Вы находитесь на странице: 1из 14

Computers and Geotechnics 108 (2019) 61–74

Contents lists available at ScienceDirect

Computers and Geotechnics


journal homepage: www.elsevier.com/locate/compgeo

Review

A review of pile-soil interactions in isolated, thermally-activated piles T


a,⁎ a b
P.J. Bourne-Webb , T.M. Bodas Freitas , R.M. Freitas Assunção
a
CERIS, Instituto Superior Técnico, Universidade de Lisboa, Lisboa, Portugal
b
GEOTEST AG, Lausanne, Switzerland

ARTICLE INFO ABSTRACT

Keywords: A number of full- and small-scale tests and numerical studies examining the behaviour of isolated thermal piles
Shallow geothermal energy have now been published. The results have been collated and examined in terms of the pile head displacement
Energy geostructures and maximum stress changes observed during thermal loading. These parameters are complementary and are
Piles key design parameters. Taking full-scale tests as a benchmark, it is apparent that small-scale and numerical
Pile-soil interaction
models are systematically biased towards low pile restraint. With some caveats, the limits defined by perfect
Thermo-mechanical
restraint and free movement may be used in design to provide a first-order estimate of pile response.
Physical modelling
Finite element modelling

1. Introduction • Thermally-induced pile head/pile cap movement, y th,0.

The utilization of the pile foundations that support buildings as al- Cyclic thermal loading and group effects are important considera-
ternative ground-coupled heat exchangers in shallow geothermal en- tions in design but are not included here so as to maintain the focus on
ergy systems has received an increasing amount of attention, especially the key responses of isolated piles under relatively simple thermal
in the past 10–15 years, amidst efforts to establish renewable and sus- conditions.
tainable, heating and cooling systems within the construction industry. As a first estimate, it has been considered that the response of a pile
The concept was first applied in Austria [1] but now has been used to heating and cooling could be bounded by assuming that it was either:
across the Globe [2]. In many situations, its potential use is impeded
not only by a lack of awareness amongst clients and construction pro- • Perfectly restrained, i.e. the maximum stress change,
fessionals, but also a lack of knowledge regarding the impact of heating σ = α ·ΔT·E , or
th,fixed c c
and cooling on the behaviour of the foundations. A number of experi- • Perfectly unrestrained, i.e. the maximum axial deformation,
mental studies both in the field and laboratory, supplemented by nu- yth,free = αc·ΔT·L
merical studies, have been undertaken to examine these technical is-
sues. In the following, this information has been collated and is used to where αc is the linear coefficient of thermal expansion, ΔT the change
provide some clarity regarding the potential impact on performance in temperature, Ec the Young’s modulus of the pile and L the initial
when isolated pile foundations are thermally-activated. length of the pile. The limit defined by σth,fixed seems to be supported by
In relation to thermally-activated piles, the thermally induced in- observational data published to date; thermal stress data taken from
teraction between piles and the soil is characterized by a balancing [20,21] and presented in [5], Fig. 1(a) & (b) and from [6], Fig. 1(c),
between movement and the alteration of internal stresses within the show how the mobilized thermal stress changes in various test piles lie
pile. If movement is restrained, internal stresses will increase and vice at or below the limit defined by perfect restraint of the pile.
versa, in a similar manner to passive pile loading conditions such as
“down drag” [3,4]. The expression of this interaction is examined in 2. Framework for pile-soil thermal interaction
terms of the pile head movement and changes in internal stress due to
thermal loading in isolated piles. In the following, the pile thermal Before examining these interactions further, what is happening in a
response will be characterised by two parameters, i.e. thermally-activated pile should be considered. Using the same frame-
work described by [7], Fig. 2 illustrates, in a simplified way, the
• Maximum thermally-induced stress change in a pile, σ th,max; thermal responses mobilized in a floating pile of length, L embedded in


Corresponding author.
E-mail address: peter.bourne-webb.co.uk@tecnico.ulisboa.pt (P.J. Bourne-Webb).

https://doi.org/10.1016/j.compgeo.2018.12.008
Received 23 April 2018; Received in revised form 20 November 2018; Accepted 10 December 2018
0266-352X/ © 2018 Elsevier Ltd. All rights reserved.
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

Fig. 1. Thermally induced axial stress change in energy pile tests and limit implied by perfect pile restraint with Δσ = αc·ΔT·Ec; (a) and (b) after Amatya et al. [5] &
(c) after Murphy et al. [6].

D yth,0/yth,free
yth,0
F 0.5 1.0

Hn,2H
Heating
Hn,1 Cooling 2H 2H
Fth,2H
Hn,2C
L Fth,1 Fth,1 Hn
1C 1H 0.5
Heating L

Cooling Fth,2C
2C 2C

L L 1.0

Fig. 2. Model of changes in thermal axial force and pile head movement in thermally-activated pile.

a thermally-inert soil providing a constant shaft resistance and negli- axial load, Fth,2 will depend on the value of F and how this is
gible base resistance. Two situations are represented: transferred to the soil along the pile (e.g. in this case, as the shaft
resistance is constant, the load transfers linearly with depth as in-
(1) No external mechanical load is present (Fig. 2(b)) – the pile expands dicated by the grey-shaded area in Fig. 2(c)).
or contracts about a neutral point which with a constant shaft re-
sistance is at a depth Hn,1 = L/2. As there is no base resistance Additionally,
mobilised, the pile expands/contracts equally about this point and
yth,0/yth,free ≤ 0.5 (Point 1H & 1C, Fig. 2(d)), the value of yth,0 de- (3) If the soil restraint increases with depth (the more usual situation in
pending on the restraint of the pile shaft imparted by the ground. nature) with F = 0, the initial depth to the neutral point Hn,1 will be
An associated thermal axial load, Fth,1 which is a maximum at the greater than L/2, the associated value of yth,0/yth,free will be greater
neutral point (yielding σth,max) is also mobilised. than 0.5 but the relative effect of external loading, and heating and
(2) External mechanical load, F > 0 is applied (Fig. 2(c)) – now, be- cooling will be the same as discussed above.
cause of the load added at the head of the pile, the neutral point (4) If either end is restrained in any way, then a reaction opposing the
must move in order to achieve force equilibrium along the pile thermal deformation will develop that will tend to draw the neutral
shaft. When the pile is heated the neutral point will rise (Hn,2H) and point toward the restrained end of the pile. For example, if re-
when cooled, descend (Hn,2C). Thus, yth,0/yth,free is less than 0.5 for sistance is mobilised on the base of the pile, the depth to the neutral
heating and greater than 0.5 for cooling. The associated thermal point will increase and so will yth,0/yth,free.

62
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

Fig. 3. Pile head movement during thermal loading, yth,0 reported from various field & model tests, normalised with respect to limit defined by free expansion of pile
– yth,free = αc·ΔT·L0. (Note that each result is plotted in a separate bin along the x-axis.)

In addition to the soil being considered thermally-inert, the above The results from Laloui et al. [21] clearly illustrate how the pro-
discussion also assumes that there are no other external thermal loads gressive construction of an overlying building between each thermal
involved; both these assumptions are discussed in Section 5. heating test (numbered 1–7) served to suppress the observed pile head
movement, with initial values of yth,0/yth,free in excess of 0.5 when just
the substructure was present (Test 1), reducing to values much less than
3. Observation of pile-soil thermal interaction 0.5 as each floor was constructed and pile head restraint increased (Test
2–7). Fig. 4 illustrates how the axial stress changes mobilized in the
Fig. 3 collates pile head deformation observations normalised with tests presented in [21] are complementary and increase as the pile head
respect to yth,free from full- and small-scale thermal tests on single piles restraint increases (while yth,0/yth,free reduces).
[6,8–21]. With only a few exceptions, the movement limit defined by It is also possible that in some of the tests with yth,0/yth,free values
the pile head being perfectly unrestrained appears a reasonable upper less than 0.5, the thermal loading may not have reached steady state,
limit, and an average value for yth,0/yth,free of around 0.6 is obtained. e.g. [18–20]. Overall, however, the behaviour seen in the heating tests
Note that while [22,23] form part of the data set, they do not report is consistent with the interaction mechanism described in Fig. 2, in the
yth,0. In this representation, as best as can be ascertained, the data relate presence of an external mechanical load, during heating, the neutral
to alterations in the pile response during the thermal testing and all point will be in the upper part of the pile and pile head deformation will
initial effects, i.e. external loading, residual stresses due to curing or be less than about 0.5yth,free.
driving and self-weight, have been zeroed out. Some of the data appears to suggest that the pile expands/contracts
Only limited data for pile cooling are available (denoted with either more than the theoretical free expansion value, this may be due to (i)
open-symbols or (C) in Fig. 3) but it is interesting to note how the yth,0/ errors in the measurements, (ii) the pile coefficient of thermal expan-
yth,free ratio from the cooling tests are consistently larger than those sion (CTE) used to estimate the free thermal expansion, yth,free being
from the heating tests. The discussion of Fig. 2 highlights that during incorrect (noting that many authors use a value of αc = 10 με/°C based
cooling and in the absence of head restraint/load, the neutral point will on suggested values in e.g. [24] for reinforced concrete and not a ma-
be at or below the mid-depth of the pile and yth,0/yth,free will be greater terial specific value), (iii) thermal effects in the surrounding soil mass
than 0.5; these data are consistent with this. [25,26,38], i.e. the soil mass expands/contracts leading to greater pile
When it comes to heating, most of the results lie below yth,0/ movement than just the pile expansion/contraction.
yth,free = 0.5. Those tests where yth,0/yth,free is greater than 0.5 tend to Regarding this latter possibility, it can be postulated that if the shaft
have zero or near-zero head load/restraint, [8,9]. In Ng et al. [13], resistance on a pile is fully mobilized by external loading, then it is
there appears to be some thermal soil compaction mechanism in play, possible that when cooled, the apparent thermal settlement of the pile
resulting in a gradual reduction of yth,0/yth,free as the test proceeded. head could exceed the free thermal movement, because the pile will
For the results from [6,17] a range of yth,0/yth,free is shown as the pile settle downwards as the pile toe tries to contract upwards. This appears
movements were reported as the total axial extension based on in- evident in some model tests, e.g. [9,10] where the reported settlement
tegrating axial strains from the pile base to its head, from which yth,0 during cooling is approximately equal to or greater than the free
has had to be estimated, see note (2) in Fig. 3.

63
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

Fig. 4. Thermal axial stress change, σth,max & pile head movement, yth,0 normalised with respect to limits defined by perfect restraint, σth,fixed = αc·ΔT·Ec & free
expansion, yth,free = αc.ΔT·L0.

thermal contraction, Fig. 3. In the second test, an external load was introduced and then heating
Where available, these observations of thermally-induced pile head (H10) followed by a thermal recovery period and then cooling (C10)
movement have been collated with reported maximum thermal stress were applied, as in the first test. Large changes in yth,0/yth,free develop
changes, σth,max (normalised with respect to σth,fixed) in the expectation with only rather small changes in σth,max/σth,fixed. With the external
that there should be some compatibility between the values as dis- loading present, during heating the depth to the neutral point and thus,
cussed earlier, Fig. 4. It is apparent that with the exception of You et al. yth,0/yth,free are expected to reduce (as per the earlier discussion) and, as
[19], the full-scale test results illustrate the complementarity of thermal the pile base can apparently resist tension, during cooling the neutral
deformation restraint and mobilized stress change, i.e. larger pile head point should move down and yth,0/yth,free increase; this is what is seen
movements are associated with lower internal stress changes. in the results.
The tests reported by [19] appear to be contrary in that the max- Examining the thermal stress changes during the tests reported by
imum mobilised thermal stresses increase in spite of a reported increase [15] is rather more difficult, due to the significant base reaction that
in the pile head movement. It is thought this discrepancy might be was mobilised and because significant residual stresses appear to have
explained by either an error in the reported results or the pile loading been locked into the response during each stage of the test. What is
arrangements, which involved a gravel pack between the pile head and apparent is that these tests and results from other small-scale tests on
the loading plate where the displacement measurements were taken, floating piles whether in the centrifuge [11–13] or at 1-g [14–16], yield
and which could have led to erroneous movement observations. The similar results, with rather low values of σth,max/σth,fixed being mobi-
interpretation of this test is further hindered by the absence of any lized while rather large changes in yth,0/yth,free occur. This is not com-
instrumentation in the bottom third of the pile shaft. patible with the field results. Chen and McCartney [29] noted a similar
The results from small-scale centrifuge tests in [11,12] demonstrate discrepancy when comparing model parameters, after back-analysing
the effect of a rigid end-bearing layer (the base of the soil tank). As large and small scale thermally-activated pile tests using the load-
discussed in Section 2, as the pile base reaction becomes stiffer the transfer method.
neutral point moves down, and this is what is seen [11] with the ratios The reduced variation in σth,max/σth,fixed is thought to be due to
of yth,0/yth,free and σth,max/σth,fixed both trending towards a value of 1 (in there being low confining stresses and therefore limited shaft resistance
this case both are about 0.8). The neutral point is at the pile toe thus the being available. Low confinement on the pile shaft could be due a
maximum stress occurs here, and the pile cannot expand downwards, so number of reasons:
the thermal expansion due to heating is largely expressed at the pile
head. An interesting point here is the comparison between floating piles (a) The way in which the soil is placed within the test tanks, although
and end-bearing piles highlighted in these two studies, especially how if the results from [12] for a compacted silt are similar to those in
the neutral point moves to the end of the pile then the potential max- other tests in sand which used air-pluviation [13–16];
imum stress change is far higher than if the pile was expanding about a (b) The limited depth of burial in 1-g models, though the results for the
neutral point part-way down the pile because the whole soil profile is tests seem to be similar whether at 1-g or multiple-g values.
then resisting the pile expansion. (c) The tests with the lowest σth,max/σth,fixed response were in saturated
In Wang et al. [15], a first test sequence involved heating (H0) sand [13] at 40 g and [16] at 1 g, where the effective confining
followed by a thermal recovery period and then cooling (C0) without pressures would half those in an equivalent dry sand. The dry sand
any applied load. In this test, very little alteration of either yth,0/yth,free tests generally yielded σth,max/σth,fixed values more than twice those
or σth,max/σth,fixed between heating and cooling was observed which is in the saturated sand tests.
consistent with the discussion of Fig. 2 when F = 0. In the test, a sig-
nificant load reaction was provoked at the pile base during heating and Another factor that could be acting in combination with this are the
cooling. Tensile stress developed at the pile base during cooling but rather small thermal displacements (e.g. in [15], yth,free is about 0.3 mm
how this was possible is not clear. Because of this base reaction, during during heating and 0.1 mm in cooling for a 1.6 m long model pile and
both heating and cooling the neutral axis therefore moved down and many of the test piles are less than half this length), which if you as-
yth,0/yth,free is greater than 0.5 as expected from the earlier discussion. sume that full mobilization of shaft resistance requires a few

64
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

Table 1
Coefficient of thermal expansion (CTE) for soil, ratio relative to pile CTE, and soil Young’s Modulus used in numerical studies examining energy pile behaviour.
Source Linear CTE, αs/αc Soil Modulus, Surface BC
αs (με/K) (–) Es (MPa)

Single piles
[30] Wang et al. 11.7 1.46 12 Const. T = T0
[31] Suryatriyastuti et al. 5.0 0.5 10 Adiabatic
[32] Bodas Freitas et al. 0, 5, 20 0, ½, 2 30, 60 T = T0; Adiabatic
[33] Ma et al. – Sand 10 1.17 36 Not stated
– Clay 10 1.17 –
[34] Suryatriyastuti et al. 5.0 0.32 10 N/A
[35] Ozudogru et al. 0.0 0.0 12.5, 25, 50 Not stated
[36] Saggu & Chakraborty 100 10.0 39 Const. T = T0
[37] Wang et al. 8.7 0.54 – Const. T = T0
[38] Bourne-Webb et al. 0, 5, 20, 40 0, ½, 2, 4 30 Const. T ≠ T0
[39] Khosravi et al. 6.0 0.46 1823 Not stated
[40] Gawecka et al. 17 2.0 f(strain) Const. T = T0

Single piles & pile groups


[41] Di Donna et al. 6.7 0.56 201 Const. T ≠ T0
[42] Salciarini et al. 30 3.0 382 Const. T = T0
[43] Dupray et al. 40 3.33 30 Const. T ≠ T0
[44] Jeong et al. – Sand 10 1.0 50 Not stated
– Clay 10 1.0 5.0
[45] Di Donna & Laloui 6.0 0.5 251 Const. T ≠ T0
[46] Rotta Loira et al. 10 0.45 11 Const. T = T0
[47] Tsetoulidis et al.5 – Fill 5.6 0.66 20 Const. T = T0
– Clay 10.0 1.18 1056
[48] Suryatriyastuti et al. 5.0 0.4 10 Not stated
[49] Salciarini et al. 10, 35, 704 1, 3.5, 7.0 30 Adiabatic
[50] Rotta Loria & Laloui 10 0, 0.5, 1, 2 60–90 Adiabatic
[51] Alberdi-Pagola et al. 15 0.5 30 Const. T = T0

Notes:
1
Estimated from nonlinear parameters quoted, value at pile mid-depth.
2
Weighted average along pile.
3
Bulk & Shear modulus not compatible with Poisson ratio quoted.
4
Soil skeleton values quoted (αeff/αc ratios are 9, 10 & 13).
5
Also single pile considered.
6
E = 800 cu with average cu of 131 kPa.

millimetres of displacement would lead to associated small thermally- exceptions, consistently predicted values of σth,max/σth,fixed less than
induced changes in mobilized shaft and base resistance. It is known that about 0.5, whereas the full-scale observations tend to lie above about
on soil-structure interfaces, the displacement to mobilize the ultimate 0.30 with most between about 0.4 and 0.9. The numerical studies that
shearing resistance does not scale or at least the scaling is not well reported larger values of σth,max/σth,fixed were those where the adopted
understood [27,28], and it is possible that this effect in combination soil parameters had been calibrated based on a full-scale test, e.g.
with small thermal displacements, could reduce further the mobilized [39,47], resulting in the adoption of large stiffness values in the ana-
thermal axial stress response. lyses compared with other numerical studies, Table 1.
When the pile is cooled there is a possibility that the pile could Further to the earlier comments regarding the development of ten-
develop tensile stress. This was demonstrated in [19,20,23] but has sile forces in the pile shaft, numerical studies such as [40,47] have also
perhaps not been reported in more tests as (a) the cooling loads have predicted tension however this likely due to their having been based on
not been as large or (b) the restraint of the pile is lower. parameters back-analysed from [20]. Otherwise, tension has been
The effect of the available shaft resistance in limiting the thermal predicted under special circumstances only, e.g. in [51], the pile head
stress changes that can be mobilized is explored further in Section 5. was fixed and when cooled the pile went into tension with the max-
imum value at the point of fixity. With the current state of knowledge, it
is quite difficult to state reliably when tension might arise as it will be
4. Numerical analysis of pile-soil thermal interaction
highly dependent on the particular conditions being considered. The
lack of analyses reporting tension during cooling could also be due to
In recent years, a number of studies have also been published where
apparent lack of pile restraint captured in the studies, as discussed
numerical modelling was applied to the analysis of isolated thermally-
above.
activated piles, Table 1 [30–51]. In each of these studies, models of
Bodas Freitas et al. [32], noted that increasing the soil stiffness by a
varying complexity have been implemented. The results from these
factor of two led to an increase in the axial thermal stress changes, (E)
studies are compared with the observed responses in terms of the pre-
and (2E) Fig. 6. Young’s modulus values from the numerical studies are
dicted yth,0/yth,free and σth,max/σth,fixed in Figs. 5 and 6, respectively.
summarised in Table 1. Compared with e.g. [39,47], the stiffness values
Broadly speaking, the numerical analyses report similar variations in
adopted in the remaining numerical studies are significantly lower,
pile head restraint to the field and small scale tests, including values of
which may explain in part why many of the numerical studies are
yth,0/yth,free in excess of 1.0, Fig. 5. To be compatible with the large- and
producing predictions commensurate with moderate to large yth,0/
small-scale test results, the responses from the numerical analyses
yth,free ratios and small σth,max/σth,fixed ratios, Fig. 6.
presented here and later, are only those due to the influence of thermal
It is clear that while numerical models may reproduce field and
loading.
model tests in a qualitative way, there is still some work required to
It is interesting to note that in terms of thermal stress changes
understand these differences and ensure that the full range of possible
provoked in the pile (Fig. 6), the numerical studies, with only a few

65
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

Fig. 5. Observed and numerically predicted values for normalised pile head movement.

soil restraint is examined. changes due to heating becoming less compressive and in the case of
Bodas Freitas et al. [32] also report that while the results in terms of larger CTE ratios, becoming tensile. These effects are also a function of
thermally induced axial stress changes were broadly in line with ob- the pile length and the surface temperature boundary condition imposed
servations and other numerical studies, the temperature field around by the overlying structure. Regarding the latter, it is apparent in Table 1
the pile, in combination with the thermal characteristics of the pile and that most numerical studies have assumed that either the upper
soil also appeared to play major roles in determining the final pile-soil boundary temperature is unchanged (Constant T = T0) or is adiabatic;
interaction response. This finding was examined in greater detail by [32] highlighted the differences in analysis results that these two as-
[38] who confirmed the importance of these parameters in the devel- sumptions would cause and [38] expanded on this by examining cases
opment of internal stress changes derived from thermal loading, Fig. 6. with different constant temperature surface boundary conditions (Con-
This effect has also been illustrated for pile groups by [50]. stant T ≠ T0), the results of which can be seen in Fig. 6.
Comparing the results of [38] to other numerical predictions (Fig. 6), Bourne-Webb et al. [38] focused solely on the internal stress changes
in terms of σth,max/σth,fixed it is apparent that some quite different beha- (Figs. 6 and 7) and did not discuss in detail the complementary de-
viours are being predicted. When the ratio of the CTE for the soil to that formation behaviour of the thermally-activated pile, largely because the
of the pile, αs/αc is less than 1, the results are comparable to other nu- authors were interested in examining the limitations of the “framework”
merical studies (where in most cases αs/αc was also less than 1, Table 1); presented in [7]. In the next section, the issue of soil-pile thermal inter-
stress changes during pile heating are compressive and increase in action is revisited in terms of the pile deformation that complements the
magnitude with pile length (see also Fig. 7). However, as αs/αc increases, thermal axial stress changes reported in [38] and puts these into context
there is a distinct change in the predicted response with axial stress with the observations and other numerical studies summarised above.

Fig. 6. Observed and numerically predicted values for normalised thermal stress change.

66
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

Fig. 7. Effect of pile length and ratio of soil to concrete CTE (αc = 10 με/°C) on pile response to heating in terms of maximum mobilized axial stress, σth,max, after
Bourne-Webb et al. [38].

5. Complementary pile-soil interaction analysis Table 2


Material properties of soil and concrete used in analyses.
5.1. Basis Soil Concrete

The numerical analyses presented here were undertaken on the Density, kg/m 3
1600 2450
same basis as those reported by [38]. Steady-state thermal conditions Young’s modulus, E (MPa) 30 30,000
Poisson’s ratio, ν 0.3 0.2
were modelled, it is assumed that no excess pore water pressures were
Cohesive shear strength, c (kPa) 75 –
generated by the thermal loading, and the only thermo-mechanical Pile-soil interface adhesion, a (kPa) 75 –
coupling considered was that between temperature and volumetric Linear CTE, α (με/°C) 0, 5, 20 & 40 10
change, via the CTE. It is acknowledged that by assuming steady-state CTE Ratio, αs/αc 0, 0.5, 2, 4 –
Thermal conductivity, λ (W/m°C) 1 2
conditions and disregarding hydraulic coupling, the predicted response
will be biased towards greater movement and lower stress changes. This
effect is highlighted by [40], who showed how pile head movement
play and to allow comparison with the framework in [7].
increased and axial thermal stress change reduced with time and that
The pile-soil interface was modelled by attributing to the contact
this effect is more important for stress change (20–40%) than pile
between the two materials, a shearing resistance that increases linearly
movement (c. 20% change).
with relative displacement, reaching a maximum of 75 kPa at 0.0015 m
A single 1 m diameter floating pile was modelled “wished-in-place”
relative displacement. This latter value is not an intrinsic property and
with a length of either 15, 30 or 45 m (length to diameter ratios, L/D of
was arrived at by simulating pile static load tests numerically and se-
15, 30 & 45), and axis-symmetry was assumed with the bottom
lecting the value which yielded a reasonable load-displacement re-
boundary at a depth of 90 m (fixed vertically & horizontally) and the
sponse [52].
side boundary at a radius of 60 m (fixed horizontally). These boundary
In this study, only soils that expand linearly when heated or cooled
dimensions were established to ensure that their presence did not affect
were considered. As in [38], in order to consider the potential range in
the thermal-mechanical response of the pile [52].
soil CTE for moderately to highly OC clay and granular soils, the ratio of
The entire problem domain was assumed to have a constant initial
soil to concrete CTE, αs/αc has been varied such that it is either zero,
temperature of 15 °C and thermal boundary conditions were set such
0.5, 2 or 4 (Table 2). The likely range of values for the CTE of concrete,
that this temperature was maintained on the bottom and side bound-
αc is about 6–14 με/°C, depending on the aggregate type used, and a
aries, while the line of symmetry was adiabatic. The effect of the surface
value of 10 με/°C was adopted in this study which is in line with sug-
boundary temperature was examined on the basis that the pile was
gested values for the determination of thermal action effects in e.g.
located centrally within the footprint of a wide building. Imposed
[24].
surface temperatures of 12 °C, 18 °C and 24 °C were applied, where the
latter two were considered to be the likely extremes for a human oc-
cupied, climate controlled structure [52] while the former does not 5.2. Pile-soil interaction response
represent a particular condition but acts as a bound for the results,
allowing resulting trends to be examined. It was assumed that the heat A parametric numerical study based on the above assumptions was
exchanger pipes were arranged around the circumference of the pile, as undertaken in order to examine the impact of variations in the pile
though attached to a reinforcing cage, and that the thermal loading length, soil CTE and surface temperature on the interaction between the
could be modelled by applying a uniform temperature change to all the pile and the soil. The results are summarised in Figs. 8–12.
pile elements [52]. Here the results for a uniform pile temperature Fig. 8 is complementary to Fig. 7, and shows the normalised pile
change of +30 °C with respect to the initial temperature are presented head movement due to the combined effect of the applied surface
in detail; cooling was also investigated however the patterns of re- temperature boundary condition and the pile thermal loading. What is
sponse are broadly the same as for heating, as discussed later. immediately apparent is that the shorter the pile, the hotter the surface
Material properties assumed for the soil and concrete are detailed in and the higher the CTE ratio, the greater the likelihood that yth,0 will
Table 2. The pile is assumed to behave elastically while the soil is exceed yth,free. It is possible for this to occur because there are two
modelled as linear elastic-perfectly plastic with a Tresca type failure thermal actions acting in harmony, i.e. when the change in surface
criterion, i.e. purely cohesive. Initial stress conditions assumed the temperature is positive with respect to the initial temperature, the soil
coefficient of earth pressure at-rest, K0 was 1.0. The adopted soil model & pile expand then, as the pile is heated, both the pile and the soil
parameters were not intended to represent a particular geotechnical expand further, and the combined effect may exceed yth,free. It is also
condition but were used in order to control the number of variables at apparent that as the pile length increases, the impact of the surface
temperature and CTE ratio is reduced significantly in terms of pile

67
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

Fig. 8. Predicted values for normalised pile head movement yth,0/yth,free as a function of pile length, surface temperature and CTE ratio.

movement, however the associated thermal stress changes increase,


Figs. 7 and 9.
In Fig. 9, the interaction between movement and internal stress
changes is clearly demonstrated, with small stress changes associated
with greater pile movement. As previously noted, while the interaction
between pile movement and stress change brought about by thermal
loading appear compatible, numerically predicted mobilized stress
changes seem rather small compared to those reported in field tests,
Fig. 6.
As was discussed earlier, this discrepancy may be associated with
the adopted soil parameters. In Fig. 10 this hypothesis is examined by
considering the flowing three scenarios:

(i) Doubling the soil stiffness,


(ii) Quadrupling the stiffness,
(iii) Quadrupling the stiffness & doubling the soil strength and pile-soil
interface adhesion.

As the soil stiffness and/or pile-soil adhesion is increased, there is a Fig. 10. Values for normalised thermal response parameters, σth,max and yth,0
progressive increase in the mobilised σth,max values, Fig. 10. These re- (pile length 30 m; surface temp. 18 °C) as a function of increased soil stiffness,
sults are as expected but still represent values at the lower end of those Esoil and cohesive strength, cu.
inferred from field testing. It is also interesting to note that for a given
CTE ratio, the changes in soil stiffness had very little impact in terms of
the pile and available pile-soil interface resistance also impose a limit
the movements at the pile head (yth,0); there is only a very small change
on the maximum axial stress change that can be caused by thermal
in yth,0 as the soil stiffness and/or pile-soil adhesion were increased,
movements (Appendix A and [53]).
denoting that there is very little additional restraint to the pile move-
In Appendix A, limits are defined for a floating pile with negligible
ment as a consequence of these changes.
end restraint, based on one of two assumptions:
While attempting to understand this effect better, it was concluded
that not only is the potential thermal stress change limited by that
(1) The adhesion on the pile-soil interface is assumed to be fully mo-
generated when the pile is perfectly restrained (σth,fixed), the length of
bilized and constant along the pile shaft, either side of the neutral

Fig. 9. Values for normalised thermal response parameters, σth,max and yth,0 as a function of pile length, surface temperature and CTE.

68
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

point. Because the adhesion is constant, the neutral point is at the Having made a rather simple assumption about the load-transfer
pile mid-depth and consequently, for the values adopted, the mechanism at the pile-soil interface, it is apparent that these latter
maximum axial thermal stress change that can occur is 25%, 50% limits are a close approximation for the maximum stress changes pre-
and 75% of σth,fixed for the 15, 30, and 45 m long piles respectively. dicted by the FE numerical analysis (both compressive and tensile) for
In this case, ignoring pile base restraint has a less than 10% effect the 15 m long pile and are approximately double those predicted for the
on the ratio of σth,max/σth,fixed, as depending on pile length, the base 30 and 45 m long piles, Fig. 9. This also serves to explain the reduction
resistance represents only 5% to 15% of the total resistance. Also, it in yth,0/yth,free and increase in σth,max/σth,fixed as pile length increases,
can be seen from the analyses that the yth,0/yth,free ratio is ap- shown in Fig. 9.
proximately 0.5 when the CTE ratio is zero (Fig. 8), implying near Further, it also provides an explanation for why the analyses pre-
symmetrical expansion either side of the neutral point and thus, sented here and in e.g. [31,46] which modelled sand with low values
negligible base resistance is being mobilised. for interface friction angle (δ = 28° & 30° respectively) and coefficient of
(2) Acknowledging that the actual axial deformations are unlikely to earth pressure at-rest (K0 = 0.5 & 1 − sin φ′ respectively), do not mo-
result in the full mobilization of the shaft resistance over the full bilise large proportions of σth,fixed. However, those that involve cases
length of the pile, a simple linear load-transfer function based on with large shaft resistance do mobilise large proportions of σth,fixed, e.g.
5 mm of relative movement on the soil-pile interface being required [39,47] replicated field cases where the ground is stiff, i.e. profiles
to fully mobilize the shaft resistance is used and it is assumed that dominated by sandstone and firm to stiff London Clay respectively. This
the thermal expansions are unimpeded. This leads to the maximum observation may also be applied to the small-scale tests discussed ear-
axial stress change that can occur reducing to about 5.6%, 23% and lier.
47% of σth,fixed for the 15, 30, and 45 m long piles respectively (with Another aspect of the results that needs to be considered further is
the full shaft resistance only being mobilized on part of the 45 m the coupling of the effect of the imposed surface boundary condition
long pile, and not at all on the others). and the pile thermal loading. In reality, both are linked, as the internal
Here, allowing that the ultimate base resistance would only be climate conditions within the overlying structure will determine the
mobilised at displacements in excess of 20% of the pile diameter surface boundary condition, and are controlled by the operation of the
(typical for bored piles), then at a displacement of yth,free/2 only a climate control system which may rely in part or entirely on the energy
very small proportion of the resistance would be mobilised and the foundations. However, it is interesting to examine the two effects se-
ratio σth,max/σth,fixed, would only change by a few-tenths of a per- parately.
cent. Obviously, this might change for displacement type piles and Fig. 11(a) shows the results in terms of yth,0/yth,free and σth,max/
should be investigated further. σth,fixed for a 30 m long pile with only an overlying climate-controlled

Fig. 11. Values for normalised thermal response parameters, σth and yth comparing application of surface temperature only, to the combined effect with pile thermal
loading (ΔT = 30 °C) and the incremental effect (difference) of pile thermal loading for 30 m long pile.

69
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

structure (open symbols) versus one where in addition, the pile is and that of the pile thermal load oppose each other. The application of
thermally-activated (filled symbols). It is apparent that simply altering the surface temperature causes the pile head and the soil close to the
the surface temperature boundary condition can result in significant surface to expand and move upwards, its magnitude increasing with
thermal effects within the piles that may augment the effect of pile CTE ratio. Then, when the pile is cooled, the pile and the surrounding
heating/cooling or mitigate it. These effects will happen irrespective of soil contract (due to the temperature field created by the pile cooling).
whether the foundation is thermally-activated or not, and currently are For the case when the surface temperature is set at 18 °C, overall,
not considered in design, even though they may be of a similar mag- the soil close to the pile head contracts, resulting in a (downward) pile
nitude to the effect of thermally-activating a pile foundation. For the head movement larger than that for the case when the soil CTE is equal
15 m and 45 m long piles, the application of the surface temperature to zero (its value increasing with CTE ratio).
boundary condition alone leads to similar effects to those illustrated For the case when the surface is set at 24 °C, the opposite occurs.
here. However, with increasing pile length associated yth,0/yth,free ratios Overall, the soil close to the pile head dilates and the pile head has a
reduce, σth,max/σth,fixed ratios increase and the influence of the CTE ratio smaller (downward) movement than that for the case when the soil CTE
becomes less pronounced. is equal to zero, becoming negative (upward) for a CTE ratio equal to 4
It appears that there is an almost uniform difference between the (which is why this result plots as a negative Yth,0/yth,free ratio). Because
two sets of results shown in Fig. 11(a). This difference represents the the movement of the soil close to the pile head is in the opposite di-
additional change in pile response due to the thermal-activation of the rection to that of the pile which contracts, the soil acts as a restraint and
pile, and on examination (Fig. 11(b)), the thermal deformations and there is an increase in the tensile axial stress change (that becomes
thermal stress changes vary rather uniformly as the CTE ratio increases. more significant as the CTE ratio increases).
There remains some difference between the analyses with differing Referring back to Fig. 9(b), when instead the 30 m long pile was
surface temperature boundary conditions; these are small however and heated and the surface cooled in relation to the initial temperature field
are thought to be due to local interactions in the temperature field close (T0 = 15 °C), the induced thermal load due to the application of the
to the pile and in particular at the pile head. surface BC is small (ΔT = −3°C) when compared with the pile thermal
Bourne-Webb et al. [38] suggested that the discrepancies in- load (ΔT = +30 °C) and therefore, its effect is not so noticeable.
troduced by using models such as the load-transfer technique (that in
effect assume the soil CTE is zero), might be able to be corrected, at 6. Conclusions
least for the conditions explored in the reported analyses. Examining
Fig. 11(b), this appears to hold true for the response due to pile thermal Based on an extensive review of reported large- and small- scale
loading without inducing too much error. However, while it needs to be model tests, and analytical studies, the following conclusions have been
studied in greater depth, external thermal loads clearly have a sig- drawn regarding the thermal response of isolated thermally-activated
nificant effect that cannot be readily incorporated in load-transfer piles:
models.
The preceding discussion has been based on the pile being heated by • The interaction problem associated with energy pile thermal loading
30 °C; cooling was also considered and Fig. 12 illustrates the response of is such that it is reasonable to expect that there should be some
a 30 m long pile when the pile is cooled by 15 °C with respect to the relationship between the internal thermal stress changes that are
initial temperature. The response for pile cooling when the surface was generated and pile deformations. While full-scale testing confirms
set to 12 °C essentially parallels the results presented for pile heating this, a number of results, especially those from small scale tests are
with surface temperatures of 18 °C and 24 °C (Fig. 9(b)). It is thought inconsistent with this view.
this is because in the former the surface and the pile were both cooled, • While small scale tests often have advantages over full scale tests in
and in the latter case they are both heated, i.e. the thermal impacts act terms of cost, time, and control of conditions, in the tests reported
in the same sense. to-date the mobilised shaft restraint appears to be rather low. This
In the cases when the surface temperature boundary condition is set reflects that most of the tests were undertaken in sand and future
to 18 °C or 24 °C (i.e. the surface is heated) but the pile is cooled, the studies should focus on testing in stiffer materials to ensure that a
effects of the application of the surface temperature boundary condition full range of behaviours are examined.
• Numerical studies have been able to broadly reproduce the types of
response observed in field tests. Where conditions in field studies
have been approximated, very similar behaviours in terms of in-
ternal stress changes and pile movement are predicted. However,
many numerical studies referenced here, have modelled conditions
that result in rather low pile restraint and as a result, predicted
limited changes in internal stress and larger pile movements, re-
lative to those seen in field testing.

The numerical results from the study reported herein fall into the
group of studies where the conditions modelled result in limited pile
restraint and mobilise rather small internal stress changes. However,
these results have been explored and the following insights may be
derived from them:

• It may be rather obvious but bounds to the thermal stress changes in


floating energy piles may be formulated making some simple as-
sumptions regarding the pile-soil load-transfer interactions, and
these demonstrate that the internal stress changes are limited as a
function of the pile length, temperature change, thermal properties
(and therefore imposed deformations on the pile-soil interface), and
Fig. 12. Values for the normalised thermal response parameters, σth,max and the interface resistance-displacement response.
yth,0 for a 30 m long pile undergoing 15 °C cooling.
• The values derived from this simple assessment are compatible with

70
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

the maximum values predicted by the numerical analyses presented, where simplified analytical design approaches are proposed but are
and the method can be readily applied to other conditions. based exclusively on numerical validation, e.g. [50].
• These results highlight how the thermally-induced deformation of • When assessing the impact of thermal-activation of pile foundations,
the pile head should be compatible with the internal stress changes, only considering the pile thermal loading may not be conservative.
and how the soil thermal properties (primarily CTE in this case) and The effect of heat flow from the building (which occurs in any case)
imposed surface temperature boundary conditions modify this in- must also be taken into account.
teraction. • The use of σth,fixed as a safe bound for estimating the internal
• Ensuring that there is compatibility in terms of movement and in- thermal stress changes seems reasonable with most observed values
ternal stress changes can serve as a check for observed or calculated lying between 40% and 80% of this value.
responses in this type of problem. • The use of yth,free as a bound on pile head movements, is not so clear
• If the foundations of a building are thermally-activated, the thermo- cut. Some situations have been identified as possibly leading to
mechanical response seen in the foundations will be due to the su- movements exceeding this value, i.e. during cooling in conjunction
perposition of the heat flow from the structure and the thermal with the shaft resistance having been fully mobilised, and when the
loading within the foundation elements. The former can result in soil has a larger CTE than the pile and the surface temperature
thermal effects similar in magnitude to those generated by thermal changes significantly. This is also likely to be an issue in pile groups
operation of the energy foundation alone, and are not currently [50].
considered in design. • Tension may also occur in some instances, i.e. physical testing has
• The effect of transient heat flow is not considered in this set of demonstrated this effect when the pile undergoes strong cooling, as
analyses and is an important consideration in understanding when have numerical modelling – when the pile ends are restrained, and
the most critical design situations arise. There is some evidence (e.g. in situations where the ratio of soil to pile CTE is high and the pile is
see [40]) that maximum stress changes will occur at the start of a heated. To provide better design guidance, future studies should
thermal cycle and that the change in pile head movement will in- address the situations where tensile stresses may arise and their
crease throughout the thermal cycle, as the heating/cooling front likely magnitude.
penetrates the soil mass.
This article has only considered isolated thermally-activated piles
This review of physical testing and analytical case studies has undergoing monotonic thermal loading so as to clearly establish a
highlighted a number of aspects which require further investigation consistent set of pile-soil interaction behaviours for a relatively simple
and/or which should be considered in the design of thermally-activated set of conditions. The impact of thermal load cycles and pile group
pile foundations, including: effects have not been included here due to the additional complexity of
the interactions but are to be addressed in a separate review [54].
• Due to the complementary interactions between internal force
changes and movement, physical testing and numerical modelling
studies need to report their results in as complete and consistent Acknowledgements
manner as possible. In physical testing, material properties should
be fully defined, especially (but not exclusively) the thermal char- This work was undertaken as part of the project DEEPCOOL (PTDC/
acteristics. All thermal boundary conditions should be measured in ECI-EGC/29083/2017) financed by the Fundação para a Ciência e a
physical tests and detailed in all studies. Tecnologia (FCT), Portugal. The authors would also like to acknowl-
• Numerical studies should be utilised to examine a full range of po- edge the interactions with other researchers from across Europe en-
tential soil types, and hence restraint, in order to explore the full abled by COST Action TU1405 - European network for shallow geo-
range of possible response of thermally-activated piles, especially thermal energy applications in buildings and infrastructures.

Appendix A. Thermal axial stress limits for floating pile

In terms of thermally induced stress changes in a column, an upper bound value may be readily defined based on the assumption that the column
(or pile) is perfectly restrained and thus, the mobilized stress change will be:
th,fixed = c T·Ec (A1)
where αc is the CTE of the pile, ΔT is the imposed temperature change and Ec is the pile Young’s modulus.
Further limits to the possible stress change within a pile embedded in the ground and subject to thermal load may be defined based on the
available shaft resistance, qs,u, how this is mobilised and the pile dimensions (length, L & diameter, D).
For this study a simple soil model, consistent with the FEA presented in Section 5, was used whereby the soil and the pile-soil interface strength
were assumed to be cohesive (with a value of 75 kPa) and constant with depth. It is straightforward to apply the same evaluation in the case of non-
uniform (i.e. varying with depth) shaft resistance, the neutral point will move however. Assuming that the temperature change is applied uniformly
to the pile and thus, the axial deformations are uniform, and assuming that the expansion/contraction is not impeded at the ends of the pile (i.e. it is
a floating pile with negligible base resistance), it is straight forward to define these limits based on two assumptions regarding the mobilisation of the
shaft resistance, i.e.

(a) Assuming that the full shaft resistance is mobilised

In this case, the mobilised shaft resistance, qs,mob is equal to the ultimate shaft resistance, qs,u and the shear stress has opposing direction either
side of the neutral point (see grey shaded areas in Fig. A.1), which is at the mid-depth of the pile because the resistance is uniform. Based on this
distribution of qs,mob it is then possible to evaluate the restraining load that can be generated at the level of the neutral point and thus, the thermal
stress change limit imposed by the soil resistance (Table A1.):
L
Qs, max = D q
2 s, u (A2)

71
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

Qs, max
th, max =
Ap (A3)
where Ap is the pile cross-sectional area.
In this case, ignoring possible base restraint has a less than 10% effect on the ratio of σth,max/σth,fixed, as the base resistance represents only 5–15%
of the total resistance, depending on pile length.

(b) Assuming a simple load transfer scheme to model shaft resistance mobilisation

This model recognises that the axial deformations are unlikely to be sufficiently large to fully mobilize the shaft resistance along the pile, and
instead assumes that the thermal deformations are unimpeded (a major assumption) and that they are uniform about the neutral point, i.e. when
heated, the pile head expands upwards the same amount as the pile toe moves downwards. The unimpeded thermal movements associated with a
column/pile of length, L are:
yth,free = c T·L (A4)
And thus, the assumed pile head movement is ½yth,free. It is further assumed that this reduces linearly to zero at the neutral point (pile mid-depth,
in this case). It is then a simple task to evaluate the mobilised shaft resistance whereby, at the pile head:
If ½yth,free is less than yu, then
½yth,free
q s,mob = q s,u
yu (A5)
where yu is the displacement on the pile-soil interface required to fully mobilise qs,u, see Fig. A.1.
L qs, mob
Qs, max = D
2 2 (A6)
and, if ½yth,free is greater than yu, then
yu L
Lm = ·
½yth,free 2 (A7)

Fig. A1. Schematic showing distribution of shaft adhesion for (a) assumption that qs,u is fully mobilised (grey shading) and (b) using a simple load-transfer model as
indicated (diagonal hatching), and simple pile-soil interface load-transfer model used in b).

72
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

L
Lu = Lm
2 (A8)
q s,u
Qs,max = D L u·q s,u + L m·
2 (A9)
Here, allowing that the base resistance would only be mobilised at displacements in excess of 20% of the pile diameter (typical for bored piles),
then at a displacement of yth,free/2 only a very small proportion of the resistance would be mobilised and the ratio σth,max/σth,fixed, would only change
by a few-tenths of a percent.
Table A1 summarises the calculations applicable to thus study. It can be seen that with yu = 5 mm (0.5%D) the shaft resistance was full mobilised
only in the case of the 45 m long pile and then only over part of the pile shaft based on the thermal expansion associated with a 30 °C temperature
change.

Table A1
Limits of axial stress change as function of pile length and mobilized shaft adhesion.
Pile details
L (m) = 15 30 45 30
D (m) = 1.0 1.0 1.0 1.0
Ap (m2) = 0.79 0.79 0.79 0.79
Ec (GPa) = 30 30 30 30
αc (με/K) = 10 10 10 10

Maximum thermal effects


ΔT (K) = 30 30 30 30
αcΔT (με) = 300 300 300 300
yth,free (mm) = 4.5 9.0 13.5 9.0
σth,fixed (kPa) = 9000 9000 9000 9000

(a) Axial stress change with fully mobilized shaft adhesion


qs,u (kPa) = 75 75 75 150
Qsu (kN) = πDLcu
= 3534 7069 10,603 14,137
Qsu/2 (kN) = 1767 3534 5301 7069
σth,max (kPa) = 2250 4500 6750 9000
th,max
= 25% 50% 75% 100%
th,fixed

(b) Axial stress change with simple load-transfer model


yu (mm) = 5.0 5.0 5.0 5.0
yth,free/2 (mm) = 2.3 4.5 6.8 4.5
yth,free/2yu = 0.45 0.90 1.35 0.9
Lm (m) = 16.67 16.67 16.67 16.67
Lu (m) = −9.17 −1.67 5.83 −1.67
qs,mob (kPa) = 33.8 67.5 75.0 135.0
Qs,mob (kN) = 397.6 1590.4 3337.9 3180.9
σth,max (kPa) = 506.3 2025.0 4250.0 4050.0
th,max
= 5.6% 23% 47% 45%
th,fixed

References 10.1061/(ASCE)GT.1943-5606.0001061.
[12] Goode III JC, McCartney JS. Centrifuge modeling of boundary restraint effects in
energy foundations. ASCE J Geotech Geoenviron Eng 2015;141(8). https://doi.org/
[1] Brandl H. Energy foundations and other thermo-active ground structures. 10.1061/(ASCE)GT.1943-5606.0001333.
Géotechnique 2006;56(2):81–122. [13] Ng CWW, Shi C, Gunawan A, Laloui L, Liu H-L. Centrifuge modelling of heating
[2] Bourne-Webb P. Observed response of energy geostructures. In: Laloui L, Di Donna effects on energy pile performance in saturated sand. Can Geotech J
A, editors. Energy geostructures. Hoboken (NJ, USA): John Wiley & Sons; 2013. p. 2015;52(8):1045–57.
45–77. [14] Liu H-L, Wang C-L, Kong G-Q, Ng CWW, Che P. Model tests on thermo-mechanical
[3] Fellenius BH. Results from long-term measurement in piles of drag load and behavior of an improved energy pile. Eur J Environ Civ Eng 2016. https://doi.org/
downdrag. Can Geotech J 2006;43(4):409–30. 10.1080/19648189.2016.1248792. 16 pages.
[4] Poulos HG. A practical design approach for piles with negative friction, Proc. of the [15] Wang C-L, Liu H-L, Kong G-Q, Ng CWW, Wu D. Model tests of energy piles with and
Institution of Civil Engineers. Geotech Eng 2008;161(GE1):19–27. without a vertical load. Environ Geotech 2016;3(4):203–13.
[5] Amatya BL, Soga K, Bourne-Webb PJ, Amis T, Laloui L. Thermo-mechanical beha- [16] Huang X, Wu J, Peng H, Hao Y, Lu C. Thermomechanical behavior of energy pile
viour of energy piles. Géotechnique 2012;62(6):503–19. embedded in sandy soil. Math Probl Eng 2018. https://doi.org/10.1155/2018/
[6] Murphy KD, McCartney JS, Henry KS. Evaluation of thermo-mechanical and 5341642. Article ID 5341642, 11 pages.
thermal behavior of full-scale energy foundations. Acta Geotech [17] Murphy KD, McCartney JS. Seasonal response of energy foundations during
2015;10(2):179–95. building operation. Geotech Geol Eng 2015;33(2):343–56.
[7] Bourne-Webb PJ, Amatya B, Soga K. A framework for understanding energy pile [18] Santiago C, Pardo de Santayana F, de Groot M, et al. Thermo-mechanical behavior
behaviour. ICE Proc Geotech Eng 2013;166(GE2):170–7. of a thermo-active precast pile. Bull Chem Commun 2016;48(Special Issue
[8] Kalantidou A, Tang AM, Pereira J-M, Hassen G. Preliminary study on the me- E):41–54.
chanical behaviour of heat exchanger pile in physical model. Géotechnique [19] You S, Cheng X, Guo H, Yao Z. Experimental study on structural response of CFG
2012;62(11):1047–51. energy piles. Appl Therm Eng 2016;96(March):640–51.
[9] Yavari N, Tang A-M, Pereira J-M, Hassen G. Experimental study on the mechanical [20] Bourne-Webb PJ, Amatya B, Soga K, Amis A, Davidson C, Payne P. Energy pile test
behaviour of a heat exchanger pile using physical modelling. Acta Geotecnica at Lambeth College, London: geotechnical and thermo-dynamic aspects of pile re-
2014;9(3):385–98. sponse to heat cycles. Géotechnique 2009;59(3):237–48.
[10] Ng CWW, Shi C, Gunawan A, Laloui L. Centrifuge modelling of energy piles sub- [21] Laloui L, Moreni M, Vulliet L. Comportement d’un pieu bi-fonction, foundation et
jected to heating and cooling cycles in clay. Géotech Lett 2014;4(4):310–6. éschangeur de chaleur. Can Geotech J 2003;40(2):388–402.
[11] Stewart MA, McCartney JS. Centrifuge modeling of soil-structure interaction in [22] Sutman M, Olgun G, Laloui L, Brettmann T. Effect of end-restraint conditions on
energy foundations. ASCE J Geotech Geoenviron Eng 2014;140(4). https://doi.org/ energy pile behavior. Geotechnical frontiers 2017: geotechnical materials

73
P.J. Bourne-Webb et al. Computers and Geotechnics 108 (2019) 61–74

modeling, and testing, ASCE GSP 280. 2017. p. 165–74. [39] Khosravi A, Moradshahi A, McCartney JS, Kabiri M. Numerical analysis of energy
[23] Allani M, Van Lysebetten G, Huybrechts N. Experimental and numerical study of the piles under different boundary conditions and thermal loading cycles. Proc. 3rd
thermo-mechanical behaviour of energy piles for Belgian practice. In: Ferrari A, European conference on unsaturated soils, “E-UNSAT 2016”. 2016. 6 pages https://
Laloui L, editors. Advances in laboratory testing and modelling of soils and shales doi.org/10.1051/e3sconf/20160905005.
(ATMSS) 2017. p. 405–12. https://doi.org/10.1007/978-3-319-52773-4_48. [40] Gawecka KA, Taborda DMG, Potts DM, Cui W, Zdravkovic L, Haji Kasri MS.
[24] BS EN 1991-1-5:2003. Eurocode 1: Actions on structures - Part 1-5: General actions Numerical modelling of thermo-active piles in London Clay. Proc Institution Civ
– thermal actions. BSi; 2010. 52 pages. Eng Geotech Eng 2017;170(GE3):201–19.
[25] McCartney JS, Murphy KD. Investigation of potential dragdown/uplift effects on [41] Di Donna A, Dupray F, Laloui L. Effect of thermo-plasticity of soils on the design of
energy piles. Geomech Energy Environ 2017;10(June):21–8. energy piles, EGC2013. Proc European geothermal congress, Pisa, Italy, Paper SG4-
[26] Rotta Loria AF, Laloui L. Thermally induced group effects among energy piles. 03. 2013. 10 pages.
Géotechnique 2017;67(5):374–93. [42] Salciarini D, Ronchi F, Cattoni E, Tamagnini C. Thermomechanical effects induced
[27] Garnier J, Gaudin C, Springman SM, et al. Catalogue of scaling laws and similitude by energy piles operation in a small piled raft. Int J Geomech 2015;15(2):1–14.
questions in geotechnical centrifuge modelling. Int J Phys Modell Geotech [43] Dupray F, Laloui L, Kazangba A. Numerical analysis of seasonal heat storage in an
2007;7(3):1–23. energy pile foundation. Comput Geotech 2014;55(January):67–77.
[28] Bezuijen A. Similitude in soil-structure and soil-soil interaction. Proc 3rd European [44] Jeong S, Lim H, Lee JK, Kim J. Thermally induced mechanical response of energy
conference on physical modelling in geotechnics, Nantes, France, vol. 1. 2016. p. piles in axially loaded pile groups. Appl Therm Eng 2014;71(October):608–15.
43–8. [45] Di Donna A, Laloui L. Numerical analysis of the geotechnical behaviour of energy
[29] Chen D, McCartney JS. Parameters for load transfer analysis of energy piles in piles. Int J Numer Anal Meth Geomech 2015;2015(39):861–88.
uniform nonplastic soils. ASCE Int J Geomech 2016;17(7). https://doi.org/10. [46] Rotta Loira AF, Gunawan A, Shi C, Laloui L, Ng CWW. Numerical modelling of
1061/(ASCE)GM.1943-5622.0000873. energy piles in saturated sand subjected to thermo-mechanical loads. Geomech
[30] Wang W, Regueiro RA, Stewart M, McCartney JS. Coupled thermo-poro-mechanical Energy Environ 2015;1(April):1–15.
finite element analysis of an energy foundation centrifuge experiment in saturated [47] Tsetoulidis C, Naskos A, Georgiadis K. Numerical investigation of the mechanical
silt. Proc geo-congress: state of the art and practice in geotechnical engineering, behaviour of single energy piles and energy pile groups. In: Wuttke F, Bauer S,
ASCE GSP 225. 2012. p. 4406–15. Sánchez M, editors. Energy Geotechnics. CRC Press; 2016. p. 569–75.
[31] Suryatriyastuti ME, Mroueh H, Burlon S. Understanding the temperature-induced [48] Suryatriyastuti ME, Burlon S, Mroueh H. On the understanding of cyclic interaction
mechanical behaviour of energy pile foundations. Renew Sustain Energy Rev mechanisms in an energy pile group. Int J Numer Anal Meth Geomech
2012;16(5):3344–54. 2016;40(1):3–24.
[32] Bodas Freitas TM, Cruz Silva F, Bourne-Webb PJ. The response of energy founda- [49] Salciarini D, Ronchi F, Tamagnini C. Thermo-hydro-mechanical response of a large
tions under thermo-mechanical loading. Proc of the 18th intl conf on soil mech and piled raft equipped with energy piles: a parametric study. Acta Geotech
geot Engg, Paris. 2013. p. 3347–50. 2017;12(4):703–28.
[33] Ma X, Qiu G, Grabe J. Numerical simulation of an energy pile using thermo-hydro- [50] Rotta Loria AF, Laloui L. The interaction factor method for energy pile groups.
mechanical coupling and a visco-hypoplastic model. Geotech Eng J SEAGS AGSSEA Comput Geotech 2016;80(December):121–37.
2014;45(2):12–6. [51] Alberdi-Pagola M, Madsen S, Lund Jensen R, Erbs Poulsen S. Numerical in-
[34] Suryatriyastuti ME, Mroueh H, Burlon S. A load transfer approach for studying the vestigation on the thermo-mechanical behavior of a quadratic cross section pile
cyclic behaviour of thermo-active piles. Comput Geotech heat exchanger. IGSHPA technical/research conference and expo, Denver. 2017. p.
2014;55(January):378–91. 134–43 [accessed 15 January 2018].
[35] Ozudogru TY, Olgun CG, Senol A. 3D numerical modeling of vertical geothermal [52] Assunção RM. Thermal and thermal-mechanical analysis of thermo-active pile
heat exchangers. Geothermics 2014;51(July):312–24. foundations. Instituto Superior Técnico, University of Lisbon; 2014. [MSc thesis,
[36] Saggu R, Chakraborty T. Thermal analysis of energy piles in sand. Geomech Geoeng 100 pages].
2014;10(1):10–29. [53] Bourne-Webb PJ. An overview of observed thermal and thermo-mechanical re-
[37] Wang W, Regueiro RA, McCartney JS. Coupled axisymmetric thermo-poro-me- sponse of piled energy foundations. European geothermal congress, Pisa, Italy.
chanical finite element analysis of energy foundation centrifuge experiments in 2013. 8 pages.
partially saturated silt. Geotech. Geol. Eng. 2015;33(2):373–88. [54] P.J. Bourne-Webb, T.M. Bodas Freitas, Thermally-activated piles and pile groups
[38] Bourne-Webb PJ, Bodas Freitas TM, Assução RM. Soil-pile thermal interactions in under monotonic and cyclicthermal loading - a review, Renew. Energy, doi:10.
energy foundations. Geotechnique 2016;66(2):167–71. 1016/j.renene.2018.11.025.

74

Вам также может понравиться