A piled raft case study in Malaysia Patrick Wong Coffey Geotechnics Pty Ltd, Sydney, Australia Keywords: piled raft, pile group settlement interaction, pile load testing ABSTRACT This paper presents a piled raft case study in Penang, Malaysia. The building footprint is about 90m by 76m with a 4m deep basement. The central office tower is 21 storeys in height and occupies an area of 41m by 35m, with a 2 storey retail podium over the remaining area. The site is underlain by deep marine sediments (>120m), and the original foundation design required the use of large diameter bored piles to carry column loads that are in excess of 20MN. An innovative and more economical foundation solution, comprising a piled raft was adopted. It comprised a combination of 500mm and 600mm diameter prestressed driven spun piles supporting a raft having a thickness of 0.5m in the podium area and 1.7m in the tower area. The solution adopted resulted in significant cost and time savings compared to the original design. The piled raft design approach is described and settlement predictions compared with field performance. 1 SITE CONDITIONS AND PILE DESIGN PARAMETERS The site is located on Lot 131, Jalan Sultan Ahmad Shah, Georgetown in Penang, Malaysia. It covers an area of approximately 0.785 hectares, and is situated on the northern coastline of Georgetown, adjacent to the Strait of Malacca. A review of the 1:500,000 Geological Map of the Peninsular of Malaysia reveals that the regional geology of the city of Georgetown consists of unconsolidated Quaternary deposits. These deposits constitute marine and continental deposits of clays, silts, sands and gravels to greater than 120m depth. Based on a programme of investigation comprising boreholes, cone penetration tests and laboratory testing, the adopted geotechnical model for the piled raft design is summarised in Table 1. The design parameters were originally based on empirical correlations of skin friction and end bearing resistance with SPT and CPT results, and were later refined based on a pile load testing program discussed in Section 2. Table 1: Adopted Geotechnical Model and Pile Design Parameters Unit Depth Typical Typical Adopted Pile Design Parameters Interval SPT CPT Cone Elastic Ult. Ult. End (m) Value Tip Modulus Skin Bearing Resistance (MPa) Friction (kPa) (MPa) (kPa) Soft clay/silt (treated 0.0 - 8.0 0-2 <0.5 3.5 (1.8) 20 (17) - by lime piles) Firm clay/silt 8.0 - 13.5 2-5 0.7 – 1 15 (8) 35 (30) - Stiff clay/silt 13.5 - 24.5 5 – 10 1.5 – 1.8 20 (14) 45 (40) - Sandy gravel/gravelly 24.5 - 29.0 15 – 35 Not tested 100 (50) 60 (50) 6000 sand (6000) Stiff sandy clay and 29.0 - 54.0 20 – 35 1.5 – 3.5 30 (30) 55 (40) 2200 silt (2200) Very stiff to hard clay 54.0 – 120 30 - >50 Not tested 85 to 200 140 7500 and silt (85) (125) (6000) Below 120 200 (200) Note: Values in brackets were original parameters adopted for Class A Prediction (i.e. Prediction made of pile capacity and stiffness prior to load testing, using the method described in Poulos 1980) The parameters for the upper soft clay/silt layer took into account the effect of ground improvement carried out by chemical lime piling to facilitate design and construction of the basement excavation, improve trafficability by construction machinery at bulk excavation level, and to improve the stiffness of the material immediately below the foundation raft. A description of the chemical lime piling ground treatment carried out at this site is presented in Wong (2004). In brief, the undrained shear strength of the upper soft layer was increased from an average of about16kPa to between 27kPa and 36kPa. 2 PILE LOAD TESTING PROGRAMME To enable refinement of the original geotechnical model for final design of the piled raft for the project, an extensive program of pile load testing was carried out. In particular, the ultimate load capacity and stiffness of 600mm diameter spun piles (pile type preferred by the client) were of critical interest. The pile load testing program comprised: • Static load tests – one using Constant Rate of Penetration (CRP) and two using Maintained Load Test (MLT) methods. • 7 dynamic load tests using CAPWAP signal matching techniques • The test piles had a wall thickness of 80mm, and were driven to penetrations of between 27m and 57m. Pile settlement during test loading was measured using 4 dial gauges (A, B, C and D) located around the pile perimeter at 90o apart, and a central survey point (Reference 1) located on the pressure jack. The central point was measured by survey against a remote bench mark and is considered to be more accurate compared to the dial gauge readings which were affected by movement of the ground and reference beams during testing. The results of the pile load testing are summarised in Tables 2 and 3. An example of the static pile load testing on test pile TP8 is shown in Figure 1. 6000 5000 Load (kN) 4000 3000 2000 1000 0 0 10 20 30 40 50 60 Settlement (mm) TP8(2) - Gauge A TP8(2) - Gauge B TP8(2) - Gauge C TP8(2) - Gauge D Reference 1 Figure 1 – Static Pile Load Test (TP8) The following observations and conclusions were made from the test results: • The end bearing resistance did not appear to have been fully mobilised in both the static load and dynamic load tests. • The dynamic load tests provided relatively close match with the static load test results. • The Class A prediction provided relatively good prediction of the pile ultimate load, although the parameters were probably conservative in view of the fact that the test piles were not fully mobilised as indicated above. • The shorter pile TP2 had an initial axial stiffness that was practically the same as the significantly longer test pile TP4. The structural stiffness of the relatively slender piles was found to be important with respect to limiting settlement. During final design, the wall thickness of the piles beneath the tower was increased to 100mm, and it was decided to fill the centre of the hollow spun piles with mass concrete to increase their structural stiffness. • Although the actual failure load was assessed to be greater than the predicted ultimate load, it was decided to limit the operational ultimate load of the tower piles to 6000kN, at which the settlement corresponds to about 10% of the pile diameter. Table 2: Summary of Dynamic Pile Load Test Results (600mm dia. Spun Piles) Test Pene- Date Date Mobilised Mobilised Mobilised Settlement Pile tration Installed Tested Skin End Total at (m) Friction(1) Bearing(1) Resistance Maximum (kN) (kN) (kN) Load (mm) TP2 27 16-9-00 25-9-00 2030 670 2700 15 TP4 57 13-9-00 25-9-00 4870 1860 6730 40 TP5 55 12-9-00 15-9-00 3330 1150 4480 26 12-9-00 25-9-00 4970 1210 6180 31 TP6 44 12-9-00 15-9-00 2740 950 3690 20 48 15-9-00 25-9-00 3590 1050 4640 26 TP7 55 15-9-00 25-9-00 4190 1660 5850 32 TP8 55 16-9-00 25-9-00 4180 1570 5740 32 (1) Mobilised skin friction and end bearing resistance were based on signal matching CAPWAP analysis Table 3: Static Load Test Results And Comparison With Dynamic Load Tests Pile Ultimate Load (kN) Settlement at Ultimate or Maximum Load (mm) Test Pile Class A Static Load Dynamic Load Static Load Test Dynamic Load Prediction Test Test Test TP2 (MLT) 3329 > 2700 2700 20 15 TP4 (CRP) 6533 > 6500 6730 > 77(2) 40 TP8 (MLT) 6062 > 5540(1) 5740 > 40(2) 32 Notes: 1. Maximum load could not be sustained due to equipment problem 2. Pile continues to creep slowly at maximum applied load 3 PILE GROUP SETTLEMENT INTERACTION FACTORS In addition to dial gauges mounted on the test pile, 6 survey reference points were also established at various distances from the pile. Reference 1 was located on the pressure jack (i.e. zero distance from the pile) and 5 other references located at distances of 2m, 5m, 10m, 15m, 20m and 25m from the pile. This was an attempt to assess pile group settlement interaction effects, although the survey points established would at best only measure the ground settlement influence due to pile settlement. Figure 2 shows the measured ground settlement profiles at various distances away from test pile TP4. The pile to pile settlement interaction was expected to be less than the ground to pile settlement interaction and therefore the adopted values were those assessed using program DEFPIG (Poulos, 1991), which are also shown in Figure 2. However, DEFPIG calculates the interaction of two adjacent piles in a group, and does not take into account the presence of intermediate piles that may provide shielding effects which tend to reduce the settlement interaction. Furthermore, when the distance between two piles increases, the operating soil modulus between the two piles becomes higher due to small strain effects. Therefore, for large pile groups, the use of theoretical pile interaction factors from single piles has been known to over-predict the pile group settlement as discussed in Poulos (1993). This aspect will be further discussed later in Section 5 of this paper. 1 0.9 Settlement Interactor Factor 0.8 TP4 Load Test 0.7 (Measured 0.6 Ground/Pile Settlement Ratio) 0.5 0.4 Pile to Pile Interaction Assessed 0.3 Using DEFPIG 0.2 0.1 0 0 10 20 30 40 50 spacing to diameter ratio, s/d Figure 2 – Pile Group Settlement Interaction Factors 4 PILED RAFT DESIGN 4.1 Piled Raft Design Concept For piled raft design, it is a well recognised and accepted practice that the piles do not need to be designed to have conventional geotechnical factors of safety. The piles could be regarded as “settlement reducers” and as long as the entire foundation system has a satisfactory factor of safety, and the system performs satisfactorily with respect to serviceability criteria, some of the piles can be designed to “yield” under ultimate load conditions. If a conventional pile foundation design were to be adopted, over 400 piles would have been required for this project. Using the piled raft design concept, only 293 piles were required. The design involved a number of iterations to optimise the raft thickness and piling layout. The final solution adopted comprised a raft having a thickness of 0.5m in the Podium area and a thickness of 1.7m beneath the Tower over an area of about 49m by 43m that extends 4m beyond the edge of the actual tower footprint. The extension of the thicker raft beyond the edge of the Tower was to enable piles to be strategically located at the corners of the Tower where load concentration was found to occur based on the analysis results (see Section 4.2). 293 nos. prestressed spun piles were adopted in the final design as summarised in Table 4 below: TABLE 4: DETAILS OF PROPOSED SPUN PILES Adopted Wall Design Design Allowable Location Pile Ultimate thickness Length Structural Capacity Diameter Geotechnical Numbers (mm) below (kN) (mm) Capacity Base of Compression Tension (kN) Raft (m) Podium 500 2,100 87 80 23.5 2,000 760 Raft 600 2,700 34 80 23.5 2,800 910 Tower 600 6,000 76 110 to 17m 3,700 930 Columns 80 thereafter 53 Core 600 filled 6,000 96 110 to 29m 3,700 930 with 80 thereafter 53 Concrete mass infill concrete ignored 4.2 Design Approach and Analysis Results Initially, hand calculations were carried out to satisfy the Strength Limit State requirements to approximate the number of piles required, such that the overall design capacity of the piled raft φgRug = φgRug (raft) + φgRug (piles) is greater that the design action effect, Rs*. Geotechnical strength reduction factors, φg, of 0.71, 0.67, and 0.74 were adopted for the raft, podium piles, and tower piles respectively, based on the extent of site investigation data and pile load testing. For the detailed design, numerical analyses of the proposed piled raft system were carried out using computer program GARP (Poulos, 1994) to assist the structural designer in the following: • SLS - assess whether the predicted settlement and differential settlements are acceptable, and how these may impact on the structural design and/or construction sequence. • ULS - to obtain the design action effects for design of the structural elements of the foundation system (i.e. raft and piles in that they must have adequate structural capacity against the predicted loads/bending moments). For both of these conditions, φg values of 1.0 are applicable. It should be pointed out that for the calculation of structural design action effects, the use of φg values of less than 1.0 could result in lower structural action effects being calculated, which could result in an unsafe structural design. For the ULS analysis, 11 load cases were analysed with different partial load factors to model various load combinations including wind loads and uplift from buoyancy effects during construction. Bending moments and shear forces were computed to enable structural design to be carried out, and pile loads were computed for checks to be made regarding the structural capacity of the piles. With respect to pile loads under the ULS cases, the following findings were of particular interest: • The analysis indicated a stress concentration at the corner of the core, due to the “rigidity” of the central core. The 1.7m thick raft beneath the core was artificially increased to 25m in the analysis to account for the structural stiffness of the building core. This analysis has enabled the piles to be more efficiently located around the edge and corners of the core. • Under the individual tower columns, maximum working and ultimate pile loads were calculated to be 3,400kN and 5,653kN. The maximum working and ultimate pile loads for the piles beneath the core (mainly at the corners) were calculated to be 3,722kN and 5,987kN respectively under over-turning wind loading. These are slightly above the allowable structural capacity and close to the ultimate geotechnical capacity of the 53m long 600mm diameter spun piles (see Table 4). Under conventional pile design conditions, this would not have been allowed. However, as the entire foundation system has satisfied the design requirements, overloading of some of the piles was considered to be acceptable. For the SLS case, the computed maximum settlement and differential settlements were as follows: Settlement beneath core 130mm Local Raft Rotation 1:257 Differential settlements between the Tower column and the core 1:360 Differential settlement between Tower columns 1:430 Although some of these computed values were relatively high, the design was considered to be acceptable on the acknowledgement that the pile group settlement prediction was probably conservative due to likely over-estimation of pile group interaction effects in this large piled raft due to pile shielding and small strain effects away from the loaded pile. As a precaution, the structural engineer incorporated a construction joint between the Tower and Podium areas to enable rotation to occur to reduce the impact of potential differential settlement. As it turned out, actual settlements and differential settlements were significantly lower than those compared as discussed in Section 5. 5 FIELD PERFORMANCE Following construction of the raft (completed May 2001), 54 survey points were established at the top of the raft for subsequent settlement monitoring. 23 of the survey points were located within the Tower area and the remainder located within the Podium area. The tower structure was completed in November 2001 and monitoring continued until March 2003. The following settlement monitoring results were obtained: • Settlement of the Podium ranged from 0mm to 2mm. • Settlement of the Tower ranged from 15mm to 28mm with an average of about 21mm. • There was practically no increase in settlement between completion of the Tower in November 2001 and March 2003. The maximum settlement observed was only 21.5% of the predicted settlement, and this was surprising even though it was recognised during the design stage that the estimated settlement was likely to be over-predicted due to the conservative estimates made on pile settlement interaction factors. Besides the difficulties associated with estimating pile settlement interaction factors for large pile groups, a possible cause to the relatively small settlement observed may be that the live load component of the structure has not yet been fully effected during the monitoring period. If the live load component (which was about 26.3% of the total working load) were removed, then the predicted maximum settlement would have been about 96mm only. This is still over 3 times the observed maximum settlement. Another possible cause in over-estimating the pile settlement interaction factors for this project is the fact that the piles used are relatively slender, and about 10mm to 15mm of the settlement on the heavily loaded piles was due to elastic compression of the pile shaft. It is probable that the elastic compression of the pile shaft should not have been included in the assessment of pile group settlement interaction. 6 CONCLUSIONS Although the settlement of the piled raft was significantly over-predicted, the project demonstrated the successful use of the innovative piled raft solution. The following benefits were derived: • More economical, driven spun piles were used instead of the original intended large diameter bored piles. • Fewer piles were possible compared to conventional pile foundation design. • Satisfactory over-all performance was achieved even though some of the piles could be loaded close to their ultimate geotechnical capacity under ultimate load conditions. • Design confidence in this project was achieved via a detailed geotechnical investigations and a prototype pile load testing programme, with settlement monitoring during and after construction of the building. The significant over-prediction of settlement observed supports the general view that pile settlement interaction factors may be over-estimated for large pile groups due to small strain and pile shielding effects. The author also speculates that for slender piles, perhaps the elastic compression component should be ignored in assessing pile settlement interaction factors. These aspects warrant further research. REFERENCES Poulos, H.G. (1980) User Guide to DEFPIG – Deformation Analysis of Pile Groups, School of Civil Engineering, University of Sydney. Poulos, H.G. (1993) Settlement prediction for bored pile groups, Deep Foundations on Bored and Auger Piles, Van Impe (ed.) 1993, Balkema, Rotterdam, 103 - 117. Poulos, H.G. (1994) Alternative Design Strategies for Piled Raft Foundations, 3rd Int. Conf. Deep Foundations, Singapore, 239 – 244. Wong P. K. (2004) Ground Improvement Case Studies – Chemical Lime Piles and Dynamic Replacement, Australian Geomechanics Jnl. Vol. 39, No. 2 June 2004, 47-60.