This paper presents a formulation for coupled heat and moisture transfer in a deformable partially saturated soil. The research is based on a mechanistic phase interaction model coupled to a state surface approach. The method takes into account the coupling effect of temperature gradient and deformations on flows in porous media. Pore water pressure, pore air pressure, temperature and displacement are treated as the primary unknowns. A numerical solution of the formulation is then achieved via the finite element method. An example of the use of the new model is then presented.

A numerical model of the stress–strain behaviour of unsaturated soil is presented. An extended critical state elastoplastic constitutive model based on net stress and suction is solved via the finite-element method. Numerical techniques are proposed which allow for the development of an elastoviscoplastic solution algorithm within the context of a transient finite-element analysis. The simultaneous solution of the coupled partial differential equations provides variations of net stress, suction, specific volume and degree of saturation in an unsaturated soil. Several examples are then analysed which illustrate typical features of the behaviour of unsaturated soil samples. Expected patterns of behaviour are reproduced and good correlation is obtained between numerical results and an alternative solution of the theoretical model. Comparisons of numerical predictions with results obtained from two suction-controlled experiments on compacted kaolin and sandy clay are then presented. Good correlation is again achieved, yielding confidence in the validity of the numerical algorithm. Finally, the solution of a boundary value problem is attempted, namely the prediction of seasonal ground movement. Comparison of numerical results with field observed data shows good correlation is achieved. The ability of the new approach to solve boundary problems of geotechnical significance is thus illustrated. Nous présentons une maquette numérique du comportement contrainte–déformation d'un sol non saturé. Une maquette constitutive élastoplastique d'état critique étendue basée sur une contrainte nette et une succion est résolue par la méthode d'éléments finis. Nous proposons des techniques numériques qui tiennent compte du développement d'un algorithme de solution élastoviscoplastique dans le contexte d'une analyse d'éléments finis transitoire. La solution simultanée des équations différentielles épardelles couplées donne des variations de la contrainte nette, de la succion, du volume éspécifique et du degré de saturation dans un sol non saturé. Nous analysons alors plusieurs éééexemples qui illustrent les caractéristiques type du comportement d'échantillons de sol non saturé. Nous reproduisons les modes attendus de comportement et nous obtenons une bonne corrélation entre les résultats numériques et la solution alternative du modèle théorique. Nous comparons les prédictions numériques avec les résultats de deux expériences à succion sur le kaolin compacté et Pargile sableuse. Encore une foie, une bonne corrélation est obtenue, confirmant la validité de Palgorithme numérique. Enfin, nous essayons la solution d'un probléme à valeur limite, nommément la prédiction des mouvements saisonniers du sol. La comparaison des résultats numériques avec des données observées sur le terrain montre qu'on obtient une bonne corrélation. Nous illustrons ainsi Pefficacité de la nouvelle méthode à résoudre les problèmes limites à signification géotechnique.

No abstract

SUMMARYThis paper focuses attention on the development of a numerical model of the hydro/thermo/mechanical behaviour of unsaturated clay and its consequent verification and validation. The work presented describes on-going collaboration between the Cardiff School of Engineering and Atomic Energy of Canada. The model development, which was carried out at Cardiff, can be described as being based on a mechanistic approach to coupled heat, moisture and air flow. This is then linked to a deformation analysis of the material within a 'consolidation' type of model. The whole is solved via the finite element method to yield a computer software code named COMPASS (COde for Modelling PArtly Saturated Soil). Some aspects of verification and validation of the model have been addressed in-house. However, the purpose of current AECL work is to provide an independent, rigorous, structured programme of validation and the paper will also explore the further validation of COMPASS within this context.

No abstract