EFFECT OF PILE LAYOUT ON THE BEHAVIOUR OF CIRCULAR PILED RAFT ON SAND

IGC 2009, Guntur, INDIA EFFECT OF PILE LAYOUT ON THE BEHAVIOUR OF CIRCULAR PILED RAFT ON SAND V. Balakumar Senior Consultant, Simplex Infrastructures Limited, Chennai 600 008, India. E-mail: vb_kumar2002@yahoo.com K. Ilamparuthi Professor and Head, Divn. of Soil Mechanics and Foundation Eng., Anna University, Chennai 600 025, India. E-mail: kanniilam@gmail.com ABSTRACT: Piled raft foundation system is increasingly becoming an alternate to deep piles in the case of structures with raft, when raft alone cannot satisfy the settlement requirement. Among the various structures, storage tanks are more sensitive for settlements. Hence the piled raft can become a viable alternate system, when the raft (which forms the base of the tank) is seated on a favorable ground from bearing capacity point of view. For such cases the design economy depends upon the optimized pile design. The layout and the configuration become very important to produce the desired settlement reduction and load sharing with minimum required piles. This paper presents the effect of pile configuration and the pile raft area ratio on the behavior of piled raft on sand based on the results of 1g model studies. 1. INTRODUCTION By tradition structures that are sensitive for settlements (total or differential) have always been supported on pile foundations ignoring the presence and contributions of raft in load sharing. Among the various types of structures storage tanks have more rigorous settlement requirements. Hence effective control of settlement of the foundation system supporting storage tanks becomes paramount importance particularly in soft deposits. In the recent past piled raft foundation which takes into account the contribution of raft has gained the status of an alternative foundation system when raft cannot satisfy the settlement (both total and differential settlements) requirements for the set loading conditions. The economics of design in the case of piled raft is a function of the pile group optimization. One very important aspect to be noted while designing the piled raft for tanks is that the raft will be placed close to the ground surface unlike most of the buildings, where the raft is located at deeper depths due to the presence of basements. In such cases the raft of the piled raft loses the advantages of the relief in the overburden as far the settlement is concerned. Studies on the work done so far indicate that these important aspects have not been covered adequately. The extensive research work done through numerical modeling (Clancy 1993; Gandhi & Maharaj, 1996; Prokoso & Kulhawy 2001; Small & Poulos, 2007) centrifuge model studies (Horikoshi & Randolph 1996) and 1g model tests (Weisner & Brown, 1978) have brought out a very valuable information on the behavior of piled raft and the effect of various parameters on the behavior. But these studies are mostly related to control of differential settlement only. Also most of these works are on over consolidated clay bed. Further the results of numerical studies are subject to numerous simplifications and assumption and such simplification resulted in lot of variations in the outcome as shown by Polous (2001). Moreover adequate importance has not been given for overall settlement reduction behavior. The studies on the behaviour of piled raft on sand have been carried out by Turek & Katzenbach (2003) with 1g models. Although the results are very limited it had brought out the important aspect that the load sharing behavior is dependent on settlement. The results obtained by monitoring the prototype piled raft represent realistic behavior as shown by Katzenbach et al. (1998); Polous (2008). But these structures are very heavily loaded with the raft located at deeper depths and are very thick (thickness varying 1.5 m to 4 m); the supporting strata in almost all the cases is over consolidated clay. The most important step in the design of piled raft, particularly for tanks is the design of pile layout, and studies on this aspect appear to be limited. Optimization studies have been carried out by Kim et al. (2002) relating to the layout of the piles for square piled raft using genetic algorithm and had conducted 1g model studies using steel plates of 3 mm and 6 mm thickness and polycarbonate pipes as piles. Studies through different layouts had established that piles have to be concentrated in the centre to reduce the average settlement. However the results are compared at a settlement level of 3 mm (1% of the pile length) and at such small level of settlement the piled raft behavior is elastic and the pile group shares major part of the load (Balakumar & Ilamparuthi 673

2006). As seen from the review a study on the effect of pile configuration from settlement reduction and load sharing behaviour is essential particularly for piled raft on sand for tank foundations where the raft is located close to ground level. 2. MODEL STUDIES Although 1g model studies may not reflect the true field conditions, such studies have been found to be very useful in understanding general behavior pattern as well as a guide for further studies. With the above in mind a series of small scale 1g model studies were conducted on a circular piled raft. The raft diameter was taken as 200 mm. The model raft had a thickness of 8 mm. For a scale of 1:0, this represents a tank pad of 20 m dia and 800 mm thick. Perspex was used as model material for the pile and raft. Poorly graded fine to medium Palar sand (classified as SP) was rained from a precalibrated height in layers of 0 mm and tamped with a designed tamper to achieve the density. Tests were conducted in sand beds of three different densities. The test set up and the procedure adopted are explained in detail elsewhere (Balakumar & Ilamparuthi 2006). 3. SELECTION OF PILE LAYOUT Figure 1 presents the pile layout commonly used for circular tanks. The first arrangement (Figure 1a) has piles placed radially with a radial angle (RA) of 36, with one pile at the centre of the raft. The total number of piles (N) is 21 with a spacing of 4d (d = diameter of pile) along the radial direction. The second arrangement (Figure 1b) is termed as square grid layout, in which piles are placed at a spacing of 4d and here also the number of piles is 21. In both the cases, the parameters relating to the piles and raft were kept the same, except the arrangement of piles. The diameter and length of the pile were 8 mm and 160 mm respectively and the area ratio (A r ) of piled raft (A r = A p /A, where A p is total cross sectional area of piles and A is the area of raft) was kept as 5.2%. (a) No. OF PILES = 21 RADIAL ANGLE 36 & 4d SPACING (b) No. OF PILES = 21 SQUARE GRID SPACING 4d Fig. 1: Layout of Piles in Circular Piled Raft Figure 2 presents the load-settlement response of plain and piled raft. A study of the curves indicates that the settlement of piled raft is lesser than the plain raft for a given load irrespective of the pile arrangement. It was found that for the given area ratio both the radial and square grid arrangements exhibited almost identical load-settlement behaviour. However, the square grid layout carried a marginally higher load, when the settlement was approaching 20 mm particularly from the settlement level of 14 mm. The maximum variation in the resistance between the two pile layouts is 3% for the maximum piled raft settlement of 20 mm. SETTLEMENT, mm 0 2 4 6 8 12 14 16 18 20 22 24 LOAD, kn 0 1 2 3 4 5 6 7 8 9 PLAIN RAFT RADIAL GRID Piled raft SQUARE GRID t = 8 mm L = 160 mm N = 21 R.A. = 36º D = 200 mm Bed = Dense Fig. 2: Comparison of Load Settlement Response of Plain and Piled Raft Radial and Square Grid Tests were also conducted on piled raft with square and radial arrangement for piles in other densities and compared. The results of tests on other two densities showed virtually no difference in load settlement response between the two pile arrangements of piled raft. Since the radial arrangement is more commonly used in practice for tank pads, it was decided to carry out rest of the studies on piled raft with the piles arranged in the radial directions. 4. GENERALIZED LOAD SETTLEMENT RESPONSE OF PILED RAFT In order to study the load-settlement response of piled raft with minimum number of piles, 11 piles were provided in a radial form. Figure 3 represents the load-settlement response in a characteristics form for piled raft with radial grid for piled raft area ratio of 2.75% and compared with the characteristic response of plain raft tested in medium dense bed. It is seen from the curves that the load taken by the piled raft is higher than the plain raft irrespective of the magnitude of settlement. The settlement of piled raft in the initial stages of loading is much lesser than plain raft. For the settlement of 2 mm the load on piled raft is 2.0kN which is 1% higher than the plain raft load, whereas for the settlement of 20 mm, the piled raft load is 35% higher. In other words the ratio between the load taken by the piled raft and the plain raft progressively reduces with the settlement. At the settlement of 20 mm the excess load taken by the piled raft varies between 30% and 35% for the three densities of sand. The trend seen above indicates that, in the initial stages of 674

loading, the addition of piles makes the system stiffer. The piles function as settlement reducer, and the combined interaction between pile-raft-soil makes the raft to take higher load under reduced settlement. However as the load increases, the settlement of piled raft increases; this is due to the reduction in soil-pile stiffness. This indicates that the provision of piles to the raft is effective when the settlements are less, and in particular settlement less than 2% of the raft size tested. The load-settlement response explained above is seen in sand of all the three densities tested. The load settlement response has three well defined phases. It is also seen in this case that as the loading increases the difference between the capacity of the plain raft and the piled raft reduces indicating that the influence of piles provided reduces at higher load and remains constant beyond a particular level. 5. EFFECT OF PILE CONFIGURATION ON S R AND α pr Having studied the load settlement response of piled raft with radial grid, the test were carried out in three different configurations as shown in Figure 4 with equal number of piles to bring out the effect of pile layout on load-settlement response. In the configuration A, B and C, ten piles are arranged in outer ring, inner ring and both inner and outer ring (alternate ring) respectively and one pile in the centre. The results of the piled raft with area ratio 2.75% is compared with the piled raft of area ratio 5.25% having 21 piles uniformly arranged as shown in Figure 1a (Configuration D). LOAD, kn x -2 00 0 A B Piled 11 Piles raft Plain Plain raft Raft O 0 2 4 6 8 12 14 16 18 20 22 24 SETTLEMENT, mm C d = 8mm L = 200mm L N = = 200mm 11 R.A. = 45º N D = 11 200mm Bed=Medium dense Fig. 3: Characteristic Response of Plain Raft and Piled Raft (Area ratio 2.75%) It can be seen that beyond a settlement level of around 3% of the raft dimension, the soil piled raft stiffness reduced to a low value and the piled raft settles as that of plain raft. This indicates that, even when the area ratio is quite small the pile group enhances the load carrying capacity of the raft as a combined system. However, at higher settlement even though the enhancement in the capacity is less, the pile group contributes to settlement reduction. The first phase of the curve up to a settlement level of around 2 mm represents the elastic behaviour of the entire system. The second phase shows (upto 6 mm settlement) gradual loss of system stiffness (the pile group loses its elastic behaviour) and beyond this stage the loss of stiffness is rapid and at 20 mm (the maximum settlement at which all the tests were terminated) the stiffness is close to that of plain raft. In other words, beyond a settlement level of 3% of the least lateral dimension of the raft, the piled raft system behaves more like plain raft. (a) Configuration A (b) Configuration B (c) Configuration C Fig. 4: Configurations Studied with 11 piles Figure 5 represents the load-settlement response as measured in 1g model tests on piled raft with the piles arranged in different configurations (Fig. 4). It is seen that upto a settlement level of 3 mm the variation in the load taken is quite small as the settlement increases the load taken by the piled raft with piles concentrated in the centre reduces compared to the other two configurations. The load taken by the piled raft with piles evenly distributed increases progressively indicating that at higher level of settlement (around 3% of the least lateral dimension of the raft) the performance of piled raft will be for better for uniformly distributed load when the piles are evenly distributed. It is also seen that the load taken by the piled raft is higher in the initial stages of settlement (upto 3 mm) when the piles are concentrated in the centre. This is perhaps due to the increase in the confining pressure around the piles when the piles are concentrated. This confirms the views of Kim et al. (2002), as stated earlier. Load, kn 0 1 1 2 2 3 3 4 4 5 5 6 6 7 0 2 4 6 8 12 14 t = 8mm L = 160 16 N = 21 Plain raft 18 R.A. = 36º Inner D = 200mm Outer 20 22 Bed = MD Evenly distributed Fig. 5: Load-Settlement Response of Piled Raft with Various Pile Configurations 675

αpr The performance of piled raft is quantified by two other parameters namely settlement reduction ratio S r and load sharing ratio α pr. They are defined as, S R Sr S pr = (1) S r q q pr r α PR = q (2) pr where, S r is settlement of plain raft at a particular load and S pr is settlement of piled raft at the same load q pr is load carried by piled raft for any given settlement q r is load carried by plain raft for the same settlement The influence of pile configuration on load sharing, α pr and settlement reduction, S R was arrived. The variation of α pr with the settlement is compared for all the four configurations of piles in Figure 6. In the case of piled raft with the piles distributed symmetrically and equally to the entire area of the raft (i.e. in configuration D); the α pr value is high irrespective of the settlement when compared with the other configurations namely A, B and C. This response is obvious and is attributed to reduction in the number of piles. Further it can be observed from the figure that the configurations adopted in this study show some difference in the α pr value despite all the configurations (A, B, and C) have same number of piles. Among the configurations A, B and C the configuration C shows higher α pr value than the other two configurations irrespective of the magnitude of settlement excepting for settlements less than or close to critical settlement. The least value is for the configuration B (piles in the inner ring), which is two to four times lesser than the α pr value of the configuration D (area ratio 5.25%). The higher reduction in the α pr value is for the settlement more than mm. For the configuration B the α pr values lie between the pile configurations of B and C. However for the settlements less than the critical settlement the difference in α pr value between the three configurations (A, B, and C) is small. 1.00 0.80 0.60 0.40 0.20 OUTER INNER ALT. FULLY 0.00 0 4 8 12 16 20 24 SR Fig. 6: Effect of Configuration on the Variation of S R and α pr with Settlement In the same figure, the settlement reduction, S R is compared for the configurations of A, B, C, and D. Settlement 1.00 0.80 0.60 0.40 0.20 OUTER INNER ALT. FULLY 0.00 0 4 8 12 16 20 24 reductions are compared for the given settlement of the raft. This figure also shows the trend as seen in the case of relation between load sharing ratio (α pr ) and settlement. From the comparison made it can be said that among the three configurations (A, B and C), the configuration C (piles arranged in alternate ring) has performed better in load sharing and settlement reduction. Thus the piles distributed evenly over the entire area are best choice than concentrating them over a specific area in the case of piled raft subjected to uniformly distributed load. Having established that the performance of piled raft is better when the piles are evenly distributed the tests were repeated for 3 different area ratios namely 9.2%, 5.25% and 4.25%. The characterized loadsettlement response is presented in Figure 7. Table 1 presents the variation of stiffness in three stages. It is seen that the stiffness reduces progressively and reaches to the value equal to that of plain raft. This implies that irrespective of pile-raft area ratio, when piles are evenly distributed addition of even a small number of piles can produce considerable settlement reduction. Load, LOAD, kn Table 1: Variation of Piled Raft Stiffness Area ratio Stiffness at various phases % Phase OA Phase AB Phase BC 9.25 2900 420 280 6.25 2600 390 220 4.25 1600 340 170 0.1 1 O A B Area ratio 9.25% 6.25% 4.25% 0.0 5.0.0 15.0 20.0 25.0 Settlement, SETTLMENT, mm D = 200mm t = 8mm L = 160mm Medium dense Fig. 7: Characterisation Curves for Various Area Ratios 6. CONCLUSION The above study has brought out an important aspect that even when the raft is located close to the ground level, addition of a small number of piles (pile-raft area ratio 2.75%) can reduce the settlement considerably in the case of piled raft on sand. The studies have further proved that from the overall settlement reduction, the performance of piled raft is better when the piles are evenly distributed. It was also seen that when the piles are concentrated in the center, the piled raft had higher stiffness in the initial stages (settlement level 1.5 to 2% of the diameter of the raft used in the test) C 676

and thereafter the loss of stiffness was very rapid. Further the observation has shown that when the piles are evenly distributed the performance of piled raft is identical; but when the area ratio increases beyond 6% the tendency of the system is to behave as fully piled foundation. REFERENCES Balakumar V. and Ilamparuthi K. (2006). Performance of Model Piled Raft on Sand, Proc. Indian Geotechnical Conference 2006, Chennai, India, pp. 463 466. Clancy P. (1993). Numerical Analysis of Piled Raft Foundations, University of Western Australia, PhD Thesis. Gandhi S.R and Maharaj D.K. (1996). Analysis of Piled Raft Foundations, 6th International Conference on Piling and Deep Foundations, Bombay, pp. 1.11.1 1.11.7. Horikoshi K. and Randolph M.F. (1996). Centrifuge Modeling of Piled Raft Foundations on Clay, Geotechnique, Vol. 46, No.4, pp. 741 752. Katzenbach R., Arslan V. and Moorman Ch (1998). Design and Safety Concept of Piled Raft Foundations, Proc. of 3rd Int. Conference on Deep Foundations on Bored and Auger Piles, pp. 439 448. Kim H.T., Yoo H.K. and Kang I.K. (2002). Genetic Algorithm Optimum Design of Piled Raft Foundations with Model Tests, Journal of South East Asian Geotechnical Society, pp. 1 9. Poulos H.G. (2001). Piled Raft Foundation: Design and Application, Geotechnique, Vol. 51, No. 2, pp. 111 113. Poulos H.G. (2008). The Piled Foundation for the Burj Dubai Design and Performance, IGS Ferroco Terzaghi Oration. Prokoso W.A. and Kulhawy F.H. (2001). Contribution of Piled Raft Foundation, Journal of Geotechnical and Geoenvironmental Engineering, ASCE, pp. 17 24. Small. J. and Poulos, H.G. (2007). A Method of Analysis of Piled Raft, th Australia Newzeland Conference on Geo Mechanics, pp. 555. Turek J. and Katzenbach R. (2003). Small Scale Model Tests with Combined Piled Raft Foundations, Proceedings of the 4th International Seminar on Deep foundations on Bored and Augured Piles, Ghent, Belgium, pp. 409 413. Weisner T.J. and Brown P.T. (1978). Laboratory Tests on Model Piled Raft Foundations, Research Report 318, Sydney University. 677