Thermo-mechanics of energy piles: fine-grained soils, cycles, and interfaces

In the serviceability lifespan of thermo-active geo-structures such as energy-piles, soils surrounding these structures are exposed to a combination of mechanical and thermal loads. These loads are often complex (including cycles) and, depending on the state of the soils, the response of the surrounding soil to these loads may differ. Since the performance and safety of the soil-structure system directly depends of the response of the surrounding soil, it is important to understand and quantify the thermomechanical behaviour of soils. These objectives can be achieved by performing laboratory-scale element tests to gain knowledge on the fundamental response of the material and by developing numerical tools which can be used to simulate the complete soil--structure system under various complex load paths.


To date, many laboratory test have been conducted to study the thermomechanical behaviour of soils. A large portion of these tests have been triaxial tests and many thermomechanical constitutive models for soils are developed based on the phenomenological findings from these tests. While these models have been seen to be capable of capturing the general thermomechanical behaviour of soils, none have been formulated to ensure that they unconditionally satisfy the principles of thermodynamics. Therefore, under complex loading paths certain phenomena may not be captured/predicted, and other phenomena may be spuriously predicted. On the other hand, only a very limited number of tests have been conducted on soil-structure interfaces. Therefore the available knowledge on the thermomechanical behaviour of soil-structure interfaces until this time has been limited.


The objective of this thesis is to fill-in the gaps mentioned above by investigating and exploring the main mechanisms governing the thermomechancial behaviour of soils and soil-structure interfaces, as well as developing thermomechanical constitutive models constructed from a sound foundation (i.e. thermodynamics) and numerical algorithms that can be used in boundary-value solvers such as finite-element methods.


First, the phenomenological temperature effects observed in laboratory-scale tests are combined with principles of thermodynamics to develop a "{\it base}" thermomechanical constitutive model, defined in triaxial stress space, that can capture the main thermomechanical behaviour of fine-grained soils. This base model has a single flexibly shaped yield surface. The base model is then upgraded to a "{\it two surface/bubble}" thermomechanical model by introducing an additional yield surface. The additional yield surface translates within the admisible stress space via a temperature-dependent kinematic rule, which enables the model to capture additional thermomechanical features such as the shakedown behaviour of soils when subjected to thermal cycles, which the single yield surface constitutive model was not able to capture or predict.


The main value of constitutive models is achieved when they are efficiently embed within boundary-value solvers, such as a finite-element method solver. One such efficient method is to use the implicit stress integration scheme. However, many constitutive models fail to converge within these schemes. One possible reason, as demonstrated in this thesis, is the existence of undesired elastic nuclei or domains with erratic divergence. A new yield function (which can also be used as a plastic potential function) is proposed, which is flexible and unique, and overcomes the aforementioned drawbacks. The single surface thermomechanical model (defined in triaxial space) is then modified by incorporating the newly proposed yield surface formulation with the addition of Lode angle dependency and generalisation to three-dimensional stress space, prior to being implemented in a finite-element context. Since the non-linear thermo-elastic relationships of the model were derived from a Gibbs-type energy potential, a new numerical algorithm was designed to accommodate this feature when implementing the model in a finite-element context using an implicit stress integration scheme.


The thermomechanical behaviour of soil-structure interfaces is experimentally investigated using a temperature-controlled direct shear apparatus. Several thermomechanical stress paths, with a wide ranges of stresses, temperatures and boundary conditions, analogous to those an interface element experiences in the serviceability life-time of an energy-pile, were designed and performed. Unique observations including the coupling effect of initial shear stress and thermal cycles were recorded, which enhanced the knowledge of thermoemchanical behaviour of soil-structure interfaces. The main impact on soil-concrete interfaces was seen to be the mechanical cyclic loads arising due to the heating and cooling of the concrete pile, rather than direct thermal impacts. Thermal creep was identified as a novel phenomen which had not been previously identified.