Академический Документы
Профессиональный Документы
Культура Документы
CFDpredictionoflocalscourholearoundbridgepiers
ZHU Zhiwen()1,LIU Zhenqing()1, 2
1.CollegeofCivilEngineering,HunanUniversity,Changsha410082,China
2.DepartmentofCivilEngineering,UniversityofTokyo,Tokyo,1138656,Japan
CentralSouthUniversityPressandSpringerVerlagBerlinHeidelberg2012
Abstract: In order to predict the local scour hole and its evaluation around a cylindrical bridge pier, the computational fluid
dynamics (CFD) and theories of sediment movement and transport were employed to carry out numerical simulations. In the
numericalmethod,thetimeaveragedReynoldsNavierStokesequationsandthestandardkmodelwerefirstusedtosimulatethe
threedimensionalflowfieldaroundabridgepierfixedonriverbed.Thetransientshearstressonriverbedwastreatedasacrucial
hydrodynamic mechanism when handling sediment incipience and transport. Then, riverbed volumetric sediment transport was
calculated, followed by the modification of the river bed altitude and configuration. Boundary adaptive mesh technique was
employedtomodifythegridsystemwithchangedriverbedboundary.Theevolutionoflocalscouraroundacylindricalbridgepier
waspresented.Thenumericalresultsrepresenttheflowpatternandmechanismduringthepierscouring,withagoodpredictionof
themaximumscourholedepthcompared with testresults.
Keywords: localscour bridgepier computationalfluiddynamics sedimenttransport
1Introduction
Scour is defined as the erosion of streambed
sedimentaroundanobstructioninaflowfield[1].Local
scour occurs at a bridge site when the local flow field
near the bridge pier is strong enough to remove bed
material. The obstruction to the flow caused by the
bridge pier is of primary importance in scour process.
Theprincipalfeaturesoftheflowincludethedownflow
aheadofthepier,thehorseshoevortexatthebaseofthe
pier, the surface roller ahead of the pier and the wake
vorticesdownstreamof thepier.
Forbridgeengineeringpractice,accurateprediction
oflocalscour,suchasthemaximumdepthofscourhole
around the bridge piers, is critical for bridge design,
maintenance and evaluation. Fail to present an accurate
maximum depth of scour may either lead to an
uneconomical design of substructures, orresult in scour
damageorevenbridgefailure.Thismayleadtomassive
lose on life and economy, and always result in serious
impact on local traffic. Duringthe past several decades,
there were lots of bridge scour damages in China. In
1963,thetorrentialraininHebeiProvincedamaged209
bridgesduetolocalscour.In1994,localscourdamaged
49railway bridges in China, andtherailway traffic had
Foundation item: Project(50978095) supported by the National Natural Science Foundation of China Project(IRT0917) supported by the Program for
ChangjiangScholarsandInnovativeResearchTeaminChineseUniversityProjectsupportedbyChinaScholarshipCouncil
Receiveddate:20110411 Accepteddate:20110921
Correspondingauthor:ZHUZhiwen,ProfessorTel:+8673188821424Email:zwzhu@hnu.edu.cn
274
2Numericalmodel
2.1Governingequationofflow
The fluid equations for unsteady, incompressible,
threedimensional viscous water flow around bridge
piers can be solved approximately by the Reynolds
averagedNavierStokesequationsasfollows:
ui
= 0
xi
(1)
ujui
ui
p
+r
= rFi +
[2mSij - r uiuj]
t
xj
xi xj
(2)
2 xj
directionandzerosotherwise. Sij = 1 ui +
uj
,isthe
xi
x j xi
kui
m t k
m +
+ Gk - re
=
xi x j
s k x j
e ui
m t e
e
e 2
m +
+ C1e Gk - C2e r
=
xi x j
s k x j
k
k
(4)
(5)
u u u
where Gk = m t i + j i ,is the production term
x
j xi x j
ofturbulentkineticenergyinducedbythemeanvelocity
gradientC1,C2,C,k, aretheempiricalconstants,
with value of C11.44, C21.92, C0.09, k1.0,
1.3,respectively[10].
The standard k turbulent model is based on the
hypothesisof fullydevelopedturbulentflow.Inorderto
avoid fine grids among sublayer and save computer
power, wall function is applied to obtaining wall shear
stress.Thiscanberealizedbyusing
t wall =G wallU p /d p
(6)
if
d p+
>11.63 ,then
if
d p+
/U p+
(7)
(8)
where tistheturbulentviscosity.
There are great varieties of turbulent models
available in industrial CFD simulation. Although the
standard k turbulent model comprises weakness in
practical application, it remains the workhouse of
industrial computation because of its relatively high
resolutionandlowCPUtimerequirement.Sincepresent
studies deal with riverbed and pier surfacebounded
threedimensional unsteady flow, the standard k
turbulent model will beadesired selection on PCbased
CFDsimulations.
For the standard k turbulent model, t=Ck2/,
where C is the empirical constant kuiu j 2, is the
m ui ui
, is the
r xk xk
dissipationrateofturbulentkineticenergy.
kand canbeobtainedfromthefollowingtransport
equations[10]:
turbulent kinetic energy =
U2
360g
(9)
whereUisthemeanapproachflowvelocity.Ifthemean
sediments diameter is larger than the critical diameter,
J.Cent.SouthUniv.(2012)19: 273281
275
t i
h
- Cqb
t
xi
CL/CD=0.85.istheanglebetweentheflowdirectionat
sedimentlocationand the slope direction projected onto
ahorizontalplaneisthe slopeangle tothehorizontal
plane,asshowninFig. 1.
(10)
whereqb,i isthecomponentoftimeaveragedvolumetric
sedimenttransportrateperunitwidthisthebedshear
stress vector with three components of i (i=x, y, z)
C=1.5,is anempiricalconstant h is thebedelevation.
qb is the timeaveraged volumetric sediment
transportrateperunitwidthonhorizontalbed,whichcan
beobtainedfromthe followingequations[11]:
1.5 -0.3 2.1
qb = 0.053D 0.5 g 0.5 d 50
D* T (T <2.5)
(11)
(12)
Asforsedimentsonhorizontalbed,thecriticalbed
shearstresscanbeobtainedfrom
whered50 isthemediansedimentgrainsize=(s)/,
is the nominal density s is the sediment density
D*=d50[(s1)g/2]1/3, is the dimensionless sediment
parameterss/.
T and Tm canberecoveredfromEqs. (13)and(14):
T = ( t -t b,cr ) /t b,cr
(13)
l t - t b,cr
t b,cr
(14)
Tm =
t b,cr
t b,cr
CD
CL
cos q + sin q sin a cot 2j + cotj
C
D
C
cos2 q - sin 2 q cot 2 j + L sin2q +
CD
CL
sin 2q sin a + sin 2 q sin 2 a cot2j
CD
1/ 2-1
(15)
whereistheangleofreposeforsedimentsCL andCD
donate the sediments lift and drag coefficients,
respectively. WEI and YE [13] suggested the value of
t b,cr = rq cr ( s -1)gD*
(16)
(17)
where n isthesedimentsvoidage.
2.3 grid
With scouring development, sediments on the bed
close to piers may be moved away gradually, and the
plane bed may not be maintained. This will result in
inclined bed surface, and gradually form the scour hole
around the bridge pier. On application of CFD to the
calculation of flow field around the pier, computational
grids should be remeshed since the computational
domain is changed concerning the bed scouring. Since
the scour hole may be of complex configuration,
remeshing of the computational domain with high
qualitymaypresent achallenge.
There are three remeshing strategies which are
widelyusedtofitthecomplexterrainafterscouring,i.e.,
ZlevelgridShavedgridandgrid,asshowninFig.2.
Onecanseethattherepresentedriverbedwillbezigzag
if the Zlevel grid system is applied, which does not
capturetherealbedconfiguration.TheShavedgridmay
fit the complex terrain well. However, it may be very
difficulttorefinethemeshclosetotheriverbedsurface.
Inthiswork,thegridisemployedtofitthescoured
276
Fig.2 Comparisonbetweenthreegridssystems
s=
z- z
H +z
Fig.3 Schematicplotoftestarrangement
(18)
3Casestudy
In orderto validatethe effectiveness of the present
method, numerical simulations of flow field as well as
river bed scouring around a cylindrical pier are carried
out based onavailablereferencereports.The simulation
parameters are almost the same as those reported by
MELVILLE [15], except for the domain length,
boundaryconditionsimposedonthesidesurfacebecause
of the available computer resource. The test was
conducted in water flume of 19 m long, 45.6 cm wide,
with water depth of 15 cm. A cylinder withdiameter of
D=5.08 cm was simulated as a scaled bridge pier fixed
ontheriverbed.Thebedsedimentsinvolved weresand
with relatively uniform size, and the median grain size
was 0.385 mm. The approach mean flow velocity was
0.25 m/s. Water was discharged at the rate of 0.017 12
m3/s with a bed slope of 1/10 000. Figure 3 shows the
schematicplotofthetestarrangement.
3.1Computationaldomainandnumericalmethods
The computational domain, as shown in Fig. 4, is
0.15mdeep and45.6cmwide, with theinlet boundary
locates atadistance of 7D upstream of the pierandthe
outletboundarylocatesatadistanceof10Ddownstream
Fig.4 Computationaldomainandboundaryconditions
J.Cent.SouthUniv.(2012)19: 273281
277
278
Fig.9 Gridsinvicinityofpierat10min
Fig.7VelocitydistributionatplaneBB:(a)Numericalresults
(b)Experimentalresults
Fig.10Gridsinvicinityofpierat 20 min
Fig.8 Shearstresscontoursonbedsurface
leading area of the pier also gets low stress value. The
high stress regions are located at both sides of the pier
facing the flow, indicating significant velocity gradient
amongtheregions,especiallyclosetothepier,wherethe
scour may first initiate when sediments satisfy the
scouringconditions.
3.2.2Livebedsimulation
In this work, the mean approach flow velocity is
0.25m/s,withsedimentsaveragediameterof0.385mm,
Figure12comparesthecomputedscourholeshape
with the experimental result at the scouring time of 30
min.Themaximumscour depth observed in experiment
is 4 cm and locates at the area close to the pier nose
whilethemaximumscourdepthfromthepresentstudyis
4.1 cm, and locates at pier sides. Hence, the presented
J.Cent.SouthUniv.(2012)19: 273281
maximumscourdepthagreeswellwiththetest.However,
thescourholeshapeandlocationbetweenthenumerical
simulation and the test are quite different. One can see
that the predicted scouring at the area immediately
adjacent to the pier nose is not so intensive as that
indicated in the test, i.e., there is a doublesided sharp
slopeimmediatelyclosetopiernoseinnumericalresults,
whichmaycollapseinlocalavalanchesofsedimentinto
theerosionzoneundertheactionofturbulentflowinthe
test. Such kind of collapse may change the shape of
scourholeand certainly resultindifferent flow patterns
around the pier. However, such kind of slope collapse
may be quite different to model, and thus is not
consideredinthepresentstudy.
279
Fig.14 Azimuthalplanesarrangement
280
direction,normaltotheflowdirectionandbetweenthem,
respectively (see Fig. 14). The horizontal axis x is the
radial distance from the pier surface, while the vertical
axis indicates the scourhole depth, in which zero atthe
verticalaxis correspondsto theriver bed altitude before
scouring.ItisshowninFig.15thatthemaximumdepth
(4.1cm)occursat30minatpointc.Thisalsoindicates
that remarkable bed scouring occurs at the region
roughlyoneradiumawayfromthepiersurface,withthe
maximum scour depth close to the pier surface line on
theplacenormaltotheflowdirection.
Fig.16Scourholetopographyafter30min scouring
4Conclusions
1) The CFD methods, with the powerful flow
visualization, show the ability of flow representation
during local scouring. This helps to recognize scour
mechanismanddevelopmentwithscourtime.
2) It appears that the present threedimensional
RANS equations combined with the standard k
turbulence model can predict the complicate flow field
aroundbridgepiers.Theaccuratecaptureofshearstress
valueontheriverbedisacrucialstepforbedscouring.
3) The maximum scour hole depth, which is the
mostimportantparameterinscourprediction,isingood
agreement with the test results. From engineering point
of view, the present numerical method can be used to
predict local scour hole and helps to carry out scour
designofbridgepiers.
4) The location of the maximum scour hole depth
and the configuration predicted by present method are
differentfromthosereportedbytests.Thismayattribute
to the fact that slope collapse to the erosionzoneinthe
testisnotrecognizedbynumericalmethod.Suchkindof
collapse may change the intermediate shape of scour
holeandresultindifferentflowpatternsaroundthepier,
andfinallyfailtoreproducethecorrectscourholeshape.
Such kind of influence should be modeled in the future
studies.
References
[1]
Fig.15 Scourholeprofilesevolutiononthreeazimuthalplanes:
(a) aa plane(b) bbplane(c) cc plane
CHANGHH.Fluvialprocessesinriverengineering[M].NewYork:
JohnWileyandSons,1988:2035.
[2]
J.Cent.SouthUniv.(2012)19: 273281
WaterResourcesPublication,LLC,2000:814.
[3]
NationalTransportationSafetyBoard(NTSB).CollapseofNewYork
JournalofHydrodinamics(A),2008,23(1):3341.(inChinese)
BARBHUIYAAK,DEYS.Measurementsofturbulentflowfieldat
NileRiver,Egypt[J].JHydraulEng,2004,130(6):488450.
[12] WANG Qiang. The finite element ocean model and its aspect of
verticaldiscretization[D].Bremen:UniversityofBrement,2007.
[13] WEIYanji, YE Yincan.3D numerical modeling of flowand scour
aroundshortcylinder[J].JournalofHydrodynamicsA,2008,23(6):
ASCE,1992,118(8):10911106.
PENG Jing, NOBUYUKI T, YOSHIHISA K. Three dimensional
numerical simulation for sour around the head of spur dikes [J].
JournalofSedimentResearch,2002,1:2529.
[8]
[M].Berlin:Springer,2002:292302.
[11] ABDELFATTAH S, AMIN A, VAN RIJN L C. Sand transport in
atbridgepiersandabutments[J].JHydraulRes,1998,36:183197.
variation of scour around cylindrical bridge piers [J]. J Hyd Engrg
[7]
rectangularchannel[J].FlowMeasInstrum,2004,15:8796.
[6]
[9]
281
655661.(inChinese)
[14] QIAN Ning, WAN Zhaohui. Sediment transport mechanics [M].
Beijing:SciencePress,2003:233237.(inChinese)
[15] MELVILLE B W. Local scour at bridge sites [R]. Auckland: The
UniversityofAuckland,1975.
(EditedbyYANGBing)