflow is in Region3.

10.1 Calculatethe channelcoherence,COH,usingthe equationgivenfor


the Qo calculation,btrtforthe specifiedflow depth,H, insteadof F|/.

10.2 Calculatethe Region3 dischargeadiustmentfactor,

DISADF3= 1.567 - 0.667COH

10.3Calculatethe Region3 adjusteddischargefor the specifiedwater


level,

Qng = Q**XDISADFg

sR329 l7l03/93
Step 11. Determineif Q", is the actualdischarge.

lf QR2 . Q*, then Q = Qne

lf Q = Qo the calculations are complete for the specified water level, unless
a skew conection(step 14) is required.lf ORz> Q^. the actualdischargeis still
unknown;in this case proceed with step 12.

Step 12. Adjust Qo-," assuming flow is in Region 4.

12.1 Define the Region 4 discharge adjustmentfactor. This is equal to


the channelcoherencefor the specifiedflow depth,H, as calculated
above for Region 3, i.e.

DISADF. = COH

12.2 Calculate the Region 4 adjusted discharge for the specified water
level,

QRr = Qo."tXD|SADF4

Step 13. Determinewhich of Q". and Q*o is the actual discharge.

lf Qm > QF4 then Q = Qns

lf Qm < QR4 then Q = Qn+

Dischargecalculationsare now completefor the specifiedwater level, unless


a skew conection is required. lf so, proceed with step 14.

Step 14. Apply the skew correction if the main channel is not aligned
with the flood plains. This is done as follows and applies for
angles of skew up to 1d.

14.1 Measurethe angle of skew (in degrees)betweenthe main channel


and the flood plain (O) on a suitable map.

14.2 Calculatethe dischargedeficiencyfrom the resultsalreadyobtained,

DISDEF=Qu""r-Q

14.3 Correct the discharge deficiency to account for skewness,

D|SDEF"k. = DISDEFx (1.03 + 0.074@)

14.4 Recalculatethe actual discharge,

Q=Q*"o-DlSDEF"k"w

Q is the actual dischargefor the specifiedflow depth, H.

sR 329 17103193
3, Procedure for separation of main channel and
flood plain discharges
lf dischargesfor the mainchanneland flood plains are requiredseparately,
they can be estimatedas follows.Thiswill be necessaryif fr is to be estimated
from measureddata. The procedurehas not been verifiedfor skewed main
channelsand shouldbe appliedwith cautionfor such cases.

Step 1. Determinethe actual,adjusted,total dischargefor the required


water level, as describedin Section 2.

Step 2. ldentifythe flow region and calculatethe separatedischarges.

2.1 lf the actualdischargeis in Region1, i.e. Q = Qnr,determinethe


separatedischargesusingthe resultsfrom the predictivemethod
describedin the section2, i.e.

Qe = Q.*"o - Q'r" (V"' Vt) H h ARF

for the mainchannel,and

Q r = Q r u " " -o Q r r ( V " - V t ) H h A R F

for eachfloodplain.

2.2 lf the actualdischargeis in one of Regions2,3 or 4, assumethat


the flood plain dischargesare unaffectedby the interaction,and
allocateallthe adjustment to the mainchanneldischarge,i.e.

Q"=Q"*"i"'DISDEF

Qp = Qro.","

4. Procedurefor estimationof boundary shear


sfress
Boundary shear stresses are required for predicting locations of scour,
designingscour protection,and estimatingsedimenttranspoil rates. These
issueswill be addressedby future research.The followingstepscan be used
for obtainingprovisionalestimatesof the averageshearstress on the main
channel bed and the average and maximumshear stresseson the flood
plains.

Step 1. Catcutatethe average shear stress on the bed in the main


channel.

1.1 Calculate the average boundary shear stress, ignoring the


interactioneffects,

toc = PgR.S

in which p is the density of water (1000 kg/m3),and

sR329 1710993
Rc is the hydraulic radius of the main channel,
excludingthe bank lines from the wetted perimeter.

1.2 Galculatethe dischargeadjustmenlfactor for the main channel,

DISADF. = Q"/Q"*"o

in which Qc is the actual main channel discharge,as calculated


in Section 3, and
Q"o""oisthe basic main channel discharge,as
calculatedin Section 2, step 5.

1.3 Calculatethe correctedaverage boundaryshear stress,accounting


for the interactioneffect,

r*' = r,c (DlsADFc)2

Step 2. Calculate the average shear stress on the surface of the flood
plain, ignoring the interaction effects,

ror = Pg(H-h)S

This will apply on the flood plain surface beyond the zone of interactionwith
the main channel flow. Allow for a maximum local value of 5 t* within a
distanceof 3 h from the bank line.

5, Reference
Ackers,P. (1991)The hydraulicdesignof straightcompoundchannels,Report
SR 281, HR Wallingford,December.

sB 329 17lO3/93
6. Notation
A cross-sec{ionalarea
&"u. area of main channel below bank elevation,from surueyed
cross-section
A. ratio Arlfu
B half the total compoundchannel width for two flood plains;
width of flood plainplus half main channelwidthfor one flood
plain
b halfthe bottomwidthof the mainchannel
COH channelcoherence
DISADF adjustmentfactor appliedto basic dischargeto accountfor
interactioneffects;subscriptwill indicateappropriateregion
DISDEF dischargedeficit,i.e. differencebetweenactual and basic
discharges
DISDEF"k"* dischargedeficit,accountingfor mainchannelskew
f Darcy'Weisbachfrictionfactor,= SgRSA/2
f. ratio trlf"
G parameterin Region1 dischargedeficitprediction
g gravitationalacceleration
H depthof flow in rnainchannel
H. ratioof flow depthson floodplainand mainchannel,i.e.
(H-h)/H
H' shiftedflowdepthin mainchannel(for Region2 prediction)
h depthof mainchannelbed belowriverbank elevation
NF numberof flood plains,1 or 2
n Manning'sroughnesscoefficient
P wettedperimeter
P. ratio P"/P"
O actualdischarge,unsubscriptedfor wholecompoundchannel
Q*." zonal dischargeignoringbank lines from wetted perimeter,
unsubscripted for sum of main channeland floodplainvalues
QF dischargeas adjustedto account for interactioneffects in
regionirdicatedby numeriealsubscript
Q.2 dischargedeficitnormalizedby (V"-Vr)Hh
R hydraulicradius,= NP
S hydraulicgradientof channel
q surveyedchannelgradient
sc side slopeof main channelbank, horizontalfuedical
shift additionto main channelflow depth in Region2 adiustment
prediction
V averageflow velocity
wc hatfwidthof mainchannelbetweenbank lines
p densityof water
to averagebed shearslress
r/ averagernainchannelbedshearstressadjustedforinteraction
effect
O angle of skew betweenmain channeland flood plains

subscripts:

C mainchannel
F flood plain
L left bank

sR329 l7lo993
R right bank
1,2,3,4 region of flow behaviour

sR ges rzloryse
Appendix4

Datafrom laboratorystudiesinto
meandering flow
Appendix 4 Data from laboratory studies into
meandering flow
The stage dischargedata availablefrom the various laboratorystudies are
listedbelow. The data includes:

SERCFCF 820, 821, 825, 8,26, 831, B,32, 833, 834,


838, 839, 843, 846, B.47, 848, 849, 850.

Aberdeen A8100, AB100A,AB1ol, A8102, A8103, AB1O4,A8105.

Vicksburg V8201, VB2O2, V8203, V8204, V8205, \/EP6,


vB.207, V8208, V8209, V8210, V8211.

Kiely K1300,301

Sooky SK401, SK402, SK403, SK4O4, SK405, SK46,


sK407, SK408, SK409, SK410, SK411.

The file formatis as follows:

Thefirstten linesat the top of eachfile are commentlines. Information


about
the data is listedhere.

The firsttwo numbersafterthis are:

1 The test seriesnumber


2 The numberof data points

The stagedischargedatafor eachdischargepointfollows.An Exampledata


line is shownbelow.

Date discharup depttrl deptr2 slope tailgate temp


29 1189 0.01975 593/t 59.32 0.996 667.84 tl.4

Dateis givenas threeintegers: day,monthand year.

Dischargeis givenin cubicmetresper second.

Depthl is the depthin millimetresas recordedin the originaldata files.

Depth2is the depthin millimetresas conectedto the channe!bed slope. This


valueof depthshouldbe used in any analysis. h is wofih notingthat only the
SERC FCFdata was adiustedin this way. Depthl and Depth2for the other
data sets are identical.

valleyslopeof the flume. Thesevatuesshouldbe


Slopeis the longitudinal
multipliedby 1/1000.

Tailgateis the tailgatesettingfor the SERCFCF. This data was retainedin


the filesbut shouldnot be used.

sR 329 r7l0993
Temp is the temperaturein degreescentigradeof the waterin the flumeas
recordedby the investigators.Where the temperaturewas not recordeda
defaultvalueof 15 oChas beenassumed.

sR329 r7l0993
I sERc FcF slage Dlscharge Daia Phase B Meanderlng case

3 F1IE NAME (ASSIGNED FOR ME"ANDR} : SDB2O


4 Pl.an geomet.ry (angle of cross over) : 60
Maln Channel X-sn : Trapezolda I
6 Floodplaln xldth : Standard
7 Floodplaln roughness : smooth
I
9 DATE DISCHARGE DEPTH AS DEPIH AS SLOPE TAILGATE TEMP
10 m3 sec. RECORDED mrn PIOTTED mm

20 l'l

29 11 89 0.01975 59.34 59.32 0. 996 667-84 lt.4


2'l tt 89 0.02512 67.t6 5't.26 o.996 663.84 11.0
17 l1 89 0.02654 7 1. 8 9 71.98 0.996 661.81 12.4
27 tt a9 0.03056 7'r.55 77.t4 0.996 658 -12 10.8
r? 11 89 0 .03308 82.43 82.63 o.996 656.58 t2.4

29 l1 89 0.03630 85.06 86.08 0.996 653.81 11.6


30 11 89 0.04015 91.70 o.996 650.83 t1.5
01 12 89 0.04425 97-19 9'' -55 o.996 648.15 11.6
o7 L2 89 0.04?08 101.92 101.92 6 46 . 0 0 12.7
30 ll 89 0.04782 102.99 103. 02 0.996 644.92 11 . 5

29 lr 89 0.05015 105.29 105.93 0.996 643.39 11.?


27 Lt 89 0.04974 105,68 105.96 o-996 543.49 10.9
20 rl 89 0.04953 106.9? 105.80 o.996 643.22 12.4
q?
16 11 89 0.0546? 111 0. 996 64t -72 13.7
16 ll 89 0.05?02 113.94 113.94 0. 9969 6 40 . 0 0 12."t

t5 ll 89 0.06035 118.81 119.O0 0. 996 636.81 13.6


qn
t6 11 89 0. o?073 131.21 111 c. 996 629.6't !2 -9

1 sERc FcF Stage Discharge Da:a Phase B !,ieandering case


2
3 F i I e n a m e ( A S S I G N E DF O R M E A N D R ) : sDB21
Plan geometry (angle of cross over) : 60
Main Channel x-sn : Trapezoidal
6 FLoodplain wldth : Standard
1 Floodplaln roughness : Smooth
8
9 DATE DISCHARGE DEPTH AS DSPTII AS SLOPE TAILGATE TEMP
10 m3/sec. R E C O R D E Dm PLoTTED m

27 16

28 t1 89 0.08240 1<1 Al 1 6 4 .1 3 0.9960 562.3't 11.5


28 11 89 0.08576 16 5 . 5 6 165.84 0.9960 561.56 11.5
28 11 89 0.09753 169.85 1?0.11 0.9660 560.13 1 1. 4
24 \t 89 0.10960 t7 2.'13 0. 9560 559.'12 11.6
24 rL 89 0 . 1 19 8 0 175.48 i ?q tn 0.9560 559.42 11.6

02 ll 89 0.14940 181.49 0.9560 5 s 9 .b O 1 4. 6


01 11 89 0.20390 191.98 1 9 1 .9 8 0.9986 55?.00 14.0
06 L1 89 0 . 2 49 6 0 199.61 t99.73 0.9950 556.00 12.e
23 11 89 0.30228 207.65 207.95 0.9950 556.90 L2.2
01 11.89 0. 30300 204.46 204.46 0.9903 555.00 14.5

16 11 89 0.44020 227.59 227.63 0.9960 552.00 13.2


16 11 89 0.48501 232.90 232.90 1.0069 551.41 14.4
28 11 89 0.49360 235.02 2 35 . 0 0 0.9960 551,74 11.0
10 02 90 0. ?5670 264.44 264.44 o.9978 546.00 L2.3
10 02 90 0. 8?861 277.48 2'r7.95 0.9960 543.40 12.3

10 02 90 0.98939 2 89 . 1 1 284-13 0,9950 541.40 12.3

1 SERC FCF Stage Dlscharge Dara Phase B lleanderlng case

3 FiIe name (ASSIGNED FOR MEANDR} : SDB25


4 Plan geomelry (angle of cross over) : 60
5 Main channel x-sn : Natural lnbank
o Floodplaln eldth : Standard
'1
FloodpLain roughness : Srnooth
I
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE IAIIGATE TEMP
10 m3 sec. RECORDED m PLOTTED m c.
25 10

1,7 7 90 0.01019 99.57 99.57 0.9960 610.00 16.1


1??90 0.0120? 1 0 3 .5 9 1 0 3 . 59 0. 9954 610.03 15.1
16 ? 90 0.01442 108.40 108.60 0.9560 506.67 16.1
L6790 o .01512 1.12.36 rr2.25 0.9960 605.31 16.0
16 7 90 0.01806 116.50 116.48 0.9960 603.21 16.1
16 7 90 0 . 0 2 15 0 123 -20 123.16 0.9960 500.20 16.0
19 7 90 o.02288 LZJ.IJ t25.66 0. 9960 599.10 15.0
t7790 o.02498 L29.09 L29.23 0.9960 591.59 16.2
18 7 90 o.02646 131.04 131.65 0.9960 596.53 16.0
06 I 90 0.03341 L 4 2. 3 4 132.34 o.997't 592.00 17.1
1 SERC FcF Stage Discha:9e Data Phase B Meandering case
2
3 File name {ASSIGNED FoR MEANDR) : SDB26
4 Plan geom€try (angle of cross over) : 60
5 Main Channel X-sn : Naeural Over bank
FLoodplain wldth : Standard
7 Floodplaln roughness : smooth
I
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPS TAIIGATE TEMP
10 m3 /sec. RICORDED nm PLOTTED m c.

15

12 0? 90 0.03993 752.'15 r52 .64 t62-O6 L6.2


06 08 90 0.04965 162.89 r62 -89 o.9942 557.00 17.1
10 10 90 0.05'r91 1 6 5. 2 3 165.23 1.0030 557-00 13.5
13 0? 90 0.0605r 166.93 167.11 c. 9960 555.53 15.9
10 07 90 0.10310 771.9I 1?7.89 0.9960 55s.10 14.7

12 0? 90 0.16018 189.58 189.58 4.9962 :55.00 15.9


1C 0? 90 0.20447 ) , 9 6. 6 4 rvo-fJ 0.9960 >54.87 14.4
13 07 90 0.26144 206.84 206.44 0.9980 553.88 16.0
10 0? 90 0.30?25 2t2 . 06 2r2.26 0.9960 553.56 14.?
12 07 90 0.38133 223 -35 0.9960 552.31 15.5

11 0? 90 0. 53963 z4!.2! 241.38 0.9960 549.?0 14.9


11 0? 90 0. 54651 2 5 3. 4 2 253.94 0.9950 517.39 15.5
1l. 07 90 0.75203 2 6 1. 5 9 264 -96 0.9960 545.60 15.6
11 0? 90 0.858?3 2 ' t1 . 8 2 zt>.zJ 0.9960 4 4 4. 4 ' t 1 5 .?
27 10 90 0.97?58 t6t- tb 287 -69 0.9960 t12.59 13 . 9

21 tO 90 1.09296 296.46 296.46 1,0167 3 41 . 3 3 14.2

sERC FCF Stage Discharge Data Phase B Meandering case

F 1 I E N A M E ( A S S I G N E DF C R M E A N D R ) : SDB31
Plan geoneiry (angle o: cross over) : 60
Main Channel X-sn : Nacural
Floodplain widch : Narrow
Floodplain roughness : Snoo[h

DATE DISCHARGS DE?TH AS DEPTH AS SLOPE TA]LGATE TEMP


C m 3/ s e c . R E C O R D E Dm PLoTTED mm

31 14

22 9 90 0.03871 159-80 159.83 o.9960 = 6 2. 5 7 1 4- 0


22 9 90 0.050?9 rb6,)r 168.s1 o.9913 13.9
19 9 90 0.05?66 l't c .2I 1?0.15 0 . 9960 :59.68 t4 -4
2\ 9 90 0 . 0 6 6 16 1?3.64 1 ? 3 .6 4 0. 9948 :60.0 t4.4
18 9 90 0.08538 l en o1 180.05 0. 9960 t4 .0

ql
18 9 90 0.11231 l tq 188.75 14.1
2i 9 90 0.16120 202 .'16 o ooql 359.0? L4.4
21 9 90 0.19296 249.1 6 209 - 94 0.9960 t58.2L 1 4. 4
18 9 90 0.22834 218.89 219.18 0.9960 556.68 1 4. 3
t9 9 90 0.28208 230.53 230.85 0.9960 554.39 L4.7

21 9 90 0.29029 232.A6 232.46 0. 9968 5 55 . 0 0 1 5, 4


20 9 90 0.38009 251.55 251. s8 0.9950 552.93 15.6
20 9 90 0.4?314 264.45 264 .68 0 - 9960 550.59 l5 .3
20 9 90 0.57135 282.52 242 -'t8 0.9960 547 - 47 1 5. 0

1 SERC FcF stage Discharge Data Phase B Meanderlng case


2
Flle name (ASSIGNED FOR MEANDR) SD832
Plan geometry (angle of cross over) 60
5 Maln Channel X-sn Natura 1
Floodplaln width Standa rd
1 Floodplaln roughness Breeze blocks slmulatlng piers
8
9 DATE DISCHARGE DEPTH A5 D E P T HA S STOPE TAILGATE TEMP
10 m3/sec. R E C O R D E Dm PLOTTEDmm

I3

10 01 91 0.04333 1s9.91 158.85 0.9950 558.62 12.3


10 01 91 0.05430 165.26 16 5 . 3 7 0.9960 555.50 L2.2
10 01 91 o.o9961 178.41 1?8.54 0.9960 5 5 4. 5 5 13.4
10 01 91 0.13331 185.85 186.00 0.9960 5 5 3. 6 2 1 3. 3
1r 01 91 0.19814 198.50 0-9960 533.3? t3-4

10 01 91 o.267L4 2 1 0. 9 2 2L0.92 0.9973 550.00 13.3


11 01 91 0.33568 221..92 22r.87 0.9960 54?.56 13.0
11 01 91 0.395?6 231.31 231.88 0.9960 545.49 13.4
26 10 90 0.45946 2 3 8. 2 3 23A.23 0.9959 543.00 75.2
26 10 90 o.5120'l 2 5 4. 9 3 254.-17 0.9960 539.20 15.0

26 tO 90 0.68564 2 6 9. 3 5 269.52 0.9960 535.29 14-5


26 tO 90 0.80040 2 A 4. 6 7 2a4.43 0.9960 53t.22 14.2
26 10 90 0.91E32 297.51 298.55 0.9960 5 2 - .t 5 3 13.9
I SERC FCF Slage Dlscharge Data Phase B Meandering case
2
Flle name (ASSIGNED FOR MEANDR) : SD833
4 Plan geometry (angle of cross over) : 60
Maln Channel X-sn : Natura I
5 Floodplain wldth : standa rd
FloodpLain roughness : PartlaIIy roughened dowel rods
I
9 DAIE DISCHARGE DEPTH AS DEPTH AS SLOPE TAII"CATE TEMP
l0 m3/sec. RECORDEDriln PLOTTED mm c.

33 LZ

05 11 90 0.04161 L>6.t t 158.68 0. 9950 5 5 8 . 15 L2.2


05 11 90 0.06651 169.86 169.83 0. 9960 553.06 12.2
0l 11 90 0.08653 176.51 176.51 1. 0012 557.-12 t2.A
01 ll 90 0.11244 1E3.6? 183.60 0. 9960 550.1? 13.0
01 t1 90 0.16988 tvD.J) 196.83 0. 9960 546.61 13.0

05 l1 90 0.22181 207.55 20?.80 0. 9660 543.63 12.1


05 11 90 0.27I'14 218.23 2r8.23 o.9966 540.00 11 , 8
1l 90 o.32182 228.39 228.50 0.9960
3 1 l0 90 0.49790 258.89 2 5 ' t. 8 7 0. 9960 524.22 13.4
31 10 90 0.55938 271.'19 21L.95 0. 9960 5t1.16 13.0
1l l0 90 0. 6?561 289.11 289.22 0. 9960 509.89 13.3
3 1 10 90 0.75534 305.01 305.01 0. 99?5 503.00 13.3

1 SERC FCF Slage Dlscharge Data Phase B Meandering case

3 Flle name (ASSIGNED FOR MEANDRI S D 8 34


4 Plan geometry (angle of cross over)
5 Main Channel X-sn Natura I
5 Floodplain width St anda rd
'l
Floodplain roughness Roughened with Dowel Rods
I
9 DATE DISCHARGE D E P T HA S DEPTHAS SLOPE TAI LGATE TEMP
I 0 m3/sec. R E C O R D E Dm PLOTTED mm

34 18
16 11 90 0.04015 158.61 158.58 0.9960 557.03 15.0
16 11 90 0.05445 7 5 1. 5 4 t61-64 0.9951 552.01 14.5
16 11 90 0.06742 1?4.00 I'14.32 0.9960 549.24 14.4
15 11 90 0.09209 184.90 184.80 0.9960 544.63 15.0
15 11 90 0.11197 1 9 2 .1 3 192.50 0.9960 54r.24 14.'1
15 11 90 0.13203 200.?9 201.01 0.9960 53?.18 15.0
15 11 90 0.15440 2 09 . 3 6 209.60 0.9960 532.63 14.8
14 11 90 0. 17485 2I7.2L 2L7.t1 0.9960 529.O5 15.4
12 11 90 0.1985? 2 2 4. 4 8 224.45 0.9960 525.O9 13.8
12 11 90 0.23395 238.71, 238.65 0.9960 517.06 L4.2

14 11 90 0.26401 250.38 250.83 0.9960 510.00 15.0


12 11 90 0.2?588 255.26 254.83 0.9960 50?.97 14.8
14 11 90 0.30182 255.00 265.35 0.9960 501.62 15.0
12 tl 90 0.32655 272.95 2 ' t 3. 4 0 0.9960 491.L2 14.5
13 1l 90 0.34158 2'18.44 2 79 . 0 8 0.9960 493.31 14.8

14 11 90 0.3?602 292.09 292.25 0.9960 485.36 L4.6


L3 11 90 0.41000 301.68 301.96 0.9960 4 1 9. 6 4 15. 4
13 11 90 0.45527 31?.31 317.55 0.9960 469.64 15.0

1 SERC FCF Stage Dlscharge Data Phase B Meanderlng case


2
3 Flle name (ASSIGNED FOR MEANDR) : SDB38
4 Plan geomecry (angle of cross over) : 110
5 Maln Channel X-sn : Natural Inbank
6 Floodplaln wldth : standard
Floodplain roughness : Smooth
8
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
10 m3/sec. RECORDEDnn PLOTTED MM

t1

23 4 9L 0_01135 109.80 110.06 1.0210 604.L7 10.4


31 7 91 0-Ot322 115.16 115.16 1.0144 601.00 15.0
24 4 9L 0.01533 1 2 0 .3 0 1 2 0. 3 0 1.0218 598.9s 10.5
31 ? 91 0.01560 120.73 1 2 0. 7 3 t.o2L7 5 9 7. 9 2 14.4
01 8 91 0.01699 r24.20 L24.20 1.0161 596.00 14.1

23 4 91 0.01873 127.91 L2't.97 .0205 595.O0 10.3


30 ? 91. 0.02006 1 3 0 .? 2 130.72 .0161 594.22 15-9
01 8 91 0.02206 135.65 135.66 -0262 s90.50 1 4. 9
23 4 9L 0.02342 138.82 138.54 .0210 588.59 10.3
24 4 9t 0.02432 140.08 !40.2'l .0210 587.69 10.5

23 4 9L O.O2178 1.46.72 1 4 6. 7 2 .0210 584.46 r0.4

DEPTHS SET FOR DR I G U Y M E RF O R D I S P E R S I O N T E S T S


I SERC FCF Slage Dlscharge Data Phase B Heandering case
z
3 FIIe name (ASSICNED FOR MEANDRI : SDB39
4 Plan geomelry (angle of cross over) : llo
5 Maln Channel X-sn : Natural
6 Floodplatn Hldth : sEandard
7 Floodplaln roughness : Smooch
6
9 DISCHARGE DEPTH AS DEPTH AS SLOPE TAI I6ATE TEMP
1o m3/sec. RECORDEDm PIOTTED rutr c.

39 14

05 ? 91 0.03815 165.14 165.14 1.0177 557.00 15.5


22 4 9t 0.0?693 118.92 t't8.92 1.0210 553.50 10.3
22 4 97 0.09912 183.58 183.76 1.0210 553.35 10.3
23 5 91 0.14208 19 3 . 6 s I 93. 43 1.0210 552.56 12.7
08 5 91 0.1?92s 200.61 200. 61 l -0221 552.00 11.4

23 5 91 0-25282 2 7 4- O 7 2t4.15 1.0210 5 49 . 6 3 t2.9


03 7 9t 0.32461 225.2r 225.20 1.0210 :48.26 14.5
03 7 91 0.39138 235.01 235.13 !.vz!v r16.61 15.4
03 7 91 0.4451? 242-66 242.'t1 1.0210 315.20 15.3
0? 5 91 0.55351 256.86 256.86 1.0201 541.07 tt.2

07 5 91 0.5613? 2'to.o5 2 70 . 3 3 1.0210 538.6 1l .6


07 5 91 0.?7988 284.50 2 8 4. 4 8 1.0210 335.60 12.0
24 5 91 0.88127 296.44 296-17 t -vztv 532.73 13.2
24 5 97 0.94356 302.82 302. 88 332.43 14.5

1 SERC FCF Stege Di.scharge Data Phase B l.teanderlng case

FiIe name (ASSIGNED FOR MEANDR} : S D 8 43


PIan geometry (angle of cross over) : 110
Maln Channel X-sn : Natura I
Floodplaln wldth : s! anda rd
1 Floodplaln roughness 3 Roughened with cjorel rods
I
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE :AILGATE TEMP
10 m3/sec. R E C O R D E Dm m PLOTTED rnm c.

43 t5

02 9 9L 0.03252 160.54 160.47 1.0210 558.13 16.6


23 8 91 0.03688 165.98 165.98 1.0124 556 . 00 16.'t
15 I 91 0.05451 L75.O2 t /).uo 1.0210 :50.42 t6.3
l6 I 91 0.08361 l8?.39 18?.60 1.0210 314.68 16. 3
15 8 91 0.10803 197.19 101 da 1.0210 =10.21 16,1

29 A 9L 0.1168s 200.85 200.85 1.0205 539.00 L6.1


21 8 91 0.14392 2lt.',3 27t.75 1.0210 s33.30 16.9
21 I 91 0.1737s 2 2 ) ,- 7 5 221,.93 1.0210 521.26 L6.7
19 8 91 0.201t2 231.55 zJL-v5 1.0210 523.13 15.3
21 8 91 0.24342 24't.40 2 4 7. 3 5 1.0210 5'14.40 15.4

19 8 91 0.2784A 259.26 259.2'l 1.0210 5C7.51 16.5


19 8 91 0.316t6 212.21 t?1 a o 1.0210 5C0.31 16.3
20 I 91 0.34940 283.40 2A3.42 493.42 16.5
20 I 91 0.38851 296.31 296.60 486.25 17.0
20 8 91 0.43331 311.19 310.53 477.62 L6.4

1 sERc FcF stage Discharge Dala Phase B Meandering case

3 FlIe name (ASSIGNED FOR MEANDR) : SDB45


4 Plan g€ometry (angle of cross over) : 110
5 Maln Channel X-sn : Natural
6 Floodplain wldLh : . Standard
1 Floodplaln roughness : Roughened wlth breeze blocks
I
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
t- 0 m3/sec. RECORDED m PLOTTED MM

46 14

17 9 91 0.03525 1,62.88 t62.43 1.0210 5 59 . 0 7


17 9 91 0.05185 171.30 171.30 t.0224 1 5. 7
16 9 91 0.10054 184.99 185.18 1.0210 552-42 15.7
19 9 91 0.12617 190.98 19 0 . 8 9 1.0209 5 5 2 .O O 15.4
19 9 91 0.157?5 t97 .96 1 , 9 7- 9 6 1.0203 555.00 1 5. 3

l? 9 91 0 . 16 8 1 ? 200.65 200- 64 L.VZLU 5 4 9 -5 4 15.8


19 9 91 0.21985 2ro.96 2L0.77 1.0210 547.34 15.4
19 9 91 0.2524-t 2t6 .8 4 2 1 ?- 1 5 1.0210 545-6? 1 5 .?
l7 9 91 0.31685 229.81 2 2 9. 4 5 1.0210 542.43 1 6. 0
20991 0.36295 237 -38 2 3 1- 5 5 1.0210 5 3 9 .? 8 L4.?
q?1 ql
20991 0.46613 254.90 254.95 1.0210
t8 9 91 0.53525 zoJ. t6 2 6 3. 6 0 1.0210 5 3 L. 2 1 15.8
18991 0.54489 280.09 rad 1q 1.0210 5 2 6. 6 3 16.6
lE 9 91 o.75129 296.26 2 9 6. 3 5 1.0210 520.8? 16.3
1 SERC FCF Stage Discharge Data Phase B Meanderlng case

J F i I C N A N E ( A S S I G N E DF O R M E A N D R ) : SDB47
4 PIan geometry (angle o! cross over) : 110
Maln channel X-sn : Natural
6 Floodplaln uidth : Narrow
Floodplain roughness : Snooth
8
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
l0 m3/sec. R E C C R D E Dm PLOTTED nm

4'l 14

03 10 91 0.03559 1 6 1. 8 4 t54 .94 1.0210 556.44 t3.9


01 10 91 0.05023 1?1.12 1.0206 555.02 14.3
30 09 91 0.07456 r /v-bu 179.50 1 .0210 553.62 14.2
30 09 91 0.09496 185.23 185.30 1.0210 553,50 13.8
07 10 91 0.13586 t95.66 19 5 . 6 3 1.0210 551.44 13.2

30 09 91 0.16491 2 0 2. 4 8 202.65 1.0210 549.50 14.0


01 10 91 0.220s0 ttJ.)) 213 -54 1.0210 547.00 13.6
03 10 91 0.25933 zzt -b) 227.65 1.0210 544.1A 14.7
01 10 91 0.32792 231.20 2 3 4- t 2 1.0210 540.55 14.1
03 l0 91 0.38649 244.52 1.0210 538.20 14.0

07 10 91 0.48556 259.96 259 - 63 1.0210 534.10 13.3


04 10 91 0.53429 261.95 2 6 4. 9 3 1.0210 532.85 14.5
08 10 91 0.6984? 2 A A. 4 2 2 A A- 4 6 1.0210 526.23 14.3
08 10 91 0.750s5 2 9 4- 1 5 294 .60 1.0210 524.10 13.4

1 SERC FCF stage Discharge Data Phase B Meanderlng case

F l l e n a m e ( A S S I G N E DF O R M E A N D R ) S D 8 48
PIan geometry (angIe of cross over) 110
5 Main channel x-sn Natura I
6 FloodpLain uidth Standard
1 Floodplain roughness smoo!h with walls
8
9 DATE DISCHARGE DE?TH AS DEPTHAS SLOPE TAI LGATE TEMP
10 m3/sec. R E C o R D E Dm PLOTTED mm

48

I5 10 91 0.03353 170.30 169.85 1.0210 560.46 13.5


14 10 91 0.03499 l 7 4. 9 5 175.05 1.0210 5 55 . 9 0 14.0
14 10 91 0.0406s 189.95 189.95 1.0210 5 41 . 8 0 14.0
11 10 91 0.04540 202.60 2 0 2. 6 0 1.0210 534.r2 13.9
14 10 91 0-04945 2 \ 2. 6 8 2r2.68 1.0210 )z>.)u lJ..

l5 10 91 0.05538 232.44 t.0223 51C.00 13.4


11 10 91 0.05659 233.15 t12 2q 1.0210 5C9.45 13-6
15 10 91 0.06312 2 4 2. 3 5 242.O5 1.0210 502.40 13.2
I SERC FCF Scage Discharge Data Phase B Meanderlng case
2
File name : SDB49
4
Maln Channel X-sn : trapezoldal
6
1 Floodplaln roughness : rod roughness Phase A orientatlon
I
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
10 m3/sec. R E C o R D E Dm PLOTTSD mm

49

03 01 92 o.02252 29.5A6 29.64 t.o224 540.0 1 1. 1


03 01 92 0.04432 50.544 50.57 1.0130 5 2 7. O 10.9
02 01 92 0.06r83 65.082 65.08 1.0210 518.0 10.5
05 01 92 0.07971 8 4. 1 3 0 83.78 1.0152 505.0 11.6
02 01 0.10305 110.895 109.25 0.9897 485.0 10.8

05 01 92 0.13002 t34.722 135.52 1.0281 470.0 12.1


03 01 92 0 . 1 5 35 2 152.180 163.43 1. 0 3 6 3 453.0 1 1. 4

1 SERC FCF Stage Dlscharge Data Phase B Meanderlng case

FiIe nane : sDB50


4
5 Mai.n Channel X-sn : traPezoidal
6
'1
Floodplaln roughness : rod roughness Phase B orientatlon
8
DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
10 m3/sec. RECORDED m P L O T T E Dm m c.

10 01 92 o. c2225 2 8. 2 4 1 2 A. 2 4 1.O210 542 12.3


09 0r 92 0 . 0 41 3 4 15.233 45.42 l.o2'11 )JJ IJ.]
08 01 92 0.06159 62 - 602 62.00 1.0038 522 12.9
08 01 92 o.o1942 84.135 83.40 1.0060 505 13.1
08 01 92 0.1c015 1Ci.256 101.98 1.0439 495 13.3

09 01 92 0.13394 132.168 132.11 1.0190 4'75 13.3


09 01 92 0.16028 153.969 154.10 1.0238 460 13.1
1 VICKSBURGDATA SET 2Tt WIDE MAIN CHANNEL
2
3 FlIe name (ASSIGNED FoR MEANDR) : SDVB201
4 Plan geometry (SINUOSITY) : 1.5?1
5 Main Channel X-sn : Trapezoidal
6 Floodplaln widrh 2 4.877
? Floodplain roughness : 0.012
8
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
10 m3/sec. R 0 C O R D E Dm PLOTTeD mn C

20r 3

01 01 01 0.0?02 182.88 182.88 1.0000 0.00 15.0


01 01 01 0.1424 213-36 213.36 1.0000 0.00 15.0
01 01 01 0.2237 243.84 243.84 1.0000 0.00 15.0

1 VICKSBURC DATA SET 2fI WIDE MAIN CHANNEL

3 F l I e n a m e ( A S S I G N E DF O R M E A N D R } : SDVB2O2
4 Plan geometry {SINUOSITY} : 1.571
5 Main channel X-sn : Trapezoidal
5 Floodplain widrh : 4.87'l
7 Floodplaln roughness : 0.025
I
9 DATE DISCHARGE DEPTHAS DIPTH AS SLOPE TAILGATI TEMP
10 m3/sec. R E C o R D E Dm P L O T T E Dm m C

01 01 01 0.0430 182.88 182.88 1.oo00 0.00 15.0


01 01 01 0.087s 2L3.36 213.36 1.0000 0.00 15.o
01 01 01 0.1546 243.84 243.84 1.0000 0.00 15.0

VICKSBURGDATA SET 2ft WIDE MAIN CHANNEL

F i l e n a m e ( A S S I G N E DF O R M E A N D R ) : SDVB2O3
Plan geomerry (SINUoSITY) : 1.571
Main Channel X-sn : Trapezoidal
6 Floodplain width : 4.87't
7 Floodplain roughness : 0.035
8
9 DATE DISCHARGE DTPTH AS DEPTH AS SLOPE TAILGAT: TEMP
10 m3/sec. R E C O R D E Dm P L O T T E Dm m

203 3

01 01 01 0.0396 182.88 182.88 1.0000 0.00 15.c


01 01 01 0.0733 213.36 213.36 1.0000 0.00 15.0
01 01 01 0.1283 243-84 243.84 1.0000 0.00 15.0

1 VICKSBURG DATA SET 2ft WIDE MAIN CHANNEL

3 FlIe nane (ASSIGNED FOR MEANDR) : SDVB204


4 Plan geomerry (SINUOSITY) : 1.400
5 Main Channel X-sn : Trapezoj.dal
6 FloodPlaln widrh . 4.a77
7 FloodpLain roughness : 0.012
I
9 DATE DISCHARGE DEPTH AS DEPTII AS SLOPE TAILGAT' TEMP
10 m3/sec. R E C o R D E Dr n n PLoTTED mm C

204 3

01 0L 0r 0.0830 182.88 182.88 1.0000 0.00 15.0


01 01 01 0.1560 2t3-36 213.36 1.0000 0.00 15.0
01 01 0t 0.2430 243-84 243-84 1.0000 0.00 15.0

1 VICKSBURG DATA SET 2ft WIDE MAIN CHANNEL


2
3 FIIE name (ASSIGNED FOR MEANDR) : SDVB2O5
4 Plan geometry (SINUOSITY) : 1.400
5 Maln channel x-sn : Trapezoidal
6 Floodplaln uldth | 4.877
7 Floodplaln roughness : 0.025
8
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
10 m3/sec. R E C o R D E Dn u n PLoTIED mm c

205 3

01 01 01 0.0490 182.88 182.88 1.0000 0.00 15.0


01 01 01 0.0985 2t3.36 213.36 1.0000 0.00 15.c
01 01 01 0.1679 2$.8q 2$.84 1.0000 0.00 15.0
1 VICKSBURG DATA SET 2fC TIDE MAIN CHANNEL
2
3 FlIe name (ASSIGNED FOR !,!EANDR) : SDVB2O6
4 Plan geonetry {SINUOSITY) : 1.400
5 Maln channel x-sn : Trapezoidal
6 Floodplain wtdth | 4.87'l
7 Floodplaln roughness : 0.035
8
9 DATE DISCHARGE DEPTH AS DEPTII AS SLOPE TAIICATE TEMP
10 n3/sec. R E C O R D E Dm PLOTTED mm C

206 3

01 01 01 0.0439 182.88 182.88 1.0000 0.00 15.0


01 01 01 0.0832 213.36 213.36 1.0000 0.00 15.0
01 01 01 0.13?3 243.84 243.84 1.0000 0.00 15.0

1 VICKSBURG DATA SET 2ft WIDE MAIN CI]ANNEL

3 F l I e n a m e ( A S S I G N E DF O R M E A N D R ) : SDVB207
4 Plan geometry (SINUOSITY) : 1.2C0
5 Mai.n channel X-sn : Trapezoidal
6 Floodplain vldth : 4.8'l'l
7 Floodplain roughness : 0.012

9 DATE DISCHARGE DEPTS AS DE?TH AS SLOP| TAITGATE TEMP


10 m3/sec. R E C o R D E Dm P L O T T E Dm m c

01 01 01 0.0915 i82.88 182.88 1.0000 0.00 15.0


01 01 01 0.t't92 213-35 213.36 1.0000 0.00 15.0
01 01 01 0.27'12 243.84 243.84 1.0000 0.00 15.0

1 VICKSBURG DATA SET 2ft WIDE MAIN CHANNEL


2
3 F1le nane (ASSIGNED FOR MEANDR) : SDVB208
4 Plan geometry (SINUOSITY) : 1.200
5 Maln Channel. X-sn : Trapezoldal
6 Floodplain wldth : 4.871
7 Floodplain roughness : 0.025
8
9 DATE DISCHARGE D E P T i . IA S D'?TH AS SLOP' TAILGATE TEMP
10 m3/sec. RECORDED m P L C T T E Dn m C

208 3

01 0l 01 0.0550 182.88 182.88 1.C000 0.00 15.c


01 01 01 0.1090 213.36 213.36 1.0000 0.00 15.c
01 01 01 0.1798 243.81 243.A4 1.0000 0.00 15.0

1 VICKSBURG DATA SET 2fI WIDE MAIN CHANNEI

3 FIIE NA:NC (ASSIGNED FOR MEANDR} : SDVB2O9


4 PIan geometry (SINUOSITY) : 1.200
5 Main channel x-sn : Trapezoldal
6 Floodplain xldt.h 2 4-877
7 Floodpl.ain roughness : 0.035
8
9 DATE DISCHARGE DEPTH AS DE?TH AS SLOPE TAILGATE TEMP
10 m3/sec. R E C o R D E Dm PLOTTED mm c

209 3

01 01 01 0.0484 182.88 182.88 1.0000 0.00 1s.0


0t 01 01 0.0864 213.35 213.36 1.0000 0.00 15.0
01 01 01 0.143? '243.a4 243.84 1.0000 0.00 15.0

1 VICKSBURG DATA SET 2fT WIDE MAIN CHANNEL


2
3 Flle name (AsslcNED FoR MEANDR! : sDVB2lO
4 Plan geometry (SINUOSITY) : 1.200
5 Main Channel X-sn : Trapezoldal
6 Floodplaln wldth 2 9.\44
7 Floodplaln roughness : 0.035
8
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
10 m3/sec. R E C o R D E Dm i PLoTTED m C

2LO 3

01 01 01 0.0614 182.88 182.88 1.0000 0.00 15.0


01 01 01 0.1399 213.36 213.36 1.0000 0.00 1s.0
0t 01 01 0.2449 ?43.84 2q3,84 1.0000 0.00 15.0
1 VICKSBURGDATA SET 2fT WIDE MAIN CHANNEL

3 F l I e n a m e ( A S S I G N E DF O R M E A N D R ) : SDVB211
4 PIan geometry (SINUOSITYI : 1.571
5 Maln Channel x-sn : Trapezoidal
6 Floodplaln width : 9.144
? Floodplaln roughness : O.O35
8
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAIICATE TEMP
10 m3/sec. R E C O R D E Dm PLOTTED mm C

2It 3

0L 01 01 0.0507 182.88 182.88 r.0000 0.00 1s.0


01 01 01 0.1178 213.36 213.36 1.0000 0.00 15.0
01 01 01 0.2226 243.84 243.84 1.0000 0.00 15.0

1 KIELY (UNIV COLLEGE CORK) DATA SET

3 F i L e n a m e ( A S S I G N E DF o R M E A N D R ) : SDKI301
4 Plan geomerry (sINUosITy) | 7.224
5 Main channel X-sn : RECTANGULAR
6 Floodplain widrh : 1.200
7 Floodplain roughness, : Smooch
8
9 DATE DISCHARCE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
10 m3/sec. R E c o R D e Dm PLoTTED mn C

301 5

01 01 0r 2.438-3 51.1 54.1 1.0000 0.00 15.0


01 01 01 3.10E-3 60.0 60.0 1.0000 0.00 15.0
01 01 01 6.708-3 69.'t 69.'t 1.0000 0.00 15.0
01 01 01 11.18-3 8C.0 80.0 1.0000 0.00 15.0
01 01 01 16.3E-3 A9.4 89.4 1.0000 0-00 15.0

P O I N T S B E L O WA R E T H R E E I N B A N K RESULTS AND THE BANKFULL POINT FROM THE


RATING CURVE

01 01 01 0.6698-3 20.0 20.0 1.0000 0.00 15.0


01 01 01 1.303E-3 28.4 2A.4 1.0000 0.00 15.0
01 01 01 2-O42E-3 40.0 40.0 1.0000 0.00 15.0

01 01 01 2.324E-3 s0.0 5 0. 0 1.0000 0.00 15.0

NOTE THESE DATA HAVE BEEN SCALED OFF OF A PLOT


I T O E B E S + S O O K Y( S O O K Y ' S T I . i E S I S ) D A T A S E T

3 Fll.e name {ASSIGNED FOR MEANDR} : sDsK401


Plan geomeEry ( S I N U O SI T Y } : 1.I (GIVEN BY SOOKY}
5 Maln ChanneL X-sn : RECT 1.5" DEEP
6 Floodplaln rrldth : 3.886' 1.1845n SL = 0.675E-3
Floodplain roughness : Smooth
8
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAII,GATE TEMP
l0 m3/sec. R E C O R D E Dm PLOTTED nm

401

01 01 01 6.309E-3 63.6 63.6 0.675 0.00 15.0


01 01 01 ?.886E-3 70.6 70. 6 0.675 0.00 15.0
01 01 01 9.4638-3 73.5 0.6?5 0.00 15.0
01 01 01 1 1 . 0 4 1E - 3 71.5 0.6?5 0.00 15.0
01 01 01 12.618E-3 80. 6 80. 6 0.6?5 0.00 15.0

1 ( S O O K Y . ST H E S I S ) D A T A S E T
TOEBES+SOOKY

Flle name (ASSIGNED FOR MEANDR) : sDsK402


PIan geometry ( 5 I N U O SI T Y } : 1.1 (GIVEN BY SOOKY)
Main Channel X-sn : RECT ]..5" DE:P
6 Floodplaln uldth : 3.886' 1.1845n SL - O.87E-3

7 Floodplaln roughness : Snoo!h


8
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
10 m3/sec. RECORDED m PLOTTED mm a

402

01 01 01 6.3098-3 6 2. 2 0.8? 0.00 15.0


01 01 01 ?.886E-3 b/-6 0.87 0.00 15.0
01 01 01 9.4 638-3 70.3 70.3 0.87 0.00 15.0
'14.I 1A 1
01 01 01 11.0418-3 0.87 0.00 15.0
01 01 01 12.618E-3 75.1 15.1 0.87 0.00 15.0

01 01 01 14.195E-3 80.4 80. 4 0.8? 0.00 15.0

1 (SOOKY'STHESIS) DATA SET


TOEBES+SOOKY
2
3 F i l e n a m e ( A S S I G N E DF O R M E A N D R ) : sDsK403
q Plan geometry ( s I N U O SI T Y ) : 1.1 (GIVEN BY SOOKY}
5 Maln Channel X-sn : RECT 1.5" DE:P
Floodplain widc h : 3.886'1.1845: SL = 1.6E-3
'1
Floodplain roughness : Snoolh
I
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
10 m3/sec. RECORDED m PLOTTED mm

403

01 01 01 6.309E-3 1.60
01 01 0t ?.886E-3 6 1. 3 51.3 1.60 0.00 15.0
0l 01 01 9.4638-3 62.9 6 2. 9 1.60 0.00 15.0
01 01 01 11.041E-3 65.9 65.9 I.60 0.00 15.0
01 01 01 12.6188-3 64.2 68.2 1.60 0.00 15.0

01 01 01 14.1958-3 71.9 ?1 0 1.60 0.00 15.0

1 ?oEBES+SOOKY(S@KY'S THESIS) DATA SET

3 File name (ASSIGNED FOR MEANDR) : sDsK4o4


Plan geometry ( S I N U O SI T Y ) : 1.1 (GIVEN BY SOOKY)
5 Maln Channel X-sn : RECT 1. 5" DESP
Floodplain vidth : 3.886' 1.1845n SL - 3.6?E-3
7 Floodplaln roughness : smooth
8
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
'
10 m3/sec. R E C o R D E Dm PLOTTED mm C

404 6

01 01 01 6.3098-3 54.5 54.5 3.61 0.o0 15.0


01 0L 01 7.8868-3 s5.1 3.67 0.00 15.0
q<
01 01 01 9.463E-3 56.9 o 3.61 0.00 15.0
01 01 01 t1.04lE-3 59.0 3.6'1 0.00 15 . 0
o1 0t 01 12.6188-3 6 1. 8 61.8 3.6'l 0-00 1s.0

01 01 01 14.1958-3 62.1 oz- t 3.61 0.00 15.0


1 T O E B E S + S O O K Y( S O O K Y ' S T H E S I S } DATA SET

F l l e n a m e ( A S S I G N E DF O R M E A N D R ' : sDsK40 5
4 Plan geometry (SINUOSITY) : 1.1 (GMN BY SOOKY)
5 Maln Channel x-sn : RECT 3.0" DEEP
6 Floodplaln uldt.h : 3.8E6' 1.1845m SL - O.3E-3

? Floodplaln roughness : smooth


I
9 DATE DISCHARGE DEPTI{ AS DEPTH AS STOPE TAILGATE TEMP
10 m3/sec. R E C O R D E Dm PLOTTED mm

405

01 01 01 4.100E-3 88.6 88.6 0. 30 0.00 1 5 .0


0l 01 01 4.'r32E-3 tl.J 91.3 0. 30 0.00 1 5 .0
01 01 01 6.309E-3 98.6 98. 6 0. 30 0.00 15.0
01 0l 01 9.463E-3 I07.9 107.9 0.30 0.00 15.0
01 01 01 12.6188-3 i14.3 114.3 0.30 0.00 1 5 .0

1 T O E B E S + S O O K {YS O O K Y ' S T H E S I S ) D A T A S E T

3 F I I e n a n e ( A S S I G N E DF O R M E A N D R } : sDsK4o6
Plan qeomegry (SINUOSITY) : 1.1 (GIVEN BY SOOKY}
Main channel X-sn : RECT 3.0" DEEP
6 Floodplaln width : 3.885' 1.1845m SL - 0.575E-3

Floodplain roughness : SilooLh


8
9 DATE DTSCHARGE DEPTH AS DEPTH AS SLOPE TAITGATE TEMP
10 n3lsec. R E C O R D a Dn n PLOTTED mm
'l
406

01 01 01 4.1328-3 88.4 88.4 0.6?5 0.00 15.0


01 01 01 6.3098-3 94.1 9 4- 1 0.6?5 0.00 15.0
01 01 01 7.8868-3 o9 q 98. 9 0.675 0.00 15.0
0l 01 01 9.453E-3 103.8 103.8 0.675 0.00 15.0
0r.01 01 11.041E-3 106-4 106.4 0.675 0.00 15.0

0l 01 01 12.6188-3 10?.6 107. 6 0.6?5 0.00 1 5 .0


01 01 01 14.195E-3 109.0 109.0 0.675 0.00 15.0

T O E B E S + S O O K (YS O O K Y ' S T H E S I S ) D A T A S E T

FlIe nane (ASSIGNED FOR HEANDR) s D s K 40 7


PIan geomeEry { S I N U O SI T Y ) 1.1 {GTVEN BY SOOKY.}
Main channel X-sn RECT 3.0'DEEP
Floodplain wldth 3.886' 1.1845m SL = 0.87E-3
7 Floodplain roughness Smooth
8
9 DATE DISCHARGE DEPTH AS DEPTH AS STOPE TAI LGATE TgMP
lo n3/sec. REcoRDeD m PLOTTED mm

40? 7

01 01 0l 4.7328-3 86.5 86.5 0.8? 0.00 15.0


01 01 01 6.309E-3 9 2- 5 92.5 0.87 0.00 15.0
'7.886E-3 0.00 15.0
01 01 01 96.5 96.5 0.87
01 01 01 9.463E-3 100.7 1 0 0 .7 0.87 0.00 15-0
01 01 01 11.041E-3 to{.2 104.2 0.8? 0.00 15.0

01 01 01. 12.618E-3 105.0 0.87 0.00 r - 5. 0


01 01 01 14.195E-3 t 06.2 0.8? 0.00 15.0

1 T O E B E S + S O O K Y( S O O K Y ' S T H E S I S ) DATA SET

J FiIE NAME (ASSIGNED FOR MEANDR) : SDSK4OS


4 Plan geomerry (SINUOSITY) : 1.1 {GMN BY SOOKY}
Main channel x-sn : RECT 3.0" DEEP
6 Floodplain wldrh : 3.885' 1.1845n SL = 1.08-3
1 Floodplaln roughness : snooth
I
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE ?AIIGATE TEMP
10 n3/sec. R E C o R D E Dm PIOTTED mm

408

01 01 01 4.416E-3 6Z.A 82.8 1.00 0.00 1 5 .0


01 0l 01 4.?328-3 85.3 85.3 1.00 0.00 15.0
01 01 01 6.3098-3 9C.3 90. 3 1.00 0.00 15.0
01 01 01 9.4538-3 9A.1 9 A- 1 1.00 0.00 1 5 .0
01 01 01 12.618E-3 I04.0 104.0 1.00 0.00 15.0
1 T O T B E S + S O O K Y( S O O K Y ' S T H E S I S } D A T A S E T
2
3 FlIe name (ASSIGNED FOR MEANDR) : sDsK409
4 Plan geometry (SINUOS ITY} : 1.1 (GIVEN BY SOOKY)
Main Channel X-sn : RECT 3.0'DEEP
6 Fl.oodplaln width : 3.886' 1.1845m SL = 1.68-3
Floodplaln roughness : Smooth
8
9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP
n3/sec. R E C O R D E Dn u t r PLOTTED mm

409 o

qq
01 01 01 6.3098-3 88. 1 1 .60 0.00 15.0
01 01 01. 7.886E-3 91.6 91.6 .60 0.00 15.0
01 01 01 9.4 63E-3 94.1 94.7 .50 0.00 15.0
01 0t 01 11.0418-3 97.2 9 7- 2 .60 0.00 15.0
01 01 01 12.618E-3 9 9. 4 99.4 .60 0.00 15.0

01 01 01 1 4 . 19 5 E - 3 99-9 9 9. 9 .60 0.00 15.0

I (SOOKY.STHESIS) DATA SET


TOEBTS+SOOKY
)
FlIe n a m e ( A S S I G N E DF O R M E A N D R ) : SDSK410
plan geometry (siNuosrry) : 1.1 (GIVEN BY sooKy)
Main Channel X-sn : RECT 3.0" DESP
Floodplain widrh : 3.886, 1.1845m SL = 3.08-3
1 FioodpLain roughness : Smooth
I

9 DATE DISCHARGE DEPTH AS DEPTH AS SLOPE TAILGATE TEMP


10 m3/sec. R E C O R D E Dm PLOTTED mm a

410 5

01 01 01 6.940E-3 83.8 83.8 3.00 0.00 15.0


01 01 01 7.886E-3 86.1 86,I 3.00 0.00 15.0
01 01 01 9.4638-3 8 9. 6 89. 6 3.00 0.00 15.0
01 01 01 12.6188-3 95.1 95.1 3.00 0.00 15.0
01 01 01 14,1958-3 9?.0 9?.0 3.00 0.00 15.0

I T O E B E S + S O O K (YS O O K Y ' S T H E S I S ) D A T A S E T
2
3 F i I e n a m e ( A S S I G N E DF O R M E A N D R ) : sDSK411
4 PIan geometry (srNUosITY) : 1.1 (cIVEN BY SOoKY)
5 Main Channel X-sn : RECT 3. 0" DEEP
Floodplain widrh : 3.886'1.1845n SL = 3.6'7S-3
'l
Floodplain roughness : Smooth
8
a DATS DISCHARCE SLOPE TAILGATE TEMP
DEPTHAS DEPTH AS
10 m3/sec. RECORDEDm PLOTTED mm C

41"1

01 0t 01 7.886E-3 85.4 85.4 5.0t 0.00 15.0


o1 01 01 9.4 63E-3 88.? 88.7 3 .67 0.00 15.0
o1 01 01 11.0418-3 90.6 90. 6 J.bl 0.00 15.0
o1 01 1< n
01 12.6188-3 93.0 93.0 3 .67
01 01 01 14.19sE-3 94.3 9 4- 3 164 0.00 15.0
Appendix5

Summaryof the AckersRod


Roughnessmethod

sR 329 l7lo393
Appendix 5 Summary of the Ackers Rod
Roughnessmethod
Basicresistancecalculationsfor rod roughnessas in the Wallingfordtestsmay
be baseduponthe followingset of formulae,whichallowfor differentnumbers
of rods in ahematerows:

F, = (1-n,zdA)-z

F" = (1-nrzd/A)'2

For1.75<Z<6.6:

oCo = 1.184- 0.277Z +',/1o.szs


Z'0.843)

elseaOo = 0.95

lA/fs =2.ollog(RefJfj - 1.38

fro, FrN.)dzlP + (p, + pJfr/2


=8gRSA/2=4oCo(FrN,+

where

Re = Reynoldsnumberof blockedchannel= 2 V R 1{8,+{B.yv


Fl,, = blockageeffect,i.e. squareof area ratiosfor altematerows
Dr,e = numberof rodsof dia d acrosszone rows1 and 2
Nl,. = numberof rodsper unit lengthof zone,rows1 and 2
t = depthof flow
A = crosssectionalarea of zone underconsideration
fs = frictionfactordue to smoothboundary
fror = overallfrictionfactor
V = nominalvelocitygiven by componentdischarge/A
cco = effectivedrag coefficientof rods
L =zJd
R = hydraulicmeandepth NP, tor zone underconsideration
S = hydraulicgradient(watersurfaceslope)

The verticalrodswere mountedin woodenframes(2.46mby 2.33m)to atlow


them to be lifted in and out of the flume. During PhaseA the frames were
placedin the flumewith lheir shottdimensionperpendicular to the flow and
during phase B the frameswere usedwith the long dimensionperpendicular
to the flow direction,FigureA5.1.

Thusthe appropriatevaluesof n and N etc are slightlydifferentfor PhaseA


and B.

PhaseA
Zone width = 2.33m
Lengthof frame -- 2.4Em
lateralrod spacing = 0.315m
longitudinalrodspacing = 0.537m
nr., = 8, 7
N,z = nt, / longitudinal
rod spacing
rod diameter= O.O25m

sR 329 17r'0993
Phase B
Zone width = 2.46m
Length of frame = 2.33m
lateral rod spacing = 0.537m
longiludinalrod spacing = 0.315m
D12 = 5,5
Nr,e= n,,r/ longitudinal
rod sPacing
rod diameter= 0.025m

Thusthe Ackersprocedurewas appliedto a singleframewhencomputingrod


roughnessfrictionfactors. The methodwas developedbasedon calibration
datacollectedduringphaseA and it wasconsiderednecessaryto carryout an
independent checkon the methodfor the phaseB orientation
of the frames.

In orderto carryoutan independent assessment of the Ackersrod roughness


methodtwo extra sets of stage dischargetests were carriedout as pail of
HR's internalresearchprogramme.The FCF had alreadybeen infilledand
etfectivelyturnedin to a rectangularchannelwith width 10m. The portable
side wallsvrlereusedto narrowthe flumeand the extratestswere canied out
in a trapezoidalchannelwith bottomwidth4.6078mand side slopesof 1:1.
The longitudinal bed slopewas assumedto be the PhaseB 110 flood plain
slopeof 1.021x1O3.

Two setsof stagedischargeresultsweremeasuredandthe importantaspects


of eachare listedbelow:

849 Stage Dischargeresultsin trapezoidalchannelwith the roughness


framesorientedas duringPhaseA.

850 Stage Dischargeresultsin trapezoidalchannelwith the roughness


framesorientedas duringPhaseB.

Seven individualvalues of stage and dischargewere measuredfor each


roughnessconfiguration.Thesetwo setsof data are listedin TableA5.1.

The stagedischargedata were analyzedin two ways:

1 Valuesof frictionfactorwerebackcalculatedfromthe measuredflowsand


the knownchannelgeometryand comparedwiththe valuesobtainedfrom
the Ackersrod roughnessmethod. The resultsof this analysisare shown
in TableA5.2

2 The Ackersrod roughnessmethodwas used to calculatethe flow in the


channelandthis was comparedwiththe measuredflows. The resuftsof
this analysisare shownin TableA5.3.

The cafculatedDarcy friction factors are in error by between -9.47to and


23.80%.The data pointwhichgavethis largeerrorwas judgedto be suspect
and the meanerrorwas calculatedfor PhaseB caseboth includingthis point
and omittingit. The meanerrorsin the calculatedfrictionfactors were 2.9y"
for the phaseA case and4.51"for the phaseB case. As can be seen the
Ackers methodgave mean errors in dischargeof O.7%and '2.O"hfor the
PhaseA and B roughnesspatternsrespectively.The standarddeviationsfor
these resultsare 4.87oand2.9"/o.The fairlywide rangeof errorsis probably

sR 32{' r7l03/93
due to the fact that a wider tolerance was allowed on the measured water
sudace slopes in these measurementsthan during either Phase A or B.

The Ackers rod roughness method has been tested against two independent
sets of stage discharge data and reproduced the measured discharges and
total friction factors to an acceptable level of accuracy. Hence the Ackers rod
roughnessmethodmay be used in allfuture analysisof rod roughenedSERC
FCF data.

sR 329 17103193
Table A5.1 Stage Discharge Measurements

Date Discharge Depth Temp


cumecs mm "C

B49 Phase A Roughness

030192 0.02252 29.64 11.1


030192 0.04432 50.57 10.9
420192 0.06183 65.08 10.5
050192 0.07971 83.78 11.6
020192 0.10315 109.25 10.8
050192 0.13002 135.52 12.1
030192 0.16352 1 6 3 . 4 31 1 . 4

B50 Phase B Roughness

100192 o.02225 28.24 12.3


090192 0.04134 45.42 13.1
080192 0.06159 62.00 12.9
080192 0.07982 83.40 13.1
080192 0.10015 1 0 1 . 9 81 3 . 3
090192 0.13394 1 3 2 . 1 11 3 . 3
090192 0.16028 1 5 4 . 1 01 3 . 1

Notes
1 Depthsare adjustedfor flood plain slope 1.021x1O3

sR 329 r7l03v93
Table A5.2 Friction Factor Analysis

Discharge Depth ActualValue Calculated 7o Error in


cumecs mm Darcyf Darcy f Darcy f

PhaseA Roughness
0.02252 29.64 0.0874 o.0767 13.95

o.o4432 50.57 o.1121 0.1105 -1.43

0.06183 65.08 0.1230 o.1400 13.82

0.07971 8s.78 0.1579 o.1729 9.50

0.10315 109.25 0.2100 0.2101 0.05

0.13002 135.52 o.2523 0.2373 -5.95

0.16352 163.43 o.2798 0.2533 -9.47

MEAN% ERROR 2.92

PhaseB Roughness
0.02225 28.24 o.o774 o.0732 -5.43

0.04134 45.42 o.0934 0.0964 3.21

0.061592 62.00 o.1074 0.1326 23.80


'to.23
0.07982 83.40 0.1554 0.1713

0.10015 101.98 0.1808 0.1996 10.40

0.13394 132.11 o.2200 0.2341 6.41

0.16028 154.10 o.2M2 0.2494 2.13

MEAN % ERROR 7.25s

MEAN % ERROR 4.494

Notes

1 Y" Error= 1oo"(Calc- Meas)/Meas


2 Thisdatapointis out of sequenceand is suspect
3 MeanErrorincludingsuspectpoint
4 MeanErrorwith suspectpointomitted

sR 329 r7l0g93
TableA5.3 FIow Analysis

Discharge Depth Calculated 7" Error in


cumecs mm flow Calc Flow
PhaseA Roughness
o.02252 29.64 0.02405 6.78
o.M432 50.57 o.44467 0.80
0.06183 65.08 0.05795 -6.27

0.07971 83.78 o.07622 -4.38

0.10315 109.25 0.10304 -0.01

0.13002 135.52 0.13002 3.10


0.16352 163.43 0.16352 5.18
MEAN% ERROR o.74
PhaseB Roughness
o.02225 28.24 0.02289 2.88
0.04134 45.42 0.04070 -1.55

o.07982 83.40 0.07604 4.73


0.10015 101.98 0.09530 -4.84

0.13394 132.11 0.12989 -3.O2

0.16028 154.10 0.15867 -1.00

MEAN% ERROR -2.O52

Notes

1 % Error='100*(Calc-
Meas)/Meas
2 MeanErrorwith suspectpointomitted

sR 329 l7l0ry93
O o o

o o o

a a o

o o a

O o o

o o o

o o o

----+ Flowdirection PhaseA

AI Flowdirestion PhaseB
I

a Dowelrod,25mmdiameter

FigureA5.1 Schematicof rod roughnessframe

You might also like