INTERNATIONAL SOCIETY FOR SOIL MECHANICS AND GEOTECHNICAL ENGINEERING

This paper was downloaded from the Online Library of the International Society for Soil Mechanics and Geotechnical Engineering (ISSMGE). The library is available here: https://www.issmge.org/publications/online-library

This is an open-access database that archives thousands of papers published under the Auspices of the ISSMGE and maintained by the Innovation and Development Committee of ISSMGE. A performance based design for the coastal structures and breakwaters,

J.D. Adrichem & A.G. Wiggers Royal HaskoningDHV, Nijmegen, The Netherlands Dr. E. Güler ELC Group Inc. & Boğaziçi University, Civil Eng. Dept., , Turkey G. Buitenhuis Royal HaskoningDHV, Dubai, United Arab W.J. Karreman Van Oord Dredging and Marine Contractors B.V., Rotterdam, The Netherlands

ABSTRACT The Deira Islands project is located between Port Rashid and Mamzar Beach, at the northern border of Dubai, and comprises four islands of sand reclamation. The project included the finishing of revetments and quay walls, adapting structures into beach and the construction of a number of new structures including revetments, groynes and breakwaters. In addition, two offshore breakwaters with a total length of 2 kilometers were to be constructed. As the Dubai region is located in a moderately active seismic zone, seismic stability was of importance. The generation of excess pore pressures and potential liquefaction of the foundation soils and/or reclamation material, consisting of (siliceous) carbonate sands, created a challenging environment for which the edge structures and breakwaters were to be designed.

The revetment and breakwater designs were governed by seismic performance criteria in terms of allowable deformations. The breakwater designs were particularly challenging, as their geometries were to include an untreated, hydraulically placed sand core. The performance criteria did however allow for large vertical deformation of the breakwater crests, which could even allow for displacements induced by lateral spreading. By adopting a fit for purpose performance-based design approach, optimal use was made of the opportunities provided by the performance criteria to achieve the most economical design.

1 INTRODUCTION 2 THE PROJECT

The Deira Islands are a group of artificial islands situated 2.1 History adjacent to the Deira district in Dubai, . The region is located in a moderately active The Deira Islands project is located between Port Rashid seismic zone, which made seismic stability and and Mamzar Beach, at the northern border of Dubai, and liquefaction risk important factors to be considered during comprises four islands of sand reclamation. The Deira the design and construction of the islands and their Islands were reclaimed between 2005 and 2008 as part of coastal structures. the development, at which time the edge The design of two rubble mound breakwaters was structures were largely completed. particularly challenging, as their geometries were to In 2015, Van Oord Dredging and Marine Contractors include an untreated, hydraulically placed sand core that was awarded the Deira Islands project. They sub- would be susceptible to liquefaction as a result. The contracted Royal HaskoningDHV to provide the detailed performance criteria for earthquake loading did however design for the 25 kilometre long Deira Islands Coastline - allow for relatively large permanent vertical deformations Stage 1 construction works, based on a master plan of the breakwaters. proposed by the master developer. The reclaimed islands The importance of adopting a performance-based will be the future location of new hotels, serviced design for the Deira Islands project was recognized in an apartments and marinas accommodating more than 500 early stage of the tendering process and turned out to be yachts and boats. vital in securing the tender. This paper focusses on the subsequent and successfully completed seismic 2.2 Scope geotechnical design for the project and provides an overview of the applied performance-based design The scope of the project included the revitalization of the approach. The approach provided an economical design coastline of two islands, referred to as Islands A and B. by making optimal use of the opportunities provided by Partially constructed quay walls and revetments were the prescribed performance criteria. completed, new revetments, groynes and 8.5 kilometers

of beaches were designed and constructed and an Serviceability Limit State (SLS) conditions, the specified existing land bridge between Deira Corniche and Island A design PGA at bedrock level was 0.05g. was removed. In addition to the works on the coastline, the scope 3.2 Ground conditions included the design and construction of two breakwaters with a total length of 2 kilometers between Islands C and The near-surface geology of Dubai is dominated by D. Figure 1 includes an illustrative overview of the scope aeolian dune sand deposits of Holocene to Pleistocene for the Deira Islands project. age. In the nearshore coastal zone of the project area, the superficial deposits are relatively thin and consist of 2.3 Execution shelly, (slightly) silty (siliceous) carbonate sands, which can contain random bands or lenses of sandy silt. Below Reclamation works for the Deira Islands were executed a shallow, loose, sand layer, the relative density typically by trailing suction hopper dredger. The sand used for the increases to dense. The carbonate content of these reclamation works was obtained from an earlier deposits generally ranges between 82% and 95%. reclamation at the Trunk Island, located between Islands The degree of cementation is variable and generally C and D. The majority of the rockworks for the revetments increases with depth, such that the variably cemented and breakwaters were performed using floating barges, sand grades to predominantly calcareous sandstone and while the rock armor was partly placed from barges and calcarenite. At many locations, moderately strong to partly from land. strongly cemented layers, commonly referred to as caprock, are present within the superficial deposits. The dredged sand fill material has the same properties as the in-situ superficial sand deposits because of their similar origin. The fill was classified as slightly silty to silty, slightly gravelly to gravelly, fine to coarse sand. The sand particles are generally of low sphericity and sub-rounded to angular in terms of angularity. The Deira Islands structures are located up to a distance of 5 kilometers from the original coast line, where Breakwaters C & D are constructed. Original seabed level at the breakwaters varies between -13m and -15m Dubai Municipality Datum (DMD).

3.3 Ground investigations

Both terrestrial and marine geotechnical ground investigations were carried out for the Deira Islands project. The scope of the investigations included boreholes with an exploration depth of 5 m into bedrock. Figure 1. Overview of the Deira Islands project site and Cone Penetration Tests (CPTs) were performed in the structures. superficial deposits and were, at various locations, continued below caprock after pre-drilling. Additional vibrocores were carried out to sample the fill material for 3 SITE CONDITIONS the breakwater core. A correction to the CPT records was necessary, as 3.1 Seismic hazard research has shown that the relative density derived from CPT records in carbonate sands is underestimated when The Dubai region is located in a moderately active using correlations developed for silicate sands. Wehr seismic zone. Seismic activity in this region is controlled (2005) established a specific shell correction relationship by the collision between the Arabian Plate and the for sands, which was deemed appropriate Eurasian Plate, creating the Zagros collision zone in Iran for the Deira Islands project. The shell correction factor is and the Makran subduction zone in the Gulf of Oman. The to be multiplied with the measured cone penetration nearest proximity of the Zagros Belt to Dubai is resistance qc before applying conventional silicate sand approximately 140 km. The distance between the Makran based correlations between CPT records and relative subduction zone and Dubai is approximately 220 km. density such as developed by Baldi et al. (1986) or Bellotti The project performance specifications prescribed that & Jamiolkowski (1990). the earthquake loadings and seismic analyses for the Downhole seismic tests were performed to determine marine structures should be determined in accordance shear wave velocity profiles. Besides standard with Eurocode 8. For Ultimate Limit State (ULS) classification tests, CD-triaxial tests and shearbox tests conditions, the design peak horizontal ground were carried out to determine shear strength parameters acceleration (PGA) at bedrock level was specified as for the carbonate sands. The advanced laboratory testing 0.15g based on a 475 year return period. A design program included cyclic triaxial tests and resonant column earthquake magnitude, MW, of 6.0 is used. For

tests to investigate liquefaction susceptibility and dynamic thethe breakwaters breakwaters should should have have a a relative relative density density of of behavior. generally not less than 40%, with horizons of not more thanthan 0.5m0.5m inin thicknessthickness with a relative density of not less than 35%. 4 PERFORMANCE CRITERIA

4.1 Revetments 5 GROUND RESPONSE

The performance specifications specified maximum 5.1 Earthquake record selection earthquake induced displacement criteria based on Newmark sliding block analyses for three points along the Records were selected from the PEER/NGA database revetments, as summarized in Figure 2 and Table 1. that matched the design moment magnitude (MW) and PGA as good as possible. In choosing the appropriate records, the following selection criteria were used: MW A. Crest line of crest wall between 5.5 and 6.5; PGA between 0.135 and 0.165; B. Toe of revetment monitoring instruments on rock or shallow stiff soil. A C. Toe of berm This selection resulted in 10 pairs of time histories, which were used for the one dimensional ground responseresponse analyses.analyses. TheThe averageaverage response of these time histories lies slightly below the Eurocode 8 design spectrum, as illustrated in Figure 3. These response C B spectra show the pseudo spectral acceleration for 5%

Figure 2. Reference points for revetment damping. displacement criteria.

Table 1. Revetment displacement criteria.

Point SLS displacements [m] ULS displacements [m] Horizontal Vertical Horizontal Vertical A <0.15 <0.10 <0.30 <0.20 B <0.50 <0.25 <1.00 <0.50 C <1.00 <0.25 <2.00 <0.50

In addition to the Newmark deformations, the “shake- down” settlement of the reclamation fill for partial liquefaction and dry soil particle re-arrangement is determined. The maximum permissible settlement from Figure 3. Pseudo spectral acceleration chart for selected these combined effects was 150 mm for ULS and 50 mm earthquakes. for SLS conditions. The post-earthquake stability check (static conditions after the earthquake) included fully developed excess For the finite element modelling (FEM), a sub- pore pressures at the end of the earthquake. The selection of three earthquakes was made, consisting of: prescribed Factor of Safety for stability of the slope for the Whittier Narrows-01 Earthquake, LA-Baldwin Hills Station post-earthquake condition is FSpost EQ = 1.05, which would record (EQ 627); Chichi Taiwan 06 Earthquake, CHY028 ensure that no excessive displacements would develop Station record (EQ 3268) and Chichi Taiwan 06 immediately after the occurrence of the earthquake. Earthquake, CHY035 Station record (EQ 3274).

4.2 Breakwaters 5.2 Ground profile

The seismic performance specifications for the The measured shear wave velocities in the various breakwaters were less stringent in comparison to those layers were: 250 m/s to 337 m/s for reclamation sand fill; for the revetments. 446 m/s to 562 m/s for natural (cemented) sand and 557 In SLS, the maximum permissible crest settlement as m/s to 718 m/s for the calcarenite bedrock. 1000mm and damage repairable within 6 months. In ULS, G/Gmax degradation curves and strain dependent the requirement for the crest to remain at or above Mean damping ratios (for the sands were determined by Sea Level (MSL) resulted in a maximum permissible crest resonantresonant columncolumn teststests. Selected parameters are given in settlement of 3650mm with repairable damage. Table 2. The sand samples were prepared in three In addition to the displacement criteria, the different relative densities (Dr) to represent the conditions specifications required that the sand fill in the sand core of of various sand layers.

Table 2. Dynamic parameters from resonant - Consolidated Drained (CD) triaxial tests and column tests. direct shear box (DS) tests show that the Deira Islands sands do have a relatively high critical 1 2 Dr Gmax 0.7  state friction angle compared to silicate sands. (%) (MPa) (%) (%) Friction angles vary from 36 to 49 degrees for a 12 89.9 1.2 2 range of relative densities (Figure 4). 28 111.1 0.8 3 - CD triaxial tests hardly show any effect of shear 52 142.2 0.7 4 strength reduction at high strain levels (beyond 1 the peak strength). shear strain at G/Gmax = 0.7 2 damping ratio at  = 1% - Cyclic triaxial tests results on Deira Islands sands show that the carbonate sands in this region have a relatively high resistance against 5.3 Soil amplification factors liquefaction compared to literature on silicate sands of similar relative density. This finding is For the revetment structures, one-dimensional equivalent discussed in more detail below. linear ground response analyses were performed with SHAKE2000, resulting in soil amplification factors between 1.3 and 1.6. For the breakwaters, both one- dimensional equivalent linear ground response analyses and non-linear time history analysis were performed to account for soil amplification. Section 7 of this paper provides a comparison of results.

6 LIQUEFACTION SUSCEPTIBLITY OF CARBONATE SANDS

6.1 Carbonate sands

The sand deposits overlying cap rock in the Deira Islands are predominantly carbonate sands with a carbonate content of about 90%. Studies from different parts of the Figure 4. Friction angles for different relative densities world on geotechnical characteristics of carbonate sand based on CD triaxial and DS tests on Deira sands. samples show distinct differences between static and seismic strength properties of carbonate sands and 6.2 silicate sands. Liquefaction susceptibility Carbonate sands generally have high critical state friction angles. Values in excess of 40 degrees are not Liquefaction is often one of the design determining unusual, compared to 33 degrees on average for silica geotechnical risks for earth structures and shallow sands (Semple, 1988). The relatively high values (when foundations. During the design of the Deira Islands compared to silica sands) are a result of the relatively coastal structures, the liquefaction potential of the high angularity (lower roundness) and irregular particle carbonate sands was assessed in two different ways: shapes (lower sphericity). These particle characteristics are responsible for several particle-level mechanisms that 1. Via cyclic triaxial tests on carbonate sand increase shear resistance, such as hindered rotation, samples slippage and ability for particle rearrangement (Cho et al., 2006). 2. Via correlation between CPT records and cyclic Because of their relatively high compressibility, the shear resistance using the shell correction factor strain levels required to reach peak shear strength are from Wehr (2005) much larger for carbonate sands in comparison to silica sands. Concurrently, the shear strength does not fall back 6.3 Liquefaction analysis based on cyclic triaxial tests to a significantly lower (residual) strength at large strain. In the case of slope stability analyses, this also means Results of cyclic triaxial test results on Dubai sands were that progressive failure along a failure plane is less likely used to establish a relationship between the number of to occur in carbonate sands when in comparison to silica cycles to liquefaction (Nliq) and the Cyclic Stress Ratio sands. (CSR) for a relative density of approximately 30%. The Considering natural variability in carbonate deposits tests were carried out on samples of carbonate sand that between different geological environments, it is not were prepared at different levels of relative density and advisable to rely on literature only when assessing the different consolidation stress with a minimum B factor of characteristics of carbonate sands. The Deira Islands 0.95.. Figure 5 presents the results of the cyclic triaxial project included an extensive laboratory testing program, teststests onon DeiraDeira sands.sands. which revealed:

be applied when the shell correction factor is used to correct the measured cone resistance for carbonate content. InIn thethe casecase ofof thethe DeiraDeira IslandsIslands project,project, tthe factors of safety obtained from the liquefaction analysis via CPT base curves were lower than the factors of safety that resultedresulted fromfrom thethe anaanalysis via cyclic triaxial tests. This was the case, even though the shell correction factor was applied to the measured cone resistance. It is therefore discouraged to use well-established silicate sand correlations with in-situ tests to determine static and cyclic shear strength parameters for carbonate sands. Ignoring the typical nature of carbonate sands will result in over-conservative designs and ground Figure 5. Pore water pressure build-up from cyclic triaxial improvementimprovement schemes schemes for for large-scale reclamations, tests on Deira sands for a range of relative densities. breakwater foundations and revetments.

7 DESIGN

7.1 Revetment design

The scope of the project included 15 revetment sections on Island A and 12 revetment sections on Island B. Some of the revetments along Deira Islands A and B had been completed during earlier construction works, whilst others werere partially partially completed. completed. The The partially partially completed completed rockworkrockwork structures structures were to be finished in accordance with the updated performance specifications. Of the partially completed revetments, several were adapted to include minor changes to the design such as a slight increase of reclamation level and/or crest level. Figure 6. Relationship Nliq and CRS for different relative Others were optimized by using newly acquired soil densities based on cyclic triaxial tests Dubai sands. investigation data in the geotechnical design. Two different types of revetment design were applied forfor Deira Deira Islands: Islands: a a single single slope slope revetment revetment and and a a Figure 6 presents the relation between CSR and Nliq revetmentrevetment with with a a submerged submerged berm berm ( (Ockeloen et al., based on cyclic triaxial test results on Dubai sands. The 2012). A single slope revetment was preferable at all factor of safety against liquefaction (FSliq) can be derived times,times, due due to to lower lower costs. costs. However, However, at at locations locations where where from results of cyclic triaxial tests, as FSliq is related to the overtopping and low crest level requirements make it ratio between Nliq and the equivalent number of cycles impossibleimpossible to to design design a a single single slope slope revetment, revetment, a a (N), where N represents the design earthquake. The submerged berm is applied. value of N can be estimated from relationships with Geotechnical stability analyses were performed for the moment magnitude of the earthquake (Mw) and the cross sections that were developed based on hydraulic duration of the earthquake (Seed et al., 1975 and Idriss & demands for revetment armor stability and wave Boulanger, 2008). During the design of the Deira Islands overtopping. Basis of the geotechnical design was that coastal structures, the effect of pore pressure generation potential deformations during and after an earthquake was explicitly taken into account using relationships would remain within prescribed limits, while all normal between FSliq and the ratio between excess pore pressure geotechnical safety levels would be achieved during staticstatic and initial effective stress (ru) for a certain relative density. conditions. It was chosen to adopt simplified performance based analyses for the revetments, to be carried out by 6.4 Liquefaction analysis based on CPT records using straightforward limit equilibrium models and empirical relations. Standard procedure for the determination of the Cyclic Cyclic strength degradation was taken into account Resistance Ratio (CRR) is the use of normalized SPT and based on the project specific relation for generation of CPT base curves, based on case histories of observed excess pore pressures. The effect of excess pore liquefaction. This standard procedure of liquefaction pressures was introduced by a reduction in shear strength analysis via CPT or SPT base curves includes stress parameters (i.e. with equivalent effective friction angles). normalization and corrects for fines content and plasticity, 100% of the maximum expected excess pore pressures but it is not directly applicable for sands with high were used in the post-earthquake analyses, whereas 50% carbonate content as the case histories predominantly of the maximum expected excess pore pressures were include silicate sands. The CPT base curves can however used in the co-seismic slope stability analyses.

As a first step, the hydraulically approved geometry To ensure the most economical design, it was desired was checked for its post-earthquake stability, since a that the hydraulically placed sand core could be left slope that will start to deform even without base untreated while still complying with the seismic accelerations will not meet the performance criteria. The displacement criteria for the breakwaters. This design post-earthquake stage considers the condition of the basis would form the main challenge for the breakwater slope just after the earthquake, when maximum excess design, as without densification measures the sand fill pore pressures will have developed. would be susceptible to liquefaction. Subsequently, the earthquake induced displacements The performance criteria for ULS required that the of the slope were checked. The vertical component of breakwater crest remained at or above Mean Sea Level both the ULS and SLS ground acceleration was taken as (MSL). With a hydraulically determined crest height of +/- 50% of the horizontal acceleration. Differentiation was +4.9m DMD and a design MSL of +1.25m DMD, this made between the following displacement types: requirement allowed for large vertical deformations of up to 3.65 m for the breakwater crests. 1. Kinematic movements of sliding soil. In a limit The challenging design basis required a detailed equilibrium model, these displacements can be understanding of the shear strength characteristics of the estimated by the Newmark sliding-block analysis. sand core material. Fortunately, a significant number of Use was made of SLAMMER (Jibson et al., CD triaxial and DS test results was available, based on 2013), a program that can be freely downloaded which a project specific relation between relative density from the USGS website, to estimate co-seismic and angle of internal friction could be established for the landslide displacements using rigid sliding-block sand fill (Figure 4). analyses. Although the project specifications required that the sand fill in the sand core of the breakwaters should have 2. Cyclic shake-down of dry material and a relative density of not less than 40%, initial analyses liquefaction induced volumetric strain. These showed that a lower relative density profile could also deformations were estimated based on empirical fulfill compliance with the performance criteria. A design relations as developed by Tokimatsu and Seed relative density of 35% for the sand core was therefore (1987) and Ishihara and Yoshimine (1992). proposed in an alternative design. For the breakwaters, simplified pseudo-static After successful verification of the seismic calculation models were combined with dynamic finite displacements, verification of static geotechnical stability element analysis to provide the most realistic of the revetments was considered a formality and did not displacement analyses. The simplified analysis enabled lead to any design changes. the designers to check sensitivities, relations between relative density requirements and minimum required 7.2 Breakwater design geometry, differences in ground conditions, etc. The dynamic analyses were performed as a final evaluation of The original tendering scope was extended to include the the expected displacements for selected geometries. design and construction of two new breakwaters with a Non-linear time history analyses were performed in total length of 2 kilometers between the islands C and D two-dimensional plane-strain with PLAXIS FEM software (Figure 1). The objective was to provide an operationally for geotechnical applications. Use was made of the safe wave climate and safer marine access to vessels, Hardening Soil Small Strain constitutive model, which providing protection against the threats of coastal erosion allows for the degradation of stiffness from very small to along the internal edges of Deira Islands A and B. large strains in shear. The model does not include The breakwater design consisted of a sand core degradation of the soil’s stiffness and strength as a protected by layers of quarry run and armor stones, as function of excess pore pressures generated by cyclic illustrated in Figure 7. The sand for the sand core was to loading, which was accounted for manually. Standard be dredged from an earlier reclamation for the Trunk viscous absorbent side boundaries were used in the Island close to the breakwaters. calculations, with a compliant base as lower boundary.

Figure 7. Breakwater cross-section.

The dynamic model was set up with appropriate ground motion records. This is a clear result of the dynamic soil parameters and the effect of pore pressure variation in frequency content. Significant differences are was reflected in reduced shear strength values similar to observed between results of simplified and FEM analyses the approach followed in the pseudo-static models. A for the same ground motion records as well. Closer cohesion of 0.1 kPa was assigned to all soil units for examination of the results reveals that the time histories numerical stability. with dominant frequencies in the 0.2 s to 0.3 s range (EQ Appropriate levels of Rayleigh damping were 627 and EQ 3268) give the highest response in the FEM attributed to the different ground and fill materials, as analyses. The SHAKE2000 results are for a 1D column summarized in Table 3. Rayleigh damping provides and independent of the fundamental period of the additional damping - in addition to hysteretic damping breakwater structure, which produces opposite results to included in the Hardening Soil Small Strain constitutive those of the FEM analyses (i.e. EQ 3274, with dominant model - in order to obtain realistic material damping for all frequencies lower than 0.2 s, giving the highest strain levels. response).

Table 3. Dynamic parameters FEM. Table 4. Comparison of soil amplification factors for two different breakwater cross-sections. 1 Characteristics G0 (MPa) C (%) Cross- Rock Armour 202 1 Earthquake SHAKE2000 PLAXIS Sections Quarry Run 202 1 EQ.627 0.93 1.43 Sand with Dr =35% 120 2 Caprock 410 1 EQ.3268 1.27 1.42 Section 1 Sand with Dr =40% 127 2 EQ.3274 1.60 1.20 Bedrock 410 1 Average 1.26 1.35 1 Raleigh damping EQ.627 0.93 1.47 EQ.3268 1.20 1.37 Figure 8 provides a comparison of response spectra Section 2 obtained from one of the breakwater cross-sections. EQ.3274 1.73 1.13 When comparing the response spectrum for the Average 1.29 1.32 breakwater crest with the bedrock and free field spectra, the influence of the breakwater presence can clearly be seen. High frequencies are damped out by the breakwater’s inertia, while amplification takes place Naturally, the FEM analyses can be regarded far around a period of 0.2 s to 0.3 s, which corresponds to superior to the one-dimensional SHAKE2000 analyses, as the fundamental period of the breakwater structure. the latter do not include the mass inertia of the breakwater and topographical effects and only approximate non-linear behavior. On average, the calculated soil amplification factors from SHAKE2000 show good agreement with those from PLAXIS. This exemplifies the importance of analyzing a series of ground motion records when adopting a performance based design. Regarding the calculated co-seismic displacements for the breakwater crest, the Newmark displacements (10- 12 cm) compared reasonably well to the displacements calculated in PLAXIS (8-20 cm). Displacements due to cyclic shake-down and liquefaction induced volumetric strain were added to the calculated co-seismic displacements. It was assumed that the constructed filter layer would avoid any void redistribution between the sand core and quarry run material. The total sum of calculated displacements was Figure 8. Response spectra for Chichi Taiwan 06 subsequently checked against the performance criteria, to Earthquake, CHY028 Station record (EQ 3268). which all design sections complied.

7.3 Optimization during breakwater construction Table 4 provides a comparison between soil amplification factors obtained from one-dimensional During construction, the relative density requirements equivalent linear ground response analyses and non- were verified by CPTs after placement of the fill. CPT data linear time history analysis. The results indicate a revealed that the as-placed relative density varied per significant range in values obtained between different location along the breakwater alignment and over depth.

The average measured relative density was about 48%  By adopting a fit for purpose performance-based from seabed level to DMD -10.5 m and about 43% from design approach, optimal use was made of the DMD -10.5 m until refusal depth at approximately -15 m opportunities provided by the performance DMD. criteria to achieve the most economical design. As some lenses with relative density below 35% were observed, additional checks were performed assuming a continuous lower bound profile with 30% relative density 9 ACKNOWLEDGEMENTS as a worst case scenario. Simultaneously, crosschecks were performed to assess the design implications of The authors would like to thank Royal Haskoning DHV gentler side slopes and a higher sand core level, based and Van Oord Dredging and Marine Contractors B.V. for on as-built surveys of the core. supporting the realization of this paper. It was concluded that liquefaction could potentially occur in the sand core underneath the breakwater berm. Associated displacements could be allowed as long as 10 REFERENCES these would not exceed the performance specifications for the breakwater crest. Lateral spreading displacements Baldi, G., Bellotti, R., Ghionna, V., Jamiolkowski, M. and at the breakwater berm were assessed based on Zhang Pasqualini, E. 1986. Interpretation of CPTs and et al. (2004). Subsequently, the breakwater crest CPTUs - 2nd Part: Drained Penetration of Sands, 4th deformation was conservatively estimated based on a Int. Geotechnical Seminar Field Instrumentations and volume loss calculation, assuming that the laterally In-Situ Measurements, Nanyang Technological spreading fill material underneath the berm would be Institute, Singapore, 143-156. (partially) replaced by non-liquefied sand core material Bellotti, R. and Jamiolkowski, M. 1990. Evaluation of CPT from underneath the crest. The calculations resulted in a and DMT in Crushable and Silty Sands. Second total estimated vertical crest deformation of 2 m, within Interim Report to US Army Corps of Engineers, the performance requirements for ULS conditions. Waterways Experimental Station, Vicksburg. Cho, G., Dodds, J. and Santamarina, J. C. 2006. Particle shape effects on packing density, stiffness and 8 CONCLUSIONS strength: Natural and crushed sands. Journal of Geotechnical and Geoenvironmental Engineering, 132 Based on the work done for the Deira Islands project, the (5): 591-602. following conclusions can be drawn: Idriss, I.M. and Boulanger, R.W. 2008. Soil Liquefaction  Seismic stability and liquefaction susceptibility During Earthquakes, EERI Monograph, MNO-12, controlled the detailed design process in the Oakland, CA, USA. Deira Islands project. Ishihara, K. and Yoshimine, M. 1992. Evaluation of  Ground response analyses produced a reduction settlements in sand deposits following liquefaction in ground motions in comparison to soil during earthquakes, Soils and Foundations, 32(1): amplification factors from the local code. 173-188.  Seismic slope deformations for the revetments Jibson, R.W., Rathje, E.M., Jibson, M.W., and Lee, Y.W. were successfully analyzed with simplified 2013. SLAMMER - Seismic LAndslide Movement analyses using the Newmark sliding block Modeled using Earthquake Records (ver.1.1, method. The revetments generally had sufficient November 2014): U.S. Geological Survey Techniques resistance against earthquake loading, without and Methods, book 12, chap. B1, unpaged. the need for additional measures. Ockeloen, W.J., Adrichem, J.D. & van der Lem, J.C.  For the breakwaters, research of characteristic 2012. The berm breakwater concept – ‘natural’ th shear strength properties and liquefaction structures in the Middle East. 8 Int. Conf. on Coastal susceptibility of the carbonate sand fill material and Port Engineering in Developing Countries, IIT enabled the implementation of significant Madras, , India. optimizations in the design. Seed, H.B., Martin, P.P. and Lysmer, J. 1975. The  Ignoring the typical nature of carbonate sands Generation and Dissipation of Pore Water Pressures may result in over-conservative designs for large During Soil Liquefaction. EERC 75-26. University of scale reclamations, breakwater foundations and California, Berkeley. revetments in these environments. Semple, R.M. 1988. State of The Art Reports: The  A comparison of 1D equivalent-linear and 2D Mechanical Properties of Carbonate Soils. Int. Conf. non-linear ground response analyses revealed a on Calcareous Sediments, Balkema, Perth, Australia, significant divergence in the 1D analysis results 2: 807-836. for a breakwater. This is mainly caused by Tokimatsu, K. and Seed, H.B. 1987. Evaluation of disregard of the effect of the fundamental period settlements in sands due to earthquake shaking. J. of the breakwater structure. Geot. Engineering, 113 (8): 861-878. Wehr, W.J. 2005. Influence of the Carbonate Content of  Allowance of local liquefaction potential enabled th the incorporation of an untreated hydraulically Sand on Vibro Compaction, 6 Int. Conf. on Ground placed sand core inside breakwater, while still Improvement Techniques, CI-Premier, Coimbra, complying with the performance criteria. Portugal, 625-632.

Zhang, G., Robertson, P.K. and Brachman R. W. I. 2004. Estimating Liquefaction Induced Lateral Displacements Using the Standard Penetration Test or Cone Penetration Test, Journal of Geotechnical and Geoenvironmental Engineering, ASCE, August, 2004, 861-871.