You are on page 1of 10

Geo-Frontiers 2011 ASCE 2011

470

A method for the geotechnical design of heat exchanger piles H. Peron, C. Knellwolf and L. Laloui Ecole polytechnique fdrale de Lausanne, Laboratory of Soil Mechanics, CH-1015 Lausanne, Switzerland; PH (+41) 21 6932315; FAX (+41) 21 6934153; email: lyesse.laloui@epfl.ch ABSTRACT There is currently a lack of established calculation method for the geotechnical design of heat exchanger piles. Thermo-mechanical effects are ignored and large overall security factors are therefore applied. This paper presents a new geotechnical numerical design method for heat exchanger piles, based on the load transfer approach. The method is validated on the basis of two in situ measurements. It is shown how the pile design could be adapted and optimized with respect to concrete resistance and the mobilization of the pile shaft friction during the operation of the heat exchange system. INTRODUCTION This paper considers a new sustainable technology for the intermittent storage of energy in soils, namely heat exchanger piles. Heat exchanger piles take advantage of the ground as an energy storage system. The heat exchange system consists of absorbing and transporting ground thermal energy to buildings via a fluid circulating in pipes placed within the piles. In the case of a hollow pre-cast pile, the pipes are placed in the hollow part in contact with the inner wall of the concrete. In the case of cast-in-place piles, the pipes are fixed to the inner side of the metallic reinforcement. Actually, any kind of foundation can be used as a heat exchanger, such as retaining walls, slabs, anchors, etc. With this geothermal use of geostructures, buildings can be economically cooled and heated with a heat pump (Figure 1). The heat exchanger pile technology, although very successful in Europe faces a lack of rational knowledge of the thermal effects on the behavior of the foundations. In particular, no design method is yet available to consider the complex interactions between thermal storage and the mechanical behavior of these geostructures. Therefore, for years, the dimensioning of heat exchanger piles has been based on empirical considerations (Bonnec 2009). In order to err on the safe side, the safety factors usually employed for typical piles are considerably increased. This may lead to considerable extra costs and non-standard construction skills. In situ experience shows that applying a thermal load induces a significant change in the static behavior of a foundation pile. In this paper, a new geotechnical design method for heat exchanger piles is described and validated on the basis of in situ data. Finally, representative cases for which thermal loads could lead to failure are discussed.

Geo-Frontiers 2011 ASCE 2011

471

Building

Heat pump
Layer 1 Layer 2

Heat exchanger pile


Soil

Figure 1. Representation of a heat exchanger pile system (Laloui et al., 2006) A NEW GEOTECHNICAL DESIGN METHOD FOR HEAT EXCHANGER PILES Origin of thermal stresses in heat exchanger piles In service conditions, the pile is heated or cooled due to the circulation of the heat exchanger fluid within the pipes cast in the concrete. The temperature generally lies between +4 C (to avoid freezing of the pile and of the ground) and +30 C. Please Note that in the UK energy piles are designed to temperatures of -1 C to +35 C. However, in some situations (bad operation or even external thermal recharge from solar panels), increase in the temperature up to +40 or +50 C is conceivable (SIA 2005, Silvani et al. 2008). Basically, the heating of a pile induces expansion, while the cooling induces contraction. If the pile is unrestrained, the change in temperature T induces a uniform free strain th,f = .T, where is the coefficient of thermal expansion of the pile. In the general case, a part of the strain th,f is blocked by the surrounding soil and the structure, so that only the observed strain th,o is finally produced. The fact that the soil and the structure restrain the pile in its movements introduces some additional stresses in the pile. A convenient way to cope with the process and to assess the additional stresses in the pile is to use the degree of freedom of the pile (denoted n). The degree of freedom of the pile is defined by the ratio between the free and observed axial strains th,f and th,o (Bochon 1992, Laloui et al. 2003). The degree of freedom is theoretically 0 when the pile is completely blocked and 1 when the pile is completely free to move. In the general case, n ranges from 0 to 1 due to the variable shaft friction mobilization and restraint at the two extremities of the pile. The observed strain then reads th , o = n T The blocked strain th,d is the difference between free and observed strain. Assuming a linear elastic behavior of the pile, the additional stress th due to thermal loading is proportional to th,d. Presently, in situ strain and stress measurements along pile while heated or cooled down are very scarce. Complete set of data are available in Laloui et al. (2003) and Bourne-Webb et al. (2009). Basic assumptions of the proposed method The method relies on the following basic assumptions: 1) The pile displacement calculation is done using a one dimensional finite difference scheme (only the axial displacements are considered). The radial displacements and

Geo-Frontiers 2011 ASCE 2011

472

their mechanical interactions with the soil are neglected (such interactions are considered small with regards to the effects of the axial displacements). 2) The pile behavior is considered linear and elastic. The properties of the pile, namely its diameter D, Young modulus Epile and coefficient of thermal expansion , remain constant along the pile and do not change with temperature. 3) The relationships between the shaft friction/shaft displacement, head stress/head displacement and base stress/base displacement are known. Upwards movements are taken as positive; downwards movements are negative. Compression stresses are taken as negative. Pile displacement calculation The pile displacement for a mechanical load P is done by the load transfer method (Coyle and Reese 1966). In this method, the pile is subdivided into several rigid elements, which are connected by springs representing the pile stiffness. Each rigid element experiences along its side an elasto-plastic interaction with the surrounding soil. The pile base element is supported by the reaction of the substrate, the pile/soil interaction being elasto-plastic as well. The relation between the shaft friction and pile displacements, as well as the relation between the normal stresses and the pile displacements at the base, are described by load transfer functions. In the following, the load-transfer functions proposed by Frank and Zhao (1982) are used for the mobilized shaft friction and base reaction, respectively, versus pile displacements. The curves feature two linear parts and a plateau equal to the ultimate value. Ks and Kb are the slopes of the shaft and base load transfer functions for the first linear branches, respectively. Ks and Kb are related to the Menard pressuremeter modulus EM. In the present study, an unloading branch has been added, accounting for the irreversible behavior of the soil. The load transfer functions described above are chosen for convenience. Other forms are indeed possible. The load transfer curves describe the mobilized stress for a given displacement. The originality of the present approach is to consider the pile-supported structure interaction, represented by the spring constant Kh. Kh ultimately depends on many factors, such as the supported structure rigidity, the type of contact between the pile and the foundation raft, the position and the number of heat exchanger piles. On the basis of the soil-pile interaction laws defined above, the calculation of the thermomechanical response of the heat exchanger pile is made as follows. First, the stress state and the pile displacements induced by the imposed mechanical loading are calculated; this state is further referred to as the initialization state and corresponds to effects due to the weight of the building. Then, from the initialization state, the pile response due to the thermal loading (heating or cooling occurring during heat exchange) is calculated. Each element i of the pile has a length hi, diameter D and section A. Thermal loading When a pile is heated or cooled, it dilates or contracts about a null point (Bourne-Webb et al. 2009). The null point is situated at that depth NP where the sum

Geo-Frontiers 2011 ASCE 2011

473

of the mobilized friction along the upper part plus the reaction of the structure is equal to the sum of the mobilized friction along the lower part plus the reaction at the base. In order to assess the blocked strain, an iterative procedure is applied following the method described below. 1) Choice of a starting value for the observed deformation: to compute a first set of mobilized resistance (mobilized shaft friction and resistance at the extremities), the pile is initially assumed to be totally free to move. The first displacement calculations are therefore done with th = th,f = T. 2) Displacement calculation: the thermal displacement calculation is done from a non zero initialized displacement and strain state induced by the mechanical loading. By definition, there is no displacement at the null point. Using the t-z curve, a first set of mobilized reaction stresses is obtained. The axial stress in the pile th,i induced by the thermal free displacement of the pile is the sum of all the external forces divided by the pile section A. In the case of unloading (uplift), the stress path follows the unloading branch. 3) Calculation of the blocked thermal strain th,b (from the mobilized stress). 4) Calculation of the observed strain (the blocked strain minus the free one). Steps 2 to 4 must be repeated with the new set of observed strains th = th,o. The observed strain will converge to the actual values of the blocked and observed strain. Related parameters, such as pile displacement, internal axial stresses, mobilized shaft friction and mobilized reaction at the base and head of the pile, are then deduced. More details can be found in Knellwolf et al. (2010). Numerical implementation The above numerical method has been coded in the Java language and validated against an analytical calculation of the pile deformations for a mechanical loading. Several soil layers can be considered, each of them with different parameters. For each layer, specific soil proprieties can be defined. The bearing capacity can either be calculated by the code from analytical expressions or set directly by the user. In order to set the load transfer function proposed by Frank and Zhao (1982), one can enter the Menard pressuremeter modulus, the ultimate shear stress and bearing capacity at the base. The interaction between the pile and the supported structure is modeled by an elastic law, the stiffness of which is directly defined by the user. Pile geometry as well as material parameters (Young modulus and thermal expansion coefficient) are set to be constant with depth. The weight of the building (i.e., the mechanical load) and the change in temperature with respect to pile depth (i.e., the thermal load) are both defined by the user. The verification of the static behavior of the pile is further done by comparing the total axial stress to the resistance of the concrete pile on the one hand, and the total mobilized bearing forces to the ultimate bearing capacity on the other hand. VALIDATION OF THE METHOD In order to validate the above method, experimental data on the stresses and strains experienced by a heat exchanger pile are required. The validation is

Geo-Frontiers 2011 ASCE 2011

474

undertaken using the results of two comprehensive full scale in situ tests: one carried out at the EPFL in Lausanne, Switzerland (Laloui et al. 2003, 2006) and another one undertaken at Lambeth College in London, UK (Bourne-Webb et al 2009). In-situ energy pile at EPFL (Switzerland) A 25 m long heat exchanger pile that was part of a pile raft supporting a four storey building was equipped with load cells, extensometers and temperature sensors (Laloui et al. 2003, 2006). This pile was subjected to a thermal load, generated by a heat carrying fluid circulating in polyethylene pipes embedded in the concrete pile. The Young modulus of the pile was estimated from laboratory tests and cross-hole ultrasonic transmission tests, yielding the value Epile = 29.2 GPa. The pile crosses four layers of sandy and silty gravels (layers A1, A2, B and C) while the bottom of the pile rests on a stiff layer (molasse, layer D). The soil geotechnical parameters were obtained on the basis of geotechnical investigations and two static pile loading tests. The groundwater table was found to be very close to the ground surface. The behavior of the pile was measured for seven successive construction stages. Test 1 was done before the construction of the building. The strains were therefore only due to the thermal load. In Test 7, the whole building was built and was acting on the pile. Varying changes in temperature were applied (up to 21.8 C in Test 1 and 14.3 C for Test 7). The complete set of soil parameters used for the validation of the modeling approach is listed in Table 1. Table 1. Soil parameters used for modeling the EPFL pile. Soil Layer A1 A2 B C D Ks [MPa/m] 16.7 10.8 18.2 121.4 qs [kPa] 102 70 74 160 Kb [MPa/m] 6681335 qb [kPa] 11000 The pile section is considered to be constant. Both experimental and modeled pile axial strains for Test 1 are shown in Figure 2, for one average temperature increments. As mentioned above, Test 1 was done before the construction of the building; the mechanical load and pile head-structure contact stiffness Kh are therefore set to zero. For Test 7, the experimental and numerical results are shown in the form of the degree of freedom. The soil parameters are the same as in Test 1. Because the building is completely constructed (representing a mechanical load of P = 1000 kN), a stiffness Kh = 1.5 GPa/m is further imposed at the contact pile headstructure. Kh, as the only parameter for which no information is available, is chosen in order to match the measured degree of freedom. The excellent fitting of the method results with the experimental data (see Figure 2) shows that the new approach is able to reproduce the observed behavior, either in the case of thermal loading alone or in the case of both thermal and mechanical loads.

Geo-Frontiers 2011 ASCE 2011

475

Figure 2. Modeled strains versus measured ones for successive changes in temperature, Test 1: left, T =17.4 C, right, modeled degree of freedom of the pile versus measured one, Test 7, T = 14.3C (experimental data from Laloui et al, 2003). Kb(X) stands for Kb of Layer X In-situ energy pile at Lambeth College (UK) A full scale test was undertaken on a pile located in a construction site (Lambeth College) in London by Bourne-Webb et al. (2009). Most of the pile is installed in the London Clay formation, which extends well below the toe level of the pile. The mechanical load was applied on the pile head with a loading frame. The strains were measured with vibrating-wire strain gauges (VWSG) and with fiber-optic sensors (OFS). The test stages of interest were as follows: an initial mechanical loading stage (two loading-unloading cycles at 1200 and 1800 kN, respectively), a cooling stage (with a 1200 kN mechanical load and T = -19 C) and a heating stage (maintaining the 1200 kN mechanical load, while T = +10 C). The test data showed significant strain variations in the upper part of the pile, partly due to bending effects, suggesting that little resistance was mobilized in this zone (Bourne-Webb et al 2009). Hence, the shear resistance of the upper 6.5 m is not considered. The Young modulus Epile is equal to 40 GPa. For the numerical validation exercise made in the following, typical London Clay geotechnical parameters are used (Marsland and Randolph 1977). In compression, the rigidity on the pile head Kh is taken equal to 10 GPa/m (based on estimates of the beam profile of the loading frame and span values). In tension Kh is set to 0.1 GPa/m: this case is experienced during cooling, and thus the beam rigidity should not interfere. The complete set of soil parameters are listed in Table 2 (four layers are distinguished). Table 2 - Soil parameters used for modeling the Lambeth College pile Layer 1 2 3 4 Depth [m] 0-6.5 6.5-10.5 10.5-16.5 16.5-22.5 EM [MPa] 0 45 45 45 qs [kPa] 0 60 70 80 qb [kPa] 460

Geo-Frontiers 2011 ASCE 2011

476

The comparison of the measured and modeled strain profiles of the Lambeth College pile demonstrates that the method is able to quantitatively reproduce the effects of mechanical and thermal loadings. The occurrence of tensile axial strain (and axial stress) during the cooling phase in the bottom part of the pile is well predicted (see Figure 3). In particular, the decrease of shaft friction in the bottom part of the pile and increase in the upper part are accurately reproduced. In addition (not represented here), the noticeable additional compressive axial strain and stress increase observed during the heating phase within the whole pile is well assessed. In the case of heating, the increase in shaft friction mobilization is well reproduced below six meters in depth.

Figure 3. Modeled strains versus measured ones for thermal test at the end of cooling (left), and modeled and measured profiles of mobilized shear stress during pile cooling (experimental data from Bourne-Webb et al, 2009) STUDY OF REPRESENTATIVE CASES In this section, critical situations are examined using the above model, for which the temperature changes in the pile lead to structural failure of the pile element or to serviceability limit state failure as well as ultimate bearing resistance failure. A pile with a given geometry is considered, which is 10 m in length and 0.5 m in diameter. The coefficient of thermal expansion of the pile is = 1x10-5 C-1 and the Youngs modulus is Epile = 30 GPa. The soil is homogenous (1 layer). The other model parameters are adapted for each case. Case 1: Floating pile A floating pile is such that almost the entire weight of the building is transferred to the ground through friction along the pile shaft; no or little weight is supported by the base of the pile. If the initial mechanical load (building weight) is chosen such that the mobilized friction is already near the ultimate value, the model shows that the heating is likely to increase the mobilized shear stress, initiated by the weight of the building, up to the bearing capacity. In the same time, the additional compression in the pile depends on the friction resistance qs. If qs is small, the

Geo-Frontiers 2011 ASCE 2011

477

expansion of the pile is slightly constrains, stresses remain well below the structural resistance. The pile response would actually be different (with more significant compressive stresses but less chance to reach soil bearing capacity) in the case when a soil with a larger friction resistance is considered. When the pile is cooled, the results show negative shear stress can be experienced, depending on the soil parameters. This situation is favorable in terms of bearing capacity, but it may generate tension stresses in the pile. The pile therefore needs to be designed to resist tension. The cooling reduces the mobilized bearing forces, down to values less than the building weight. The difference is transmitted via the raft (if any) to adjacent piles. In the case of increased cooling, the tension could act up to the head of the pile; this would mean the pile pulls on the building. Case 2: Semi-floating pile A semi-floating pile is understood as a pile that supports the weight of the building both at its base and through friction along its lateral surface. This situation is encountered in most practical cases. In the present case, conditions such that the structural resistance in compression is reached are examined. For this purpose, a stiff soil is considered, with a relatively high strength (qs = 250 kPa and qb = 38.2 MPa); moreover, both large mechanical and thermal loads are applied. In particular, the applied temperature variation (T = 50C) is beyond the usual functioning range. The results of this simulation are shown in Figure 4. A noticeable increase in mobilized shear stress is observed after heating. More importantly, the displacement of the pile being significantly restrained, an additional compression develops within the pile. In Figure 4 (left) the axial compression exceeds a typical pile resistance of fcd = 20 MPa. Due to the high end-bearing resistance, the ultimate bearing capacity is not problematic. In the same situation, and for the same soil characteristics, hollow precast piles would be much more likely to experience structural failure than cast-inplace concrete piles because the same axial load would be applied on a smaller section. In the case of cooling, the pile head-structure interaction and the rigid soilpile shaft interaction completely constrain the displacements in the upper 5 m (n = 0). The behavior is in this case close to the one of a floating pile.
0 2
Building Weight

0 2

Building Weight

Depth [m]

Depth [m]

4 6 8 10 25
After Heating Resistance

4 6 8 Semi-floating pile 10
After Heating Bearing Capacity

Semi-floating pile 20 15 10 5

5000

10000

15000

Axial stress [MPa]

Forces [kN]

Figure 4. Changes in axial stresses (left) and forces (right) within the pile for a semi-floating heat exchanger pile (T=50 C)

Geo-Frontiers 2011 ASCE 2011

478

Below the depth of 5 m, when the pile starts to move, Figure 5 shows that the unloading of shear stresses is observed first, and the direction of shear stresses is reversed in the lowest part of the pile as well. The development of negative shear stresses is supplemented with tensile axial stresses. Here, the same phenomena of the reduced bearing forces, already discussed for the floating pile example, are observed. Case 3: End-bearing pile In this case, the load transfer is done via axial stress down to the base. The degree of freedom and thus the strain and axial stress due to the thermal loading are constant in depth and only depend on the stiffness of the bedrock and the upper structure. This third example yields the following comment: for the design of a conventional pile resting on a hard substrate at its base, neglecting the shaft friction is conservative. However, if a heat exchanger pile is designed, doing so can be problematic. The axial stresses due to thermal loading directly depend on the friction resistance. There is the risk that pile structural failure will be reached due to the contribution of ignored shear stresses (in particular tensile stresses).
0 2
After Cooling

0 2
Building Weigth

Depth [m]

4 6 8 Semi-floating pile 10 20 15 10
Building Weight

Depth [m]

4
Bearing Capacity

6 8 10
After Cooling

Semi-floating pile

Axial stress [MPa]

5000

10000

15000

Forces [kN]

Figure 5. Changes in a) axial stresses and b) forces within the pile for a semifloating heat exchanger (T=-50 C) CONCLUSION In spite of the existence of hundreds of heat exchanger pile installations, no design method is available to consider the complex interactions between thermal storage and the mechanical behavior of geostructures. This paper presents a new geotechnical design method, which assesses the main effects of thermal loading on heat exchanger pile stress and strain response. The proposed method is based on the load transfer method and considers the shear resistance of the surrounding soil and the tip resistance of the soil at the bottom of the pile. The interaction between the pile and the supported structure, decisive in the case of thermal loading, is also taken into account. A simplified scheme can be drawn: the heating of the pile induces additional compression in the pile and increases the mobilized shear stress. The cooling can induce a release of mobilized shear stress, possibly leading to the reversal of shear stress sign and the development of tensile stress in the pile.

Geo-Frontiers 2011 ASCE 2011

479

There is interplay between the changes in friction mobilization on the one hand and the additional efforts within the pile on the other hand, caused by the changes in temperature and the prevailing soil-pile-supported structure interactions. This deserves a careful analysis in each case. The proposed method is believed to furnish adequate analyses for user-defined problems. ACKNOWLEDGMENTS This work was partly funded by swisselectric research. REFERENCES Bonnec, O. 2009. Piling on the Energy, Geodrilling International, March 2009: 25-28. Bochon, A. 1992. Les mesures de dformation des structures hyperstatiques: le tmoin sonore. Revue franaise de gotechnique, 60: 41-50. Bourne-Webb, P.J, Amatya, K., Soga, K., Amis, T., Davidson C. and Payne, P. 2009. Energy pile test at Lambeth College, London: geotechnical and thermodynamic aspects of pile response to heat cycles. Gotechnique, 59: 237-248. Coyle, H. M. and Reese, L. C. 1966. Load transfer for axially loaded piles in clay. Journal of the Soil Mechanics and Foundations Division, ASCE, 92(SM2): 126. Frank, R. and Zhao, S.R. 1982. Estimation par les paramtres pressiomtriques de lenfoncement sous charge axiale de pieux fors dans des sols fins. Bulletin de Liaison des Laboratoires des Ponts et Chausses, 119: 17-24. Knellwolf, C., Peron H. and Laloui L. 2010. Geotechnical Analysis of Heat Exchanger Piles. ASCE Journal of Geotechnical and Geoenvironmental Engineering, submitted. Laloui, L., Moreni, M. and Vulliet, L. 2003. Comportement dun pieu bi-fonction, fondation et changeur de chaleur. Canadian Geotech. J., 40: 388-402. Laloui, L., Nuth, M., and Vulliet, L. 2006. Experimental and numerical investigations of the behaviour of a heat exchanger pile. Int. J. for Num. and Analyt. Methods in Geomechanics, 30: 763-781. Marsland, A. and Randolph, M.F. 1977. Comparisons of the results from pressuremeter tests and large in situ plate tests in London Clay. Gotechnique, 27: 217-243. SIA. 2005. Nutzung der Erdwrme mit Fundationspfhlen und anderen erdberhrten Betonbauteilen, Swiss Association of Engineers and Architects, SIA, Dokumentation D 0190. Silvani, C., Nuth, M., Laloui, L. and Peron, H. Understanding the thermomechanical response of heat exchanger piles. Proc. First Int. Symposium on Computational Geomechanics (COMGEO I), Juan-les-Pins, France, 589-596.

You might also like