Professional Documents
Culture Documents
2022 Hu Et Al.-Finite Element Analysis of The Lateral Load Response of Monopiles in Layered Sand
2022 Hu Et Al.-Finite Element Analysis of The Lateral Load Response of Monopiles in Layered Sand
2022 Hu Et Al.-Finite Element Analysis of The Lateral Load Response of Monopiles in Layered Sand
Abstract: Most of the existing design methods for laterally loaded monopiles were developed for uniform soil profiles. The application of
such design methods in cases involving layered sand profiles can lead to incorrect capacity estimations. In order to investigate the impact of
the soil layering on the response of monopiles to lateral loads and to develop a design method that is applicable to layered soil profiles, a series
of three-dimensional (3D) finite-element (FE) analyses were performed on laterally loaded monopiles embedded in layered sand profiles
using an advanced two-surface-plasticity constitutive model. The analyses take into consideration a wide range of pile geometries (pile
diameter, slenderness ratio, and wall thickness), load eccentricities, sand types and relative densities, and layered soil profiles. The study
examines the change of the lateral capacity of a monopile in a two-layer sand profile as the top layer thickness varies, as well as the impact of
the presence of a thin loose-sand layer in a dense-sand profile on the pile lateral capacity. Based on the results of these analyses, a set of design
equations is proposed to estimate the lateral capacity and load-rotation response of monopiles in layered sand profiles. The proposed method
accounts for the 3D pile–soil interaction and provides estimates of the lateral capacity that are in close agreement with those obtained from the
3D FE analyses. DOI: 10.1061/(ASCE)GT.1943-5606.0002745. © 2022 American Society of Civil Engineers.
for Ottawa sand and Toyoura sand were summarized by Loukidis and
Salgado (2009). Both sands were considered in the present study. The
Mudline steel pipe pile was modeled as a linear-elastic material with Young’s
Pile
(ground surface)
modulus Ep equal to 200 GPa and Poisson’s ratio ν p equal to 0.2.
L
The FE analyses were performed in Abaqus/Explicit (ABAQUS
B 2014). Since the selected advanced constitutive model is not avail-
able in Abaqus, it was coded as a user-defined material subroutine
(VUMAT) in Fortran (Han 2017). Fig. 2 shows an example mesh
Fig. 1. A typical monopile subjected to a lateral load H applied at a
prepared for the FE analysis of a 2-m-diameter, 10-m-long pipe pile
height of h from the mudline (ground surface).
embedded in a two-layer sand profile. Due to symmetry of the
boundary-value problem, only half of the whole soil-pile domain
analyses for uniform soil profiles directly into the p-y analyses of was modeled using 47,940 linear, 8-node, hexahedral elements
the same monopiles in a layered soil profile, the error was reduced with reduced integration. Because the results of FE analyses of
to 18%. Recently, Burd et al. (2020) developed p-y curves for mo- laterally loaded piles are less sensitive to mesh size (as detailed
nopiles based on the results of FE analyses and the PISA load tests in Hu et al. 2021) than those of FE analyses of axially loaded piles
and have indicated that the modified p-y curves proposed by them (Han et al. 2017b, 2021), a mesh with the smallest element size of
for uniform soil profiles can also be used for layered soil profiles. 4 cm was used in the analyses. The width and length of the soil
In this paper, we present results of a series of 3D FE analyses of domain were taken as 20 times of the pile diameter B, while the
laterally loaded monopiles in layered sand profiles using an ad- thickness of the soil domain was 2 times of the pile length L to
vanced two-surface-plasticity sand model. We performed the analy- avoid any boundary effects. Ideally, the steel pile–sand interface
ses for three typical layered sand profiles: a two-layer sand profile, should be modeled using the contact-pair approach (Han et al.
a thin loose-sand layer sandwiched by two dense-sand layers, and a 2018), allowing slippage between the pile and the surrounding
general three-layer sand profile. The analyses cover a wide range of sand. However, Hu et al. (2021) showed that the difference between
design scenarios, considering pile diameters B ranging from 1 to the lateral pile capacity obtained from FE analyses using the
10 m, slenderness ratios L=B ranging from 2 to 20, wall thickness contact-pair approach and that obtained by using the perfect-
tw ranging from 1=100B to 1=50B, load eccentricities h ranging contact approach is less than 2.5%. For this reason, the perfect-
from 15 to 30 m, and two types of sands (Ottawa sand and Toyoura contact approach was used in the present study, assuming the
sand) with relative density DR ranging from 40% to 95%. The pa- common nodes of the soil and the pile are tied to each other with
per aims to provide quantitative insights into the impact of soil respect to all degrees of freedom. The perfect-contact approach has
layering on the response of monopiles to lateral loads. Based on more often been used to model nondisplacement piles (Han et al.
the data produced from the FE analyses, design equations are pro- 2019; Salgado et al. 2017).
posed to obtain the load-rotation response of monopiles in layered The FE analysis starts with a geostatic step, in which equilib-
sand profiles. A web application has been developed that can be rium is obtained in the pile-soil domain subjected to gravity load-
used to easily apply the proposed method. ing, producing only negligible displacement or deformation in the
domain. More details regarding the geostatic step were provided
by Hu et al. (2021). After that, the lateral load is applied by impos-
Finite-Element Modeling ing a horizontal velocity of 6 mm=s at a height h above the mudline
(Fig. 1). The quantity h is known as the load eccentricity. To min-
The mechanical response of sand is highly nonlinear and is depen- imize any dynamic effect, the imposed horizontal velocity is
dent on the initial stress state and the relative density of the sand. smoothly increased from zero to 6 mm=s in the first 1 s, and then
The global response of a monopile to lateral loads is closely related it remains constant until the end of the loading. Since large-
to the local stress–strain response of sand at different locations diameter monopiles are mostly used offshore, the sands were con-
along the pile. It is critical to select a realistic constitutive model sidered to be fully saturated in the analyses. The effective unit
that is able to accurately capture sand behavior under different weight of the fully saturated sands was calculated from a value
Pile
Soil (Dunkirk sand) at the test site, the FE analysis was performed using
the model parameters for Ottawa sand, except that the value of the
slope M cc of the critical-state line (CSL) in q–p 0 space was set to
Fig. 2. Mesh configuration for the three-dimensional FE analysis of 1.28 for Dunkirk sand. Fig. 3 shows that the FE analysis closely
a pile with B ¼ 2 m, L ¼ 10 m, and h ¼ 15 m in a two-layer sand predicts the load-deflection response of the test pile. More details
profile with the thickness of the top layer d1 ¼ 3B. regarding the validation case study can be found in Hu et al. (2021).
Fig. 4 shows three typical layered soil profiles that are often
encountered in practice: a two-layer sand profile with a thickness
d1 for the top layer, a thin loose-sand layer with a thickness of
1B sandwiched by two dense-sand layers, and a general three-layer
sand profile consisting of three different sands (i.e., sand A, sand B,
and sand C in the figure). To develop design equations for lateral pile
capacity estimation, a series of FE analyses (summarized in Table 2)
were performed considering three pile diameters (1, 4, and 10 m), a
range of values (2–20) for the slenderness ratio L=B, two values (50
and 100) for the wall thickness ratio B=tw , eccentricity h ¼ 15 m,
and two types of two-layer sand profiles (DR ¼ 40% over 80%
and DR ¼ 80% over 40%). The developed design equations were
then verified against results of FE analyses that are different from
those used to develop the equations (summarized in Table 2).
Analysis Results
d1 d2
Sand A Sand A
Loose sand 1B
L Sand B
Sand C
B
Sand B
Fig. 4. The layered soil profiles considered in this paper: (a) two-layer sand profile; (b) three-layer sand that consists of a loose sand layer sandwiched
by two dense sand layers; and (c) general three-layer sand.
d1 DR=40% d1 DR=80%
DR=40%
DR=80%
(a) (b)
Fig. 5. Load-rotation curves for monopiles with B ¼ 4 m, L ¼ 40 m, tw ¼ 1=50B, and h ¼ 15 m in: (a) loose-over-dense Ottawa sand (40%–80%);
and (b) dense-over-loose Ottawa sand (80%–40%) with varying top-layer thicknesses d1 .
H 80% of the pile in a uniform dense sand (DR ¼ 80%) when the same as that in a uniform dense (DR ¼ 80%) sand, which is shown
top-layer thickness d1 approaches zero. If the thickness d1 of as the top curve in Fig. 5(a). As the thickness d1 of the top loose
the top loose sand layer were to approach infinity, then the two- sand layer increases, the load-rotation response of the pile becomes
layer sand profile would transition from a uniform dense sand softer and approaches that of the pile in a uniform loose (DR ¼
to a uniform loose sand, and, consequently, H40%−80% would de- 40%) sand deposit [shown as the bottom curve in Fig. 5(a)].
crease from H80% to the pile lateral capacity H40% that would result The load-rotation curve of the pile almost coincides with that for
for the pile installed in the uniform loose sand. To investigate how a uniform loose sand when d1 reaches 5B. Further decrease in the
the pile lateral capacity in a two-layer sand profile changes as a relative density (from 80% to 40%) of the sand below a depth of 5B
function of the top-layer thickness d1 , we performed a series of has a negligible effect on the lateral load response of the pile. Sim-
FE analyses simulating the response of monopiles to lateral loads ilar trends were observed for the same pile (B ¼ 4 m, L ¼ 40 m,
in two representative two-layer sand profiles (DR ¼ 80% over h ¼ 15 m) in 80%–40% Ottawa sand, except that the load-rotation
DR ¼ 40% and DR ¼ 40% over DR ¼ 80%) with varying d1 . response evolves from the lower bound (the case of the uniform
loose sand with DR ¼ 40%) to the upper bound (the case of the
Load-Rotation Response uniform dense sand with DR ¼ 80%) as d1 increases [Fig. 5(b)].
When laterally loaded, a short, stiff pile experiences mainly rigid- Short Pile in Two-Layer Sand Profile. Similarly, the load-
body rotation and rigid-body translation, whereas a long, flexible rotation response of a pile with L=B ¼ 5 (B ¼ 4 m, L ¼ 20 m,
pile undergoes bending in addition to rigid-body rotation and trans- h ¼ 15 m) in a 40%–80% sand profile transitions from that of a
lation. The different deformation modes for the two types of piles uniform dense (DR ¼ 80%) sand to that of a uniform loose (DR ¼
result in their different load responses to lateral loads. Therefore, 40%) sand as d1 increases [Fig. 6(a)]. Only when d1 reaches 5B,
both short, stiff piles and long, flexible piles are considered sepa- meaning that the relative density of sand along the entire length of
rately in this section. the pile becomes 40%, does the load-rotation curve become almost
Long Pile in Two-Layer Sand Profile. Fig. 5(a) compares the re- identical to that for the uniform loose sand. Similar results were
sponses (in terms of lateral load versus pile rotation at the mudline) observed for the same pile in an 80%–40% Ottawa sand profile
of a laterally loaded monopile with L=B ¼ 10 (B ¼ 4 m, L ¼ with varying d1 , as shown in Fig. 6(b). Comparing the results
40 m, h ¼ 15 m) in 40%–80% Ottawa sand with the top-layer for the long pile (L=B ¼ 10) and the short pile (L=B ¼ 5) in Figs. 5
thickness d1 ranging from 0 to infinity (note that as d1 approaches and 6, the gap between the upper and lower bounds within which
infinity, the soil profile is modeled as a uniform sand with DR ¼ the load rotation curve varies for the short pile is greater than that
40%). When d1 ¼ 0, the load-rotation response of the pile is the for the long pile. The gradual change of thickness of the top layer
d1 DR=40% d1 DR=80%
DR=80% DR=40%
(a) (b)
Fig. 6. Load-rotation curves for monopiles with B ¼ 4 m, L ¼ 20 m, tw ¼ 1=50B, and h ¼ 15 m in: (a) loose-over-dense Ottawa sand (40%–80%);
and (b) dense-over-loose Ottawa sand (80%–40%) with varying top-layer thicknesses d1 .
d1 DR=40%
DR=80%
d1 DR=80%
DR=40%
(a) (b)
Fig. 7. The relationship between the lateral capacity ratio H=Hjd1¼0 and the normalized top-layer thickness d1 =B for long piles: (a) pile with
B ¼ 4 m, L ¼ 40 m, tw ¼ 1=50B, and h ¼ 15 m in 40%–80% Ottawa sand; and (b) pile with B ¼ 4 m, L ¼ 40 m, tw ¼ 1=50B, and h ¼ 15 m
in 80%–40% Ottawa sand.
has a greater impact on the lateral capacity of a short pile than of a Long Pile in Two-Layer Sand Profile. Figs. 7(a and b) show the
long pile. lateral capacity ratio H=Hjd1¼0 for a 4-m-diameter pile with L=B ¼
10 as a function of the normalized top-layer thickness d1 =B in two
Lateral Load Capacity Ratio two-layer sand profiles. In Fig. 7(a), the pile lateral capacity Hjd1¼0
To quantify the change of the lateral capacity H of a monopile in a when d1 ¼ 0 is equal to the lateral capacity H80% of the pile in a
two-layer sand profile as the top layer thickness d1 increases from uniform dense sand (DR ¼ 80%). Starting with a value of 1, the
zero to infinity, it is useful to define a lateral capacity ratio lateral capacity ratio H=Hjd1¼0 monotonically decreases as the
H=Hjd1¼0 using the pile lateral capacity Hjd1¼0 when d1 ¼ 0 as thickness d1 of the top, loose sand layer increases from 0 to infinity.
a reference. The lateral pile capacities H and Hjd1¼0 are defined Theoretically, when d1 → þ∞, the sand profile is a uniform loose
corresponding to a pile rotation θ ¼ 0.5° or θ ¼ 1° at the mudline; (DR ¼ 40%) sand, and the lateral pile capacity H is that (H 40% ) in
these are commonly used serviceability limit state criteria for the the uniform loose sand. Accordingly, the lateral capacity ratio
design of laterally loaded monopiles (Arany et al. 2017; Doherty H=Hjd1¼0 approaches a limit value that is equal to H40% =H 80% .
and Gavin 2012; Hu et al. 2021). For the case (B ¼ 4 m and L=B ¼ 10) shown in Fig. 7(a), the limit
DR=80%
Downloaded from ascelibrary.org by "Indian Institute of Technology, Roorkee" on 05/02/23. Copyright ASCE. For personal use only; all rights reserved.
d1 DR=80%
DR=40%
(a) (b)
Fig. 8. The relationship between the lateral capacity ratio H=Hjd1¼0 and the normalized top-layer thickness d1 =B for short piles: (a) pile with
B ¼ 4 m, L ¼ 20 m, tw ¼ 1=50B, and h ¼ 15 m in 40%–80% Ottawa sand; and (b) pile with B ¼ 4 m, L ¼ 20 m, tw ¼ 1=50B, and h ¼ 15 m
in 80%–40% Ottawa sand.
values of H=Hjd1¼0 are 0.73 and 0.7 (shown as horizontal reference where α defines the rate at which the H=Hjd1¼0 ðH=H B Þ ratio ap-
lines in the figure) for pile rotations θ ¼ 0.5° and θ ¼ 1° at the mud- proaches its limit value Hjd1¼∞ =Hjd1¼0 (¼ HA =HB , with H A and
line, respectively. In the limit, the pile lateral capacity H decreases HB being the pile lateral capacity in uniform sand A and uniform
by about 30% when the thickness of the top loose sand layer ap- sand B, respectively). Despite small differences, the H=Hjd1¼0 ratio
proaches infinity. Fig. 7(a) shows that the lateral capacity ratio for piles with slenderness ratios L=B ≥ 5 approaches the limit value
H=Hjd1¼0 is close to its asymptotic value when d1 reaches the 4 − Hd1¼∞ =Hd1¼0 at similar rates, reaching values that are sufficiently
5B range, suggesting that for d1 values greater than 5B, there is close to the limit value when d1 =B ¼ 5 (Figs. 7 and 8). Thus, a
minimal impact on the lateral capacity of the pile. unique value for the rate α can be used for piles with L=B ≥ 5.
A similar trend of H=Hjd1¼0 as a function of d1 =B is observed In contrast, for shorter piles—with L=B < 5—the H=Hjd1¼0 ratio
for the pile (B ¼ 4 m and L=B ¼ 10) in an 80%–40% sand profile approaches the limit value when the top-layer thickness d1 becomes
[Fig. 7(b)]. Note that the limit value of H=Hjd1¼0 ð¼ H 80% =H 40% Þ greater than the pile length L (i.e., d1 =B ≥ L=B, with L=B < 5). For
for the 80%–40% sand, as d1 approaches infinity, is the reciprocal example, when L ¼ 3B, the H=Hjd1¼0 ratio approaches the limit
of the limit value (¼ H40% =H80% ) for the 40%–80% sand. value when d1 ¼ 3B (d1 =B ¼ 3). This requires the H=Hjd1¼0 ratio
Short Pile in Two-Layer Sand Profile. As shown in Figs. 8(a to approach the limit value at a greater rate as the pile slenderness
and b), the trends of the lateral capacity ratio H=Hjd1¼0 versus ratio L=B decreases further. Based on the FE results for both Ot-
d1 =B for a short pile (B ¼ 4 m and L=B ¼ 5) are similar to those tawa sand and Toyoura sand, the following equation was obtained
for the long pile (L=B ¼ 10): the pile lateral capacity H decreases or for α:
increases monotonically with increasing top-layer thickness d1 . The 8
lateral capacity H of the short pile is more sensitive to the change of >
< −0.23; when L=B ≥ 5
top-layer thickness d1 than the long pile [shown in Figs. 7(a and b)]. α¼ L ð2Þ
For example, the lateral capacity H (for θ ¼ 1°) of the 4-m-diameter >
:1 −1.07 þ 1 ; when L=B < 5
B
short pile (L=B ¼ 5) in an 80%–40% sand profile increases by
about 2.3 times as the thickness of the top dense sand layer
(DR ¼ 80%) increases from zero to 5B, whereas it increases by only Table 3 compares the limit values of the H=Hjd1¼0 ratios ob-
1.5 times for the 4-m-diameter long pile (L=B ¼ 10). tained from the FE analyses for piles of different slenderness ratios
Figs. 7 and 8 show a similar relationship between the lateral in two-layer sand profiles. The limit value of the H=Hjd1¼0 ratio is
capacity ratio H=Hjd1¼0 versus the normalized top-layer thickness the ratio of the lateral capacities of the pile in the uniform sand
d1 =B for long and short piles in two-layer sand profiles (sand A profiles with DR values equal to those of the two-layer sand profile
over sand B): H=Hjd1¼0 varies from 1 to the limit value being considered. The limit values (¼ Hjd1¼∞ =Hjd1¼0 ) are signifi-
H d1¼∞ =Hjd1¼0 as d1 increases from 0 to infinity. Thus, an equation cantly different from 1 for piles with slenderness ratios less than 5,
is proposed to describe the dependency of the lateral capacity ratio suggesting that a change of the relative density of the uniform sand
H=Hjd1¼0 on the top-layer thickness d1 for two-layer (A over B) has a great impact on their lateral capacity. Taking the 1-m-diameter
sand cases: piles as an example, the lateral capacity H of the pile with L=B ¼
20 for θ = 0.5° decreases by 23% (corresponding to a ratio
H Hjd1¼∞ Hjd1¼∞ Hjd1¼∞ =Hjd1¼0 ¼ 0.77 in the table) but that of the pile with L=B ¼
¼ þ 1− exp½αðd1 =BÞ1.5 5 decreases by 65% (corresponding to a ratio Hjd1¼∞ =Hjd1¼0 ¼
Hjd1¼0 Hjd1¼0 Hjd1¼0
0.35 in the table) when the relative density changes from 80%
H H
¼ A þ 1 − A exp½αðd1 =BÞ1.5 ð1Þ to 40%. Figs. 9(a–d) compare the profiles of the pile lateral deflec-
HB HB tion and unit lateral soil resistance for the piles with slenderness
4 20 5 0.46 0.43 2.18 2.34 be obtained from Eqs. (1) and (2). We have developed, in Hu et al.
1 5 5 0.35 0.33 2.85 3.07 (2021), the equations that can be used to estimate the lateral
Note: The limit values (=H40%/H80%) for the 40%–80% sand profile are the capacities H A and H B in uniform sand profiles A and B, as sum-
reciprocals to those (=H80%/H40%) for the 80%–40% sand profile. Load marized in Fig. 15 [Eqs. (7)–(10)]. The equations [Eqs. (1) and (2)]
eccentricity h ¼ 15 m for all cases in the table. tw =B ¼ 1=100 for piles for the estimation of the lateral capacity ratio were developed based
with B ¼ 10 m, and tw =B ¼ 1=50 for piles with B < 10 m. on results of FE analyses of monopiles in 40%–80% and 80%–40%
sand profiles. To verify the accuracy of the proposed equations and
ratios of 5 and 20. Because of the bending of the pile with their applicability to layered sand profiles with other combinations
L=B ¼ 20, the effective pile length—the distance from the ground of relative densities, the pile lateral capacities were calculated using
surface to the zero-deflection point—is greater for the pile in the the proposed equations [Eqs. (7)–(10) and Eqs. (1) and (2)] and
loose sand than in the dense sand [Fig. 9(b)]. This allows the pile in compared with the FE results for a number of test cases that were
the loose sand to mobilize positive soil resistance along a longer not used in the development of Eqs. (1) and (2); the results for both
Deflection (mm)
-20 -10 0 10 20 30 40
0 0
1 1 DR = 80%
DR = 80% DR = 40%
DR = 40%
2 2
3 3
4 4
5 5
6 6
(a) (b)
2 2
3 3
4 4
5 5
6 6
(c) (d)
Fig. 9. Profiles of pile deflection and unit soil resistance for short and long piles for θ ¼ 0.5°: (a) pile deflection for the short pile (B ¼ 1, L ¼ 5 m,
tw ¼ 1=50B, and h ¼ 15 m); (b) pile deflection for the long pile (B ¼ 1, L ¼ 20 m, tw ¼ 1=50B, and h ¼ 15 m); (c) unit soil resistance for the short
pile (B ¼ 1, L ¼ 5 m, tw ¼ 1=50B, and h ¼ 15 m); and (d) unit soil resistance for the long pile (B ¼ 1, L ¼ 20 m, tw ¼ 1=50B, and h ¼ 15 m).
Note: H estimated is the lateral capacity calculated using the proposed equations; H FE is the lateral capacity obtained from the 3D FE analysis; and Error is
defined as ðH estimated − H FE Þ=H FE ; tw =B ¼ 1=100 for piles with B ¼ 10 m, and tw =B ¼ 1=50 for piles with B < 10 m.
160000
Proposed equation
140000
3D FE analysis
60000
d1 DR=40% 40000
d1 DR=80%
DR=80% 20000 DR=40%
0
0 0.5 1 1.5
Fig. 10. Lateral load-rotation responses obtained from FE analyses and the proposed equations for: (a) a 10-m-diameter pile with L ¼ 80 m and
tw ¼ 1=100B, and h ¼ 15 m in 40%–80% Ottawa sand; and (b) a 10-m-diameter pile with L ¼ 50 m and tw ¼ 1=100B, and h ¼ 15 m in 80%–40%
Ottawa sand.
Ottawa sand and Toyoura sand are shown in Table 4. The predic- of laterally-loaded monopiles (Arany et al. 2017; Doherty and
tions from the proposed equations match closely the FE results. Gavin 2012; Hu et al. 2021); a pile rotation θ ranging from 2°
According to Hu et al. (2021), the lateral pile capacity H in- to 5° (Ahmed and Hawlader 2016; Klinkvort and Hededal 2013;
creases with increasing pile rotation θ (in degrees) at the mudline Luo et al. 2018) or a pile deflection at mudline equal to 10%B
following: (Zdravković et al. 2015), which corresponds to a pile rotation θ
of 1.6° to 2° for the cases considered in this study, may be used
θ as ultimate limit state (ULS) criteria. With Eqs. (3) and (4), a full
H¼ ð3Þ
k þ ηθ load-rotation curve (for θ ≤ 2°) can be obtained for the designer to
consider both SLS and ULS criteria. For convenient application of
where the coefficients k and η control the magnitude of H and the the proposed equations to calculate the lateral pile capacity and to
rate at which H increases with respect to θ. The initial stiffness obtain the load-rotation curve, a web-based application was devel-
of the pile at θ ¼ 0° is equal to 1=k. The two coefficients can oped using the design equations proposed in this paper. The link to
be determined from the values of the lateral pile capacity Hj0.5° the web-based application is provided in the Appendix.
and Hj1° for θ ¼ 0.5° and θ ¼ 1° using
DR=80% DR=80%
(a) (b)
d2 DR=40% 1B d2 DR=40% 1B
DR=80% DR=80%
(c) (d)
Fig. 11. The effect of the loose-thin-layer on load-rotation curves and the corresponding relationships between pile lateral capacity ratio H=H uniform
versus the d2 =B ratio for: (a and c) the pile with B ¼ 4 m, L ¼ 40 m, and tw ¼ 1=50B; and (b and d) the pile with B ¼ 4 m, L ¼ 20 m, and
tw ¼ 1=50B.
Fig. 11(a) compares the load-rotation responses of a pile when the thin loose sand layer is located near the ground surface
(B ¼ 4 m, L ¼ 40 m, and h ¼ 15 m) in soil profiles with the thin (i.e., d2 ≤ 2B), and the capacity ratio H=H80% reaches 1 when d2
loose sand layer located at different depths d2 from the ground sur- becomes greater than 5B. The lateral capacity of the pile is not af-
face to the bottom of the loose-sand layer. The lateral load-rotation fected by the presence of the loose sand layer that is located at
response of the same pile in a uniform dense sand (DR ¼ 80%) is depths greater than 5B.
also plotted in the figure for reference. The lateral capacity of the As shown in Figs. 11(b and d), the presence of the loose sand
pile in dense sand with a loose sand layer in the middle is slightly layer reduces the lateral capacity of the pile with L=B ¼ 5 more
less than that in the uniform dense sand when d2 varies from 1B to than that of the pile with L=B ¼ 10 [Figs. 11(a and c)]. The loose
3B. As the value of d2 becomes greater than 4B, the load-rotation sand layer has the greatest impact on the lateral capacity H of the
response of the pile in the layered soil is almost identical to that in pile with L=B ¼ 5 when it is located near the ground surface
the uniform dense sand. (within a depth of 2B) or near the pile base. This is because the
To quantify the decrease of the lateral load capacity by the pres- relatively short pile (L=B ¼ 5) undergoes mainly rigid-body rota-
ence of the loose sand layer, the lateral capacity ratio is defined as tion when subjected to lateral loads, experiencing the largest
the pile lateral capacity H in the layered soil profile (consisting of a amount of lateral deflection near the ground surface and near
thin loose sand layer in dense sand) to that (H80% ) in uniform dense the pile base and mobilizing most of its lateral resistance from these
sand. As shown in Fig. 11(c), the decrease of the lateral load capac- two regions. Fig. 11(d) shows that the lateral capacity of the pile
ity of the pile due to the presence of the loose sand layer is insig- with L=B ¼ 5 is not affected by the loose sand layer located below
nificant overall: the lateral capacity H decreases by about 10% the base of the pile.
A A B B
B A A B
= − +
C A A C
Fig. 12. Estimation of the lateral capacity H A-B-C of a monopile in a three-layer sand (A-B-C) from the linear combination of the lateral capacities
H A-A-A ; H B-A-A , H B−B−C in three uniform and two-layer sand profiles.
10 Layer 3 2 10 B=4m 2
15 15 L = 40 m
h = 15 m
20 20
Rsoil II−Rsoil III Rsoil II−Rsoil III
25 Rsoil IV 25
2 2 Rsoil IV
30 30
35 B=4m 35
L = 40 m
40 h = 15 m 40
Ottawa sand DR(thickness)
45 45
DR (thickness)
50 Soil profile I: 80%(1B)-40%(1B)-80% 50
55 55
(a) (b)
Fig. 13. Profiles of the soil resistance mobilized for a monopile (B ¼ 4 m, L ¼ 40 m, tw =B ¼ 1:50, h ¼ 15 m) in: (a) three-layer Ottawa sand
80%ð1BÞ − 40%ð1BÞ − 80%; and (b) uniform dense sand with DR ¼ 80%, two-layer sand profiles 40%ð1BÞ − 80% and 40%ð2BÞ − 80%. The
soil resistance is obtained for a pile rotation θ ¼ 0.5° at the mudline.
Lateral Load Capacity of Piles in Three-Layer Sands HA-B-C ¼ HA-A-A − HB-A-A þ HB-B-C
Recognizing that the lateral capacity of a monopile in a layered- ¼ ðR1A þ R2A þ R3A Þ − ðR1B þ R2A þ R3A Þ
sand profile equals the summation of the lateral soil resistances mo-
bilized along the entire pile length in all sand layers, we propose a þ ðR1B þ R2B þ R3C Þ
novel method that can be used to calculate the pile lateral capacity ¼ R1A þ R2B þ R3C ð6Þ
in three-layer (or multiple-layer) sand profiles. As illustrated in
Fig. 12, the lateral capacity HA-B-C of a monopile in a three-layer With Eqs. (7)–(12), summarized in Fig. 15, we can calculate the
sand profile (A-B-C) is the summation of the soil resistances R1A , pile lateral capacities (H A-A-A ; HB-A-A , and HB-B-C ) in uniform and
R2B , and R3C mobilized in these three layers two-layer sand profiles and further obtain the lateral capacity
HA−B−C of the pile in the three-layer sand profile.
H A-B-C ¼ R1A þ R2B þ R3C ð5Þ To demonstrate the use of the proposed method [Eq. (6) and
Fig. 12], a case of a 4-m-diameter, 40-m-long pile in a three-layer
sand is investigated. The three-layer sand consists of a 1-B-thick
Assuming that the lateral soil resistances for a pile mobilized in dense sand (DR ¼ 80%) top layer, a 1-B-thick loose sand (DR ¼
the same layer of the same sand in two different soil profiles are 40%) middle layer, and a dense sand (DR ¼ 80%) bottom layer. For
approximately the same (e.g., the soil resistances R2A mobilized in convenience, this soil profile is named as 80%ð1BÞ − 40%ð1BÞ −
the middle layer of soil profile II and soil profile III in Fig. 12 are 80%. Fig. 13(a) shows the unit lateral soil resistance mobilized
approximately the same), the pile lateral capacity HA-B-C in a three- along the pile length in the three-layer sand when the pile rotation
layer sand (A-B-C) can be obtained as a linear combination of the θ ¼ 0.5° at the mudline. The lateral capacity H 80%−40%−80% of the
lateral pile capacities in three uniform and two-layer sand profiles pile in the three-layer sand profile is equal to the integration of the
(A-A-A, B-A-A, and B-B-C, as shown in Fig. 12) unit soil resistance along the pile length (represented by the area
Note: B = pile diameter; L = pile length; h = load eccentricity (the distance from the lateral load to the mudline); θ = the mudline pile rotation corresponding to
which the lateral capacity H of the pile is determined; and Error is defined as ðH estimated − H FE Þ=H FE .
between the unit soil resistance curve and the vertical axis at zero embedded in different soil profiles for θ ¼ 0.5° and 1° were calcu-
soil resistance). lated using the proposed method and compared with those ob-
According to the proposed method [Eq. (6) and Fig. 12], the pile tained from the FE analyses. As Table 5 shows, the proposed
lateral capacity H80%−40%−80% in the three-layer sand profile can method provides close estimates of the pile lateral capacity in
be obtained as a linear combination of the lateral pile capacities three-layer sand profiles. The lateral load rotation curves for two
H 80% in a uniform dense sand (DR ¼ 80%) and H40%ð1BÞ−80% cases listed in Table 5 were also obtained using the proposed
and H 40%ð2BÞ−80% in the two-layer sand profiles: H80%−40%−80% ¼ method and compared with those obtained from 3D FE results, as
H 80% − H 40%ð1BÞ−80% þ H 40%ð2BÞ−80% . Fig. 13(b) shows the pro- shown in Figs. 14(a and b).
files of the unit lateral soil resistance mobilized along the pile
length in these three soil profiles. The unit soil resistance in soil
profile II (DR ¼ 80%) is shown as the thin solid black line, whereas Design Method and Example
the unit soil resistance in soil profile III [40%ð1BÞ − 80%] is shown The proposed method, which consists of a set of equations (sum-
as the dotted line. Because the difference between these two lines marized in Fig. 15), can be used to estimate the lateral capacity
below a depth of 1B is negligible, the difference in the pile and the load-rotation response of monopiles in uniform, two-layer,
capacities mobilized in soil profiles II and III can be represented and three-layer sand profiles under fully saturated conditions.
by zone 1—the shaded area with vertical stripes—in Fig. 13(b). The method accounts for the pile geometry (pile length and cross
The pile capacity H40%ð2BÞ−80% mobilized in soil profile IV section), load eccentricity, and soil profile characteristics (sand type,
(40%ð2BÞ − 80%) is represented by the shaded area with horizontal layer profile, and relative density). The main advantage of the pro-
stripes (zone 2) in Fig. 13(b). As shown in Fig. 13(a), the super- posed method is that an engineer does not need to have access to a
position of the two zones (zone 1 and zone 2) is almost identical to finite element code with an advanced constitutive model imple-
the area representing the pile capacity H80%ð1BÞ−40%ð1BÞ−80% mobi- mented in it to obtain equivalent results. A design example is pro-
lized in the three-layer sand 80%ð1BÞ − 40%ð1BÞ − 80%). vided below to illustrate the steps required for calculation of the
To verify the accuracy of the proposed method, the lateral lateral capacity and load-rotation curve of a monopile using the pro-
capacity for piles of different diameters and slenderness ratios posed method. A 10-m-diameter, 80-m-long monopile with a wall
(a) (b)
Fig. 14. The lateral load–rotation curves obtained from the proposed method and the FE analyses for (a) a 10-m-diameter monopile with L ¼ 80 m
and h ¼ 15 m in an 80%ð1BÞ − 40%ð3BÞ − 65% Ottawa sand; and (b) a 4-m-diameter monopile with L ¼ 40 m and h ¼ 30 m in a 40%ð2BÞ −
60%ð2BÞ − 80% Ottawa sand.
DR I 0.25
p L crit is the critical pile length beyond
L crit I p0.25 12.5 3ln 37.9
Step 1: Calculate the 0.5° 100% LR which the pile lateral capacity no longer
critical pile length Lcrit (7) increases.
DR Ip0.25
for = 0.5 and 1 . L crit I p0.25 12 3.3ln 40.3 is the pile rotation at the mudline.
1° 100% LR Ip B 4-Bi4)/64; Bi = inner diameter,
0.72
Step 2: Calculate the H crit Ip Hcrit is the pile lateral capacity when the
a DR, h 4 b
critical pile lateral pA L2R LR pile length is equal to or greater than Lcrit.
(8)
capacity Hcrit for = h DR h The values of the coefficients a1, a2, a3,
Downloaded from ascelibrary.org by "Indian Institute of Technology, Roorkee" on 05/02/23. Copyright ASCE. For personal use only; all rights reserved.
1 L
sin 0.5 , when L L crit
Step 3: Calculate the H Hcrit 2 L crit 2
lateral capacity ratio 1 , when L L crit (9)
H/Hcrit for = 0.5 and
DR, I p 0.084 L crit I p0.25 1.2
1 0.5°
DR, I p 0.08 L crit I p0.25 1.25
1°
Step 4: Calculate the H The critical capacity Hcrit is obtained
H 0.5° H crit 0.5°
lateral capacity H in a Hcrit 0.5° from Eq. (8)
(10)
uniform sand for = H The lateral capacity ratio H/Hcrit is
H 1° H crit 1°
0.5 and 1 Hcrit 1° obtained from Eq. (9)
Two-layer sand profile
HA B HA HA 1.5 HA-B
1 exp d1 B d1 is the thickness of the top layer
Calculate the pile lateral HB HB HB d1 A HA and HB are the lateral capacities of the
capacity HA-B for a two- 0.23 , when L B 5 (11)
pile in uniform sand A and B,
layer (A over B) sand L B
1 1.07
B
1 , when L B 5 respectively (obtained from step 1-4).
Three-layer sand
HA B C HA A A HB A A HB B C
HA-A-A HB-A-A HB-B-C HA-A-A is the pile lateral capacity for a
Calculate the pile lateral HA-B-C
A A B B uniform sand A.
capacity HA-B-C for a
B A A B (12) HB-A-A and HB-B-C are the pile lateral
three-layer (A-B-C)
sand C
=
A
−
A
+
C
capacities for two-layer sand profile
(obtained from (11)
Soil profile I Soil profile II Soil profile III Soil profile IV
Load-rotation curve
Note: The equations for uniform sand were developed by Hu et al. (2021)
Fig. 15. Steps to calculate the lateral capacity of monopiles in uniform and layered sand profiles. (Data from Hu et al. 2021.)
thickness tw ¼ 1=100B (I p ¼ 38.11 m4 ) is loaded at h ¼ 15 m in Step 1.2. Calculate the critical pile lateral capacity H crit using
an 80%ð1BÞ − 40%ð3BÞ − 65% three-layer Ottawa sand profile. Eq. (8) in Fig. 15 and the coefficients in Table 6:
Step 1 Calculate the lateral capacities H80% ; H 40% , and H65% in For θ ¼ 0.5° at the mudline:
uniform sand profiles (calculation steps for H 80% are shown as an 0.72
example): Ip
Hcrit j0.5° ¼ aðDR ; hÞj0.5° 4 þ bj0.5° pA L2R
Step 1.1. Calculate the pile critical length Lcrit using Eq. (7) LR
in Fig. 15: ¼ f½ð−0.95 × 15 þ 49.00Þ × 0.8 þ ð−0.34Þ
For θ ¼ 0.5° at the mudline:
0.25 × 15 þ 25.80 × 38.110.72 þ ð−0.01Þg × 100ðkNÞ
D Ip
Lcrit j0.5° ¼ −12.5 R − 3 ln þ 37.9 I 0.25
p ¼ 66,689ðkNÞ
100% LR
¼ ½−12.5 × 0.8 − 3 × lnð38.110.25 Þ þ 37.9 × 38.110.25 ðmÞ For θ ¼ 1° at the mudline, H crit j1° ¼ 122,731 kN.
¼ 62.5ðmÞ Step 1.3. Calculate the lateral capacity ratio H=H crit using
Eq. (9) in Fig. 15:
For θ ¼ 1° at the mudline, Lcrit j1° ¼ 68.8 m. Since L > Lcrit , then ðH=H crit Þj0.5° ¼ 1, and ðH=Hcrit Þj1° ¼ 1.
= – +
DR = 65% DR = 80% DR = 80% DR = 65%
Fig. 16. Estimation of the pile lateral capacity H 80%ð1BÞ−40%ð3BÞ−65% in an 80%ð1BÞ − 40%ð3BÞ − 65% sand profile from the linear combination of
the pile lateral capacity H 80% ; H 40%ð1BÞ−80%, and H 40%ð4BÞ−65% .
profiles. The method proposed in this study is only applicable nology Conference.
Ahmed, S. S., and B. Hawlader. 2016. “Numerical analysis of large-
to: (1) monotonic lateral loading conditions; and (2) input param-
diameter monopiles in dense sand supporting offshore wind turbines.”
eters in the range of those considered in the analyses. Note that the Int. J. Geomech. 16 (5): 04016018. https://doi.org/10.1061/(ASCE)GM
effects of pile installation and cyclic loading on the static lateral .1943-5622.0000633.
capacity of monopiles are not considered in the development of Anderson, J. B., F. C. Townsend, and B. Grajales. 2003. “Case history
the equations proposed in this paper. evaluation of laterally loaded piles.” J. Geotech. Geoenviron. Eng.
129 (3): 187–196. https://doi.org/10.1061/(ASCE)1090-0241(2003)
129:3(187).
Appendix. Web Application API (American Petroleum Institute). 2014. API recommended practice
2A-WSD. Washington, DC: API.
The web application can be found through the link: Arany, L., S. Bhattacharya, J. Macdonald, and S. J. Hogan. 2017. “Design
https://script.google.com/macros/s/AKfycbwveAcQBPbghcXK of monopiles for offshore wind turbines in 10 steps.” Soil Dyn. Earth-
quake Eng. 92 (Jan): 126–152. https://doi.org/10.1016/j.soildyn.2016
9WHrktAFisBe8OfDIRvU-WfP-nLwT60XXi3i923u/exec.
.09.024.
Burd, H. J., et al. 2019. “PISA design model for monopiles for offshore
wind turbines: Application to a marine sand.” Géotechnique 70 (11):
Data Availability Statement 1048–1066. https://doi.org/10.1680/jgeot.18.P.277.
Burd, H. J., et al. 2020. “Application of the PISA design model to monop-
Data that support the findings of this study are available from the iles embedded in layered soils.” Géotechnique 70 (11): 1067–1082.
corresponding author upon reasonable request. https://doi.org/10.1680/jgeot.20.PISA.009.
Byrne, B. W., et al. 2019a. “PISA design methods for offshore wind turbine
monopiles.” In Proc., Offshore Technology Conf. Boca Raton, FL: CRC
Notation Press.
Byrne, B. W., H. J. Burd, L. Zdravković, R. A. McAdam, D. M. G.
The following symbols are used in this paper: Taborda, G. T. Houlsby, R. J. Jardine, C. M. Martin, D. M. Potts, and
K. G. Gavin. 2019b. “PISA: New design methods for offshore wind
B = outer diameter of pile; turbine monopiles.” Revue Française de Géotechnique 158 (1): 3.
DR = relative density of sand; https://doi.org/10.1051/geotech/2019009.
d1 = the depth from the ground surface to the bottom of the Byrne, B. W., R. Mcadam, H. J. Burd, G. T. Houlsby, and C. M. Martin.
first soil layer; 2015. “New design methods for large diameter piles under lateral
d2 = the depth from the ground surface to the bottom of the loading for offshore wind applications.” In Proc., 3rd Int. Symp. on
second soil layer; Frontiers in Offshore Geotechnics, 705–710. Pathumthani, Thailand:
Asian Institute of Technology. https://doi.org/10.1201/b18442-96.
Ep = Young’s modulus of pile; Choi, Y. S., D. Basu, R. Salgado, and M. Prezzi. 2014. “Response of
H = pile lateral capacity; laterally loaded rectangular and circular piles in soils with properties
HA-B = pile lateral capacity in a two-layer sand profile (sand A varying with depth.” J. Geotech. Geoenviron. Eng. 140 (4): 04013049.
over sand B); https://doi.org/10.1061/(ASCE)GT.1943-5606.0001067.
Hcrit = the critical pile lateral capacity when the pile length Choo, Y. W., and D. Kim. 2016. “Experimental development of the p-y
reaches the critical pile length Lcrit ; relationship for large-diameter offshore monopiles in sands: Centrifuge
tests.” J. Geotech. Geoenviron. Eng. 142 (1): 04015058. https://doi.org
Hjd1¼0 = lateral capacity of a pile in a two-layer sand when the /10.1061/(ASCE)GT.1943-5606.0001373.
top layer thickness d1 ¼ 0; DNV (Det Norske Veritas). 2014. Design of offshore wind turbine struc-
Hjd1¼∞ = lateral capacity of a pile in a two-layer sand when the tures. DNV-OS-J101. Bærum, Norway: DNV.
top layer thickness d1 → ∞; Doherty, P., and K. Gavin. 2012. “Laterally loaded monopole design for
Hj0.5° = pile lateral capacity corresponding to a pile rotation offshore wind farms.” Proc. Inst. Civ. Eng. Energy 165 (1): 7–17.
θ ¼ 0.5° at mudline; https://doi.org/10.1680/ener.11.00003.
Georgiadis, M. 1983. “Development of P-Y curves for layered soils.” Int. J.
Hj1° = pile lateral capacity corresponding to a pile rotation
Rock Mech. Min. Sci. Geomech. Abstr. 22 (5): 158. https://doi.org/10
θ ¼ 1° at mudline; .1016/0148-9062(85)92248-X.
h = load eccentricity; Han, F. 2017. Axial and lateral resistance of non-displacement piles.
I p = area moment of inertia of pile; West Lafayette, IN: Purdue Univ.
L = pile length; Han, F., E. Ganju, R. Salgado, and M. Prezzi. 2018. “Effects of interface
Lcrit = critical pile length beyond which the pile lateral roughness, particle geometry, and gradation on the sand-steel interface
capacity no longer increases; friction angle.” J. Geotech. Geoenviron. Eng. 144 (12): 04018096.
https://doi.org/10.1061/(ASCE)GT.1943-5606.0001990.
LR = reference length ¼ 1 m; Han, F., M. Prezzi, and R. Salgado. 2017a. “Energy-based solutions for non-
pA = reference stress ¼ 100 kPa; displacement piles subjected to lateral loads.” Int. J. Geomech. 17 (11):
tw = wall thickness of pile; 04017104. https://doi.org/10.1061/(ASCE)GM.1943-5622.0001012.
04019027. https://doi.org/10.1061/(ASCE)GT.1943-5606.0002050. cability of conventional p-y relations to the analysis of piles in laterally
Hearn, E. N., and L. Edgers. 2010. “Finite element analysis of an offshore spreading soil.” J. Geotech. Geoenviron. Eng. 137 (6): 557–567. https://
wind turbine monopile.” In Proc., GeoFlorida 2010, 1857–1865. doi.org/10.1061/(ASCE)GT.1943-5606.0000468.
Reston, VA: ASCE. Reese, L. C., W. R. Cox, and F. D. Koop. 1975. “Field testing and analysis
Hu, Q., F. Han, M. Prezzi, R. Salgado, and M. Zhao. 2021. “Lateral load
of laterally loaded piles in stiff clay.” In Proc., Annual Offshore
response of large-diameter monopiles in sand.” Géotechnique 1–16.
Technology Conf., edited by L. Reese, 671–675. Houston: Society of
https://doi.org/10.1680/jgeot.20.00002.
Petroleum Engineers.
Hu, Z. B., and Y. X. Yang. 2018. “Study on p-y curves of large-diameter
Salgado, R., F. Han, and M. Prezzi. 2017. “Axial resistance of non-
steel pipe piles for offshore wind farm in sand based on in-situ tests.”
displacement piles and pile groups in sand.” Rivista Italiana di Geotecn-
J. Appl. Sci. Eng. 21 (2): 171–178. https://doi.org/10.6180/jase.201806
ica 51 (4): 35–46.
_21(2).0004.
Klinkvort, R., and O. Hededal. 2010. “Centrifuge modelling of offshore Suryasentana, S. K., and B. M. Lehane. 2016. “Updated CPT-based p–y
monopile foundation.” In Frontiers in offshore geotechnics II, formulation for laterally loaded piles in cohesionless soil under static
581–586. London: CRC Press. loading.” Géotechnique 66 (6): 445–453. https://doi.org/10.1680/jgeot
Klinkvort, R., C. Leth, and O. Hededal. 2010. “Centrifuge modelling of a .14.P.156.
laterally cyclic loaded pile.” In Physical modelling in geotechnics, two Suzuki, P. T., D. Maloney, and L. Hamre. 2019. “Influence of layered soil
volume set, 959–964. London: CRC Press. profiles on the application of p-y curves for large diameter monopiles.”
Klinkvort, R. T., and O. Hededal. 2013. “Lateral response of monopile sup- In Proc., 1st Vietnam Symp. on Advances in Offshore Engineering,
porting an offshore wind turbine.” Proc. Inst. Civ. Eng. Geotech. Eng. 415–420. New York: Springer.
166 (2): 147–158. https://doi.org/10.1680/geng.12.00033. Yang, Z., and B. Jeremić. 2005. “Study of soil layering effects on lateral
Lesny, K., and J. Wiemann. 2005. “Design aspects of monopiles in loading behavior of piles.” J. Geotech. Geoenviron. Eng. 131 (6): 762–
German offshore wind farms.” In Proc., 1st Int. Symp. on Frontiers 770. https://doi.org/10.1061/(ASCE)1090-0241(2005)131:6(762).
in Offshore Geotechnics: Frontiers in Offshore Geotechnics, ISFOG Zdravković, L., et al. 2015. “Numerical modelling of large diameter piles
2005, 383–389. London: Taylor & Francis. under lateral loading for offshore wind applications.” In Proc., 3rd Int.
Lesny, K., and J. Wiemann. 2006. “Finite-element-modelling of large Symp. on Frontiers in Offshore Geotechnics, 759–764. Pathumthani,
diameter monopiles for offshore wind energy converters.” In Proc., Thailand: Asian Institute of Technology. https://doi.org/10.1201/b18
GeoCongress 2006, 1–6. Reston, VA: ASCE. 442-105.