My research focuses on quantifying how geologic setting and infrastructure design interact. This involves numerical models implemented in software that is capable of capturing the multiscale nature of these interactions. I am especially interested in natural hazard modeling and describing how infrastructure responds to this type of loading as an aggregated system. I incorporate the latest advances made in computing technology into open source tools that can be used by the research community as well as practitioners. Additionally, I work on ways to utilize advances made in sensing and UAV technology such that the data acquired from these technologies can be quickly processed and used to update model predictions.
Dynamic Rock-Fluid Interaction
Scour of rock by hydraulically induced plucking of rock blocks from the stream bed is an important geomorphic process in rock channel evolution and, on a practical level, this process is a critical issue facing many of the world’s dams at which excessive scour of the dam foundation or spillway can compromise their performance. Numerical modeling of rock scour is a challenging and interesting problem that combines rock mechanics and hydraulics of turbulent flow. Our approach has been to directly model the solid and fluid phase by representing the individual polyhedral blocks using the Discrete Element Method (DEM) and using the lattice Boltzmann method (LBM) to represent water. The multiscale nature of rock-fluid interaction requires adaptive meshing in a parallel computing environment so that fluid flow both through joints and fractures, and over the rock surface are captured. Currently, we are working on applying the coupled DEM-LBM approach on a High Performance Computing platform such the the numerical model is able to capture the multiscale nature of the fluid-solid interaction with sufficient fidelity to understand mechanistically what governs plucking.
Future potential applications include direct evaluation of the effect of water pressure inside the fractured rock mass, along potential sliding planes, and can be extended also to rock falls and slides into standing bodies of water, such as lakes and reservoirs.
Three-Dimensional Model Generation
Generating a realistic representation of a fractured rock mass is a first step in many different analyses. Field observations need to be translated into a 3-D model that will serve as the input for these analyses. The block systems can contain hundreds of thousands to millions of blocks of varying sizes and shapes; generating these large models is very computationally expensive and requires significant computing resources. We have taken advantage of advances made in big data analytics and Cloud Computing and developed an open-source program—SparkRocks—that is able to generate real-world scale block systems containing millions of blocks in minutes and achieve orders of magnitude speedup. Importantly, the same application is able to run on many different computing environments—from a laptop, to desktop, to cluster on the Cloud—so computational requirements can be seamlessly scaled up as required.
The next phase of this research is focusing on two aspects: 1.) Developing the ability to process various input data types, such as UAV (drone) or LiDAR surveys, to automatically extract rock mass characteristics and their variability and, 2.) generating stochastic realizations of the three dimensional rock mass given the variability extracted from the input data. The functionality of the data stream processing and stochastic generation is being developed to be extensible and agnostic of the application with the long-term goal of developing a research tool that is capable of generating three-dimensional models for various types of problems and that is portable to many different computing platforms.