Modelling of Cyclic Soil Degradation

Modelling of cyclic soil degradation
Development of a cyclic accumulation model

and the application to a gravity based foundation
Author
C. Rasch
Delft University of Technology

Faculty of Civil Engineering and Geosciences
Department of Geoscience & Engineering
Section Geo-Engineering
Picture on front page taken from baminnovation.eu/nl
Modelling of cyclic soil degradation
Development of a cyclic accumulation model and the application to a gravity based foundation
MSc thesis
by
C. Rasch
in partial fulfilment of the requirements for the degree of

Master of Science in Civil Engineering
October 2016
Graduation committee
Chairman
Prof. dr. M. A. Hicks

TU Delft, Geo-Engineering
Daily supervisor
Dr. Ir. R. B. J. Brinkgreve

TU Delft, Geo-Engineering, Plaxis bv
Supervisor BAM Infraconsult
Dr. Ir. S. E. J. Spierenburg

BAM Infraconsult
External supervisor
Ir. J. S. Hoving
Offshore Engineering
An electronic version of this thesis is available at http://repository.tudelft.nl/

Acknowledgement
This thesis is the final assignment of the master Civil Engineering at the Delft University of Technology. This research
would not have been possible without the help of many people. This thesis has been performed at the section of Geo-
Engineering of the Delft University of Technology, BAM Infraconsult in Gouda, and Plaxis bv in Delft. This thesis has been
supervised by the committee, who I would like to thank for their time and guidance during the meetings.
I would like to thank Bart van Paassen for the opportunity to graduate at BAM Infraconsult in Gouda. My gratitude
goes to Sjoert Spierenburg, for his guidance in my project and the many discussions we conducted. I am also grateful for
the opportunity to experience the process of the development of gravity based foundations.
I would like to thank Anita Laera for the time and effort she put into guiding me in the inspection of the new soil model
at Plaxis. During my time at Plaxis she was always willing to listen and help, if I had any questions or comments. Also a
special thanks to Ronald Brinkgreve, for the opportunity to work at Plaxis and the time he made for the discussions on my
findings.
C. Rasch
Gouda, October 2016
I
Abstract
This thesis entitled "Modelling of cyclic degradation - Development of a cyclic accumulation model and the application
to a gravity based foundation" presents a new soil model to analyse the behaviour of undrained soils subjected to cyclic
loading. Cyclic loading on offshore support structures for wind-turbines results in complex design challenges, and current
codes and standards offer limited guidance on the assessment of undrained cyclic soil behaviour. To assess the undrained
cyclic behaviour, a new procedure is being developed within the finite element software Plaxis, the UDCAM-S procedure.
This new procedure is designed to asses the undrained cyclic behaviour of clays in the design of offshore support structures.
This study focuses on the development of this procedure and the application to a gravity based foundation for a large
wind-turbine placed on clay. Large cyclic wind and water loads acting on the turbine are transferred through the structure
to the foundation, resulting in cyclic shear stresses in the clay. These shear stresses lead to an increase in pore pressure due
to the restriction of pore pressure dissipation, and thereby a decrease in effective stress. With each cycle, shear strains are
developed due to the decrease in effective stress: the average, cyclic and permanent shear strains. Permanent shear strains
accumulate as a result of accumulated pore pressures due to cyclic shearing of the soil, the irreversible deformations of the
soil. With the accumulation of pore pressures and plastic shear strains the soil degrades in terms of the undrained shear
strength and the stiffness of the soil.
The degradation of the soil as a result of a design storm is determined with the UDCAM-S procedure. The design
storm is simplified by analysing the cyclic loads in terms of load parcels, to use the load data in the strain accumulation
procedure (Andersen, 1976). With the result of the strain accumulation procedure, the degraded behaviour of the soil
can be determined by means of scaled contour diagrams for the behaviour in direct simple shear, triaxial compression,
and triaxial extension. Parameters of the NGI-ADP soil model in Plaxis are fitted onto the degraded stress-strain curves,
resulting in a parameter set accounting for the degradation of the soil by the given design storm.
The results of the UDCAM-S procedure have been used to model the influence of cyclic degradation on the stability
of gravity based foundations. The investigated cases gave insight in the influence of the different steps in the UDCAM-S
procedure on the outcome of a finite element analysis. It has been found that VHM capacity of gravity based foundations
can be reduced significantly by large equivalent number of cycles for a design storm. As the soil strength decreases with
increasing equivalent number of cycles, the maximum moment resistance is mobilized at lower vertical loads. This results
in a large reduction of the HM capacity at higher vertical loads. The same effect has been found on the stability of the
foundation for different widths of the foundation. The equivalent number of cycles is the dominant factor influencing the
stability of the foundation. This is due to the accumulation of large strains due to a design storm, resulting in a softer clay.
III
Symbols & Abbreviations
ABBREVIATIONS
CSR Cyclic stress ratio
DAF Dynamic Amplification Factor
DSS Direct Simple Shear test
GBF Gravity Based Foundation
NGI Norwegian Technological Institute
OCR Overconsolidation ratio
OTM Overturning Moment
SLS Serviceability Limit State
TXC Triaxial Compression test
TXE Triaxial Extension test
UDCAM UnDrained Cyclic Accumulation Model
UDCAM-S Simplified UnDrained Cyclic Accumulation Model
ULS Ultimate Limit State
VHM Failure diagrams in terms of the Vertical load, Horizontal load, and Overturning moment
SYMBOLS
χ Scaling factor as a result of the strain accumulation procedure [-]
xx Horizontal strain [-]
yy Vertical strain [-]
γa Average shear strain [%]
γfC Shear strain at failure in the triaxial compression test [%]
γfE Shear strain at failure in the triaxial extension test [%]
γfDSS Shear strain at failure in the direct simple shear test [%]
γp Plastic shear strain [%]
V
γcy Cyclic shear strain [%]
γxy Shear strain [-]
ν Poisson’s ratio [-]
σ0 Effective normal stress [kPa]
σ1 Main principal total stress [kN/m2 ]
σ3 Minor principal total stress [kN/m2 ]
σxx Horizontal total stress [kN/m2 ]
σyy Vertical total stress [kN/m2 ]
τ Shear stress [kPa]
τ0 /SuA Initial mobilisation of the shear stress with respect to the undrained shear strength in active failure
[-]
τa Average shear stress [kPa]
τcy Cyclic shear stress [kN/m2 ]
τcy /τa Cyclic stress ratio [-]
Fcy /Fa Cyclic load ratio [-]
Fcy /Fcy,max Cyclic load ratio of the load parcels [-]
G/SuA The ratio of the degraded shear modulus (in MPa) to the degraded undrained shear strength (in
kPa) [-]
qc Cone resistance of the CPT cone [MPa]
SuA The degraded undrained shear strength in active failure in the UDCAM-S procedure [-]
SuE /SuA The ratio of the degraded shear stress at failure in the triaxial extension test to the degraded
undrained shear strength in active failure in the UDCAM-S procedure [-]
SuDSS /SuA The ratio of the degraded shear stress at failure in the direct simple shear test to the degraded
undrained shear strength in active failure in the UDCAM-S procedure [-]
A
Su,inc Increase of shear strength with depth [kN/m2 /m]
yref Reference depth [m]
B Width of the foundation [m]
Fa Average load [kN]
Fcy Cyclic load [kN]
H Horizontal load [kN]
K0 The neutral earth pressure coefficient [-]
M Overturning moment [kNm]
N Number of cycles [-]
Neq Equivalent number of cycles as a result of the strain accumulation procedure [-]
V Vertical load [kN]
VI
Contents
List of Figures IX
List of Tables IX
1 Introduction 1
1.1 Research questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Readers manual . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Gravity based foundations 5

2.1 Design of Gravity based foundations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 Standard offshore foundation design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Combined loading on an offshore gravity based foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2.1 Cyclic loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.2 VHM capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3 Undrained cyclic soil behaviour 11

3.1 Undrained behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1.1 Undrained shear strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1.2 Pore pressure built-up in cyclic loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2 Cyclic soil strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2.1 Stress path . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2.2 Combination of average and cyclic shear stresses in the soil . . . . . . . . . . . . . . . . . . . . . . . 14
3.2.3 Loading history of the soil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2.4 Strain compatibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.3 Equivalent overconsolidation hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.4 Current available options for the modelling of the undrained cyclic soil behaviour . . . . . . . . . . . . . . . 16
3.4.1 Undrained behaviour in Plaxis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.4.2 Available advanced soil models for undrained cyclic soil behaviour . . . . . . . . . . . . . . . . . . . 16
3.4.3 Overview of the soil models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4 The Simplified UnDrained Cyclic Accumulation Model 19

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2 Transformation of stresses and strains for the input of Plaxis . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2.1 Triaxial Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2.2 Triaxial Extension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2.3 Direct Simple Shear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.3 Step 1 - Determination of the equivalent number of cycles of a design storm . . . . . . . . . . . . . . . . . . 22
4.3.1 Strain accumulation for different OCR values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.4 Step 2 - Determination of the degraded stress-strain response . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.4.1 Resulting contour diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.4.2 Determination of the stress path . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
VII
4.5 Step 3 - Parameter determination of the NGI-ADP soil model . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.5.1 Initial values of the laboratory tests for the optimisation . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.5.2 Optimization of the stiffness parameter G/SuA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.5.3 Recommended procedure for the optimisation of the NGI-ADP parameters . . . . . . . . . . . . . . . 32
4.5.4 Softening in compression for normally consolidated clays . . . . . . . . . . . . . . . . . . . . . . . . 32
4.6 Verification with the Case-Study presented by Jostad et al. (2015) . . . . . . . . . . . . . . . . . . . . . . . 33
4.6.1 Strain accumulation procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.6.2 Idealized stress path . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.6.3 Parameter determination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5 Modelling the influence of cyclic degradation on the stability of a gravity based foundation 39
5.1 Finite element model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
5.1.1 Foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
5.1.2 Mesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.1.3 Influence of vertical and horizontal loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.1.4 Verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.1.5 Displacement controlled . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5.1.6 Python scripting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5.2 Parameter determination using the UDCAM-S procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.2.1 Influence of an increasing Neq for OCR = 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.2.2 Influence of an increasing Neq for OCR = 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.2.3 Influence of an increasing Neq for OCR = 1 assuming isotropic behaviour . . . . . . . . . . . . . . . 53
5.2.4 Influence of the cyclic load ratio for OCR = 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.3 Resulting VHM failure diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.3.1 VHM diagrams for OCR = 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.3.2 VHM diagrams for OCR = 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.4 The influence of the foundation width . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.4.1 Influence of the foundation width for overconsolidated Drammen Clay . . . . . . . . . . . . . . . . . 62
5.4.2 Influence of the foundation width for OCR = 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
6 Discussion 67
7 Conclusions & Recommendations 69

7.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
7.1.1 Answers to the formulated sub-questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
7.1.2 Main conclusions of this research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
7.2 Recommendations for further work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Appendices 77
A Design codes for offshore gravity based foundations: DNV-GL (2016), API (2002), and API (2011) A-1
B NGI-ADP model B-1
C Python Script for automatic Plaxis Calculations C-1
D VHM failure contour diagrams D-1
VIII
List of Figures
1.1 The distance to the shore and the water depth of the constructed wind farms (Andrew et al., 2016) . . . . . 1
1.2 Development of the wind farm locations considering the distance to the shore and the water depth (Andrew
et al., 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Different offshore foundation types (adapted from Bhattacharya, 2014) . . . . . . . . . . . . . . . . . . . . 2
1.4 Readers manual to the different chapters of this thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1 A gravity based foundation for wind-turbines placed on the seabed . . . . . . . . . . . . . . . . . . . . . . . 5

2.2 Standard simplification of the soil-structure interaction of offshore foundations (Zaaijer, 2007) . . . . . . . . 6
2.3 Loads on the super-structure and the transmission of the loads to the foundation (Byrne, 2011) . . . . . . . 7
2.4 Example of the development of the overturning moments for a time domain of 600 seconds . . . . . . . . . 8
2.5 Example of the development of the horizontal loads for a time domain of 600 seconds . . . . . . . . . . . . 8
2.6 Displacement contours plots for pure loading of a gravity based foundation without embedment allowing
tension at the soil-structure interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.7 Sign convention for loads in the VHM failure diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.8 Failure locus found for a circular footing subjected to moment and horizontal loading (with V = 0) by
Ukritchon et al. (1998) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.9 Difference in soil movements between different loading situations regarding the direction of the horizontal
load and the overturning moment (Taiebat and Carter, 2000) . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.1 Generation of pore pressures and strains in a cyclic triaxial test (Recreated from Andersen, 2009) . . . . . . 12
3.2 Classification of the stress states underneath a gravity based foundation in three laboratory tests (Andersen,
2015) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.3 Contour diagram for a triaxial test for 10 cycles for overconsolidated Norwegian Drammen Clay . . . . . . . 14
4.1 Example of the resulting load parcels from the structural loads in a design storm . . . . . . . . . . . . . . . 22
4.2 Principe of the strain accumulation, including the incremental shear strain with an increment of the cyclic
shear stress (Recreated from Andersen, 2015) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.3 Increment of the cyclic shear strain γcy,i due to an increment of the cyclic shear stress (Recreated from
Andersen, 2015) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.4 Effect of the overconsolidation ratio on the normalized stress-strain relationship of Drammen Clay with
overconsolidation ratios of 1, 4, and 40 (Recreated from Andersen, 2015) . . . . . . . . . . . . . . . . . . . 24
4.5 End points of the strain accumulation procedure for each scaling factor for the set of load parcels in table 4.2
for the different OCR values. The blue lines represent the cyclic shear strain contours, with the outermost
contour defining failure (γcy = 15%). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.6 Resulting triaxial contour of OCR 1 for Neq 5 and 50 with the average strain contours (blue contour lines),
and the cyclic strain contours (red contour lines) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.7 Determination of the stress path using the DSS contour diagram in case of a jack-up structure and the
resulting target stress-strain diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.8 Triaxial test set-up for the optimisation in the third step of the procedure . . . . . . . . . . . . . . . . . . . . 30
4.9 Optimized stress-strain curves with a large optimisation range for G/SuA . The blue line is the target stress-
strain curve, the orange line is the optimized stress-strain curve. . . . . . . . . . . . . . . . . . . . . . . . . 30
4.10 Optimization of G/SuA using only a selection of the target stress-strain points for the triaxial extension test . 31
4.11 Influence of the stiffness parameter on the stress-strain response in direct simple shear for a soil with γfDSS
= 10% and SuDSS = 100kPa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
IX
4.12 Softening in triaxial compression with the actual stress path, represented by the continuous black line, and
the idealized stress path in the UDCAM-S, represented by the dashed black line (adapted from Andersen,
2015) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.13 Geometry of the spudcan and the parameters of the soil beneath the spudcan by Jostad et al. (2015) . . . . 33
4.14 Locus of end points from the strain accumulation procedure for all scaling factors . . . . . . . . . . . . . . . 34
4.15 Visual representation of the 2D contour diagrams for Neq = 5 for OCR = 4 with the defined stress paths for
DSS, TXE, and TXC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.16 Resulting stress-strain responses for the cyclic and total analyses for OCR = 4 by Jostad et al. (2015) . . . . 36
4.17 Resulting laboratory tests fitted on the target stress-strain curves for the cyclic analysis for OCR = 4 by the
UDCAM-S procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.18 Resulting laboratory tests fitted on the target stress-strain curves for the total analysis for OCR = 4 by the
UDCAM-S procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
5.1 Finite element mesh for the analysis of the gravity based foundation . . . . . . . . . . . . . . . . . . . . . . 40
5.2 Meshes for four different coarseness factors for the soil close to the foundation . . . . . . . . . . . . . . . . 41
5.3 Mesh dependency of the ultimate loads for different mesh refinements of the inner soil surface for Neq = 5,
and 10 with OCR = 4 and τcy /τa = 1.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.4 Developed plastic strains (γp ) in pure horizontal and vertical loading at failure . . . . . . . . . . . . . . . . 42
5.5 Validation of the load controlled finite element analysis with Taiebat and Carter (2002) . . . . . . . . . . . . 43
5.6 Failure mechanisms in terms of the total displacements [m] for four different load ratios of h/m with a
vertical load of v = 2.6 V/(A/Su ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.7 Pseudo code of the python script for automatic Plaxis calculations using Excel, Plaxis 2D and Python . . . . . 44
5.8 Fixed outcomes of the strain accumulation procedure to Neq = 1, 5, 10, and 20 for the Drammen Clay with
OCR = 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.9 Resulting direct simple shear diagrams for N = 1 with idealized stress path inclinations of τcy /τa = 1.5 in
DSS, for TXC and TXE respectively 1.72 and -17.0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
DSS, for TXC and TXE respectively 1.62 and -15.5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
DSS, for TXC and TXE respectively 1.575 and -14.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.12 Resulting triaxial contour diagrams for N = 20 with idealized stress path inclinations of τcy /τa = 1.5 in
DSS, for TXC and TXE respectively 1.49 and -15. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.13 Resulting target stress-strain curves for the idealized stress paths in figure 5.9 - 5.12 . . . . . . . . . . . . . . 48
5.14 Stress-strain curves from the fitted NGI-ADP parameters on the target stress-strain curves from figure 5.13 . 48
5.15 Fixed outcomes of the strain accumulation procedure to Neq = 1, 5, 10, and 20 for OCR 1 of the Drammen
Clay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
DSS, and TXC -4.2 TXE -0.35 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
DSS, and TXC -2.6 TXE -0.37 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
DSS, and TXC -2.15 TXE -0.38 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.19 Resulting triaxial contour diagrams for N = 20 with idealized stress path inclinations of τcy /τa = 1.5 in
DSS, and TXC -1.95 TXE -0.36. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.20 Resulting target stress-strain curves for the idealized stress paths in figure 5.16 - 5.19 . . . . . . . . . . . . . 51
5.22 Resulting direct simple shear contour diagrams for OCR 1 with idealized stress path inclinations of τcy /τa
= 1.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.23 Target stress-strain curves resulting from the stress paths in figure 5.22 assuming isotropic behaviour . . . . 54
5.24 Stress-strain curves from the fitted NGI-ADP parameters on the target stress-strain curves from figure 5.13
assuming isotropic behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.25 Contour diagrams with the idealized stress paths for OCR 4 with Neq = 5 and varying cyclic stress ratio’s . . 55
5.26 Resulting target stress-strain curves for the idealized stress paths in figure 5.25 . . . . . . . . . . . . . . . . 56
5.28 HM failure diagrams for the parameter sets shown in table 5.4 normalised to the undrained shear strength
for the soil after N = 1 with applied vertical loads of 3482 kN/m (0.3 Vult,N =1 ), 5803 kN/m (0.5 Vult,N =1 ),
and 8124 kN/m (0.7 Vult,N =1 ). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
X
5.29 VM diagram for H = 0 for the Neq values of 1, 5, 10, and 20. The lines are plotted by modifying the equation
of Houlsby and Purzin (1999) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.30 Influence of the cyclic stress ratio on the HM capacity diagram at a vertical load V = 0.5 Vult = 4449 kN/m 58
5.31 HM diagrams for the parameter sets shown in table 5.5 with applied vertical loads of 3068 kN/m (0.3
Vult,N =1 ), 5113 kN/m (0.5 Vult,N =1 ), and 7159 kN/m (0.7 Vult,N =1 ), normalised to the undrained shear
strength for the soil after Neq = 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.32 HM diagrams for the parameter sets shown in table 5.6 with applied vertical loads of 3068 kN/m (0.3
Vult,N =1 ), 5113 kN/m (0.5 Vult,N =1 ), and 7159 kN/m (0.7 Vult,N =1 ), normalised to the undrained shear
strength for the soil after N = 1 from case 2 (non-isotropic). . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.33 Normalised ultimate limit loads in 3D space for Neq = 1 (continuous line) and Neq = 10 (dotted line) for
both anisotropic (blue lines) an isotropic shear strengths (red lines) for normally consolidated Drammen Clay 61
5.34 Differences in width of the foundation slab with an identical superstructure . . . . . . . . . . . . . . . . . . 62
5.35 Total deformations for three different foundations widths with equally applied total loads . . . . . . . . . . 63
5.36 Resulting overall safety factors for Neq of 1, 5, and 10 for OCR = 4 . . . . . . . . . . . . . . . . . . . . . . . 64
5.37 Relative overall safety factor with respect to the soil strength after Neq = 1 from figure 5.36 . . . . . . . . . 64
5.38 Influence of the cyclic stress ratio (τcy /τa ) for OCR = 4 after Neq = 5 . . . . . . . . . . . . . . . . . . . . . 65
5.39 Difference in overall safety factor between the overconsolidation ratios of the Drammen clay at different
widths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.40 Difference in overall factor of safety for isotropic and non-isotropic undrained shear strength for OCR = 1
for different widths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
7.1 Results for the overall safety factor for different widths of the strip foundation for a homogeneous soil of
OCR = 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
7.2 VM diagram for H = 0 for the Neq values of 1, 5, 10, and 20. The lines are plotted by modifying the equation
of Houlsby and Purzin (1999) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
A.1 Load centre and the effective area of a GBF (DNV-GL, 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . A - 2
A.2 Effective area according to the API (API, 2002) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A - 3
B.1 Visual representation of the NGI-ADP failure criterion in the π-plane (Brinkgreve et al., 2016) . . . . . . . . B - 1
D.1 VHM failure contour diagram for N = 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D-1

D.2 VHM failure contour diagram for N = 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D-1
D.3 VHM failure contour diagram for N = 10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D-2
D.4 VHM failure contour diagram for N = 20 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D-2
D.5 Applied vertical load of 3068 kN/m (0.3 Vult,N =1 ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D-3
XI
List of Tables
1.1 Number of total connected wind turbines in 2014 and 2015 by different countries (Andrew et al., 2016) . . 2
3.1 Performance of the different advanced soil models for undrained cyclic soil behaviour . . . . . . . . . . . . . 17
4.2 Set of load parcels used in the strain accumulation procedure for all OCR values . . . . . . . . . . . . . . . . 24
4.3 Results of the strain accumulation procedure for the load set in table 4.2 . . . . . . . . . . . . . . . . . . . . 24
4.4 Load sets as input for the strain accumulation procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.5 Resulting equivalent number of cycles and scaling factors for load sets 2, 3, and 4. . . . . . . . . . . . . . . . 26
4.6 Input of the load parcels for the strain accumulation in terms of the overturning moment (OTM) in percent-
age of the maximum overturning moment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.7 Results from the strain accumulation procedure for OCR = 4 for the load set in table 4.6 . . . . . . . . . . . 34
4.8 Comparison of the resulting cyclic parameters of the paper and the UDCAM-S . . . . . . . . . . . . . . . . . 35
5.1 Parameters of the plate model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

5.2 Parameters of the soil model for the concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.3 Mesh dependency of the ultimate loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.4 Optimized NGI-ADP parameters for the stress-strain curves in figure 5.14 . . . . . . . . . . . . . . . . . . . . 49
5.8 Applied loads on the foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
D.1 Applied loads on the strip foundation in the 2D Plaxis calculations . . . . . . . . . . . . . . . . . . . . . . . D - 4
XIII
1.
Introduction
The demand for energy in the world is pressing, and the view on fossil fuels in society deteriorates. Renewable energy
seems the best solution to the problem of depleting fossil fuels. The demand for renewable energy is also stated in the
Kyoto protocol and the recent agreements of the Paris conference. Several options are available for the generation of
renewable energy, and wind turbines play a key role in this matter. The construction of offshore wind turbines is a solution
to the limitation of space onshore, and the strong aversion in society to the construction of onshore wind turbines. The
construction of offshore wind turbines enables the creation of large wind farms, establishing enormous possibilities for the
capacity of energy generated by wind.
The water depth in the North sea is irregular, resulting in different

conditions for the turbines. Figure 1.1 shows the relation between the
water depth and the distance to the shore of constructed wind farms
for the United Kingdom, the Netherlands, and Germany (Andrew et al.,
2016). The overall trend of new wind farm locations is the movement
further away from the shore. In figure 1.2 the consented wind farms
in the North Sea are shown, which are to be build in the coming years
(Andrew et al., 2016). The construction of wind turbines in deeper
water results in larger forces due to the water on the structures.
To increase the energy production of the wind turbines, the turbines
increase significantly in size. Ten years ago a wind turbine producing
2 MW would have been a large turbine, nowadays turbines with a ca- Figure 1.1: The distance to the shore and the water depth of
pacity from 8 to 10 MW are being installed. The consequence of larger the constructed wind farms (Andrew et al., 2016)
turbines is a challenging construction of the foundation. The loads on the structure increase significantly due to the in-
crease of the wind load by the increase of the turbine size, and the increase of the water load due to greater water depth.
This makes the design of offshore wind turbines a challenging process.
Different options for the foundation of offshore support structures are

shown in figure 1.3. The foundation with a monopile is the predomi-
nantly used foundation type for offshore wind turbines. The structure is
fixed to the bottom and can bear large forces from the structure. How-
ever, every monopile has to be hammered individually into the soil, which
results in significant installation costs as part of the total costs. Doubt ex-
ists if monopiles are capable to bear the large forces in turbines that are
expected to be built in the coming years, generating up to 20MW.
An alternative for the monopile foundation is the gravity based foun-
dation. This type of foundation ensures the stability of the structure by
the weight of the foundation. The base of the structure is placed directly
onto the seabed, without penetration of the soil. The advantage of this
method is a relative simple installation procedure compared to the instal-
Figure 1.2: Development of the wind farm locations consid-
ering the distance to the shore and the water depth (Andrew lation of monopiles. This enables the creation of large wind farms at high
et al., 2016) construction rates. The gravity based foundation offers more advantages,
in terms of production rate and applicability to different soil conditions. The environmental impact of the installation is
smaller than the impact of the installation of piled foundation, as noise due to hammering of piles has a negative impact
on the environment.
1
CHAPTER 1. INTRODUCTION
Table 1.1: Number of total connected wind turbines in 2014 and 2015 by different countries (Andrew et al., 2016)
Country Total number of connected wind farms Total number of connected turbines
2014 2015 2014 2015
Netherlands 5 6 124 184
United Kingdom 24 27 2488 3230
Germany 16 18 258 792
Gravity based foundations can be installed at locations with hard soils, for which piled foundations encounter installa-
tion problems. The drive-ability of piles in very dense sand can be limited, resulting in damage in the pile. Gravity based
foundations can be placed on top of the dense sand, which provides great stability for the structure.
The number of connected wind turbines in the North Sea increased rapidly over the past few years. The number of
connected wind turbines for the United Kingdom, the Netherlands, and Germany are shown in table 1.1. The table also
shows the increase in wind farm size, as the number of connected turbines increased significantly more than the number of
connected wind farms. The application of gravity based foundation enables a cut in the offshore construction time of these
wind farms, as the installation of the foundation is a relative quick process. The turbines are built onshore and transported
to the project site, where they are placed on the seabed by immersion.
Monopile Monopod Gravity base Tetrapod caisson Assymetric-tripod Jacket Tripod Tripod
caisson caisson caisson caisson pile
Figure 1.3: Different offshore foundation types (adapted from Bhattacharya, 2014)
However, the challenge to ensure life-long stability of the turbine remains with gravity based foundations. Offshore
winds can be strong, resulting in large forces on the structure and the foundation due to the wind loads and generated
waves. On top of that, larger turbines are accompanied with larger wind loads, and thus a bigger challenge to ensure the
stability of the structure. However, the large loads itself are not the greatest challenge in the design of offshore founda-
tions. The biggest design challenge lies within the cyclic character of the loads. In the third ISSMGE McClelland lecture
Andersen (2015) describes the problem of degradation of the soil by cause of cyclic offshore wind and wave loads.
The degradation of the soil is initiated by the accumulation pore pressures in the undrained clay underneath the
foundation due to cyclic shearing of the soil. As the volumetric deformation of the clay is restricted due to undrained
behaviour of the clay, the effective stress in the soil decreases as a result of an increase of the pore water pressure. The
reduction in effective stress leads to shear strains. This process causes a decrease in strength and stiffness of the soil, called
the cyclic degradation of the soil strength and stiffness.
With severe offshore storms, the degradation of the soil can endanger the stability of the foundation. The existence
and risks of cyclic soil degradation is known for a long time. However, the modelling of this phenomenon is still a complex
matter, dealing with many uncertainties. Many lab tests have been performed with cyclic loads on scale models of offshore
foundations, including gravity based foundation. However, the determination of the exact response of offshore foundation
to environmental loads remains difficult. The time domain of these loads are very large, and the amplitudes of the loads
are highly irregular. Therefore it is important to simplify a design storm (Andersen, 1976), before it is possible to model
the effect of a storm on the stability of an offshore foundation.
To assess the cyclic behaviour of the soil, a new complex constitutive cyclic accumulation model has been developed
by the Norwegian Geotechnical Institute (NGI) (Jostad et al., 2014). This constitutive model is validated against small
scale model test results, and shows the power of this model for the assessment of cyclic degradation. However, this model
is very complex and is therefore not suitable for ’common’ geotechnical design. A simplified version of this constitutive
2
1.1. RESEARCH QUESTIONS
model is being developed for the finite element software Plaxis, which could be more convenient for engineering practice.
The new model is a procedure, that is performed before using finite element software. This procedure is the UnDrained
Cyclic Accumulation Model (UDCAM-S), which produces a set of soil parameters that takes the effect of cyclic degradation
on the soil strength and stiffness into account. This study presents this new procedure, which is tested and validated in this
research before applying it in finite element software. Afterwards the influence of the cyclic degradation is investigated for
the design of a gravity based foundation.
1.1 Research questions
The main objective of this research is to quantify the influence of the degradation of soil strength and stiffness due
to cyclic loading on a gravity based foundation. To determine this degradation, a good understanding of the cyclic soil
behaviour is required. With this research objective, the main research question is stated as:
What is the effect of taking the degradation of soil strength and stiffness as a result of cyclic loading
into account on the stability of a gravity based foundation?
To be able to answer this question, this research has been divided into different parts. Sub questions have been
established to create a well-founded answer to the main research question. The sub questions are answered in the different
parts of this research.
Part 1: Literature study
The literature study of this research is focused on the concept of gravity based foundations and the cyclic soil behaviour
initiated by the loads on the foundation. A good understanding of these two subjects is required in the succeeding steps of
this research. The sub questions for the literature study are:
• What design codes are applicable for an offshore gravity based foundation and how do these codes deal with cyclic
soil degradation?
• What is the behaviour of soils subjected to cyclic loading and how is the degradation of soil strength and stiffness
influenced?
Part 2: Model investigation

The model that is used in this research is the UDCAM-S procedure, a new procedure to take the cyclic degradation of soil
strength and stiffness into account. A part of this research is focused on the understanding and development of this new
procedure. As this procedure is being developed for the finite element software Plaxis, it is required to determine the
actions that have to be performed for cases in which the new procedure is used. The sub questions involving the model
investigation are:
• How does the UDCAM-S procedure analyse the cyclic degradation of the strength and stiffness of the soil?
• What actions have to be performed to implement the UDCAM-S procedure in a real project?
Part 3: Modelling
The influence of cyclic soil degradation on the stability of gravity based foundations is investigated by means of modelling
in the finite element software Plaxis 2D. The result of the UDCAM-S procedure is a parameter set, which is specifically
defined for the NGI-ADP soil model. Therefore it is required to investigate this soil model with respect to the UDCAM-S
before drawing conclusions on the results of the modelling of cyclic soil degradation in Plaxis. The result of the modelling
has been visualized by VHM failure envelopes, which show the ultimate loads for a foundation in terms of the vertical load
(V), horizontal load (H), and the overturning moment (M). The sub questions for the modelling part are:
• What is the sensitivity of the input soil parameters in the NGI-ADP model in Plaxis on the soil behaviour regarding
the simplified UDCAM procedure?
• What is the effect of degradation in strength and stiffness of the soil on the stability of a gravity based foundation in
terms of VHM failure envelopes?
3
CHAPTER 1. INTRODUCTION
1.2 Readers manual
This thesis is divided into seven chapters. After the introduction (chapter 1), the chapters can be split into four parts:
the literature study, the model investigation, the modelling phase, and the outcome of this research. This is shown in
figure 1.4. Chapter 2 and chapter 3 present the result of the literature study on respectively gravity based foundations
and cyclic soil behaviour. With the theory of the literature study the new procedure is presented in chapter 4, the model
investigation. In chapter 5 the new procedure is applied to different case studies by modelling in finite element software,
the modelling phase. The results of these case studies are discussed in chapter 6. With the results and the discussion, the
conclusions are drawn and recommendations for further work are suggested in chapter 7.
Chapter 2
Gravity based
foundations
Chapter 4
Chapter 5
Chapter 1 The Simplified
Modelling the
Introduction UnDrained
influence of cyclic
Cyclic Accumu-
soil behaviour
Chapter 3 lation Model
Cyclic soil Model investigation Modelling & Results
behaviour
Literature study
Chapter 7
Conclusions Chapter 6
& Recom- Discussion
mendations
Outcome of this research
Figure 1.4: Readers manual to the different chapters of this thesis
4
2.
Gravity based foundations
This chapter presents an introduction to the design of gravity based

foundations for offshore wind turbines. The design considerations are
discussed, together with the loads acting on the structure.
2.1 Design of Gravity based foundations
The main difference between gravity based foundations and other

offshore foundation types is the fact that gravity based foundations are
not fixed to the seabed. This gives a huge advantage in the installation
of the foundation, as the foundation does not have to be placed by
hammering or suction installation.
Figure 2.1 shows an example of a wind turbine on a gravity based

foundation. The foundation itself consists only of the foundation slab,
which is placed either directly on the seabed, or on a layer of gravel.
However, a large concrete caisson is required to ensure enough self
weight of the structure to withstand the loads on the structure without
toppling over. Furthermore, this caisson provides the transition of the
loads from the tower towards the foundation.
In the design of gravity based foundations it can be decided to in-

stall skirts at the bottom of the foundation. Skirts protect water from
flowing under the foundation when contact between the soil and the
foundation is lost, to avoid erosion of the soil underneath the founda-
tion. Skirts increase the resistance to horizontal loads as well. How-
ever, with the application of skirts under the foundation, the advantage
of the installation decreases due to a more complicated installation pro-
cess. In this research only gravity based foundations without skirts are
considered. Figure 2.1: A gravity based foundation for wind-turbines
placed on the seabed
2.1.1 Standard offshore foundation design

In figure 2.2 a simplification of a monopile foundation placed in the seabed can be seen (Zaaijer, 2007). The soil is
represented by springs, which are either normal or rotational springs. The simplification of the soil behaviour by springs
is a commonly used method, as this is also used in the design codes by the API (2002) and DNV-GL (2016). The stiffness
of the soil is then described by a spring stiffness k. In the middle of figure 2.2 the resulting loads are described by the
horizontal displacements (u) and rotational displacements (θ).
The right image in figure 2.2 shows the simplification of the soil-structure interaction for a gravity based foundation.
The foundation behaviour is represented by three springs: a rotational spring, a vertical spring and a horizontal spring.
Values for these springs can be found in design codes as DNV-GL (2016). Simplifying the soil behaviour with springs
instead of treating the soil as a continuum neglects important effects of the soil behaviour. For the design of monopile
foundations it is common practice to use established load displacement curves (P-Y curves), designed by API (2002).
5
CHAPTER 2. GRAVITY BASED FOUNDATIONS
Figure 2.2: Standard simplification of the soil-structure interaction of offshore foundations (Zaaijer, 2007)
Design codes
In appendix A the design codes by the DNV-GL (2016), the API (2002), and the API (2011) are described. These codes
are very simplified approaches to describe the soil behaviour. The eccentricity of loads is dealt with by placing the load in
the load centre and using an effective area of the foundation by scaling factors. This is the effective width method, which
was introduced by Vesic (1975).
Bhattacharya (2014) describes the complexity in the design of foundations for offshore wind turbines and concludes
that the design codes for the oil and gas industry are not to be applied directly to offshore wind turbines foundations. The
stiffness of offshore wind turbines is significantly lower than the stiffness of the structures used in the oil and gas industry,
for which the codes are designed. Due to the lower stiffness, the structures are more sensitive to dynamic loading, and
thus the stability of the foundation is more influenced by the dynamic loads. This is however not captured in the design
codes.
Bhattacharya (2014) states that the design codes from API (2011) lack as well in the assessment of long term cyclic
loading on the offshore wind turbines by: "The widely used API model is calibrated against response to a small number of
cycles (maximum 200 cycles) for offshore fixed platform applications. In contrast, for a real offshore wind turbine 107 –108
cycles of loading are expected over a lifetime of 20–25 years."
2.2 Combined loading on an offshore gravity based foundation
With the design of offshore structures many different loads have to be taken into account. According to DNV-GL (2016)
the loads on an offshore structure can be divided into:
• Permanent loads
– Mass of the structure

– Mass of permanent ballast and equipment
– External and internal hydrostatic pressures
– Tower base reaction on the three loads above
• Variable functional loads
– Actuation loads
– Loads on access platforms and internal structures
– Ship impacts
– Crane operational loads
• Environmental loads
– Wind loads
– Hydrodynamic loads
– Earthquake loads
– Current-induced loads
6
2.2. COMBINED LOADING ON AN OFFSHORE GRAVITY BASED FOUNDATION
– Tidal effects
– Marine growth
– Snow and ice loads
For the design of an offshore foundation the aforementioned loads have to be considered in different load cases. How-
ever, in this research only the degradation due to the wind and hydrodynamic loading is investigated, as these loads are
the dominant cyclic loads on the structure.
The stated loads are all loads on acting on the structure. It must be noted that the loads on the foundation are dominant
for the foundation design. These loads are different because of the dynamic behaviour of the structure (Zaaijer, 2007).
The loads on an offshore foundation are never pure horizontal loading, pure vertical loading, or pure moment loading.
The resulting load on these type of foundations is always a combination of a vertical component, a horizontal component
and an overturning moment.
In figure 2.3 by Byrne (2011) it can be seen that the wind loads act at a large distance from the foundation level. In the
example of figure 2.3(a), this distance can vary between 90 and 120m. As the distance between the load and the founda-
tion level is extensive, the resulting overturning moment (shown in figure 2.3(b) with 'M') at the base of the wind-turbine
can be enormous for large wind loading.
(a) Example of wind and (b) Transmission of the loads to the foundation for (c) Frequencyies of the structure due to external loads and the
wave loads on a wind- a monopile foundation (left) and a tripod founda- Eigen frequencies (1P and 3P) that must be avoided
turbine tion (right)
Figure 2.3: Loads on the super-structure and the transmission of the loads to the foundation (Byrne, 2011)
The blades of a wind-turbine rotate with a certain speed. The main excitation frequencies (i.e. the natural frequencies
of the structure) are called 1P and 3P, as shown in figure 2.3(c). The frequency of one period (1P) represents the rotation
frequency of the blades, the frequency of one third of the period (3P) the blade passing frequency. This can be seen in the
graph in figure 2.3(c), in which the horizontal axis represents the frequency, and the vertical axis represent the Dynamic
Amplification Factor (DAF). The Dynamic Amplification Factor represents the ratio between the dynamic response of the
structure and the static deformation of the structure. The loading frequencies coinciding with the natural frequencies of
the turbine must be avoided, as this will lead to resonance. Resonance in the structure leads to large deformations, which
will endanger the stability of the wind turbine.
2.2.1 Cyclic loads

Cyclic loading on the structure results in cyclic behaviour in the soil. An important aspect is the behaviour of soils in
cyclic loading, which is significantly different than in static loading. This is described in chapter 3. In this research, cyclic
loading is related to offshore foundations, but this behaviour can also occur onshore, for example cyclic loads induced by
earthquakes.
The loads on an offshore structure can be described by (Norén-Cosgriff et al., 2015)
F (t) = Fa (t) + Fcy (t), (2.1)

in which Fa is the average component of the loads, and Fcy the cyclic component. The loads on an offshore structure
are always three dimensional, and therefore the load F is a load vector described by
7
h i
3D F = Fx , F y , F z , M x , M y , M z
h i (2.2)
2D F = Fx , F y , M z
However, in many calculations the problem is defined in 2D, and only Fx , Fy , and Mz are used. Examples of the
development of the overturning moment and the horizontal load for a period of 600 seconds on the foundation of a gravity
based foundation are shown in respectively figure 2.4 and figure 2.5. It can seen that many different amplitudes of loads
are present in a storm, and the periods of the two loads differ significantly. The data used for these figures is adjusted
from the actual data for 8 MW wind turbines, and only intended to illustrate the development of the cyclic loads. To be
able to model the results of an offshore storm, it is required to simplify a storm and transform the actual storm data to a
simplified load composition, as described by Norén-Cosgriff et al. (2015). Modelling every cycle individually would lead
to enormous calculation times.
×10 5
4
Overturning moment [kNm]
-2
-4
-6
-8
0 100 200 300 400 500 600
Time [s]
Figure 2.4: Example of the development of the overturning moments for a time domain of 600 seconds
×10 4
3
2
Horizontal Load [kN]
-1
-2
-3
0 100 200 300 400 500 600
Time [s]
Figure 2.5: Example of the development of the horizontal loads for a time domain of 600 seconds
Sterigiou et al. (2015) and Gourvenec (2007) state that the overturning moment and the horizontal loads are the
dominant loading types for wind-turbines. Especially the overturning moment can cause difficulties in the design of the
foundation. This is due to the fact that the wind loads act with a long distance from the foundation surface, resulting
in large overturning moments with a cyclic load character. The vertical load for wind-turbines however can be assumed
constant compared to the horizontal load and overturning moment, as the vertical load is predominantly a result of the
self-weight of the structure.
2.2.2 VHM capacity

The loads on an offshore gravity based foundation are, as mentioned, a combination of vertical loading (V), horizontal
loading (H), and moment loading (M). The bearing capacity of an offshore gravity based foundation subjected to combined
loading is often described with VHM failure envelopes. These failure envelopes show directly the influence of the three
different loading types on the stability of the foundation. The approach of describing the stability of a foundation with
VHM failure envelopes fall within the yield surface approach (API, 2011).
The ultimate loads for pure vertical loading (Mult ), pure moment loading (Mult ), and pure horizontal loading (Hult )
are required for the determination of VHM failure envelopes. The contours of the failure mechanisms for pure vertical
8
2.2. COMBINED LOADING ON AN OFFSHORE GRAVITY BASED FOUNDATION
loading, pure horizontal loading, and pure overturning moment for a gravity based foundation with suction between the
soil and the base are shown in figure 2.6.
(a) Pure Vertical Loading (b) Pure Horizontal Loading (c) Pure Moment Loading
Figure 2.6: Displacement contours plots for pure loading of a gravity based foundation without embedment allowing tension at the soil-structure interface
However, the failure mechanism for combined loading depends on the individual influence of the three loading types.
As the vertical load on the foundation for wind-turbines can be assumed quasi-static (Gourvenec, 2007), the 3D failure
diagrams can be represented in 2D H-M diagrams for predefined values of the vertical load (V). This allows for a simplifi-
cation of the 3D failure envelope to a clear 2D failure envelope.
For a gravity based foundation it is possible that suction is developed between the base of the foundation and the
undrained clay by loading. The vertical bearing capacity of a strip foundation on an undrained clay subjected to inclined
loading, without base suction, is defined by Vesic (1975) as

Q 2e 1
α ≤ α0 = Nc0 1 −
BSu B cos(α) + 2 sin(α)
(2.3)
Q 2e 1
α > α0 = 1−
BSu B cos(α)
in which α0 is the angle at which the undrained shear strength is mobilized for all points along the footing interface,
and the theoretical bearing capacity factor (Nc0 ) is defined as 2 + π. The theoretical bearing capacity of a strip foundation
Q
in pure vertical loading is thus 2 + π ( BSu
≈ 5.14).
The sign convention for the VHM loads is shown in figure 2.7 as proposed by Butterfield et al. (1997). This sign con-
vention is applicable for the displayed VHM failure envelopes.
RP M (+) H (+)
V (+)
Figure 2.7: Sign convention for loads in the VHM failure diagrams
Ukritchon et al. (1998) investigated the effect of allowing tension between the foundation and the soil at the soil-
structure interface. The result of this analysis is shown in figure 2.8, in which the failure contours are shown for a gravity
based foundation with suction between the foundation and the soil (upper figure), and a gravity based foundation without
suction between the foundation and the soil (lower figure). The failure loads are expressed in normalized loads to the
undrained shear strength of the soil (Su ). The main difference between these two figures is the skewed failure contour
at smaller vertical loads. If base suction is taken into account, the foundation can bare significant larger moments if the
horizontal load is applied in the same direction as the moment load (i.e. both positive values). It must be noted that in
reality both the horizontal load and the overturning moment act in almost all cases in the same direction, as both are a
result of the wind and wave loads on the structure.
9
Figure 2.8: Failure locus found for a circular footing subjected to moment and horizontal loading (with V = 0) by Ukritchon et al. (1998)
The explanation for the skewed form of the HM failure envelope is given by Taiebat and Carter (2000). The authors
show the difference in the soil movement between the case in which the horizontal load acts in the same direction as the
overturning moment and the case in which the horizontal load acts in the opposite direction of the overturning moment.
The soil movement for the two load situations is shown in figure 2.9. The rotation point of the movement of the foundation
is below the interface of the soil and the foundation if the horizontal load and the overturning moment act in the same
direction (figure 2.9(a)). If the horizontal load acts in the opposite direction of the overturning moment, the rotation
point is above the interface of the soil and the foundation (figure 2.9(b)). The difference in the soil movement results in
a difference in the generated plastic zone. For the case in which the two loads act in the same direction, the plastic zone
extends through deeper soil than in the case in which the two loads act in the opposite direction. It is thus the difference
in failure mechanism that results in a skewed HM failure envelope.
(a) Horizontal force and overturning moment acting in the same direc- (b) Horizontal force and overturning moment acting in opposite direc-
tion tions
Figure 2.9: Difference in soil movements between different loading situations regarding the direction of the horizontal load and the overturning moment
(Taiebat and Carter, 2000)
10
3.
Undrained cyclic soil behaviour
This chapter presents the theory of undrained behaviour of clays and the behaviour of clays due to cyclic loading. In
this chapter the introduction of the theory of cyclic degradation. Both of the subjects are written regarding the undrained
cyclic behaviour of clay. This chapter forms the background for the chapter of the cyclic accumulation model in chapter 4.
3.1 Undrained behaviour
As the seabed is below the water level, the pores of the soil are filled with water. If the loads on the soil act at a fast
rate, consolidation of the soil cannot be initiated, and the behaviour of the soil is called undrained (Verruijt, 2012). By
Andersen (2015) it is stated that the behaviour of clays in offshore cyclic loading can be assumed undrained, as the excess
pore pressures cannot dissipate due to the quick succession of the load cycles. The built up of excess pore pressure (∆p) is
described by Skempton (1948) as
∆p = B [∆σ3 + A (∆σ1 − ∆σ3 )] , (3.1)

in which A and B are the Skempton parameters, which are to be determined in undrained triaxial tests, and ∆σ1 and
∆σ3 the principal total stresses.
An important aspect regarding the pore fluids filled with water, is shearing of the soil. Water can only sustain normal
stresses, but cannot sustain shear stresses (Brinkgreve, 2015b). However, pore pressures may increase as a result of
shearing: the shear induced pore pressures.
3.1.1 Undrained shear strength

According to Verruijt (2012) the undrained shear strength in a triaxial test is determined as half of the difference of the
two main principal stresses at failure. In a triaxial test the principal stresses act at an angle of 0% and 90% with respect to
the horizontal on the soil specimen. In triaxial compression the main principal stress (σ1 ) is equal to the cell pressure in
combination with the resulting pressure from the compression of the sample, and the minor principle stress (σ3 ) is equal
to the cell pressure. The undrained shear strength at failure (Su ) can then be described as
1
Su = (σ1 − σ3 ) , (3.2)
2
The undrained shear strength is defined by Ladd and Foott (1974) as the relation between the overconsolidation ratio
0
(OCR), the initial effective stress level (σyy,0 ), the undrained shear strength ratio at a normally consolidated state (S), and
a material depending exponent (m) by
Su
0
= S · OCRm . (3.3)
σyy,0
11
CHAPTER 3. UNDRAINED CYCLIC SOIL BEHAVIOUR
Undrained shear strength from cone penetration tests by Lunne et al. (1997)
Lunne et al. (1997) suggest an empirical correlation between the cone resistance of the CPT and the undrained shear
strength of the soil. The authors propose three possible relations: an estimation using the total cone resistance, an
estimation using the effective cone resistance, and an estimation using the excess pore pressure. The relation for undrained
shear strength at failure and the total cone resistance is described by
(qc − σv0 )
Su = , (3.4)
Nk
in which Su is the undrained shear strength in kPa, qc the measured cone resistance from the CPT in kPa, σv0 the total
vertical stress in kPa, and Nk an empirical cone factor. This relation is a common used method in the assessment of the
undrained shear strength of the soil by means of CPT results.
3.1.2 Pore pressure built-up in cyclic loading

As mentioned, the behaviour of clays can be assumed undrained during a design storm. This means that excess pore
water pressures cannot dissipate during cyclic loading, and the excess pore pressure will accumulate with every cycle. An-
dresen et al. (2007) state that the cyclic shear stresses generate the pore pressure build-up due to a compressive behaviour
of the clay in shearing. This is comparable to the behaviour of sands subjected to wave loads, as described by Spierenburg
(1987).
The process of excess pore pressure build-up in cyclic loading is shown in figure 3.1. Figure 3.1(a) shows the devel-
opment of the pore pressures over time. The shear stress, the pore pressure and the shear strains each consist of two
components: an average component and a cyclic component. It is shown that without an increase of the average shear
stress (upper figure), the pore pressures (middle figure) and the shear strains (lower figure) increase.
In figure 3.1(b) this process is shown in the σ 0 − τ plane. The red line represents the stress conditions in a cyclic
undrained triaxial test, and the blue line the stress conditions in a monotonic undrained triaxial test. In a monotonic
triaxial test the soil sample is compressed until the sample collapses, indicated by the black inclined failure envelope. In
a cyclic triaxial test the stress is varying with the cyclic shear stress (τcy ) with respect to the average shear stress (τa ).
Due to the increase in pore pressure in an undrained cyclic triaxial test without an increase of the maximum shear stress,
the effective stress decreases. In figure 3.1(b) this is phenomenon is shown, as the stress paths move to the left. This is
continued until the stress path reaches failure at cycle N. The same soil can thus fail at a lower shear stress in cyclic loading
than in monotonic loading.
τ
τ
τcy
τa
time
u τcy
ucy
up
time up τa σ,
γ
γcy
γp Nth cycle 1st cycle
time
(a) Pore pressure built up and shear strain accumu- (b) Decrease in effective stress due to cyclic loading (red line) and monotonic loading
lation during cyclic loading (blue line)
Figure 3.1: Generation of pore pressures and strains in a cyclic triaxial test (Recreated from Andersen, 2009)
12
3.2. CYCLIC SOIL STRENGTH
3.2 Cyclic soil strength
The soil strength in static undrained conditions can be described by the undrained shear strength of the soil, which
is dependent on the stress level and the material. However, by cyclic loading of the soil, the soil strength is not only
determined by the undrained shear strength of the soil, but depends on the value of the combination of the average and
cyclic stresses in the soil, the loading history of the soil, and the stress path (Randolph and Gourvenec, 2011, p. 140). The
undrained cyclic shear strength is described by Andersen and Lauritzsen (1988) as
τf,cy = (τa + τcy )f , (3.5)

in which τa is the average shear stress at failure, τcy the cyclic shear stress at failure, and τf,cy the cyclic undrained
shear strength. The cyclic shear strength of the soil is thus not a parameter only depending on the stress situation and the
soil type, but a resulting strength from the conditions influencing the soil behaviour. It is therefore written as τf,cy instead
of the static undrained shear strength Su .
3.2.1 Stress path

Stress anisotropy is an important aspect regarding the stability of a shallow foundation. This means that the stresses
in the soil are not homogeneous, and thus have to be accounted for. As shown in figure 3.2, the soil behaviour in the soil
along a potential failure surface, indicated by the dashed line, can be described by three laboratory tests:
• Direct Simple Shear test (DSS)

- Symmetric loading (τa = 0)
- Asymmetric loading (τa 6= 0)
• Triaxial Compression test (TXC)
• Triaxial Extension test (TXE)
Simplified versions of the shear stress development over time for the three laboratory tests are shown in figure 3.2.
The type of soil behaviour depends on the angle at which the shear stresses act. The shear stresses in direct simple shear
act at 0◦ with respect to the horizontal, in triaxial compression and extension at ± 45◦ with respect to the horizontal.
The undrained shear strength of the soil is thus dependent on the angle of the shear stresses in the soil. Grimstad et al.
(2011) propose a model, that can model this shear strength anisotropy in clay: the NGI-ADP model. This model is an
improvement on previous modelling of the shear strength anisotropy of clays. The anisotropic formulation from Davis and
Christian (1971) is bound to plane strain conditions. The improvement of the NGI-ADP model, with respect to the model
from Davis and Christian (1971), is the modelling of anisotropy in the stiffness of the soil. The model formulation from
Prévost (1978) does not include hardening of the soil in shearing, the proposed model from Grimstad et al. (2011) does
include hardening in both the strength and stiffness of the clay.
Figure 3.2: Classification of the stress states underneath a gravity based foundation in three laboratory tests (Andersen, 2015)
13
Although the stress states described by three laboratory tests the soil behaviour is simplified, Andersen and Lauritzsen
(1988) conclude that by extensive laboratory testing with different combinations of average and cyclic stress ratios, the
undrained cyclic shear strength can be accurately described.
Jostad et al. (2015) state that the average and cyclic stresses in the soil can be assumed equal to the ratio of the average
and cyclic loads from the structure, if the average loads cyclic loads act in the same direction. For an offshore gravity based
foundation this is an acceptable assumption, as the average and cyclic loads are both induced by the wind and wave loads.
3.2.2 Combination of average and cyclic shear stresses in the soil

From many lab tests it has been found that the value of the maximum shear stress is not the driving factor for cyclic
soil behaviour, but the ratio of the cyclic shear stress with respect to the average shear stress in the soil (Andersen and
Lauritzsen, 1988; Andersen et al., 1988; Andersen, 2015). It has been shown that the cyclic strength of the soil decreases
if the ratio of the cyclic stress with respect to the average stress increases. The soil thus collapses at a lower number of
cycles, if a higher cyclic stress ratio is applied in cyclic laboratory tests.
The cyclic behaviour of clays is described by many researchers using contour diagrams. An example of a contour
diagram is shown in figure 3.3, in which the ratio of the cyclic shear stress with respect to the static undrained shear
strength in compression (τcy /SuC ) is plotted against the ratio of the average shear stress with respect to the static undrained
shear strength in compression (τa /SuC ). A contour diagram is representative for one type of clay for a certain number of
cycles.
In the contour diagram the contours are shown for different values of the average shear strain (γa ), drawn in figure 3.3
with the semi-vertical continuous blue lines, and the cyclic shear strain (γcy ), drawn in figure 3.3 with the semi-horizontal
red lines. In offshore geotechnical engineering, failure for undrained clays is defined as 15% shear stress, either the average
strain or the cyclic strain. Therefore the contour diagram is bound by three contours: ± 15% average shear strain and 15%
cyclic shear strain.
0.8
0.6
τ cy / SA
u
0.4
0.2
0
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
τ a / SA
u
Figure 3.3: Contour diagram for a triaxial test for 10 cycles for overconsolidated Norwegian Drammen Clay
These contour diagrams can be established for every soil type after every number of cycles by performing many lab-
oratory tests at different stress conditions. Each diagram represents the behaviour of the soil after a certain number of
cycles for all possible stress situations. In figure 3.3 it can be seen that the developed shear strains are dependent on the
combination of shear stress in the soil, as the contour lines for both the cyclic shear strains as the average shear strains
are curved lines. It is thus important to know the composition of the stresses in the soil to assess stress strain behaviour in
undrained cyclic loading.
3.2.3 Loading history of the soil

The loading history of the soil is an important factor for the cyclic undrained shear strength of the soil. As mentioned
in section 3.1.2, the pore pressure and the shear strains increase with every cycle. It is thus important to know the number
of cycles and the combinations of shear stresses of these cycles in a design storm to asses the behaviour of the soil.
Norén-Cosgriff et al. (2015) investigated the influence of the procedure of counting the minima and maxima of the
cyclic loads. They show in their paper that the way in which the cycles are treated can have a significant effect on the cyclic
degradation of the soil. They have presented a new counting method, the NGI counting method, which is an alternative
for the commonly used Rain-flow counting method (Downing and Socie, 1982). However, there is no evidence that one
counting method is better than the other.
14
3.3. EQUIVALENT OVERCONSOLIDATION HYPOTHESIS
3.2.4 Strain compatibility

Andersen and Lauritzsen (1988) state that it is not possible that zones along a potential failure surface develop dif-
ferent shear strains. This is based on the theory that zones with large developed cyclic shear strains and small developed
average shear strains cannot be on the same failure surface as zones with large developed average shear strains and small
developed cyclic shear strains. This is the theory of the strain compatibility. Due to redistribution of average shear strains
along a potential failure surface the assumption of strain compatibility would always be satisfied. Andersen and Lauritzsen
developed a method to determine the factor of safety for a gravity based foundation subjected to cyclic loads. With the
back-calculation of a model test they have validated this method. However, this is an iterative and extensive process to
determine the factor of safety, and therefore not suitable for common practice.
3.3 Equivalent overconsolidation hypothesis
Another approach of dealing with cyclic degradation of soils is the equivalent overconsolidation hypothesis (Matsui
et al., 1991, 1999; Yasuhara, 1994; Cai, 2008). This theory is based on the hypothesis that due to the pore pressure
build-up in the soil, the effective stress of the soil decreases, which results in an overconsolidated state of the soil. This
means that the effective stress of the soil is lower than the maximum effective stress in the stress history of the soil. Using
empirical relations between the overconsolidation ratio of the soil, the undrained shear strength after a certain amount of
load cycles of the soil can be predicted. The problem with the proposed empirical relations, as the authors conclude, is
the fact that these relations are formulated for a specific type of clay, and therefore not to be automatically used to predict
the behaviour of other types of clay. Laboratory tests first have to be performed to establish the same type of relations for
another clay type or to test the applicability of existing relations.
15
3.4 Current available options for the modelling of the undrained cyclic soil behaviour
3.4.1 Undrained behaviour in Plaxis

For an undrained analysis, multiple options are available within the geotechnical finite element software Plaxis 1 , which
is used in this research. The options to model undrained behaviour in Plaxis are undrained type A, B, or C. Depending on
the drainage type, the soil is modelled with a total stress analysis or an effective stress analysis (Brinkgreve et al., 2016).
• Undrained (A)
Undrained type A models the undrained behaviour with effective strength and stiffness parameters. A large bulk
stiffness is defined for the pore water to model the soil as a whole undrained. The undrained shear strength is then
an outcome of the program. The undrained A option gives an effective stress analysis.
• Undrained (B)
Undrained type B models the undrained behaviour with undrained strength parameters and drained stiffness param-
eters. Analogous to the undrained type A, a large bulk stiffness is defined for the pore water.
• Undrained (C)
Undrained type C models the undrained behaviour with undrained strength and stiffness parameters. This results in a
total stress analysis, which means that there is no distinction between the generated pore pressures and the effective
stress. In Plaxis this type is restricted to the Linear elastic model, the Mohr Coulomb model and the NGI-ADP soil
model.
These three options are available for modelling the undrained behaviour of soils. However, care has to be taken when
modelling undrained behaviour. As the Plaxis manual mentions:
"The modelling of undrained behaviour is even more complicated than the modelling of drained behaviour.
Therefore, the user is advised to take the outmost care with the modelling of undrained soil behaviour"
(Brinkgreve et al., 2016).
3.4.2 Available advanced soil models for undrained cyclic soil behaviour
Multiple advanced soil models are available to model the undrained behaviour of the soil. In some extend these models
can model the cyclic behaviour as well. In this section a short overview is given for advanced soil models that could be an
option to model the desired behaviour of the soil.
Hardening soil model

A commonly used advanced soil model in geotechnical design is the Hardening Soil Model (HS model). The HS model
models the soil with a stress-dependent stiffness. An extension of this model is the Hardening Soil model with small
strain stiffness, the HSsmall model. This model captures the significant stiffer behaviour at very small strains. This model
is described extensively in the PhD thesis of Benz (2007). The hardening soil model is not developed to model cyclic
behaviour, but it is a very good option for the modelling of the undrained behaviour.
UBCSand model
For earthquake Engineering the UBCSand model is a common used soil model. This model can model the pore pressure
build-up in sands and model liquefaction of the soil (Brinkgreve, 2015a). This model is very applicable for earthquake
engineering, as in those situations the number of cycles are very limited with respect to the number of cycles that have to
be taken into account in offshore geotechnical engineering.
NGI-ADP model
A model to describe the anisotropy in the behaviour of undrained materials is the NGI-ADP model (Grimstad et al.,
2011). It describes the anisotropic behaviour of the soil by the stress-strain relations in three laboratory tests: Triaxial
compression, Triaxial Extension, and Direct Simple Shear. This model is ideal to model the anisotropy in the undrained
stress-strain behaviour, but does not incorporate cyclic behaviour of the soil.
Hypoplastic model
One of the more complicated advanced soil models is the hypoplastic model. The hypoplastic model is a constitutive
model based on the hypoplasticity theory, relating the stress rate to the strain rate (Von Wolffersdorff, 1996).
Lupea (2014) did research for her master thesis to model the pore pressure build up in sand due to cyclic loading for
a suction caisson foundation with the Hypoplastic model. She concluded that the modelling of many cycles can result in
extensive calculations. However, with modelling only a limited number of cycles the uncertainty increases significantly
with the extrapolation of the found results for many cycles.
1
www.plaxis.com
16
3.4. CURRENT AVAILABLE OPTIONS FOR THE MODELLING OF THE UNDRAINED CYCLIC SOIL BEHAVIOUR
UnDrained Cyclic Accumulation Model

A recently developed model is the UnDrained Cyclic Accumulation Model (UDCAM). This model is presented by Jostad
et al. (2014) and verified by back-calculations of model tests performed on a gravity based structure. The UDCAM is
capable of modelling the cyclic degradation of the soil, without modelling each cycle individually. This leads to significant
less calculation time with respect to models that simulate every cycle individually. The results of this model are very
promising. However, as input for the input data from many cyclic laboratory tests are required. NGI has performed many
laboratory tests on the Norwegian Drammen Clay, but for another type of clay the same amount of data is not always
available.
Simplified UnDrained Cyclic Accumulation Model

A new model that has the potential to be used in common practice in offshore geotechnical design is the Simplified
UnDrained Cyclic Accumulation Model. This model is a simplified version of the UDCAM from NGI, and is at the time of
this research being developed by Plaxis with the cooperation of NGI.
The difference with the complex UDCAM is that with the simplified UDCAM soil parameters are derived for the NGI-
ADP model accounting for cyclic degradation, and the calculation in Plaxis is done using a static load. This results in
relatively simple calculations with respect to the complete UDCAM, but could be reliable within certain boundaries.
3.4.3 Overview of the soil models
Table 3.1: Performance of the different advanced soil models for undrained cyclic soil behaviour
HS & HSs UBCSand NGI-ADP Hypoplastic model UDCAM UDCAM-S

Undrained behaviour ++ ++ ++ + + +
Cyclic behaviour ± + - ± ++ +
Cyclic degradation - ± - - ++ +
17
18
4.
The Simplified UnDrained Cyclic Accumulation Model
4.1 Introduction
The Simplified UnDrained Cyclic Accumulation Model (UDCAM-S) is a simplified constitutive model to deal with
undrained soil behaviour in cyclic loading. The UDCAM-S procedure is developed to make the more complex constitu-
tive model UDCAM (Andresen and Jostad, 2009) from NGI available for ’common’ engineering practice. UDCAM-s is a
kind of pre-processing procedure. The result of this procedure is a parameter set, based on the NGI-ADP constitutive model.
This model has been described in Appendix B. This chapter presents the application of the theory of cyclic degradation in
the UDCAM-S procedure.
In the NGI-ADP soil model the soil strength is characterised by three shear strengths to model the stress anisotropy
in the soil: The undrained shear strength in triaxial compression, the undrained shear strength in triaxial extension, and
the undrained shear strength in direct simple shear. The UDCAM-S procedure to obtain these NGI-ADP parameter set
accounting for cyclic degradation can be divided into three steps. The steps that have to be performed to obtain the
parameter set are:
1 Determination of the equivalent number of cycles given a set of load parcels and a strain contour diagram,
2 Calculation of the cyclic degraded stress-strain response using a specified idealized stress path in strain diagrams for
DSS, TXE, and TXC interpolated for the equivalent number of cycles,
3 Determination of the optimisation ranges for the degraded parameters using the stress-strain response from step 2
and performing the optimisation procedure to obtain the degraded parameters for the NGI-ADP model.
The three steps of the procedure result in a parameter set that takes into account the effect of cyclic loading of the
foundation. The degraded cyclic parameters for the NGI-ADP model that are determined with the UDCAM-s procedure are:
• G/SuA [-] The ratio of the degraded shear modulus (in kPa) to the degraded undrained shear strength
(in kPa)
• G/SuA [-] The ratio of the degraded shear modulus (in kPa) to the degraded undrained shear strength
(in kPa)
• SuA [kPa] The degraded undrained shear strength in active failure
• γfC [%] The degraded shear strain at failure in the triaxial compression test,
• γfE [%] The degraded shear strain at failure in the triaxial extension test,
• γfDSS [%] The degraded shear strain at failure in the direct simple shear test,
• SuE /SuA [-] The ratio of the degraded shear stress at failure in the triaxial extension to the degraded
undrained shear strength in active failure.
• SuDSS /SuA [-] The ratio of the degraded shear stress at failure in the direct simple shear test to the de-
graded undrained shear strength in active failure.
• τ0 /SuA [-] The initial mobilized shear stress to the degraded undrained shear strength in active failure.
• K0 [-] The neutral earth pressure coefficient
19
CHAPTER 4. THE SIMPLIFIED UNDRAINED CYCLIC ACCUMULATION MODEL
In the UDCAM-S procedure the undrained shear strength in active failure (SuA ), is the same value as the undrained
shear strength in compression (SuC ) (Jostad et al., 2015). This is different with respect to the NGI-ADP soil model, in which
the undrained shear strength in compression is set to 0.99 times the undrained shear strength in active failure as default.
By taking into account the degradation of the soil parameters used in the design of an offshore foundation, the calcula-
tion can be performed using a static approach. This results in relative simple and fast calculations, saving in time and costs.
Within the UDCAM-S procedure three types of analyses are available. The type of analysis is chosen in the second step
of the procedure determining the degraded stress-strain response of the soil. The three available analyses are:
• Cyclic Analysis
• Average Analysis
• Total Analysis
The cyclic analysis is used to obtain soil parameters for dynamic structural analyses. These are required for serviceability
limit state calculations. The average analysis is not an individual analysis, but it is used to determine the response in the
total analysis. The total analysis is the cumulative result of the average and cyclic analysis, and used in quasi-static
structural analyses (Jostad et al., 2015). With the total analysis the ultimate limit state calculations can be performed.
4.2 Transformation of stresses and strains for the input of Plaxis
The stresses and strains calculated by the UDCAM-S procedure are not directly applicable for the Plaxis convention.
These stresses and strains have to be transformed to stresses and strains, that can be handled by the Plaxis software.
Although this is done by UDCAM-S procedure in the background, it is worth noticing the difference in values between the
input parameters and those handled by Plaxis. The shear stresses in the three laboratory tests are calculated using the
0
active undrained shear strength at a depth where the vertical effective stress is equal to 300 kPa (σyy = 300 kPa). This
shear strength is calculated with the static input parameters of the NGI-ADP model defined by the user as
Suc (d) = c
Su,ref c
+ (d − yref ) · Su,inc , (4.1)
in which
0
d = σyy /γ 0 = 300/ (γsat − γw ) ,
c
Su,ref = The reference undrained active shear strength at reference depth yref ,
c
Su,inc = The incremental undrained active shear strength per unit of depth.
0
The effective horizontal stress (σxx ) in the laboratory tests are calculated by
0 0
σxx = K0 · σyy , (4.2)
in which K0 is the neutral earth pressure coefficient. The value of K0 is soil specific and therefore depends on the OCR
value of the Drammen clay. The derivations of the correct values for the stresses in the laboratory tests are presented in
section 4.5.1.
4.2.1 Triaxial Compression

In triaxial compression the shear strains from the output of the procedure have to be transformed to strains in the
y-plane (yy ). The average shear strain from the output is always positive, and therefore has to be multiplied by ’-1’,
as compression in Plaxis is defined as negative strain. For the same reason the cyclic shear strains have to be converted
to negative vertical strains. The vertical strain (yy ) in triaxial compression is not equal to the shear strain (γcy ). The
ratio between the vertical strain and the shear strain in triaxial compression is 2/3. The strains in the y-plane for triaxial
compression are calculated by
yy,av = γav/100 · 2/3,

(4.3)
yy,cy = γcy/100 · 2/3.
In triaxial compression the shear stresses from the output of the procedure are positive ratios of the shear stress to
the active undrained shear strength. In Plaxis, compression is defined as a negative pressure, and therefore has to be
20
4.2. TRANSFORMATION OF STRESSES AND STRAINS FOR THE INPUT OF PLAXIS
multiplied by ’-1’. The shear stress ratios have to be multiplied by the active undrained shear strength at a depth where
0
σyy = 300 kPa.

(σyy − σxx ) τav
= c
· Suc (d) · (−1)
2 S u
av (4.4)
(σyy − σxx ) τcy c
= · Su (d) · (−1)
2 cy Suc
4.2.2 Triaxial Extension

In triaxial extension the shear strains from the output of the procedure have to be transformed to strains in the y-plane
(yy ). The average shear strain from the output is always negative in triaxial extension, and therefore has to be multiplied
by ’-1’, as extension in Plaxis is defined as positive strain. The cyclic shear strain from the output of the procedure is always
positive, and thus does not have to be transformed. Analogous to triaxial compression, the ratio between the vertical strain
and the shear strain in triaxial extension is 2/3. The strains in the y-plane for triaxial extension in Plaxis are calculated by
yy,av = γav/100 · 2/3,

(4.5)
yy,cy = γcy/100 · 2/3.
The triaxial extension shear stresses from the output of the procedure are ratios of the shear stresses to the active
undrained shear strength. These values thus have to be multiplied by the reference active shear strength at a depth where
0
σyy = 300 kPa. The cyclic shear stress in triaxial extension is always positive. The sign of the shear stress thus does not
have to be changed. The average shear stress is always negative from the output. As in triaxial extension this is tension,
and in Plaxis tension is denoted as positive, this value has to be multiplied by ’-1’. The shear stresses for triaxial extension
in the x-y plane are calculated by

(σyy − σxx ) τav
= c
· Suc (d),
2 S u
av (4.6)
(σyy − σxx ) τcy c
= · S u (d).
2 cy Suc
4.2.3 Direct Simple Shear

In direct simple shear the strains from the output of the procedure are correct for the sign convention in Plaxis.
However, the strains have to be converted to dimensionless strains instead of percentages by
γxy,av = γav/100,
(4.7)
γxy,cy = γcy/100.
The shear stresses in direct simple shear from the output of the procedure are ratios between the cyclic and aver-
age shear stress and the active undrained shear strength. Therefore these values have to be multiplied with the correct
0
undrained shear strength at the depth where σyy = 300 kPa by
τav
τav = · Suc (d),
Suc
τcy (4.8)
τcy = · Suc (d).
Suc
21
4.3 Step 1 - Determination of the equivalent number of cycles of a design storm
As mentioned in chapter 2, the result of a design storm on the soil behaviour is made by determining an equivalent
number of cycles for a given maximum amplitude, which has the same effect on the soil behaviour regarding the devel-
oped cyclic shear strains in the design storm. To be able to determine the equivalent number of cycles, the storm has to
be simplified. This simplification is made by arranging the cyclic loads into load parcels. The load parcels indicate the
occurrence of a ratio of the cyclic load to the maximum cyclic load. The assumption is that the cyclic shear stress in the
soil is equal to the ratio of the cyclic load on the foundation (Jostad et al., 2015). Figure 4.1 shows the resulting load
parcels from determined overturning moments in the structure. These load parcels are always applied in ascending order
for defined load ratio’s. The table in figure 4.1 is thus applied from the lowest cyclic load ratio (Fcy /Fcy,max = 0.15) to
the highest cyclic load ratio (Fcy /Fcy,max = 1.00).
Figure 4.1: Example of the resulting load parcels from the structural loads in a design storm
With the load parcels, an equivalent number of cycles can be determined using the cyclic shear strain accumulation
procedure, as described by Andersen (1976). This equivalent number of cycles is an equivalent number of load cycles with
a constant cyclic shear stress, that gives the same effect on the soil behaviour as the design storm in terms of the load
parcels (Andersen and Lauritzsen, 1988).
3.0
τcy [kPa]
τcy [kPa]
%
1.5
%
0.75
%
B
D E γc
τcy y =
0.5% C 15
.0
A
τcy %
A B
∆γcy,i
γcy [−] Number of cycles [-]
Figure 4.2: Principe of the strain accumulation, including the incremental shear strain with an increment of the cyclic shear stress (Recreated from Andersen,
2015)
The principle of the strain accumulation procedure is shown in figure 4.2 . The procedure starts with the first load
A
parcel at point A, with a cyclic shear stress of τcy . From this point, with a horizontal line, the number of cycles in the first
load parcel is added to point A, resulting in point B. This point shows the generated strains within the first load parcel.
From point B the contour of the cyclic shear strain is followed, until the cyclic shear stress of the second load parcel is
B
reached (τcy ), which is at point C. The assumption is that the strain generated by the last load parcel, is the same strain at
the start of the next parcel. However, the next load parcel determines an increase of shear stress. This effect is shown in
22
4.3. STEP 1 - DETERMINATION OF THE EQUIVALENT NUMBER OF CYCLES OF A DESIGN STORM
figure 4.3. Due to an increase of the cyclic shear stress, the cyclic shear strain increases, regardless of the number of cycles.
This incremental shear strain (γcy,i ) has to be accounted for in the shear strain accumulation procedure, which is shown
B
in figure 4.2 as the step from point C to point D. From point D the number of cycles at the cyclic shear stress ratio τcy are
added, resulting in point E. From there onwards the rest of the load parcels are added in the same way up to τcy /τcy,max
= 1.0.
N=1
τ [kPa]
∆γcy,i B
Cycle N when τcy = τcy
A
Cycle N when τcy = τcy
B
τcy
A
τcy Cycle N + ∆N
γ [-]
∆γcy,∆N
∆γcy,i
∆γcy,N +∆N
Figure 4.3: Increment of the cyclic shear strain γcy,i due to an increment of the cyclic shear stress (Recreated from Andersen, 2015)
The aforementioned procedure is repeated for different scaling factors. The scaling factor (χ) defines the linear rela-
tionship between the ratio of the applied forces, and the ratio of the acting shear stress to the maximum shear stress in
direct simple shear by

τ F
= χ . (4.9)
SuDSS max Fmax
The strain accumulation procedure is first performed for a small scaling factor. If the end point of the strain accumu-
lation procedure does not reach 15% cyclic strain, the procedure is repeated with a larger scaling factor. This process is
performed until failure is reached. The last applied scaling factor determines the factor that has to be multiplied by the
maximum load ratio, to reach cyclic failure with the resulting equivalent number of cycles. As the maximum load ratio is
always equal to one, the resulting scaling factor defines the stress ratio τ /SuDSS .
The result of this strain accumulation procedure is thus an equivalent number of cycles (Neq ) with a corresponding
cyclic shear stress ratio in direct simple shear (τcy /SuDSS ). The combination of the equivalent number of cycles and the
scaling factor of the shear stress multiplied by the maximum shear stress leads to cyclic failure.
23
4.3.1 Strain accumulation for different OCR values

To investigate the differences between the results of the strain accumulation procedure for the six available OCR values,
a set of load parcels has been created. This set of load parcels is shown in 4.2, and is only for the numerical investigation.
This set of load parcels has no physical meaning, and thus does not represent an actual design storm.
Table 4.2: Set of load parcels used in the strain accumulation procedure for all OCR values
Load Parcel Fcy /Fcy,max N

1 0.1 1000
2 0.3 500
3 0.5 20
4 0.7 5
5 0.9 1
6 1.0 1
The strain contour diagrams for the six OCR values of the Drammen Clay with the results of the strain accumulation
procedure are shown in figure 4.5. The numerical results from the strain accumulation are shown in table 4.3. The blue
lines in figure 4.5 represent the cyclic shear strain contours for different values of the cyclic shear strain. The outermost
contour line defines the failure contour representing 15% cyclic shear strain. In red the end points of the strain accumu-
lation procedure for every scaling factor (χ) is shown. It can be seen that for this set of load parcels the scaling factor
decreases and Neq increases for an increasing OCR value.
Table 4.3: Results of the strain accumulation procedure for the load set in table 4.2
Resulting cyclic shear

OCR Neq Scaling factor
strain [%]
1 2.62 1.12 14.78
2 2.16 1.09 14.83
4 2.56 0.97 14.95
10 2.91 0.96 14.98
20 3.87 0.91 14.97
40 5.14 0.79 14.96
The effect of the decreasing scaling factor with increas-

ing OCR value can be seen from the strain contour dia- τ /SuDSS
grams, as the strain contours move downwards on the y- 1.0
axis (τ /SuDSS ) for increasing OCR values. For OCR = 1
0.8
the 15% cyclic shear strain at 100 cycles is reached with
Drammen clay DSS
a cyclic stress ratio of 0.68 (τcy /SuDSS ), as for OCR = 40
0.6 OCR = 1
the 15% cyclic shear strain at the same number of cycles is
DSS
reached at cyclic stress ratio of 0.43 (τcy /Su ). Overcon- OCR = 4
0.4
solidated soil thus reaches the same cyclic shear strain at OCR = 40
a lower cyclic stress ratio. This corresponds with Andersen 0.2
γ [%]
(2015), who states that "the shear strain at a given de- 0
gree of shear strength mobilization, τ /Su , increases with 5 10 15
increasing OCR, meaning that the normalized secant mod-
Figure 4.4: Effect of the overconsolidation ratio on the normalized stress-strain
ulus, G/Su , decreases with increasing OCR for a given de- relationship of Drammen Clay with overconsolidation ratios of 1, 4, and 40
gree of strength mobilization." This effect is shown in fig- (Recreated from Andersen, 2015)
ure 4.4. It must be noted that it is not the absolute shear stiffness that decreases. "The absolute value of the stiffness
will increase with increasing OCR for given consolidation stress and strength mobilization, both for clay and sand, but
significantly less than the increase in shear strength" (Andersen, 2015).
24
4.3. STEP 1 - DETERMINATION OF THE EQUIVALENT NUMBER OF CYCLES OF A DESIGN STORM
ocr = 1 ocr = 2
1.2 1.2
1 1
τcy / SDSS
τcy / SDSS
0.8 0.8
u
u
0.6 0.6
0.4 0.4
0.2 0.2
0 0
10 0 10 1 10 2 10 3 10 0 10 1 10 2 10 3
N N
(a) OCR = 1, Neq = 2.62 (b) OCR = 2, Neq = 2.16
ocr = 4 ocr = 10
1.2 1.2
1 1
τcy / SDSS
τcy / SDSS
0.8 0.8
u
0.6 0.6
0.4 0.4
0.2 0.2
0 0
10 0 10 1 10 2 10 3 10 0 10 1 10 2 10 3
N N
(c) OCR = 4, Neq = 2.56 (d) OCR = 10, Neq = 2.91
ocr = 20 ocr = 40
1.2 1.2
1 1
τcy / SDSS
τcy / SDSS
0.8 0.8
u
0.6 0.6
0.4 0.4
0.2 0.2
0 0
10 0 10 1 10 2 10 3 10 0 10 1 10 2 10 3
N N
(e) OCR = 20, Neq = 3.87 (f) OCR = 40, Neq = 5.14
Figure 4.5: End points of the strain accumulation procedure for each scaling factor for the set of load parcels in table 4.2 for the different OCR values. The
blue lines represent the cyclic shear strain contours, with the outermost contour defining failure (γcy = 15%).
25
Influence of the OCR value on Neq

From the results for the loads in table 4.2 it was found that with an increasing OCR value, the Neq increases and the
scaling factor decreases. To investigate if this trend is achieved for different load parcels, the set of load parcels from table
4.4 have been used for the strain accumulation procedure. Load set 2 and 3 are generated numerical load sets, and load
set 4 is obtained from Norén-Cosgriff et al. (2015).
Table 4.4: Load sets as input for the strain accumulation procedure
Load 2 Load 3 Load 4

Load Parcel
Fcy / Fcy,max N Fcy / Fcy,max N Fcy / Fcy,max N
1 0.1 2000 0.1 2000 0.05 151
2 0.2 1000 0.3 500 0.15 609
3 0.3 500 0.5 200 0.25 967
4 0.4 400 0.7 50 0.35 1049
5 0.5 200 0.9 2 0.45 940
6 0.6 100 1.0 1 0.55 699
7 0.7 50 - - 0.65 317
8 0.8 10 - - 0.75 122
9 0.9 2 - - 0.83 26
10 1.0 1 - - 0.88 8
11 - - - - 0.93 4
12 - - - - 1.0 5
The resulting equivalent number of cycles and scaling factors from the load sets in table 4.4 are shown in table 4.5. It
can be seen that the trend found in table 4.3 is found in these results as well. For all three load sets the equivalent number
of cycles increases and the scaling factor decreases with an increasing OCR value. The only Neq that falls out of this trend
is in load 2 for OCR 2, which is lower than the value for OCR 1. In general it can be concluded that with an increasing
OCR value the equivalent number of cycles increases and the scaling factor decreases. This means that the effect of the
same storm on the cyclic behaviour is larger for more overconsolidated clay.
Table 4.5: Resulting equivalent number of cycles and scaling factors for load sets 2, 3, and 4.
Load 2 Load 3 Load 4

OCR value
Neq Scaling Neq Scaling Neq Scaling
1 8.61 0.94 5.53 1.01 29.95 0.79
2 7.48 0.88 5.92 0.92 34.83 0.72
4 9.20 0.77 6.26 0.83 39.41 0.63
10 14.71 0.72 8.32 0.80 51.69 0.59
20 19.54 0.68 12.42 0.74 59.68 0.54
40 32.28 0.54 20.80 0.60 107.32 0.42
Comparing the results of load 4 with the results of load 2 and 3, it can be concluded, that an increase in the number
of cycles of the larger load ratios leads to a significant increase in the equivalent number of cycles. Load 2 and 3 have a
resulting Neq in the range of 5 to 10 cycles for the OCR values of 1, 2, and 4. Load 4 has for the same OCR values significant
larger values for Neq . Especially load 4 has a higher number of cycles than the other two load parcels for the higher load
ratios. A design storm with predominantly small cyclic loads and only a few large cyclic loads will have a smaller impact
than a design storm with predominantly large cyclic loads.
26
4.4. STEP 2 - DETERMINATION OF THE DEGRADED STRESS-STRAIN RESPONSE
4.4 Step 2 - Determination of the degraded stress-strain response
The stress-strain behaviour for the three different laboratory tests for a resulting equivalent number of cycles are the
result of the contour diagrams. The translation of the contour diagram to a stress-strain curve is the determination of an
idealized stress-path in the resulting contour diagrams from the equivalent number of cycles. To obtain correct and reliable
behaviour, the inclination of the idealized stress path is of great importance.
Physical meaning of the inclination of the idealized stress path in the contour diagrams
To use the contour diagrams and draw the idealized stress paths in the contour diagrams, a good understanding of the
physical meaning of these stress paths is required. The inclination of the stress path defines the ratio between the cyclic
shear stress over the average shear stress (τcy /τa ) for the three laboratory tests. In a direct simple shear test this ratio
is thus between the shear stress developed by cyclic horizontal shearing in the specimen and the shear stress due to the
vertical pressure on the specimen. The determination of the stress path is dependent on the investigated case.
4.4.1 Resulting contour diagrams

The resulting contour diagrams are the result of the aforementioned strain accumulation procedure. The result of the
strain accumulation procedure is an equivalent number of cycles. For every type of Drammen Clay the contour diagrams
for Neq of 1, 10, 100, and 1000 are established. These contour diagrams are interpolated for the Neq that comes out of
the strain accumulation procedure.
In figure 4.6 the resulting interpolated triaxial contour diagrams are shown for Drammen Clay with OCR 1 for Neq
values of 5 and 50. The horizontal axis represents the ratio of the average shear stress to the static undrained shear
strength in active failure (τa /SuA ). The vertical axis represents the ratio of the cyclic shear stress to the static undrained
shear strength in active failure (τcy /SuA ). The semi-vertical blue lines define the average shear strain contours, the semi-
horizontal red lines define the cyclic shear strain contours. The triaxial contour diagram is bounded by -15% average shear
strain (outermost left semi-vertical strain contour), 15% cyclic shear strain (uppermost semi-horizontal red strain contour),
and 15% average shear strain (outermost right semi-vertical strain contour). It can be seen that Neq predominantly impacts
the cyclic stress-strain behaviour.
1 1
0.9 0.9
0.8 0.8
0.7 0.7
0.6 0.6
τcy / S Au
τcy / S Au
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
-0.5 0 0.5 1 -0.5 0 0.5 1
τa / S Au τa / S Au
(a) Neq = 5 (b) Neq = 50
Figure 4.6: Resulting triaxial contour of OCR 1 for Neq 5 and 50 with the average strain contours (blue contour lines), and the cyclic strain contours (red
contour lines)
Scaling of the resulting contour diagrams

In the UDCAM-S procedure it is possible to scale the resulting direct simple shear and the triaxial contour diagrams
in a vertical direction. This allows for a stronger or weaker stress-strain behaviour with the existing contour diagrams of
the Drammen Clay. With this method the UDCAM-S procedure can be used for other type of clays that have a different
stress-strain behaviour. With a limited number of cyclic laboratory tests the data from the tests can be fitted by scaling the
contour diagrams.
27
4.4.2 Determination of the stress path

The determination of the degraded stress-strain behaviour can be divided into three steps using the idealized stress
path. The steps to obtain the correct target stress-strain diagrams from the scaled contour diagrams are:
1 Determine the ratio between the cyclic and average load on the foundation (Fcy /Fa or Mcy /Ma ). This can result
from the structural analysis of the loads on the foundation.
2 In the DSS contour diagram, the idealized stress path starts at zero percent cyclic and average strains and follows
the inclination defined by Fcy /Fa or Mcy /Ma and ends at the failure contour of the DSS contour diagram.
3 Draw the idealized stress paths in the Triaxial contour diagram for the TXE and TXC starting from zero percent strains
and resulting in the same combination of average and cyclic strains as in the DSS contour diagram, to maintain the
assumption of strain compatibility.
Determination of the stress path for Jack-up structures

The determination of the stress path is different when performing an analysis for jack-up structures. With jack-up
structures, the shear strength is partly mobilised by the weight of the structure. Therefore this initial strength mobilisation
τa,mob /SuA has to be taken into account. The steps to determine the correct idealized stress path including the initial

strength mobilization are:
1 Define the initial average shear stress mobilisation τa,mob /SuA

2 Determine the ratio between the cyclic and average load on the foundation (Fcy /Fa or Mcy /Ma )
3 Draw in the DSS contour diagram interpolated for the correct Neq a line starting from τa,mob /SuA with an inclination
of Fcy /Fa (or Mcy /Ma )
4 The idealized stress path starts at zero percent strains and ends at the intersection between the line drawn in step 3
and the failure contour of the DSS contour diagram
5 Draw the idealized stress paths in the Triaxial contour diagram for the TXE and TXC starting from zero percent strains
and resulting in the same cyclic strains as the in DSS.
This process is shown in 4.7(a). The process starts at τa,mob /SuA , and with the inclination of Fcy /Fa to the point at
which the failure contour is reached, indicated with γcy |γa . This line is drawn with the dashed line. The idealized stress
path for the DSS test, indicated with the continuous line, is drawn from zero percent strains to the same point at the
contour. The point (γcy |γa ) is then used in the triaxial contour for the triaxial extension and the triaxial compression test
to reach the same value of γcy by the idealized stress path.
The result of the stress path is shown in figure 4.7(b), in which the resulting target stress-strain diagram is shown. The
points in figure 4.7(b) are the result of the drawn stress path, and found by interpolation of the strain contour lines from
the corresponding strain contour diagram.
1
1
0.9
0.9
0.8
0.8
0.7
0.7
(τ cyclic + τ average ) / SA
u
0.6
0.6
τ cy / SA
u
0.5 ( γcy | γa )
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
( τ a,mob / SA
u
)
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18
τ a / SA
u
γcy + γa
(a) Determination of the stress path (b) Resulting target stress-strain diagram for a total analysis
Figure 4.7: Determination of the stress path using the DSS contour diagram in case of a jack-up structure and the resulting target stress-strain diagram
28
4.5. STEP 3 - PARAMETER DETERMINATION OF THE NGI-ADP SOIL MODEL
4.5 Step 3 - Parameter determination of the NGI-ADP soil model
The parameters of the NGI-ADP soil model are determined in the UDCAM-S by simulating the three determining
laboratory tests: the undrained triaxial compression test, the undrained triaxial extension test, and the undrained direct
simple shear test.
4.5.1 Initial values of the laboratory tests for the optimisation

The initial values that are used for the laboratory tests in the parameter optimisation in the UDCAM-s are to be deter-
mined correctly to describe the soil behaviour in the right way. As aforementioned, the laboratory tests are simulated with
a cell pressure of 300 kPa. The parameters that are used in the optimisation, such as the undrained shear strength and
average mobilisation, must be correct for the simulated stress conditions in the laboratory tests.
A common used relation between the effective vertical stress and the undrained shear strength in compression is the
SHANSHEP equation (Ladd and Foott, 1974). The SHANSHEP equation relates the undrained shear strength with respect
to the effective vertical stress to the overconsolidation ratio, as shown in equation (4.10). It can be concluded that the
ratio of the undrained shear strength to the vertical effective stress is a constant for one type of the Drammen Clay. Using
the SHANSHEP equation with a correct value of m, the value of the undrained shear strength can be determined.
Su
0
= S · OCRm (4.10)
σyy,0
As the initial mobilisation follows from the contour diagram for the triaxial compression and triaxial extension test in
0
the second step of the procedure, the correct values for the earth pressure coefficient K0 and vertical effective stress σyy
are to be determined for the optimisation step in the procedure. In the UDCAM-S procedure the vertical cell pressure is
fixed to 300 kPa within the optimisation. To satisfy the condition set in equation (4.10), the correct values for K0 have
to be derived. The derivation of these values is achieved by equation (4.11), in which SuA is the static undrained shear
strength in active failure.
σyy
τ0 /SuA · SuA = (1 − K0 ) (4.11)
2
The correct K0 value for the optimisation is then calculated by
τ0
2· · SuA
!
A
Su
K0 = 1− (4.12)
σyy
Using this transformation the contour diagrams can be used in the optimisation for other cell pressures than those
which have been used to establish the contour diagrams of the Drammen Clay. These diagrams are constructed from labo-
ratory tests at certain cell pressures. However, these contour diagrams are applicable for other cell pressures if the initial
mobilisation and the undrained shear strength are scaled to the correct cell pressure.
0 0
A triaxial test sample is shown in figure 4.8 The correct vertical and horizontal cell pressures (σyy and σxx ) are thus
expressed in terms of the initial mobilisation (τ0 /SuA ) and the earth pressure coefficient (K0 ) by
2τ0
σyy =
1 − K0
2τ0 (4.13)
σxx = K0 · σyy = · K0
1 − K0
29
σyy = 300 kPa
σxx = σyy · K0
Figure 4.8: Triaxial test set-up for the optimisation in the third step of the procedure
4.5.2 Optimization of the stiffness parameter G/SuA

The stiffness of the soil in the UDCAM-S is described by the stiffness parameter G/SuA . As the strength parameters
(SuC /SuA ,
SuE /SuA , and SuDSS /SuA ) and the failure strains (γfC , γfE , and γfDSS ) are determined separately for the three
laboratory tests (TXC, TXE, and DSS), the stiffness parameter is optimized for all three tests simultaneous. G/SuA is thus
optimized for all available points generated with the target stress-strain curves.
Figure 4.9 shows a case in which all parameters are fixed except for G/SuA . The optimisation is performed using the
optimisation tool within the Plaxis software. All three target stress-strain curves from laboratory tests (blue curves) are
used in the optimisation of G/SuA . It can be seen that the overall fit of the stress-strain curves for the optimized parameters
(orange curve) on the target stress-strain curves is not optimal. At strains larger than 5% the stiffness of the optimized
curve is underestimated with respect to the target stress-strain curve. However, for very small strains the fitted curves
match the target curves accurately. However, the overall fit of the stiffness is underestimated.
(a) TXC (b) DSS (c) TXE
A
Figure 4.9: Optimized stress-strain curves with a large optimisation range for G/Su . The blue line is the target stress-strain curve, the orange line is the
optimized stress-strain curve.
To optimize the stiffness parameter G/SuA better for the complete target stress-strain curve, it is possible to use only
one of the three target stress-strain curves for the optimisation. This process is shown in figure 4.10. Figure 4.10(a) shows
the selection of the target stress-strain points larger than 0.1% vertical strain (yy ). Only these data points are now used
in the optimisation of G/SuA .
The fitted curve in figure 4.10(b) matches the target stress-strain curve better with respect to the fit in figure 4.9(c).
However, it should be noted that for very small strains, the stiffness is now overestimated, as can be seen in figure 4.10(c).
For vertical strain strains smaller than 0.06%, the optimized stress-strain curve is larger than the target stress-strain curve.
The difference in the stiffness parameter between the fit of figure 4.9 and figure 4.10 is significant. The optimized value
of G/SuA by using all data points of the three target stress-strain curves is 373.6. The optimized value of G/SuA using the
larger strains in only the triaxial extension target stress-strain curve is 731.5. This means that a relatively small difference
of the curve (figures 4.9(c) and 4.10(c)) can result in a doubled stiffness of the soil.
30
4.5. STEP 3 - PARAMETER DETERMINATION OF THE NGI-ADP SOIL MODEL
(a) Selection of the target stress-strain points (b) Fitted curve for the TXE test (c) Fit at very small strains for the TXE
A
Figure 4.10: Optimization of G/Su using only a selection of the target stress-strain points for the triaxial extension test
From Jostad et al. (2015) it can be concluded that a first guess of the stiffness parameter can be obtained using the
initial shear stiffness of the soil (G0 ) and the degraded undrained shear strength in active failure (SuA ). The stiffness
parameter is then determined by dividing the initial shear stiffness of the soil by the degraded undrained shear strength in
active failure. In the previous described case this would have led to a value of G/SuA of 761, which is close to the optimized
value using the larger strains (731.5).
In figure 4.11 the influence of an increasing value of the stiffness parameter on the stress-strain response in the NGI-
ADP model is shown. The stress-strain curve is shown for an undrained direct simple shear test. The soil is modelled with
the NGI-ADP soil model with an undrained shear strength in direct simple shear of 100 kPa and a failure shear strain in
direct simple shear of 10.0 %. It can be seen that the curves for G/SuA of 100 and 200 can be distinguished. However, the
difference in stress-strain behaviour between the values of 800 and 900 for G/SuA is almost negligible.
It can also be seen in figure 4.11 that the influence of G/SuA on the stress-strain behaviour is only visible at smaller
shear strains. At larger strains in figure 4.11 (shear strains larger than 5%) the differences between the lines representing
the stress-strain behaviour are negligible.
Figure 4.11: Influence of the stiffness parameter on the stress-strain response in direct simple shear for a soil with γfDSS = 10% and Su
DSS
= 100kPa
It must be decided, based on the project, which order of strains are of the most importance, and thus at which strain
level the optimized stress-strain curve should fit the target stress-strain curve the best. The sensitivity of this parameter
should be taken into account when the stability of a structure is analysed using UDCAM-S.
31
4.5.3 Recommended procedure for the optimisation of the NGI-ADP parameters

The optimisation tool in Plaxis is implemented in such a way that it does not find the best possible fit in one time with
large ranges. Ukritchon et al. (1998) propose equations that are able to fit the parameters of the NGI-ADP soil model
with high accuracy using a solver routine. Unfortunately not all equations are proposed in their paper, and therefore their
method is not reproducible. Therefore it is recommended to perform the optimisation of the parameters in the following
order:
1 Determine the undrained shear strength parameters (SuC /SuA , SuE /SuA , and SuDSS /SuA )
These parameters are independent, as they are all defined with respect to the static undrained shear strength in
active failure, and thus can be optimized simultaneously.
2 Determine the failure shear strains (γC , γE , and γDSS )

Same as the strength parameters, the failure strains are independent, and thus can also be optimized simultane-
ously.
3 Optimize the stiffness parameter (G/SuA )

As mentioned, the optimisation of the stiffness parameter is project dependent. It must also be kept in mind
that the curvature of the stress-strain response does not change significantly with a changing stiffness parameter
(G/SuA ) as shown in figure 4.11, but the change in shear stiffness (G) is significant with a changing stiffness
parameter.
4.5.4 Softening in compression for normally consolidated clays

The initial stress (i.e. 0% cyclic shear strain and 0% average shear strain) for normally consolidated Drammen Clay
(OCR = 1) in the triaxial contour lies at a ratio of 0.76 (τa /SuDSS ).
Neq = 50
1
0.9
0.8
0.7
0.6
τcy / S Au
0.5
0.4
0.3
0.2
0.1
0
-0.5 0 0.5 1
τa / S Au
Figure 4.12: Softening in triaxial compression with the actual stress path, represented by the continuous black line, and the idealized stress path in the
UDCAM-S, represented by the dashed black line (adapted from Andersen, 2015)
In cases in which the cyclic shear stress is much larger than the average shear stress (τcy /τa >> 1), one should be aware
of the possibility that the cyclic strength is lower than the initial stress in triaxial compression. This situation would lead
to softening in the triaxial compression, and thus the behaviour of the soil is described by a softening stress-strain curve.
The complex UDCAM has the ability to cope with this process according to Hans Petter Jostad (personal communication,
June 2016). However, as the UDCAM-S idealizes the stress path with a straight line through the triaxial contour diagram,
it is not possible to produce a correct stress-strain curve describing this behaviour. Possible solutions to produce the correct
cyclic strength with the UDCAM-S for normally consolidated clays is to reduce the initial stress in triaxial compression or
to draw an idealized stress path in triaxial compression to the left side of the initial mobilization.
32
4.6. VERIFICATION WITH THE CASE-STUDY PRESENTED BY JOSTAD ET AL. (2015)
4.6 Verification with the Case-Study presented by Jostad et al. (2015)
Jostad et al. (2015) presented a case study for a FE calculation for a Spudcan stability analysis with a simplified ap-
proach of the UDCAM in "A FE procedure for calculation of fixity jack-up foundations with skirts using cyclic strain contour
diagrams", similar to the implementation of the UDCAM-s procedure in Plaxis. The results of this case study are used to
compare the results of the two approaches in terms of the resulting NGI-ADP soil parameters.
The geometry of the spudcan and the soil beneath the foundation is shown in figure 4.13. The soil profile in the paper
is divided into two layers, one of an overconsolidated clay with OCR = 4, and one of an overconsolidated clay with OCR =
40. In this case only the parameter determination is compared for the lower layer with an overconsolidation ratio of OCR
= 4. For this layer a constant active shear strength is considered, resulting in values of yref and Su,inc of the NGI-ADP
model, equal to zero.
Spudcan geometry
D = 22 m
hs =2m
Loads
Preload = 160 MN
VSWL = 100 MN
hs =2 m
OCR = 40
s uC = 130 kPa D = 22 m
Gmax = 400 * suDSS= 35 880 kPa
7.0 m
OCR =4 Both layers
s uC = 130 kPa suDSS /suC = 0.69
Depth
Gmax = 750 * suDSS= 67 275 suE /suC = 0.56
below
kPa γ’ = 10 kN/m3
seabed,
z (m)
Figure 4.13: Geometry of the spudcan and the parameters of the soil beneath the spudcan by Jostad et al. (2015)
4.6.1 Strain accumulation procedure

In the case of the spudcan in the paper, the loads in the design storm are defined as overturning moments (OTM)
instead of horizontal loads. The strain accumulation procedure is the same as for horizontal loads, as the OTM in each
load parcel is defined as percentage of the maximum OTM. The input in the strain accumulation procedure is shown in
table 4.6.
Table 4.6: Input of the load parcels for the strain accumulation in terms of the overturning moment (OTM) in percentage of the maximum overturning
moment
OTM in % of maximum 2 11 26 40 51 62 75 89 100

Number of cycles 2371 2877 1079 163 64 25 10 3 1
The result of the strain accumulation procedure is shown in figure 4.14 and table 4.7. The resulting Neq of the procedure
for the load set is 5.27, with a scaling factor χ of 0.87. In the figure, the locus of the end points for all the scaling factors
are shown. The corresponding cyclic shear strain of the last scaling factor is 13.64 %.
33
ocr = 4
1.2
τcy / SDSS
0.8
u
0.6
0.4
0.2
0
10 0 10 1 10 2 10 3
N
Figure 4.14: Locus of end points from the strain accumulation procedure for all scaling factors
Table 4.7: Results from the strain accumulation procedure for OCR = 4 for the load set in table 4.6
UDCAM-S Jostad et al.

procedure (2015)
Neq 5.3 5.0
χ 0.87 0.86
Jostad et al. found that a value of 5.0 was representative for this load set. This value corresponds reasonably with the
values found using the UDCAM-S procedure. The resulting scaling factor of the UDCAM-S is in accordance with the paper
as well.
4.6.2 Idealized stress path

The outcome for the strain accumulation procedure has been set to Neq = 5.0 for OCR = 4, to be able to compare the
results more accurately. However, it should be noted that the results for the following steps are slightly different if the
outcome of the UDCAM-S strain accumulation would have been used.
The initial average shear stress mobilisation τa,mob /SuDSS is defined by Jostad et al. (2015) as 0.4. However, in the

contour diagrams of the UDCAM-S, the shear stress are scaled with respect to the active undrained shear strength (SuA ).
Therefore the initial mobilisation from the paper has to be transformed with respect to the active undrained shear strength.
The ratio between the undrained shear strength and the active shear strength by Jostad et al. (2015) is defined as 0.69.
The required initial mobilisation becomes
)
τa,mob /SuDSS = 0.40
τa,mob /SuA = 0.28. (4.14)
SuDSS /SuA = 0.69
The ratio of the cyclic overturning moment to the average overturning moment is defined as 3.0, resulting in failure at
2.2% cyclic strain and 15% average strain for direct simple shear. The idealized stress path for the DSS test is drawn in
figure 4.15(a). In the triaxial contour diagram the stress paths are chosen in a way to reach the same cyclic and average
shear strain. These stress paths are shown in figure 4.15(b) for OCR = 4. Although the y-axis of the direct simple shear
contour diagram differs from the y-axis form the paper, the resulting shear strain combination corresponds with the values
from the paper.
34
4.6. VERIFICATION WITH THE CASE-STUDY PRESENTED BY JOSTAD ET AL. (2015)
1 1
0.8 0.8
0.6 0.6
τ cy / SA
τ cy / SA
u
u
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
τ a / SA
u
τ a / SA
u
(a) DSS contour (b) TX contour
Figure 4.15: Visual representation of the 2D contour diagrams for Neq = 5 for OCR = 4 with the defined stress paths for DSS, TXE, and TXC
The resulting target stress-strain relations for the triaxial compression and extension test, and the direct simple shear
test are shown in Figure 4.17.
4.6.3 Parameter determination

The NGI-ADP model parameters are fitted on the target stress-strain curves using the parameter optimisation in Plaxis.
The resulting parameters by the paper and the fitted parameters by the UDCAM-S are shown in Table 4.8. The difference
in the parameters in the cyclic analysis are negligible. The only small difference between the two methods is the definition
of the failure strains in the three laboratory tests. This difference will result in a different stress-strain behaviour, but the
failure shear stress is in both methods the same. However, when calculating the response with a total analysis, the soil
behaves softer in the UDCAM-S than in the resulting stress-strain behaviour defined by the paper.
Table 4.8: Comparison of the resulting cyclic parameters of the paper and the UDCAM-S
Results by Jostad et al. (2015) UDCAM-S optimisation

Parameter
cyclic total cyclic total
Suc [kPa] 88.4 149.5 88.5 144.8
Gur /scu [-] 761 450 760 446
SuE /Suc [-] 0.84 0.57 0.842 0.571
SuDSS /Suc [-] 0.66 0.70 0.688 0.758
γfC [%] 2.0 7.0 2.59 6.236
γfE [%] 2.0 7.0 2.98 6.990
γfDSS [%] 2.5 7.0 2.51 9.595
The stress-strain responses by Jostad et al. (2015) for the layer corresponding with OCR = 4 are shown in Figure 4.16.
For the cyclic analysis the NGI-ADP fitted parameters give the same result in TXE, TXC, and DSS as the target stress-strain
results. In the total analysis the NGI-ADP fitted parameters give a different result than the target stress-strain results.
35
Figure 4.16: Resulting stress-strain responses for the cyclic and total analyses for OCR = 4 by Jostad et al. (2015)
The resulting stress-strain responses and the fitted curves for the cyclic analysis are shown in Figure 4.17. The scatter
points are the resulting target stress-strain points from the idealized stress paths in the three laboratory tests. On the
horizontal axis the shear strain in the corresponding tests is plotted, on the vertical axis the ratio of the shear stress to the
undrained shear stress in compression.
0.7 0.5 0.6
0.6 0.5
0.4
0.5
0.4
0.3
0.4
τ DSS /suC [ − ]
[−]
[−]
0.3
τ E /suC
τ C /suC
0.3
0.2
0.2
0.2
0.1
0.1
0.1
target_TRXC target_DSS target_TRXE
optimized_TRXC optimized_DSS optimized_TRXE
0.0 0.0 0.0
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 0 2 4 6 8 10 12 14 16
γC [%] γ DSS [%] γE [%]
Figure 4.17: Resulting laboratory tests fitted on the target stress-strain curves for the cyclic analysis for OCR = 4 by the UDCAM-S procedure
The resulting stress-strain responses and the fitted curves for the total analysis are shown in Figure 4.18. The scatter
points are the resulting target stress-strain points from the idealized stress path. Comparing the results in Figure 4.18
with Figure 4.16, it can be seen that the total analysis parameters do not fit as well as the cyclic parameter on the target
stress-strain results from the procedure, especially for the DSS results.
1.2 1.0 0.7
0.6
1.0
0.8
0.5
0.8
0.4
0.6
τ DSS /suC [ − ]
[−]
[−]
0.6 0.3
τ E /suC
τ C /suC
0.4 0.2
0.4
0.1
0.2
0.2
0.0
target_TRXC target_DSS target_TRXE
optimized_TRXC optimized_DSS optimized_TRXE
0.0 0.0 −0.1
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18
γ C
[%] γ DSS [%] γ E
[%]
Figure 4.18: Resulting laboratory tests fitted on the target stress-strain curves for the total analysis for OCR = 4 by the UDCAM-S procedure
The results by Jostad et al. in the total analyses differ more from the target stress-strain curves, as the same failure
strain has been used for TXC, DSS and TXE in the paper. However, as can be seen from figure 4.18 the stress-strain curves
36
4.7. SUMMARY
have a better fit when different failure strains are chosen for TXC, DSS, and TXE, although this does not comply with the
strain compatibility theory.
4.7 Summary
In this chapter the Simplified UnDrained Cyclic Accumulation Model has been presented. This model takes the effect
of cyclic degradation into account of the soil by the cyclic load history. This effect is included by the assessment of the
accumulated plastic shear strains in the strain accumulation procedure, as presented by Andersen (1976). The result of this
procedure is an equivalent number of cycles (Neq ), at a specific cyclic shear stress ratio in direct simple shear (τcy /SuDSS ).
A direct simple shear test with this equivalent number of cycles at the given cyclic shear stress ratio has the same effect on
the soil behaviour as the given load history.
Given the equivalent number of cycles from the strain accumulation procedure, the degraded stress-strain response of
the soil in triaxial compression, triaxial extension, and direct simple shear is determined. These curves are derived from
the strain contour diagrams in direct simple shear, triaxial extension, and compression using an idealized stress path. This
stress path is defined by the ratio of the cyclic load to the average load in the cycle with the largest load amplitude (Jostad
et al., 2015). These loads are the structural loads as a response on the external loads in the design storm, i.e. the wind
and wave loads on the structure (Zaaijer, 2007). From the stress path a decision can be made on how to evaluate the
cyclic stress-strain response of the soil (SLS) or the total stress-strain response of the soil (ULS). The cyclic response is used
in dynamic structural calculations, the total response is used to evaluate the stability of the foundation (Jostad et al., 2015).
The parameters of the NGI-ADP soil model can be derived from the degraded stress-strain response in the three lab-
oratory tests. For this part it is important to keep in mind the level of shear strains which is governing the design of the
foundation, a perfect fit with the NGI-ADP parameters on the degraded stress-strain curves is not always possible.
The database of UDCAM-S consists of the Drammen Clay data with overconsolidation ratios of 1, 2, 4, 10, 20, and 40,
based on many laboratory tests performed by the NGI. Other types of clay can be used by interpreting the soil behaviour
and comparing the results from laboratory tests with the data from the Drammen Clay. For an accurate assessment of the
soil behaviour of another type of clay it would be best to develop contour diagrams for this type of clay. However, this
would lead to many cyclic and static lab tests to obtain enough data to construct the contour diagrams. A well working
solution seems is to interpret the behaviour of a specific clay with a limited number of laboratory tests, and find the best
fit with the data by scaling the contour diagrams of the Drammen Clay. However, this is only possible for the contour
diagrams, the strain accumulation contour diagrams cannot be modified. These diagrams can be constructed with limited
symmetric direct simple shear tests, resulting in an accurate description of the soil behaviour of the clay. The strain contour
diagrams of the Drammen clay can then be scaled to match the behaviour of the specific clay.
37
5.
Modelling the influence of cyclic degradation on the
stability of a gravity based foundation
In this chapter the influence of cyclic degradation for a gravity based foundation is presented. The UDCAM-S procedure
is used to obtain the degraded parameters for different cases, to simulate different outcomes of the UDCAM-S procedure.
All finite element analyses are performed in the finite element software Plaxis 2D. The results of the calculations are
presented in VHM diagrams, to show the influence of the different input parameters of the UDCAM-S on the capacity of
the foundation. Afterwards the influence of the width of the foundation is investigated with respect to the determined soil
parameters.
5.1 Finite element model
The design of an offshore foundation is a complex 3D problem. The final design of an gravity based foundations should
be performed using 3D calculations. However, for this research the problem has been modelled in 2D, in accordance with
Taiebat and Carter (2000), who investigated the difference between modelling a shallow foundation in 2D and 3D. The
modelling in 2D is also used for common practice, as described by API (2011).
The used model is a plane strain finite element model with 15 noded elements. An axisymmetric model would have
been useful for the 2D representation of a circular footing. However, as the failure mechanism in combined loading is not
symmetric, axisymmetric modelling of the foundation is an incorrect assessment of the stability.
A homogeneous value of 100kPa has been assigned to the static undrained reference shear strength in active failure
A
(Su,ref ) as an input of the UDCAM-S procedure, with a saturated and dry unit weight (γs and γd ) of 20kN/m3 .
5.1.1 Foundation
The foundation has been modelled as a stiff plate with a concrete block on top. The plate has no bearing capacity in the
x-direction of the model, and therefore the option "end-bearing" has been disabled. The plate is modelled as an isotropic
linear elastic material. The concrete is modelled with a large stiffness, assuring that negligible deformations occur in the
foundation, and the self-weight of the concrete is set to zero. The concrete is modelled using a non porous linear elastic
soil model. The parameters of the applied soil model can be found in table 5.2.
Table 5.1: Parameters of the plate model
Material type EA1 EI ν γunsat / γsat

2
kN/m kNm /m - kN/m3
Elastic 1.0 · 109 1.0 · 109 0.00 -
39
CHAPTER 5. MODELLING THE INFLUENCE OF CYCLIC DEGRADATION ON THE STABILITY OF A GRAVITY BASED
FOUNDATION
Table 5.2: Parameters of the soil model for the concrete
Soil model E G ν γunsat / γsat

2 2
kN/m kN/m - kN/m3
Linear-elastic 1.0 · 109 5.0 · 1011 0.00 0.00
The interface between the foundation and the soil has been modelled using an interface allowing zero tension between
the concrete and the soil. The decision for no tension is made because of the absence of skirts in the foundation. This
method is also applied in the paper of Ukritchon et al. (1998), in which they found a failure envelope for combined loading
of strip foundations. According to the API (2011), "engineering judgement is required to determine whether suctions will
be generated and the duration over which they can be maintained." Assuming zero suction between the soil and the
foundation is a conservative approach.
The modelled foundation is placed directly on the seabed. The effect that by tilting of the structure water can flow
under the foundation and initiate erosion of the soil is acknowledged, but neglected in this research.
In the NGI-ADP soil model a specific traction criterion is defined for interfaces. However, it is not possible to use a
tension cut-off in this soil model. This would lead to an over-estimation of the moment capacity of the foundation, as
shown in figure 2.8. Therefore the interface has been chosen of a Mohr-Coulomb type with an undrained shear strength
defined as the mean of the three undrained shear strength parameters of the NGI-ADP model.
5.1.2 Mesh
The mesh that has been used in the analysis is shown in figure 5.1. A refined mesh has been applied under the
foundation, which allows for more accurate calculations without increasing the mesh refinement for the entire mesh. The
size of this refined mesh has been chosen such that the failure mechanisms of pure loading, as shown in figure 2.6, are
incorporated within this surface.
Figure 5.1: Finite element mesh for the analysis of the gravity based foundation
Mesh dependency
To investigate the mesh dependency of the analysis, the soil has been chosen as OCR = 4 with the resulting parameters
from Neq = 1, and τcy /τa = 1.5. The corresponding parameters can be found in the first case (section 5.2.1). The mesh
dependency is investigated for both the ultimate horizontal and the ultimate vertical load. In table 5.3 the results can be
found for the mentioned soil. The coarseness factor for the refined mesh underneath the foundation has been chosen as
0.5.
40
5.1. FINITE ELEMENT MODEL
Table 5.3: Mesh dependency of the ultimate loads
Coarseness Number of
Vult /ASu Hult /ASu
factor elements
0.80 467 5.2456 0.9992
0.60 745 5.2364 0.9996
0.50 881 5.2231 0.9979
0.40 1103 5.2258 0.9983
0.30 1489 5.2214 0.9974
0.20 2621 5.2036 0.9961
0.10 8139 5.1718 0.9926
(a) Coarseness factor 0.8 (b) Coarseness factor 0.5
(c) Coarseness factor 0.3 (d) Coarseness factor 0.1
Figure 5.2: Meshes for four different coarseness factors for the soil close to the foundation
To investigate the consistency of the results from the load controlled finite element analysis the same analysis as in
table 5.3 has been performed for a soil profile with OCR 4 with the resulting parameters from Neq = 5, and τcy /τa = 1.5,
which can be found in case 1. The ultimate horizontal and vertical loads for a strip foundation according to Vesic (1975)
are 1.00 (H/BSu ) and 5.14 (V /BSu ) respectively. The results for both analyses can be found in figure 5.3. It can be seen
that, although the results are not consistent over the different mesh sizes, the difference between the two soil profiles is
minimal. The results found with the load controlled finite element method are very accurate for the horizontal ultimate
load. The theoretical value of 1.0 for H / BSu is reached in almost all of the results, as shown in figure 5.3.
Figure 5.3: Mesh dependency of the ultimate loads for different mesh refinements of the inner soil surface for Neq = 5, and 10 with OCR = 4 and τcy /τa
= 1.5
41
FOUNDATION
5.1.3 Influence of vertical and horizontal loading

The fact that the ultimate vertical load decreases with increasing elements, which is in contrast to the constant ulti-
mate horizontal load with increasing elements, can be explained by the failure mechanism. The failure mechanism of the
maximum vertical load is according to Brinch Hansen (Verruijt, 2012). The start of this mechanism can be seen in figure
5.4(a), in which the plastic strains are shown at failure. This failure mechanism is very deep in contrast to the horizontal
failure mechanism, as shown in figure 5.4(b). With the increasing number of elements the resulting ultimate vertical load
converges to the actual ultimate vertical load for a strip foundation, which is found as V/ASu = 2+π (Vesic, 1975; Taiebat
and Carter, 2002).
In figure 5.4 it can be seen that the influence depth is different for each of the loading types. The influence of vertical
loading for the generation of plastic shear strains is approximately 1.5 times the width of the foundation. The influence of
the pure horizontal loading is very shallow in terms of plastic strains, as the failure occurs at a shallow failure surface.
(a) Plastic shear strains in pure vertical loading
(b) Plastic shear strains in pure horizontal loading
Figure 5.4: Developed plastic strains (γp ) in pure horizontal and vertical loading at failure
42
5.1. FINITE ELEMENT MODEL
5.1.4 Verification
With a load-controlled analysis different problems arise. To ensure that the results found in this research can be
evaluated, the load-controlled method is verified with published research. The normalized loads for vertical (v), horizontal
(h), and moment loading (m) are defined by
v = V / BSu
h = H / BSu (5.1)
2
m = M / B Su
The sign convention for the loads, as shown in figure 2.7, is used in the analysis for the VHM failure envelopes. The
ultimate loads are defined with respect to the width of the foundation (B) and the average undrained shear strength (Su ).
In figure 5.5 the difference between the lower bound solution from Houlsby and Purzin (1999) and the performed load
controlled finite element method is shown with respect to the results found by Taiebat and Carter (2002). The vertical
load has been applied as a constant line load over the width of the foundation. The moment is applied at the reference
point according to figure 2.7.
Figure 5.5: Validation of the load controlled finite element analysis with Taiebat and Carter (2002)
From figure 5.5 it can be concluded that the load displacement finite element method coincides with the results found
by Taiebat and Carter (2000). From figure 5.3 it can be seen that the vertical bearing capacity with the load controlled
finite element method is slightly overestimated for all the mesh sizes. However, as mentioned, the result is consistent for
different soil profiles. Therefore the load controlled finite element method can be used in this research for the development
of VHM failure envelopes.
(a) h/m = -4.0 (b) h/m = -0.25 (c) h/m = 0.25 (d) h/m = 4
Figure 5.6: Failure mechanisms in terms of the total displacements [m] for four different load ratios of h/m with a vertical load of v = 2.6 V/(A/Su )
43
FOUNDATION
5.1.5 Displacement controlled

A displacement controlled finite element analysis would be ideal for the analysis on the limit states of combined
loading on gravity based foundations. The behaviour such as shown by Gourvenec (2007), Bransby and Yun (2009), and
Gottardi et al. (1999), in which the yielding of the soil in combined loading can be analysed, is not possible with load
controlled finite element analysis. It is possible to apply prescribed displacements in Plaxis 2D and in this way achieve
correct displacement and rotation ratios corresponding to the method by Gourvenec (2007), but no working solution has
been found yet to measure the separate resisting horizontal and vertical force and the overturning moment. Therefore the
analysis is performed using the load-controlled method.
5.1.6 Python scripting

To visualize VHM failure envelopes, many individual calculations are required. To minimize the calculation time a
Python1 script has been written to change the load combinations for every calculation and retrieve the results from the
output of Plaxis. The remote scripting server in Plaxis has been used to be able to control the Plaxis software from the
Python environment. At first the ultimate loads are calculated for every analysis. Based on the values for Hult , Vult , and
Mult load combinations are determined. These load combinations are read from an Excel file into Matlab to be processed.
The pseudo-code of the python script is shown in figure 5.7.
Calculate load
Determine Write the loads
combinations
ultimate loads from the Excel
based on the
(Vult and Hult ) file into Plaxis
ultimate loads
Plaxis 2D Excel
Calculation
Write the of maximum
calculation results loads with Plaxis
to an Excel file for the load
combinations
Python Plaxis 2D
Figure 5.7: Pseudo code of the python script for automatic Plaxis calculations using Excel, Plaxis 2D and Python
Determination of the load

For every analysis different load combinations are considered for different constant values of the vertical load. To
obtain the entire VHM failure envelope, 10 individual load ratios for the normalised loads (h/m) are considered besides
the pure loading of the vertical load and the horizontal load in both directions. This results in 13 individual calculations
for every value of the vertical load.
1
The software and documentation of the open source software PythonTM can be found on https://www.python.org/
44
5.2. PARAMETER DETERMINATION USING THE UDCAM-S PROCEDURE
5.2 Parameter determination using the UDCAM-S procedure
In this section the parameters are determined that take cyclic degradation into account. Four cases are presented to
investigate the influence of the different steps in the UDCAM-S procedure. The four cases of investigated are:
• The influence of an increasing Neq for OCR = 4

• The influence of an increasing Neq for OCR = 1
• The influence of an increasing Neq for OCR = 1 assuming isotropic behaviour
• The influence of the cyclic load ratio for OCR = 4
With these four cases the influence of the different steps of the procedure should become clear, in terms of the resulting
NGI-ADP parameters, which are then applied to the stability of a gravity based foundation.
5.2.1 Influence of an increasing Neq for OCR = 4

The first case is an investigation on the influence of accumulation of shear strains in the soil due to an offshore storm.
As discussed in chapter 4, the effect of a storm is expressed as the equivalent number of cycles (Neq ). This number of cycles
is determined with the strain accumulation procedure, as presented by Andersen (1976). As discussed by Norén-Cosgriff
et al. (2015), different method of simplifying a design storm, in terms of load parcels, can lead to significant different
results in the equivalent number of cycles. It is therefore important to investigate the influence of a change in equivalent
number of cycles for the stability of a gravity based foundation.
In a real project the equivalent number of cycles is the result of a design storm that is measured at the project site.
However, for this research, the values of the equivalent number of cycles are fixed to 1, 5, 10, and 20. This range of values
is representative for possible outcomes of the strain accumulation procedure.
The soil profile consists of a homogeneous overconsolidated clay. The soil parameters result from the UDCAM-S pro-
cedure. The cyclic stress ratio in direct simple shear (τcy /τa ) is fixed at 1.5. The outcome of the strain accumulation
procedure has been fixed to 1, 5, 10, and 20. These results are shown in strain contour diagram for the Drammen clay
with OCR = 4 in figure 5.8. The outcome of the strain accumulation procedure of Neq = 1 is used as the reference soil
strength.
ocr = 4
1.2
0.8
τcy / SDSS
u
0.6
0.4
0.2
0
10 0 10 1 10 2 10 3
N
Figure 5.8: Fixed outcomes of the strain accumulation procedure to Neq = 1, 5, 10, and 20 for the Drammen Clay with OCR = 4
45
FOUNDATION
Resulting contour diagrams

The outcome of the strain accumulation procedure is thus an equivalent number of cycles. This number of cycles is
used to scale the contour diagrams for direct simple shear and triaxial compression and extension. The scaled contour
diagrams describe the stress-strain behaviour for the given clay after the specified number of cycles in the corresponding
laboratory tests.
In figure 5.9 - 5.12 the resulting direct simple shear and triaxial contour diagrams are shown for Neq = 1, 5, 10, and
20. It can be seen that in all four cases the idealized stress path reaches the failure contour close to 15 % cyclic shear
strain, which is defined by the uppermost semi-horizontal contour line. As discussed in chapter 4, the stress paths in
triaxial compression and triaxial extension are chosen such that the stress path intersects the failure contour at the same
combination of cyclic shear strains (γcy ) and average shear strains (γa ) as found in the direct simple shear contour diagram.
1 1
0.8 0.8
0.6 0.6
τ cy / SA
τ cy / SA
u
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
τa / SA
u
τ a / SA
u
(a) Direct Simple Shear (b) Triaxial
Figure 5.9: Resulting direct simple shear diagrams for N = 1 with idealized stress path inclinations of τcy /τa = 1.5 in DSS, for TXC and TXE respectively
1.72 and -17.0.
1 1
0.8 0.8
0.6 0.6
τ cy / SA
τ cy / SA
u
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
τa / SA
u
τ a / SA
u
1.62 and -15.5.
46
1 1
0.8 0.8
0.6 0.6
τ cy / SA
τ cy / SA
u
u
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
τ a / SA
u
τ a / SA
u
1.575 and -14.5
1 1
0.8 0.8
0.6 0.6
τ cy / SA
τ cy / SA
u
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
τa / SA
u
τa / SA
u
Figure 5.12: Resulting triaxial contour diagrams for N = 20 with idealized stress path inclinations of τcy /τa = 1.5 in DSS, for TXC and TXE respectively
1.49 and -15.
From the idealized stress paths the target stress-strain curves are derived, in the same way as shown in figure 4.7.
These curves are derived by interpolation of the intersected strain contours by the drawn stress path, and they describe the
degraded stress-strain behaviour in the three laboratory tests, given the equivalent number of cycles. In this case a total
stress analysis is performed, and thus the total stress-strain curves are obtained. The total stress-strain curve is a combi-
nation of the cyclic and average stress-strain curves, and thus these curves are obtained by summation of the average and
cyclic stresses and strains at each intersection of the strain contours and the drawn stress path.
In figure 5.14 the resulting target stress-strain curves are shown for the total analysis, derived from the stress paths in
figure 5.9 - 5.12. As mentioned in chapter 4, the definition of the strains differs between the three laboratory tests. The
strains in the triaxial test are defined in terms of the vertical strains (yy ), the strains in the direct simple shear test are
defined in terms of the shear strains (γxy ).
It can be seen that the difference in total shear stress at failure, differs significantly between the three laboratory tests.
This is caused mainly by the difference in average shear stress at failure. In figures 5.9 - 5.12 it can be seen that for each
equivalent number of cycles the stress path for direct simple shear and triaxial extension reaches the failure contour at
approximately the same cyclic shear stress ratio (τcy /SuA ) for all three laboratory tests. However, all three laboratory tests
differ on the average shear stress ratio (τa /SuA ) at which the stress path intersects the failure contour. It is thus mainly the
average stress-strain behaviour that causes the difference in total stress-strain behaviour for the direct simple shear test
and the triaxial extension test for OCR = 4.
47
FOUNDATION
150 150 150

N=1 N=1 N=1
N=5 N=5 N=5
N = 10 N = 10 N = 10
N = 20 N = 20 N = 20
| (σyy - σxx) / 2 | [kN/m 2 ]
(σyy - σxx) / 2 [kN/m 2 ]

100 100 100
τ xy [kN/m 2 ]
50 50 50
0 0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15
ǫyy [-] γxy [-] ǫyy [-]
(a) Triaxial compression (b) Direct simple shear (c) triaxial extension
Figure 5.13: Resulting target stress-strain curves for the idealized stress paths in figure 5.9 - 5.12
Resulting parameters
With the resulting target stress-strain curves from figure 5.13, the NGI-ADP parameters can be fitted to match the
degraded stress-strain behaviour in direct simple shear, triaxial compression, and triaxial extension. In accordance with
the method from Jostad et al. (2015), the failure strains (γf ) have been chosen equally for the behaviour in the three
laboratory tests. As a result, not all three target stress-strain curves can be fitted in the most optimal way. However, the
shear stress at failure can be fitted as best as possible for each individual laboratory test, which is dominant in this analysis.
This is in accordance with the Plaxis manual, as it states that "for pure capacity and stability (e.g. factor of safety) analyses
the values for shear strains at failure are not important and one may set all three values equal to e.g. 5% for simplicity"
(Brinkgreve et al., 2016). This differs for cases in which the deformations of the foundation are assessed, for which the
failure strains are of great importance.
150 150 150

N=1 N=1 N=1
N=5 N=5 N=5
N = 10 N = 10 N = 10
N = 20 N = 20 N = 20
| (σyy - σxx) / 2 | [kN/m 2 ]
(σyy - σxx) / 2 [kN/m 2 ]
100 100 100

τ xy [kN/m 2 ]
50 50 50
0 0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15
ǫyy [-] γxy [-] ǫyy [-]
Figure 5.14: Stress-strain curves from the fitted NGI-ADP parameters on the target stress-strain curves from figure 5.13
The stress-strain curves representing the fitted NGI-ADP parameters are shown in figure 5.14. The corresponding NGI-
ADP parameters of these stress-strain curves are shown in table 5.4. It can be seen that the undrained shear strength in
active failure (SuA ) decreases significantly with an increase of the equivalent number of cycles. For the equivalent number
of cycles of 5, 10, and 20, the undrained shear strengths in active failure are respectively 79.1%, 68.7%, and 63.1% of the
undrained shear strength after one cycle. The decrease in undrained shear strength in active failure for the soil after 20
cycles with respect to the soil after one cycle is thus almost 40%.
From table 5.4 it can also be seen that the shear strength anisotropy remains almost constant with an increase of the
equivalent number of cycles, which is in accordance with the Andersen (2015) paper. The shear strength ratio of the
undrained shear strength in triaxial extension to the undrained shear strength in active failure (SuE /SuA ) ranges from 0.55
to 0.58. The initial mobilisation is equal to zero, as the zero percent strain contours of the triaxial contour diagrams for
OCR = 4 intersect at a cyclic and average stress ratio of 0.0, and thus as well the starting point of the idealized stress path
in the triaxial compression and extension.
48
Table 5.4: Optimized NGI-ADP parameters for the stress-strain curves in figure 5.14
Neq G/SuA SuA SuE /SuA SuDSS /SuA γfT XC γfT XE γfDSS τ0 /SuA K0
[-] [-] [kPa] [-] [-] [%] [%] [%] [-] [-]
1 300 144.8 0.57 0.77 10.0 10.0 10.0 0.0 1.00
5 300 114.5 0.55 0.78 8.0 8.0 8.0 0.0 1.00
10 300 99.5 0.55 0.79 6.0 6.0 6.0 0.0 1.00
20 300 91.3 0.58 0.80 5.5 5.5 5.5 0.0 1.00
5.2.2 Influence of an increasing Neq for OCR = 1

In the first case, the influence of the equivalent number of cycles as a result of the strain accumulation procedure
on the stability of a gravity based foundation is investigated for the Drammen Clay with an overconsolidation ratio of 4.
The database of the UDCAM-S procedure consists of the Drammen Clay for different overconsolidation ratios. The same
influence is investigated in this case as in the previous case, but for the Drammen Clay with an overconsolidation ratio of
1, thus a normally consolidated soil.
The soil profile in this case consists of a homogeneous normally consolidated clay. The parameters of the soil result
from the UDCAM-S procedure. The cyclic stress ratio in direct simple shear (τcy /τa ) is fixed at 1.5.
The parameters of the NGI-ADP model have been derived with the UDCAM-S procedure. Equal to the first case, the
outcome of the strain accumulation procedure has been fixed to 1, 5, 10, and 20. These results for a normally consolidated
Drammen clay are shown in figure 5.15. The resulting soil after N = 1 is used as the reference soil strength.
ocr = 1
1.2
0.8
τcy / SDSS
u
0.6
0.4
0.2
0
10 0 10 1 10 2 10 3
N
Figure 5.15: Fixed outcomes of the strain accumulation procedure to Neq = 1, 5, 10, and 20 for OCR 1 of the Drammen Clay

With the fixed outcomes of the strain accumulation procedure, the scaled contour diagrams for direct simple shear and
triaxial compression and extension are shown in figure 5.16 - 5.19. The stress paths in triaxial compression and triaxial
extension are chosen in such a way that the stress path intersects the failure contours at the same combination of cyclic
shear strains (γcy ) and average shear strains (γa ) as found in the direct simple shear contour diagram.
The difference with the resulting contour diagrams for OCR = 4 in case 1 for the same equivalent number of cycles is
significant. Even though the inclination of the stress path in the direct simple shear is equal for all cases (corresponding to
the stress path in case 1), the resulting combination of cyclic and average shear strains is significantly different for OCR =
1 with respect to OCR = 4. For OCR = 4 the stress path with an inclination of τcy /τa = 1.5 in direct simple shear intersects
the failure contour almost at the intersection point between the failure contour of the average shear strains (representing
49
FOUNDATION
γa = 15%) and the failure contour of the cyclic shear strain (representing γcy = 15%). However, for OCR = 1 the stress
path with the same inclination in direct simple shear reaches the failure contour at a lower value of the cyclic shear strain.
For all scaled contour diagrams in figure 5.16 - 5.19, the stress path reaches the failure contour at approximately 1% cyclic
shear strain.
As the inclination of the stress path (τcy /τa ) follows from structural load data of a design storm, the same design storm
will result in the same inclination in direct simple shear for different soils. This means that the same design storm has
a greater effect on the accumulated cyclic shear strains in an overconsolidated soil with OCR = 4 than for a normally
consolidated soil with OCR = 1.
A significant difference with the stress paths in the triaxial contour diagrams for OCR = 1 in figure 5.16 - figure 5.16
with respect to the stress paths for OCR = 4 is the starting point of the stress path. This starting point is determined by the
initial mobilization (τ0 /SuA ), which is at the intersection point of zero percent cyclic shear strain contour and zero percent
average shear strain contour. For the normally consolidated Drammen Clay this initial mobilization is at an average shear
stress ratio (τ0 /SuA ) of 0.7617.
1 1
0.8 0.8
0.6 0.6
τ cy / SA
τ cy / SA
u
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
τa / SA
u
τ a / SA
u
(a) N = 1 (b) N = 1
Figure 5.16: Resulting direct simple shear diagrams for N = 1 with idealized stress path inclinations of τcy /τa = 1.5 in DSS, and TXC -4.2 TXE -0.35
1 1
0.8 0.8
0.6 0.6
τ cy / SA
τ cy / SA
u
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
τ a / SA
u
τ a / SA
u
50
1 1
0.8 0.8
0.6 0.6
τ cy / SA
τ cy / SA
u
u
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
τ a / SA
u
τ a / SA
u
(a) N = 10 (b) N = 10
1 1
0.8 0.8
0.6 0.6
τ cy / SA
τ cy / SA
u
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
τa / SA
u
τ a / SA
u
(a) N = 20 (b) N = 20
Figure 5.19: Resulting triaxial contour diagrams for N = 20 with idealized stress path inclinations of τcy /τa = 1.5 in DSS, and TXC -1.95 TXE -0.36.
In figure 5.20 the resulting target stress-strain curves are shown for the stress paths in figure 5.16 - figure 5.19. It can
be seen that the initial mobilisation (τ0 /SuA ) for the triaxial compression and extension stress paths results in a non-zero
starting point of the target stress strain curves in the corresponding laboratory tests. As the static undrained shear strength
in active failure is chosen as 100kPa, the initial mobilisation is 76.2kPa.
150 150 150

N=1 N=1 N=1
N=5 N=5 N=5
N = 10 N = 10 N = 10
N = 20 N = 20 100 N = 20
| (σyy - σxx) / 2 | [kN/m 2 ]
(σyy - σxx) / 2 [kN/m 2 ]
100 100
τ xy [kN/m2 ]
50
0
50 50
-50
0 0 -100
0 0.05 0.1 0.15 0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15
ǫyy [-] γxy [-] ǫyy [-]
Figure 5.20: Resulting target stress-strain curves for the idealized stress paths in figure 5.16 - 5.19
51
FOUNDATION
The resulting stress-strain curves from the corresponding fitted NGI-ADP parameters to the target stress-strain curves in
figure 5.20 are shown in figure 5.21. The corresponding parameters are shown in table 5.5. In this case it has been chosen
to optimize the failure strains in this case to the behaviour in triaxial compression. However as mentioned in section 5.2.1,
for this analysis it is important to optimize the undrained shear strength parameters of the NGI-ADP soil model.
150 150 150

N=1 N=1 N=1
N=5 N=5 N=5
N = 10 N = 10 N = 10
N = 20 N = 20 100 N = 20
| (σyy - σxx) / 2 | [kN/m 2 ]
(σyy - σxx) / 2 [kN/m 2 ]

100 100
τ xy [kN/m2 ]
50
0
50 50
-50
0 0 -100
0 0.05 0.1 0.15 0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15
ǫyy [-] γxy [-] ǫyy [-]
In table 5.5 it can be seen that the anisotropy of the undrained shear strength remains constant with an increasing
value for Neq . However, the decrease in undrained shear strength in active failure (SA
u ) is significant with an increase in
Neq . If Neq increases from 1 to 20, SA
u decreases by 26% (35 kPa).
[-] [-] [kPa] [-] [-] [%] [%] [%] [-] [-]
1 557 135 0.43 0.78 3.5 3.5 3.5 0.7617 0.492
5 518 115 0.43 0.74 3.0 3.0 3.0 0.7617 0.492
10 322 105 0.43 0.76 2.3 2.3 2.3 0.7617 0.492
20 322 100 0.40 0.75 2.0 2.0 2.0 0.7617 0.492
52
5.2.3 Influence of an increasing Neq for OCR = 1 assuming isotropic behaviour

In the UDCAM-S procedure it is possible to assume isotropic shear strength of the soil, if limited data is available for the
three laboratory tests. The stress-strain behaviour in the UDCAM-S procedure is then described with only the direct simple
shear test, the soil is then assumed to behave as the stress-strain behaviour in direct simple shear. The strength parameters
are derived for the direct simple shear test in the UDCAM-S procedure, and the parameters in triaxial compression and
triaxial extension are set equal to the parameters in direct simple shear.
To investigate the influence of this assumption, the same assumptions are made as the previous case, but in the param-
eter optimisation shear strength isotropy is assumed. Equal to case 2, the outcome of the strain accumulation procedure
have been fixed to the values of Neq of 1, 5, 10, and 20. The difference thus with case 2 is that the shear strength of the
soil is assumed isotropic, resulting in equal undrained shear strengths in active failure, extension and direct simple shear.
The parameters are derived by only using the contour diagrams of the direct simple shear test.

The direct simple shear contour diagrams are equal to the contour diagrams in case 1, which are shown in figure 5.16
- figure 5.19. As all parameters are derived from the direct simple shear contour diagrams, the triaxial contour diagrams
are not required to determine the degraded stress-strain response of the soil.
1 1
0.8 0.8
0.6 0.6
τ cy / SA
τ cy / SA
u
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
τa / SA
u
τa / SA
u
(a) N = 1 (b) N = 5
1 1
0.8 0.8
0.6 0.6
τ cy / SA
τ cy / SA
u
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
τ a / SA
u
τ a / SA
u
(c) N = 10 (d) N = 20
Figure 5.22: Resulting direct simple shear contour diagrams for OCR 1 with idealized stress path inclinations of τcy /τa = 1.5
The target stress-strain curves resulting from the stress-paths in direct simple shear in figure 5.22 are shown in figure
5.23. All parameters of the NGI-ADP soil model are fitted on these target stress-strain curves.
53
FOUNDATION
150
N=1
N=5
N = 10
N = 20
100
τ xy
50
0
0 0.05 0.1 0.15 0.2 0.25
γxy
Figure 5.23: Target stress-strain curves resulting from the stress paths in figure 5.22 assuming isotropic behaviour
The stress-strain curves for the optimized NGI-ADP parameters are shown in figure 5.24. The corresponding fitted
NGI-ADP parameters are shown in table 5.6. In this case the stress-strain behaviour of the target stress-strain curves can
be fitted reasonably well with the NGI-ADP soil model.
150
N=1
N=5
N = 10
N = 20
100
τ xy
50
0
0 0.05 0.1 0.15 0.2 0.25
γxy
Figure 5.24: Stress-strain curves from the fitted NGI-ADP parameters on the target stress-strain curves from figure 5.13 assuming isotropic behaviour
The optimized NGI-ADP parameters are shown in figure 5.6. As a result of the assumption of shear strength isotropy,
the shear strength ratios SuE /SuA and SuDSS /SuA are set to a value of 1. The undrained shear strength in active failure (SuA )
is a result of the fitted stress-strain curve in direct simple shear. Comparing the undrained shear strengths in active failure
between the isotropic and non-isotropic the difference is significant. The optimized value after Neq = 20 for SuA is 105
kPa assuming anisotropic shear strength, and 75 kPa assuming isotropic behaviour. However, the undrained shear strength
in extension after Neq = 20 is only 58.05 kPa, but 75 kPa assuming isotropic behaviour. It is thus not directly possible to
conclude that the behaviour of the soil assuming isotropic shear strength is stronger or weaker than the behaviour of the
soil assuming anisotropic shear strength.
54
[-] [-] [kPa] [-] [-] [%] [%] [%] [-] [-]
1 557 105.3 1.0 1.0 10.0 10.0 10.0 0.7617 0.492
5 740 87.4 1.0 1.0 8.0 8.0 8.0 0.7617 0.492
10 740 79.8 1.0 1.0 7.0 7.0 7.0 0.7617 0.492
20 740 75 1.0 1.0 6.0 6.0 6.0 0.7617 0.492
5.2.4 Influence of the cyclic load ratio for OCR = 4

The cyclic load ratio (Fcy,max /Fa ) has been found as an important aspect in the degradation of soil strength and
stiffness. The cyclic load ratio is assumed to be equal to the cyclic stress ratio (τcy,max /τa ). The cyclic stress ratio is a
result of the design storm which is taken into account in the design of the foundation, and used in the determination of
the idealized stress path in the second step of the UDCAM-S procedure. The ratio τcy,max /τa is determined by the largest
cyclic load in the design storm (Jostad et al., 2015). To investigate the effect of this stress ratio on the ultimate combined
loads of a gravity based foundation, five stress ratio’s have been used for a soil consisting of the Drammen Clay with OCR
4. The outcome of the strain accumulation procedure has been fixed to Neq = 5. The parameters of the NGI-ADP model
have been derived with the UDCAM-S procedure for the overconsolidated Drammen Clay.
1 1
τ cy/τ a = 1.0 τ cy/τ a = 1.0
τ cy/τ a = 1.5 τ cy/τ a = 1.5
0.8 0.8
τ cy/τ a = 2.0 τ cy/τ a = 2.0
τ cy/τ a = 2.5 τ cy/τ a = 2.5
τ cy/τ a = 3.0 τ cy/τ a = 3.0
0.6 0.6
τ cy / SA
τ cy / SA
u
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
τa / SA
u
τ a / SA
u
(a) Direct simple shear contour (b) Triaxial contour
Figure 5.25: Contour diagrams with the idealized stress paths for OCR 4 with Neq = 5 and varying cyclic stress ratio’s
In figure 5.26 the resulting target stress strain curves are shown for the stress paths in figure 5.25. It can be seen
that the inclination of the stress path has an influence on the stress-strain curve in direct simple shear (figure 5.26(b)).
However, for triaxial extension, the difference between the stress-strain curves is almost negligible, as shown in figure
5.26(c). This effect can be explained with 5.25(b), in which the stress paths are drawn for triaxial compression and tri-
axial extension for the varying cyclic stress ratios. The triaxial extension stress paths are placed noticeably close to each
other. This is due to the fact that the average strain contours in triaxial extension are narrowly spaced with respect to
the same average strain contours in direct simple shear. Therefore the difference in the resulting stress-strain curves from
the different strain combination is negligible. The same effect can be seen in stress-strain curve for triaxial compression
(figure 5.26(c)), although for triaxial compression there is a slight difference between the resulting stress strain curves, as
the average strain contours in triaxial compression are further apart than the same strain contours in triaxial extension.
55
FOUNDATION
150 150 150

τ cy/τ a = 1.0 τ cy/τ a = 1.0 τ cy/τ a = 1.0
| (σyy - σxx) / 2 | [kN/m 2 ]
(σyy - σxx) / 2 [kN/m 2 ]

100 100 100
τ xy [kN/m2 ]
50 50 50
0 0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15
ǫyy [-] γxy [-] ǫyy [-]
Figure 5.26: Resulting target stress-strain curves for the idealized stress paths in figure 5.25
The NGI-ADP parameters have been fitted to the target stress-strain curves in figure 5.26 as shown in figure 5.27.
The corresponding parameters are shown in table 5.7. As mentioned in the previous section, the difference in strength
parameters is limited, compared to the difference in inclinations of the stress path in direct simple shear. An increase of the
inclination of the stress path from 1.0 to 3.0, which is a significant difference in the cyclic stress ratio, leads to a decrease
of 4.5% in the undrained shear strength in active failure.
150 150 150

τ /τ = 2.0 τ /τ = 2.0 τ /τ = 2.0
cy a cy a cy a
| (σyy - σxx) / 2 | [kN/m 2 ]
(σyy - σxx) / 2 [kN/m 2 ]

100 100 100
τ xy [kN/m 2 ]
50 50 50
0 0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15
ǫyy [-] γxy [-] ǫyy [-]
From the resulting parameters shown in table 5.7 it can be seen that the ratio of the undrained shear strength in
extension to the undrained shear strength in active failure remains constant with an increasing cyclic load ratio. However,
the ratio of the undrained shear strength in direct simple shear to the undrained shear strength in active failure decreases
with an increasing cyclic load ratio.
τcy /τa G/SuA SuA SuE /SuA SuDSS /SuA γfT XC γfT XE γfDSS τ0 /SuA K0
[-] [-] [kPa] [-] [-] [%] [%] [%] [-] [-]
1.00 300 110 0.57 0.77 6.0 6.0 6.0 0.0 1.0
1.50 300 110 0.59 0.77 8.0 8.0 8.0 0.0 1.0
2.00 300 108 0.58 0.74 7.0 7.0 7.0 0.0 1.0
2.50 322 108 0.57 0.74 7.0 7.0 7.0 0.0 1.0
3.00 322 105 0.58 0.71 6.0 6.0 6.0 0.0 1.0
56
5.3. RESULTING VHM FAILURE DIAGRAMS
5.3 Resulting VHM failure diagrams
In this section the results from the different cases in the previous section are presented in terms of VHM failure
diagrams, which are explained in chapter 2. In all cases a strip foundation is analysed, with a foundation width of 20m.
5.3.1 VHM diagrams for OCR = 4

The analysis used to investigate the influence of an increasing Neq on the stability of a gravity based foundation, the
aforementioned method is used. At first, the vertical load is applied, representing the self-weight of the structure. There-
after, the horizontal load and overturning moment are applied to the foundation and increased up to failure. This analysis
has been done at vertical loads of 0.3, 0.5, and 0.7 times the ultimate vertical load of the soil after N = 1. The results of
these analyses are shown in figure 5.28.
The results in figure 5.28 are shown in the HM failure diagrams, with the normalised horizontal failure loads on the
horizontal axes and the normalised overturning moment on the vertical axes. These values are all normalised to the mean
average undrained shear strength for the soil after Neq = 1.
0.7 0.7
0.6 0.6
0.5 0.5
M/B2 S u
0.4
M/B2 S u
0.4
0.3 0.3
0.2 0.2
V = 0.3 Vult (N = 1)
V = 0.5 Vult (N = 1)
0.1 0.1
V = 0.7 Vult (N = 1)
0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
H/BS u H/BS u
(a) N = 1 (b) N = 5
0.7 0.7
0.6 0.6
0.5 0.5
M/B2 S u
0.4
M/B2 S u
0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
H/BS u H/BS u
(c) N = 10 (d) N = 20
Figure 5.28: HM failure diagrams for the parameter sets shown in table 5.4 normalised to the undrained shear strength for the soil after N = 1 with applied
vertical loads of 3482 kN/m (0.3 Vult,N =1 ), 5803 kN/m (0.5 Vult,N =1 ), and 8124 kN/m (0.7 Vult,N =1 ).
The red lines represent the limit loads at a vertical load of 0.3 times the ultimate vertical load for the soil after Neq =
1, the blue and yellow lines are respectively for vertical loads of 0.5 and 0.7 times the ultimate vertical load for the soil
after Neq = 1. It can be seen that the strength of the soil after Neq = 10 and Neq = 20 is eminently degraded, and the
foundation fails at a vertical load of 0.7 times the ultimate vertical load for the soil at Neq = 1: the yellow line in figures
5.28(a) en 5.28(b) is not present in figures 5.28(c) en 5.28(d).
57
FOUNDATION
The skewed shape of the failure envelope in the HM diagram corresponds with the shape found in literature, as dis-
cussed in chapter 2. The largest moment capacity is mobilised when a relative small horizontal load is applied to the
foundation acting in the same direction as the overturning moment.
The same results can also be visualized in terms of VM failure diagrams, which is shown in figure 5.29. The horizontal
axis represents the normalised vertical load, the vertical axis represents the normalised overturning moment. Equal to
figure 5.28, these loads are normalised to the mean undrained shear strength after Neq = 1 (Su,N =1 ).
With this figure it can be explained why the yellow line in figure 5.28 (representing 0.7 times Vult,N =1 ) is not present
for the soil after Neq = 10 and Neq = 20. Seventy percent of the ultimate limit load of the soil after Neq = 1 results in a
normalised vertical load of 3.65 (V/BSu,N =1 ), which is a larger vertical load than the vertical limit load of the soil after
Neq = 10, which is found at a normalised vertical load of 3.58 (V/BSu,N =1 ). The soil thus already fails with only the pure
vertical load applied to the foundation.
A noticeable result shown in figure 5.29, is the small difference between the limit overturning moment at small vertical
loads. All failure envelopes have the same shape in the VM space, but the maximum resistance against the overturning
moment is at different vertical loads. The maximum overturning moment after Neq = 10 is only 68.6% of the maximum
overturning moment after Neq = 1. For Neq = 20 the maximum overturning moment is decreased to 64.1% of the maxi-
mum overturning moment after Neq = 1. These maximum overturning moments are mobilised at a vertical load of 1.77
and 1.65 (in terms of V/BSu,N =1 ) respectively for the soil after Neq = 10 and Neq = 20.
0.7
N =1
0.6 eq
Neq = 5
0.5 Neq = 10
M/B2 S u,N=1
0.4 Neq = 20
0.3
0.2
0.1
0
0 1 2 3 4 5 6
V/BS u,N=1
Figure 5.29: VM diagram for H = 0 for the Neq values of 1, 5, 10, and 20. The lines are plotted by modifying the equation of Houlsby and Purzin (1999)
For the influence of the cyclic load ratio on the resulting HM diagram, the results from the parameters in table 5.7
are shown in figure 5.30. The results from τcy /τa = 1.0 are not visible, as they coincide with the results from τcy /τa =
1.5. It can be seen that the influence of cyclic load ratio is small compared to the equivalent number of cycles for the
overconsolidated Drammen Clay.
0.7
0.6
0.5
M/B2 S u
0.4
0.3
τ cy/τ a = 1.0
τ /τ = 1.5
0.2 cy a
τ cy/τ a = 2.0
τ cy/τ a = 2.5
0.1
τ cy/τ a = 3.0
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
H/BS u
Figure 5.30: Influence of the cyclic stress ratio on the HM capacity diagram at a vertical load V = 0.5 Vult = 4449 kN/m
58
5.3.2 VHM diagrams for OCR = 1

The same calculations have been performed for the normally consolidated Drammen Clay (OCR = 1). The results of
these calculations are shown in figure 5.31 and 5.32, assuming anisotropic and isotropic undrained shear strength of the
soil. In figures 5.31(c) and 5.31(d) it can be seen that, in contrast to the overconsolidated Drammen Clay (figure 5.28),
the soil after Neq = 10 and Neq = 10 does not fail at a vertical load of 0.7 Vult,N =1 . The degradation of the strength for
normally consolidated Drammen Clay is thus less influenced by the equivalent number of cycles than the overconsolidated
Drammen Clay with OCR = 4.
0.7 0.7
0.6 0.6
0.5 0.5
M/B2 S u
0.4
M/B2 S u
0.4
0.3 0.3
0.2 0.2
V = 0.3 Vult (N = 1)
V = 0.5 Vult (N = 1)
0.1 0.1
V = 0.7 Vult (N = 1)
0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
H/BS u H/BS u
(a) N = 1 (b) N = 5
0.7 0.7
0.6 0.6
0.5 0.5
M/B2 S u
0.4
M/B2 S u
0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
H/BS u H/BS u
(c) N = 10 (d) N = 20
Figure 5.31: HM diagrams for the parameter sets shown in table 5.5 with applied vertical loads of 3068 kN/m (0.3 Vult,N =1 ), 5113 kN/m (0.5 Vult,N =1 ),
and 7159 kN/m (0.7 Vult,N =1 ), normalised to the undrained shear strength for the soil after Neq = 1.
The HM diagrams for the parameters for the normally consolidated Drammen Clay assuming isotropic shear strength,
are shown in figure 5.32. The diagrams show similar results compared to the anisotropic shear strength. In figure 5.32(d)
an odd result is shown, for a vertical load of 0.7 times the ultimate limit load. After the applied vertical load, a numerical
instability results in failure at a combination of the horizontal load and the overturning moment.
59
FOUNDATION
0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
M/B2 S u
M/B2 S u
0.4 0.4
0.3 0.3
0.2 V = 0.3 Vult (N = 1) 0.2

V = 0.5 Vult (N = 1)
0.1 0.1
V = 0.7 Vult (N = 1)
0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
H/BS u H/BS u
(a) N = 1 (b) N = 5
0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
M/B2 S u
M/B2 S u
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
H/BS u H/BS u
(c) N = 10 (d) N = 20
Figure 5.32: HM diagrams for the parameter sets shown in table 5.6 with applied vertical loads of 3068 kN/m (0.3 Vult,N =1 ), 5113 kN/m (0.5 Vult,N =1 ),
and 7159 kN/m (0.7 Vult,N =1 ), normalised to the undrained shear strength for the soil after N = 1 from case 2 (non-isotropic).
The VHM failure envelopes comparing the normally consolidated Drammen Clay with isotropic and isotropic shear
strength are shown in figure 5.33 and figures D.5, D.6, and D.7.
In figure 5.33 the effect of the equivalent number of cycles is clear on the ultimate limit loads of the foundation. At a
vertical load of 30% of the ultimate vertical load for the soil after Neq = 1, the moment capacity of the soil after Neq = 10
is at more than 90% of the moment capacity of the soil after one cycle. At a vertical load of 70% of the ultimate vertical
load, the moment capacity of the soil after Neq = 10 is drastically decreased with respect to the soil after one cycle, as it
has only 50% of the original moment capacity.
60
0.8
0.7
0.6
0.5
M/B2 S u
0.4
0.3
0.2
0.1
1.5
1
0 0.5
0.3 0.35 0
0.4 0.45 -0.5
0.5 0.55 -1
0.6 0.65 -1.5
0.7
H/BS u
V/V ult
Figure 5.33: Normalised ultimate limit loads in 3D space for Neq = 1 (continuous line) and Neq = 10 (dotted line) for both anisotropic (blue lines) an
isotropic shear strengths (red lines) for normally consolidated Drammen Clay
61
FOUNDATION
5.4 The influence of the foundation width
The loads that have to be carried by the foundation, result from the structural load analysis of the wind-turbine. The
foundation slab consists of many steel rebars and cast concrete to transfer the loads from the structure to the soil. In case
of insufficient stability of the foundation, the width of the foundation can be adjusted to increase stability. Figure 5.34
shows three different widths of the foundation slab for the same structure.
(a) (b) (c)
Figure 5.34: Differences in width of the foundation slab with an identical superstructure
In this analysis the influence of the width of the foundation on the stability of the foundation with respect to the cyclic
degradation for the soil is investigated. The loads shown in table 5.8 have been chosen as the load combination on the
foundation in this analysis. These loads are used as the total load on the foundation for every investigated width of the
foundation slab. The foundation width has been varied between 12m and 24m. In reality it may not be possible to change
the width on such a large scale, but for this analysis it has been chosen to broaden the effect of a different foundation
width. The foundation is modelled, as described in the beginning of this chapter, as a very stiff soil, resulting in negligible
deformations in the foundation.
Table 5.8: Applied loads on the foundation
Vertical load V 70 000 kN

Horizontal load H 10 000 kN
Overturning moment M 300 000 kNm
The loads shown in table 5.8 are for a complete 3D foundation. The analysis is performed in 2D finite element
calculations, resulting in different loads on the strip foundation. As the loads from the structure remain constant, the load
per square metre decreases with a wider foundation. Equivalent cylindrical diameters for a 2D foundation can be calculated
following the codes from DNV-GL (2016) and API (2011). The loads from table 5.8 are scaled for the corresponding widths
to apply the correct loads in the 2D analysis. Table D.1 shows the resulting loads applied in the 2D calculations. The 3D
effect has been neglected in this analysis. However, this must be kept in mind for the practical application of the UDCAM-S
procedure.
5.4.1 Influence of the foundation width for overconsolidated Drammen Clay

The parameter set that has been used is derived in section 5.2.1 and shown in table 5.4. The used parameters are the
resulting soil parameters for the overconsolidated Drammen Clay with OCR = 4 after Neq 1, 5, and 10, and a cyclic stress
ratio (τcy /τa ) of 1.5 in direct simple shear.
Figure 5.35 shows the difference in total deformation of the soil underneath the foundation for the loads shown in
table 5.8. Figure 5.35(d) shows the legend of the displacement contours. It can be seen that for a foundation width of 16m
62
5.4. THE INFLUENCE OF THE FOUNDATION WIDTH
the deformation is significantly larger than the deformation for a foundation width of 24m. The maximum deformation
of the soil for a 16m wide foundation is 170mm, as for a 20m and 24m wide foundation the maximum deformation is
respectively 70mm and 50mm.
(a) Foundation width of 16m
(b) Foundation width of 20m
(c) Foundation width of 24m (d) Legend
Figure 5.35: Total deformations for three different foundations widths with equally applied total loads
63
FOUNDATION
The results from the applied loads in table D.1 are shown in figure 5.36 in terms of the overall safety factor found in
Plaxis (Msf). It can be seen that the foundation width has a significant impact on the overall safety factor of the foundation.
As found in section 5.2.1, the influence of the equivalent number of cycles is significant for the stability of the foundation.
From figure 5.36 it can be concluded that the beneficial effect of an increment of the foundation width at larger widths
decreases for larger values of Neq . This results from the difference in inclination of the lines.
4.5
4
Overall safety factor Msf [-]
3.5
2.5
2
Neq = 1
Neq = 5
1.5
Neq = 10
1
12 14 16 18 20 22 24 26
Width of the foundation slab [m]
Figure 5.36: Resulting overall safety factors for Neq of 1, 5, and 10 for OCR = 4
In figure 5.37 the results from figure 5.36 are plotted in terms of the overall safety factor found for the soil after Neq =
1. It can be seen that the ratio between the overall safety factors remains constant with an increasing foundation width.
The factor of safety for the soil after Neq = 5 is at 79% of the overall safety factor for the soil after Neq = 1, for Neq =
10 this value has dropped to 69%. The ratio of the overall safety factor thus remains constant, but the absolute difference
between the lines increases for larger foundation widths.
1.2
1
Relative overall safety factor [-]
0.8
0.6
0.4
N =1
eq
0.2 Neq = 5
N = 10
eq
0
12 14 16 18 20 22 24
Figure 5.37: Relative overall safety factor with respect to the soil strength after Neq = 1 from figure 5.36
The same analysis has been performed for the parameters in table 5.7 with the resulting parameters from τcy /τa = 1.5
and τcy /τa = 3.0. The results are shown in figure 5.38. The difference in overall safety factors between the two cyclic
stress ratios is small, compared to the influence of the equivalent number of cycles in figure 5.36, the same result as found
in section 5.3.1.
The difference between the overall factors of safety at a width of 16m for the two cyclic stress ratios is 0.15, at a width
of 24m this difference is increased to 0.35. The difference in figure 5.36 between Neq = 1 and Neq = 5 at foundation
64
5.4. THE INFLUENCE OF THE FOUNDATION WIDTH
widths of 16m and 24m is respectively 0.39 and 0.86. The influence of an increase in the equivalent number of cycles at
large foundation widths is thus larger than an increment in the cyclic stress ratio.
4.5
4
3.5
2.5
τ cy/τ a = 1.5 (N eq = 5)
1.5
τ cy/τ a = 3.0 (N eq = 5)
1
12 14 16 18 20 22 24 26
Figure 5.38: Influence of the cyclic stress ratio (τcy /τa ) for OCR = 4 after Neq = 5
5.4.2 Influence of the foundation width for OCR = 1

Equal to the previous analysis for the overconsolidated Drammen Clay, the influence of cyclic degradation with respect
to the foundation width for the normally consolidated Drammen Clay is analysed. The results for the influence of the
foundation width for OCR = 1 are shown in figure 5.39 and figure 5.40.
In figure 5.39 the results are shown for the influence of the foundation width on the overall factor of safety with respect
to the equivalent number of cycles for normally consolidated Drammen Clay. The parameters that are used, are shown in
table 5.5 for the equivalent number of cycles of 1 and 5. The results from the overconsolidated Drammen Clay are shown
to compare the difference in obtained factor of safety.
It can be seen that the influence of the equivalent number of cycles for the normally consolidated Drammen Clay is
small compared to the overconsolidated Drammen Clay. The normally consolidated Drammen Clay shows a decrease of
16% in the overall factor of safety after 5 number of cycles, the overconsolidated Drammen Clay a decrease of 21%. In
both cases the decrease of the overall safety factor is constant for different foundation widths.
4.5
4
3.5
2.5
2 OCR = 4 (N = 1)
eq
OCR = 4 (N = 5)
eq
OCR = 1 (N = 1)
1.5 eq
OCR = 1 (N eq = 5)
1
12 14 16 18 20 22 24 26
Figure 5.39: Difference in overall safety factor between the overconsolidation ratios of the Drammen clay at different widths
65
FOUNDATION
For the assumption of isotropic shear strength in the UDCAM-S procedure, the parameters have been found to be
slightly overestimating the bearing capacity of a gravity based foundation compared to the anisotropic shear strength pa-
rameters, as shown in figure 5.33.
In figure 5.40 the results are shown for the influence of the foundation width on the overall factor of safety with
respect to the (an)isotropic shear strength for normally consolidated Drammen Clay. The used parameters are shown in
table 5.5 and 5.7. It can be seen that the difference in overall factor of safety between the isotropic and anisotropic shear
strength remains almost constant for different widths of the foundation. For the equivalent number of cycles of 1 and 5, an
overestimation of 5% is found by assuming shear strength isotropy. The assumption of isotropic shear strength thus gives
a good indication on the stability of the foundation.
4.5
4
3.5
2.5
2 non-isotropic (Neq = 1)
isotropic (Neq = 1)
non-isotropic (Neq = 5)
1.5
isotropic (Neq = 5)
1
12 14 16 18 20 22 24 26
Figure 5.40: Difference in overall factor of safety for isotropic and non-isotropic undrained shear strength for OCR = 1 for different widths
66
6.
Discussion
This section discusses the methods and results found in this research, related to the research question of this thesis.
Comments on the found results are drawn up before the conclusions are drawn in chapter 7.
• The study of the influence of the different steps of the UDCAM-S procedure on the stability of a gravity based
foundation in chapter 5, one step of the UDCAM-S procedure was varied, while the others remained constant. As all
the steps in the UDCAM-S procedure are the result of actual storm data, these steps are expected to be correlated. It
was expected that the cyclic load ratio Fcy /Fa is correlated to the equivalent number of cycles as the outcome of the
strain accumulation procedure. The results found in chapter 5 are thus intended to see the sensitivity of the different
steps in the procedure, and do not represent actual data from a design storm.
• The result of the UDCAM-S procedure is based on the optimisation of the NGI-ADP parameters to the target stress-
strain curves. As shown for the case study in the paper of Jostad et al. (2015), the results depend on the decisions
made in the parameter optimisation. The parameter optimisation in the different cases was based on decisions made
in this thesis. Other results could be obtained when the optimisation of the parameters is performed differently.
• The results in figure 5.40 comparing isotropic and anisotropic shear strength for a normally consolidated Drammen
Clay are based on a 2D finite element analysis. By assuming isotropic shear strength, the result can be significantly
different if the problem is assessed in a 3D finite element analysis, as the contribution of the different undrained
shear strengths is different in a 3D failure surface compared to a 2D failure surface.
• This study is based on the available data from extensive laboratory testing on the Norwegian Drammen Clay at
different overconsolidation ratios by the NGI. This data forms the database of the UDCAM-S procedure. All results
found in this research are related to this type of clay. There is no guarantee that identical results follow from equal
analyses on different type of clays.
• In the finite element analysis only ultimate limit state analyses are performed in this study. The stiffness parameter
and the failure strains from the NGI-ADP model are not required in this type of analysis (Brinkgreve et al., 2016).
For analyses on the deformation of the foundation, the stiffness parameter and the failure strains are required. The
conclusions in this research are thus only viable regarding the undrained shear strength parameters as a result of the
NGI-ADP model, and not the stiffness parameter.
• All finite element calculations are performed in 2D plane strain conditions. However, it is known that for the design
of gravity based foundations significant differences arise between 2D and 3D modelling (Taiebat and Carter, 2002).
The modelling in this study is only intended to obtain insight of the influence of cyclic degradation on the stability
of gravity based foundations. The outcomes in terms of safety factors is thus only intended for this purpose, and not
meant to draw conclusions regarding a specific project.
• For the influence of the width of foundation slab on the stability of the gravity based foundation it is assumed that
the stiffness of the concrete is not of influence on the distribution of the loads. The result is that the loads from
the structure are distributed equally over the foundation slab. In reality the loads will not be distributed completely
homogeneous over the width of the foundation. There will be a limit for which an increment of the foundation slab
width will actually lead to a an increment of the overall factor of safety. This is however left outside of the scope of
this analysis, and thus not incorporated in this study.
67
7.
Conclusions & Recommendations
This thesis presents the modelling of cyclic soil degradation for a gravity based foundation by means of the Simplified
UnDrained Cyclic Accumulation Model. The implementation of this model has been developed and tested during this
research. Afterwards it has been applied to an offshore gravity based foundation. The main objective of this research is
formulated with the main research question as
What is the effect of taking the degradation of soil strength and stiffness as a result of cyclic
loading into account on the stability of a gravity based foundation?
To answer this research question, sub-questions have been formulated. The sub-questions are answered in this section,
afterwards the main conclusions of this study are presented. Additionally, recommendations for further work are presented
following from findings in this study.
7.1 Conclusions
7.1.1 Answers to the formulated sub-questions

• What design codes are applicable for an offshore gravity based foundation and how do these codes deal with cyclic soil degra-
dation?
The codes by DNV-GL (2016), API (2011), and API (2002) are commonly used codes for the design of offshore support
structures. However, the available design codes fall short in quantifying the degradation of soil strength and stiffness
due to cyclic loading for gravity based foundations. The codes give normalised bearing capacity formula, which are to be
applied in static conditions. In all three codes it is stated that the influence of cyclic loading must be taken into account
in the design of offshore foundations, but no quantification of this degradation is mentioned. These codes thus fall short
in guidance on how to take the effect of cyclic degradation into account in the design of offshore gravity based foundations.
• What is the behaviour of soils subjected to cyclic loading and how is the degradation of soil strength and stiffness influenced?
The cyclic behaviour of clays is significantly different than the static behaviour. The most important aspect of the cyclic
behaviour is the accumulation of pore pressures and shear strains in the soil. This is initiated by the cyclic repetition of
the shear stresses and the restriction of the drainage of pore pressures. The ratio of the average shear stress and the cyclic
shear stress is found to be an important parameter for the generation of cyclic and average shear strains. This ratio is
defined by the ratio between the cyclic and average loads as a result of the structural analysis (Jostad et al., 2015). A
higher ratio of the cyclic shear stress to the average shear stress, results in predominantly cyclic shear strains, a lower ratio
results in predominantly average shear strains. The accumulation of pore pressures and shear strains can result in lower
shear stresses and shear strains at failure with respect to static loading of the soil, as both the stiffness and the strength of
the soil is affected in a negative way. This phenomenon is the cyclic degradation of the soil.
• How does the UDCAM-S procedure analyse the cyclic degradation of the strength and stiffness of the soil?
The UDCAM-S procedure is developed to take the effect of cyclic degradation of the soil into account. The accumulation
of cyclic shear strains due to a design storm is determined by the strain accumulation procedure (Andersen, 1976). The
input of this procedure is a set of load parcels, which is a simplification of a design storm by counting and arranging the
cyclic load amplitudes. The result of this procedure is an equivalent number of cycles at a certain stress ratio in direct
simple shear, which gives the same result in terms of the accumulated shear strains as the load parcels from the simplified
69
CHAPTER 7. CONCLUSIONS & RECOMMENDATIONS
design storm. The equivalent number of cycles is then used to scale the contour diagrams to describe the behaviour of the
clay in direct simple shear, triaxial compression, and triaxial extension given the number of cycles. With the ratio of the
cyclic shear stress to the average shear stress in the soil a stress path is defined, resulting in degraded stress-strain curves
for direct simple shear and triaxial compression and extension. On the degraded stress-strain curves NGI-ADP parameters
are fitted, resulting in a parameter set accounting for the cyclic degradation and the shear strength anisotropy. The imple-
mentation of the model has been verified with a case study by Jostad et al. (2015), and shows promising results for the
assessment of cyclic degradation.
• What actions have to be performed to implement the UDCAM-S procedure in a real project?
This has not been shown in this report, but it has been found that the triaxial and direct simple shear contour diagrams
can effectively be scaled along the vertical axes to match cyclic laboratory tests. However, the strain contour diagram
in the strain accumulation procedure cannot be scaled to fit cyclic direct simple shear tests. For the strain accumulation
procedure one is thus restricted to the strain contour diagrams from the Drammen Clay. It is possible to construct strain
contours for a specific type of clay. However, several direct simple shear tests at zero average shear stress must be carried
out at different shear stress ratios to obtain sufficient data points to construct a strain contour diagram for a specific type
of clay.
• What is the sensitivity of the steps in the UDCAM-S procedure regarding the resulting degraded NGI-ADP parameters in
Plaxis?
It has been found that the result of the strain accumulation procedure affects the result of the UDCAM-S procedure in
terms of the resulting NGI-ADP parameters the most. As the equivalent number of cycles is determined by the design
storm simplified in terms of load parcels, it is important that the load parcels are determined correctly. This is however
depending on the counting procedure of the cyclic loads, as concluded by Norén-Cosgriff et al. (2015). The influence
of the cyclic stress ratio determining the idealized stress path on the resulting NGI-ADP parameters is significantly less
than the influence of the equivalent number of cycles. The optimisation of the parameters can also have a significant im-
pact on the result of the procedure, depending on the decisions made in the optimisation regarding the strain compatibility.
• What is the effect of degradation in strength and stiffness of the soil on the stability of a gravity based foundation?
The stability of gravity based foundations can be significantly affected by cyclic degradation of the soil. From the different
case studies it has been found that the outcome of the strain accumulation procedure is the most influencing factor on
the outcome of the UDCAM-S procedure, and thus on the influence on the stability of a gravity based foundation. The
undrained shear strength parameters of the NGI-ADP model are most affected by an increase of the resulting equivalent
number of cycles of a design storm. The influence of the inclination of the stress path has been investigated, which is
a result of the ratio between the cyclic load and the average load in the peak of the design storm. The influence of this
inclination has been found significantly smaller than the influence of the equivalent number of cycles. This is also shown in
the resulting overall factor of safety in figure 7.1, in which the results are shown for the overall safety factors for different
widths for Drammen Clay with OCR = 4.
4.5 4.5
4 4
3.5 3.5
3 3
2.5 2.5
2 2
Neq = 1
Neq = 5 τ cy/τ a = 1.5 (N eq = 5)
1.5 1.5
Neq = 10 τ cy/τ a = 3.0 (N eq = 5)
1 1
12 14 16 18 20 22 24 26 12 14 16 18 20 22 24 26
Width of the foundation slab [m] Width of the foundation slab [m]
(a) Influence of the equivalent number of cycles on the overall factor of (b) Influence of the cyclic stress ratio on the overall factor of safety of
safety of the strip foundation the strip foundation
Figure 7.1: Results for the overall safety factor for different widths of the strip foundation for a homogeneous soil of OCR = 4
70
7.1. CONCLUSIONS
7.1.2 Main conclusions of this research

With the UDCAM-S procedure the influence of the cyclic degradation on the stability of a gravity based foundation is
investigated. It has been found that the degradation of soil strength and stiffness can significantly influence the stability of
a gravity based foundation. From this study the following conclusions are drawn:
• The UDCAM-S procedure is a promising model to asses the influence of cyclic loading on undrained clay. Without
time consuming finite element codes, the degradation of the soil can be assessed by determining degraded soil
parameters for the NGI-ADP soil model in Plaxis.
When it is implemented in the Plaxis software, the UDCAM-S procedure is a clear and understandable user-friendly
procedure. A consequence is that every user of the Plaxis software can use this model, without understanding the
theory of cyclic soil behaviour. This is however also a general problem for finite element software.
• The equivalent number of cycles (Neq ) as a result of the strain accumulation procedure is the dominant factor
determining the degradation of the soil strength and stiffness of the soil. A change of the equivalent number of
cycles can have a significant effect on the stability of a gravity based foundation. It has also been found that with an
increase of the width of the foundation, the beneficial effect in terms of the overall safety factor of the foundations
decreases for larger equivalent number of cycles.
• For the same set of load parcels resulting from the simplification of a design storm, the equivalent number of cycles
as a result of the strain accumulation procedure increases with an increasing overconsolidation ratio. However, the
corresponding cyclic shear stress ratio decreases for larger overconsolidation ratios.
• In accordance with literature the maximum moment resistance of a strip foundation is found at a vertical load
approximately at half of the ultimate vertical load (V /BSu ≈ 2.6). As the undrained shear strength decreases with
an increase of the equivalent number of cycles, the maximum overturning moment is mobilised at an absolute lower
vertical load. This results in a significant decrease in moment capacity for gravity based foundation at higher vertical
loads due to large equivalent number of cycles. This effect is shown in figure 5.29. The effect on the moment capacity
for lower vertical loads is almost negligible.
0.7
Neq = 1
0.6
Neq = 5
0.5 Neq = 10
M/B2 S u,N=1
0.4 Neq = 20
0.3
0.2
0.1
0
0 1 2 3 4 5 6
V/BS u,N=1
Figure 7.2: VM diagram for H = 0 for the Neq values of 1, 5, 10, and 20. The lines are plotted by modifying the equation of Houlsby and Purzin (1999)
• The optimisation of the degraded stiffness of the soil based on the degraded stress-strain curves is difficult. The effect
of the stiffness parameter in the NGI-ADP model (G/SuA ) on the curvature of the stress-strain curves in direct simple
shear, triaxial extension, and triaxial compression is almost negligible, resulting in a large range of possible values
in the optimisation of the stiffness parameter. For calculations on the integrity of the foundation this is no problem,
but for calculations regarding the serviceability of the foundation, the effect can be disastrous for the developed
deformations.
71
CHAPTER 7. CONCLUSIONS & RECOMMENDATIONS
7.2 Recommendations for further work
This study focusses on the development of the UDCAM-S procedure and an application to a gravity based foundation.
The results from the model are promising. However, there are many more items to investigate with the modelling of cyclic
soil degradation by means of the UDCAM-S procedure. Recommendations for further work are presented as:
• In this research, the soil has been assumed homogeneous in all cases. In addition, the influence of a storm on
the cyclic soil behaviour is assumed homogeneous over depth for gravity based foundations. However, as gravity
based foundations are shallow foundations, the degradation of soil strength and stiffness will most probably not be
homogeneous over depth. The complex soil model from Jostad et al. (2014) is able to capture the degradation in
each integration point of the soil, the UDCAM-S procedure on the other hand only gives one resulting parameter
set. If this model is to be applied for foundations installed in the soil (e.g. monopiles), the effect of the cyclic soil
behaviour over depth must be known. The non-homogeneous character of the soil degradation is also shown by
Jostad et al. (2014) for piled foundations.
• This research focusses on the UDCAM-S procedure based on the current database of the NGI for the Drammen Clay.
It is however required to validate the procedure for other clays than the Drammen Clay. Laboratory tests have to
be performed to validate the use of the contour diagrams from the Drammen Clay. Vertical scaling of the contour
diagrams can be applied to match the results found in the laboratory tests.
• Large failure strains can be a result from the UDCAM-S procedure, especially in direct simple shear for heavily
overconsolidated clays. These failure strains can become larger than 30%. However, the NGI-ADP soil model has
restriction on the input of the failure strains. Aside form the large failure strains, the resulting strains from the
UDCAM-S procedure are not always compatible with the NGI-ADP soil model with regard to the ratio of the shear
strains.
The limitations on the failure strains in the NGI-ADP soil model in Plaxis are implemented to ensure convexity of the
yield surface (Brinkgreve et al., 2016). However, if the results from the UDCAM-S procedure are to be used in the
NGI-ADP soil model, a solution has to be found to be able to use the different failure strains for calculations assessing
the deformations.
• This study focuses on the cyclic behaviour of offshore undrained soils due to a design storm. However, this is not the
only situation in which cyclic soil behaviour occurs. The applicability of the UDCAM-S procedure to asses the cyclic
soil behaviour due to such an event should be investigated, such as earthquakes. The UDCAM-S procedure might
be a good option for the assessment of the soil behaviour during an earthquake, regarding stability of analysis of
structures.
72
Bibliography
Andersen, K. (1976). Behavior of clay subjected to undrained cyclic loading. In Behaviour of off-shore structures: proceedings
of the first international conference: BOSS ’76, pages 392–403. University of Trondheim, Norway.
Andersen, K., Kleven, A., and Heien, D. (1988). Cyclic soil data for design of gravity structures. Journal of Geotechnical
engineering, 114(5):517–539.
Andersen, K. H. (2009). Bearing capacity under cyclic loading — offshore, along the coast, and on land. Canadian
Geotechnical Journal, 46(5):513–535.
Andersen, K. H. (2015). Cyclic soil parameters for offshore foundation design. In Meyer, V., editor, Frontiers in Offshore
Geotechnics III, pages 5–82, Oslo, Norway. Taylor & Francis Group.
Andersen, K. H. and Lauritzsen, R. (1988). Bearing capacity for foundations with cyclic loads. Journal of Geotechnical
engineering, 114(5):540–555.
Andresen, L., Andersen, K. H., Jostad, H. P., and Rahim, A. (2007). Bearing capacity of offshore gravity platforms by 3D
FEM. In Pande, G. N. and Pietruszczak, S., editors, Proceedings of the 10th International Symposium on Numerical Models
in Geomechanics (NUMOG X), pages 509–515, Rhodes, Greece. CRC Press.
Andresen, L. and Jostad, H. P. (2009). A FE procedure for calculation of displacements and capacity of foundations
subjected to cyclic loading. In Pande, G. N. and Pietruszczak, S., editors, Proceedings of the 1st International Symposium
on computational geomechanics (ComGeo I), Juan-les-Pins, France. CRC Press.
Andrew, H., Mbistrova, A., and Corbetta, G. (2016). The European offshore wind industry - key trends and statistics 2015.
Technical Report February, European Wind Energy Association.
API (2002). Recommended Practice for Planning , Designing and Constructing Fixed Offshore Platforms — Working Stress
Design. API Publishing Services, Washinton, United States of America.
API (2011). ISO 19901-4:2003 (Modified), Petroleum and natural gas industries - Specific requirements for offshore structures,
Part 4 - Geotechnical and foundation design considerations. API Publishing Services, Washington, United States of America.
Benz, T. (2007). Small-Strain Stiffness of Soils and its Numerical Consequences Small-Strain Stiffness of Soils and its Numerical
Consequences. Phd thesis, Universität Stuttgart.
Bhattacharya, S. (2014). Challenges in Design of Foundations for Offshore Wind Turbines. Engineering & Technology
reference article, pages 1–9.
Bransby, M. F. and Yun, G. J. (2009). The undrained capacity of skirted strip foundations under combined loading.
Géotechnique, 59(2):115–125.
Brinkgreve, R. (2015a). UBCSand model [Lecture notes from Behaviour of Soils and Rocks]. Delft University of Technology.
Brinkgreve, R. (2015b). Undrained Behaviour [Lecture notes from Behaviour of Soils and Rocks]. Delft University of Tech-
nology.
Brinkgreve, R., Kumarswamy, S., and Swolfs, W. (2016). Plaxis Manual. Plaxis bv, The Netherlands, Delft.
Butterfield, R., Houlsby, G., and Gottardi, G. (1997). Standardized sign conventions and notation for generally loaded
foundations. Geotechnique, 47(5):1051–1054.
73
BIBLIOGRAPHY
Byrne, B. (2011). Foundation Design for Offshore Wind Turbines [Lecture Notes]. Geotechnique Lecture, (University of
Oxford).
Cai, F., Eiksund, G., Grimstad, G., and Jostad, H. (2014). An Anisotropic Shear Strength Model for Cyclic Accumulated
Plastic Strain of overconsolidated clay. Electronic journal of geotechnical engineering, 19:4393–4406.
Cai, Y. (2008). Modeling of stress-strain behavior of soft clay under cyclic loading. Geotechnical Engineering for Disaster
Mitigation and Rehabilitation, (1999):277–283.
Davis, E. and Christian, J. (1971). Bearing capacity of anisotropic cohesive soil. Journal of soil Mechanics and Foundation
Divison, 97:753–769.
DNV (1992). Foundations - Classification Notes NO. 30.4. Det Norske Veritas AS, Hovik, Norway.
DNV-GL (2016). Support structures for wind turbines - DNVGL-ST-0126. DNV GL AS, Norway.
Downing, S. and Socie, D. (1982). Simple rainflow couting algorithms. International Journal of Fatigue, 4(1):31–40.
Gottardi, G., Houlsby, G., and Butterfield, R. (1999). Plastic response of circular footings on sand under general planar
loading. Géotechnique, 49(4):453–469.
Gourvenec, S. (2007). Failure envelopes for offshore shallow foundations under general loading. Géotechnique, 57(9):715–
728.
Grimstad, G., Andresen, L., and Jostad, H. (2011). NGI-ADP: Anisotropic shear strength model for clay. International
Journal for Numerical and Analytical Methods in Geomechanics, 36(1):483–497.
Houlsby, G. and Purzin, A. (1999). The bearing capacity of a strip footing on clay under combined loading. Proceedings of
the Royal Society A, 455:893–916.
Jostad, H., Torgersrud, Ø., and Engin, H. (2015). A FE procedure for calculation of fixity of jack-up foundations with
skirts using cyclic strain contour diagrams. In The 15th International Conference: The Jack-up Platform, London, United
Kingdom.
Jostad, H. P., Grimstad, G., Andersen, K. H., Saue, M., Shin, Y., and You, D. (2014). A FE procedure for foundation design
of offshore structures - Applied to study a potential OWT monopile foundation in the Korean Western Sea. Geotechnical
Engineering, 45(4):63–72.
Ladd, C. and Foott, R. (1974). New design procedure for stability of soft clays. Journal of Geotechnical engineering,
100(7):763–786.
Lunne, T., Robertson, P. K., and Power, J. J. M. (1997). Cone penetration testing. E & FN Spon, Cornwall, Great Britain, 1st
edition.
Lupea, C. (2014). Long Term Effects of Cyclic Loading on Suction Caisson Foundations. Msc. thesis, Delft University of
Technology.
Matsui, T., Abe, N., and Bahr, T. (1991). Degradation of Saturated Clays after Cyclic Loading. In Prakash, S., editor, Second
International Conference on Recent Advances in Geotechnical Earthquake Engineering & Soil Dynamics, pages 41–46, St.
Louis, Missouri, United States of America. University of Missouri.
Matsui, T., Nabeshime, Y., and El Mesmary, M. (1999). Degradation in cyclic shear behavior and soil properties of saturated
clays. In Chung, J. S., Matsui, T., and Koterayama, W., editors, Proceedings of the Ninth International Offshore and Polar
Engineering Conference, volume I, pages 536–541, Brest, France. International Society of Offshore and Polar Engineers.
Norén-Cosgriff, K., Jostad, H. P., and Madshus, C. (2015). Idealized load composition for determination of cyclic undrained
degradation of soils. In Meyer, V., editor, Frontiers in Offshore Geotechnics III, pages 1097–1101, Oslo, Norway. Taylor &
Francis Group.
Prévost, J.-H. (1978). Anisotropic undrained stress-strain behavior of clays. Journal of the Geotechnical Engineering Division
(ASCE), 108:1075–1090.
Randolph, M. and Gourvenec, S. (2011). Offshore Geotechnical Engineering. Spon Press, New York, United States of
America, first edition.
74
Skempton, A. (1948). The phi = 0 Analysis of stability and its theoretical basis. In Proceedings of the 2nd International
Conference on Soil Mechanics and Foundation Engineering, pages 72–78.
Spierenburg, S. E. J. (1987). Seabed Response to water waves. Phd thesis, Technical University Delft.
Sterigiou, T., Terzis, D., and Georgiadis, K. (2015). Undrained bearing capacity of tripod skirted foundations under
eccentric loading. Geotechnik, 38:17–27.
Taiebat, H. and Carter, J. P. (2000). Numerical studies of the bearing capacity of shallow foundations on cohesive soil
subjected to combined loading. Géotechnique, 50(4):409–418.
Taiebat, H. and Carter, J. P. (2002). Bearing capacity of strip and circular foundations on undrained clay subjected to
eccentric loading. Geotechnique, 52(1):61–64.
Ukritchon, B., Whittle, A. J., and Sloan, S. W. (1998). Undrained limit analysis for combined loading of strip footings on
clay. journal of geotechnical and geoenvironmental engineering, 124(3):265–276.
Verruijt, A. (2012). Soil Mechanics. Delft University of Technology, Delft.
Vesic, A. S. (1975). Bearing capacity of shallow foundations. In Winterkorn, H. F. and Fang, H. Y., editors, Foundation
Engineering handbook, pages 121–145. Van Nostrand Reinhold, New York.
Von Wolffersdorff, P. (1996). A hypoplastic relation for granular materials with a pre-defined limit state surface. Mechanics
of cohesive-frictional materials, (1):251–271.
Yasuhara, K. (1994). Postcyclic undrained strength for cohesive soils. Journal of Geotechnical engineering, 120(11):1961–
1979.
Zaaijer, M. (2007). Loads, dynamics and structural design [Lecture Notes from Offshore Wind Farm Design]. Delft University
of Technology.
75
Appendices
77
A.
Design codes for offshore gravity based foundations:
DNV-GL (2016), API (2002), and API (2011)
For the design of offshore gravity based foundations the Norwegian and the American guidelines are generally accepted
and widely used. ’Det Norske Veritas’ and the ’American Petroleum Institute’ have guidelines for all the aspects concerning
offshore structures, including gravity based foundations. These two guidelines will be used to compare results from finite
element modelling for the bearing capacity of the gravity based foundation. The static bearing capacity is well defined by
these institutes, and can be calculated according to these guidelines. However, with dynamic loading these guidelines are
more vague and give only or little or no guidance in the design of the foundation. In this chapter the design codes for the
undrained bearing capacity and sliding resistance are presented for both design codes.
A.1 The DNV Approach
The organisation in Norway responsible for the regulations for offshore structures is ’Det Norske Veritas’ (DNV-GL).
DNV GL is a merger between Det Norske Veritas (DNV) and Germanischer Lloyd. In this research there will be referred to
this institute as DNV. DNV offers regulations for the design of offshore foundations in DNV (1992) and DNV-GL (2016).
In this part the undrained bearing capacity of a gravity based foundation is presented as it is described by in DNV (1992);
DNV-GL (2016).
A.1.1 Undrained bearing capacity

The static undrained bearing capacity of a gravity based foundation according to DNV is described by (DNV-GL, 2016)
Qu,d = su · Nc0 · s0c · i0c + p0 · Aef f ,

(A.1)
in which su is the undrained shear strength [kN/m2 ], Nc0 , s0c , and i0c shape factors [-], and p0 the overburden pressure
[kN/m2 ]. The shape factors depend on the shape of the foundation and the eccentricity of the loads (i.e. the ratio between
the vertical force and the bending moment). These shape factors are defined by
Nc0 = π + 2,
bef f
s0c = 1 + 0.2 · ,
lef f (A.2)
s
Hd
i0c = 0.5 + 0.5 · 1− ,
Aef f · sud
in which bef f and lef f are respectively the effective width and effective length of the foundation, Hd the horizontal
design force, and Aef f the effective area of the foundation. The effective area is the area of the foundation that due to non
symmetric loading is not equal to the entire foundation are, and can be determined by the eccentricity of the Load centre.
The effective area of a circular footing is calculated by

e p
Aef f = 2 R2 arccos − e R2 − e2 (A.3)
R
A-1
which represents a rectangular footing, as shown in figure A.1(b). This effective area can be used in the case of a
circular footed gravity based foundation.
r
le
lef f = Aef f · ,
be (A.4)
lef f
bef f = · be ,
le
q
be 2

in which be = 2(R − e) and le = 2R 1 − 1 − 2R . The physical meaning of the different dimensions can be found in
figure A.1. It can be seen that the effective area is significantly smaller than the total area of the gravity based foundation.
(a) Load centre (LC) of a GBF (b) Effective area of the founda-
tion
Figure A.1: Load centre and the effective area of a GBF (DNV-GL, 2016)
A.1.2 Sliding resistance

The sliding resistance of either the construction on the top layer, or sliding between two soil layers, has to satisfy the
following condition
Hd < Aef f · r · su,d , (A.5)

in which Hd is the horizontal load on the foundation, Su,d the design value or the undrained shear strength, and r the
roughness parameter [-]. This roughness parameter has a value of 1.0 when the sliding is checked for soils on soil, but it
has a lower value for sliding with structure on soil.
A.1.3 Foundation stiffness

For the determination of the foundation stiffness the design codes of the DNV give guidelines on four different types of
displacements. In all cases the soil is simplified to a spring, and the stiffness of the spring for these types of displacements
of an embedded gravity based foundation on a uniform soil layer overlying a bedrock is calculated by

4GR R D D D/H
Vertical Kv = 1 + 1.28 1+ 1 + 0.85 − 0.28
1−ν H R R 1 − D/H

8GR R 2D 5D
Horizontal KH = 1+ 1+ 1+
2−ν 2H 3R 4H
(A.6)
8GR3

R D D
Rocking KR = 1+ 1+2 1 + 0.7
3(1 − ν) 6H R H
3

16GR 8D
Torsion KT = 1+
3 3R
in which R is the radius of the circular footing, D is the embedment depth, ν the Poisson’s ratio of the soil, and H the
thickness of the soil layer.
A.1.4 Dynamic loading

The codes of the DNV do not give much guidance in the analysis of dynamic loading of the soil. For homogeneous
soil conditions the aforementioned springs can be used to model the dynamic behaviour of the soil. For non-homogeneous
soils the codes advise to use finite element software. However, there is no guidance in the assessment of the degradation
A-2
of the soil due to cyclic loading.
A.2 The API Approach
The American Petroleum Institute (API) offers guidelines for offshore structures as well. API is the association repre-
senting all aspects accompanied in the oil and gas industry. The design codes from the API for foundation engineering are
widely accepted in the oil and gas industry. In this part the undrained bearing capacity of a gravity based foundation is
presented as it is described by the API. The versions of API (2002) and API (2011) are considered.
A.2.1 Undrained bearing capacity

The undrained bearing capacity of a gravity based foundation according to the API design codes (API, 2002) is defined
by
Qu,d = (su · Nc · Kc + γ · D) · A0 . (A.7)

This formula is similar to the bearing capacity according to the DNV, but the main difference is the definition of the
shape factors and load inclination factors. At first equation (A.8) seems more simplified than equation (A.1), but the
calculation of the effective area is more complex. The effective area is presented in figure A.2. The main difference with
the effective area according to the DNV (figure A.1(b)) is the ring shape instead of filled area.
Figure A.2: Effective area according to the API (API, 2002)
The effective area, as shown in figure A.2 is defined by (API, 2002)
A0 = 2 · s = B 0 · L0 ,
r !1
2
0 R+e
L = 2·s· , (A.8)
R−e
r
0 0 R−e
B =L · .
R+e
In these formulae, the eccentricity (e) of the load is the governing parameter for the determination of the effective
area. The shape factor (s) is also influenced by the eccentricity by
πR2 h p e i
R2 − e2 + R2 sin−1
e
s=− . (A.9)
2 R
The correction factor Kc for the inclination of the loads is determined by
Kc = ic · sc · dc . (A.10)
The shape factors ic , sc , and dc for the correction factor Kc for undrained conditions are calculated by
m·H
ic = 1 −
B 0 · L0 · Su · Nc
B 0 Nq
sc = 1 + 0 · (A.11)
L Nc
D
dc = dq = 1 + 2 tan(φ)(1 − sin(φ))2
B0
A-3
A.2.2 Sliding resistance
The sliding resistance according to the design codes of the API has to satisfy the following condition:
Hd < Su · Aef f , (A.12)

in which Hd is the horizontal design load. This is a simple formula for a very complex process.
A.2.3 Dynamic loading

With dynamic loading the API guidelines are not guiding. The behaviour of the foundation due to dynamic loading is
described as
"Dynamic loads are imposed on a structure-foundation system by current, waves, ice, wind, and earth-
quakes. Both the influence of the foundation on the structural response and the integrity of the foundation
itself should be considered" (API, 2002).
The code by the API is thus not applicable for the bearing capacity in cyclic loading. The engineer has to make his
own judgement on the influence of cyclic degradation with respect to safety and stability of the foundation, but there is no
guidance by the design code.
A-4
B.
NGI-ADP model
This finite element modelling in this research is based on the NGI-ADP constitutive model. The NGI-ADP model is an
advanced soil model suitable for engineering practices on offshore undrained clays. This model has the ability to model
anisotropic stress behaviour, with isotropic elasticity (Grimstad et al., 2011). The strength is defined by the undrained
shear strength of the soil in three different stress conditions, together with the failure strains in the three stress conditions.
B.1 Formulation
The formulation of the NGI-ADP model is described by Cai et al. (2014).
Yield surface
The failure criterion of the NGI-ADP model is an adapted Tresca criterion, which means that the yield contour in the
π-plane has a hexagonal form. This yield criterion has been adapted to avoid numerical problems and fit it into the criteria
of the NGI-ADP model, as shown in figure B.1. The π-plane is a 2D representation of the 3D stress plane, in which the
angles between three axes are equal.
Figure B.1: Visual representation of the NGI-ADP failure criterion in the π-plane (Brinkgreve et al., 2016)
The yield criterion in the NGI-ADP model for plane strain is defined by (Grimstad et al., 2011)
s 2 2
SuA + SuP

σyy − σxx S A − SuP S A + SuP
F = − (1 − κ) · τ0 − κ · u + τxy · u DSS −κ· = 0. (B.1)
2 2 2 · Su 2
The yield criterion of the NGI-ADP model is a Tresca type yield criterion, which has the ability to represent anisotropic
undrained shear strength.
B-1
Hardening function
The advantage of the NGI-ADP model is thus the ability to model the anisotropy in shear strength. For the modelling
of the anisotropy in shear strains the hardening parameter κ is introduced (Cai et al., 2014). This parameter is stress-path
dependent and is defined as
q
γ P /γfP
κ=2· , for γ P < γfP , else κ = 1 (B.2)
1 + γ P /γfP
A value of 1.0 for κ represents isotropic hardening, which means that the hardening function is independent of the
stress orientation in the soil.
B.2 Parameters of the constitutive model
The input for the parameters of the NGI-ADP models is restricted to certain limit to ensure a convex yield surface. A
non convex yield surface will lead to numerical problems. The input parameters of the NGI-ADP model can be divided into
three categories: the strength parameters, the stiffness parameter, and the other parameters.
Strength parameters
A
• Su,ref - Undrained active shear strength
The active undrained shear strength is set to the same value as the undrained shear strength in triaxial compression
(SuC ).
• yref - The reference depth for the undrained shear strength

The reference depth defines the depth at which the active undrained shear strength is equal to the reference
undrained shear strength. Above yref depth the value of SuA is equal to the reference shear strength, below yref
A
the shear strength can increase with Su,inc .
A
• Su,inc - Incremental shear strength over depth
The increment of the shear strength defines the development of the shear strength below the reference depth.
• SuP /SuA - Ratio of passive shear strength over active shear strength
The undrained shear strength in triaxial extension is defined as a ratio with respect to the active undrained shear
strength.
• SuDSS /SuA - Ratio of direct simple shear strength over active shear strength
The undrained shear strength in the direct simple shear test is defined as a ratio with respect to the active undrained
shear strength.
Stiffness parameter
• Gur /sA
u - Ratio of unloading/reloading shear modulus over plane strain active shear strength,
Other parameters
• γfC - Shear strain at failure in triaxial compression
• γfE - Shear strain at failure in triaxial extension
• γfDSS - Shear strain at failure in direct simple shear
• τ0 /SuA - Initial strength mobilisation The initial mobilisation of the shear strength in the soil describes the ratio of
the shear stress before loading with respect to the undrained active shear strength.
• ν - Poisson’s ratio
B-2
C.
Python Script for automatic Plaxis Calculations
The following Python code has been used to control the Plaxis 2D software for automatic calculations. The loads are
read from an Excel file, and the results from each calculation are stored in a separate Excel file, which is used for the
visualization of the VHM failure diagrams.
"""
Python s c r i p t f o r t h e VHM a n a l y s i s o f a GBF i n P l a x i s 2D
P l a x i s 2D v e r s i o n 2015.01
a u t h o r : C . Rasch
C o m p a t i b l e w i t h t h e f i l e ’ L oa dC om bi na ti on s . x l s x ’
. U l t i m a t e l o a d s need t o be c a l c u l a t e d f i r s t and w r i t t e n i n t h e e x c e l f i l e
. L o a d r e s u l t . x l s x n e e d s t o be c l o s e d when c a l c u l a t i n g
Phases
0 − I n i t i a l Phase ( o n l y s o i l , K0 p r o c e d u r e )
1 − C o n s t r u c t i o n o f GBF ( i n c l u d i n g i n t e r f a c e s and p l a t e s )
2 − Apply v e r t i c a l l i n e load
3 − A p p l y H and M t o t h e R e f e r e n c e P o i n t
"""
p l a x i s _ p a t h = r ’ C : \ Program F i l e s ( x86 ) \ P l a x i s \ PLAXIS 2D 201501 ’

##p l a x i s _ p a t h = r ’ C : \ Program F i l e s ( x86 ) \ P l a x i s \ P l a x i s 2D\ PLAXIS 2D 201502 ’
i n p u t p o r t = 1005
from openpyxl import *

import imp
found_module = imp . find_module ( ’ p l x s c r i p t i n g ’ , [ p l a x i s _ p a t h ] )
p l x s c r i p t i n g = imp . load_module ( ’ p l x s c r i p t i n g ’ , * found_module )
from p l x s c r i p t i n g . e a s y import *
import s m t p l i b
import math
import m a t p l o t l i b . p y p l o t as p l t
import m a t p l o t l i b . p a t c h e s as mpatches
import m a t p l o t l i b . l i n e s as m l i n e s
import time
C-1
## F u n c t i o n s
def g e t p h a s e v a l u e ( p l x o b j e c t , phase ) :
return p l x o b j e c t [ phase ]
def create_phases_name_map ( phases ) :

r e s u l t = {}
f o r phase in phases [ : ] :
r e s u l t [ phase . Name . v a l u e ] = phase
return r e s u l t
## I m p o r t Load data from e x c e l f i l e
# Openpyxl − Load c o m b i n a t i o n s
wb = load_workbook ( f i l e n a m e= ’ LoadCombinations . x l s x ’ , d a t a _ o n l y=True ) # w i t h : , d a t a _ o n l y=
True −−> f o r m u l a ’ s d i s s a p e a r b u t v a l u e s can be r e a d
ws = wb[ ’ Comb1 . 1 ’ ]
wb2 = load_workbook ( f i l e n a m e= ’ L o a d R e s u l t . x l s x ’ )
ws2 = wb2[ ’ Blad1 ’ ]
# C r e a t e empty v e c t o r s
v = []
h = []
m = []
stepids = [ ]
phasenames= [ ]
vult = []
v u l t . append ( ws . c e l l ( row = 5 , column = 4) )
Vult = v u l t [ 0 ] . value
# Read t h e l o a d s from t h e e x c e l f i l e
f o r i in range (62 ,95) : #14 i s f i r s t c a l c u l a t i o n , 95 i s the l a s t c a l c u l a t i o n
v . append ( ws . c e l l ( row = i , column = 6) ) # column F
h . append ( ws . c e l l ( row = i , column = 7) ) # column G
m. append ( ws . c e l l ( row = i , column = 8) ) # column H
## PLAXIS c a l c u l a t i o n s −−−−−−−−−−−−−−−−−−−
s , g = new_server ( ’ l o c a l h o s t ’ , i n p u t p o r t )
f o r i in range ( 0 , len ( v ) ) :
V = v [ i ] . value
H = h[ i ] . value
M = m[ i ] . v a l u e
s t a r t _ t i m e = time . c l o c k ( )
nr = i+1
p r i n t ( ’−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−− ’ )
p r i n t ( ’#### C a l c u l a t i o n number ’ , nr )
C-2
print ( ’ ’ )
p r i n t ( " V = {:03 f } kN/m | H = {:05 f } kN | M = {:06 f } kNm" . format (V , H,M) )
# Berekeningen
g . gotostages ()
# V e r t i c a l load in a l l phases constant

g . s e t ( g . L i n e l o a d _ 1 _ 1 . q y _ s t a r t , g . Phases [ 2 ] , −V)
## # change M and H
g . s e t ( g . P o i n t l o a d _ 1 _ 1 .M, g . Phases [ 3 ] , −M)
g . s e t ( g . P o i n t l o a d _ 1 _ 1 . Fx , g . Phases [ 3 ] , H)
g . s e t ( g . Phases [ 2 ] . S h o u l d C a l c u l a t e , True )
g . s e t ( g . Phases [ 3 ] . S h o u l d C a l c u l a t e , True )
p r i n t ( ’ Loads Changed . . −−> C a l c u l a t i n g . . . ’ )

g . calculate ()
print ( ’ Calculation finished ’ )
## Results
g . view ( g . phase_3 )
o u t p u t _ p o r t = 10000
s_o , g_o = new_server ( ’ l o c a l h o s t ’ , o u t p u t _ p o r t )
p h a s e o r d e r = [ g_o . Phase_3 ]
phasesmap_o = create_phases_name_map ( g_o . Phases )
f o r phase in p h a s e o r d e r :
summstage = phase . I n f o . SumMStage . v a l u e
p r i n t ( ’ SumMStage ’ , summstage )
## f o r s t e p in phase . S t e p s . value :
## phasenames . append ( p h a s e . Name . v a l u e )
## s t e p . I n f o . SumMStage . v a l u e
## s t e p i d s . append ( i n t ( s t e p . Name . v a l u e . r e p l a c e ( " S t e p _ " , " " ) ) )
ws2 [ ’ B ’+s t r ( i +5)] = summstage
s_o . c l o s e ( )
print ( ’ ’ )
p r i n t ( ’ Output r e s u l t s s u c c e s f u l ’ )
print ( r ’ Reached SigmaMstage {:20 f } ’ . format ( summstage ) )
p r i n t ( time . c l o c k ( ) − s t a r t _ t i m e , " seconds " )
del summstage
wb2 . s a v e ( ’ L o a d r e s u l t . x l s x ’ )
C-3
wb2 . s a v e ( ’ L o a d r e s u l t . x l s x ’ )
C-4
D.
VHM failure contour diagrams
D.1 3D VHM contour diagrams for OCR = 4 with different values for Neq
0.6
M/ABSu,N=1
0.4
0.2
0
1
-1 1 0
H/AS u,N=1 6 5 4 3 2
V/AS u,N=1
Figure D.1: VHM failure contour diagram for N = 1
0.6
M/ABSu,N=1
0.4
0.2
0
1
-1 1 0
H/AS u,N=1 6 5 4 3 2
V/AS u,N=1
D-1
0.6
M/ABSu,N=1
0.4
0.2
0
1
-1 1 0
H/AS u,N=1 6 5 4 3 2
V/AS u,N=1
0.6
M/ABSu,N=1
0.4
0.2
0
1
-1 1 0
H/AS u,N=1 6 5 4 3 2
V/AS u,N=1
D-2
D.2 HM contour diagrams for OCR = 1 comparing anisotropic and isotropic shear strength
0.8
N = 1, Isotropic
N = 1, Anisotropic
0.7 N = 5, Isotropic
N = 5, Anisotropic
N = 10, Isotropic
N = 10, Anisotropic
N = 20, Non-isotropic
0.5
M/B2 S u
0.4
0.3
0.2
0.1
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
H/BS u
Figure D.5: Applied vertical load of 3068 kN/m (0.3 Vult,N =1 )
0.8
N = 1, Isotropic
N = 1, Anisotropic
N = 5, Anisotropic
N = 10, Isotropic
N = 10, Anisotropic
N = 20, Anisotropic
0.5
M/B2 S u
0.4
0.3
0.2
0.1
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
H/BS u
D-3
0.8
N = 1, Isotropic
N = 1, Anisotropic
N = 5, Anisotropic
N = 10, Isotropic
N = 10, Anisotropic
N = 20, Anisotropic
0.5
M/B2 S u
0.4
0.3
0.2
0.1
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
H/BS u
D.3 Input Load data in the Plaxis calculations for different widths of the foundation
Table D.1: Applied loads on the strip foundation in the 2D Plaxis calculations
Width [m] Vertical Line load [kN/m/m] Horizontal load [kN/m] Overturning moment [kNm/m]
12 486.1 833.3 25000.0
13 414.1 769.2 23076.9
14 357.1 714.2 21428.6
15 311.1 666.7 20000.0
16 273.4 625.0 18750.0
17 242.2 588.2 17647.1
18 216.0 555.6 16666.7
19 193.9 526.3 15789.5
20 175.0 500.0 15000.0
21 158.7 476.2 14285.7
22 144.6 454.5 13636.4
23 132.3 434.8 13043.5
24 121.5 416.7 12500.0
D-4

Modelling of Cyclic Soil Degradation

Uploaded by

Copyright:

Available Formats

Modelling of Cyclic Soil Degradation

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Modelling of Cyclic Soil Degradation

Uploaded by

Copyright:

Available Formats

What is the thesis about?

What is the thesis about?

What modelling approach is developed?

What modelling approach is developed?

Modelling of cyclic soil degradation

Development of a cyclic accumulation model

Delft University of Technology

in partial fulfilment of the requirements for the degree of

Prof. dr. M. A. Hicks

Dr. Ir. R. B. J. Brinkgreve

Supervisor BAM Infraconsult

Dr. Ir. S. E. J. Spierenburg

An electronic version of this thesis is available at http://repository.tudelft.nl/

CSR Cyclic stress ratio

DAF Dynamic Amplification Factor

DSS Direct Simple Shear test

GBF Gravity Based Foundation

NGI Norwegian Technological Institute

OCR Overconsolidation ratio

OTM Overturning Moment

SLS Serviceability Limit State

TXC Triaxial Compression test

TXE Triaxial Extension test

UDCAM UnDrained Cyclic Accumulation Model

UDCAM-S Simplified UnDrained Cyclic Accumulation Model

ULS Ultimate Limit State

χ Scaling factor as a result of the strain accumulation procedure [-]

xx Horizontal strain [-]

yy Vertical strain [-]

γa Average shear strain [%]

γfC Shear strain at failure in the triaxial compression test [%]

γfE Shear strain at failure in the triaxial extension test [%]

γp Plastic shear strain [%]

γxy Shear strain [-]

ν Poisson’s ratio [-]

σ0 Effective normal stress [kPa]

σ1 Main principal total stress [kN/m2 ]

σ3 Minor principal total stress [kN/m2 ]

σxx Horizontal total stress [kN/m2 ]

σyy Vertical total stress [kN/m2 ]

τ Shear stress [kPa]

τa Average shear stress [kPa]

τcy Cyclic shear stress [kN/m2 ]

τcy /τa Cyclic stress ratio [-]

Fcy /Fa Cyclic load ratio [-]

Fcy /Fcy,max Cyclic load ratio of the load parcels [-]

qc Cone resistance of the CPT cone [MPa]

yref Reference depth [m]

B Width of the foundation [m]

Fa Average load [kN]

Fcy Cyclic load [kN]

H Horizontal load [kN]

K0 The neutral earth pressure coefficient [-]

M Overturning moment [kNm]

N Number of cycles [-]

V Vertical load [kN]

2 Gravity based foundations 5

3 Undrained cyclic soil behaviour 11

4 The Simplified UnDrained Cyclic Accumulation Model 19

xx Horizontal strain [-]

yy Vertical strain [-]

yy,av = γav/100 · 2/3,

yy,av = γav/100 · 2/3,