We are surrounded by complexity – our world is a compendium of discrete entities (atoms, grains, rock pieces, etc) that interact with each other by a set of laws of which we now know a few. However, understanding those interactions does not necessarily make it easy to create a model based on them. One example is the exercise of modelling a soil. To grasp an understanding of the titanic task ahead for the interest modeller, one may consider key physics layers, relevant to the problem, such as elastic and frictional forces between the grains, the capillary bridges for the water pockets trapped in the soil and the different chemical reactions that can produce either flocculation or fragmentation of soil sub volumes. Because of this complexity, the continuum modelling approach has dominated these disciplines for decades. However, the ever growing computational power that is available for scientific computing nowadays makes it possible to deal with this complexity and consider the interacting entities individually. This is the discrete approach for modelling which has recently gained popularity in both science and engineering disciplines. In this presentation I will present some exciting developments on the use of the discrete approach to model and tackle problems related to geo-science and engineering such as the bearing capacity of soils, the conductivity of anisotropic porous media and the connectivity of fracture networks. It will be highlighted how the two approaches (continuum and discrete) need not to be exclusive but instead complement each other to create full multiscale frameworks to gain both understanding and prediction of the behaviour of complex geo-systems.

Biography:

Dr. Sergio Torres is a lecturer in applied mathematics at the University of Queensland Australia. After obtaining his Ph.D. in computational physics in 2010 he has devoted his research efforts to develop novel numerical methods for their use in applied fields with a focus on problems related to civil, environmental and geotechnical engineering. At the moment he and the Ph.D. students he works with have produced several simulation engines that are carefully validated with experiments conducted at UQ. For these achievements, he was recently awarded an Advance Queensland Fellowship (AQF). He has authored and co-authored more than 40 journal publications and is the lead developer of the open source library Mechsys which has been used for research and consultancy projects in Australia, Germany, United States, Ecuador, Colombia and China.