PLAXIS MoDeTo Validation Cowden Clay
PLAXIS MoDeTo Validation Cowden Clay
PLAXIS MoDeTo Validation Cowden Clay
Validation of PLAXIS MoDeTo based on the Cowden till PISA field tests
Eleni Minga and Harvey Burd
Oxford University, UK
6/2/2019
1. Introduction
This document describes a validation example for the PISA methodology implemented in PLAXIS
MoDeTo (Panagoulias et al. 2018) for a clay soil, based on the results obtained from the pile tests
conducted at Cowden during the PISA project.
During the PISA project, data from the Cowden pile tests were used to develop and validate the 1D
PISA design model as follows:
(a) A ground profile and constitutive model calibration was developed for the Cowden test site
based on the available site investigation data (Zdravković et al. 2019a). 3D finite element
models, based on these data, were then used to analyse the response of the test piles at
Cowden (Zdravković et al. 2019b). Comparisons between the measured response of the test
piles and the corresponding finite element results confirmed the validity of the calibration data
and the 3D finite element modelling procedures.
(b) A separate ground model was developed, for a ‘representative glacial till offshore site’. This
model was closely-based on the conditions at the Cowden site, but with certain modifications
(e.g. to the initial pore pressure distribution) to ensure that the ground model is representative
of conditions at a submerged offshore site (noting that the Cowden site is onshore). 3D finite
element calibration calculations, using the constitutive model calibration data from Step (a),
together with the representative glacial till offshore site ground profile, were then used to
determine the PISA 1D model parameters for this particular representative offshore site.
In the current document, an alternative process is adopted to demonstrate a validation of the PLAXIS
MoDeTo model as follows:
(c) Develop a ground profile and constitutive model calibration for the Cowden test site using
the NGI-ADP model (Andersen & Jostad 1999).
(d) Conduct a calibration exercise for the PLAXIS MoDeTo approach, using the calibration for
the Cowden site from Step (c). The calibration pile set consists of pile geometries that straddle
the geometric parameters of one of the medium diameter Cowden test piles (CM9, 𝐷 =
0.762m, 𝐿 = 3.98m) .
(e) Demonstrate that the calibrated PLAXIS MoDeTo model provides a close fit with the
measured response of CM9.
It should be noted that in these calibration calculations the site is onshore. Any gap that forms
between the soil and the pile as it is loaded is assumed not to fill with water; as a consequence, the
unit weight adopted for the soil is the saturated bulk unit weight. This contrasts with an offshore site,
where any gaps at the soil-pile interface will immediately fill with water; in this case, the submerged
unit weight should be adopted for the soil. The modelling procedures described in the current
document (employing the saturated bulk unit weight) therefore differ from the procedures that would
be normally be employed in practice (based on the submerged unit weight).
1
Validation of PLAXIS MoDeTo for Cowden Till
A further consideration is that the soil shear strength variation at the Cowden site has a very significant
nonlinear variation with depth in the top 5m. Since the calibration piles all lie in this surface region,
the performance of the MoDeTo calibration process is likely to be slightly less accurate than in practical
design cases in which the piles are typically significantly longer than the length scale of any spatial
variations with depth in the soil strength profile.
The first part of this document describes the calibration of the NGI-ADP model for the Cowden till soil
profile. This calibration is based on (i) site investigation data obtained before the PISA field tests and
(ii) subsequent triaxial compression (TXC) tests performed by Imperial College during the PISA project.
This ground model calibration is demonstrated by analysing, using PLAXIS 3D (Brinkgreve et al. 2018),
three of the Cowden test piles and comparing the computed results with the field test data. The
document then describes the calibration of a PLAXIS MoDeTo 1D model for the Cowden site.
Parameter Description
γsat (kN/m3) Saturated unit weight
K0total Lateral earth pressure coefficient for total stresses
suA (kPa) Undrained shear strength with linear distribution within a given soil layer
su,incA (kPa/m)
Gur/suA Shear modulus ratio (where 𝐺𝑢𝑟 is interpreted as having the same meaning as
the small strain shear modulus 𝐺𝑜 )
suC/suA Ratio between undrained shear strength for triaxial compression (TXC) and suA
suP/suA Ratio between undrained shear strength for triaxial extension (TXE) and suA
suDSS/suA Ratio between undrained shear strength for direct simple shear (DSS) and suA
τ0/suA
γf C (%) Ultimate plastic strain for compression
E
γf (%) Ultimate plastic strain for extension
γf DSS (%) Ultimate plastic strain for direct shear
The saturated unit weight of Cowden till has been estimated at 𝛾𝑠𝑎𝑡 = 21.19 𝑘𝑁/𝑚3 (Zdravković et
al. 2019a).
Profiles of the undrained shear strength, 𝑠𝑢𝐶 for triaxial compression (TXC), small strain shear modulus,
𝐺0 , and the lateral earth pressure coefficient in terms of effective stresses, 𝐾0 , at Cowden were
established during the PISA project (Zdravković et al. 2019a). These profiles are shown in Fig. 1. This
ground model was used, directly, to determine the values of suC, 𝐺𝑢𝑟 (assumed to be equivalent to 𝐺𝑜 )
and 𝐾0𝑡𝑜𝑡𝑎𝑙 (i.e. the value of 𝐾0 for the total stresses) for the NGI-ADP soil model. [Note that Fig. 1
provides data on the value of 𝐾0 for the effective stresses. Values of 𝐾0𝑡𝑜𝑡𝑎𝑙 for use in the NGI-ADP
model are determined from these data using Equations 1-5]. The NGI-ADP model adopts a linear depth
2
Validation of PLAXIS MoDeTo for Cowden Till
variation of suc and Gur, and a constant value of 𝐾0𝑡𝑜𝑡𝑎𝑙 in each soil layer. To obtain a reasonable
representation, a relatively large number of layers (12 layers in the top 22m of the soil profile) is
adopted in the model. Soil layers of thickness 1m were employed to a depth of 6m, to provide an
appropriate model for the significant variation of undrained shear strength close to the surface. At
greater depths, thicker layers are employed. The piecewise-linear profiles adopted in the PLAXIS 3D
ground model are indicated as ‘PLAXIS’ in Fig. 1.
Figure 1. Cowden till ground model (a) triaxial compression (TXC) undrained shear strength (b) small
strain shear modulus, 𝐺𝑜 (c) lateral earth pressure coefficient (in terms of effective stress) 𝐾𝑜 ; (d)
pore water pressure, u. Cowden site data from (Zdravković et al. 2019a,b).
3
Validation of PLAXIS MoDeTo for Cowden Till
The ratio 𝑠𝑢𝐶 ⁄𝑠𝑢𝐴 is set to the default value 0.99. The suA profile can therefore be determined directly
from the 𝑠𝑢𝐶 profile in Fig. 1(a). It is noted that the negative gradient of 𝑠𝑢𝐶 (and therefore 𝑠𝑢𝐴 ) in the
layers between depths of 2m and 4m causes a negative gradient of the simulated 𝐺𝑜 (due to the
constant ratio Gur/suA in the layer). The thickness of these layers is small, however, and so the influence
of these locally-unrealistic 𝐺𝑜 gradients is unlikely to be significant.
The K0 profile in Fig. 1(c) corresponds to the effective stresses. Since NGI-ADP is a total stress model,
and the calculations in this case are based on the bulk unit weight 𝛾𝑠𝑎𝑡 , the K0total profile needs to be
determined. Values of K0total at the mid-depth of each layer were calculated as follows:
𝜎𝑣 = 𝑧 𝛾𝑠𝑎𝑡 (1)
𝜎𝑣′ = 𝜎𝑣 − 𝑢 (2)
where u is pore pressure determined at the appropriate depth from the data in Fig. 1(d).
𝜎ℎ = 𝜎ℎ′ + 𝑢 (4)
The triaxial tests performed during the PISA project indicated that the average ultimate stress ratios
q/p’ in triaxial compression (TXC) and triaxial extension (TXE) are 1.07 and 0.9 respectively (Zdravković
et al. 2019a).
Therefore,
𝑝
𝑠𝑢 ⁄𝑠𝑢𝐶 = 0.90⁄1.07 = 0.841 (6)
This gives,
𝑠𝑢𝑃 ⁄𝑠𝑢𝐴 = (𝑠𝑢𝑃 ⁄𝑠𝑢𝐶 )(𝑠𝑢𝐶 ⁄𝑠𝑢𝐴 ) = 0.841 × 0.99 = 0.833 (7)
𝑝
The undrained shear strength for direct simple shear 𝑠𝑢𝐷𝑆𝑆 is taken as the average of 𝑠𝑢 and 𝑠𝑢𝑐 . This
gives,
1
𝑠𝑢𝐷𝑆𝑆 ⁄𝑠𝑢𝐴 = 2 (𝑠𝑢𝐶 + 𝑠𝑢𝑃 )/𝑠𝑢𝐴 = 0.911 (8)
When the value calculated by Equation (9) is negative, the τ0/suA ratio is set to zero. In the current
example, the soil has relatively high values of 𝐾0 near the surface; the value of τ0/suA determined on
this basis is either zero or, for depths greater than about 10m, a small positive value. Separate
numerical experiments were conducted with τ0/suA=0.7 for all of the layers, but the resulting PLAXIS
3D models of the Cowden field test piles were all unrealistically stiff.
4
Validation of PLAXIS MoDeTo for Cowden Till
Table 2. Specifications of the initial conditions for the Cowden TXC triaxial tests (Zdravković et al.
2019a).
Depth (m) K0 σv' (kPa) σh' (kPa) u (kPa) σv (kPa) σh (kPa) K0total
2.5 1.5 41.1 61.7 14.6 55.7 76.2 1.368
5 1.5 69.5 104.3 37.7 107.2 142.0 1.325
8.2 1.0 121.0 121.0 58.1 179.1 179.1 1.000
The values of the parameters 𝛾𝑓𝐶 , 𝛾𝑓𝐸 and 𝛾𝑓𝐷𝑆𝑆 were based on the results of TXC tests on Cowden till
samples taken from depths of 2.5m, 5m and 8.2m. These TXC tests, conducted by Imperial College
during the PISA project, are specified in Table 2. The effective stress conditions indicated Table 2 –
which are the stresses actually applied during the initial 𝐾𝑜 consolidation stage - relate approximately
to the estimated in situ stresses for each sample depth. The pore pressures, u, in Table 2 (required to
determine the appropriate total stresses to facilitate the NGI-ADP analysis of the tests) have been
determined from the pore pressure profile in Fig.1(d).
The triaxial tests were simulated in PLAXIS 3D (using a finite element model for the test rather than
the soil test facility). Calculations were conducted in two stages. Initially the sample was compressed
by applying the total stresses indicated in Table 2. Then, in the second stage, axial compression was
applied. During the calibration, the ratios between the three plastic strain parameters were kept
constant at 𝛾𝑓𝐸 = 2𝛾𝑓𝐶 and 𝛾𝑓𝐷𝑆𝑆 = 1.5𝛾𝑓𝐶 . It was considered that 𝛾𝑓𝐶 varies as a function of the ratio
𝐺𝑢𝑟 /𝑠𝑢𝐴 , based on the equation (Panagoulias et al. 2018),
𝐺𝑢𝑟
𝛾𝑓𝐶 = 𝐾𝑓 ∙ 100/( 𝐴) (10)
𝑠𝑢
where 𝐾𝑓 is the single calibration parameter that is adjusted to provide, by eye, the best fit between
the TXC results and the PLAXIS 3D results. The calibration data for the model obtained from this
process are listed in Table 3.
5
Validation of PLAXIS MoDeTo for Cowden Till
Table 3. Calibration of the NGI-ADP model for Cowden till. [Note that the water table is taken to be at
1m below the ground surface.]
Layer Depth K0 u (kPa) K0total suA (kPa) suAinc γsat Gur'/suA suP/suA suDSS/suA τ0/suA Kf γf C (%) γf E (%) γf DSS
(m) (kPa/m) (kN/m3) (%)
1 0 1.500 -4.950 1.500 50.149 56.626 21.19 289.107 0.833 0.911 0.000 190 65.720 131.439 98.580
1 106.775
2 1 1.500 4.929 1.422 106.775 56.626 21.19 290.504 0.833 0.911 0.000 150 51.634 103.269 77.452
2 163.401
3 2 1.500 14.571 1.362 163.401 -15.298 21.19 361.861 0.833 0.911 0.000 120 33.162 66.324 49.743
3 148.103
4 3 1.500 23.999 1.338 148.103 -37.555 21.19 568.968 0.833 0.911 0.000 120 21.091 42.182 31.636
4 110.548
5 4 1.500 33.237 1.283 110.548 3.263 21.19 784.927 0.833 0.911 0.000 120 15.288 30.576 22.932
5 113.811
6 5 1.305 41.274 1.197 113.811 6.736 21.19 854.320 0.833 0.911 0.000 120 14.046 28.093 21.069
6 120.547
7 6 1.117 51.200 1.077 120.547 3.711 21.19 916.135 0.833 0.911 0.000 120 13.099 26.197 19.648
8 127.969
8 8 0.992 60.110 0.994 127.969 6.056 21.19 1065.790 0.833 0.911 0.003 120 11.259 22.519 16.889
10 140.081
9 10 0.970 60.066 0.978 140.081 8.624 21.19 1224.000 0.833 0.911 0.013 120 4.902 9.804 7.353
11 148.704
10 11 0.918 45.797 0.934 148.704 8.322 21.19 1355.089 0.833 0.911 0.038 120 4.428 8.856 6.642
12 157.026
11 12 0.863 52.864 0.888 157.026 3.243 21.19 1492.755 0.833 0.911 0.071 120 4.019 8.039 6.029
16 169.998
12 16 0.816 101.182 0.862 169.998 3.379 21.19 1613.928 0.833 0.911 0.116 120 3.718 7.435 5.576
22 190.275
Figure 2 shows the comparison between the experimental TXC data and the corresponding PLAXIS 3D
simulations employing the calibrated model. A generally good agreement is obtained. The results of
the simulated TXC tests have been used to determine the degradation of the secant shear modulus
with incremental deviatoric strain, 𝛿𝜀𝑑 , where the general form of incremental deviatoric strain (in
terms of principal stresses) is (Potts and Zdravković 2001):
(11)
2
𝛿𝜀𝑑 = √ [(𝛿𝜀2 − 𝛿𝜀3 )2 +(𝛿𝜀1 − 𝛿𝜀2 )2 + +(𝛿𝜀1 − 𝛿𝜀3 )2 ]
3
6
Validation of PLAXIS MoDeTo for Cowden Till
TXC z=8.2m
300
Deviator stress q (kN/m2)
250
200
150
Lab test
100
PLAXIS
50
0
0 2 4 6 8 10
Incremental axial strain (%)
Figure 2. TXC tests; comparison of experimental result and PLAXIS 3D simulations using the NGI-ADP
model. Triaxial test data from (Zdravković et al. 2019b).
The secant shear modulus degradation data are plotted in Fig. 3 (noting that the deviatoric strain in
these plots refers to incremental deviatoric strain with reference to the sample state at the end of the
𝐾0𝑡𝑜𝑡𝑎𝑙 compression stage). These data are compared in Fig. 3 with the secant shear modulus
degradation adopted in the ICFEP model that was employed during PISA to model the Cowden test
piles. The equation of this nonlinear stiffness degradation model (in terms of the tangent shear
modulus) is given as Equation 4 of Zdravković et al. 2019b as,
(13)
0.095
𝐺𝑡𝑎𝑛 = 1100𝑝′ (0.05 + 0.987 )
𝛿𝜀
1 + ( 𝑑⁄ −5 )
9.78 × 10
The comparisons in Fig 3 indicate that, for the 2.5m and 5m deep samples, the maximum shear strain
(𝐺𝑜 or 𝐺𝑢𝑟 ) plateau extends to larger strains than is implied by the PISA model (implemented in ICFEP).
This is due to the influence of the relatively high value of 𝐾0𝑡𝑜𝑡𝑎𝑙 in these two samples. At the start of
the axial compression stage, the axial stress is the minor principal stress; relatively large axial strains
are therefore required to mobilise the yield surface for the case where the axial stress becomes the
major principal stress. This influence of 𝐾0𝑡𝑜𝑡𝑎𝑙 is not present in the nonlinear elastic model used in the
7
Validation of PLAXIS MoDeTo for Cowden Till
ICFEP model. This extended plateau does not occur in the PLAXIS model for the 8.2m deep sample
because, in this case, the axial compression stage begins with a 𝐾0𝑡𝑜𝑡𝑎𝑙 of 1.0.
Gsec (MPa)
40.00
60.00
30.00
PISA
PISA 40.00
20.00
PLAXIS PLAXIS
10.00 20.00
0.00 0.00
0.00001 0.0001 0.001 0.01 0.1 1 0.00001 0.0001 0.001 0.01 0.1 1
Deviatoric strain (%) Deviatoric strain (%)
TXC z=5.0m
120.00
100.00
80.00
Gsec (MPa)
60.00
PISA
40.00
PLAXIS
20.00
0.00
0.00001 0.0001 0.001 0.01 0.1 1 10
Deviatoric strain (%)
Figure 3. Degradation of secant shear modulus, 𝐺𝑠𝑒𝑐 . (The data labelled ‘PISA’ refer to the assumed
non-linear elastic model employed in the PISA ICFEP 3D finite element calculations, Zdravković et
al. 2019b)
The geometry of the analysed piles is specified in Fig. 4 and Table 4. Comparisons between the PLAXIS
3D results and the field test data are shown in Figs. 5-7.
8
Validation of PLAXIS MoDeTo for Cowden Till
Ground level
Table 4. Geometric characteristics of piles tested in Cowden and simulated in PLAXIS 3D (from Byrne
et al. 2019)
9
Validation of PLAXIS MoDeTo for Cowden Till
45
40
35
30
Horizontal force (kN)
25
Field test
20 ICFEP
PLAXIS 3D
15
10
0 0.1 D
0 20 40 60 80 100 120
Ground level displacement (mm)
Figure 5. Pile CM2 (D=0.762m, L=2.24m): Comparison between the field data and PLAXIS 3D results.
(Field test data from Byrne et al. 2019; ICFEP data from Zdravković et al. 2019b)
The PLAXIS 3D simulations (Figs. 5-7) provide reasonable representations (similar in quality to the
ICFEP calculations developed during the PISA project) of the observed response of the test piles up to
the assumed ultimate limit state pile ground level displacement of 𝑣𝐺 = 0.1𝐷. The numerical response
of the (shorter) pile CM2 is stiffer than the field test data (Fig. 5), suggesting that the soil model
adopted for the superficial layers may not be optimal. However, the response of piles CM9 and CL1,
which extend to deeper layers, is accurately reproduced.
10
Validation of PLAXIS MoDeTo for Cowden Till
140
120
100
Horizontal force (kN)
80
Field test
60 ICFEP
PLAXIS 3D
40
20
0 0.1 D
0.00 20.00 40.00 60.00 80.00 100.00 120.00
Ground displacement (mm)
Figure 6. Pile CM9 (D=0.762m, L=3.98m): Comparison between the field data and PLAXIS 3D results.
Field test data from Byrne et al. 2019; ICFEP data from Zdravković et al. 2019b
2500
2000
Horizontal force (kN)
1500
Field test CL1
PLAXIS 3D
1000 ICFEP
500
0.1 D
0
0.00 50.00 100.00 150.00 200.00 250.00 300.00
Ground level displacement (mm)
Figure 7. Pile CL1 (D=2m, L=10.35m): Comparison between the field data and PLAXIS 3D results. Field
data from Byrne et al. 2019; ICFEP results from Zdravković et al. 2019b
11
Validation of PLAXIS MoDeTo for Cowden Till
The soil properties given as input in PLAXIS MoDeTo for the calibration of the 1D model, based on the
data in Table 3) are provided in Table 5. The PLAXIS MoDeTo 1D model is calibrated using the
‘calibration set’ specified Table 6. The selected parameter space spans the parameters for pile CM9
(although this particular pile geometry is not included in the calibration set). The models GeoDS_1 –
GeoDs_8 are generated automatically in PLAXIS MoDeTo.
Layer Top (m) Bottom Saturated unit G0 (kN/m2) su,top su,bottom 𝑲𝒕𝒐𝒕𝒂𝒍
𝟎
(m) weight (kN/m3) (kN/m2) (kN/m2)
1 0.000 -1.000 21.19 22684 50.15 106.8 1.500
2 -1.000 -2.000 21.19 39243 106.8 163.8 1.422
3 -2.000 -3.000 21.19 56361 163.8 148.1 1.362
4 -3.000 -4.000 21.19 73582 148.1 110.5 1.338
5 -4.000 -5.000 21.19 88053 110.5 113.8 1.283
6 -5.000 -6.000 21.19 100108 113.8 120.5 1.197
7 -6.000 -8.000 21.19 113837 120.5 128.0 1.077
8 -8.000 -10.000 21.19 142842 128.0 140.1 0.994
A PLAXIS 3D model for each of the calibration piles was generated and calculated in PLAXIS MoDeTo.
Each pile was subjected to a lateral force H that was sufficient for the pile ground level displacement,
𝑣𝐺 , to reach a value of between 2.3 × 𝐷/10 and 2.8 × 𝐷/10. This magnitude of displacement is
necessary for reliable determination of the ultimate resistance parameters for the 1D model. The
parameterisation procedure was then launched and the depth variation functions for the soil reactions
were computed.
The accuracy of the calibrated 1D model was initially checked by comparison with the calibration set.
Plots showing the 𝐻 − 𝑣𝐺 responses from the PLAXIS 3D model and the PLAXIS MoDeTo 1D model for
each of the calibration piles are shown in Figs. 9 and 10. The plots in the left column correspond to
𝑣𝐺 in the range 0 to D/10,000 (referred to as the ‘small displacement’ range); the plots in the right
column show the computed response for 𝑣𝐺 in the range 0 to D/10 (referred to as the ‘large
12
Validation of PLAXIS MoDeTo for Cowden Till
displacement’ range). The accuracy of the 1D model is quantified using an ‘accuracy metric’ which is
defined,
Aref H
H 3D FE results Aref
Adiff 3D FE results
Adiff
1D model 1D
model
vG D/10,000 vG
0.1D
Large displacement Small displacement
Figure 8: Accuracy metric
The accuracy metric η was calculated for each pile for both displacement ranges; these values are
listed in Table 7. The quality of fit metric is greater than 0.9, except for four small displacement cases,
in indicating a successful calibration.
Although a reasonable match between the 1D model and the 3D calibration data is obtained, a key
aspect of this particular ground model needs to be noted. The profile of undrained shear strength in
the top 5m of the soil at the Cowden site is highly variable with depth. The length of the piles in the
calibration set all lie in this highly non-uniform soil. As a consequence, the general shape of the
strength profile relevant to pile GeoDS_1 (length 2.25 m, shear strength increasing approximately
linearly with distance along the pile) is significantly different than, for example, pile GeoDS_8 (length
5.5m with a positive followed by a negative shear strength gradient along the pile). Calibrating the 1D
model over a significant range of shear strength profile ‘shapes’ presents a significant challenge for
the optimisation procedures in the 1D model. For current wind turbine applications, piles tend to be
of lengths between about 20m and 50m. In these cases (which are consistent with the calibration
space considered in the PISA project) the influence of any non-uniform surface layers is likely to be
less significant.
13
Validation of PLAXIS MoDeTo for Cowden Till
Figure 9. Comparison between the response of the PLAXIS 3D model and the response of the calibrated
1D model for piles GeoDS_1 to GeoDS_4
14
Validation of PLAXIS MoDeTo for Cowden Till
Figure 10. Comparison between the response of the PLAXIS 3D model and the response of the
calibrated 1D model for piles GeoDS_5 to GeoDS_8
15
Validation of PLAXIS MoDeTo for Cowden Till
Table 7. Accuracy metric for the calibration piles in the small and large displacement ranges
140
120
100
Horizontal force (kN)
80
60 Field
PLAXIS3D
40 1D model
20
0 0.1 D
0.00 20.00 40.00 60.00 80.00 100.00 120.00
Ground level displacement vg (mm)
Figure 11. Response of 1D model of Pile CM9 compared to the field rest results and the PLAXIS 3D data.
Field test data from Byrne et al. 2019
16
Validation of PLAXIS MoDeTo for Cowden Till
Table 8. Accuracy metrics for pile CM9 in the small and large displacement range
6. Conclusions
This document describes an application of the PLAXIS MoDeTo system to the analysis of the PISA test
monopiles at the Cowden test site. Initially, a calibration for the NGI-ADP model is developed, based
on geotechnical data collected during the PISA project. PLAXIS 3D is then employed, using the
calibrated NGI-ADP model, to analyses the monotonic loading performance of three of the PISA test
piles; the close comparison between the field test data and the PLAXIS 3D results confirms the validity
of the NGI-ADP model calibration.
A separate exercise is then conducted in which the PLAXIS MoDeTo is calibrated using a set of eight
calibration piles, based on the soil conditions at the Cowden site but not including the dimensions of
the piles that were actually tested. The calibrated model is then used to predict one of the PISA test
piles; the close fit between the PLAXIS MoDeTo response and the field data (accuracy metric values
greater than 0.9) demonstrates the effectiveness of the PLAXIS MoDeTo calibration process.
The analyses described above relate to an onshore site. For practical offshore design applications,
however, the site will be submerged. In these cases, to provide a correct representation of the
influence of any gap that forms between the pile and the ground, the submerged using weight should
be employed for the soil (rather than the bulk unit weight as employed in the calculations described
in this document).
References
Andersen, L., Jostad, H.P. (1999). Application of an anisotropic hardening model for undrained
response of saturated clay. In Proc. Numerical Models in Geomechanics - NUMOG VII, Graz,
Austria.
Brinkgreve, R.B.J., Kumarswamy, S. and Swolfs, W.M. (2018). PLAXIS manual 2018. Plaxis bv, Delft, the
Netherlands.
Byrne, B.W., McAdam, R.A., Burd, H.J., Beuckealers, W.J.A.P., Gavin, K., Houlsby, G.T., Igoe, D., Jardine,
R.J., Martin, C.M., Muir Wood, A., Potts, D.M., Skov Gretlund, J., Taborda, D.M.G. and
Zdravković, L. (2019) Monotonic laterally loaded pile testing in a stiff glacial clay till at Cowden
Géotechnique (in press).
Panagoulias, S., Brinkgreve, R.B.J., Zampich, L. (2018). PLAXIS MoDeTo manual 2018. Plaxis bv, Delft,
the Netherlands.
Potts, D. M., & Zdravković, L. (2001). Finite element analysis in geotechnical engineering:
application (Vol. 2). Thomas Telford.
Zdravković, L., Jardine, R.J., Taborda, D.M.G., Abadias, D., Burd, H.J., Byrne, B.W., Gavin, K., Houlsby,
G.T., Igoe, D., Liu, T., Martin, C.M., McAdam, R.A., Muir Wood, A., Potts, D.M., Skov Gretlund, J.
and Ushev, E. (2019a) Ground characterisation for PISA pile testing and analysis Géotechnique
(in press).
Zdravković, L, Taborda, D.M.G., Potts, D.M., Abadias, D., Burd, H.J., Byrne, B.W., Gavin, K., Houlsby,
G.T., Jardine, R.J., Martin, C.M., McAdam, R.A., Emil Ushev, E. (2019b) Finite element modelling
of laterally loaded piles in a stiff glacial clay till at Cowden. Géotechnique (in press).
17