505
1 BACKGROUND
OfficiallyinauguratedonFebruary2nd,1892,Santos
Port is the most important port in Latin America,
beingresponsiblefor25%oftheBrazilianbalanceof
trade.Onceitislocatedinanestuarinearea,naturally
deep,thefirstcapitaldredgingwasperformedonlyin
1964,72yearsafter
theportinauguration.Inthistime,
itwasintendedtoestablishadesigndepthof14.8m
(CD) in the access channel. However, the design
depth was never reached, due to the inefficiency of
dredging proceedings, and the access channel depth
wasmaintainedinabout12.5m(CD).
In 2010,
another capital dredging was made in
SantosPort,inordertodeepentheaccesschanneltoa
designdepthof15m(CD).Oncemore,againdueto
the inefficiency of dredging procedures, the design
depth was never reached. In July 2017, it was
established a maximum operational draft in Santos
Port of 12.6 m, after depths lower than 14 m were
observedinbathymetricsurveys.
Considering an expected enlargement of
commercial vessels all around the world, associated
withtheroleofhubportcurrentlyplayedbythePort
ofSantos,itisindispensabletodeepenandmaintain
design depths. In
this context, the main goal of this
studyistoanalyzesedimentdepositioninSantosPort
Access Channel and predict annual maintenance
dredging volumes, considering current bathymetric
survey and design depths of 15, 16 and 17 m (CD),
based on a numerical hydrodynamic and
morphologicalmodel.
Dredging Volumes Prediction for the Access Channel of
Santos Port Considering Different Design Depths
L.M.Pion
HydraulicTechnologicalCenterFoundation,SaoPaulo,Brazil
J
.C.M.Bernardino
UniversityofSaoPaulo,SaoPaulo,Brazil
ABSTRACT:SantosisthemostimportantBrazilianport,handlingabout114millionoftonsin2016.In2010,
therewasagreatcapitaldredginginordertodeepentheAccessChannelto15mdeep(ChartDatum‐CD).This
depthwasnotachieved, due to
inefficiency ondredging procedures.As deepeningand maintaining design
depthsareindispensable,thisstudypresentsananalysisofsedimentdepositioninSantosPortAccessChannel
andanannualdredgingvolumesprediction,consideringcurrentbathymetricsurveyanddesigndepthsof15,
16and17m(CD).Anumericalhydrodynamicandmorphologicalmodel
wasdevelopedfortheinterestarea,by
usingDelft3D®,calibratedwithwaves,currentsandwaterleveldatamea s uredwithinSantosPortadjacen‐cies.
Sedimenttransportmodelwascalibratedwithsuspendedsedimentdataandhistoricseriesofdredgedvolumes
fromSantosPortAccessChannel.Twodifferentscenariosweresimulatedforeachdesign
depth,accordingto
theregionalenvironmentalcharacteristics.Forcurrentbathymetricscenario,themodelestimatesthatitwould
benecessarytodredgeanannualaverageofabout4,325,000m³fromSantosPortaccesschan‐neltomaintain
currentdepthcondition.Regardingdesigndepthsof15,16,17meters,itwouldbe
anin‐creaseof15%,55%,
and80%.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 12
Number 3
September 2018
DOI:10.12716/1001.12.03.09
506
2 STUDYAREADESCRIPTION
SantosPortislocatedinBraziliansoutheasterncoast,
(Figure1).ThecityofSantosislocatedatSaoVicente
Island, within a very complex estuarine system,
where there are more than 60 river outfalls. The
SantosPortissituatedinbothmarginsoftheestuary
outfall.
TheSantosPortareaisshowninFigure2.Its
access channel is divided into four different areas,
suchasillustratedinFigure3.
Figure1.LocationofSantosPort (GoogleEarth).
Figure2.SantosPortArea(GoogleEarth)
Figure3. Scheme of Santos Port Access Channel (Google
Earth)
Thisarea’sclimatecharacteristicsarewelldefined
by two different seasons: a rainy period, usually
betweenspringandsummer(fromOctobertoMarch),
during which around 70% of annual rainfall takes
place;anda dry one,betweenApril andSeptember.
These periods are generally designated as summer
and winter, respectively. Also,
winter period is
knownfor high significantwave heights, dueto the
occurrenceofcoldfrontsgeneratedintheoceanicarea
(stormsurges),whichdonotfrequentlyoccur in the
summerperiod.
The tideis semidiurnal, with amplitudes varying
between0.27m,intheneaptide,and1.23m,inspring
tides.
However, considering meteorological effects,
the water level can reach up to 1.83 m. Maximum
flow speed is about 1m/s along the access channel,
neartheestuaryoutfall,butitdoesnotexceed0.5m/s
inmostofSantosbay.
InAreaI,bottomsedimentisbasicallysand(70%)
anditsfraction
decreasesinwardtheestuary.InArea
II,thesandfractionisabout50%,whileinAreaIIIit
is about 40%. In Area IV, the most inner, bottom
sedimentisbasicallyfinesediment(siltandclay).
Forareaswhicharenotsubjecttowaveaction(II,
IIIandIV),sediment
depositionisdirectlyrelatedto
rain seasonality, due to higher river flow and,
consequently, higher total sediment transport load.
Hence,sedimentdepositioninAreasII,IIIandIVis
expectedtobemoreintenseinthesummerperiod.
AsAreaIisexposedtowaveactionandlocatedin
an area
characterized by low current speeds. Its
sedimentdeposition patternis associated withwave
climate.Generally,inthewinterperiod,whenwaves
are higher, sediment deposition is more intense in
Area I. In periods characterized by higher waves,
sediment is removed from the beaches and tends to
settleatthechannel
area.
3 DATABASE
For the study development, waves, currents, and
waterlevelmeasuredwithinSantosBayandEstuary
wereused.Also,formodelcalibrationandboundary
conditions,waves,wind,andtidedatawereextracted
fromglobalmodels,WaveWatchIII,NCEP/CFSRand
TPXO, respectively. Figure 4 illustrates all
measurementandglobalmodel
pointsusedformodel
boundariesandcalibration.
Wave
Current
Water Level
Wind
Figure4.Measurementandglobalmodelpoints(UTM23S–
WGS1984)
Also, bed representation was based on a
bathymetric survey realized in March 2016 at the
507
SantosPortAccessChannelandNauticalChartsfrom
Brazilian Navy data (CHM, 2016). Sediment
characteristicsandriverflowusedinmodelingwere
obtainedfromtheStudyofEnvironmentalImpactof
the last capital dredging (Fundação Ricardo Franco,
2008).
4 MODELREPRESENTATION
4.1 ModelBackground
Delft3D hydrodynamic module simulates non‐
uniform
flowsandtransportphenomenausingwater
level variation, river discharges or meteorological
forcing variables, including density gradient effects,
calculatedfromsalinityandtemperaturedistribution.
This model can be used to predict flow patterns in
shallow regions, coastal, estuarine or lake areas
(Deltares, 2014). In this case, numerical modeling is
based
on the continuity equation, the momentum
conservation equation (Navier‐Stokes) and the
transportequationssolution.
The model solves Navier‐Stokes equations for an
incompressible fluid, considering the Boussinesq
approximation, in which fluid density is considered
to be constant, except for the baroclinic term, which
represents flow variations due to vertical density
gradients. Moreover, the Boussinesq approximation
does not account for vertical flow acceleration,
considering hydrostatic pressure. This hypothesis is
valid when the horizontal extension is much larger
thanflowdepths(Deltares,2014).
Navier‐Stokes equations are simplified using
Reynolds average, which means deriving these
equations from va riables decomposition in time
average
and turbulent components, which are equal
to zero when integrated on time by definition
(Versteeg and Malalasekera, 2007). Numerical
simulations are performed through the finite
differences method, and space is divided into cells
from a computational grid. Delft3D uses orthogonal
curvilinearcoordinates.Inthiscase,theflowspeedis
calculated according
to the orientation of grid cell
faces.Thewaterleveliscalculatedatthecenterofthe
cells(Deltares,2014).
The Delft3D wave module (SWAN) is a spectral
wave model, able to reproduce wave propagation,
wave generation by wind and non‐linear wave
interactionsanddissipationindeep,intermediateand
finite waters (Deltares, 2014). The model solves the
energy balance equation for wave energy transport,
includingwavesgenerationbywind,non‐linearwave
interactions,bottomfriction,depth‐inducedbreaking,
andenergydissipationbywhitecapping.
The two modules shall be coupled in order to
accuratelyrepresent hydrodynamic conditions atthe
interestarea.
Thecouplingaccountsseveralimportant
processes due to wave‐current interaction, such as
enhancementofverticalmixingduetowave‐induced
turbulence and enhancement of the bed shear stress
bywaves.Forthiscase,Fredsoe(1984)wave‐current
interactionmodelwasused.
Inordertoproperlyrepresentsedimenttransport,
it
wasnecessary toconsider fine sediment and sand
transport,duetobedmaterialcharacteristics.Hence,
it was necessary to use two different equation for
sediment transport. Generally, transport of
suspended sediment is defined by the advection‐
diffusion equation for suspended sediment
concentration,asshowninEquation(1):
0
s
xy z
wwc
cuc vc c c c
tx y z x xy y z z
(1)
where:
c–suspendedsedimentconcentration(kg/m³);
u,vandw–flowvelocitycomponents,indirections
x,y,andzrespectively(m/s);
w
s–settlingsedimentspeed;
ε
x,εyeεz–eddydiffusivityindirectionsx,yandz.
For fine sediment erosion and deposition, the
Partheniades‐Kroneformulations(Partheniades,1965)
were used, as described below in Equations (2) and
(3). The model considers that fine sediment is only
transportedinsuspension.
e
EMS
(2)
s
d
DwcS
(3)
where:
E–erosionflux(kg/(m²s));
M–erosionparameter(kg/(m²s));
D–depositionflux(kg/(m²s));
w
s–settlingsedimentspeed;
c–finesedimentconcentrationnearbottom;
S
e–erosionstepfunction;
S
d–depositionstepfunction.
The step functions are calculated as exposed in
Equations(4)and(5).
1,
0,
cre
e
cre
cre
when
S
when
(4)
1 ,
0,
crd
e
crd
crd
when
S
when
(5)
where:
τ–bedshearstress;
τcre–criticalerosionshearstress;
τcrd–criticaldepositionshearstress.
Non‐cohesive sediment transport was computed
by Van Rijn (1993) formulation. Bedload transport
andsuspended‐loadaredistinguishedbyareference
height,abovewhichsedimenttransportisconsidered
as suspended‐load
and below which sediment
transport is considered bedload. The interaction
between bedload and suspended transport is
computed by using a reference concentration,
calculatedasshowninEquation(6),whichisimposed
inthewatercolumnatthereferenceheight:
508
1.5
50
0.3
*
0.015
as
DT
c
aD
(6)
where:
c
a–sedimentconcentrationatreferenceheight;
ρ
s–sedimentspecificdensity;
T–non‐dimensionalbedshearstress;
a–referenceheight;
D
50–mediansedimentdiameter;
D*‐non‐dimensionalparticlediameter.
Bedloadtransportrateiscomputedas(7):
0.5 0.7
50
0.006
bss e
SwDMM
(7)
where:
S
b–bedloadtransport(kg/(ms));
M–sedimentmobilitynumber;
M
e–excesssedimentmobilitynumber.
MandMearenon‐dimensionalparameters,given
by(8),(9)and(10):
2
50
1
eff
v
M
sgD
(8)
50
²
1
eff cr
e
vv
M
sgD
(9)
2
²
eff
vvU
(10)
where:
V
cr–criticalflowspeedforparticlemotion,basedon
shieldscurve;
V–depth‐averagedspeed;
U – near‐bed peak orbital velocity, due to wave
action,basedonsignificantwaveheight.
4.2 Modeldescription
Two different grids were used to perform the
simulationsfortheinterestarea.Figure5
presentsthe
grid used for wave propagation simulation and
Figure 6 shows the grid for flow and sediment
transportsimulation. At theinterest area, both grids
wererefinedtoaresolutionofabout20m.Asalready
mentioned, wave and wind boundary conditions
were forced with WaveWatch III and NCEP/CFSR
data,
respectively. Also, flow model boundary was
forced by TPXO global tide model harmonic
constituents, complemented with NCEP/CFSR data
for mean sea level elevation, in order to adequately
representmeteorological effects inwater levelat the
interest area. Bottom roughness was set considering
bedmaterialcharacteristicsandadjustedaccordingto
hydrodynamic
model results’ precision. In addition,
river flow data and total load were inserted in the
model domain as sources, considering each outfall
location.
For sediment transport simulation, two different
sedimentfractionswereconsidered:afineoneanda
coarserone(sand),whichavailabilityandpositioning
withinmodel domainweredefinedaccording
tothe
proportionofsedimentdatapresentedintheStudyof
EnvironmentalImpactoftheAccessChannelCapital
Dredgingcollectedattheinterestarea.
Figure5.Wavegrid
Figure6.FlowandSedimenttransportgrid
4.3 Modelaccuracyobtained
The accuracy of waves representation by the model
was evaluated considering the comparison between
model results and wave measurements at Palmas
(Figure 4) point. RMSE (Root Mean Squared Error)
index was used as a statistical indicator of model
accuracy. The closer its value is from zero, better
is
modelaccuracy. Also,Figure 8,Figure 9 and Figure
10 show the comparison between model results and
field data for current speeds at Praticagem, Palmas
and CPSP points (Figure 4), respectively. Finally,
Figure 11 presents water level comparison between
field measurements and model results. The
comparison between field data
and model results
reveals overall model’s good representation of the
hydrodynamicfieldattheinterestarea.
509
Palmas - Significant Wave Height – RMSE = 0.045 m
Field
Model
Palmas – Peak Wave Period – RMSE = 1.3 s
Time
Significant Wave Height (m)
Time
Peak Wave Period (s)
Field
Mod el
Figure7.FieldXModelComparison‐Waves–Palmas
Praticagem – W/E Component – RMSE = 0.22m/s
Praticagem – N/S Component – RMSE = 0.03m/s
Praticagem – Flow Speed – RMSE = 0.19m/s
Speed (m/s)
Speed (m/s)
Speed (m/s)
Field
Model
Field
Model
Field
Model
Time
Time
Time
Figure8.FieldXModelComparison‐CurrentSpeed–Praticagem
510
Palmas – W/E Component – RMSE = 0.09m/s
Palmas – N/S Component – RMSE = 0.02m/s
Palmas – Flow Speed – RMSE = 0.07m/s
Speed (m/s)
Speed (m/s)
Speed (m/s)
Field
Mod el
Fie ld
Model
Field
Model
Time
Time
Time
Figure9.FieldXModelComparison‐CurrentSpeed–Palmas
CPSP – W/E Component – RMSE = 0.12m/s
CPSP – N/S Component – RMSE = 0.18m/s
CPSP – Flow Speed – RMSE = 0.17m/s
Speed (m/s)
Speed (m/s)
Speed (m/s)
Field
Mod el
Time
Time
Time
Field
Model
Field
Mod el
Figure10.FieldXModelComparison‐CurrentSpeed–CPSP
511
Field
Mod el
Praticagem – Water Level– RMSE = 0.08m/s
Water level (m)
Time
Figure11.FieldXModelComparison–Waterlevel–Praticagem
Sedimenttransportmodelaccuracywasevaluated
by comparing suspended sediment concentration
measured within Santos estuary by DHI (2008) in
March2006.Thisdatawascollectedalongthewhole
accesschannel,extendinguptotheinnerareaofthe
estuary,in11differentsections.Suspendedsediment
concentrationmeasuredwithinSantosestuaryvaries
between0.005kg/m³and0.06kg/m,where0.02kg/m³
is the average suspended sediment concentration.
Figure 12 shows a comparison between field
measurements and modeled data for the mentioned
period, indicating the good relation between model
andreality.Unfortunately,therewerenoconsecutive
bathymetric surveys performed in periods without
dredging
activitiesavailableattheaccesschannel,in
ordertoobtainanidealcalibrationofmorphological
updating.
Figure12. Field X Model Comparison‐Suspended
SedimentConcentration
5 SIMULATIONSPERFORMED
Inordertorepresentsummerandwinterperiod,two
different environmental scenarios were selected for
simulation performance. The adopted scenarios
definition was based on wave statistics analysis,
whichaimedtoselectasummerandawintermonth
thatcouldbestrepresentaveragewaveconditionfor
eachperiod.Figure
13andFigure14exposeselected
wave conditions for summer and winter periods
simulations.Also,forthesummermonthsimulation,
riverflowconditionwasconsideredtobemaximum,
while average river discharge was considered for a
wintermonth.
Monthly simulations were performed to save
computational time. The estimative of annual
dredgingvolumeswas simplifiedby consideringsix
typical months of winter and six typical months of
summer.Theenvironmentalscenariosweresimulated
foreachdesigndepth:currentbathymetry,15,16and
17m.
512
Summer Waves – Total Series (1979-2016)
Selected Month(January1982)
Figure13‐SelectedSummerPeriod
Winter Waves – Total Series (1979-2016)
Selected Month(May 2013)
Figure14‐SelectedWinterperiod
6 RESULTSANDDISCUSSION
6.1 Currentaccesschannelsituation
Figure 15 presents model results of monthly winter
and summer simulations and Figure 16 shows the
annual sediment deposition distribution in each
accesschannelarea.
172,000
18,000
9,000
50,000
249,000
102,000
55,000
27,000
109,000
293,000
0 50,000 100,000 150,000 200,000 250,000 300,000 350,000
AreaI
AreaII
AreaIII
AreaIV
Total
Se d i mentD eposi tio n( m³)
Monthly Simulation ‐ SummerXWinter ‐ Santos PortAccessChannel
Summer Winter
Figure15. Sediment deposition at Santos Port Access
Channel‐SummerXWinter
513
AreaI
1,644,000
51%
AreaII
438,000
13%
AreaIII
216,000
7%
AreaIV
954,000
29%
ANNUALSEDIMENT DEPOSITIONATSANTOSPORTACCESSCHANNEL‐ CURRENT
SITUATION(m³)
Figure16. Annual sediment deposition distribution at
SantosPortAccessChannel
Consideringcurrentbathymetry,modelingresults
indicatedthatthereisanaveragesedimentdeposition
of about 3,250,000 m³ in the Santos Port Access
Channel. About 51% of total sediment accumulation
is observed in Area 1, predominantly in the winter
period,whenwaveactionismoreintense.Withal,in
the summer period, sediment
deposition occurs
predominantly in the inner areas of the access
channel, mainly in Area IV, the closer to rivers
outfalls.InAreasIIandIII,sedimentdepositionrate
is lower, due to higher current speeds caused by
confinedflow.
According to Alfredini (2004), average annual
dredging volume between 1978 and
2002 was about
2,500,000 m³ (in situ). Regarding that, after this
period, the channel was deepened, it is possible to
assume that the model estimative is consistent with
reality.Still,whenanalyzingonlyAreasII,IIIandIV,
which are not subjected to wave action, 70% of
sediment deposition is observed
in the summer
period,followingtherainfallregimeofSantosregion.
The relation between dredged volumes inside
cisternanddredgedvolumesinsituisabout1.33for
Santos Port Access Channel (Alfredini, 2004).
Therefore, according to simulations results, in order
tomaintaincurrentdepths,annualdredgingvolume
shallbeabout
4,325,000m³.
6.2 Designdepthscomparison
Figure 17 exposes the model results for annual
sedimentdepositionconsideringdesigndepthsof15,
16 and 17m (CD) for each area of the Santos Port
AccessChannel.
The model results indicate that sediment
depositionisexpectedtoincrease15%,55%and80%,
considering channel
deepening for 15, 16 and 17m,
respectively.Ahigherincreaseofsedimentdeposition
isexpectedtooccurinAreaI,wherewasobservedan
increaseupto140%whenthechannelwasdeepened
to17m(CD).LessincreasewasobservedinAreasII
and III, where higher current speeds
prevent
sedimentsettling.
Table 1 summarizes modeled results for each
designdepth.
1,644,000
438,000
216,000
954,000
3,252,000
2,088,000
438,000
210,000
1,068,000
3,804,000
3,186,000
438,000
234,000
1,164,000
5,022,000
3,942,000
450,000
276,000
1,236,000
5,904,000
0 1,000,000 2,000,000 3,000,000 4,000,000 5,000,000 6,000,000 7,000,000
AreaI
AreaII
AreaIII
AreaIV
Total
Sedimentdeposit(m³)
AreaI AreaII AreaIII AreaIV Total
17
3,942,000 450,000 276,000 1,236,000 5,904,000
16
3,186,000 438,000 234,000 1,164,000 5,022,000
15
2,088,000 438,000 210,000 1,068,000 3,804,000
Current
1,644,000 438,000 216,000 954,000 3,252,000
Annualsedimentdeposition ‐ Designdepths comparison
17 16 15 Current
Figure17.AnnualSedimentDeposition–Designdepthscomparison
514
Table1. Annual sediment deposition and dredging
predictionforeachdesigndepth
_______________________________________________
Design AnnualSediment AnnualDredging Increase
Depth Deposition(m³) Volumes(m³) (%)
_______________________________________________
Current 3,252,0004,325,000‐
Depth
15m 3,804,0005,059,000 15%
16m 5,022,0006,679,000 55%
17m 5,904,0007,852,000 80%
_______________________________________________
7 CONCLUSIONS
Santos Port is the most important hub port in Latin
America and one of the most important ports
worldwide.Responsiblefor25%ofBrazilianbalance
oftrade,itisalsothemainsourceofincomesofthe
city, thereafter heavily contributing for employment
generation.Therefore,maintainingandimproving
the
harborconditionsinordertoattendtherequirements
forallowingtheentranceoflargervesselsisessential
to keep the port economic attractiveness. In this
context,capitalandmaintenancedredgingstandout
as indispensable activities, although they had been
neglected and poorly managed during Santos Port
history.
The study
hereby presented aimed to provide an
estimativeofsedimentdepositionandannualaverage
maintenance dredging volumes for different
conditionsoftheSantosPortAccessChannel,through
hydrodynamic and morphological numerical model
simulations of the main environmental scenarios.
Althoughthe dataset usedfor modelcalibration can
be considered incomplete for obtaining proper
accuracy, the results presented good relation with
reality,representingoverallmainsedimentdeposition
tendenciesinthedifferentareasoftheaccesschannel.
Themodelindicatesthat,inordertomaintainthe
accesschannelcurrentdepth(March2016),anannual
maintenancedredgingvolumeofabout4,325,000m³
isnecessary.Fordeeper
designdepths,suchas15,16
and 17 m (CD), increases of 15%, 55%, and 80%,
respectively,areexpected.
It is remarkable that, in addition to dredging
volumes forecast, it is necessary to financially
evaluate best design depth for the access channel,
considering international economic scenario and
vessels’ size demand. Also,
whereas that deepening
the channel implies in a significant increase of
sediment deposition, it can be economically or
operationally impracticable to maintain the aimed
design depth without any structural intervention,
such as parallel jetties, which could improve access
channelmaintenancebytidalscour.
REFERENCES
Alfredini, P. 2004. Estudo sobre avaliação das taxas de
evolução do fundo do canal de acesso ao Porto de
Santosnoperíodode1997a2003.TechnicalReport.
CHM. 2016. Centro de Hidrografia da Marinha – Cartas
Náuticas disponíveis.
http://www.mar.mil.br/dhn/chm/box‐cartas‐
raster/raster_disponiveis.html
DELTARES. 2014. Delft 3D‐FLOW‐ Simulation of multi‐
dimensional
hydrodynamic flows and transport phenomena,
includingsediments.UserManual.712p.Delft.
DHI,2008.PortodeSantos–ModelagemdoTransportede
Sedimentos.TechnicalReport.
Fredsoe,J. 1984.Turbulentboundarylayerinwave‐current
interaction.JournalofHydraulicEngineering110:1103‐
1120.
Fundação Ricardo Franco. 2008. EIA da Dragagem de
Aprofundamento
do Canal de Navegação, Bacias de
EvoluçãoeBerçosdeAtracaçãodoPortoOrganizadode
Santos.
Partheniades,E.1965. ErosionandDepositionofCohesive
Soils.JournaloftheHydraulicsDivision,ASCE91:105‐
139.
Van Rijn, L. C. 1993. Principles of Sediment Transport in
Rivers, Estuaries and Coastal Seas. Aqua
Publications,
TheNetherlands.
Versteeg,H.K.;Malalasekera,W.2007.AnIntroduction
toComputationalFluidDynamics–TheFiniteVolume
Method.Harlow,England:PearsonEducation,517p.
