Charles A. Taylor- Stanford University
Title: Patient-specific Finite Element Modeling of Blood Flow and Vessel Wall Dynamics
Abstract: Hemodynamic factors including shear stress and pressure provide the stimuli for many acute and chronic changes in the vascular system and contribute to the initiation and progression of congenital heart diseases and acquired vascular diseases such as atherosclerosis and aneurysms. Furthermore, knowledge of hemodynamic variables is essential for properly assessing the severity of many vascular diseases and devising an appropriate therapeutic strategy.
While advances in cardiovascular imaging have provided unprecedented insight into vascular anatomy, noninvasive methods for quantifying physiologic variables are not as broadly applied, with the notable exceptions of phase contrast MRI and Doppler ultrasound. Yet, these imaging methods, no matter how advanced, can only provide data about the present state and do not provide a means to predict the outcome of an intervention or evaluate alternate prospective therapies. We have developed a computational framework for computing blood flow in anatomically relevant vascular anatomies with unprecedented realism. This framework includes methods for (i) creating subject-specific models of the vascular system from medical imaging data, (ii) specifying boundary conditions to account for the vasculature beyond the limits of imaging resolution, (iii) generating finite element meshes including anisotropy, adaptivity and boundary layers, (iv) assigning blood rheological and tissue mechanical properties, (v) simulating blood flow and vessel wall dynamics, and (vi) visualizing simulation results and extracting hemodynamic data. Such computational solutions of blood flow offer an opportunity to predict potential hemodynamic benefits of treatment strategies. An entirely new era in medicine could be created whereby doctors utilize simulation-based methods, initialized with patient-specific anatomic and physiologic data, to design improved treatments for individuals based on optimizing predicted outcomes. Methods and applications of modeling blood flow and vessel wall dynamics in the aorta, the lower extremities, the cerebrovasculature, coronary and pulmonary arteries in children and adults will be discussed. New challenges related to extracting time-dependent patient-specific geometry from 4D image data and the application of such data to image-based fluid-solid interaction will be presented.
Biography: Charles A. Taylor received his B.S. degree in Mechanical Engineering in 1987 from Rensselaer Polytechnic Institute. He then joined the Engineering Physics Laboratory at GE Research & Development Center in Schenectady, New York. He received his M.S. degree in Mechanical Engineering in 1991 and his M.S. Degree in Mathematics in 1992 from Rensselaer Polytechnic Institute. He entered the Ph.D. program in the Division of Applied Mechanics at Stanford in 1992 and earned his Ph.D. degree in 1996 for his work on finite element modeling of blood flow. He was co-advised by Professor Thomas J.R. Hughes in the department of Mechanical Engineering and Professor Christopher K. Zarins in the department of Surgery.
Dr. Taylor joined the faculty at Stanford University in 1997 and is currently an Associate Professor in the Departments of Bioengineering and Surgery with courtesy faculty appointments in the Departments of Mechanical Engineering and Radiology. He is internationally recognized for the development of computer modeling and imaging techniques for cardiovascular disease research, device design and surgery planning. His early research contributions included the first three-dimensional simulations of blood flow in the human abdominal aorta and the first simulations of blood flow in vascular models created from medical imaging data. He started the field of predictive, simulation-based medicine by describing the application of computational fluid dynamics to predict outcomes of cardiovascular interventions in individual patients. In recent years, his lab has developed techniques to model blood flow and wall motion in large, patient-specific models at realistic levels of pressure and has used these methods in patient applications ranging from congenital heart malformations in children to hypertension and aneurysms in adults. Dr. Taylor received the “Young Investigator in Computational Mechanics Award” in 2002 from the International Association for Computational Mechanics and the “R.H. Gallagher Young Investigator in Computational Mechanics” from the United States Association for Computational Mechanics in 2003. In 2007 he was elected as a Fellow of the American Institute for Medical and Biological Engineering (AIMBE)
Chris Johnson - University of Utah
Biography: Chris Johnson directs the Scientific Computing and Imaging (SCI) Institute at the University of Utah where he is a Distinguished Professor of Computer Science and holds faculty appointments in the Departments of Physics and Bioengineering. His research interests are in the areas of scientific computing and scientific visualization. Dr. Johnson founded the SCI research group in 1992, which has since grown to become the SCI Institute employing over 160 faculty, staff and students. Professor Johnson serves on several international journal editorial boards, as well as on advisory boards to several national and international research centers. Professor Johnson has received several awards, including the the NSF Presidential Faculty Fellow (PFF) award from President Clinton in 1995 and the Governor's Medal for Science and Technology from Governor Michael Leavitt in 1999. He is a Fellow of the American Institute for Medical and Biological Engineering (AIMBE), a Fellow of the American Association for the Advancement of Science (AAAS) and a Fellow of the Society of Industrial and Applied Mathematics (SIAM). (http://www.cs.utah.edu/~crj/)
Robert Haimes - Massachusetts Institute of Technology
Title: The use of Parametric CAD for Multidisciplinary Analysis and Design
Abstract: Multidisciplinary Analysis and Design requires the use of a consistent representation of a configuration that can be shared amongst the various Computer-Aided Engineering (CAE) disciplines. A natural choice for this geometric hub is a Computer-Aided Design (CAD) system. Unfortunately, each vendor's CAD system is unique; therefore an engineering organization must decide on a single system that interfaces with all the various CAE analyses to be utilized. This has been a major impediment to the deployment of multidisciplinary analysis and design systems throughout industry in that customized CAD API software must be generated as part of the CAE discipline and additionally maintained across CAD releases. Or worse, the geometry is translated to some neutral form usually losing much information (including the design intent) and resulting in a fragmented and fragile process.
Modular software that provides a vendor-neutral setting for directly accessing a variety of CAD systems through a unified and simple programming interface is critical for building a general analysis and/or design framework. This removes the direct CAD API (or translation) requirement and more easily allows for the coupling of best in class CAE. It should also be noted that the ability to generate meshes in a “hands-off” manner is important in analysis and crucial for optimization. In design settings the ability to change Parameters and regenerate components is essential. The software CAPRI has been developed to provide an appropriate foundation for all of these tasks.
Geometry is presented through CAPRI with a dual-view: the normal Boundary Representation (BRep) and a tightly associated discrete watertight tessellation. The BRep is comprised of a set of topological entities such as Nodes, Edges, and Faces (which refer to the geometric entities points, curves and surfaces, respectfully). The topological entities provide both connectivity information as well as handles into the CAD system's geometry kernel that are used to access the geometric data. The discipline-based grid generator can use either or both views to satisfy geometric and fidelity requirements of the solver. Access to the geometry can be performed through evaluations (and their inverse) as well as conversions to NURBS. The vertices of the tessellation contain owning information as well as geometric parameter values (t for curves, [u,v] for surfaces) to facilitate augmentation by evaluation.
While CAPRI has been quite successful at providing a vendor-neutral CAD interface, it does not manipulate the structure of the BRep (except for the splitting of periodic curves and surfaces). The BRep held by CAD (and reflected through CAPRI) contains essential configuration information but also artifacts of the model-building operations. As a result, small changes in design variables can cause discontinuous changes in the resultant new BRep even though the geometry may have smoothly morphed. These abrupt changes in topology will then cause discontinuous changes in the resulting grid (if meshed directly upon the BRep) and may propagate through the solver to the predicted physics. This is a serious impediment to automated design, and in the case of gradient-based optimization may force termination of the entire design procedure.
The work presented in this talk extends CAPRI's methodologies so that a BRep can be manipulated outside of CAD. A merge operation allows for the removal of topological entities such that the new BRep mirrors the original design intent (without construction artifacts). A split operation can divide parts of the BRep so that automatic grid generation can be directly supported. These operations are made possible due to the use of a lightweight Ferguson spline modeler that generates a thin skin, which is directly coupled and associated back to the CAD system's geometric data. This talk describes this simple topological algebra.
With the ability to manipulate, it is now possible to generate a new BReps whose topology remains fixed as long as the overall morphology of the component does not change. Additionally, the capability of being able to “drill down” to the owning CAD geometry removes the traditional problems associated with translation. Overall, this allows for the generation a continuously varying engineering representation of a configuration from a (possibly discontinuous) model that is produced through CAD system regeneration. Automation is a consideration throughout this work.
Biography: Robert Haimes is a Principal Research Engineer in the Aerospace Computational Design Laboratory of the Department of Aeronautics & Astronautics at the Massachusetts Institute of Technology. Bob’s major research focuses have been (1) scientific visualization for the results from CFD simulations, (2) parallel, distributed and High-Performance Computing, (3) applied computational geometry and (4) the use of geometry in conceptual through final design. Bob has had a number of projects in these areas funded by NASA, DoD, and industry. The research commonly has a component that produces software that is more generally useful than the original investigation. The products Visual3 and pV3 (for CFD-style visualization) are in active use worldwide (even though they were introduced more than a dozen years ago). CAPRI is an expression of the research in applied computation geometry and design (through analysis). This software has been adopted by a number of National Labs as well as the CFD meshing and the Structural Analysis software industry.
Mark S. Shephard - Rensselaer Polytechnic Institute
Title: Mesh Adaptation Considering Curved Meshes and Parallel Execution
Abstract: The generation and adaptive control of anisotropic meshes for large scale simulations presents a broad set of challenges that range for ensuring the appropriate consideration of curved geometry representation to dealing with mesh adaptation on massively parallel computers. This particularly true when applying higher that linear spatial discretization methods to problems on curved domains. Procedures to generate adaptively defined curved anisotropic meshes including controlled mesh configurations that address issues such as boundary layers and thin sections will be discussed.
Massively parallel simulations are performed on meshes that are partitioned on computers with over 100,000 computing cores. The ability to adapt meshes in those cases requires the capability to execute all steps in parallel on a distributed mesh. Building on previously developed parallel mesh adaptation procedures, extensions are being developed to support effective parallel mesh adaptation on very large core counts. The extension to be discussed include more scalable communication tools, new incremental dynamic load balancing procedures and supporting multiple parts per core.
Meshes and simulation results will be shown to demonstrate the effectiveness of the capabilities presented.
Biography: Mark S. Shephard is the Samuel A. and Elisabeth C. Johnson, Jr. Professor of Engineering, and the director of the Scientific Computation Research Center at Rensselaer Polytechnic Institute. He holds joint appointments in the departments of Mechanical, Aerospace and Nuclear Engineering; Civil and Environmental Engineering; and Computer Science. Dr. Shephard has published over 250 papers. He is a fellow in and the past President of the US Association for Computational Mechanics, a fellow and member of the General Council of the International Association for Computational Mechanics, a fellow of ASME and an Associate Fellow of AIAA. He is the editor of Engineering with Computers and on the editorial board of six computational mechanics journals. He is a co-founder of Simmetrix Inc., a company dedicated to the technologies that enable simulation-based engineering.
Eric P. Kronstadt - IBM T. J. Watson Research Center
Title: Three more orders of magnitude: The road to Exascale Computing
Abstract: The very first PetaFlop systems started appearing a little over a year ago, and already people are beginning to seriously consider how we get to an ExaFlop (1018 Floating Point Operations/second). Although it is likely to take close to a decade to do this, it is no pipe dream. Several government laboratories and government agencies as well as system venders and University researchers are beginning to explore what it might take to get to this level of performance. This talk will give a brief overview of the challenges and potential of achieving this goal.
Biography: Dr. Kronstadt was graduated from Brown University in 1967, and received his Ph.D. in mathematics from Harvard University in 1973. He was T. H. Hildebrandt Assistant Professor of mathematics at the University of Michigan from 1973 to 1978, where he conducted research in the area of several complex variables and functional analysis. He joined the Computer Science department of the IBM T. J. Watson Research Center in 1978. From 1978 to 1983 he helped develop software and architectural extensions for the Yorktown Simulation Engine (YSE). In 1983, he joined the VLSI Systems group, and became manager of that group in June 1983. In that capacity he was involved in the design and specification of a number of high performance experimental RISC microprocessors, as well as the development of a standard cell design system. From 1986 to 1988 he was manager of the Microsystems, Analysis and Design Department with responsibility for experimental microprocessor design, advanced VLSI chip design, and circuit design and analysis tools. After an assignment to the Research Division
Technical Planning Staff, he became manager of the RISC Systems Department in January, 1990. Subsequently he was named Director of Advanced RISC Systems in January, 1994, Director of Personal Systems Solutions in May, 1995. Responsibilities in these positions included, PowerPC based architecture, microprocessor design, compilers and operating systems, the IBM Anti-virus product, development of advanced handwriting recognition techniques and prototypes, development of MPEG encoding and decoding hardware and software, and the development of wireless and mobile computing environments. In 1996, Dr. Kronstadt became Director of VLSI Systems where his responsibilities continue to include development of the PowerPC architecture, research in microprocessor implementation and micro-architecture, as well as CAD development. Since 2004, he has been Director of Deep Computing Systems, with responsibility for advanced operating systems research, high performance computing architectures including BlueGene, and emerging high performance applications including computational biology.
Dr. Kronstadt is a member of the IBM Academy of Technology and was awarded two IBM Outstanding Technical Achievement awards for his work on the YSE software. He holds three patents in microprocessor design, and is a Fellow of the IEEE.
Kazuhiro Nakahashi - Tohoku University
Title: Block-Structured Cartesian Mesh Approach for Near-Future Peta/Exa-Scale CFD
Abstract: Currently unstructured-mesh CFD has become an indispensable tool for analyzing and designing aircrafts because of the capability to treat complicated configurations. However, it still has some critical issues for practical engineering use. In this talk, demands for next-generation CFD are described with an expectation of near future Peta/Exa-scale computers. Then, a Cartesian-mesh approach, as a promising candidate for next-generation CFD, is discussed by comparing it with the current unstructured-mesh CFD. It is concluded that the simplicity of Cartesian-mesh CFD from the mesh generation to the post processing will be a big advantage in the days of Peta/Exa-scale computers.
Biography: Kazuhiro Nakahashi is a professor in the department of aerospace engineering at Tohoku University in Sendai, Japan. He has developed CFD algorithms including a solution-adaptive grid method using tension-torsion spring analogy, FDM-FEM hybrid method, prismatic grid method, and overset unstructured grid method. During the last decade, he has been working on developing an unstructured-grid CFD code named TAS (Tohoku University Aerodynamic Simulation). Currently, the TAS-code is used for aerospace applications at JAXA (the Japan Aerospace Exploration Agency), industries, and universities in Japan. He is a fellow of the Japan Society of Mechanical Engineers and the Japan Society of Fluid Mechanics. He is currently the president of the Japan Society for Aeronautical and Space Sciences.