A failure surface for caisson foundations in undrained soils

H.A. Taiebat
University of Technology Sydney, NSW, Australia

J.P. Carter
The University of Sydney, NSW, Australia

ABSTRACT: This paper presents the results of a series of numerical analyses of caisson foundations embedded in a homogeneous soil deforming under undrained conditions. The performance of a typical caisson foundation under separate axial, torsional and lateral forces is investigated, followed by the interaction of these forces with each other. The lateral force is applied at various points along the skirt of the caisson so that the effects of overturning moments are also included in the analyses. The ultimate capacity of the caisson under combined loading is presented in the form of failure envelopes in the axial-lateral, axial-torsional and lateral-torsional loading planes and the axial-lateral-torsional loading space. The results of this study show that although the capacity of the caisson under lateral load depends on the location of the padeye along the caisson skirt, a unique failure envelope, in a non-dimensional form, can be presented for the caisson regardless of the location of the padeye.

INTRODUCTION

Over the last two decades caisson foundations have been used increasingly in marine environments for temporary or permanent mooring of floating offshore facilities, for tension leg platforms as well as for gravity based structures. Recently, their use has also been proposed for foundations for offshore wind turbines. They are considered as particularly reliable and cost-effective alternatives to more conventional mooring systems in deep and ultra deep waters. They have advantages over other conventional offshore anchoring systems because of their large bearing capacity and the efficiency of their installation. Caisson foundations have typically a large diameter and a wide range of length-to-diameter ratios, so they provide relatively large lateral and axial capacity. A caisson can partially penetrate into the soil under its own weight. Further penetration is usually facilitated by pumping water out of the caisson chamber, thus applying suction inside the caisson. The difference between the external and internal fluid pressures acts as an external surcharge pushing the caisson into the soil. This simple installation procedure is probably the greatest advantage of caisson foundations over pile foundations. A suction caisson can be withdrawn later by applying a positive pressure inside the chamber to help pull it out of the soil. In the marine environment caisson foundations are subjected to various combinations of axial, lateral and torsional loading. Acting as a part of mooring system,

a caisson foundation is predominantly subjected to axial and lateral loading, transferred from a mooring line to a padeye on the caisson wall. Any misalignment of the padeye or any change in the direction of the mooring line also induces torsional (twisting) loads on the caisson. Caisson foundations are often subjected to overturning moments; the effects of which are often considered by appropriately selecting the point of application of the horizontal load along the skirt of the foundations. The response of caisson foundations to axial and lateral loads has been studied previously. The focus of some earlier research (e.g. Bransby & Randolph 1997, 1998, Bransby 2001) has been on the capacity mobilized at the tip of the caisson, ignoring the resistance of its skirt. In other research (e.g. Sukumaran et al. 1999, Zdravkovic et al. 2001, Deng & Carter 1999, 2002, Deng et al. 2001) resistances of both the tip and the skirt of the caissons have been considered. The effects of torsional loads on the axial and lateral capacities of caissons have also been investigated by Taiebat & Carter (2004), where it was shown that, in general, torsional loads reduce the axial capacity and lateral resistance of cylindrical caisson foundations. In this study a series of finite element analyses was performed in order to obtain an insight into the interaction of axial, lateral and torsional loading and the capacity of caisson foundations under combinations of these loads. Since the main aim of this study was to investigate the overall interaction of these forces, the

289

Copyright 2005 Taylor & Francis Group plc, London, UK

problem was solved primarily for a typical caisson with a length-to-diameter (aspect) ratio of 2. This is within the typical range encountered in practice. A limited number of analyses was also performed for a caisson with an aspect ratio of 4, to investigate the effects of the aspect ratio on the overall behaviour of the caisson foundations. 2 FINITE ELEMENT MODEL
3.5L

The typical caisson foundation considered here has a diameter D and a length L, and is embedded in a homogeneous soil that deforms under undrained conditions. It was assumed that the soil obeys the Tresca failure criterion. It has a uniform undrained shear strength of su and an undrained Youngs modulus Eu 300 su. A Poissons ratio of 0.5 ( 0.49) was assumed for the soil to approximate the constant volume response under undrained conditions. The rigidity index G/su is therefore equal to 100, where G is the elastic shear modulus of the soil. The caisson material has a Youngs modulus of Ec 1000 Eu. The geometry of the problem under investigation is axi-symmetric, but generally the loading is of course non-symmetric. The axi-symmetric nature of the geometry was exploited to achieve economies in obtaining the finite element solutions. The finite element mesh used in the analyses has 12 wedges of elements in the circumferential direction. Each wedge of the typical caisson consists of 304 isoparametric (20 nodes) brick elements. A thin layer of elements has been used around and under the caisson in order to capture the effects of shearing close to the foundation. A schematic representation of six of the wedges, i.e. half of the threedimensional finite element mesh used in the analyses of the typical caisson, is shown in Figure 1, which also defines the overall geometry of the finite element model. Of course, a finer mesh may be used to obtain a more accurate finite element solution to the problem in hand. However, as the main aim of this study was to find the overall interaction of the forces at foundation failure, this finite element mesh was considered satisfactory. The small strain formulation used for these analyses is based on the semi-analytical approach in finite element modelling described by Zienkiewicz & Taylor (1989), which is an efficient tool for three-dimensional problems. A semi-analytical finite element method, based on representation of nodal variables in terms of discrete Fourier series, has been integrated into the finite element code AFENA (Carter & Balaam 1995). Application of this method in the analyses of threedimensional problems has shown a considerable reduction in computational time, compared with a straightforward 3-D analysis. Details of the semianalytical method used in this study may be found in Taiebat & Carter (2001).

Downloaded by [Delft University of Technology] at 05:57 17 March 2014

8D

Figure 1. Finite element mesh and geometry of the problem.

All analyses were performed under displacementdefined conditions, where a vertical or torsional displacement was applied to the foundation at the ground level, or a lateral displacement was applied at various locations along the skirt of the foundation, above or below the ground level. A combination of two or three components of displacement was also applied to the foundations to investigate their behaviour under combined loading. It is assumed that the loads are applied at a rate sufficiently rapid that the surrounding soil deforms under undrained conditions. No provision has been made to model any separation or de-bonding that may occur between the soil and the foundation. 3 FINITE ELEMENT RESULTS

The resistance of the caisson foundation under each individual component of loading, i.e. axial, lateral or torsional loading, is presented in this section. 3.1 Axial resistance

The axial bearing capacity of a caisson foundation in undrained soil, Vu, can be approximated using the conventional method of bearing capacity calculation, e.g. from the equation suggested by Vesic (1975), as: (1) where A is the plan area of the caisson, Nc is the bearing capacity factor for a strip footing corresponding to the undrained shear strength of the soil, s and d are factors that include the effects of the shape of the footing and the effects of embedment of the foundation.

290

Copyright 2005 Taylor & Francis Group plc, London, UK

30 4 Vertical load / (A . su) 25 20 D 15 10 5 0 0 10 20 30 40 Vertical Displacement . G / (D . su) 50 L/D=4 L/D=2 -V + L 0 0 5 10 Displacement . G / (L . su) 15 20 H / (D. L. su) 3 2 1 L/D=4 L/D=2 L L D D H

Figure 3. Caisson response under lateral loading. Figure 2. Caisson response under axial loading.

Downloaded by [Delft University of Technology] at 05:57 17 March 2014

The bearing capacity factor for undrained soil Nc (2 ). The shape factor for a circular footing is usually suggested as s 1.2. The embedment factor for a circular footing has been suggested as d 1 0.4 tan1(L/D); for the case of L/D 2, the embedment factor is s 1.443 and for L/D 4, the embedment factor is s 1.530. Therefore, for the special cases considered here the axial load capacity of a buried circular footing is given by the conventional method as Vu 8.9 A su for L/D 2 and Vu 9.4 A su for L/D 4. The effects of the adhesion developed on the caisson skirt are not included in this method. Deng & Carter (1999) recognised the effect of adhesion developed on the skirt of the foundation and suggested an equation for the uplift capacity of caisson foundations in an undrained homogeneous soil as: (2) where Np 9.0 is the uplift capacity factor and ce 1 0.4 (L/D). Equation (2) results in an uplift capacity of 19.44 A Asu and 28.08 A su for caissons with L/D of 2 and 4, respectively. It should be noted that the uplift capacity problem of a caisson in soil deforming under undrained conditions is a reverse bearing capacity problem and can be treated similarly (e.g. Anderson et al. 1993). Therefore, Equation (2) can equally be used to calculate the compressive bearing capacity of caisson foundations. The results of the finite element analysis also do not indicate any significant change in the axial capacity when the direction of loading is reversed. The results of the finite element analyses of the caissons under axial loads are presented in Figure 2. The response is approximately linear at the beginning of loading where about 65% of the ultimate axial resistance is mobilized. After a rapid bend in the loaddeflection curve, the rate of increase in the resistance

reduces significantly. At a relatively large displacement the increase in the resistance becomes insignificant and the ultimate load is approached. The ultimate axial resistances of the caisson predicted by the finite element analysis, at vertical displacements of about 50% of the caisson diameter, are 18.4 A su for the caisson with L/D 2 and 26.6 A su for L/D 4. These values are slightly smaller than those predicted by Equation (2). If the skirt adhesion, DLsu, is added to the capacity predicted by the conventional method, equation (1), the results will be 16.9 A su and 25.4 A su for L/D of 2 and 4, respectively. These values are closer to those predicted by the finite element analyses. 3.2 Lateral resistance The lateral resistance of a caisson foundation in an undrained homogeneous soil can be given as: (3) where Hu is the ultimate lateral capacity of the caisson and Nh is defined as the lateral capacity factor. Deng & Carter (1999) suggested a lateral capacity factor that is a function of the point where the load is applied. For lateral load applied at the surface Deng & Carter suggested Nh 4.8. The results of the finite element analysis of the caisson subjected to lateral loading applied at the ground level are presented in Figure 3. Lateral capacity factors of Nh 4.0 and 4.20 are obtained for L/D of 2 and 4, respectively. Note that the lateral displacements are normalised here with respect to L, the skirt length, indicating a larger final lateral displacement for the caisson with L/D 4. The resistance of a caisson against lateral load depends on the location of the padeye, as any eccentricity of the lateral load applies an overturning moment on the caisson. The variation of the lateral resistance of caissons with the point of load application is presented in Figure 4. In this figure z represent the coordinate of the load application point, with its negative

291

Copyright 2005 Taylor & Francis Group plc, London, UK

2.0

7 6 5 T / (D3. su) D T

1.5

4 3 L 2 1 L/D=4 L/D=2 T

1.0

z/L

0.5

L/D=2 L/D=4

0 0 1 2 3 Rotation . G / su 4 5

Figure 5. Caisson response under torsional loading.

0.0

Downloaded by [Delft University of Technology] at 05:57 17 March 2014

-0.5

are 3.40 su D3 and 6.53 su D3 for L/D of 2 and 4, respectively, in excellent agreement with the theoretical values obtained using Equation 4. 4
0 2 4 6 8 H / (D. L. su) 10 12

-1.0

FAILURE ENVELOPES

Figure 4. Lateral resistance versus the load application point.

sign indicating that the point is below the ground level. The responses of the two caissons considered here are very similar; as L/D increases the depth of the maximum lateral resistance shifts slightly upward, toward the ground level. The maximum lateral resistances occur when the lateral load is applied at about 0.6 L below the ground level. 3.3 Torsional resistance

In order to evaluate the interaction between different components of loading, a series of finite element analyses was performed using different ratios of the vertical and lateral displacements and rotation of the caisson. For each loading case the ultimate axial capacity is reached at a vertical displacement equal to about 50% of the diameter of the caisson, while the ultimate torsional and lateral capacities are obtained at displacements equal to approximately 20% of the diameter. The results of the finite element analyses will be presented in the form of two-dimensional failure loci in axiallateral, axial-torsional, lateral-torsional loading planes, followed by a non-dimensional 3-D failure surface. 4.1 Axial-lateral failure plane

Assuming the full value of the undrained shear strength of the soil is mobilized as a shear stress at the caisson-soil interface, the theoretical value for the ultimate torsional capacity of a caisson foundation, Tu, can be calculated as the sum of the base resistance and the shaft resistance, i.e. (4) Equation (4) gives the ultimate torsional capacities of 3.402 su D3 and 6.545 su D3 for caissons with L/D of 2 and 4, respectively. The load deflection curves predicted by the finite element analyses of the caissons under torsional loading are presented in Figure 5. The response of each caisson is virtually elastic-perfectly-plastic. The torsional resistance increases linearly to its ultimate value at a rotation of about 0.35 to 0.45, after which the resistance remains constant. The ultimate torsional resistances predicted by the finite element analyses

Failure loci for caissons with L/D 2 subject to combinations of axial and lateral forces, applied at different points along the caisson skirt below and above the ground level, are presented in Figure 6. As explained previously, the lateral capacity of the caisson depends on the point of application of the load. A non-dimensional form of the failure loci can be obtained using the maximum lateral capacity at any point, Hmax, and the maximum axial capacity, Vmax, as the normalisation factors. These non-dimensional failure loci are presented in Figure 7. Included in this figure is a normalised failure locus for the caisson with L/D 4, where the lateral load is applied at the ground level. It can be seen that all the failure loci are very close to each other and the discrepancies between different failure loci are almost insignificant. Therefore it may be concluded that for most practical purposes and for any caisson a unique normalised failure locus in the axial-lateral loading plane can be found

292

Copyright 2005 Taylor & Francis Group plc, London, UK

20 16 Vu / (A.su) 12 8 L 4 0 0 2 4 6 Hu / (D . L . Su) 8

Vu / Vmax

D +V z H

z /L = 1.0 z /L = 0.5 z /L = 0.0 z / L = -0.6 z / L = -1.0

1.0 0.8 0.6 0.4 0.2 0.0 +V T

10

12

0.0

0.2

0.4

0.6 Tu / Tmax

0.8

1.0

Downloaded by [Delft University of Technology] at 05:57 17 March 2014

Figure 6. Failure locus in the axial-lateral loading plane for caissons with L/D 2.
1.0

Figure 8. Non-dimensional failure locus in the axialtorsional loading plane.

1 0.8 0.8 Vu / Vmax 0.6 0.4 L 0.2 0.0 0.0 0.2 0.4 0.6 Hu / Hmax 0.8 1.0 0 0 0.2 0.4 0.6 Tu / Tmax 0.8 1 D +V z H z/L= 1.0, L/D=2 z/L= 0.5, L/D=2 z/L= 0.0, L/D=2 z/L=-0.6, L/D=2 z/L=-1.0, L/D=2 z/L= 0.0, L/D=2 z Hu / Hmax 0.6 L 0.4 0.2 T H z / L= 0.0 z / L=-0.6 z / L=-1.0

Figure 7. Non-dimensional failure locus in the axiallateral loading plane.

Figure 9. Non-dimensional failure locus in lateraltorsional loading plane.

which is representative of all failure loci for caissons with L/D 2. Furthermore, this failure locus may also be applicable to caissons with other aspect ratios, although this point warrants further investigation. 4.2 Axial-torsional failure plane A normalised failure locus for the caisson with L/D 2 subjected to combinations of axial and torsional forces is presented in Figure 8. In this figure, the axial and torsional forces are normalised by their maximum values, Vmax and Tmax, obtained under either pure axial or pure torsional loading. It is noted that these are small strain results and in reality the ultimate axial capacity may increase slightly at larger vertical displacements. The failure locus presented in Figure 8 shows that torsional forces have a very significant effect on the axial capacity of the caisson. When much of the shearing strength around and under the caisson is mobilized by torsion, the axial capacity of the foundation reduces significantly. For axial loads lower than about 0.6 Vmax, torsional displacements govern the failure mechanism of the caisson foundation.

4.3

Lateral-torsional failure plane

The ultimate response of the caisson foundation, with L/D 2, to different combinations of lateral and torsional deformations is presented in Figure 9 as failure loci for 3 different points of lateral load application. The ultimate lateral and torsional forces are normalised by their maximum values, Hmax and Tmax, which can be obtained for pure torsional and pure lateral loading at different points of load application. Figure 9 shows that as the torsional force increases to its maximum value, the lateral resistance of the foundation decreases to about 60% of its maximum value. For lateral loads lower than about 0.6 Hmax, torsional displacements govern the failure mechanism of the caisson foundation. 4.4 Axial-lateral-torsional failure surface

A series of finite element analyses was performed on the typical caisson with L/D 2 in order to obtain a general failure surface for the foundation. Various combinations of axial, lateral and torsional displacements were applied to the foundation. In all the analyses

293

Copyright 2005 Taylor & Francis Group plc, London, UK

1.0 0.8 Tu 0.6 Tmax 0.4 0.2 0.0 -0.8 -0.4 0.0 Vu/Vmax 0.4 0.8 -0.8 -0.4 0.4 0.0 Hu/Hmax 0.8

the failure surface obtained for z/L 0, and shown in Figure 10, can be used for other values of z/L. Furthermore, based on the results of the limited study performed here on the effects of the aspect ratio of the caisson foundation, it was shown that the nondimensional failure planes of the caisson with L/D 4 are very similar to those of the caisson with L/D 2. Therefore, the shapes of the failure envelopes for foundations with various aspect ratios are probably very similar to the non-dimensional 3-D failure envelope shown in Figure 10. 5 CONCLUSIONS

Figure 10. Three-dimensional failure envelope in nondimensional load space.

Downloaded by [Delft University of Technology] at 05:57 17 March 2014

1.0 Tu/Tmax 0.8 0.6

0.95 0.9 0.6 0.5 0.0 0.7 0.8

0.4
0.99

0.2 Vu Vmax 0.0 -0.2 -0.4 -0.6 -0.8 -1.0 -1.0 -0.8 -0.6 -0.4 -0.2 0.0
1.0

0.2

0.4

0.6

0.8

1.0

Hu/Hmax

Results of a numerical study of the capacity of caisson foundations under axial, lateral, and torsional loads have been presented. Failure envelopes were presented in non-dimensional form and it has been shown that a single non-dimensional failure envelope can be obtained for the foundation in the different loading planes. The limited study performed here did not indicate significant changes in the shape of the failure envelopes for caisson foundations with different aspect ratios. It can be tentatively concluded that the nondimensional failure surface obtained in this study may be useful for caissons with various aspect ratios. Having the maximum resistances of a caisson foundation against purely axial, lateral and torsional loading, the resistance of the foundation against any combination of the axial, lateral and torsional loading can be obtained using the general failure envelope presented here. ACKNOWLEDGEMENT The research described in this paper was conducted as part of the work of the Special Research Centre for Offshore Foundation Systems, established and supported under the Australian Research Councils Research Centres Program.

Figure 11.

Failure loci in the non-dimensional VHT space.

the horizontal load was applied at the ground level, i.e. z/L 0. The three-dimensional failure surface was constructed based on a triangulation scheme of linear interpolation between computed data points. A three-dimensional image of the failure envelope for the caisson in the axial-lateral-torsional loading space is presented in Figure 10. Only half of the failure envelope is shown as the uplift capacity problem of a caisson in clay soils deforming under undrained conditions is merely a reverse bearing capacity problem. Representation of the failure loci in the VHT loading space is shown in Figure 11, where contours of equal torsional load are presented. In these figures all loads are normalised by their maximum values, Vmax, Hmax and Tmax, obtained under purely axial, lateral and torsional loading. As shown in the previous sections, a single nondimensional failure envelope in the HT or VT loading planes can be used for the foundation regardless of the point of the application of the lateral load. Therefore,

REFERENCES
Andersen, K. H., Dyvik, R., Schoder, K., Hansteen, O. E., & Bysveen, S. 1993. Field test of anchors in clay, II: Prediction and interpretation. Journal of Geotechnical Engineering. ASCE, 119, 5321549. Bransby, M. F. & Randolph, M. F. 1997. Finite element modelling of skirted strip footings subjected to combined loadings. Proc. 7th Int. Offshore and Polar Engineering Conference. Honolulu, USA, 791796. Bransby, M. F. & Randolph, M. F. 1998. Combined loading of skirted foundations. Gotechnique. 48(5), 637655. Bransby, M. F. 2001. Finite element analysis of jacket structures with bucket foundations. Proc. 10th Int. Conference on Computer Methods and Advances in Geomechanics.

294

Copyright 2005 Taylor & Francis Group plc, London, UK

Carter J. P. & Balaam N. P. 1995. AFENA users manual. Version 5, Centre for Geotechnical Research, The University of Sydney, Australia. Deng, W. & Carter, J. P. 1999. Analysis of suction caissons in uniform soils subjected to inclined uplift loading. Report No. R798, Department of Civil Engineering, The University of Sydney, Australia. Deng, W., Carter, J. P. & Taiebat, H. A. 2001. Prediction of the Lateral Capacity of Suction Caissons, Keynote Lecture, Proc.10th Int. Conference on Computer Methods and Advances in Geomechanics, 1, 3338. Deng, W. & Carter, J. P. 2002 A theoretical study of the vertical uplift capacity of suction caissons. Int. J. of Offshore and Polar Engineering., 12(2), 8997. Sukumaran, B., McCarron, W. O., Jeanjean, P. & Abouseeda, H. 1999. Efficient finite element techniques for limit analysis

of suction caissons under lateral loads. Computers and Geotechnics. 24, 89107. Taiebat, H. A. & Carter, J. P. 2001. A semianalytical finite element method for three-dimensional consolidation analysis. Computer and Geotechnics. 28, 5578. Taiebat, H. A. & Cater. 2004. Effects of torsion on caisson capacity in clay. Proc. 9th Australia New Zealand Conference on Geomechanics. 1, 130136. Vesic, A. S. 1975. Bearing capacity of shallow foundations, Foundation Engineering Handbook, Eds Winterkorn & Fang, Van Nostrand Reinhold, New York, 121147. Zdravkovic, L., Potts, D. M. & Jardine, R. J. 2001. A parametric study of the pull-out capacity of bucket foundations in soft clay. Gotechnique. 51(1), 5567. Zienkiewicz, O. C. & Taylor, R. L. 1989. The Finite Element Method. 4th Edition, McGraw-Hill, New York.

Downloaded by [Delft University of Technology] at 05:57 17 March 2014

295

Copyright 2005 Taylor & Francis Group plc, London, UK