Geomaterials, i.e., soil and rock, are quite prevalent in nature and they are part of numerous engineering problems and solutions. All civil engineering structures have to be ultimately supported on soils or rocks. Soils and rocks play an integral part in the behavior of civil engineering structures such as buildings, bridges, dams, levees, and aqueducts during catastrophic loading events such as earthquakes. Soils are also major part of environmental problems and solutions. Production of petroleum hydrocarbons involves rocks at large depths. Geomaterials are multiphase porous media consisting of a solid skeleton and number of fluids within the pore space. Geomaterials exhibit highly nonlinear and hysteretic behavior during various loading situations. Traditionally the engineers have treated the above-mentioned processes separately and often with simplifying assumptions. With the recent advances made in Information Technology (IT) in areas such as scalable parallel computer algorithms, distributed computer systems, high speed networks, and visualization it is now possible to realize a unified geo-analysis tool for solving current and future problems involving geomaterials where unnecessary simplified assumptions need not be made and the full 3-D model of a real world system can be analyzed.
A key component in achieving this unified analysis tool is a framework-based finite element application. A framework represents a collection of common software components for building applications in some common domain. The basic premise behind the use of a framework is the recognition of a common set of tasks that must be accomplished in writing the applications. These tasks can be factored out of the application codes and collected into a single set of components. Essentially a framework separates physics of a problem from the computer science aspects of solving that problem.
This project is directed toward developing a scalable, parallel geo-analysis tool using a framework-based finite element application in a three-way collaboration between the University of Oklahoma (OU), the TeraScale, LLC, and the Lawrence Livermore National Laboratory. This tool and the underlying framework will be readily available to the academic community and industry at a fraction of the cost of similar monolithic codes. This analysis tool will initially be capable of analyzing static and dynamic behavior of dry, unsaturated, and saturated soils (fluid flows and solid skeleton deformations). However, as opposed to the "closed" architecture of the proprietary monolithic codes, the "open" architecture of the framework will provide users with easy extendibility to include other multi-physics behavior in geomaterials.
The resulting tool will be verified first using closed-form solutions of simple problems in saturated and unsaturated soils. Then the tool will be validated using centrifuge model tests results for boundary value problems. Computational experiments will be conducted to test the analysis tool's capabilities under various distributed and parallel computing environments. On the educational front the TeraScale framework will be integrated into courses such as the Introduction to Finite Element Method and thereby introducing students to this new paradigm in engineering analysis and IT concepts such as scalable parallel computing. Using Oklahoma's OneNet, High School students in Oklahoma will be provided an opportunity to manipulate and view the results from the proposed geo-analysis tool as well as conduct some simple analyses through the Internet. Seeing and interacting with visualization of exciting problems such as the earthquake loading of a dam will help recruit more High School students to science and engineering.
It is expected that the geo-analysis tool developed will not only make the current advances in IT, such as the distributed computing, widely accessible to engineering community, but also advance various aspects of IT research itself in a synergetic manner. For example, the proposed geo-analysis tool will serve as a test case for evaluating the efficiency of current and future distributed computing facilities.
