Sie sind auf Seite 1von 14

Analytical Solutions for Ultimate Limit State

Design of Thermal Piles


Chiara Iodice 1; Raffaele Di Laora 2; and Alessandro Mandolini 3
Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

Abstract: This work aimed at providing analytical closed-form solutions for the design of thermal piles. To this end, a model in which
a cylindrical pile is attached along the shaft to a series of distributed vertical springs representing soil stiffness is proposed. The pile has
constant section and elastic properties; the restraints provided by the superstructure and base stiffness are represented through concentrated
springs. The model allows derivation of exact solutions for homogeneous, two-layer soil and soil with linearly increasing stiffness with depth.
In addition, approximate energy solutions are derived via the principle of virtual work for more general subsoil conditions with spring
stiffness calibrated through finite element results. Expressions for the axial force and shear stress at the pile–soil interface are provided
for typical soil stiffness distributions. A successful comparison to literature studies, involving complex transient-coupled numerical analyses
and two field tests, corroborate model reliability. The proposed analytical solutions provide insight into the behavior of thermally loaded piles
and can be used as a simple tool for ultimate limit state design. DOI: 10.1061/(ASCE)GT.1943-5606.0002204. © 2020 American Society of
Civil Engineers.
Author keywords: Thermal piles; Ultimate limit state design; Analytical solution; Finite element analysis.

Introduction herein the thermally induced stress state within the pile will be
investigated.
Energy piles are capable of collecting heat from the surrounding Because of the restraint imposed by the surrounding soil, a
soil and transferring it to the superstructure for heating during change in pile temperature translates to a change in pile stress both
winter; during summer, this transfer is reversed, allowing piles along its shaft and at its base (Laloui et al. 1999; Bourne-Webb
to provide heat seasonally and cyclically. The exchange of energy et al. 2009; McCartney and Murphy 2014; Stewart and McCartney
is possible because piles interact via thermal conduction with the 2014; Ng et al. 2014b; Goode III and McCartney 2015; Murphy
ground at depths suitable for the exploitation of low-enthalpy geo- and McCartney 2015). When the pile is heated, an additional
thermal energy. For this purpose, they are equipped with closed compressive axial load develops. When it is cooled, the overall
loop pipes through which a heat exchanger is circulated using a axial force may decrease until resulting in a negative load at some
heat pump. depths. In some cases, the thermally induced variation in the axial
This foundation has therefore the double role of transferring force may even be of the same magnitude as the mechanical load
both structural loads (from the structure to the ground) and heat imposed at the top (e.g., Laloui et al. 1999). For a free-head pile, the
(from the ground to the structure and vice versa). During recent heat exchange always implies a mere redistribution of load between
years, such a foundation system has been employed in a number the shaft and base; nevertheless, considering that in reality the pile
of public and private applications (Brandl 2006). is always restrained by the superstructure, the total load at the pile
The main geotechnical issues concerning energy piles refer to head may also significantly vary. In the latter case, the ultimate
the additional settlements and stresses caused by the cyclic temper- limit state (ULS) design with reference to the axial-bearing capacity
ature variation inside the pile and surrounding soil. The first issue
could lead to an increase in pile size to comply with thermal effects.
has been explored in several works (Laloui and François 2009;
Furthermore, regardless of the restraint at the pile head, the ULS
Ng et al. 2014a, 2016; Di Donna and Laloui 2015a, b; Ma et al.
design from a structural viewpoint will be detrimentally affected by
2017) and will not be further discussed in this paper. Instead,
the thermally induced axial forces along the pile.
When extraction or recharge of heat begins, the difference
between the pile and the surrounding soil temperatures, which
1
Ph.D. Candidate, Dept. of Engineering, Università della Campania understandably represents a key parameter controlling the magni-
“Luigi Vanvitelli”, via Roma 29, Aversa, CE 81031, Italy. Email: chiara
tude of thermally induced effects, is maximum. As the process
.iodice@unicampania.it
2
Assistant Professor, Dept. of Engineering, Università della Campania continues, owing to heat conduction, the ground temperature ap-
“Luigi Vanvitelli”, via Roma 29, Aversa, CE 81031, Italy (corresponding proaches that of the pile leading to a decrease in the pile thermal
author). ORCID: https://orcid.org/0000-0002-9993-5353. Email: raffaele load. To assess the variation in the axial load within the pile, a re-
.dilaora@unicampania.it markable simplification of the analysis may be achieved referring
3
Professor, Dept. of Engineering, Università della Campania “Luigi to the scenario in which heat conduction has not yet occurred. The
Vanvitelli”, via Roma 29, Aversa, CE 81031, Italy. Email: alessandro validity of this assumption was confirmed by Rammal et al. (2018).
.mandolini@unicampania.it
The authors performed a series of finite element (FE) analyses in
Note. This manuscript was submitted on April 22, 2019; approved on
September 17, 2019; published online on March 5, 2020. Discussion period which the pile was subjected to temperature changes and thermal
open until August 5, 2020; separate discussions must be submitted for cycles along with a simple mechanical calculation (i.e., imposing
individual papers. This paper is part of the Journal of Geotechnical only the volumetric strain on the pile) neglecting the heat conduc-
and Geoenvironmental Engineering, © ASCE, ISSN 1090-0241. tion. They showed that the latter simple scenario, reasonable for

© ASCE 04020016-1 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


soils that are not subject to thermal plastic strains, well reproduces
the additional thermal axial forces.
Notwithstanding this simplification of the problem, which
avoids calculations in the transient regime, currently the analysis
of energy piles must be necessarily carried out through the use
of numerical techniques (Knellwolf et al. 2011; Suryatriyastuti
et al. 2012, 2014; Pasten and Santamarina 2014; Yavari et al.
2014; Di Donna and Laloui 2015a; Chen and McCartney 2016;
Di Donna et al. 2016; Rotta Loria and Laloui 2017; Adinolfi et al.
2018; Rui and Yin 2018; Rammal et al. 2018; Rui and Soga 2019).
Rotta Loria et al. (2018) highlight that for piles subjected to a
thermal load “the need to restore to numerical analyses : : : results
Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

from the lack of closed form solutions : : : , in contrast to their avail-


ability for isolated piles subjected to a mechanical load.”
This work aimed to eliminate this lack of closed-form solutions.
Exact analytical solutions are developed for the problem at hand,
with reference to homogeneous and two-layered soils as well as
subsoils whose stiffness linearly varies with depth. For more gen-
eral subsoil conditions, approximate energy solutions were derived.
These solutions provide accurate results compared to the numerical
and experimental data, and thus can aid practitioners in designing Fig. 1. Proposed model.
thermal piles by means of simple hand calculations.

Proposed Model 2πr0 τ ðzÞ ¼ −kðzÞ · up ðzÞ ð4Þ


Consider a cylindrical pile subjected to a uniform temperature
where r0 = the pile radius.
change ΔT. If the pile is not subjected to any restraint from the
The derivative of the axial load is related to the shear stress as
surrounding soil, it experiences a constant axial strain along its
follows:
length equal to the following:
εΔT ðzÞ ¼ αΔT ð1Þ dNðzÞ
¼ 2πr0 τ ðzÞ ¼ −kðzÞ · up ðzÞ ð5Þ
dz
where α = the coefficient of the thermal expansion of the pile
material and the expansion is taken as positive. However, the soil Upon substituting Eq. (3) into Eq. (5), a second-order differen-
acts as a mechanical restraint that opposes the elongation or short- tial equation is obtained as follows:
ening of the pile by applying shear stresses upon its external sur-
face. The resulting thermal strain is therefore lower than εΔT ðzÞ and d2 up ðzÞ kðzÞ
can be expressed as follows: − · up ðzÞ ¼ 0 ð6Þ
dz2 EA
dup ðzÞ NðzÞ
εp ðzÞ ¼ ¼ αΔT − ð2Þ The solution to the aforementioned equation can be found im-
dz EA
posing proper boundary conditions. In the following sections, the
where E and A = the Young’s modulus and cross-sectional area solution for the case of homogeneous and two-layer subsoils is
of the pile, respectively; up = the pile vertical displacement; and presented; the details regarding the integration of the differential
N = the axial load induced by the temperature. The compressive equation can be found in the Appendix.
axial loads are taken as positive.
Eq. (2) can be written as follows:
dup ðzÞ Exact Solutions
NðzÞ ¼ αΔTEA − EA ð3Þ
dz
Constant Stiffness
To account for the restraining effect of the surrounding soil, a
If soil stiffness is constant with depth, Eq. (6) can be written
linear elastic model is proposed in which the soil is represented
as follows:
through distributed springs of stiffness kðzÞ (dimensions: FL−2 ),
and the pile base and pile top are connected to concentrated springs
d2 up ðzÞ
of stiffness kb and kt (dimensions: FL−1 ) (Fig. 1). Note that this is − λ2 up ðzÞ ¼ 0 ð7Þ
exactly the same model adopted by Randolph and Wroth (1978) dz2
for mechanical axial loads at the pile top. The calibration of the
springs’ stiffness as a function of pile and soil properties is dis- where λ (dimensions: L−1 ) = the pile-soil wavenumber:
cussed later; however, k is on the order of twice the soil shear rffiffiffiffiffiffiffi
modulus and this first-approximation value is adopted in the sub- k
λ¼ ð8Þ
sequent graphs. EA
The shear stress τ ðzÞ acting upon the pile surface is therefore
proportional to the pile displacement according to the following The pile axial load distribution can be found solving Eq. (7) by
equation: imposing proper boundary conditions as follows:

© ASCE 04020016-2 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


 
coshðλðL − zÞÞ þ Ω½sinhðλðL − zÞÞ þ Θ coshðλzÞ þ sinhðλzÞ
NðzÞ ¼ αΔTEA 1 − ð9Þ
ð1 þ ΘΩÞ coshðλLÞ þ ðΘ þ ΩÞ sinhðλLÞ

where L = the pile length and the dimensionless parameters Ω and or soil stiffness approaches infinity and when the axial stiffness
Θ are equal to the following: disappears; (2) the presence of a restraint at the pile top always
results in an increase in the axial force; (3) for a given pair
k EAλ (λL, Θ), the axial force is practically independent of the pile–soil
Ω¼ ¼ ð10Þ
λkb kb stiffness ratio.
The depth at which the axial load is a maximum corresponds to
Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

kt k
Θ¼ ¼ t ð11Þ the depth at which the shear stress and pile displacement disappear,
Ωkb EAλ and can be found through the following equation:
and physically represent the ratios between the base or top stiffness  λL 
1 e ½ΩðeλL − 1Þ þ eλL þ Θ
and shaft stiffness (integrated over a characteristic pile-soil wave zN¼Nmax ¼ zτ ¼0 ¼ ln ð12Þ
length), respectively. Note that values of Θ greater than 50 corre- 2λ ΩðeλL − 1Þ þ ΘeλL − 1
spond, from an engineering viewpoint, to a perfect restraint.
It is very useful to define a mechanical slenderness, λL, encom- Fig. 2(b) depicts the normalized depth of the maximum axial
passing both the pile geometry and pile–soil stiffness ratio. In the force against mechanical slenderness; regarding the axial force,
absence of restraints at the pile ends, this would be the unique there is no dependence on the pile–soil stiffness ratio. For a fully
dimensionless parameter controlling the maximum axial force in- restrained pile (Θ ¼ ∞), it is simple to verify that the maximum
duced by the thermal load. The presence of the base and top springs axial load is at the pile head (zN¼Nmax ¼ 0). As the mechanical
introduces two additional parameters. slenderness ratio approaches infinity, the maximum axial load is
In Fig. 2, the maximum thermal load along the pile shaft, N max , at the pile half length (zN¼Nmax ¼ L=2).
normalized by the (maximum) axial force the pile would have been
subject to if fully restrained at the ends, N fixed ¼ αΔTEA, is plot-
Two-Layer Soil
ted against the mechanical slenderness λL for different values of Θ
and two extreme values of the pile–soil stiffness ratio (with G being When a two-layer soil is considered, the axial load distribution
soil shear modulus). The following aspects are noteworthy: (1) the has the following expression (the solution of the differential equa-
thermally induced axial force increases with λL, tending to the tion, as well as the expressions of settlement and shear stress, are
value for a bar fully restrained at the ends when the pile length reported in the Appendix):

 
ρ½a3 ða5 þ Ω1 a6 Þ þ Ω1 ða2 þ Θa1 Þ þ a4 ða6 þ Ω1 ρ2 a5 Þ
N 1 ðzÞ ¼ αΔTEA 1 −
ρa3 ½a7 ð1 þ ΘΩ1 Þ þ a8 ðΩ1 þ ΘÞ þ a4 ½a7 ðΩ1 ρ2 þ ΘÞ þ a8 ð1 þ ΘΩ1 ρ2 Þ
 
a9 ½a3 þ Ω1 ða8 þ ρa4 þ Θa7 Þ − a10 ½a4 þ Ω1 ρða3 − a7 þ Θa8 Þ
N 2 ðzÞ ¼ αΔTEA 1 − ρ ð13Þ
ρa3 ½a7 ð1 þ ΘΩ1 Þ þ a8 ðΩ1 þ ΘÞ þ a4 ½a7 ðΩ1 ρ2 þ ΘÞ þ a8 ð1 þ ΘΩ1 ρ2 Þ

where detrimental; yet, it may be of lesser importance if compared to


sffiffiffiffiffiffi the (high) mechanical loads.
G2 The depth of the maximum axial force is depicted in Fig. 4.
ρ¼ ð14Þ Notably, for most cases encountered in practice, N max occurs in
G1
the second (stiffer) layer.
encompasses the ratio between the shear modulus of the deeper
layer (G2 ) and that of the shallower layer (G1 ). The expressions Linear Stiffness Profile (Gibson Soil)
of the dimensionless terms ai can be found in the Appendix. They
involve the dimensionless parameter λ1 L (i.e., the mechanical Eq. (6) for a soil whose stiffness proportionally varies with depth
slenderness referred to the first layer) and the dimensionless inter- can be written as follows:
face depth defined as follows:
d2 up ðzÞ
h − μ3 zup ðzÞ ¼ 0 ð16Þ
ζ¼ 1 ð15Þ dz2
L
where
In Fig. 3, the normalized axial load is plotted for a free and fully
restrained head pile by varying the problem parameters. N max rffiffiffiffiffiffiffi
understandably increases with increasing λ1 L and ρ values and μ¼
3 k̄
ð17Þ
for decreasing values of ζ. Note that for end-bearing piles EA
(ρ > 2, ζ ≅ 1) of an ordinary length (λ1 L > 1), the induced axial
force is at least 80% of the maximum value (see graph c) and is the pile-soil wave number (dimensions: L−1 ) for a Gibson soil
thereby the thermal load for this category of piles is particularly profile (i.e., assuming k ¼ k̄ · z).

© ASCE 04020016-3 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


The axial loads at the pile head and base are provided as follows:
n h   i  o
3 pffiffiffiffiffiffi
2
1
36 Θ Ω̄ I −23 23 ðμLÞ2 μL − Γð1Þ3p3 ffiffiffiffi ffi þ I 1=3 23 ðμLÞ32 Þ
μL
N z¼0 ¼ αΔTEA 1 h   pffiffiffiffiffiffi i h 3  pffiffiffiffiffiffi 
1
 i
36 Θ I −23 3 ðμLÞ2 Ω̄ μL þI 13 ð3 ðμLÞ2 þ Γð1Þ2 · I −43 23 ðμLÞ2 Ω̄ μL þI −13 23 ðμLÞ3 pΩ̄ffiffiffiffi ffiþ1
2 3 2 3 2π 3

3
μL
h   i  
1 3 3
36 I 13 3 ðμLÞ2 Θ − Γð1Þpffiffiffiffi
2 3
μL
ffi þ 1 2 · I −1 3 ðμLÞ2

Γð Þ
2
N z¼L ¼ αΔTEA 1 h   pffiffiffiffiffiffi  i 3
h  3  3 pffiffiffiffiffiffi   i ð18Þ
36 Θ I −23 3 ðμLÞ2 Ω̄ μL þI 13 3 ðμLÞ2 þ Γð1Þ2 · I −43 3 ðμLÞ2 Ω̄ μL þI −1=3 23 ðμLÞ2 pΩ̄ffiffiffiffi ffiþ1
2 3 2 3 2π 2 3 3

3
μL
Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

where is totally cancelled out because the spring stiffness is directly cali-
brated in these simpler solutions.
k̄ EAμ
Ω̄ ¼ ¼ ð19Þ
μkb kb
Approximate Solutions
Θ has the same expression for the case of homogeneous soil except
using Eq. (19) for Ω̄; I i ð23 ðμLÞ3=2 Þ is the first type of modified With the aim of obtaining an approximate solution for the problem
Bessel function of i-th order and argument 23 ðμLÞ3=2 ; and Γð13Þ is at hand, it is possible to refer to a simplified displacement profile
the Gamma function evaluated at 1=3. and then exploit the principle of virtual work.
In Fig. 5(a), the normalized axial load is plotted against the The displacement of the pile induced from the thermal load is
mechanical slenderness μL for different values of Θ and two here approximated through a linear function of the depth. It is con-
extreme values of pile–soil stiffness ratio (Ḡ is the gradient of venient to express the pile displacement as follows:
the soil shear modulus with depth). As for the homogeneous soil
up ðzÞ ¼ Lūl ΦðzÞ ð20Þ
case, it can be seen that the mechanical slenderness successfully
merges the geometrical and mechanical properties of the pile where ūl = the displacement at pile tip normalized by pile length;
and soil. The same statement holds for the normalized depth plotted and ΦðzÞ = the shape function:
against the mechanical slenderness μL [Fig. 5(b)].
The aforementioned solution, on the one hand possesses
z
− z̄N
ΦðzÞ ¼ L
ð21Þ
the undeniable advantage of being exact while, on the other hand, 1 − z̄N
it may face some reluctance from practitioners who are often not
familiar with Bessel functions that, however, are implemented even with z̄N = the depth corresponding to zero pile displacement
in common spreadsheets. normalized by the pile length.
It would be straightforward to report a simple iterative pro- In the realm of the principle of virtual work, both the virtual and
cedure to derive the equivalent homogeneous soil leading to the real displacements are assumed to be described by Eq. (20).
same maximum axial force as that of the Gibson soil. However, To enlarge the domain of potential practical applications, refer-
this is not done here as a more general approximate solution is de- ence is made to a general subsoil with a stiffness profile provided
veloped for practical use. In this sense, the exact solutions previ- by the following relation:
  n

ously reported are prone to use as a benchmark for future research z


because a comparison to models involving nonlinear springs would GðzÞ ¼ GL a þ ð1 − aÞ ð22Þ
L
not require a specific selection of spring stiffnesses as a function of
the soil shear modulus. In contrast, the approximate solutions are where GL = the soil shear modulus at the pile tip and a and n are
more suitable for practical applications, considering that they are two coefficients regulating the stiffness profile. The principle of
developed for more general subsoil and that even a small error virtual work ensures that the work done by the top, base, and shaft

Fig. 2. Constant stiffness soil. Variation in the (a) normalized axial load; and (b) normalized depth with the mechanical slenderness for different
values of Θ.

© ASCE 04020016-4 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Two-layer soil. Variation in the normalized pile axial load with the mechanical slenderness for different values of ζ and ρ: (a) Θ ¼ 0;
(b) Θ ¼ ∞. Variation in the normalized pile axial load with the dimensionless interface depth for different values of λ1 L and ρ: (c) Θ ¼ 0;
and (d) Θ ¼ ∞. In all graphs E=G ¼ 1; 000.

loads equals that done by the axial force due to the variation in the Appendix. Once the displacement profile is known, it is possible
pile temperature, both associated with the displacement pattern in to derive the axial load distribution as follows:
Eq. (20). The latter statement can be written for the problem at
hand, in a dimensionless form, as follows:
Z L   n
NðzÞ ¼ EAλL Lūl
z n h z io
ΘL ΩL ū20 þ ū2l þ λL ΩL a þ ð1 − aÞ ū2l Φ2 ðzÞdz 2ΘL z̄N − λL z a Lz − 2z̄N þ 2ð1 − aÞ Lz n Lð2þnÞ − 1þn
z̄N
0 L
·
Φ0 2ð1 − z̄N Þ
¼ ΩL ūl ðαΔT − Lūl Φ 0 Þ ð23Þ
λL ð25Þ

where ū0 = the displacement at the pile head normalized with re-
spect to the pile length; λL , ΩL , and ΘL are obtained from Eqs. (8), Expressions for special cases are reported in the Appendix in
(10), and (11), respectively, using Eq. (22) for the soil shear modu- terms of axial force and shear stress, along with the extension
lus; and Φ 0 = the derivative of the shape function with respect to z. of the aforementioned method to a multi-layer soil.
In addition, the equilibrium along the vertical direction en- Energy solutions have been obtained for axially loaded (Crispin
sures that et al. 2019) as well as laterally loaded (Karatzia and Mylonakis
Z L   n
2017) piles. However, the application of this method to thermally
z
ΘL ΩL ū0 þ ūl þ λL ΩL a þ ð1 − aÞ ūl ΦðzÞdz ¼ 0 loaded piles leads to much more accurate solutions, because, con-
0 L trary to the case of piles loaded at the top, approximating the dis-
ð24Þ placement profile via a linear function results in an error of minor
concern. This is evident in Fig. 6, in which a comparison between
Eqs. (23) and (24) can be simultaneously solved for the two the FE-derived displacement profile and that of Eq. (20) is reported
unknowns, ūl and z̄N , whose expressions are reported in the for different subsoil types.

© ASCE 04020016-5 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


Spring Stiffness

In the proposed model, the soil surrounding the pile is represented


through a series of linear elastic springs whose elastic constant kðzÞ
is proportional to the soil shear modulus GðzÞ through a coefficient
δ as follows:
kðzÞ ¼ δ · GðzÞ ð26Þ

Spring stiffness has been widely investigated for the case of


mechanical axial loads at the pile top (Coyle and Reese 1966;
Randolph and Wroth 1978; Mylonakis 2001); such Winkler-type
Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

models well reproduce continuum solutions.


To calibrate the value of δ for thermally loaded piles, a number of
FE analyses were carried out by means of the commercial software
ANSYS® (ANSYS, Canonsburg, Pennsylvania). Different con-
figurations were considered, varying the soil stiffness distribution
and pile size. Both free- and restrained-head piles were analyzed.
Fig. 4. Two-layer soil. Variation in the normalized depth with the
With reference to the latter case, both pile and soil vertical displace-
dimensionless interface depth for different values of λ1 L and ρ.
ments were restrained at the surface. A total of 480 analyses were

Fig. 5. Gibson soil profile: Variation in the (a) normalized axial load; and (b) depth with the mechanical slenderness μL for different values of Θ.

Fig. 6. Energy solution. Comparison between finite element method (FEM) data and proposed solution for free-head pile: (a) constant stiffness soil;
(b) Gibson soil profile; (c) proportional stiffness profile; and (d) two-layer soil.

© ASCE 04020016-6 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Spring stiffness calibration. FEM data, proposed equations and expression from Randolph and Wroth (1978), for (a) free head pile;
and (b) restrained-head pile.

Fig. 8. Comparison between FE and analytical energy solution for constant stiffness soil with E=GL=2 ¼ 1; 000: (a) free-head and ΘL ¼ 0; and
(b) restrained-head pile and ΘL ¼ ∞.

carried out, with L=d ¼ 20, 30, 40, and 50 (where d is the pile ½−0.25·Ld·ðG E Þ−1=3 
diameter); E=GL=2 ¼ 250, 500, 1,000, and 2,000 (with GL=2 being δ ¼ 3.5 · e L=2 for restrained-head piles ð27bÞ
the soil shear modulus at z ¼ L=2); Poisson’s ratio ν ¼ 0.3, 0.4,
and 0.49; and four different subsoils (n ¼ 0; a ¼ 0 and n ¼ 1;
Fig. 7 shows a comparison between the numerically derived
a ¼ 0 and n ¼ 0.5; a ¼ 0.5 and n ¼ 1). Two-dimensional axisym-
δ and the aforementioned expressions. Also shown in the graphs
metric 8-noded elements were employed to mesh the model. As an
are the values obtained by Randolph and Wroth’s expression,
outcome of a sensitivity study, the element’s vertical size was set to
emphasizing the need for a calibration of δ on the specific problem
0.25 d, while the model dimensions were 100 diameters in height
of thermal load.
and 200 diameters in width. Perfect bounding between pile and soil
Employing the aforementioned values for δ, the results from the
was considered. The lateral boundary of the model is restrained
energy solution are compared to the rigorous FE analyses for the
against vertical displacement.
cases of a constant stiffness and stiffness proportional to depth
By imposing the equality between the maximum axial force
(Figs. 8 and 9).
derived by the numerical analyses and that furnished by the afore-
The role of δ in the accuracy of the solution is instead explored
mentioned energy solutions, simple expressions for δ were found as
in Fig. 10, with reference to values equal to one-half and twice the
follows:
optimal value δ̄ from Eq. (27). From such graphs, it is inferred that δ
½−0.075·Ld·ðG E Þ−1=5 
(as well as the soil shear modulus) has a major effect on the ther-
δ¼4·e L=2 for free-head piles ð27aÞ mally induced axial force, yet the energy solutions well reproduce

© ASCE 04020016-7 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Comparison between FE and analytical energy solution for Gibson soil with E=GL=2 ¼ 1; 000: (a) free-head and ΘL ¼ 0;
and (b) restrained-head pile and ΘL ¼ ∞.

Fig. 10. Effect of δ on the analytical solutions for free-head piles: (a) constant stiffness soil; (b) proportional stiffness soil profile; and (c) two-layer
soil. All the cases have the same average soil stiffness.

numerical data when the proposed formulae for spring stiffness are The pile tested in the London clay was 23 m in length with a
employed. The prominent role of stiffness for axially loaded piles diameter of from 0.61 to 5 m and a remaining pile length of 0.55 m.
as opposed to the case of a lateral load is due to the exponent of the The London clay was surmounted by a 4 m-thick layer of made
soil stiffness in the pile-soil wave number (= 0.5 for the axial and ground and river terrace deposits (Fig. 11). Details of the test
0.25 for the lateral load). set-up can be found in the original paper. The pile had no restraint
at its head while the initial ground temperature was 20°C. A con-
stant mechanical vertical load of 1,200 kN was first applied.
Comparison to Experimental Data The cooling phase occurred over 31 days followed by a heating
phase of 12 days; the pile was cooled to approximately 0°C and
The performance of the analytical solutions was verified via a heated to approximately 30°C. Finally, daily thermal cycles were
comparison to field test data. Notably, two tests were considered also performed over 3 days. The temperature distribution along
as follows: one was performed at Lambeth College, London, UK, the pile shaft was recorded using thermistors and optical fiber sen-
in clay soils (Bourne-Webb et al. 2009) whereas the other was sors (OFSs) at the end of the first cooling and heating phases.
executed in sandy soils at the Ecole Polytechnique Fédérale To capture both the mechanical and the thermal behavior of the
de Lausanne (EPFL), Lausanne, Switzerland (Laloui et al. 2003, pile, OFSs and vibrating wire strain gauges (VWSGs) were used.
2006). The available strain data refer to measurements immediately after

© ASCE 04020016-8 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Lambeth College test: (a) comparison between the free and thermal strain at the end of the first cooling phase, OFS data; and (b) comparison
between the field data and energy solution.

the application of the constant load of 1,200 kN (solely mechanical


strains) and of the first cooling and heating phases (thermomechan-
ical strains). The distribution of the thermal strain is obtained taking
the difference between the aforementioned two profiles (Amatya
et al. 2012). The axial load can be deduced using Eq. (3), in which
the first term is the thermal strain that the pile would be subject to if
free to displace, εΔT , and the second is the actual thermal strain
developed during the loading process, εp [Fig. 11(a)]. Following
the first cooling phase, the measured data showed that in the first
4 m below the ground surface the pile was subject to thermal strain
greater than εΔT ; the data related to this depth were neglected. Note
that the temperature was not constant along the pile. The energy
solution was employed to reproduce the results of the first cooling
phase. Because of the lack of information regarding soil stiffness,
data from the literature were used. In particular, according to
Vardanega and Bolton (2011), the clay shear modulus at a low
strain has been considered to proportionally vary due to the un-
drained shear strength su , GðzÞ ¼ 320.7 · su . For pile stiffness Fig. 12. Lausanne test: comparison between the field data and results
and soil strength parameters, reference is made to Amatya et al. from numerical and analytical solutions.
(2012). The pile diameter was considered to be 0.55 m for all
the pile lengths. A value of δ equal to 2 was derived from Eq. (27a).
Fig. 11(b) shows the prediction of the analytical solution as com- 2.45 and 1.75, respectively. The shear modulus was evaluated from
pared to the OFS measurements. Notably, the simple proposed the bulk modulus data reported by the authors using a Poisson’s
approach is capable of capturing the overall behavior of the pile ratio of 0.3. The results were compared to the field data and to that
under the thermal load. obtained via a fully coupled thermo-hydromechanical analysis
The tests performed at the EPFL involved a building under con- (Laloui et al. 2006) using an FE software (Fig. 12). Despite its
struction founded on a piled raft foundation in which one pile, simplicity, the energy approach is proven to be effective, given
equipped with heat exchanger pipes, underwent a series of thermo- that the experimental data occur between that of the free- and
mechanical tests. The pile group was installed in a multilayer soil restrained-head conditions. Note that the simple analytical results
composed of five different strata of sandy deposits (Fig. 12). The are very similar to the complex FE results.
pile length and average diameter were 25.8 and 1.05 m, respec-
tively. Further details are reported in the original work. Reference Conclusions
is made here to test number 7 in which the temperature of the pile
was increased to 13.4°C. The tested pile was fully instrumented In this work, analytical solutions for the design and analysis of
with optical fibers and extensometers. The thermal loads were ap- thermally loaded piles were developed.
plied in the presence of a mechanical load of approximately The problem was analyzed through a simple model in which a
1,300 kN; the pile head was restrained by the presence of the cylindrical pile of a constant section and Young’s modulus is con-
building connected to other piles. A heating process over 12 days nected to a series of distributed springs at the shaft; soil stiffness at
followed by 16 days of passive cooling was applied. To consider the base and the restraint offered by the superstructure were also
the degree of restraint at the pile head, the proposed energy solution modeled as two concentrated springs.
was applied considering two extreme cases: a free- and restrained- Such a model allowed for the derivation of exact solutions in
head pile. The values of δ derived from Eqs. (27a) and (27b) are dimensionless terms for pile displacement, axial load, and shear

© ASCE 04020016-9 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


stress at the pile–soil interface in homogeneous, Gibson, and two- the spring stiffness was calibrated for these solutions via compari-
layer soil. It was found that the maximum thermally induced axial son to FE analyses. Novel expressions for evaluating spring proper-
force is mainly controlled by a single dimensionless parameter. ties as function of the pile–soil stiffness ratio and pile geometry
This parameter represents a mechanical slenderness encompassing were proposed for the specific problem of thermally loaded piles.
the pile geometry and Young’s modulus and soil stiffness. The pres- The successful match with rigorous fully coupled thermo-
ence of a restraint at the pile top has a detrimental effect because the hydromechanical FE analyses as well as with two field tests avail-
maximum thermal axial force increases and moves upward where able in the literature proved that the proposed solutions provide
the axial force resulting from the existing mechanical load is higher. reliable results via hand calculations.
To allow application in routine design, approximate solutions However, notwithstanding the practical appeal of the present
were derived by estimating the displacement profile of the pile work, it should be highlighted that the proposed formulae may as-
and exploiting the principle of virtual work. Solutions for a soil sist design of thermally loaded piles only with regard to the ultimate
Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

with a three-parameter stiffness distribution as well as generaliza- limit state (ULS). The assessment of settlements due to cyclic
tion to a multilayer soil were presented. The approximation in- loading is another aspect of pile design not captured by the model
volved in the simplified assumption was cancelled out because considered in this study.

Appendix. Derivation of Exact and Approximate Solutions

Exact Solutions
1. Homogeneous soil
The general solution of Eq. (7) is as follows:

up ðzÞ ¼ C1 eλz þ C2 e−λz ð28Þ

with λ being expressed through Eq. (8). To obtain the two unknowns, Ci , the two following boundary conditions are necessary:


dup ðzÞ kt up ðzÞjz¼0
¼ αΔT þ
dz z¼0 EA

dup ðzÞ kb up ðzÞjz¼L
¼ αΔT − ð29Þ
dz z¼L EA

Once one knows the two constants Ci , the displacement up ðzÞ can be derived using Eq. (28) as follows:

αΔT ΩfcoshðλzÞ − cosh½λðL − zÞg − sinh½λðL − zÞ þ ΘΩ sinhðλzÞ


up ðzÞ ¼ · ð30Þ
λ ð1 þ ΘΩÞ coshðλLÞ þ ðΩ þ ΘÞ sinhðλLÞ

Combining Eq. (30) with Eqs. (3) and (4), the axial load [reported in section “Constant Stiffness”, Eq. (9)] and the shear stress
distributions are obtained. The latter has the following expression:

αΔTEAλ Ωfcosh½λðL − zÞ − coshðλzÞg þ sinh½λðL − zÞ − ΘΩ sinhðλzÞ


τ ðzÞ ¼ · ð31Þ
2πr0 ð1 þ ΘΩÞ coshðλLÞ þ ðΩ þ ΘÞ sinhðλLÞ

In the following, the general expressions of ūl and z̄N along with the expressions of the axial load and shear stress [Eqs. (42a), (42b),
(25), and (43), respectively] are written for the four different profiles of shear stiffness.
2. Two-layered soil
When the two-layer soil is considered, it is necessary to write the two differential equations:

d2 up1 ðz1 Þ
− λ21 up ðz1 Þ ¼ 0
dz21
d2 up2 ðz2 Þ
− λ21 up ðz2 Þ ¼ 0 ð32Þ
dz22

where λ1 = the pile-soil wave number of the first layer. The general solutions of Eq. (32) are as follows:

up1 ðz1 Þ ¼ C1 eλ1 z1 þ C2 e−λ1 z1


up2 ðz2 Þ ¼ C3 eρλ1 z2 þ C4 e−ρλ1 z2 ð33Þ

© ASCE 04020016-10 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


To obtain the four unknowns, Ci , the following four boundary conditions are necessary:

dup1 ðzÞ kt up1 ðzÞjz1 ¼h1
¼ αΔT −
dz1 z1 ¼h1 EA

dup2 ðzÞ kb up2 ðzÞjz2 ¼h2
¼ αΔT −
dz2 z2 ¼h2 EA
up1 ðzÞjz1 ¼0 þ up2 ðzÞjz2 ¼0 ¼ 0

dup1 ðzÞ dup2 ðzÞ
− ¼0 ð34Þ
dz1 z1 ¼0 dz2 z2 ¼0
Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

Using the constants Ci determined from the aforementioned boundary conditions and Eq. (33), the displacement expression can be
obtained for both layers as follows:

αΔT ρ½a3 ða6 þ Ω1 a5 Þ − Ω1 ða1 þ Θa2 Þ þ a4 ða5 þ Ω1 ρ2 a6 Þ


u1 ðzÞ ¼ ·
λ1 ρa3 ½a7 ð1 þ ΘΩ1 Þ þ a8 ðΩ1 þ ΘÞ þ a4 ½a7 ðΩ1 ρ2 þ ΘÞ þ a8 ð1 þ ΘΩ1 ρ2 Þ
αΔT a9 ½−a4 þ Ω1 ρða7 − a3 þ Θa8 Þ þ a10 ½a3 þ Ω1 ða8 þ ρa4 þ Θa7 Þ
u2 ðzÞ ¼ · ð35Þ
λ1 ρa3 ½a7 ð1 þ ΘΩ1 Þ þ a8 ðΩ1 þ ΘÞ þ a4 ½a7 ðΩ1 ρ2 þ ΘÞ þ a8 ð1 þ ΘΩ1 ρ2 Þ
Considering that the shear stress, τ i ðzÞ, is proportional to the displacement through Eq. (4), it is possible to obtain the following
expressions of τ i ðzÞ:

αΔTEAλ1 ρ½a3 ða6 þ Ω1 a5 Þ þ Ω1 ða1 þ Θa2 Þ − a4 ða5 þ Ω1 ρ2 a6 Þ


τ 1 ðzÞ ¼ ·
2πr0 ρa3 ½a7 ð1 þ ΘΩ1 Þ þ a8 ðΩ1 þ ΘÞ þ a4 ½a7 ðΩ1 ρ2 þ ΘÞ þ a8 ð1 þ ΘΩ1 ρ2 Þ
αΔTEAλ1 a9 ½a4 − Ω1 ρða7 − a3 þ Θa8 Þ − a10 ½a3 þ Ω1 ða8 þ ρa4 þ Θa7 Þ
τ 2 ðzÞ ¼ · ð36Þ
2πr0 ρa3 ½a7 ð1 þ ΘΩ1 Þ þ a8 ðΩ1 þ ΘÞ þ a4 ½a7 ðΩ1 ρ2 þ ΘÞ þ a8 ð1 þ ΘΩ1 ρ2 Þ

The expressions of the axial load, Eq. (13), are found com- up ðzÞ ¼ C1 AiryAiðμzÞ þ C2 AiryBiðμzÞ ð40Þ
bining Eq. (35) with Eq. (3).
The parameter ρ is expressed through Eq. (14), while the with μ being expressed through Eq. (17) and AiryAi and
other parameters in Eqs. (35), (36), and (13) have the following AiryBi being Airy functions.
expressions: As in the case of constant stiffness, the boundary condi-
tions are:
k1 EAλ1
Ω1 ¼ ¼ ð37Þ dup ðzÞ kt up ðzÞjz¼0
λ 1 kb kb ¼ αΔT þ
dz z¼0 EA
h2 1 − ζ
ξ¼ ¼ ð38Þ
dup ðzÞ kb up ðzÞjz¼L
h1 ζ ¼ αΔT − ð41Þ
dz z¼L EA
a1 ¼ coshðλ1 zÞ Once one determines the two constants Ci , the displacement
a2 ¼ sinhðλ1 zÞ up ðzÞ can be derived using Eq. (40); accordingly, the axial load
distribution is obtained using Eq. (3). The expressions of the
a3 ¼ coshðξρh1 λ1 Þ axial load at the pile head and base are reported in section
a4 ¼ sinhðξρh1 λ1 Þ “Linear Stiffness Profile (Gibson Soil)”, Eq. (18).

a5 ¼ cosh½λ1 ðh1 − zÞ


Approximate Energy Solutions
a6 ¼ sinh½λ1 ðh1 − zÞ
1. Soil with a continuous stiffness variation
a7 ¼ coshðλ1 h1 Þ The principal of virtual work and the vertical equilibrium,
a8 ¼ sinhðλ1 h1 Þ Eqs. (23) and (24), respectively, are written in terms of two
unknowns ūl and z̄N ; their expressions are as follows:
a9 ¼ cosh½ρλ1 ðz − h1 Þ
a10 ¼ sinh½ρλ1 ðz − h1 Þ ð39Þ ūl ¼ αΔT6ΩL ð2 þ nÞð3 þ nÞfλL L½2 þ anð3 þ nÞ
þ 2ΘL ð1 þ nÞð2 þ nÞg
with ζ being expressed through Eq. (15) and Θ having the same
expression for the case of homogeneous soil except using · fΩL λ3L L3 ½12 þ anð4 þ anÞð7 þ nð4 þ nÞÞ
Eq. (37) for Ω1. h1 and h2 are the thicknesses of the first
þ 4λ2L L2 ð2 þ nÞ½6 þ 11an þ 6an2 þ an3
and second layers, respectively.
3. Linear stiffness profile (Gibson soil) þ ΘL ΩL ð1 þ nÞð2 þ nÞð3 þ anÞ
When stiffness is varying proportionally to depth, the þ 12λL Lð2 þ nÞ2 ð3 þ nÞðΩL þ anΩL þ ΘL þ nΘL Þ
displacement is expressed through Eq. (16), whose general
solution is: þ 12ð1 þ nÞð2 þ nÞ2 ð3 þ nÞð1 þ ΘL ΩL Þg−1 ð42aÞ

© ASCE 04020016-11 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


1 ð1 þ nÞ½2ð2 þ nÞ þ ΩL λL Lð2 þ anÞ EAλL λL L½a þ ð1 − aÞðLz Þn ðLz − z̄N Þ
z̄N ¼ · τ ðzÞ ¼ − · ūl ·
2 ð2 þ nÞ½1 þ n þ ΩL λL Lð1 þ anÞ þ ΘL ΩL ð1 þ nÞ 2πr0 1 − z̄N
ð42bÞ ð43Þ

Eqs. (42a) and (42b) allow one to evaluate the axial load and In the following, the general expressions of ūl and z̄N
shear stress. The expression of the axial load is reported in along with the expressions of the axial load and shear stress
section “Approximate Solutions” Eq. (25) whereas the shear [Eqs. (42a), (42b), (25), and (43), respectively] are written
stress has the following expression: for the four different profiles of shear stiffness.
Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

• Constant stiffness profile (a ¼ 1, n ¼ 0)

6ΩL ðλL L þ 2ΘL Þ


ūl ¼ αΔT ð44Þ
12 þ λL L½12ΩL þ λL Lð4 þ ΩL λL LÞ þ 4ΘL ½3ΩL þ λL Lð3 þ ΩL λL LÞ

1 2 þ ΩL λL L
z̄N ¼ · ð45Þ
2 1 þ ΩL λL L þ ΘL ΩL

2z̄N ðΘL þ λL L Lz Þ − λL LðLz Þ2


NðzÞ ¼ EAλL Lūl · ð46Þ
2ð1 − z̄N Þ

EAλ2L L z
− z̄N
τ ðzÞ ¼ · ūl · L ð47Þ
2πr0 1 − z̄N

• Proportional stiffness profile (a ¼ 0, n ¼ 1)

12ΩL ðλL L þ 6ΘL Þ


ūl ¼ αΔT ð48Þ
72 þ λL L½36ΩL þ λL Lð6 þ ΩL λL LÞ þ 18ΘL ½4ΩL þ λL Lð4 þ ΩL λL LÞ

2ð3 þ ΩL λL LÞ
z̄N ¼ ð49Þ
3ð2 þ ΩL λL L þ 2ΘL ΩL Þ

3z̄N ½λL LðLz Þ2 þ 2ΘL  − 2λL LðLz Þ3


NðzÞ ¼ EAλL Lūl · ð50Þ
6ð1 − z̄N Þ

EAλ2L L ð − z̄N Þ
z z
τ ðzÞ ¼ · ūl · L L ð51Þ
2πr0 1 − z̄N

• Linear stiffness profile (n ¼ 1)

12ΩL ½λL Lð1 þ 2aÞ þ 6ΘL 


ū ¼ ð52Þ
ΩL λ3L L3 ½1 þ að4 þ aÞ þ 6λ2L L2 ½1 þ 3a þ ΘL ΩL ð3 þ aÞ þ 36λL LðΩL þ aΩL þ 2ΘL Þ þ 72ð1 þ ΘL ΩL Þ

6 þ ΩL λL Lð2 þ aÞ
z̄N ¼ ð53Þ
3½ΩL λL Lð1 þ aÞ þ 2ð1 þ ΘL ΩL Þ

3z̄N ½λL LðLz Þ2 ð1 − aÞ þ 2ðΘL þ aλL L Lz Þ − λL LðLz Þ2 ½3a þ 2 Lz ð1 − aÞ


NðzÞ ¼ EAλL Lūl · ð54Þ
6ð1 − z̄N Þ

EAλ2L L ½að1 − Lz Þ þ Lz ðLz − z̄N Þ


τ ðzÞ ¼ · ūl · ð55Þ
2πr0 1 − z̄N

© ASCE 04020016-12 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


• Power law stiffness profile (a ¼ 0) Notation
1 The following symbols are used in this paper:
ūl ¼ αΔTΩL ð2 þ nÞð3 þ nÞ½2λL L þ 2ΘL ð1 þ nÞð2 þ nÞ
2 A = pile cross-sectional area;
× fΩL λ3L L3 þ λ2L L2 ð2 þ nÞ½2 þ ΘL ΩL ð1 þ nÞð2 þ nÞ AiryAi , AiryBi = airy functions;
a = coefficient regulating soil stiffness profile;
þ λL Lð2 þ nÞ2 ð3 þ nÞðΩL þ ΘL þ nΘL Þ
ai = dimensionless parameter for two-layer soil;
þ ð1 þ nÞð2 þ nÞ2 ð3 þ nÞð1 þ ΘL ΩL Þg−1 ð56Þ Ci = integration constants;
d = pile diameter;
ð1 þ nÞð2 þ n þ ΩL λL LÞ E = pile Young’s modulus;
z̄N ¼ ð57Þ G, G1 , G2 , Ḡ, GL , GL=2 = soil shear moduli;
ð2 þ nÞ½1 þ n þ ΩL λL L þ ΘL ΩL ð1 þ nÞ
Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

h1 , h2 = thickness of the shallower and deeper layer for the


  nþ1 two-layer soil;
z̄ 1 z I i ðxÞ = first type of modified Bessel function;
NðzÞ ¼ EAλL Lūl · ΘL N þ λL L
1 − z̄N 1 − z̄N L k, k̄, kL , ki = spring stiffness along the pile shaft;


z̄N z kb , kt = spring stiffness at the pile ends;


× − ð58Þ
1 þ n Lð2 þ nÞ L = pile length;
N, N max , N fixed = pile thermal axial loads;
N c , N q = pile bearing capacity factors;
EAλ2L L ð z Þn ð z − z̄N Þ
τ ðzÞ ¼ · ūl · L L ð59Þ n = coefficient regulating soil stiffness profile;
2πr0 1 − z̄N
Q = pile mechanical axial load;
r0 = pile radius;
2. Multilayer soil
su = soil undrained shear strength;
Referring to a multilayer soil with constant stiffness within
each layer, the virtual work and the vertical equilibrium equa- up = pile displacement;
tions can be written as follows: ūl , ū0 = dimensionless pile displacements;
zi , zi−1 = depth of the bottom and top interface of layer i;
ū2l Xi
zm , zm−1 = depth of the bottom and top interface of layer m;
kt L2 ū20 þ kb L2 ū2l þ km ½ðzm−1 − Lz̄N Þ3
3ð1 − z̄N Þ2 m¼1 zN¼Nmax , zτ ¼0 = depth corresponding to the zero shear stress
 and pile displacement;
Lū EA ūl
− ðzm − Lz̄N Þ3  ¼ l αΔT − ð60Þ z̄N = depth corresponding to the zero pile displacement
1 − z̄N 1 − z̄N (energy solution);
α = coefficient of thermal expansion of the pile material;
ūl ΓðxÞ = gamma function;
kt Lū0 þ kb Lūl þ
2ð1 − z̄N Þ ΔT = temperature variation;
X
i δ = coefficient for spring stiffness calibration;
× km ½z2m − z2m−1 − 2Lz̄N ðzm − zm−1 Þ ¼ 0 ð61Þ εΔT , εp = thermal axial strain;
m¼1 ζ = dimensionless interface depth;
Θ, ΘL = ratios between the top and shaft stiffness;
where the third term = the contribution of the shear stress acting
along the pile shaft for each i-th layer; and zm and zm−1 = the λ, λ1 , λL = pile-soil wavenumbers for constant stiffness;
depths of the bottom and top interfaces of layer m. μ = pile-soil wave number for Gibson soil;
By solving Eqs. (60) and (61), it is possible to obtain the two ν = Poisson’s ratio;
unknowns ūl and z̄N ; the axial load and shear stress distributions ξ = ratio between the thickness of the deeper and
have the following expressions: shallower layers;
 ρ = ratio between the shear modulus of the deeper and
ūl
N i ðzÞ ¼ −kt Lū0 − k ½z2 − z2i−1 − 2Lz̄N ðz − zi−1 Þ shallower layers;
2ð1 − z̄N Þ i τ = shear stress at the pile–soil interface;
X
i−1  Φ, Φ 0 = shape function and derivative of the shape function
þ km ½z2m − z2m−1 − 2Lz̄N ðzm − zm−1 Þ ð62Þ with respect to z; and
m¼1
Ω, Ω̄, ΩL = ratio between the shaft and base stiffness.

ūl
τ i ðzÞ ¼ − k ðz − zi−1 − Lz̄N Þ ð63Þ
2πr0 ð1 − z̄N Þ i References

where zi−1 = the depths of the top interface of layer i. Adinolfi, M., R. M. S. Maiorano, A. Mauro, N. Massarotti, and S. Aversa.
2018. “On the influence of thermal cycles on the yearly performance of
an energy pile.” Geomech. Energy Environ. 16 (Dec): 32–44. https://doi
.org/10.1016/j.gete.2018.03.004.
Data Availability Statement Amatya, B. L., K. Soga, P. J. Bourne-Webb, T. Amis, and L. Laloui. 2012.
“Thermo-mechanical behaviour of energy piles.” Géotechnique 62 (6):
Some or all data, models, or codes generated or used during the 503–519. https://doi.org/10.1680/geot.10.P.116.
study are available from the corresponding author by request (finite Bourne-Webb, P. J., B. Amatya, K. Soga, T. Amis, C. Davidson, and
element results). P. Payne. 2009. “Energy pile test at Lambeth College, London:

© ASCE 04020016-13 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016


Geotechnical and thermodynamic aspects of pile response to heat Murphy, K. D., and J. S. McCartney. 2015. “Seasonal response of energy
cycles.” Géotechnique 59 (3): 237–248. https://doi.org/10.1680/geot foundations during building operation.” Geotech. Geol. Eng. 33 (2):
.2009.59.3.237. 343–356. https://doi.org/10.1007/s10706-014-9802-3.
Brandl, H. 2006. “Energy foundation and other thermo-active ground struc- Mylonakis, G. 2001. “Winkler modulus for axially loaded piles.” Geotech-
tures.” Géotechnique 56 (2): 81–122. https://doi.org/10.1680/geot.2006 nique 51 (5): 455–461. https://doi.org/10.1680/geot.2001.51.5.455.
.56.2.81. Ng, C. W. W., A. Gunawan, C. Shi, Q. J. Ma, and H. L. Liu. 2016.
Chen, D., and J. S. McCartney. 2016. “Calibration parameters for load “Centrifuge modelling of displacement and replacement energy piles
transfer analysis of energy piles in uniform soils.” Int. J. Geomech. constructed in saturated sand: A comparative study.” Géotech. Lett.
17 (7): 04016159. https://doi.org/10.1061/(ASCE)GM.1943-5622 6 (1): 34–38. https://doi.org/10.1680/jgele.15.00119.
.0000873. Ng, C. W. W., C. Shi, A. Gunawan, and L. Laloui. 2014a. “Centrifuge
Coyle, H. M., and L. C. Reese. 1966. “Load transfer for axially loaded piles modelling of energy piles subjected to heating and cooling cycles in
in clay.” J. Soil Mech. Found. Div. 92 (2): 1–26. clay.” Geotech. Lett. 4 (4): 310–316. https://doi.org/10.1680/geolett
Downloaded from ascelibrary.org by Indian Institute of Technology Madras on 03/09/20. Copyright ASCE. For personal use only; all rights reserved.

Crispin, J., H. Su, K. Round, and G. Mylonakis. 2019. “Energy formulation .14.00063.
for axial pile head stiffness in inhomogeneous soil.” In Sustainability Ng, C. W. W., C. Shi, A. Gunawan, L. Laloui, and H. L. Liu. 2014b.
Issues for the Deep Foundations: Proc., GeoMEast 2018. Sustainable “Centrifuge modelling of heating effects on energy pile performance
Civil Infrastructures, edited by H. El-Naggar, K. Abdel-Rahman, in saturated sand.” Can. Geotech. J. 52 (8): 1045–1057. https://doi
B. Fellenius, and H. Shehata, 181–190. New York: Springer. .org/10.1139/cgj-2014-0301.
Di Donna, A., and L. Laloui. 2015a. “Numerical analysis of the geotech- Pasten, C., and J. Santamarina. 2014. “Thermally induced long-term
nical behaviour of energy piles.” Int. J. Numer. Anal. Methods displacement of thermoactive piles.” J. Geotech. Geoenviron. Eng.
Geomech. 39 (8): 861–888. https://doi.org/10.1002/nag.2341. 140 (5): 06014003. https://doi.org/10.1061/(ASCE)GT.1943-5606
Di Donna, A., and L. Laloui. 2015b. “Response of soil subjected to thermal .0001092.
cyclic loading: Experimental and constitutive study.” Eng. Geol. Rammal, D., H. Mroueh, and S. Burlon. 2018. “Impact of thermal solic-
190 (May): 65–76. https://doi.org/10.1016/j.enggeo.2015.03.003. itations on the design of energy piles.” Renewable Sustainable Energy
Rev. 92 (Sep): 111–120. https://doi.org/10.1016/j.rser.2018.04.049.
Di Donna, A., A. F. R. Loria, and L. Laloui. 2016. “Numerical study of the
Randolph, M. F., and C. P. Wroth. 1978. “Analysis of deformation of
response of a group of energy piles under different combinations of
vertically loaded piles.” J. Geotech. Geoenviron. Eng. 104 (12):
thermo-mechanical loads.” Comput. Geotech. 72 (Feb): 126–142.
1465–1488.
https://doi.org/10.1016/j.compgeo.2015.11.010.
Rotta Loria, A. F., and L. Laloui. 2017. “Thermally induced group effects
Goode III, J. C., and J. S. McCartney. 2015. “Centrifuge modeling of
among energy piles.” Géotechnique 67 (5): 374–393. https://doi.org/10
end-restraint effects in energy foundations.” J. Geotech. Geoenviron.
.1680/jgeot.16.P.039.
Eng. 141 (8): 04015034. https://doi.org/10.1061/(ASCE)GT.1943-5606
Rotta Loria, A. F. R., A. Vadrot, and L. Laloui. 2018. “Analysis of the ver-
.0001333.
tical displacement of energy pile groups.” Geomech. Energy Environ.
Karatzia, X., and G. Mylonakis. 2017. “Horizontal stiffness and damping of
16 (Dec): 1–14. https://doi.org/10.1016/j.gete.2018.04.001.
piles in inhomogeneous soil.” J. Geotech. Geoenviron. Eng. 143 (4):
Rui, Y., and K. Soga. 2019. “Thermo-hydro-mechanical coupling analysis
04016113. https://doi.org/10.1061/(ASCE)GT.1943-5606.0001621.
of a thermal pile.” Proc. Inst. Civ. Eng. Geotech. Eng. 172 (2): 155–173.
Knellwolf, C., H. Peron, and L. Laloui. 2011. “Geotechnical analysis https://doi.org/10.1680/jgeen.16.00133.
of heat exchanger piles.” J. Geotech. Geoenviron. Eng. 137 (10): Rui, Y., and M. Yin. 2018. “Investigations of pile–soil interaction under
890–902. https://doi.org/10.1061/(ASCE)GT.1943-5606.0000513. thermo-mechanical loading.” Can. Geotech. J. 55 (7): 1016–1028.
Laloui, L., and B. François. 2009. “ACMEG-T: Soil thermoplasticity https://doi.org/10.1139/cgj-2017-0091.
model.” J. Eng. Mech. 135 (9): 932–944. https://doi.org/10.1061 Stewart, M. A., and J. S. McCartney. 2014. “Centrifuge modeling of
/(ASCE)EM.1943-7889.0000011. soil-structure interaction in energy foundations.” J. Geotech. Geoen-
Laloui, L., M. Moreni, A. Fromentin, D. Pahud, and G. Steinmann. 1999. viron. Eng. 140 (4): 04013044. https://doi.org/10.1061/(ASCE)GT
“Heat exchanger pile: Effect of the thermal solicitations on its mechani- .1943-5606.0001061.
cal properties.” Bull. d’Hydrogéologie 17 (1): 331–340. Suryatriyastuti, M. E., H. Mroueh, and S. Burlon. 2012. “Understanding
Laloui, L., M. Moreni, and L. Vulliet. 2003. “Comportement d’un pieu bi- the temperature-induced mechanical behaviour of energy pile founda-
fonction, fondation et échangeur de chaleur.” Can. Geotech. J. 40 (2): tions.” Renewable Sustainable Energy Rev. 16 (5): 3344–3354. https://
388–402. https://doi.org/10.1139/t02-117. doi.org/10.1016/j.rser.2012.02.062.
Laloui, L., M. Nuth, and L. Vulliet. 2006. “Experimental and numerical Suryatriyastuti, M. E., H. Mroueh, and S. Burlon. 2014. “A load transfer
investigations of the behaviour of a heat exchanger pile.” Int. J. Numer. approach for studying the cyclic behavior of thermo-active piles.” Com-
Anal. Methods Geomech. 30 (8): 763–781. https://doi.org/10.1002 put. Geotech. 55 (Jan): 378–391. https://doi.org/10.1016/j.compgeo
/nag.499. .2013.09.021.
Ma, Q. J., C. W. W. Ng, D. Mašín, and C. Zhou. 2017. “An approach for Vardanega, P. J., and M. D. Bolton. 2011. “Predicting shear strength
modelling volume change of fine-grained soil subjected to thermal mobilization of London clay.” In Vol. 1 of Proc., 15th European
cycles.” Can. Geotech. J. 54 (6): 896–901. https://doi.org/10.1139/cgj Conf. on Soil Mechanics and Geotechnical Engineering, 487–492.
-2016-0459. Amsterdam, Netherlands: IOS Press.
McCartney, J. S., and K. D. Murphy. 2014. “Strain distributions in full- Yavari, N., A. M. Tang, and J. M. Pereira. 2014. “A simple method
scale energy foundations (DFI Young Professor Paper Competition for numerical modelling of mechanical behaviour of an energy pile.”
2012).” DFI J.-J. Deep Found. Inst. 6 (2): 26–38. https://doi.org/10 Geotech. Lett. 4 (2): 119–124. https://doi.org/10.1680/geolett.13
.1179/dfi.2012.008. .00053.

© ASCE 04020016-14 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2020, 146(5): 04020016

Das könnte Ihnen auch gefallen