Proceedings of ICE Civil Engineering 162 November 2009 Pages 25–33 Paper 09-00036

doi: 10.1680/cien.2009.162.6.25 Keywords geotechnical engineering; foundations; piles & piling Foundation design for the tower in , UAE

Kamiran Ibrahim MSc, PhD The Pentominium tower in Dubai, UAE will be the tallest residential is regional technical director at Hyder Consulting, Middle East, building in at over 100 storeys tall when completed Dubai, UAE in 2012. This paper describes the design of ’s piled raft foundation in the local carbonate soils and rock. Geotechnical investigations are outlined, along with how the effect of the proposed tower on neighbouring structures, single-pile response and impact of cyclic degradation were assessed. A description

Grahame Bunce of the numerical analyses used to evaluate the overall piled raft MSc, CEng, MICE response under various static and wind loading combinations is is technical director at Hyder Consulting (UK) Ltd, Guildford, UK presented as well as some of the techniques used to optimise the foundation design, including preliminary pile testing.

The Pentominium residential development is the Jumeira Palm in the located approximately 500 m to the east of (UAE) (Figure 1). The development comprises and south of the beach near the construction of a tower over 100 storeys tall

Catherine Murrells 0 m 200 MA, MEng, CEng, MICE is principal geotechnical engineer Beach at Hyder Consulting (UK) Ltd, Guildford, UK American College

Dubai international Marine Club

Marina 23 tower Metropolitan Club Pentominium tower

Forte Grand Hotel Dubai Marina

Sheik Zayed Road

Figure 1. Location of the Pentominium tower on the

Delivered by ICEVirtualLibrary.com to: CIVIL ENGINEERING IP: 93.109.117.51 25 Mon, 07 Jun 2010 07:09:48 Ibrahim, Bunce and Murrells

and associated podium structure (Figure 2). It Geology Geotechnical investigation and testing is set to be the tallest residential building in the world when completed in 2012. The geology of the UAE and the Persian The ground investigation was undertaken All site levels are related to Dubai Municipal- Gulf area has been substantially influenced by ACES and consisted of sinking eight cable ity datum (DMD) and original ground level is at by the deposition of marine sediments associ- percussion boreholes with rotary follow-on about 5 m DMD. There are six basement levels ated with numerous sea level changes during methods, in-situ testing and laboratory testing and the structure is supported by a piled raft relatively recent geological times. With the (including specialist testing) on selected sam- system comprising large-diameter bored piles exception of mountainous regions shared ples. The boreholes were drilled to 80–125 m cast in situ. The pile cut-off levels are founded with Oman in the north-east, the country is deep with standpipe piezometers installed to about 24 m below existing ground level at an relatively low-lying, with near-surface geol- monitor the groundwater table. The scope of elevation of −19.4 m DMD. ogy dominated by deposits of Quaternary to in-situ testing is summarised as follows The area of the site is very limited with the late Pleistocene age, including mobile Aeolian structure extending to all boundaries. In addi- dune sands, sabka/evaporite deposits and n standard penetration testing tion, construction on the adjacent Marina 23 marine sands. n packer permeability testing tower is underway, with a number of levels of Dubai is situated towards the eastern extrem- n pressuremeter testing at 3 m intervals in superstructure completed before piling started ity of the geologically stable Arabian tectonic three of the boreholes on the Pentominium tower in 2008. plate and is separated from the unstable Iranian n geophysics (cross-hole, cross-hole tomog- The client for the project is Trident Inter- fold belt to the north by the Persian Gulf. It raphy and down-hole testing). national Holdings and the architect is Aedas. is therefore considered that the site is located Hyder Consulting carried out the detailed design within a moderately seismically active area. Disturbed, undisturbed and split-spoon of the foundation, substructure and superstruc- However, it was indicated from the structural samples were obtained from the boreholes for ture. Arab Centre for Engineer Studies (ACES) analysis that the wind effect was more critical laboratory testing purposes. The undisturbed carried out the ground investigation and Swiss- than the seismic effect as is typical for structures samples were obtained using double-tube- boring Limited was the piling contractor. of this size in Dubai. core barrels from which 92 mm nominal core

+10 metres DMD Very loose to loose sand BH02 +5 BH06 BH03 0 Medium dense to very dense sand –5 Very dense silty sand with sand stone fragments –10 Sandstone –15 Gypsiferous sandstone

–20 Calcisiltite/conglomerate/conglomeritic calcisiltite –25 –30 –35 –40 –45 –50 –55 –60 –65 Sandstone –70 Calcisiltite –75 –80 –85 –90

–95 Interbedded siltstone and gypsum –100 –105 Assumed geological unit based –110 on existing boreholes –115 Base of borehole data –120 Groundwater strike

Figure 2. Artist’s impression of the Pentominium tower, which, at over 100 storeys tall, will be the highest residential building in the world when completed in 2012 Figure 3. Geological long-section through three of the boreholes

Delivered by ICEVirtualLibrary.com to: 26 ProCeedings of the Institution of Civil Engineers – CIVIL ENGINEERINGIP: 93.109.117.51, 2009, 162, No. CE6 issn 0965 089 X Mon, 07 Jun 2010 07:09:48 Foundation design for the Pentominium tower in Dubai, UAE

diameters were recovered. The laboratory Pressuremeter initial Resonant column 0.001% Percentage values testing included the following standard and Pressuremeter reload 1Cyclic triaxial 0.1% relate to the level Pressuremeter reload 2 Cross-hole geophysics of strain at which specialist tests Stress path 0.01% Strata the samples are Stress path 0.1% Design line at small strain tested n standard classification testing 0 n chemical testing –5 n unconfined compression tests –10 n cyclic undrained triaxial –15 n cyclic simple shear

n stress path triaxial testing m DMD vation : –20

n resonant column Ele –25 n constant normal stiffness testing. –30 –35 The standard ground investigation testing –40 was carried out following British standards BS –45 59301 and BS 1377.2 Four preliminary trial pile tests were also –50 carried out to determine single-pile load– –55 settlement behaviour and to assess the pile –60 capacity in skin friction. –65 –70 Geotechnical conditions and parameters –75

The ground conditions comprise a horizon- –80 tally stratified sub-surface profile. Three layers –85 of sand, varying from very loose to medium –90 dense to dense as elevation decreases, overlie –95 layers of very weak to weak sandstone, gyp- –100 siferous sandstone, calcisiltite, conglomerates –105 and calcareous siltstones. An idealised ground profile used for the –110 whole site is presented in Table 1 and a sec- –115 tion through three of the boreholes is shown –120 in Figure 3. –125 The geotechnical stiffness parameters for 0 1000 2000 3000 4000 5000 the design of the foundation were determined Elasticity modulus: MPa from the tests carried out on the strata at dif- Figure 4. Young’s modulus E' values derived from geotechnical testing ferent strain levels. It is presented in Mayne and Schneider3 that rock behaviour for defor- mation analyses, which would include piled Table 1. Idealised ground profile and geotechnical parameters rafts, ranges between strain values of approxi- Sub- Level at Unconfined Undrained Drained mately 0.01–0.1%, which correlate with Strata strata Material top of Thickness: compressive modulus* modulus* number number stratum: m strength: E : MPa E': MPa testing results from the pressuremeter, stress m DMD MPa u path triaxial, resonant column, cyclic triaxial Very loose to loose +1.40 to 1a slightly silty sand with . 2.70 – – 2 testing and geophysics which are presented on occasional sandy silt +5 59 Figure 4. 1 Medium dense to very −0.13 to It should be noted that the design line 1b dense slightly silty to . 9.50 – – 36 silty sand +5 59 shown on Figure 4 is for small strain design Very dense silty sand with . . at 0.1% strain. The stiffness values provided 1c sandstone fragments −7 50 3 00 – – 75 in Table 1 are for the larger strain levels of Very weak to weak calcarenite/calcareous . . . approximately 1% determined from standard 2 2 sandstone interbedded −10 50 2 20 0 8 125 100 correlations with unconfined compressive with cemented sand strength results as presented in Tomlinson Very weak to weak . . . 3 3 gypsiferous sandstone −12 70 5 30 0 8 125 100 and Woodward:4 drained Young’s modulus at 4a Very weak to moderately −18.00 2.50 0.8 125 100 large strain E' = M j q , where M is the ratio . . . r u r 4 4b strong calcisiltite/ −20 50 950 3 0 350 280 of elastic modulus of intact rock to its uniaxial conglomerate/ . . . 4c conglomeritic calcisiltite −30 00 34 00 2 4 250 200 compressive strength, j is the mass factor and 5 5 Weak sandstone −64.00 2.60 2.4 250 200 q is the uniaxial compressive strength. u 6 6 Very weak to moderately −66.60 17.40 4.1 250 200 Non-linear stress–strain curves were devel- strong calcisiltite . oped for the rock strata based on all the Very weak to moderately >38 00 weak claystone/siltstone . (proven . ground investigation data. The curves were 7 7 interbedded with gypsum −84 00 to base of 3 0 250 200 fitted to the data using hyperbolic functions as layers boreholes) 3 presented by Mayne and Schneider and are * Note that Eu and E' values relate to large strain level (about 1%) of the strata Delivered by ICEVirtualLibrary.com to: issn 0965 089 X ProCeeIP:dings 93.109.117.51 of the Institution of Civil Engineers – CIVIL ENGINEERING, 2009, 162, No. CE6 27 Mon, 07 Jun 2010 07:09:48 Ibrahim, Bunce and Murrells

presented in Figure 5. Marina and the main coast for this to be pos- results from projects completed or under con- The single-pile load–settlement behaviour was sible. Information from other ground investiga- struction, such as the Burj Dubai as presented estimated using the Cemset method as developed tion works in the neighbouring sites indicated by Poulos and Bunce6 and the by Fleming5; experience was also utilised from a groundwater level ranging between −2.1 and as presented by Poulos and Davids.7 previous pile tests carried out and assessed in the −2.9 m DMD. Taking into account the informa- The main model was set up using the geo- region, which includes tests around the Dubai tion from adjacent sites and the fact that no ten- technical finite-element analysis program Midas- Marina area as well as the rest of Dubai. The sion loading is applicable for the development, a GTS8 with the model being developed by TNO estimated load–settlement behaviour is shown conservative groundwater level of −1.5 m DMD Diana in the Netherlands. Other models were in Figure 6 and is compared with the measured was taken for design purposes. run to correlate and validate the results from the results from the preliminary pile testing. finite-element analysis including standard pile- The groundwater levels encountered in the Geotechnical models and analyses group analysis using Repute9 and equivalent raft boreholes varied widely between −5.91 m DMD analysis using Vdisp.10 In addition, the results to −14.57 m DMD. However, a large number of A number of analysis techniques have been were compared with those obtained from the projects are under construction in the adjacent used to assess the piled raft foundation response model set up in Strand 7 which was used for area and it is considered that the associated for the Pentominium tower. The foundation the structural design. dewatering has artificially lowered the ground- design methodology has been developed using A number of foundation options were consid- water. The site is sufficiently remote from Dubai previous experience in Dubai and a number of ered for the Pentominium tower including 1.5 m and 2.2 m diameter pile systems and a barrette 9 solution. The basement levels extended to 25 m 8 below existing ground level. However, due to

a 7 the presence of the relatively high groundwater

MP 6 level, it was proposed that the piles/barrettes : : 5 were installed from approximately −4 m DMD,

stress to reduce flow of water from the pile bores dur- 4 Stratum 3 ing construction. This was approximately 10 m 3 Stratum 4a and b Stratum 4c below existing ground level and 15 m above

Normal 2 Stratum 5 pile/barrette formation level. From a prelimi- 1 Stratum 6 nary assessment of the generated pile/barrette 0 load distributions, it was determined that the 0 2468101214161820 Strain 1.5 m diameter pile solution would be the most effective in terms of constructability as well as Figure 5. Non-linear stress strain curves for rock strata load distribution, particularly considering that

Load: kN 55.4 m 020 000 40 000 60 000 80 000 100 000 120 000 0 Piles 1.2 m thick raft

3 m 5 m thick raft

. –001 3 m

–0.02 m : : 62 . 55 m

lement Working load . tt –003 Se

–0.04 Shaft Base Elastic shortening Superstructure Total –0.05 Repute 1.2 m thick raft Preliminary pile test 1 Preliminary pile test 3 Preliminary pile test 5 Preliminary pile test 6 –0.06 Shoring and outer wall Figure 7. Layout of the 1.2–1.5 m diameter 32–42 m deep piles under the Figure 6. Single-pile load–settlement behaviour foundation raft (see Table 6 for pile details)

Delivered by ICEVirtualLibrary.com to: 28 ProCeedings of the Institution of Civil Engineers – CIVIL ENGINEERINGIP: 93.109.117.51, 2009, 162, No. CE6 issn 0965 089 X Mon, 07 Jun 2010 07:09:48 Foundation design for the Pentominium tower in Dubai, UAE

the plant for the construction of the required include the 5 m thick raft or the effect of the . lengths of 2 2 m diameter piles or barrettes was 1. δrigid = 1/3 (2δcentre + δcorner)flexible superstructure above and therefore the values not easily available in Dubai. are not realistic. The adopted pile layout is shown on Figure 7. From the geotechnical finite-element analysis, The raft within the tower area is 5 m thick and The computed settlements from all the when non-linear soil is assumed, it is indicated 1.2 m thick within the podium area. Prelimi- analyses are presented in Table 2. It is indicated that while the overall settlement increases by nary analyses determined that the longest piles that the results from the finite-element analysis, 15 mm, the differential settlement only increases proposed were 56 m long, however after value Repute and Vdisp correlate relatively well for by 3 mm, which is within tolerances for design engineering; these pile lengths were reduced the same soil profile used. It is noted that large of the raft. A contour plot of the settlements significantly resulting in an optimised design. differential settlements have been calculated under dead plus live load from the MidasGTS The entire raft plan area for the Pentominium from the Vdisp analysis, however this does not run using the smaller strain soil stiffness profile tower is approximately 60 m by 53 m and the resulting finite-element analysis model was 250 m × 250 m in plan and 220 m vertically to avoid boundary effects. Within the raft, the To wer structure average mesh size was 1.5 m with the mesh size increasing to 30 m at the model boundary. A summary of the set up of the model is as follows.

n Structur

Soil strata – modelled as solid elements gr

with parameters defined for E' and Pois- ound le

son’s ratio for the elastic linear runs and e abo

using a user-defined material within Midas- ve l GTS for the non-linear runs. ve n Piles – modelled as beam elements with each node of the beams connected to the e nearest soil solid elements using pile inter- face elements. n Superstructure – modelled as beam and plate elements to represent the columns, walls and slabs. The six basement levels were included as well as nine levels of structure Basement structur above ground level. Within the structure, (including diaphragm walls) there are large inclined columns to spread the load from the centre of the tower to the outer edges. The superstructure included in the model is presented on Figure 8. n Loadings. These were applied at level 9 due to the presence of the large inclined Piles columns such that the distribution of load through those columns could be simulated. Self-weight of the superstructure elements were included to level 9. Hydrostatic load- ing was included as an uplift pressure on the raft foundation.

The foundation was also assessed by structur- al engineers through modelling the soil as brick elements in Strand 7. The settlements obtained Figure 8. The finite-element model included nine levels of the superstructure as well as the six basement levels from the geotechnical MidasGTS model were incorporated into the Strand model to calibrate Table 2. Computed settlements under dead and live lodaing from analyses the pile stiffness values such that the behaviour Computer package Soil parameters Maximum settlement: mm of the raft could be determined and its effect on Flexible Rigid Differential the superstructure above. MidasGTS Linear, small strain 77 69 20 Foundation design results MidasGTS Non-linear 92 85 23 Vdisp Linear equivalent raft, 217 198 32 large strain The estimated settlements from the finite- Vdisp Linear equivalent raft, 102 78 37 element analysis model and from Vdisp have small strain been converted from those for a flexible pile cap Vdisp Non-linear equivalent 147 115 97 to those for a rigid pile cap for comparison with raft the Repute model outputs using the following Repute Linear, large strain – 166 – general equation for a rectangle Repute Linear, small strain – 87 – Delivered by ICEVirtualLibrary.com to: issn 0965 089 X ProCeeIP:dings 93.109.117.51 of the Institution of Civil Engineers – CIVIL ENGINEERING, 2009, 162, No. CE6 29 Mon, 07 Jun 2010 07:09:48 Ibrahim, Bunce and Murrells

Pile settlement: mm Length along section Y–Y: m . 0 10 20 30 40 50 60 70 1 2 m thick raft 45

50 Raft edge Raft edge Piles 55

60

76 65 mm 74 5 m thick raft : 72 70 Settlements –

70 lement

tt linear soil 68 75 66 Se 64 80 62 . . 60 1 2 m thick 1 2 m thick 58 85 raft raft 56 Settlements – 54 90 non-linear soil 52 1.2 m thick raft 5 m thick raft 95 Cross section Y–Y Figure 10. Cross section Y–Y of settlement under dead and live loading from Figure 9. Contour plot of settlement under dead and live loading from MidasGTS MidasGTS

is presented in Figure 9. A cross-section of set- account of the stiffness of raft and superstruc- pality regulations. The purposes of the tests tlement through the foundation is presented in ture as well as the wall and column locations at were to validate the design assumptions made Figure 10, which compares the settlements from which the foundation is loaded. From the con- during the design including the load-settlement the MidasGTS model for linear and non-linear tour plots, it is observed that the pile axial loads response of the piles and the ultimate skin fric- soil stiffness profiles. are concentrated towards the outer left and tion mobilised along the pile shaft. Pile axial forces were determined by com- right sides of the foundation as this is where the Loading of the piles was achieved using paring the results from the MidasGTS model, inclined columns in the superstructure spread Osterberg cells installed part-way down the structure’s Strand model and Repute. The the applied load. piles. Pile displacements were measured above resultant contour plots of the pile axial loads and below the Osterberg cells and an equiva- are presented in Figure 11. The standard pile- Pile load testing lent top-loaded pile load–settlement curve was group analysis program, Repute, assumes that assessed. The single-pile load–settlement curves the pile cap is infinitely rigid and therefore, the Four preliminary trial pile tests were carried from the preliminary pile testing are compared pile loads are concentrated towards the outside out as summarised in Table 3. In addition, static with those determined from theory on Figure of the group. The pile axial load distribution and dynamic working pile tests are specified to 6. It is noted that the single-pile behaviour obtained from MidasGTS and Strand, on the be carried out on the piles prior to construction observed from the preliminary pile testing is other hand, are similar and both programs take of the raft in accordance with Dubai Munici- stiffer than that predicted. However, at the

Repute Strand MidasGTS

1. 2 m 1.2 m thick thick 6000 kN 10 000 kN raft raft 10 000 kN

10 000 kN 16 000 kN 12 000 kN

kN

N 0

k

0

kN

0

0

0

0

15 00 kN 000 16

0

0

1

2

1

kN

0

0

kN

0

0

N

0

8

k 5 m 0 5 m 2 10 000 kN

kN 0

6

0 thick 1 thick

kN 5 00

0 1

24 00 raft 0 raft

0 12 000 kN 10 000 kN 15 000 kN 4 16 000 kN 2 15 000 kN 15 000 kN 10 000 kN 10 000 kN 6000 kN 1.2 m 1.2 m thick thick raft raft

Figure 11. Comparison of pile axial loading under dead and live loading

Delivered by ICEVirtualLibrary.com to: 30 ProCeedings of the Institution of Civil Engineers – CIVIL ENGINEERINGIP: 93.109.117.51, 2009, 162, No. CE6 issn 0965 089 X Mon, 07 Jun 2010 07:09:48 Foundation design for the Pentominium tower in Dubai, UAE

working load of the pile of 32 500 kN, the set- Table 3. Preliminary test pile (PTP) configuration tlements calculated from theory are very similar PTP1 PTP3 PTP5 PTP6 to those measured during testing. Max test load: MN 61.80 67.50 59.20 56.30 Strain gauges were installed at eight levels Working load: MN 30 30 30 30 along the length of each pile such that the skin . . . . friction mobilised along each section of pile Diameter: m 1 50 1 50 1 50 1 50 . . . . could be calculated. The skin friction values Ground elevation: m DMD +4 50 +3 50 +3 50 +3 50 were originally calculated using theory based on Top of concrete: m DMD –15.10 –15.15 –15.60 –15.23 the design recommendations given by Horvarth Design cut-off level: m DMD –19.10 –19.10 –19.10 –19.10 and Kenney11 as

. . 0.5 Table 4. Pile ultimate compressive skin friction values (see Table 1 for strata details) 2. ƒs = 0 25 to 0 33 (qu) Sub-strata number Ultimate unit shaft friction from Ultimate unit shaft friction theory, ƒs: kPa recommended from pile tests, ƒ : kPa where ƒs is the ultimate unit shaft resistance s

and qu is the uniaxial compressive strength in 2 215 215 2 MN/m . 3 215 215 The skin friction values from theory are 4a 215 215 compared with those calculated from the pre- 4b 415 415 liminary pile testing in Table 4. Based on results 4c 372 415 from the preliminary pile testing, the idealised 5 372 415 rock profile has been simplified which also takes into account potential pile–rock interface 6 486 415 degradation. The skin friction values have been 7 415 415 improved from those calculated from theory and . the factor of safety used for design was 2 5. It is Net unit skin friction measured: kPa noted that, due to the number of consistent pile 0200 400600 8001000 1200 1400 1600 1800 2000 test results received, the factor of safety could 0 have been reduced to 2, however the pile–soil Strata levels block mechanism proved to be critical to the Design values from design and therefore the factor of safety was unconfined compressive strength maintained at 2.5. –10 Preliminary test pile 1 Preliminary test pile 3 The results from the preliminary pile tests are Preliminary test pile 5 shown in Figure 12. Ultimate skin friction has Preliminary test pile 6 not been mobilised, at the pile locations furthest Design for construction –20 from the Osterberg cell and as such none of the piles were tested to failure. Substantial movements at the base of the pile would be required before the ultimate base –30 bearing capacity can be mobilised, particularly if relatively soft materials are left at the base of the piles. From previous experience in the region it has been found that it is very difficult to clean –40 mDMD sufficiently the base of such large diameter O-cell preliminary long piles. Therefore, taking this into account, n: test piles 3 and 6 together with the structural serviceability limits O-cell preliminary vatio in terms of settlements, the pile base resistance e –50 test piles 1 and 5 El contributions were excluded in the estimate of the pile bearing capacity. Cyclic loading analysis –60

Several cyclic loading tests were carried out on both the rock mass and the pile–rock inter- –70 face. In each case, in situ stresses were applied. Skin friction values at the top and bottom Constant normal testing was carried out of the pile are low as these locations are on samples of the rock mass sheared against furthest from the O-cell concrete that had been profiled by a water- –80 jet technique to simulate the in situ pile–soil interface. Samples were sheared both monot- onically and cyclically. From the results, it was –90 indicated that the peak shear strength was not affected by cyclic shearing, however the Figure 12. Pile skin friction values from theory and pile testing residual shear strength was reduced by 15%. Delivered by ICEVirtualLibrary.com to: issn 0965 089 X ProCeeIP:dings 93.109.117.51 of the Institution of Civil Engineers – CIVIL ENGINEERING, 2009, 162, No. CE6 31 Mon, 07 Jun 2010 07:09:48 Ibrahim, Bunce and Murrells

Therefore, the maximum mobilised skin fric- the pile diameter. Therefore, one of the failure As construction of the Marina 23 superstruc- tion values as measured from the preliminary mechanisms of the piles could be a block move- ture is significantly ahead of the Pentominium pile testing have not been adopted in the final ment of the piles and soil. structure, a large amount of settlement will design to allow for any potential degradation Taking into account only the side frictional occur before the superstructure construction of effects due to cyclic loading. resistance along the perimeter of the soil–pile the latter commences. Therefore in terms of tilt In addition, during the preliminary pile testing block to the depth of the pile toe levels, an of the structure, Pentominium may tilt towards programme, ten cycles of loading were applied to acceptable factor of safety of 2.5 was achieved Marina 23, however, this can be addressed each pile to determine whether cyclic degradation against applied vertical loading. When the later- during the construction of Pentominium. The at the pile–rock interface occurred. Through a al resistance of the block is considered, the slid- Marina 23 tower does not tilt towards Pento- Timeset analysis12 of the results obtained, which ing frictional resistance between the raft and the minium and the differential settlement for both can be used to assess long-term degradation, underlying rock was found to be far in excess of tower foundations is significantly reduced at the it was indicated that no degradation had been the applied lateral loading. boundary. observed. These results are also compared with Considering the overturning failure mecha- It should be noted that a number of simplifi- those from the constant normal testing above. nism and assuming the most onerous point of cations have been made during the assessment Cyclic triaxial testing and cyclic simple shear rotation of the block, a factor of safety of 2 is as the stiffness of rafts and superstructure were testing were conducted on samples of the soil achieved. Therefore, it is considered that the not included in the analysis. The bored pile mass collected during the ground investiga- overturning block movement of the piles is shoring wall which has been installed at the tion works. From the cyclic triaxial testing, it critical to the design of the pile length. Hence, boundary of the Pentominium and Marina 23 was determined that after cyclic loading, the despite the possibility for reducing the factor of sites to enable excavation of the basement levels stiffness of the samples was similar to those safety on the ultimate skin friction capacity of has also not been included in the analysis. These samples tested monotonically during the stress an individual pile, as discussed previously, this elements will assist in reducing the differential path triaxial testing. This can be observed from has not been implemented and the pile lengths settlement. Figure 4 where stiffness values from the cyclic were not shortened due to the concern of over- triaxial testing and monotonic stress path triax- turning block movement failure. Final pile design ial tests have been determined at the same strain level of 0.1% and it is shown that the stiffness Effect on adjacent structure The final pile design was determined based values from both tests were similar. The cyclic on an assessment of all the ground investiga- simple shear tests, which were carried out at a The Pentominium tower is being constructed tion and preliminary pile data as well as all higher strain level, indicated that some degrada- immediately adjacent to the Marina 23 tower, the results from the equivalent raft analysis, tion due to cyclic loading could occur at these which is currently under construction. A plan of the standard pile group analysis and the geo- larger strains. Young’s modulus design values at the locations of the two structures is shown in technical and structural finite-element analysis smaller strain have thus been determined based Figure 13. A study was carried out to determine models. From the preliminary design carried on a lower bound to take account of any degra- the effect of the two structures being con- out, the maximum pile axial working loads dation of the soil mass. structed concurrently. The two foundations were were 36 000 kN and the proposed pile lengths modelled using the equivalent raft technique were 56 m long. However, as a result of the Overall stability assessment in Vdisp and the construction staging was esti- detailed structural and geotechnical design mated as presented in Table 5. A cross-section through assessment of the geotechnical param- The minimum centre-to-centre spacing of the of the estimated settlement of the foundations of eters and finite-element analyses, the working piles adopted in the design is about 2–2.3 times both structures is shown in Figure 14. load was reduced to 32 500 kN. In addition,

Table 5. Construction staging of Pentominium and Pentominium Marina 23 towers Pentominium 23 Marina Stage progress progress 3 basement levels and 1 – 3 storeys 48 storeys above 2 Raft complete basement 69 storeys above 3 6 basement levels basement 90 storeys 29 storeys above above basement 4 basement (construction complete) 52 storeys above 5 basement 74 storeys above 6 basement 97 storeys above 7 basement 02m 0 Bored pile wall at boundary 23 Marina tower foundation 120 storeys above basement 8 (construction Figure 13. Plan of Pentominium and adjacent Marina 23 towers complete)

Delivered by ICEVirtualLibrary.com to: 32 ProCeedings of the Institution of Civil Engineers – CIVIL ENGINEERINGIP: 93.109.117.51, 2009, 162, No. CE6 issn 0965 089 X Mon, 07 Jun 2010 07:09:48 Foundation design for the Pentominium tower in Dubai, UAE

ultimate skin friction values were increased Conclusion tive and optimised foundation design solution. based on the results from the preliminary pile testing. From a combination of these two fac- A substantial amount of testing has been Acknowledgements tors, the pile lengths could be decreased to 42 carried out for the design of the Pentominium m. The final pile design details are presented tower foundation, which has included soil and The authors would like to thank Trident Inter- in Table 6. rock testing as well as a comprehensive pile- national Holdings, in particular Adelfried Saide- Monitoring of the structure will be carried testing programme. ly, for the support provided during the design of out during construction (Figure 15) such that In addition, complex geotechnical finite- the foundation of the Pentominium tower. The the design assumptions used can be verified and element analysis has been carried out, which has authors would also like to thank Louise Baker, the experience can be used by future designers been validated using standard geotechnical calcu- Tayo Roberts, Andy Davids and Darko Popovic to ensure the optimisation of the design for such lation techniques. The application of such testing of Hyder Consulting for their invaluable contri- prestigious structures in the region. and analysis approach has resulted in a cost-effec- bution to the design of the foundation.

Distance along centre line of Pentominium foundation: m References Pentominium 23 . Br i t i s h St a n d a r d s In s t i t u t i o n . Code of Prac- –201–10 0 02030405060708090100 110120 130 tice for Site Investigations. BSI, Milton Keynes, 0 Stage 1 1999, BS 5930:1999. r i t i s h t a n d a r d s n s t i t u t i o n Stage 2 2. B S I . Methods of 10 Test for Soils for Civil Engineering Purposes. BSI, Milton Keynes, 1990, BS 1377:1990. 20 Stage 3 3. Ma y n e P. W. and Sc h n e i d e r J. A. Evaluating axial drilled shaft response by seismic cone. In 30 Foundations and Ground Improvement (Br a n - Stage 4 d o n T. L. (ed.)). American Society for Civil 40 Engineers, Reston, VA, USA, 2001, geotechni-

mm cal special publication 113, pp. 655–669. : : 50 4. To m l i n s o n M. J. and Wo o d w a r d J. Pile Design Stage 5 and Construction Practice, 5th edn. Taylor and 60 Francis, Abingdon, 2007. 5. Fl e m i n g W. G. K. A new method for single Stage 6 70 pile settlement prediction and analysis. Géo- Displacement technique, 1992, 42, No. 3, 411–425. 6. po u l o s H. G. and Bu n c e G. Foundation 80 Stage 7 design for the Burj Dubai – the world’s tallest building. In Proceedings of the 6th International 90 Conference on Case Histories in Geotechnical Stage 8 Engineering. Missouri University of Science 100 and Technology, Rolla, MO, USA, 2008 (CD- ROM). Figure 14. Predicted settlement of Pentominium and Marina 23 towers during construction (see Table 5 for 7. po u l o s H. G. and Da v i d s A. J. Foundation construction stages) design for the Emirates Twin Towers, Dubai. Canadian Geotechnical Journal, 2005, 42, No. 3, 716–730. 8. mi d a s IT. midasGTS Analysis Reference. Midas IT, Seoul, Korea, 2007. 9. Ge o c e n t r i x Lt d . REPUTE Version 1 Reference Manual. Geocentrix, Banstead, 2002. 10. oa s y s . VDISP 17 Geo Suite for Windows Refer- ence Manual. Oasys, London, 2001. 11. h o r v a r t h R. G. and Ke n n e y T. C. Shaft resistance of rock-socketed drilled piers. In Proceedings of a Symposium on Deep Founda- tions. American Society for Civil Engineers, New York, NY, USA, 1980, pp. 182–214. 12. en g l a n d M. A method of analysis of stress induced displacement in soils with respect to time. In Deep Foundations on Bored and Auger Piles – BAPII, Proceedings of the Second Interna- tional Geotechnical Seminar, Ghent, Belgium, 1–4 June 1993 (Va n Im p e W. F. (ed.)). Taylor and Francis, Abingdon, 1993, pp. 241–246. Figure 15. Excavation under way in July 2009, showing proximity to adjacent Marina 23 building (far right) (Imersolt.com) What do you think? Table 6. Pile design configuration If you would like to comment on this paper, Pile type Pile diameter: Pile compressive Pile moments: Pile shear Pile embedded please email up to 200 words to the editor at m working load: kN kNm force: kN length: m [email protected]. . A 1 5 32 500 3 500 400 42 If you would like to write a paper of 2000 to 3500 B 1.5 26 000 3 500 400 36 words about your own experience in this or any D 1.2 18 000 1 500 100 32 related area of civil engineering, the editor will be happy to provide any help or advice you need. E 1.2 18 000 1 500 100 32

Delivered by ICEVirtualLibrary.com to: issn 0965 089 X ProCeeIP:dings 93.109.117.51 of the Institution of Civil Engineers – CIVIL ENGINEERING, 2009, 162, No. CE6 33 Mon, 07 Jun 2010 07:09:48