Channel Protection Gabion Mattresses and Concrete Blocks: M Escarameia

Download as pdf or txt
Download as pdf or txt
You are on page 1of 58

ChannelProtection

GabionMattressesand ConcreteBlocks

M Escarameia

ReportSR 427
July 1995

I' -

- HR Wallingford

Address and RegLsteredOtfice: HR Watlingford Ltd. Hou/bery Park, Wallinglord, O<on OXl0 8BA
Tel: + 44 (0)1491 835381 Fax:.+ 44 (0)1491 832233

R.{idctcd in EnCad Nq 2561@9. HB W.llneford h s wfiolty ffi.d {bdd.ry ol HR W.lilgtod Gt@p Lld.

sq12l Oalo7fits
tr
Contract
This report describes work commissionedby the Departmentof the
Environment (DOE)under ResearchContractPECD7161290 for whichthe
nominated officerswereMr. P B Woodhead for DOEand Dr W R Whitefor HR
Wallingford. The HR job numberwas RTS 36. The repod is publishedon
behalfof the Depailmentof the Environment, but any opinionsexpressedin
thoseof thefundingDepadment.
this reportare not necessarily Theworkwas
carriedout by Ms M Escarameiaand by Ms G Lundstrom,and the projectwas
managedbyMrRWPMay.

Preparedby flr*Ae^t*-*r*--
'1name;

Approvedby Qvt?, lut++


.,aJ Re;x^*}" vvad!r-/vw-!-

Date

@ HR WallingfordLimited1995

sa427 31tO7tSS
tr
Summary

ChannelProtection
GabionMatlressesand ConcreteBlocks

M Escarameia

ReportSR 427
July 1995

Prior to the present study, HR Wallingfordwas commissionedby the


Department of the Environment to investigatethe stabilityof riprapand loose
concreteblocksfor channelprotectionin highlyturbulentflows. The resultsof
that studyare describedin HR ReponSR 313 (Escarameia and May, 1992).
HR was later commissioned to carry out a follow-upproiectwhich can be
dividedin two parts: 1. field measurements of turbulencein rivers,and 2.
laboratorytestsof gabionmattressesand cabledconcreteblocks. The first
partof thestudywas presentedin HR ReportSR 424(Escarameia et al, 1995)
and the secondpart is the subjectof the presentreport.

Tests were carriedout with loose solid concreteblocks,scale modelsof


300mmthickRenomattresses (1:5scale),and cabledand interlocking blocks
bothat a scaleof 1:8. The materialswereplacedon a flat bed and testedfor
a widerangeof flow conditions,from normalriverflow to very high levelsof
turbulence.

The resultsof the testswereanalysedwiththe sameproceduresused in HR


ReportSR 313for the analysisof riprapand looseblocks.Designequations
incorporatinga coefficientto account for the level of turbulenceare
recommended for protectionschemesusingReno mattresses,loose blocks
and cabledblockmattresses.Renomattresses werefoundto be considerably
more stablethan equivalentdprap:for example,under normaltutbulence
the requiredsizeof the fillingstonein a 300mmthickRenomattress
condilions
is approximately halfthe sizeof riprapneededunderthesameflowconditions.
The studyalsoshowedthatthe levelof tensionimposedby the cablesin block
mat revetments does not appearto havea significanteffecton theirstability.

sa./t um7l95
tr
Notation

a,b Sizeof meshopening


C Stabilitycoefficient
Dn Thicknessof blocksor sizeof the equivalentcube
dso Dimensionof stone below which 5O%of the stonesby
weightare smaller
e Voidsratio
Fr Froudenumber
g Acceleration due to gravitY
Kh Depthfactorin Pilarczyk's equation(1990)
Ks Slopefactorin Pilarczyk's equation(1990)
KT Turbulencefactorin Pilarczyk'sequation(1990)
kr equation
Coefficientin Pilarczyk's (1990)
lt,l2,ls Celldimensions
n Porosity
s Specificgravityof revetmentmaterial

rm: x')
TI Turbulenceintensity (Tl =
X

U Meanflow velocityovercross-section
Ud Depth-averaged velocity
u Streamwise velocitycomponent
Vs Volumeof solids
Vv Volumeof voids
Vb Meanvelocitynearthe bed
v Transversevelocitycomponent
w Vedicalvelocity component
Yo FlowdePth
cr Angleof bankto horizontal
A Relativedensityof mattress
e lnternalfrictionangleof stone
Ps DensitYof mattressmaterial
pw Densityof water
0 equation(1990)
Stabilityfactorin PilarczylCs
Vcr Critical shear stress parameterin PilarczyKsequation
(1sso)
x Time-averagedvalueof quantityx
x' Fluctuationaround the mean value of quantity x

sR 427 3l/07/95
tr
Contents
Page

Titlepage
Contract
Summary
Notation
Contents

Introduction.... 1

Materialstested 2
2.1 Generalconsiderations 2
2.2 Gabionmattresses 2
2.2.1 *isting information 2
2.2.2 Descriptionof maftressesfesfed 4
2.3 Blocks 6

Experimentalset-up 6
3.1 Test rig 6
3.2 lnstrumentation.. 7
3.3 Test procedure . . 8

Background information 8
4.1 General I
4.2 Turbulence... 8
4.3 HR'sdesignequations 9

Tests 10
5.1 General ....10
5.2 Renomattresses. .. 10
5.3 Looseconcreteblocks ....10
5.4 Cabledblocks 11
5.5 Interlockingblocks .......11

6 Dataanalysis... ...12
6.1 Model-prototypeconversion.. ...12
6.2 Analysisof tests with Reno mattresses . . 12
6.3 Analysisof tests with looseconcreteblocks . . . 13
6.4 Analysisof tests with cabledblocks . . . . 13
6.5 Analysisof tests with interlockingblocks . . . . . . 14

Conclusionsandrecommendations ...15
7.1 General 15
7.2 Renomattresses. ..15
7.3 Loosesolidconcreteblocks .....16
7.4 Cabledblocks .....16
7.5 blocks
Interlocking . . 17

8 Acknowfedgements ......17

I References . . 18

sR rt27 11/0S95
tr
Contents continued
Tables
Table1 Characteristicsof the Renomattresses tested
Table2 Renomattresses - Highturbulencelests
Table3 Renomattresses - Rapidflowtests
Table4 Looseconcreteblocks- Highturbulencetests
Table 5 Cabled blocks - Levelof tension A - Highturbulence
Table 6 Cabled blocks - Level of tension A - Rapidflow
Table 7 Cabled blocks - Level of tension B - Hightulbulence
Table 8 Cabled blocks - Level of tension B - Rapidflow
Table9 Cabledblocks- Not tensioned
Table10 Interlockingblocks(ARMORLOC) - Highturbulencetests
Table11 -
Reno mattresses Modeland prototypevaluesfor critical
movementcondition
Table12 Cabledand interlocking blocks- Modeland prototypevalues
for failureconditions

Figures
Figure1 Geometriccharacteristics of Renomatlresses
Figure2 Gradingcurveof fillingstonein the Renomattresses
Figure3 Geometriccharacteristics of cabledand interlocking
blocks
tested
Figure4 Generallayoutof test rig
Figure5 betweenC andthetubulenceintensityfor Reno
Flelationship
mattressescomparedwith riprap
Figure6 RelationshipbetweenC and the turbulenceintensityfor
concreteblocks

Plates
Plate1 Assemblingof Renomattresses
Plate2 Renomattresses placedin the test flume
Plate3 Renomattressesin flumeshowingpaintedbands
Plate4 Renomattresses aftera test showingmovementof stone
Plate5 Cabledblockswithframe

saln ulo7ls6
tr
1 lntroduction
In the periodof 1990to 1992an experimental studywas carriedout by HR
Wallingford the stabilityof riprapandconcreteblocksfor channel
to investigate
protectionin highlyturbulentflows;the resultsare describedin HR ReportSR
313 (see Escarameia and May, 1992).Tests were canied out with various
degreesof turbulencerangingfrom normallevelsin uniformchannelsto very
high levels downstreamof hydraulicjumps. Based on the laboratorywork,
design equationswere developedto size riprapand loose concreteblocks;
these look quantiiative account of the destabilizingeffect of the flow
turbulence,amongstother relevantparameters.The study had been partly
fundedby the Department of the Environment (DOE)who latercommissioned
a follow-upprojectwith two objectives:1. to collectfield data on typicallevels
of turbulencein riversfor use in the designequationsproposedin the first
study(firstpartof the project);and 2. to extendthe rangeof channelprotection
materialsto gabionmattresses andcabledconcreteblocks(secondpad).The
firstpartof the projectwas describedin HR ReponSR 424 (seeEscarameia
and May,1995).The presentreportdealswiththe laboratory testscaniedout
on gabionmattressesand concreteblocksand can be consideredas an
extension of HR ReportSR 313,'1992.Thetestswithgabionmattresses were
suppofiedby MaccafeniS.P.A.

The above mentionedHR ReportSR 313 gives some generalbackgound


information and on the initiationof pafiiclemovement.lt also
on turlculence
includesan extensiveliteraturereviewof design formulaefor riprapand
concreteblocksas channelprotection materials. Parlicularattentionwasgiven
to investigatingtheir suitabilityas revetmentsdownstreamof hydraulic
structures whereflowturbulence playsan impoilantrole.The literaturesearch
showed that the existingguidelinesdo not apply satisfactorilyto highly
turbulentflowsandthatthevariousdesignequationsgivewidelyvaryingsizes
for riprap.lt alsoshowedthatverylittleindependent wort hadbeencarriedout
to quantifythe per{ormance of revetmentsconsistingof concreteblocks(either
looseor cabled).

The methodfor analysingthe resultsof the presentstudycloselyfollowedthe


procedures newequations
adoptedinthe earlierstudy.Ratherthandeveloping
for the sizingof gabionmattresses andcabledconcreteblocks,the datawere
used to determineadequatestabilitycoefficientsfor each type of channel
protection.

Section2 of this report gives a descriptionof the materialstested and


summarizesthe available infornrationin the existing literature.The
experimental set-upand instrumenlation used in the tests are presentedin
Section3, and somebackground information usedin the analysisof the data
is givenin Section4. Section5 presentsthe test resultsand Section6 deals
withthe dataanalysis.The conclusionsand recommendations fromthe study
are givenin Section7.

sal?' u/0'llg6
tr
2 Materials tested
2.1 Generalconsiderations
Two types of channelrevetmentwere selectedfor investigation underthe
presentstudy:gabionmattressesand cabledconcreteblocks.Bothtypeshave
somecommoncharacteristics whichrnakethemsuitableforcontrolling erosion
andmattress-like
in riversandchannels.Becauseof theirflexibility shape,both
materialsallow coverageof banks and river beds with relativeease and
economyof installation;providedthat a regulatinglayeris laid as a basefor
the mattress,the contourof the channelcan be followedsmoothly,resulting
in a continuousrevetmentwith uniformthickness.Adequatefixing of the
mattresses of fiherlayersare considered
at the endsand installation to be of
paramountimportancelo ensurea goodperformanceof the revetment.

2.2 Gabionmattresses
Gabion mattressesare essentiallyrock-filledwire mesh boxes which have a
large surfacearea comparedto their thickness.They retainthe advantagesof
riprap in that they constitute a flexible, permeable revetment. In addition to
that, the flexibilityof the mesh boxes allows them to deform when subiected
to current or wave forces without failing, while the rockfillinside the boxes is
containedby the mesh. Failureof a revetmentcan be defined as a situation
where erosionof the underlyingmaterialoccurs. In the specificcase of gabion
mattressesthis is likely to happen when the movement of the filling stone
inside the mesh box is such that lhe stone depth is reduced to the medium
size of the rock. For the present tests, failure conditions were defined as
correspondingto incipientmovementof the stone inside the boxes.

In river engineeringlhe use of gabion mattressescan be found in a number


of situationswhich include,for example,river bed and bank protectionagainst
erosioncaused by current and wave attack,and local protectionagainst scour
around bridge piers and downstream of weirs, gates and culveds. For the
presentstudy it was decided to investigatethe stabilityof a proprietarybrand
of gabion mattresses which has a large record of applicationsworldwide,
including the UK. The revetment chosen is called Reno mattress and is
manufactured by Officini Maccaferri S.p.A., ltaly. These mattresses are
manufacturedwith a hexagonal,doubletwistedmesh made of zinc coated mild
steel wire; in conosive environmentsthe wire can be coated with PVC to
increase its resistance to chemical attack. Each mattress unit is divided into
severalcells which have the purpose of restrictingthe movementof the stone
fill within a smaller area and thus increasing the overall strength of the
mattress.On site the individualcells are usuallyassembledtogether,then the
mattressesare placed in positionand wired with adjacent mattresses before
they are filled with stone; once the stone is placed in the wire boxes the lids
of the mattresses are positioned on top and also wired down. When the
mattressesare laid underwaterthey are assembledbefore positioning.

2.2.1 Existinginformation
Severalresearchorganizationshave in the past been commissionedby
Maccaferrito carryout researchon the stabilityof Renomattresses. These
studieshave concentrated on assessingtheirperformance underwaveand
currentattackbut informationis lackingon stabilityin highlyturbulentflow
conditions.

Oneof the mostcomprehensive of thesestudieswas carriedout by Simons


et al (1984)at the EngineeringResearchCenter of the ColoradoState
University,USA.The mainobjectivesof the studywere: 1. to measurethe
hydraulicroughnessof the Reno mattresses;2. to assessrequirementsof

sBln ulo7l95
tr
granularand syntheticfilterlayers;and 3. to developsuitabledesigncriteria
undervariouscurrentflowconditions.
for stabilityof the mattresses Bothmodel
scaleand full-sizetestswereconductedin this study.

ln the 1:3scaletestsa commercial meshwas usedto reproduce the mattress


boxesbut dynamicsimilaritywas notachieved,resultingin a relativelygreater
of variousthicknesses
flexibility.The studyinvolvedthe testingof mattresses
and valuesof the criticalvelocitywere determined.This is definedas the
velocityof the flow at whichthe stonesinsidelhe mattressesbeginto move
in the maindirectionof the flow.Formattressescorresponding in the prototype
to 300mmthickness,the criticalvelocities obtainedwere4.97m/s(for Froude
numbersmallerthan1.5)and4.18m/s(forFroudenumbergreaterlhan 3).As
the flow velocityincreased,a considerable numberof stoneswould move
insidethe rnattress cellsfromthe upstreamto the downstream partof the cell.
This massof rockwouldeventually causethe meshto deformand ripplingof
the sudacewouldoccur.lt was foundthat if the reduceddepthof rock at the
upstreamend of the mattresscellswas biggerthan the mediansize of the
stone,the underlyinglayerwouldnot be eroded.The flow velocityfor this
situationwas calledthe limitingvelocity.In the Coloradoteststhe maximum
velocityachievedcorresponded to about6m/s,but evenat thisvelocityfailure
of the mattresses was not obserued.

Thefull-sizetestsused150and 225mmthickmattresses andwerecarriedout


in a flumewith a slopeof 13%.Considerable difficultiesin the measurement
of theflowdepthwere,however,encountered dueto theturbulencegenerated
by the highspeedflow overthe steeplylaid mattresses.

Other conclusionsfrom the Coloradostudy, besidesthe criticalvelocities


mentioned, were:1. the incipientmotionof the fillingstonein the mattresses
occursat about 1.4 timesthe flow velocityand twice the shearstressthat
causeincipientmotionfor looserockof the samesize;2. the mattressstability
dependsto a largeextenton the celllenglhs;3. the meshhasan insignificant
effecton the mattressroughness.

Moregenerally,Pilarczyk(1990)presentedan equationfor designof riprap


and mattressesusedin channelprotection. This equationfeaturesa number
amongstwhichone that takesqualitativeaccountof
of differentcoefficients,
the turbulencelevel of the flow. Using the coefficientsappropriatefor
mattresses,Pilarczyk'sequationis applicableto both concreteand gabion
mattressesand can be writtenas follows:

D n = ( 0 / A ) ( 0 . 0 3 5 / v " r ) K r K h K s - 1l U o z t Z g J (1)

where:

Dn is the averagethicknessof the mattress;

Qis a stabilityfactor
1.0for exposededgesof mattresses
0.5 for continuousprotection;

A is the relativedensityof the mattress


A = ( 1 - n )( p . - p * ) / p , n
P" and pn are the densityof the mattressmaterialand of the fluid,
respectively;
n is the porosity;

V", is the criticalshearstressparameter(= 0.06 - 0.10for mattresses);

SF /t27 3l/O7195
tr
K, is the turbulencefactor
213tor low turbulence,uniformflow
1.0 for normalturbulence in rivers
2.Otor high turlculence,local disturbancesand outer bends of rivers.This
value should only be used when the velocityused in the equationis the
average velocity for the whole cross-section(U), instead of the local
depth averagedvelocity (U6);

Kn is the depth factor


"
Kh =2| (log (12y0 lltil't2 for logarithmicvelocityprofile
k, = Dn for concrete blocks
k, = 2D^ for rock;
Kn = ( Vo/ D;) {'2 for partiallydevelopedvelocityprofile;
Yo= flow depth

K^
o is the slope factor
K" = (i - ( sin2cr/ sin2e;1o's
where cr is the angle of the bank to the horizontaland 0 is the internal
friction angle of the stone;

Uo is the depth-averagedflow velociV;

g is the accelerationdue to gravity.

As can be seen,this equationrequiresan estimateof the state of development


of the flow (whether it is partiallyor fully developed)and there is also some
unceilaintyregardingthe appropriatevalue of k, to use for gabion rnattresses.
However, this is still the most comprehensive equation for design of
mattressesidentifiedduring this study.

2.2.2 Descriptionof mattressesfesfed


Contactswere establishedwith Maccaferri, and moreprecisely,with one of
their UK agents,Riverand Sea GabionsLtd, to choosethe type of Reno
mattressto test in the laboratoryand decideon the necessarymodelscale.
Sincethe rangeof flow conditionscoveredby the tests would span from
normalriverflowto very high levelsof turbulence,it was decidedto simulate
0.300m thick mattresseswhich are amongst the thickest mattresses
manufactured by Maccaferri.
Considerationsregarding themaximumflowsand
velocitiesthatcouldbe achievedin the testrig ledto the adoptionof a model
scaleof 1:5.The requiredsizesof the mattregses, wiremeshandfillingstone
werethendetermined.

The scaled mesh boxes were fabricatedby Maccaferriin ltaly. The test
sectionin the laboratoryflume requiredtwo mattressesof the following
dimensions: 1m x 1m x 0.06m(length,widthandthickness); and one slightly
shorter,only0.80mlong.Althoughthetestsectionwas 1.2mwide,the scaled
mattressesonly measuredapproximately1m in width to facilitatetheir
positioning in the flume.The 0.2mgapwassetto one sideof the channeland
filledwith purposemade mattressunitsafterthe main unitswere placedin
position.Eachmattresswas dividedintoten cells0.50m(ll) by 0.20m(lr) in
planwiththeirbiggerdimensiontransverseto the directionof the flow. The
modeldimensions of the mesh,a and b, were19 and 20mm,respectively.A
sketchof the meshboxesusedin lhe testsis presentedin Figure1.

Accordingto Maccaferri'sliteratureand to specificadvicereceivedfor this


project,the stonefillingfor a 0.300mRenomattressis normallyin the range
100 to 150mm.ln terms of duo(the size of stonebelowwhich5O%of the
stonesby weightare smaller),it willtypicallybe around0.125m.At a scaleof
1:5the modelstonewouldneedto havea valueof dro of 25mmand be fairly

SR.l?7 31rc7l95
tr
nanow-graded so thatthe smallersizeswouldbe containedwilhinthe mesh.
However,some difficultieswere encounteredbeforea suitablestone was
foundfor the tests.A first batchof stonewas rejectedon the basisthat its dro
wassubstantially belowthe requiredone;itsgranulometric curuealsoshowed
too widea rangeof stonesizes.The secondbatchorderedwas also checked
by HR'sSedimentation Laboratory and revealeda valueof droof 23mmwhich,
in prototypeterms,corresponds to 0.115m.The specificweightof the stone
was determinedand foundto be equalto 2680 kdm". This batchof stone,
whichwasangularin shape,wasconsidered adequate andwasthereforeused
to fillthe mattressboxes.The granulometric curueobtainedis shownin Figure
2 and the characteristics of the mattressesare summarized in Table1.

The operationof filling the mesh boxes followedcloselythe instructions


providedby Maccaferrias it was importantto reproducethe mattressesand
wiringoperations as accuratelyas possible.lt wassoonrealizedthattheyhad
to be assembledoutsidethe flume and then positionedin the test section
becausethe lacingwas a verytime consumingand delicateoperation.Since
the mattresses couldnot be allowedto deformduringtheirpositioningin the
flume,strongsteelbaseplateshadto be madeto enableeachmattressto be
liftedon a flat base.Eachof the three Renomattresses was thereforefilled
withstoneoutsidethe flume,wellcompacted andcoveredwiththe lids.These
weretighlyfixedwithcontinuous lacingwireat the perimeterof the mattresses
and at allthe transversaldiaphragms. As requestedby Maccaferri, a "quilt"
effectwas obtained,with the centresof lhe cellshavinga greaterthickness
than the borders.The lacingwire adoptedwas flexibletinnedcopperwire
similarin diameterto that used for the mesh (approximately 0.4mm)- see
Plate1. Oncein the flume,the three mattresses and the sectionsplacedon
the side gap werealso lacedtogetheras shownin Plate2.

Cementmortarwaspouredintothe upstreamanddownstream cellsto prevent


liftingof the mattressesat lhe ends. In orderto facilitatethe observationof
stonemovementwithina cell,red paintwas sprayedon the upstreamhalfof
each cell to form bands transverseto the flow (see Plate 3). At incipient
movementconditions,red paintedstoneswouldbe seento travelalongthe
cellto the non-paintedarea.

While the mattresseswere being assembledand filled,the rock that was


placedin each cell was carefullyweighedand the cell dimensionswere
recorded. of thevoidsratio,e, whichis definedas
Thisallowedthe calculation
follows:

e=Vv/Vs (21
whereV" is the volumeof solids(ie weightof stonedividedby its specific
gravity)and Vuis the volumeof voids(definedas the differencebetweenthe
volumeof the cell and the volumeof solids).The averagevalueof the voids
ratiowas foundlo be 0.65. In generaltermsthe biggerthe stonesize the
smallerthevoidsratiois likelyto be. Bearingthisin mindandthe factthatthe
averagestonesizewas 23mm,the valuefoundfor e is somewhatlowerthan
that determinedfor riprapin the previousstudyl(forexample,a voidsratioof
1.0wasobtainedfor stonesize9.3mm,see Esc{rameiaand May,1992).This
indicatesa relativelyhighcompaction which,acbording should
to Maccaferri,
also be achieved on site applications.

sR 427 3ll07/95
tr
2.3 Blocks
Mostof the tests involvedcabledand interlockingblocksbut somepreliminary
testswerecarriedout withscalemodelsof looseconcreteblocks.Theseloose
blocks,which were specificallymade for the study describedin Escarameia
and May (1992),were solid and rnadeof a cementmortarhavinga specific
weightof 2330k9/m5; theirdimensionswere0.030m(lengh)x 0.030m( width)
x 0.008m (thickness). The loose blocks were re-testedin order to check
conformitybetweenthe presentstudy and the earlierone.

The main tests were carriedout with modelsof cabled and of interlocking
blocks. Thesewere 1:8 scale modelsof the proprietarybrandArmorflex140
andArmorlocwhichare manufactured by MMGCivil EngineeringSystemsLtd.
The rnodelsused in the tests were existingblocks kindlyloanedby Dr. R.
Baker of Salford University.They had been carefullymade for a previous
researchstudywith a specificgravrtysimilarto that of concrete. The outside
dimensionsof the Armorflexmodelswere0.040m(length)x 0.036m(width)x
0.013m(thickness) and thoseof the Armorlocmodelswere 0.049m(length)
x 0.037m(width)x 0.011m(thickness).

The Armorflexblockmat consistsof rectangularconcreteblockswith vedical


holes;the blocksare tied togetherby means of horizontalcables that run
throughtwo cableductsin eachblock. An interlockingeffectis alsoachieved
since the blocksare cabledtogetherin a staggered fashion(see Figure3).
The blocksare slightlytaperedinwardsfrom near the bottomto the top, with
the objectiveof allowingbetter articulationof the mattress.The Armorflex
mattressesare suppliedwith the standarddimensionsof 6m x2.4m and are
usuallyinstalledwitha syntheticfilterlayerundemeath.A granularlayercan
also be placed over the mattresseswhich, apafi from the environmental
benefitsof vegetationgrowth,is claimedto increasetheir resistanceto wave
attack. In many cases the mattressesare blindedwith gravelto enhance
frictionbetweenthe blocks.The shape of the Armorflexblocks as they are
producednowadaysdifferssomewhatfromthat of the modelstested;they are
now slightly more rounded.However,this small differenceis unlikelyto
accountfor majorchangesin the stabilityof the blocks.

The Armorloc mats are formed by blocks that interlock with a dovetail
configurationand are suitedto areaswhere hand installationis requireddue
to lackof sufficientspaoefor machineryto operate.Theyare also usedwhen
the area to be protectedis small,such as near culverts,drainagechannels,
etc. The blockshavetwo verticalholesand a doubletaper whichallowsfree
articulation and the transverseplanes(seeFigure3).
in boththe longitudinal
As in the caseof the Armorflexmats,a syntheticfilter layer is usuallyplaced
beneaththe blocks,and blindingof the joints is normallycarriedout with fine
granularrnaterial.

3 Experimentalset-up
3,1 Test rig
The generallayoutof the test rig is shownin Figure4. As can be seenin the
figure,a 28m longby 2.4mwideflumefittedwith threepumpswas usedfor
the tests. The rnaximumflow achievablein the rig is 0.4 m"/s and tailwater
depthsare controlledby meansof a flap gate at the downstreamend of the
channel.The widthof the flumewas reducedfrom2Am to 1.2mto enablea
wider rangeof velocitiesand flow depths.This was achievedby installinga
partitionwallwhichdividedthe flumeintotwo parallelchannelsone of which
was blockedby a bulkhead.A smmth concreteweirwith an ogeecrestprofile
was designedand buihin the test channelwiththe objectiveof creatinga large

sF 427 ll/09/!95
tr
enoughpressurehead that would producehigh velocitiesat the base of the
weir. By liftingthe tailwatergate, a hydraulicjump could be formedwhich
would generatethe higher levels of turbulencerequiredfor the tests.
Dependingon the waterlevelsdowstream,the weir wouldalso createa head
of watersufficiently largeto producesupercritical
flow in the test sectionso
thatthe stabilityof the gabionmattressesand concreteblockscouldbe studied
in rapidflow conditions.

The model materialswere placed in a flat 2.8m long test section,2.5m


downstreamof the concreteweir;the transitionbetweenthe snpoth sudace
of the weirand the modelmaterialswas achievedby a fixedroughbed.This
consistedof gravel(approximately Smmin size)whichwas cementedto the
surfaceof the concreteslabthatformedthe bedof the channel.A 0.15mhigh
end sill was builtdownstreamof the test sectionin orderto fix the hydraulic
jumpwithinthe lengthof thetestsection.Holeswerecut at thetwo extremities
of the sill to facilitatedrainageof the test section.

Different
arrangements wererequiredfortestingthegabionmaftresses andthe
concreteblocksdueto the differentthicknesses and weightsof the two types
of modelmaterial.For reasonsthat were explainedin Section2, the model
gabionmattresses had to be assembledoutsidethe test rig, on strongsteel
plateswhichwerethen liftedintothe flumeby meansof an overheadgantry.
Oncein the flume,the steelplateswiththe modelgabionssat flat on the bed
withtheirtopflushwiththe levelof the upstreamfixedbed.Sincethe concrete
blockstested were of smallerthicknessthan the model gabions,it was
necessaryto place the blockson a false woodenfloor to ensurethat the
blockswere,as the gabions,flushwiththe upstreamfixedbed.

3.2 Instrumentation
Thetestrig is equippedwitha Crumpweirand accuratewaterlevelgaugefor
measurement of the flow dischargeat the downstreamend of the flume.A
calibrationof this weir was carriedout beforethe start of the tests. Checks
using a miniaturecurrent meter to measurethe mean flow velocity in
conjunction with measurements of the waterdepthin the test sectionwere
doneregularlyto assessthe performance of the Crumpweir.Theflow depths
in thetestsectionweredetermined by scalesfixedto the transparentsections
of the wallsof the flume.

Valuesof instantaneousflow velocitywere measuredby a two-component


electromagnetic currentmeter(Columbrock EMCM).The EMCMhasa 32mm
discusheadand can be positionedin eithera vefiicalor a horizontal planeto
suitthe typeof test beingcarriedout,as will be describedlaterin item3.3. In
the two possiblepositionsthe instrumentmeasuredthe flow velocityin two
orthogonal directions:
thesewerethestreamwise andthetransverse directions
when the head was in the horizontalplane,and the streamwiseand the
verticaldirectionswhenthe headwas in the verticalplane.Witha sampling
frequencyof 125H2,the EMCMalsocontainsa filterwhichhas a 3dB cut-off
al 2OHzand a flat reponsecharacteristicto ZHz.Beforethe beginningof the
tests,the two channelsof the electromagnetic probewerecalibratedin HR's
Meter RatingTank. In additionto this, duringthe test programmeseveral
checksof the meanstreamwisevelocitywere carriedout usinga miniature
currentmeter.Previousexperiencewith this type of electromagnetic probe
indicatedthat, due to the sensitivityof the instrumentto changes in
temperature,regularmonitoringof the offsetsignalsat zero flow velocitywas
required.Therefore,the probe'soffsetswere recordedat the beginningand
end of eachtest.

The point velocity measurementsfrom the two-componentprobe were


convefiedinto meanand fluctuatingvelocitiesby a MeanVoltageMeterand

sa427 311O7t96
tr
an RMS(rootmeansquare)Meterwith two channelseach.The rmsvalues
weredetermined by the difference signalfromthe
betweenthe instantaneous
probeandthevaluegivenby the MeanVoltageMeter.Bothvalueswereread
directlyfromthe meters'displays.

3.3 Testprocedure
Twotypesof flowconditionwereconsidered inthisstudy:normalchannelflow,
and flowwith higherlevelsof turbulencewhichwereproducedby a hydraulic
jumpat the baseof the weir.ln the normalturbulenceteststhe tailwalergate
was keptfullyopen,so thatthe downstream waterlevelsweresufficientlylow
to avoid formationof a hydraulicjump, and the flow rate was gradually
increasedto causeinitiationof particlemovement.

It was soon realizedthat very rapidflow conditionsneededto be achievedif


any movementwas to be observedin the tests of the gabionmattressesand
lateron duringthe testsof cabledblocks.In thesesituations the discusof the
probewas positionedin a horizontalplanesincethe smalldepthof the flow
wouldnototherwise allowmeasurements to be takenat the requiredlevel(see
Section4). At the beginningof eachhighturbulencetestthe tailgatewas set
quitehighandwasthengraduallylowereduntilmovement of thetestmaterials
was obserued.In theseteststhe probewas positionedsuchthatthe planeof
the discus was verticaland, as in the other type of test, at about 1m
downstream of the end of the fixedbed section.

4 Background information
4.1 General
Detailed information on turbulence in flows and on initiation of particle
movement was presented in the report that described the previous study of
channel protectionrevetmenls(see Escarameiaand May, 1992).That report
also includedan extensiveliteraturereview of existingformulaefor the design
of riprap and concreteblocks under current attack.This Sectionwill therefore
concentrate on presenting and defining the most relevant concepts and
parametersused in the analysisof the presenttest results.

4.2 Turbulence
Whenthe levelsof turbulencein the flow are highthe movementof the fluid
particlesis very erraticcausingrapidand sometimeslargechangesin the
velocitydirectionand intensity.lt is generallyacceptedto co6;ider the
instantaneous velocityvectorV as the sumof twoterms:the mean,V, andthe
turbulentfluctuationsaroundthe mean,V'. Thiscan be expressedas follows:

V=V+V' (3)

ln very turbulentflows the erraticnatureof the flow paths can causethe


instantaneous velocitycomponents,u, V and w in the three orthogonal
directions(streamwise,transverse to be of thesame
andveftical,respectively)
orderof magnitude.

The importance of the turbulentfluctuationsin relationto the time-averaged


meancan be assessedby a parametercalledthe turbulenceintensitywhich
is definedas followsfor the threevelocitycomponents:

Tlu =@11t2ttt (4)

sR127 SUOT|SS
tr
Ttu = jf,z;1270 (5)

Tlw = 6efart (6)

The numeratorsare commonlyknownas the rms values (root meansquare


values)and correspondto the standarddeviationfrom the mean.

4.3 HR's design equations


The earlierstudyrecommended designequations forthe stabilityof riprapand
of concreteblocksin flows with variouslevelsof turbulence.In the analysis
then carriedout a turbulencecoefficientwas definedin termsof the tubulence
intensityin the streamwisedirection,Tl u. The electromagnelic currentmeter
usedin the presentstudyallowedvelocityreadings to be taken not only in the
streamwise or vedical
directionbutalsoin the transverse direction,depending
on the way ihe probewas positionedin the flume.Althoughnot essentialfor
the analysis,the additionaldata on the non-streamwise componentswas
collectedand consideredas an impodantindication of the flow conditions.

It was foundthat the stabilityof channelrevetmentsdependsstronglyon the


value of the near-bottom velocityVo and on the turbulenceintensityat the
samelevel.A referencelevelcorresponding to 10%of the waterdepthabove
the bed was adoptedfor the followingreasons:1. it is sufficientlycloseto the
bed to be representative of flow conditionsexperiencedby the revetment;2.
it is less affectedby errors involvingthe vertical positioningof the velocity
measuringinstrumentthan levelscloserto the bed wherethe velocitygradient
is steeper;and 3. it is suitablefor measurements in physicalmodelsand on
site.

The designequationsdevelopedfromthe previousstudywereof the form:

Dn= C ( 1/ ( s -1))( V2b/ 29) (71

where:
Dn is the size of the equivalentcube (forstone)or the thicknessof the blocks;

C is a stabilitycoefficientthat varieswith tutbulence;

s is the ratio of the densityof the revetmentmaterialto the densityof the


water;and

g is the accelerationdue to gravity.

Forsafedesignof riprapon a flat bed and bankslopesof 1V:2Hor flatterihe


coefficientC recommended was:

c = 12.3Tl - O.2O (8)

where:

Tf = Tl u = Ilrls u / Vo al 1O"/"of waterdepthabovethe bed


and Tl )s"/". lf, for-example, the turbulenceintensilyis 10%,Tl shouldbe
enteredin equation(8) as 0.10.

sR127 0U1A95
tr
For safe design of loose concretebloks on gradientsflatter than 1V:2.5Hthe
coefficientC recommendedwas:

C = 0.75 (12.3 T1 - 0.20) (s)

For riprap,the size of the equivalentcube, Dn,isrelatedlo the Duovalue of the


stone by: Dn=0.90Ds0(see Escarameiaand May, 1992).

5 Iesfs
5.1 General
ln theteststhevelocityreadingsweretakenat approximately 10%of thewater
depthabovethe bedso thatlhe measurements couldbe analysedin thesame
way as the resultsof the previousstudy.The resultsare presentedin Tables
2 to 10.Readingsof the localmeanvelocityin two orthogonal directionswere
obtainedas wellas the rmsvaluesof the velocityfluctuations.As mentioned
in Section 3, in the high turbulencelests the two odhogonalvelocity
components recordedwerethestreamwise andthevedicalcomponents; in the
rapidflow teststhey were the streamwiseand the transversecomponents.
Shownin the tablesare also the turbulenceintensitiesat 10% of the water
depth,Tl, and Tl u (or Tl *), whichwerecalculatedwith equations(a) to (6).

Theflowdepth,y6,and ihe meancross-sectionalvelocity, U (calculated from


the ratioof flow dischargeto the cross-sectional area of the flow) are also
presentedin the tables. The right hand column of the tables contains
obseruations regardingthe stateof motionof the materialstested.ln the tests
wheresomemovementor collapseoccurredthe valueof the Froudenumber
is also presentedand was calculatedas follows:

Fr=ul(gyo)o'u (10)

due to gravity.
whereg is the acceleration

5.2 Renomattresses
Theresultsof testswithRenomattresses underhighturbulence and rapidflow
conditionsare presentedin Tables2 and 3, respeclively. lt can be seenthat
the failureof the mattresses was neverreachedin spiteof using the full flow
capacityof the test rig. Furthermore,il was not possibleto achieve conditions
for whichthe mattressmeshwasdeformed by the movement of the rockfilling
insidethe cells.The velocityassociatedwiththis conditionis calledthe limit
velocity.Rockingor smalltranslatorymovemenlsof stonesinsidethe cells
were,however,obserued;the meanflow velocitycorresponding to this latter
situationis calledthe criticalvelocity.Plate4 was taken at the end of one of
suchtestsand showsthe movementof some stones inside the mattress cells.

5.3 Loose concreteblocks


The loose blocks were laid by hand in a staggeredpattern on the bed of the
flume to form a continuous mattress. Sand with a duo size of 0.7mm was
spread over the maltressto fill any gaps betweenthe blocks.

Table 4 shows the resultsof the tests, all of which were carriedout under high
turbulenceconditions.Totalcollapseof the revetmentwas reachedin test CL8'
It was, as expected,very sudden since the movementof one block causedthe
rapid and progressivemovementof the others.This type of failure had already
been observed in previoustests with loose blocks.

10 sR 427 3l/07/95
tr
5.4 Gabled blocks
It was decidedto test the cabledbbcks withthreequalitativelevelsof tension:
level A, where the cables were tightly stretched;level B, where the cables
were stretchedbut to a lesserextent;and a third levelwhereno tensionwas
appliedto the cablesalthoughthe blocksweretied together.This latterlevel,
which correspondsto no tension, is rnore likely to represent prototype
conditionsfor Armoflex and Armorlocinstallationsas cable tension is not
expecledto lastlong,accoldingto the manufacturer.Furthermore, the relative
size of the cablesand cableductsis suchthat blockscan movebeforebeing
restrainedby the cables and this would be consideredas failure. A metal
frame surroundingthe blockswas madeto allow applicationof the required
tensionto the blockmattress.This frameconsistedof two thin platesto which
the ends of the cables were fixed, and two threaded rods that could be
adjustedto give the necessarytension(see Plate5).

Mostof the testswerecaniedout with the blockmattresspositionedsuchthat


the cableswere normaltothe mainflow direction;the effectof the orientation
of the blockswas investigatedby carryingout some tests with the cablesin
the flow direction.In all tests,sand with dro equalto 0.7mmwas spreadover
the mattressin order to fill the vedical holes of the blocks and the gaps
betweenadjacentblocks.An increasedgranularinterlockingeffect mightbe
achievedwith slightlybiggergrainsize.As a resuhof previousexperience,it
was decidednot to use any type of filter layer beneaththe mattress.The
previousstudy of channel revetmentshad shown that in highly turbulent
environments the presenceof a granularfilterlayercan actuallydestabilisethe
revetment;tests using a geotextilehad also shownthat the stabilitywas no
higherthan achievedwithouta filter layer.However,the manufacturers of the
Armorflexand Armorlocmattressesrecommendthe use of a geosynthetic
fabricundemeaththe blocksto preventmigrationof fine particlesthroughthe
holesin the blocks.

The resultsof the testswith cabledconcreteblocksare summarisedin Tables


6 to 9. With level of tension A, it can be seen that no movementwas
registeredin the highturbulencetestsandthat only movementof the upstream
row was observed in the rapid flow tests (Tables 6 and 8). Srnall lifting
movementsoccurredin the mid sectionof the mattressduringtestswith rapid
flow whenthe tensionof the cableswas at level B (see Table 8, test CRB2).
Total collpaseof the revetmentwas obseryedwhen the blocks were tied
togetherbut not tensioned,as can be seen in Table 9, test CH2.

5.5 lnterlockingblocks
For the testing of the interlockingblocks it was decided to apply some
constraintto the movementof the blocksat the upstreamand downstream
endsof the mattressin orderto bettersimulateprototypeconditions.This was
carriedout by placingwoodenbattenstranwerselyto the directionof the flow
at the two ends of the mat. No constraintwas providedat the sides of the
mattress.As in the othertests with concrete blocks,O.7mmsand was used
to fillthe holesand gapsof the mattress.

The resultsof the two highturbulence tests,are presentedin Table10 where


it can be seen that the collapsewas reachedin test lB2. This collapsewas,
as expected,quite suddenand causedtotal loss of the blocksas they were
washedaway by the flow.

11 SR,t27 l1l09/95
tr
6 Data analysis
6.1 Model-prototypeconversion
The testsof Renomattresses and cabledor interlocking blockswerecanied
out with scalemodelsof these materials.As mentionedin Section 2, the
geometric scaleswere1:5for the Renomattresses and 1:8for the blocks.ln
orderto convertmodelvaluesintotheircorresponding prototypevalues,it is
necessaryto adopta suitablesimilaritylaw. Sinceforcesdue to gravityand
inertiaare dominantfactorsin open-channel flows,it is appropriate
to adopt
the Froudiansimilaritylaw.Accordingto this lawthe modelvaluesshouldbe
multipliedby the followingconversionfactorsto givethe prototypevalues:

Geometricscale 1:8 1:5

Length,width, water depth 8 5

Velocity 8o'5= 2.828 So's= 2.236

Flow discharge 8 2 ' 5= 1 8 1 . 0 52'5= 55.90

Turbulenceintensity 1 1

6.2 Analysisof tests with Reno mattresses


The meanand near-bedflow velocitiesin testswherestonemovementwas
obseruedwere convertedinto prototypevalues using the relevantfactor
presentedin Section6.1 (seeTable11).Valuesof thecoefficient C werethen
determined from Equation (7), and plottedin Figure5 against
the turbulence
intensityin the streamwisedirectionwhichfrom now onwardswill be simply
calledTl. Since Reno matlressesare essentiallyriprapwith extra strength
providedby the wire mesh,data that had been obtainedfor riprapin the
previousstudyare also shownin the figureto allowcomparisons. lt can be
seenthatthe Renomattresses testedareconsiderably morestablethanriprap
of equivalent size ( notethat allthe experimenlalpointscorresponding to the
Reno mattressesare to the right of equivalentriprap points).lt is also
noticeablein Figure5 that in tests carriedout with "normal"tulbulence,
movementof the stone inside the mallressesoccuredat much higher
turbulenceintensities(above15%)than had beenobseryedfor riprap. This
is likelyto be due to relativelysmallerratiosof waterdepthto stonesize.

Whencomparedwith tests usingriprap,the normalturbulenceresultswith


Reno mattressesindicatethat for similarflow conditionsthe stablesize of
stonein the mattressesis approximately half that of looseriprap(compare
C=0.36for riprap with C=0.17for Reno mattresses). This is in good
agreementwith the resultsof the Coloradostudy (see Simonset al, 1984)
where it was found that the Shieldscoefficientfor Reno mattressesis
approximatelydoublethat of equivalentloosestone.

Althoughthere is some scatter in the results,a good fit to the higher


turbulencedatawas obtainedwitha linearequationhavinga gradientsimilar
to thatdetermined
for riprap:

C = 12.3T1-2.17 ( 11 )
for 19%<Tl<35/"

12 sR 427 3rl07/05
tr
Undernormalturbulenceconditions,C can be givenby
C = 0.17
for Tl<19% (12)

Forsafedesignit is recommended to adoptfor the definitionof C the envelope


curverepresentedby the dashedline in Figure5, whichwas drawnto include
all the experimentalpoints:

C = 12.3Tl - 1.65 (13)


for Tl > 15%.

A comparisoncan also be madebetweenthe meanflow velocitiesmeasured


in the presentrapidflow testsand thoseobtainedin the Coloradostudywhich
causediniliationof stonemovementin 300mmthick Renomattresses(cdtical
vefocity)-see Section2 atld Simonset al, 1984.The mattresseswerefound
to be stableup to meanflow velocitiesof 4.85to 5.61m/sin the presentstudy
whereasin the Coloradostudythe criticalvelocityreached4.18m/sfor Froude
numbersbiggerthan 3.0 or 4.97m/sfor Froudenumberssmallerthan 1.5. In
the presentstudythe Froudenunber was of the orderof 2.3 whichsuggests
a greaterstabilitycomparedwith the 1984study.

The presenttests demonstratethat in highlytutbulentflows stone movement


does occurundermeancross-sectionalvelocities whichare muchlowerthan
thosementioned above.Thisis shownin Table 2 (tests RHl2 to RH15)where
meanflow velocitiesof the order of 1.6m/s(prototype value)associatedwith
turbulenceintensitiesof around307oweresufficientto causestone movement.

6.3 Analysisof tests with looseconcreteblocks


It can be seen in Table 4 that collapseof the mattressformed by loose
concreteblockswas reachedduringTestCL8.Thevalueof coefficientC given
by equation(9) was calculatedfor this test and plottedin Figure6 againstthe
measuredvalueof turbulenceintensity,Tl. Also plottedin the figureare the
resultsof the testscaniedout in the previousstudy.lt is apparentthat the new
test value is consistentwith the previousresultsand that all the tests can be
conservativelydescribed by the following design equation which was
recommended in Escarameia and May (1992):

Dn= C ( 1/ (s-1))(Vzb | 29) (14)

with
C = 0.75(12.3Tl - 0.20) (15)
for Tl ) 5%.

6.4 Analysis of tests with cabled blocks


The resultsof testswith cabledblocksindicatedthat collapsewas moreeasily
reachedunderrapidflow conditionsthan underhighlyturbulentflows,where
the turbulenceis highbut the meanand localflowvelociliescan be relatively
low.This appliesto the threedifferentlevelsof tensionimposedon the blocks:
testswith high levelsof turbulencedid not producecollapseof the mattresses
butthis situationwas reachedin severaltestswithveryfastflow.lt shouldbe
notedthat in allthe cases ocollapse" only involvedmovementof the front row
of blocksor intermittentliftingof blocksin the centralregionof the mattresses
(andthis case onlyfor the mediumlevelof tension,levelB). The positioning
of the cablesin relationto the directionof the flow does not seemto have a
strongeffecton the stabilityof the revetment;however, tests CRA2and CRAS
pointto a slightlyhigherstabilitywhenthe cabledirectionis transverseto the
mainflow direction(see Table6).

13 SFrt27 l1l0o/95
tr
Table 12 presents model and prototype velocity values for tests where
movement occurred, as well as the corresponding values of the Froude
number.The value of coefficientC was calculatedfor these tests and plotted
in Figure6 against the turbulenceintensityTl (the resuhof test CH2 was not
plotted because the value of Tl was not available);in this figure the letter L
was used to identify the tests where leading edge failure occurred as oposed
to movement of blocks at mid sections of the block panel. Also plotted are
points correspondingto tests carried out at high tuttulence levels ahhough
collapse was not reached (these tests are identified by the letter N). ln the
determinationof the values of C, Dn was taken as the ratio of the weight per
unit area and the density of the model block material . A sample consisting
of several model blocks was weighed and the average volume of the blocks
was determinedin order to obtain the density of the block material. This was
found to be 2750kg/m3. For the calculationof the weight per unit area, the
area of the blocks was taken as that given by the outside plan dimensions
(O.O40mx 0.036m). This procedure led to the value of Dn equalto O.O055m
(modelvalue)or 0.044m (prototypevalue) and is equivalentto porosityof n =
0.58, based on the overall dimensions of the blocks. In doing so, proper
account is taken of the fact that the blocks were not solid but had insteadan
open shape. The same approach was followed for the analysis of the
interlockingblocks, which had a density of 2700kg/m3. The value of Dn
determinedwas equalto 0.0050m(modelvalue)or 0.040m(prototypevalue);
this correspondsto a porosityof n = 0.55.

It can be seen in Figure 6 that, althoughfailure was noi reached in the high
turbulencetests, it is still possibleto define a relationshipbetween C and Tl
that is safe compared with all the test data. Therefore,for high tutbulence
levels it is recommendedto adopt the following equationfor design of block
maltresses(with tensionedcables):

C = 1.79Tl - O.72 (16)

For "norrnal"turbulence conditions,C can be taken as 0.05; the transition


between this value and Equation (16) can be assumed to be at Tl of about
43%, but there are no data in the range of Tl between 125% and 52o/"to
confirm this assumption. Given the above limitations,a value of C equal to
0.05 can be used up to turbulence levels of 43% and Equation (16) can be
adopted for higher values.

6.5 Analysisof tests with interlockingblocks


The testsperformedwith interlocking blocksresultedin onlyone situationof
collapseof the revetment(Test lB2 in Table 10); the mean and near'bed
velociliesreachedduringthis test were convertedinto prototypevaluesand
are presenledin Table'12.Whenplottedin Figure6, the pointcorresponding
to TestlB2 indicates, as expected, blocks
a greaterstabilityfor the interlocking
ihan for the loose concreteblocksalso tested. However,becauseof the
scarcityof the data,it is notpossibleto determine
the relationship betweenthe
stabilityfactorG and the tubulenceintensity.

14 sRln 31tO7195
tr
7 Conclusionsand recommendations
7.1 General
An experimentalstudy was carriedout to investigatethe stabilityof SOOmm
thick Reno mattresses and of cabled concrete block mattresses in
environments withvariouslevelsof turbulence.Thetestswereperformedwith
modelsof the revetmentsplaced on a flat bed at scales of 1:5 for Reno
mattressesand 1:8for the concreteblocks. Additionaltestswerealso carried
out with loose solid concrete blocks and with intedockingblocks. The
turbulencelevels varied betweennorrnalturbulence,which conespondsto
uniformflow in straightchannels,to higherlevelssuch as those occurring
downstreamof hydraulicstructures.

7.2 Renomattresses
1. The characteristicsof the Reno rnattressestested are summarisedin
Table1 and illustratedin Figure1. Theyweretestedat a scaleof 1:5.

2. Failureof the Reno rnattressesas a revetment,whichconespondsto a


siluation where enrsion of the underlyingmaterial occurs was not
achievedin the tests,but incipientmovementof stonewithinthe mattress
cellswas observed.The meanflowvelocityassociatedwiththis situation
is called the critical velocity. During the test programme incipient
movementwas obseled both in high-velocity, shallowflow tests where
the turbulencelevels were relatively low, and in high-turbulence
conditions.

3. The stabilityof the Renomattressestestedunderbothnorrnaland higher


turbulenceconditionsis considerablyhigher than that of riprap of
equivalent size.For normalturbulence,the stablesizeof the stonefilling
in Renomattressesis approximatelyhatfthe size of loose riprap.

4. Under normalturbulenceconditionsthe Reno mattressestested were


found to be stable up to mean flow velocitiesof 4.8 to 5.6m/s, with
Froudenumbersof the orderof 2. This indicatesa slightlyhigherstability
than obtainedin tests conductedby Simonset al (1984)who foundthe
to be 4.18m/s(for F;3.0) and 4.97m/s(for Fr<l.5).
criticalvelocity

5. Under highly turbulent flows stone movementwithin the mattresses


occursfor mean cross-sectionalvelocitieswhich are much lower than
those obseruedunder normalturbulenceflows (valuesof the order of
1.6m/s were obseruedcomparedwith velocitiesof around Sm/s for
normalturbulence). This findinghighlightsthe importanceof taking
accountof the turbulencelevelin design.

6. The followingbest-fitequationwas foundfor the stabilityof 300mmthick


Renomattresseson a flat bed:

tr2

D-=c 1 vb (17)
" s-1 29

where

C = 0.17 for T1 < 0.1e(ie1s%)


C = 12.3T1-2.17 for 0.19<T1 <0.35

15 aR4.n or12n5
tr
For safe design it is recommended to adopt for the definition of C the
envelopecurue shown by a dashed line in Figure 5:

C = 12.3Tl-1.65 for Tl > 0.15 (18)

fn these equations V6 and Tl are defined al 1oo/oof the water depth


above the bed.

7.3 Loose solid concrete blocks


1. The resultsof the presenttests confirmedthe findingsof a previous
research study carried out with loose solid concrete blocks (see
Escarameia andMay,1992).Therecommendation giventhenis therefore
reiteratedherein the form of the followingsafe-designequationwhichis
a combination of Equations(14)and (15):

o
t Vu- (1e)
Dn = (9.22Tt-O.15)
F Zd
for Tl > 0.05 (5%)

ln this equationV6 and Tl are definedat 10%of the waterdepthabove


the bed.

7.4 Cabledblocks
1. Testswerecarriedout withscalemodelsof cabledconcreteblocksof the
proprietaryArrnor{lextype (seeFigure3). Threedifferentlevelsof tension
were appliedto the cablesfrom high tensionto a situationwherethe
blockswere tied togetherbut not tensioned.

2. The amountof tensionappliedto the blockrnatdoes not appearto have


a strongeffecton the stabilityof a revetmentsuch as Armodlex;this is
likelyto be partlydue to the stretcherbond interlockingsystempresent
in thistypeof revetment.

3. lt was foundthat a slightlyhigherstabilityis achievedwhenthe cable


directionis tranwerse to lhe main flow directionratherthan when it is
parallelto the flow.

4. As for the Reno mattressestested,collapsewas tnore easilyachieved


undernorrnalturbulenceconditions than underhighlyturtulentflows. In
mostcasesthe collapseinvolvedfailureof the leadingedgeof the block
mattressby rotationor up-lifting;in the other cases lifting movements
were obseruedin the centreof the mattress.

5. Withno tensionin the cablesthe Armorflextyperevetmentwasfoundto


be stablefor meanflow velocitiesof at least7mls if the tubulance levels
are low (typicallyless than 12oh);in highlyturbulentflows collapsewas
reachedat a meanflow velocityof 2.8m/s.

6. The followingsafe-designequationis recommended for cabledconcrete


to those the blockstested:
blockswithsimilargeometriccharacteristics of

Dn=c+* (20)

where

16 sR427 01/12t95
tr
C = 0.05 for Tl < 0.43 (43%)
and
C = 1.79Tl-0.72 for 0 . 4 { } < T 1< 0 . 9 0

7.5 Interlockingblocks
1. Theadditionaltestscarriedout with interlockingblocks(ofthe proprietary
Armorflextype- see Figure3) showed,as expec{ed,greaterstabilitythan
that of loose blocks under similarflow conditions. This is apparentin
Figure6 whichalso showsthat the interlockingblockswere less stable
than the cabledblockstested. Due to the small numberof data it was
not possible,however,to establisha designequation.

8 Acknowledgements
The experimentalpartof this projectwascarriedout withthe assistarrceof Ms.
G Lundstrom,from ChalmersUniversitywho was at the time workingat HR
Wallingfordas a visitingstudentwith partialfundingby WITEC.

The collaborationprovidedby Mr. R Greenfordof MaccaferriUK and by Ing.


F Ferraioloof OfficineMaccaferriS.P.A.,ltaly,is greatlyacknowledged.
So is
the co-operationof Dr. R Bakerfrom SalfordUniversitywho loanedexisting
modelsof concreteblocksfor testing.The commentsprovidedby MMGCivil
Engineering SystemsLtd are also gratefullyacknowledged.

17 sA127 0ll12n5
tr
9 References

EscarameiaM, May R W P and Atkins R (1995). Field mesurements


of
tuttulencein rivers. HR RepodSR 424.

M and May R W P (1992).Channelproteclion- turbulence


Escarame'ra
of structures.HR ReponSR 313.
downstream

PilarczykK W (1990).Stabilitycriteriafor revetments.Proc. of the 1990


NationalConferenceon HydraulicEngineering, ASCE,editedby ChangH H
and Hill J C, San Diego,USA.

SimonsD B, ChenY H and SwensonL J (1984). Hydraulicteststo develop


design criteriafor the use of Reno mattresses. ColoradoState University,
USA.

18 sR427 28107195
tr

Tables

sB4,27 M7tg5
tr
Table 1 Characterisficsof the Reno mattressestested

Model Prototype Prototype


1:5 scale values Reno mattress
Mesh size (mm)*
a 19 95 80
b 20 100 100
Celldimensions
(m)*
h 0.50 2.5 2.0-3.0
l2 o.20 1.0 1.0
l3 0.06 0.3 0.3

Stonesize (mm)
dso 23 115 125
Specific weight of
stone (kg/m", 2680
Voids ratio (e) 0.65

* See Figure1

saln 28n7n5
tr
o
.E .g .c o
.c s,
'= 'ts E
(U
.-9
*t v(, J E
o c
o c, 5a? o I
c o LO
i.9 E s
6t
t
E
o
o
br
E t!-
:c - =Eo . i Is$
cEO
3 i,t
-YO
o E !Io 96 r 96 r 8r
5 E,.- trr
o
lt o .> 5- tE 5 Eut
o z. o 6
9
sE 6 E E e
:(g o
6 (g E
E E E o
a U' U' o

G
}E
o= o
(n o o ol o
(o
() lt o o o o (o o rJ)
ol u) (\l (Y) (t) u) q) u) o ol
c> ol CO t { u? u) (o ro ro (o (o (o t\ t\
H'6 ci d c; o o o o c, o c; c; ct o ct o
-<o5

c
&a
1r ,
o
(it u)
o
o o o o o o
(') u)
@
la)
@
o
@
o
l'-
o
ol
o o
t\
(t (tt (9 (t c) (o t4t c') (o (7) (1t s t $
Es
t.|-
ct ct c; o o ci ci o o o ci o o o o

.g s
o ra) \(t) \|J) q o) q \\t al \tl
o
+. F-F (o
d
lO
d
q
N o o
lJ)
o o (\l
llt
ot
$

o tr
o
q
g
T s
e5
rL.
qe
tr
q
o)
G)
q
u)
@
ol
ol
ut
t*
G)
t\
U)
ot
n?
(O
ot
\
o,l
o!
(O
ol
\
s
ol
q
t
(\l
q
(o
ol
al
(t
d
ol
q?
c')
\c,
(it

,9
\ o att (o {t (n (t)
o (t) N (o GI o, ol (9 U) t- o
u) (t o)
;
o o
c)
c;
o
o
o
o
o
o
o
o
o
$
o
o
I
ci
co
(O
o
o
N
o
o
I
ct
E
o
ct
o
o
o
o
rf
c
o
o
ci
o
v,
io
.ao
oc
o->

t
*.
G
;F
Ta i so
E oF
>6
o
't
q o
rt
ro
o)
l'.
u)
() o,l
(t
(t) o
u)
AI t
o) (') o
lrt
(o
u)
o,
o
(t) N

o
q E ci
q
o o
t
cl
ci
a?
o
ot
t
o
s
o
$
\t
ci
a
o
rO
ct
rO
o
lO
o
@
ci
1l)
N
ci
@
lo
ci
ct
o E
tr o

C\|
o 6 ol ct st q) (o tr @ o o ol (t a u)
a e I
af, (r r(f
I r
(r I
(f,
I
tL
-(r -(f, -(f, I
TE
-(f I
(r I
E
-
E
-(f
N

saln anoTlgs
tr
.g .Y$l .g
.ER o9sol i
E
=q
.Er
o
c c
o JF b 1r_ ! (\l
.o E R
E
r r E l r lEl n
(U o O ttt
*LL E .t, *lL
P o oo bE o!,
o E =o =o
o 66
ll
o b E EE b E
o z 9F EF
6 F
>E
o e= 6 F
@?
>E
o

o
=
o9
F
o (\l t\ c,
q q u? -q
F.a
oo
N (\l oi RI $t
> oo

-&-
E

rO o o o o
Eg $ ro ro (f, !f

e*
TL
cj o o ci o

F q q (o
PS
->- O) o,
q
@
(\l
N
so
F

tr
o
rL

I g
o
E
es
-.3-
ro
o)
q
o
q
o,
a
ro
(f)

(o
tr
E
$
ca
o
E
o
o
c
o
(f)
(o
(r)
o
t
q
o
\r
(o
q
o
rt
(\l
o
v, 'Ee
ba E o o I o o

10
6s
>a
$
(g E
E EX o
o Fe
>6
o
'= o,
q
@
q
O)
\
o
q
(o
a?
q E ol (\l
o (6
o
t U)

()
o o F (\l (o $ lo
a o TE CT CE cc E
F TT CE TE tr E
N

sa427 2At07E5
tr
o
c E
o r-
.9 E o(o
(U o
-q) o $3
o E Erl
-o o od-
o z

o
=
o9
i
lo (f)
o) @ @ @ (f)
@ o lo o N g, o (f)
L> q F ol a? a? a u? u?
o'6 o o o o o o o o
> oo

o o(o ro o lo o
(o lo lo
E-E, ro s to lo ro
a? C\l c\
g e*
tr
o o
c\
o o
c\
o
c!
o
c!
o
C\
o
o
o
.h.

o bs o) q
o -=-bs ro o q q q q g?
s -.iv a.i @ t (f) s lo
e5 tr
$l

e s
rhr

q eR- \
(\
F
a? q
N (\l
a?
s(\l
It?
N
a?
(r)
a
q,
lo
(o
rO (f)
sl C! ol
Ur tr
t
g (5 c\I $ (\l o rO (o
o o
q o o q
F
q
cl
q q q
o o o
I
o
I
o o o o o
a .eQ'
o
.|r
otr
o->
E
o
Oc

3rt
tr 9?
o o
o ES.9at
oF o) o
(o C! @
(o
$
o
(\l (o
o E5 =
$t
q q
lo N
c\
l-
(\,! I
@
a n
lO

a E o o o o o
(f)
lO o o
o (5
o
(r)
o L o
-t o

\t
o (\l (r) j (o
a
al,
o
F
J
o
J
o
J
o o
5
o o
J
l-
J
o
@
J
o
N

sR 427 4to7t95
tr
o
.9 E
(! 9
OE
to z9
att
-ct
o e
at
B}
o9
@ o, @ l-
s t- o,
L>
ro .=
q \ o \ q
o o o o
sto o

o
+tg (\l
(\!
lo
(\l
q
lo
lo
a?
o
ro
a
o
o,
-q
E*
lJ.
o o o o o

o
o o
a? q
tr =F8
F

o
g N ol

5
e5 3- a -q
+. PS
->-
(f) @
@ $l
_El tr
-
\
F @ c\l C\! q @

*
PS (o
C\I
o,
(\l lo
(o
@
+
C4'

tr tr
.9
u, EC) N
r+
(\l
r+
(f)

e E q q
N
q a

g o o o o
s
a
o
Io 'ge
eQ I
o
att I
$ rO
o
T-c c
o o
: >6_ E
g e F

o
a
o EF
>6
at
3
o
rO (o
t '=
o (f,
(o @ $ o
g E
o)
@
o, o)
i-
a oq C

a o o o o o o .o
o o
I U)
o
.=
E
.g
o
u) (') o o
s -o
o o
o
F $l
ro (U
a F I I I I I o
o o o o o
N (E

sR 1Zl 28107re6
tr
: E}
o c 9
c o u O trEr
c o(\l

UEgEU
.o E c 6-99
(E
e
o
6 FN o o-oAl
q)
at,
o
E 8f;5 tt Eit'r
Ir - g* = t r - ' ,499 5 = u i
.o o
o z
(r,
= E
O.=
o, @ s (f) (o
q q
E.a
oo
c.i N
\(\t @
oi C\l

> oo

-&-
-c
o [o o o o
!. O) o N C\l (f)

q
e* r r
o o o o o
Ll-

o
= =s'8
F
o
(\
a?
C!
I ,
o
E
b bS
a PF a
(o
\(o f-
s
G
q tr

+
c
;S
;s
-v

tr
o)
d
t\
o
.q
r
q
N
q
o

.9
v, (s o c!
tr
o .9 @
o
c\l
o
t 1- q q
o
o o
o
o
Io .aQ'
otr
9
o F

@
@
o
ro
o
otz o c.l
T
F

T- c o o o
>= E
F
o =ai
{ 9!
() ?x
e
a
6o
or
E5
=
o
o
b .;o o @ O) o f-. o
o I a? oq ol n C
E $l C\l (\l .9
a o
(g
F F

o
I a
o
.=
-o
.;
(o o
o -o
e
a o
F
tt
F

E
N

(r
(f)

(I
s
E
ro
tr
o
(6

o o o
N C)
o
t!7

sR4n 31n7/95
tr
o CE
*t
c oo
o>
i
.9
ct
u
39
$g
Eg
-g .rt
o -J
o vc)
-o bo
o di -o :B
63
CN

a,
= E
osz (o (\l
N (f)

f;.a
oo
t-
o
@
o
> oo

-c
o
o F-
T' -9
rO
o,
o
o.l
c
o E*
LL
o o
5
e
E F
6a c\ (o
q -F_;q o, o,
Bl =- (r)

tr
t
sPS cr)
" -.:- o;
(o
o
1r)
c tr
.9
u,
tr
*.o
L (!
o .9
N
F
lo
$
E
o
Io .aQ'
otr
o o

o->
T-
: >E
o =&
o!
{
o ?x !

s
a
16O
or
E5
qt
.9,
ril
o)
(r)
F

E =
o E o o
o
o
a
a
I
N
C\I

e
F

o
o
co co
- -
a F o o
N

sat'n 91rc7/95
tr
.t o
-:<
o ER ER
at, =co
!(9 $tr 8rr
c =r.r.' 5 L -
.o b c.i o o
(U 3rl o2
e
o I u-- P'.E 9 6
.E6
U'
-o
o
trE
(5.E =s EE - at,
o:=
ru .o Go
6
o-
E'F E'E
a: @;
f (5 (U

G
=tr
o\>
(o O) c\l
a? \ 0q
F.a
oo
$t c\l N

> oo

F- o
o
o o
! E, F N

o
F

E E*
LL
o o o

E
a.
(!
q s
9F q
ro
\
lJ)
al
ro
I

a F-

s ;e
o g?
a @
q, PS N
tr
o tr
F

o 0)
9
Io o
o
O)
f-
q
$
O)
co
i-
I-
q
c o o
-l E
I
.aQ'
()F: F
o->
ta T-
{ >E
o =&
o 9-o
a ?x
6o o
E oF .o
'= f- ro @
o E5 o? F

E c.i c\l
a o
o
I a

a
o o TD
C\I
co
q)
co
o
a F cc
o
E
o
(I
o
N

sA127 A1n7/95
tr
6 .hl

U'
gE st
(l)ct
-o?
c g6
.9 89 .tl
(5 F5 o =6
e => =Lt #B
o
o ie
= L
c:
iU, rotr
-o io
o EZ -G
do:
.9') .9'rE
I -o

o
= F
o \:, O) @
o) o)
L>
ru.=
q
@
o? n
oo o o $l

> _oe

F-
Es,
o
r\
o
s
o
@
n o? co
E*
tr
o o

F
-)^ a?
:--;s @
=-
tr

F o,
9F
-:t- o
lo
tr
E
o
q
.9
o
q
t
.9
o,
ol
o)
(f)
I-
(f)

o o c
o
q
o
q
rh. o
rh.

o .ao
otr
ot7
= 6-
>=
o
I
=&
9E
{ ix
o 6o
o oF
o
a E5 '=
o ro
C\I
o
u?
E E o
F

o
o (6
o
a a
G
o
Ctf
o o
o -
N
-
F
(I
a F o o o
fl

saln 3Jn7p5
tr
ttt
c
'oa , 9
.9
(g
to e b
oaD
a$
60
(50 btl
o
o o4-
o F

v)
= ?
o.= C\l r-
C\I N
:( >
u ,. e q q
oc) o o
> oo

ttt
v, iia o o
o ]J -9 N
q? f-
c?
o ->, o o
o iE
c
o
F
t a)^
-_}-
bs oi
o,
sf
(\I
.lra
tr
.9
I
I s q ro
FF
o
o
-.f-

tr
f-
\r
\t
I-

-l
t
o
=
t (6 r\ (o
S .o
!
o
N
q
o
*
q
o
vr .aQ'
{ ()F
o o->
o T-
>=
a =tt
Ot 9E
.s ?x
do
{ oF
o E5 o
o .o
3
lo
@
c\l
rO
E E a
o o o o
o
s L

U)

o
o .h F C\l
a o
F
(D
e
N

sA127 31lo7l{t5
tr
Table 11 Reno mattresses - Model and prototype values for critical
movement condition

Test Fr U model U prot Vo model Vs Prot


(n/s) (n/s) (n/s) (n/s)
RH12 o.344 0.656 1.47 0.558 1.25
RH13 0.335 0.681 1.52 0.689 1.54

RH14 0.359 o.720 1.61 0.759 1.70

RH15 0.333 0.715 1.60 0.567 1.27

RR3 1.79 2.17 4.85 1.79 4.00


RR4 2.22 2.51 5.61 2.OO 4.47
RR5 2.01 2.46 5.50 2.33 5.21

sR 127 Un7tg$
tr
Table 12 Cabled and interlocking blocks - Model and prototype values
for failure conditions

Test Fr U model U prot Vo model Vs prot


(n/s) (n/s) (n/s) (nds)
CRA3 2.52 2.74 7.75 1.89 5.34

CRA4 2.61 2.83 8.00 1.90 5.37

CRAs 2.27 2.56 7.24 2.47 6.99

CRBl 2.38 2.36 6.68 1.97 5.57


CRB2 2.68 2.79 7.89 2.15 6.08

CRB3 2.60 2.82 7.98 2.18 6.16

cH2 0.541 0.988 2.79 0.110 0.311


lB2 0.329 o.627 1.77 0.451 1.28

sa427 S1lO7lS5
tr

Figures

sA127 3ln7p9
tr

Flow
direction

Lacingwire

-G{
ME/lN-g5IGI

of Renomattresses
Figure1 Geometriccharacteristics
tr
cn

;
c\T

c\,

+
n.l

C\l

Og
C.I O)
N
U)

+
s. c\l
azlsJalfrrn l(racJad

Figure2 Gradingcurve of fiiling stone in the Renomattresses


tr
Cabledblocks:

ii[:]ii
tt ll

iir-l ii

lnterlocking blocks:

Figure 3 Geometriccharacteristicsof cabledand interlockingblocks


tested
tr
a
eo
E
.s
(t)

TI .9s
HH
Ss
.l
e l
'ol
-:
o
*

:
o
E
c
tIl a,l
o|I
I
ol I
o YY
_o
o

.c.
o) =
f
9E .sE
;o8
x (g
iI so
E
a
o
(,)

o_
EE
(L

Figure 4 Generallayout of test rig


tr

ro
a?
\\ O

...\ .\ \i"
9t.
"\\

ro
c.l
o

o
itr

ro
r
ci
E
o
o ootr
g Q-
EE .}}
\tr
\tr
\o
1ts
\ oo
c)
9E S$ s e \
r
o
H5a'# I3 \ "l
. q"l*
xotr
l*
". .*l<
.F

=
lru(l-s)oz'oc
Mg517-gtcT

Figure 5 RelationshipbetweenC and the turbulenceintensityfor RENO


mattressescomparedwith riprap
tr
)q
5
9^ <d)
t
(')
!t

"'.):. $A'Ec
J g
E iE Hs e*6B
EE
% E*E$;EeeEE
"'o$ar 8otr+ 5.\ zJ

"\ te'
?\.
o\

z+ zq

\
I
I
I
zo I
I
I
zo I
zo I
I
\ I
I
I
I
I
I
-rb
q*DAA6r----"" :&,
J-o
o lJ
"$d ol

e O)
o
a a?
o
c\l
o
o
=c
lru(l-s)oz'o ME/6'.gYGT

Figure6 RelationshipbetweenC and TI for concreteblocks


tr

Plates

SR /127 31/07195
Plate1 Assemblingof Renomattresses
Plate2 Renomattressesplacedin the flume
Plate3 Renomattressesin flumeshowingpaintedbands
i-\
(:
I
|l
&'r

\.)
-t

T..T
'\i
'\ - #.1
'i-\.

-t

Plate4 Renomattressaftera test showingmovementof stone


Plate5 Cabledblockswith trame

You might also like