Beruflich Dokumente
Kultur Dokumente
Y He
AECOM, Birmingham, UK; formerly Department of Engineering Science, University of Oxford, Oxford, UK;
Email: yiling.he@yahoo.com
BW Byrne and HJ Burd
Department of Engineering Science, University of Oxford, Oxford, UK
Abstract
The Pile Soil Analysis (PISA) project, established to improve design methods for laterally loaded monopiles
for offshore wind turbines (OWTs), proposes two new one-dimensional (1D) design methods: (a) a rule-based
method similar to existing design standard approaches, and (b) a numerical-based method drawing on the
results of three dimensional (3D) finite element (FE) computations. This paper examines the applicability of
the numerical-based method for analysing monopile response in layered soil conditions. Soil reaction
components were extracted from 3D FE through a calibration exercise on homogeneous soils, parameterised
to develop soil reaction curves, with these curves then incorporated into a 1D beam model for predicting pile
responses in layered soils. Good agreement in the ultimate capacity and initial stiffness between the results
obtained using the calibrated 1D model and the 3D FE simulations provided evidence that the numerical-
based method is applicable for layered soil analyses. Further insights into layering were obtained by
comparing soil reaction components extracted from layered soil analyses to those from homogeneous soil
analyses.
15
the soil parameters and details of the calibration 3D
Aspect ratio L/D
4
10 FE analyses are further discussed in He (2016).
PB PA PB Gapping at the pile-soil interface was allowed in the
2
5 clay layer, but not in the sand layer. A small
Calibration
Layered soils cohesion was adopted for the sand layer to avoid
0 0
4 6 8 10 4 6 8 10 numerical singularity issues.
Pile diameter D Pile diameter D
Figure 4: Pile parameters employed in the calibration and
layered soil analyses (PA and PB please refer to Table 1)
Table 2 Material parameters for clay layer The base horizontal force, S, and base moment, M
were derived by summing the nodal forces and the
Parameter Value
moments due to the nodal forces of the soil elements
Shear modulus, G (MPa) 37.5
located immediately below the pile base and the soil
Undrained shear strength, su (kPa) 150 plug.
Poisson’s ratio, υ 0.49
Saturated bulk unit weight, γ (kN/m3) 21.2 2.5 Parameterisation
Earth pressure coefficient at rest, K0 1.0 The extracted soil reaction components from the
Friction coefficient at pile-soil interface, μ 0.4 calibration analyses were normalised and
parameterised to produce dimensionless soil reaction
Table 3 Material parameters for sand layer curves (described further in He, 2016). The
parameterised soil reaction curves were then
Parameter Value
provided as input to the 1D beam model to predict
Shear modulus, G (MPa) 20
pile lateral responses in layered soil cases.
Effective internal friction angle, φ' (°) 35
Dilation angle, ψ (°) 5 The effectiveness of the soil reaction extraction and
Poisson’s ratio, υ 0.25 parameterisation was determined by comparing the
Submerged unit weight, γ' (kN/m3) 11 results from the 1D beam model using the
Earth pressure coefficient at rest, K0 0.426 parameterised soil reaction curves and those from
Friction coefficient at pile-soil interface, μ 0.4 3D FE analyses for (a) the ultimate lateral load
Cohesion (kPa) 1.0 defined at 0.1D ground level displacement and (b)
the initial stiffness defined at 0.0001D ground level
2.4 Extraction of soil reaction curves displacement. This is presented in Figure 5. The
After the 3D FE calibration analyses were maximum difference in the ultimate lateral load is
within 10%, while the differences in the initial
completed, the nodal force of the soil elements
stiffness are largely within 10%. Overall, the
immediately surrounding the pile, and the nodal calibrated 1D model was considered acceptable for
displacements of the pile elements, were processed exploring application to layered soil cases,
to obtain the soil reaction components. As mention particularly considering the small number of
previously, this process follows the procedures analyses from which the reaction curves are drawn
outlined in Byrne et al. (2015a). and the range of geometries to which they apply.
Improved agreement could be obtained by
The distributed soil reaction curves along the pile increasing the number of calibration analyses and by
length were derived at the reference nodes, which introducing more complex parameterisation
were defined at the centre of each quadrilateral shell expressions.
element used to model the pile. The distributed
500
lateral soil reaction, p, was calculated by dividing
k0.0001D computed by 1D analyses
Hult computed by 1D analyses
60
the sum of the horizontal nodal forces of a ring of 400 overpredicted
elements at the same depth, by the shared length of overpredicted
300
40
the ring of elements along the pile length direction.
The pile lateral displacement, v, at the reference underpredicted
200
underpredicted
20
location was interpolated from the average nodal L/D = 2 100 L/D = 2
L/D = 6 L/D = 6
displacement of the shell element at the same depth. 10% band 10% band
0 0
0 20 40 60 0 100 200 300 400 500
The distributed moment (denoted by m) was Hult computed by 3D analyses k0.0001D computed by 3D analyses
calculated by multiplying the distributed vertical (a) Hult (MN) (b) k0.0001D (MN/m)
shear forces at the pile-soil interface with their Figure 5: Comparison of the ultimate lateral load (Hult) and
initial stiffness (k0.0001D) predicted by the 1D model (with
distance along the x-axis from the neutral axis of the parameterised curves) and 3D models for the calibration
pile. The distributed vertical shear forces were analyses (solid symbols:sand, open symbols: clay)
calculated by dividing the sum of the vertical nodal
forces of an element, by the length of the element 3. Layered soil analyses
along the pile length direction. The rotation of the 3.1 Problem definition
pile cross-section, θ, was determined by regression Two simple layered soil profiles were analysed in
of the vertical nodal displacement against the this study: a uniform sand layer over a uniform clay
distance of the reference node from the central axis layer (denoted ‘S/C’), and a uniform clay layer over
of the pile cross-section, to obtain the best linear fit a uniform sand layer (denoted ‘C/S’), as illustrated
slope. in Figure 6. The input soil reaction curves for the
10
sand or clay layer were produced from the 3D
4
3D
2 1D Parameterised
1D Numerical
0
0.00 0.05 0.10 0.15 0.20 0.25
Displacement at ground level (m)
(d) PB-C/S analysis
(a) S/C (b) C/S Figure 7: Comparison of the load-displacement responses for
Figure 6: Soil profiles for the layered soil analyses 1D and 3D analyses
Parameterised) and the 3D FE results were due to 1 10 100 0 200 400 600 800
the parameterisation of soil reaction components or Hult computed by 3D analyses k0.0001D computed by 3D analyses
soil layering effects. This involved conducting 3D (a) Hult (MN) (b) k0.0001D (MN/m)
FE for the piles PA and PB in homogeneous soil Figure 8: Comparison of the ultimate lateral load (Hult) and
profiles, extracting the soil reaction components, and initial stiffness (k0.0001D) of pile response predicted by the 1D
model (with parameterised curves) and 3D models for the
then incorporated these as numerical data into the layered soil analyses (solid symbols: S/C, open symbols: C/S)
1D model at the corresponding depth for the layered
soil analyses. Figure 7 and Figure 8 indicate satisfactory
150
agreement between the 3D results and the 1D
prediction using the parameterised curves. The
Lateral load at pile top (MN)
120
maximum difference in the ultimate capacity of
90 approximately 20% is for the PB-S/C analysis of a
60 short pile embedded in sand over clay. The
30
3D differences in the initial stiffness of pile response are
1D Parameterised
1D Numerical
within 15%. Note that the maximum displacement
0 achieved in the short pile analysis was limited due to
0.0 0.2 0.4 0.6 0.8 1.0
Displacement at ground level (m) convergence issues in the FE analysis; the maximum
(a) PA-S/C analysis load was thus taken as the ultimate lateral load.
150
Lateral load at pile top (MN)
Normalised p
Normalised p
10
homogeneous soil profiles, can be satisfactorily 6
Normalised p
Normalised p
4 8
(0.1L, 0.4L, 0.6L and 0.9L), where L is pile
embedded length, with the soil reaction components 2 depth = 0.1L 4 at depth = 0.4L
extracted from the layered soil analyses compared to Homogeneous Homogeneous
Layered Layered
those obtained from the homogeneous soil analyses 0 0
0 10 20 30 40 0 4 8 12
at the same depth. Normalised v Normalised v
1.5 3
at depth = 0.6L
Normalisation of the soil reactions was adopted to Homogeneous
Normalised p
Normalised p
1.0 Layered 2
facilitate comparison between the sand and clay
layers. The distributed load was normalised by the 0.5 1 at depth = 0.9L
undrained shear strength (su) times diameter (D) for Homogeneous
Layered
the clay layer (i.e. p/suD), and by the local effective 0.0 0
stresses (σvi΄) times the diameter for the sand layer 0.0 0.1 0.2
Normalised v
0.3 0.4 0.5 0.0 0.2 0.4
Normalised v
0.6 0.8 1.0
(i.e. p/σvi΄D). The horizontal pile displacement was (b) PA-C/S analysis
normalised by suD divided by the shear modulus (G)
for clay (i.e. vG/suD), and by σvi΄D/G for sand (i.e. 10 10
Normalised p
6 6
Normalised p
are observed for large lateral displacements in both (c) PB-S/C analysis
the upper and lower layers; this may be due to 5
effects of soil layering. For the cases considered, the 4
6
Normalised p
3 4
to be higher than those in the homogeneous soils for 2
at depth = 0.1L at depth = 0.4L
sand, while they tended to be lower for clay. In 1 Homogeneous
2
Homogeneous
addition, the differences in the ultimate lateral 0
Layered
0
Layered
Normalised p
5 0.4
12
4 at depth = 0.6L 2 at depth = 0.9L
0.2
Homogeneous
Normalised p
Homogeneous
Normalised p
8 3 Layered Layered
0.0 0
2 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.5 1.0 1.5
at depth = 0.1L at depth = 0.4L Normalised v
4 Normalised v
Homogeneous 1 Homogeneous
Layered Layered (d) PB-C/S analysis
0 0
0 10 20 30 40 0 1 2 3 4
Figure 9: Comparison of the distributed lateral load curves
Normalised v Normalised v from the homogeneous and layered soil analyses
0.6 0.8
3.3.2 Distributed moment curves at depth = 0.4L
The normalised distributed moment against 0.6 Homogeneous
Normalised m
0.4 Layered
Normalised m
normalised pile rotation at different depths is shown 0.4
in Figure 10 (denoted by normalised m and θ 0.2 at depth = 0.1L
Homogeneous 0.2
respectively). The comparison indicates that initial Layered
stiffness determined from the homogeneous and 0.0
0 10 20 30 40
0.0
0 2 4 6 8 10
layered soil analyses matches well. Differences are Normalised Normalised
0.3 1.0
again observed for the ultimate response. at depth = 0.6L
Homogeneous 0.8
Normalised m
Normalised m
0.2 Layered
2.0 0.5 0.6
at depth = 0.4L
0.4 0.4
1.5 Homogeneous 0.1 at depth = 0.9L
Normalised m
Normalised m
Layered
Normalised m
0.6 0.3
lateral load and moment determined from the
0.4 0.2
at depth = 0.6L homogeneous soil and layered soil analyses
0.2 Homogeneous
Layered
0.1
illustrates the complexity of the soil layering effects.
0.0 0.0
0 1 2 3 4 5 6
However these differences appear to balance out
0 2 4 6 8
Normalised Normalised when integrated into the overall load-displacement
(a) PA-S/C analysis response. Further studies are required to investigate
0.5 1.0
soil layering effects on the individual soil reaction
0.4 0.8
at depth = 0.4L components, particularly more complex layering
Homogeneous
profiles, and therefore the overall pile behaviour.
Normalised m
Normalised m
Layered
0.3 0.6
Also, care should be employed when applying base
0.2 0.4
at depth = 0.1L shear and base moment in layered soils, as soil
Homogeneous
0.1
Layered
0.2
layers above and below the actual pile tip may have
0.0
0 5 10 15 20
0.0
0 4 8 12
influence on the soil resistance. However, due to
Normalised Normalised limited scope of this study, the potential effect of
0.25 0.4
variation of base soil layer has not been examined,
0.20
0.3 and further analysis will be needed to address this
Normalised m
Normalised m
0.15
0.2
matter.
0.10
at depth = 0.6L at depth = 0.9L
0.1
0.05 Homogeneous Homogeneous
Layered
4. Conclusions
Layered
0.00 0.0 3D FE and 1D pile analyses have been performed to
0.0 0.5 1.0 1.5 0.0 0.2 0.4 0.6 0.8 1.0
Normalised Normalised examine the application of the PISA numerical-
(b) PA-C/S analysis based method to monopile analyses in layered soils.
2.0 1.5
The comparison of results using a 1D model
incorporating homogeneous soil reaction curves and
1.5 3D finite element calculations for layered profiles
Normalised m
Normalised m
1.0
0.4 0.4
soil reaction components (normalised p and m),
0.2 at depth = 0.6L 0.2 at depth = 0.9L show differences in the ultimate response, for both
Homogeneous Homogeneous the upper and lower layers, but this appeared to even
Layered Layered
0.0 0.0 out when integrated along the 1D model. The soil
0 1 2 3 4 5 0 2 4 6 8 10
Normalised Normalised reaction component comparisons show the complex
(c) PB-S/C analysis nature of the soil laying effects. Further work is
needed to explore the application of this method to
more complex layering conditions.
5. Acknowledgements Gupta BK and Basu D. (2016). Analysis of laterally
The first author is grateful to Dr Ross McAdam, loaded rigid monopiles and poles in multi-layered
Stephen Suryasentana and William Beuckelaers for linearly varying soil. Computer and Geotechnics
their valuable discussions and suggestions. 72, 114–125.
Hald T, Mørch C, Jensen L, LeBlanc Bakmar C and
References Ahle K. (2009). Revisiting monopile design using
Abaqus. (2013). User’s manual. Dassault Systemes p-y curves results from full scale measurements
Simulia Corp. Providence; Version 6.13. on Horns Rev. DONG Energy A/S, Proceedings
Abdel-Rahman, K. and Achmus, M. (2005). Finite of European Offshore Wind Conference and
element modelling of horizontally loaded Exhibition.
monopile foundations for offshore wind energy He Y. (2016) Application of a numerical-based soil
converters in Germany. Proceedings of the reaction curve method for design of laterally
International Symposium on Frontiers in loaded monopiles. MSc (by Research) Thesis.
Offshore Geotechnic (ISFOG 2005), Perth, University of Oxford.
Australia, 391-396. Lehane, B.M., Chow, F.C., McCabe, B.A. and
API, American Petroleum Institute. (2014). Jardine, R.J. (2000). Relationships between shaft
Recommended Practice for Planning, Designing capacity of driven piles and CPT end resistance.
and Construction Fixed Offshore Platforms - Proc. Institution of Civil Engineers, Geotechnical
Working Stress Design, 22nd Edition, Dallas. Engineering 143, 93-101.
Arany L, Bhattacharya S, Macdonald JHG and John LPILE User’s Manual 2013 (2014), Ensoft Inc.
Hogan S. (2016). Closed form solution of Eigen Matlock H. (1970). Correlations for Design of
frequency of monopile supported offshore wind Laterally Loaded Piles in Soft Clay. Proceedings
turbines in deeper waters incorporating stiffness of the 2nd Annual Offshore Technology
of substructure and SSI. Soil Dynamics and Conference, OTC 1204, Houston, Texas, 577–
Earthquake Engineering 83, 18–32 588.
Ashour M, Norris G and Pilling P. (1998). Lateral Reese LC, Cox WR and Koop FD. (1974). Analysis
Loading of a Pile in Layered Soil Using the of Laterally Loaded Piles in Sand. Proceedings of
Strain Wedge Model. Journal of Geotechnical the 6th Offshore Technology Conference, paper
and Geoenvironmental Engineering 124, No.4, No. OTC 2080, Houston, Texas, 473–483.
303–315. Reese LC, Allen JD and Hargrove JQ. (1981).
Byrne BW, McAdam R, Burd HJ, et al. (2015a). Laterally loaded piles in layered soils. Proc. 10th.
New design methods for large diameter piles Int. Conf. Soil Mech. And Found. Engrg.
under lateral loading for offshore wind Stockholm: A.A. Balkema, 2, 819–822.
applications. Proc 3rd International Symposium Roesen HR, Thomassen K, Ibsen LB and Sørensen
on Frontiers in Offshore Geotechnics (ISFOG SPH. (2011). Evaluation of Small-Scale Laterally
2015). Oslo, Norway, 705–710. Loaded Monopiles in Sand. In Symposium
Byrne BW, McAdam R, Burd HJ, et al. (2015b). Proceedings: 64th Canadian Geotechnical
Field testing of large diameter piles under lateral Conference and 14th Pan-American Conference
loading for offshore wind applications. on Soil Mechanics and Engineering, 5th Pan-
Proceedings of XV European Conference on Soil American Conference on Teaching and Learning
Mechanics and Geotechnical Engineering of Geotechnical Engineering. Toronto, Ontario:
(XVECSMGE), Edinburgh. Pan-AM CGS Geotechnical Conference.
Davisson MT and Gill HL. (1963). Laterally loaded Yang Z and Jeremić B. (2002). Numerical analysis
piles in a layered soil system. Journal of the Soil of pile behaviour under lateral loads in layered
Mechanics and Foundations Engineering (ASCE) elastic-plastic soils. International Journal of
89, No.SM3, 63–94. Numerical Analytical Method Geomechanics 26,
De Vries WE and Van der Tempel J. (2007). Quick 1385–1406.
monopile design. Proceedings of the European Zdravković L, Taborda DMG, Potts DM, et al.
Offshore Wind Conference and Exhibition. (2015). Numerical modelling of large diameter
Berlin, Germany. piles under lateral loading for offshore wind
DNV-OS-J101 Design of Offshore Wind Turbine applications. Proc. 3rd International Symposium
Structures 2014. on Frontiers in Offshore Geotechnics (ISFOG
Doherty P and Gavin K. (2011). Laterally loaded 2015). Oslo, Norway, 759–764.
monopile design for offshore wind farms.
Proceedings of the ICE - Energy, 165(1): 7–17.
Georgiadis M. (1983). Development of p-y curves
for layered soil. Proc. Geotech. Practice in
Offshore Engineering (ASCE), 1, 536-545.