palm island

Abstract

The construction of an artificial island in the shape of apalm tree with a diameter of approximately 5 km infront of the coast of Dubai is nearing completion. To protect the island against wave attack, an offshorecrescent breakwater surrounding the island with a totallength of 11 km was constructed at the same time.After completion, the island will be developed intovirtually self-contained communities including marinas,shopping centre, theme parks, restaurants and so forth.

The Client is Dubai Palm Developers, a subsidiarycompany of the Dubai Ports, Customs & Free ZoneCorporation. The main contractor for the reclamationworks, totalling some 70 million m3 of sand, is VanOord ACZ. The breakwater construction was carriedout under a separate contract awarded to AchirodonOverseas. The contract was awarded to Van Oord ACZat the end of 2001 and works have to be completedend 2003.

One of the main challenges was constructing the sandfill for the island, which had to be carried out partly inunprotected sea conditions, since the breakwater wasunder construction simultaneously because of the tighttime schedule. Therefore an execution methodologywas developed aiming at an optimal schedule in termsof speed of construction and minimal risks of damageand sand losses.

First an inventory was made of the different sandtransport mechanisms i.e. long-shore, cross-shore andwash-over transport and how this would effect thework under construction taking into account a numberof possible execution strategies. From this study, theoptimal execution methodology was derived.Also optimal logistics in terms of cycle times andcombination of placement/rainbowing has beenachieved, by implementing day-to-day survey resultsinto the DGPS tracking system. In this way underwaterfilling is made possible, leaving open sufficient space tomanoeuvre the ships.

Terra et Aqua – Number 92 – September 2003

Rob E. de Jong, Mark H. Lindo, Saeed A Saeed and Jan Vrijhof

Execution Methodologyfor Reclamation WorksPalm Island 1

Figure 1. Artist impression of Palm Island 1.

Introduction

Jebel Ali Properties is developing a prestigious housingand recreation project on new land to be reclaimed inthe Gulf between Dubai City and Mina Jebel Ali. The project, aptly named Palm Island Project, comprisesan artificial palm-shaped island protected at the sea-ward side by an armoured semi-closed oval crescent(Figure 1). The area under consideration has waterdepths ranging between 8-10 m below Jebel Ali CD(tidal range is approximately CD+0.5 m to CD+1.5 m)and an almost horizontal to very mild foreshore. The island itself is built from locally dredged sand. The dimensions of Palm Island are impressive: theperimeter of the crescent is approximately 11 km long,the surface to be reclaimed is about 650 ha and the netsand volume is about 70 million m3. The total timeallowed to construct the island is two years. The required sand is acquired by hopper and cutterdredgers and is deposited in the lee of the ovalcrescent surrounding it. The contractor ArchirodonOverseas is main contractor for the construction of therock armour protected crescent, where Van Oord ACZis the main contractor for the construction of the actualisland (Figure 2).

Since the crescent breakwater and sand-filled islandwere built simultaneously due to time restrictions, theisland was partly unprotected during the first stages ofthe construction. This means that during this construc-tion period the integrity of the island was endangeredby the incoming waves, making the progress andsuccess of its construction strongly dependent uponthe progress of the crescent construction providing asheltered area.

Therefore an optimal execution schedule in terms ofmaximum speed of construction and minimal risks (of damage) was developed by cleverly scheduling theworks taking into account and combining the increasingsheltering effect of the crescent under construction,the relevant sediment transport processes and thevessel characteristics and movements.

SHELTERING EFFECT OF THE CRESCENT

The wave climate can be characterised as generallymild. The most frequent and most intense stormscome from a narrow range of directions in the W-NWsector throughout the months November to April.These are locally referred to as “Shamal” events.Typically wave events with significant wave heights(Hs) of 1-2 m occur rather frequently in this season.Storms with return periods of 5-10 year will producewaves in the order of 3.25 m whilst the 1:100 yearsdesign conditions have been set at Hs=4 m. Stormsurges are limited to approximately 0.5 m above tidallevel (MHHW = CD+1.6 m).

Execution Methodology for Reclamation Works Palm Island 1

15

Rob de Jong obtained his MasterDegree in Civil Engineering from the Technical University Delft in The Netherlands (2001). Thereafter,he joined the Van Oord ACZ Engineering Department where estimating sand loss during the construction of Palm Island was hisfirst major project.

Rob E. de Jong

After graduating in Civil Engineering,Mark Lindo joined FC de WegerInternational Consultants, where hewas involved in the design and reviewof several large-scale hydraulic andcivil engineering projects such asstorm-surge barriers and breakwaterrehabilitation projects. From 1986-1990he was Head of the R&D Departmentof ACZ Marine Contractors and alsopart-time Scientific Officer atTechnical University Delft. Since 1990he is Head of the Engineering Department VOACZ.

Mark H. Lindo

Saeed A Saeed is Director of Projectsat Palm Island Developers, Dubai.

Jan Vrijhof is head of the Estimating &Engineering Department at VOACZsince 1999. After obtaining his degreein Civil Engineering (CoastalConstruction) at the TechnicalUniversity Delft (1979), he joined thedredging industry. Over the last 24years he has worked in many positionsand locations. As project manager hewas responsible for a number of majordredging projects including one of theAirport Core Projects in Hong Kong,the West Kowloon ReclamationProject. In 2001/2002 he was appointedinterim Project Manager during thestart-up of the Palm Island Project.

Jan Vrijhof

To determine the sheltering effect of the crescentunder construction numerical wave computations havebeen carried out with the 2-dimensional numericalwave model SWAN. Since diffraction equations are notyet modelled in SWAN, an increased directional wavespreading has been applied in the SWAN wavecomputations. The solution was tuned using thediffraction examples provided in the Shore ProtectionManual [1] and gave satisfying results for this situation.

The wave computations were carried out for acombination of 6 different wave directions, 6 differentwave heights and 18 different lengths of the crescentunder construction. Thus a total of 36 (6 x 6) wavecomputations have been performed for 18 crescentlengths, hence a total of 648 computations. Thecompletion dates for the various crescent chainages andthus crescent lengths were derived from the planningof the breakwater construction (Figure 3).

The ratio between the computed wave height and theboundary wave height give so-called transformationratios. These transformation ratios were combined withthe nearshore monthly wave climates to determine themonthly wave climates for the various stages of thecrescent construction. For each phase of the crescentconstruction it was thus possible to estimate thesheltering effect on the average wave conditions bycomparing the wave climate as computed with andwithout the crescent (for each specific location,relevant month and accompanying crescent length).

SEDIMENT TRANSPORT PROCESSES

When waves attack the partially completed sand island,they will move sand out of the predefined boundariesof the fronds and trunk (Figure 1). Especially the endsof the fronds will experience losses, since they will losesand by a combination of littoral (long-shore) and

perpendicular (cross-shore) sand transport, whilst theyare the least protected by the crescent during theconstruction phase and are more vulnerable to adverse3-dimensional effects. Furthermore, there is no naturalsand supply. The removed sand is thus permanentlylost. This means that either the lost sand must bebrought back into the profile or more sand must beborrowed. It is therefore very important to estimatehow much sand will be transported outside the finalprofiles by these waves.

To be able to give a rough assessment of the anticipatedsand losses, the sand transport generated by waveswas quantified using simple but transparent morpho-logical models. It is emphasised that these models(cross-shore and long-shore) were made for uniformstraight beaches and sandbars are not valid for areassuch as the end-section of the fronds. These morpho-logical models are discussed below making a distinc-tion between two fundamentally different situations:1. Crest level lower than the wave run-up level

(wash-over transport).2. Crest level above the wave run-up level

(cross- and long-shore transport)

For the calculations use has been made of theexpertise and/or models of WL | Delft Hydraulics,Alkyon, Professor Bijker and VOACZ’s in-houseexpertise and models.

WASH-OVER TRANSPORT

When the crest level is lower than the wave run-uplevel, waves will wash over the created berm, that canthan be seen as a sand bar. This sand bank will reshapein time due to sand transport by waves and currents.Three sub-mechanisms for this wash-over transportcan be distinguished. For each of those systems thesand grains are mainly stirred up by the wave-induced

16

Figure 2. Trailing suction hopper dredger Volvox Atalanta “feeding” the Palm with the Burj-Arab Hotel in the background.

the partially constructed fronds. These increased flowvelocities, in combination with the expected local wave climate were then used to estimate sedimenttransport rates at various locations in the project area.These sediment transport rates were determined usingformulations of Van Rijn [2], which have beenimplemented in the profile model UNIBEST-TC by WL | Delft Hydraulics.

The calculations showed that the sand losses duringthe winter are dominated by the most severe storms.Especially when breaking of the wave starts the trans-port rates increase considerably. The actual durationand severity of these storms may differ considerably

orbital flows. The origin of the current that is required totransport the stirred-up sand grains, however differs.These currents are:1. Tidal current parallel to the shore2. Down-slope directed density currents3. (Breaking) wave-induced current

For the assessment of the wash-over transport thelocal bathymetry and the complete submerged Palm Island was taken into account. Two levels of the submerged island were considered: CD–4 m andCD–6 m. The breakwater under construction was nottaken into account. A 3D flow model was used to getan indication of the increased tidal flow velocities over


17

Figure 3. Typical results SWAN wave transmission calculations for 6 execution stages of the crescent (offshore significant waveheight 2.25 m, mean wave direction as indicated by the arrow).

FEB 2002 MAY 2002

AUG 2002 NOV 2002

FEB 2003 MAY 2003

= non-constructed part breakwater

= constructed part breakwater

Hs [m]

15

from one winter to the other. Changing the incomingwave height +/–10% resulted in a +/–300% change inthe calculated transport rates. This means that theassociated sand loss may differ dramatically from onewinter to the next.

The berm level also influences the number of wavesthat are forced to break. Accordingly, the calculatedtransport rates for the berm level of CD–6 m wereconsiderably lower (order 10 times) than for the bermlevel of CD–4 m.

The calculation also showed that the sand losses aredominated by cross-shore transport. Not the tidalcurrent parallel to the shoreline, but the wave drivencurrents (perpendicular to the coastline) over thepartially constructed fronds are dominant for theexpected sand losses. Unfortunately these transportrates are very sensitive for the calculated near-bottomflow velocities, which in their turn depend on themodelled (sand) bed roughness. The bed roughnesswas not exactly known. When the bed roughness wasvaried between 1 cm and 10 cm, the calculated cross-shore current velocity varied between 1 m/s-1.5 m/s.

For this range in current velocity the calculatedsediment transport rates differed a factor 10. The magnitude of the sediment transport is howeverprincipally not equal to the losses, since part of thetransported sediment will resettle in the eventuallyrequired profile.

During the construction, the reshaping of thesubmerged sand bars were monitored. The measure-ments indicated that reshaping in case of a crest levelof about CD-4 m only occurs during extreme conditionsconform theory. The reshaping for this crest level is farless than in case of a crest level above the water level.

The calculated transport rates are very dependent onthe wave height. The real wave climate outside thebreakwater during the first winter period (2001-2002)was milder than average. This mild winter waveclimate would result in considerably lower calculatedsand transport since the losses are dominated by the highest waves with only a small probability ofoccurrence. These low transport rates were indeedrecorded.

LONG-SHORE TRANSPORT

For the berm with a crest level higher than the run-up level of the waves, the waves are blocked. Two transport directions are distinguished for thissituation: transport parallel to the berm (long-shore) and transport perpendicular to the berm (cross-shore).Three methods to calculate the long-shore transportrates were compared.

CERCThe CERC formula is commonly used to estimate thelong-shore sediment transport. It is an empirical relationbetween the waves and the long-shore transport forrelatively long and straight beaches, where the along-shore differences in the breaking waves are small. The CERC formula can be given as:

(1)

S long-shore sand transport [m3/s]A dimensionless coefficient [-]Hsig significant wave height [m]c wave celerity [m/s]cg wave group velocity [m/s]n ratio cg to c [-]ϕ angle between the wave

crests with the shoreline [°]

Subscript “1” indicates that the dimensions at a waterdepth of 10 m are used. Subscript “br” indicates thatthe dimensions at the breaker line are used.

In the Shore Protection Manual, a value of A = 0.050 isderived based on measurements on beaches which canbe characterised by a D50 of about 200 µm. At the PalmIsland project location sand of about 400 µm is present.Larger grain result in lower transport rates, the value ofA was therefore adapted for the project location.

The effect of tidal current on the transport rates cannotbe incorporated in the CERC formula. The tidal currentvelocities at the project location are however limited toextremes of 0.25 m/s to 0.30 m/s, so the error ofneglecting them may be limited here.The beach slope strongly effects the distribution of thelong-shore transport across the breaker zone. The effecton the total long-shore transport is however limited,since a steeper slope means a narrower breaker zone,but on the other hand a more (energy dissipating)intensive breaker zone. The net effect is a slightincrease in the long-shore transport in case of a steeperslope (Bijker [3]). The effect of neglecting the slope at allis therefore expected to be limited as well.

BIJKER (1971) AND VAN RIJN (1993) Alkyon calculated the long-shore transport for severalincident wave directions with respect to the normal onthe coastline using the transport model UNIBEST-LT.The following input data was used:– Slope of 1:4– Constant tidal current of 0.1 m/s– A constant water level of CD+1 m– D50 = 400 µm– Bed roughness = 0.05 m

For the computations the Bijker [3] and Van Rijn [2]transport formula for sand were applied.


18

( ) ( )1112

1, sincos ϕϕ ⋅⋅⋅⋅⋅= brsig cnHAS

CROSS-SHORE TRANSPORT

In case the crest level is above wave run-up level notonly long-shore transport occurs, but also cross-shore

For several significant wave heights (Hs) with a waveapproach angle of 45° the long-shore sand transport as calculated using the CERC, Bijker and Van Rijnformulation are shown in Figure 4.

The long-shore transport rates calculated with CERCand BIJKER are of the same order of magnitude (within the morphological accuracy factor of 2 to 3).

The VAN RIJN transports are approximately 100 timeshigher than the transports calculated with the othertwo formulas. For more gentle slopes lower transportrates are found with VAN RIJN, which is in contradic-tion with the measurements by Bijker [3] that indicatethat the slope has very little impact on the total long-shore transport.

As the long-shore sediment transport rates as calculatedwith CERC and BIJKER are in good agreement and theCERC formula is simpler and faster, the CERC formulahas been used for the determination of the resultingmonthly long-shore sediment transports.


19

0

1

2

3

4

5

0 1 2 3 4 5

Hs [m]

Lon

g-sh

ores

and

tran

spor

t[m

3/s]

CERC

BIJKER

VAN RIJN

Figure 4. Calculated long-shore transport rates for a 45° waveapproach angle.

SWL

Crest line

SWL

Transition zone Active zone Backshore

ho

hm

Figure 5A. Typical cross-profile before exposure to waves. Figure 5B. Typical cross-profile after exposure to waves.

transport. In case of cross-shore transport (perpendicularto the coastline) sand will be moved from the slopedownward (and to a lesser degree upward) and thus agentler slope will develop in time. The crest line willshift in shoreward direction and sand will depositoutside the required profile.

With several cross-shore transport models, the shapeof the foreshore (slope) for various wave conditions andsand characteristics can be estimated. Also a predictionof the time-dependent development of the profile canbe made. Eventually a more or less equilibrium profilewill develop.

In case of exposure to waves, the coastal zone can be divided in 3 different zones as shown in Figure 5B.The active zone is the zone that is directly influenced by wave action. The transition and backshore zones arenot directly influenced by the waves.

The upper boundary for the active beach profile hmtheoretically equals the wave run-up level above stillwater level. As a result of the (tidal) variation of the stillwater level, the active zone varies in time.

In the breaker zone a lot of sand is in suspension andconsiderable changes in the profile may take placewithin hours or days. Seawards of the breaker zoneseasonal profiles can occur as a result of seasonalchanges in the wave climate. Therefore, the actuallower boundary (hm) for the active zone is dependenton the time scale that is considered.

The cross-shore transport in the active zone is difficult toquantify. Three models have been applied to estimatethis cross-shore transport:

SWART’S MODELIn the model of Swart it is assumed that for a certainsand grading (characterised by its median grain diameterD50) and for certain wave conditions (characterised bythe wave height and wave period) an equilibrium profilewill develop (as shown in Figure 5B). It takes sometime to develop this equilibrium profile. The rate ofchange of the profile is proportional to the difference inshape of the existing and the equilibrium profile. The larger this difference in shape, the faster initialprofile changes takes place.

Swart’s model (see [4], [5] and [6]) gives empiricalrelations to determine the equilibrium beach profile and cross-shore sand transport as a function of thewave height, the wave period and the grain size. These relations are mainly based on a large number ofsmall-scale (mainly regular wave) model test studiesbut are validated with prototype measurements.

In [5] also an empirical relation for the speed at whichthe equilibrium profile is reached, is given.

For the Palm Island project a translation was made fromthe regular wave relations as presented by Swart toirregular wave conditions. Moreover, the impacts of the(tidal) still water variations were taken into account byextending the range of the active zone (see Figure 5B).This modified Swart’s model enabled the calculation ofthe time-dependent beach profile development.

DUROSTAThe estimated erosion of the cross-shore profile wasalso calculated by Alkyon using the DUROSTA model.This model was developed for computing the offshore-directed sediment transport of a (steep) dune profileduring storm conditions. The DUROSTA model istherefore assumed suitable for computing the erosionprocess along the steep initial slopes of the Palm Island.

UNIBEST-TCUnibest-TC is the cross-shore sediment transportmodule of the Unibest Coastal Software Package, asoftware program developed by WL | Delft Hydraulics.It is designed to compute cross-shore sedimenttransports and the resulting profile changes along anycoastal profile of arbitrary shape under the combinedaction of waves, long-shore tidal currents and wind.The model allows for constant, periodic and time seriesof hydrodynamic boundary conditions to be prescribed.

Indicative calculations were made using the modifiedSWART, DUROSTA and UNIBEST-TC model tocompare the results. In this indicative calculations thefollowing profile was modelled:– Crest level CD+3 m– Flat seabed level CD–9 m– SWL at CD+1 m (no tidal variations were taken into

account)– Initial profile was assumed to have a 1:4 slope– D50 = 400 µm

In Figure 6 the calculated time-dependent regression ofthe crest line (see Figure 5A) for all three models isplotted.

From Figure 6A it can be seen that especially theestimated regression speed during the first few daysdiffers considerably. The reason for this might be thatboth the UNIBEST-TC and the SWART models are notderived for the steep initial slopes as are present at thePalm Island project. Swart [4] mentioned that the timedependent calculation is inaccurate in the situation ofvery steep slopes, but without quantifying when aninitial slope is too steep. In the situation of steep slopesin combination with smaller waves, the horizontaldimension of the breaker zone becomes small whichalso results in instabilities in the UNIBEST-TCcalculations. DUROSTA was developed to model duneregression in case of severe storms. During thisregression steep slopes are present. The initial slopefor the modelled conditions will however normally be


20

(a few hundred metres) fronds, which are affected byhead effects (see next section). This seems to beconfirmed by the fact that the sand that was washedaway from the higher parts of the slopes was notdeposited at the lower part of the slope, but was totallyremoved from the profile. The MODIFIED SWARTmodel assumes a long strait frond with constant long-shore transport. Therefore the sand balance is closedfor this model, resulting in more sand down slope.

SEDIMENT PROCESSES AROUND END

SECTIONS OF FRONDS

As mentioned before the sand transport models havebeen applied for uniform, straight slopes; no boundaryeffects have been incorporated. With some engineeringjudgement the models can be applied for the gentlycurved fronds, taking into account the changing incidentwave angle. However the erosion pattern for theunprotected ends of the palm tree fronds is morecomplicated.

far smoother (beach profile) than present here, so itcannot be guaranteed that the model is suitable for thissituation.

From Figure 6B, it can be seen that the UNIBEST-TCmodel results in a far smaller regression speed than theDUROSTA and MODIFIED SWART model. During theerosion, parts of the steep fronds will slide into sea dueto the (too) steep slopes and wave run-up. This processis not modelled in UNIBEST. Therefore the erosion ratefor high waves (where this sliding occurs frequently) canbe expected to be underestimated by UNIBEST-TC.

The results for the DUROSTA and MODIFIED SWARTare within a margin of a factor 2-3 that is usually appliedfor the accuracy for sediment transport calculations.Both methods show considerable sand loss. For practi-cal reasons the MODIFIED SWART model was used tocalculate the profile changes as a result of the localwave climates as calculated using the SWAN wavemodel. These calculations show that the smaller wavesare not of importance for the ultimate beach profilewhich develops after a month. This profile is primarilydetermined by the higher waves (Hs > 0.5 m).

As soon as the first frond emerged and a storm tookplace, the profile deformations were measured to verifythe models used and update the dump strategy ifrequired. The effects of the storm (about 6-8 Beaufort)as occurred on April 4th 2002, with an estimated dura-tion of 12 hours and with a significant wave height nearthe central top branch of about 1.25 m, was used forthis. The disadvantage of this early measurement wasthat the frond length above the water was limited to afew hundred metres. This means that no long straightuniform beach was present, resulting in head effects(see next section). Moreover, the frond of investigationwas still under construction so that newly depositedsand also influenced some of the cross-shore profiles.Nevertheless, the real cross-shore sections before andafter the storm could be schematised as presented inFigure 7. The profiles after the storm were alsocalculated using the MODIFIED SWART model (withtidal water level variation included) for the first twosituations with no overtopping this model can be usedfor (Figure 7A and 7B).

The measured cross-sections after the storm showthat the amount of sand transported from the profile fora crest level at the still water level (Figure 7C) is farlarger than when this crest level is brought up higherbefore the storm occurs (Figure 7A and 7B). Themeasurements show that considerable regressions ofthe crest line can indeed be expected in relative shorttime spans. The real deformations were in goodagreement with the predicted ones. The measuredregression exceeded the calculated regression a little.This was probably caused by the fact that themeasured profiles were taken from relative short


21

-50

-40

-30

-20

-10

00 500 1000 1500 2000

Duration of exposure to Hsig=1m waves [hours]

Cre

st li

ne r

egre

ssio

n [m

]

MODIFIED SWART

UNIBEST-TC

DUROSTA

Figure 6A. Calculated crest line regression for significant waveheight of 1 m. Figure 6B. Calculated crest line regression for significant waveheight of 3 m.

-150

-125

-100

-75

-50

-25

00 20 40 60 80 100 120 140

Duration of exposure to Hsig=3m waves [hours]

Cre

st li

ne r

egre

ssio

n [m

]

MODIFIED SWART

UNIBEST-TC

DUROSTA


22

Crest level at CD+2.5 m

-9-7-5-3-113

-40 -20 0 20 40 60 80 100Pre-storm (April 3rd) Post-storm (April 5th) Post-strom calculated

Figure 7A. Typical measured and calculated profile deformations for crest level at CD+2.5 m.


-9-7-5-3-113

-40 -20 0 20 40 60 80 100Pre-storm (April 3rd) Post-storm (April 5th) Post-strom calculated

Figure 7B. Typical measured and calculated profile deformations for crest level at CD+2.0 m.


-9-7-5-3-113

-160 -140 -120 -100 -80 -60 -40 -20 0 20 40 60 80 100

Pre-storm (April 3rd) Post-storm (April 5th)

Figure 7C. Typical measured and calculated profile deformations for crest level at CD+1.0 m.

and cross- and long-shore computations it becameclear that when building up the sand body above thewaterline in unprotected water, the cross- and long-shore transport would result in unacceptable sandlosses. During a rough winter, unprotected frondscould even break through. The profile deformations aredominated by the (short term) extreme conditions theyare exposed to.

Frond ends that were not sheltered by further offshore-located fronds were subjected to all three erosionphenomena: cross-shore, long-shore and wash-overtransport (see Figure 8). The wash-over transport takessand from the far frond ends to the frond back slopewhere some kind of sand spit will be formed. The obliquely incoming waves will generate cross-shore and long-shore transport although it is expectedthat long-shore transport from the far frond tip will beminimal. First a certain beach length is required forsand to be suspended over the water column beforetransport takes place. Therefore, it was expected thatthe frond-tips would be mainly subjected to wash-overand cross-shore transport. The result thus will be thatthe frond tips will be lowered and stretched. Further towards the spine where long-shore transportpicks up, the width of the frond will be reduced as sandis transported away and deposited in more shelteredwaters behind the previous frond.

It may be clear that those sections are very vulnerableto losses, which in turn are hard to predict. Thereforethese ends were constructed only when protectedsufficiently.

EXECUTION PHILOSOPHY

From the combined results of wave propagation studies


Figure 8. Erosion phenomena frond ends.

Figure 9. Optimum production and safety are obtained by keeping corridors and space open for manoeuvring and constantmonitoring of the progress.

When keeping the crest level sufficiently deep belowthe water level, the deformations only occur duringextreme storm conditions. When staying sufficientlydeep, the overall profile deformations owing to waveaction will be far more limited than in case of a crestabove water.

Since repair of damaged sections would have beendisproportionately expensive, an execution methodologywas developed which was first of all based onminimising the sand transport by waves and currents.In addition to this also the following requirements weretaken into account:– The inevitable sand transport should resettle within

the final profiles as much as possible.– Optimal logistics should be achieved in terms of

cycle times and combination of dumping and rainbowing taking into account ship restrictions(draft, maximum rainbow distance).

– Production capacity and planning should meet thetime of delivery.

– Safety of the operations should be ensured.

EXECUTION METHODOLOGY

Based on the combined results of wave propagationstudies and cross- and long-shore computations, the client could be convinced that the construction ofPalm Island itself should be closely related to theprogression of the breakwater, since the protectionprovided by it, was essential. Therefore the following

execution methodology was developed based on therequirements mentioned above, the computationresults and the progression schedule of the breakwater.

During first winter – Especially at the beginning of the first winter,

starting at the end of 2001, the sheltering of thepartly constructed breakwater was very limited (see Figure 10). The profiles therefore remainedbelow the CD–4 m during the first winter, since thisresults in lower transport rates then when comingabove water.

– The width of the fronds at this stage was keptslightly smaller than required to allow somereshaping without material ending up outside theeventual required profile.

– In between the sand bars, corridors and space formanoeuvring is kept in order to allow the dredgersto operate in a safe and efficient manner during theconstruction of the fronds.

– The TSHD had such dimensions that this sand couldbe dumped and did not have to be rainbowed.

After first winter– Based on scheduled progress of the breakwater and

offshore wave climate, the wave and cross- andlong-shore sand transport model were used todetermine which fronds were sheltered enough atwhat stage. This way the fronds were given free toconstruct above water one by one, starting at themost protected top end of the Palm.

– The sequence of the filling was anti-clockwise from


24

Figure 10. Satellite pictures of Palm Island showing the progress at approximately 3-month intervals.

8 DEC 2001 22 MAR 2002 26 jun 2002

22 SEP 2002 7 JAN 2003 25 MAR 2003

Van Oord ACZ determined the right strategy of stayingunderwater the first winter and only raised thosefronds above water which had sufficient protectionfrom the crescent breakwater from then on.

References

[1] Coastal Engineering Research Center (1984).Shore Protection Manual. U.S Government Printing Office,

Washington, DC, USA.

[2] Rijn, L.C. van (1993).Principles of Sediment Transport in Rivers, Estuaries

and Coastal Seas. Aqua Publications, Amsterdam,

The Netherlands.

[3] Bijker, E.W., (1971).“Longshore transport computations”, Journal of

Waterways, Harbours and Coastal Engineering Division,

Vol. 97, No. WW4.

[4] Swart,J.H., (1974). “Offshore sediment transport and equilibrium beach pro-

files, report on model investigations”. Report number M918,

Part II.

[5] Swart, J.H., (1976). “Predictive equations regarding coastal transports”,

ASCE Proceedings 15th Conference on Coastal Engineering.

Hawaii. Vol. II. Chapter 66.

[6] Graaff, J. van de, (1978).“Transport of sand perpendicular to the coast”.

Course Coastal Dynamics and Coastal Protection CT.KK5,

Foundation of Postgraduate Education for Civil Engineering.

Delft, The Netherlands. (in Dutch).

West to East as to allow the TSHD to reach thereclamation fronds via the anticipated corridor in theEastern part of the crescent breakwater with mini-mum obstruction from already constructed sand fillallowing maximum production of each dredger in asafe manner.

– Optimal logistics in terms of cycle times andcombination of placement/rainbowing has also beenachieved, by implementing day-to-day survey resultsinto the DGPS tracking system. In this way safeunderwater filling was made possible, leaving opensufficient space to manoeuvre the ships.

From the satellite pictures in Figure 10, which weretaken with a time interval of about three months, it canbe seen that the planned methodology has indeedbeen applied. On the first two pictures even the under-water berms can be seen from space. It is also clearthat the fronds are constructed one by one starting atthe top of the Palm. Only the top of the trunk wasraised above the water before schedule. This was doneat the request of the client.

Conclusions

A thorough study of the mechanisms responsible forpossible sand losses from the reclamation area prior toexecution of the works gave a good insight in the risksand provides tools for defining the best strategy for theexecution of the works.

Cross-shore transport resulted in considerable slopedeformations, once the reclamation area is abovewater even with limited wave action during a limitedtime, which is in accordance with the calculationscarried out.

These deformations are even much more severe, if thecrest level is raised to just below the wave run up level.Therefore, if going above water, the final crest levelshould be reached as soon as possible.

The frond ends experienced similar but more severedeformations (combination wash-over/long-shore/cross-shore transport) as anticipated.

Reshaping owing to wash-over transport with a crestlevel of about CD–4 m only occurred during extremeconditions conform theory. The reshaping was however far less than in case of a crest level above thewater level.

Especially at the beginning of the winter the shelteringof the partly constructed breakwater was very limitedand the exposure of the fronds would have been large.Since the real reshaping in case of a crest at least CD–4 m was indeed considerably less than in case of acrest above the water level, it can be concluded that


25

palm island

Documents

dgps tracking

van oord acz

wave propagation

technical

significant

significant

shore protection

lower transport