Proceedings of the 18 th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013 1 Numerical modelling of desiccation crack induced permeability Modélisation numérique de la perméabilité induite par la fissuration des sols R. A. Stirling, C. T. Davie & S. Glendinning Newcastle University, Newcastle-upon-Tyne, UK ABSTRACT: The development of cracking as a result of desiccation and the apparent increase in permeability of cracked fill is increasingly under investigation. Rainfall infiltration into soil surfaces that experience cracking increases due to the additional, preferential transmission of water. This in turn results in cycles of rapidly elevated pore water pressure and is widely cited as a significant mechanism for strength reduction that leads to embankment failure. A two-phase flow numerical model that allows the partially saturated behaviour of the desiccated medium to be captured is presented based on the finite difference code FLAC 2D. The material properties of the developed model, including soil stiffness and strength, are incorporated as a function of drying. The model has allowed investigation into the factors influencing the incidence and scale of cracking. RÉSUMÉ : L’infiltration des précipitations dans les sols sensibles à la dessiccation augmente comme résultat de la transmission préférentielle, additionnelle d’eau. Ce phénomène se traduit par des cycles de pression interstitielle rapidement élevée, et est largement cité comme un mécanisme important de la réduction de la résistance qui conduit à la rupture des remblais. Un modèle numérique de l’écoulement diphasique, permettant la prise en compte du comportement partiellement saturé du milieu desséché, est présenté. Ce modèle est basé sur un code de calcul de différences finies, FLAC 2D. Les propriétés du matériau du modèle, y compris la rigidité et la résistance du sol, sont incorporées comme fonction du séchage dans la description de la courbe caractéristique sol-eau. Le modèle a permis également l’évaluation des principaux facteurs qui influencent l'incidence et l'ampleur de la fissuration des sols. KEYWORDS: Numerical modelling, Unsaturated soils, Soil behaviour 1 INTRODUCTION Cracking within clay fills has been an accepted phenomenon for many decades. The engineering study of desiccation cracking has been motivated by its impact upon the effectiveness of many earth structures including liners (Philip et al 2002), foundations (Silvestri et al 1992), cuttings and embankments (Smethurst et al 2006) due to an apparent increase in water infiltration. Desiccation cracking is the product of volumetric shrinking of clays brought about by a reduction in soil-water content. Cracking initiates when tensile stresses generated by increasing suctions exceed the soil strength, which in itself, is controlled by soil water content. Variability in soil-water content is primarily the result of seasonal fluctuation in precipitation/evaporation in addition to the transient demands of vegetation and the infiltration potential of the soil surface and is therefore largely governed by climate. Predicted climate change scenarios are recognised to have the capacity to more frequently bring about conditions conducive to the increased occurrence of this behaviour because of the increased occurrence of warmer and drier summers experiencing rainfall events of shorter duration and higher intensity (Hulme et al 2002, Jenkins et al 2010). Progressive failure is thought to be largely governed by permeability which is in turn controlled by the micro- and macro-scale structure of the soil. Previous studies have established that current permeability measurement techniques produce discrepancies between both laboratory and field established values and numerically simulated pore-water pressure values (Smethurst et al 2006, Rouainia et al 2009). These differences have been identified as being caused by permeability values ranging by up to three orders of magnitude (Nyambayo and Potts 2005, Rouainia et al 2009). Albrecht and Benson (2001) identified the same increase in hydraulic conductivity of three orders of magnitude in laboratory testing of small cracked samples when compared to equivalent non- cracked samples of the same material. This supports the notion that it is the presence of pervasive cracks that results in the elevated permeability. An empirically reasoned permeability modification has been employed in the modelling of embankment pore pressures (Nyambayo et al 2004). Many researchers have attempted to model the mechanisms involved in crack initiation and propagation, particularly with respect to crack pattern. Kodikara and Choi (2006) present a simplified analytical model for laboratory cracking which has subsequently been implemented by Amarasiri et al. (2011) into a distinct element code. Their work describes the modelling of cracking behaviour in slurried clays under given laboratory boundary conditions and incorporates material changes due to drying. More recently, work has been carried out using the finite element method to investigate the development of tensile stresses associated with desiccation (Trabelsi et al 2011, Peron et al 2012). In contrast, this work models partially saturated flow throughout the medium induced by a simulated evaporation boundary and combines this mechanism with the ability to capture a fracturing geometry. 2 TWO-PHASE FLOW Modelling has been carried out using the commercial finite difference code, FLAC (Fast Lagrangian Analysis of Continua) (ITASCA, 2002). The internal programming language, FISH, has allowed material variables to be defined as a function of water content. Given the fundamental influence of water content in desiccation cracking, it is important to be able to capture the partially saturated behaviour of the medium. To do this, the