A Script-Based, Parameterized Finite Element Mesh for Design and NDE on a GPU

ABSTRACT

Finite element mesh generators exist in the public domain, a few even based on a parametric device description. The typical mesh generator requires some man–machine interaction to define the points and boundary conditions, and does not work for non-stop optimization iterations for which we need a mesh dynamically evolving through the iterations with optimization variables as changing parameters. Such mesh generators as do exist are rare, commercial, and not easily available to researchers except at great cost and never with the code to modify them to suit individual needs. We take a regular open source mesh generator and write a script-based interface as open source to run non-stop for optimization. We then use it to create a non-destructive evaluation system for army ground vehicles’ defect characterization and use it equally for machine design. A simple scheme of averaging neighbour heights gives us a smooth geometry without having to use Bezier curves. The mesh runs on the central processing unit but finite element optimization is on the graphics processing unit for speed and practicable testing times.

Keywords:

• Finite elements
• GPU
• NDE
• Optimization

Additional information

Funding

Funded in part by the US Army's TARDEC [contract number W911NF-11-D-0001].

Notes on contributors

S. Sivasuthan

Sivamayam Sivasuthan, BSc First Class Honours in Computer Science from the University of Jaffna, was born in Sri Lanka. He has been reading for a doctoral degree in Electrical and Computer Engineering at Michigan State University since January 2012. His thesis focuses on exploiting graphics processing units to parallelize finite element optimization.

E-mail: [email protected]

V. U. Karthik

Victor U. Karthik, BSc Honours in Electrical and Electronics Engineering from the University of Peradeniya, was born in Sri Lanka. He has been reading for a doctoral degree in Electrical and Computer Engineering at Michigan State University since January 2012. His thesis focuses on the genetic algorithm to optimize electro-heat problems for application in machine design and hyperthermia.

E-mail: [email protected]

A. Rahunanthan

Arunasalam Rahunanthan has a PhD degree in Mathematics from the University of Wyoming and a BSc Engineering degree from the University of Peradeniya in Sri Lanka. He is currently an instructor in the Department of Mathematics and Computer Science at the Edinboro University of Pennsylvania. His research interests include numerical methods for ODEs and PDEs, mathematical modelling of multiphase flows in multiscale porous media using graphics processing units (GPUs), and Bayesian inference for quantifying uncertainty in problems related to subsurface flows.

E-mail: [email protected]

P. Jayakumar

Paramsothy Jayakumar, is a senior research scientist, SAE fellow, and a member of the Analytics Team at the US Army Tank Automotive Research, Development, & Engineering Center (TARDEC) in Warren, Michigan. Prior to joining US Army TARDEC, he worked for BAE Systems, Ford Motor Company, Altair Engineering, and Engineering Mechanics Research Corporation in multibody dynamics software development, vehicle dynamics modelling and simulation consulting, simulation technology development, durability load simulation, vehicle instrumentation and loads measurement, and road load engineering. He has published more than 100 technical publications, including journal articles and conference papers. His research in terramechanics and multibody dynamics won the best paper awards at the National Defense Industrial Association's Ground Vehicle Systems Engineering and Technology Symposium in 2011 and 2012. He holds a US patent for a system for virtual prediction of road loads and tire modelling. He was also instrumental in developing seven SAE standards for tire testing for the purpose of tire modelling for which he received the SAE 2014 James M. Crawford Technical Standards Board Outstanding Achievement Award. He is a member of the US Army Acquisition Corps, an honourary fellow of the Department of Mechanical Engineering at the University of Wisconsin – Madison, and an associate editor for the ASME Journal of Computational and Nonlinear Dynamics. He received his MS and PhD degrees in structural dynamics from Caltech, and B.Sc. Eng. (Hons, First Class) from the University of Peradeniya, Sri Lanka.

E-mail: [email protected]

R.S. Thyagarajan

Ravi Thyagarajan currently serves as deputy chief scientist at the US Army Tank Automotive Research, Development and Engineering Center (TARDEC), and was selected to the Researcher Review Board as a senior technical specialist in June 2012. His research pursuits are in underbody blast modelling and design, occupant protection and fast-running modelling methodologies. He received his PhD in Applied Mechanics from Caltech, and has over 15 years of prior experience in the automotive industry at Ford and Visteon, where he was involved in all aspects of product development of automotive interiors. His automotive experience includes leadership roles in concept-to-launch product design, human factors/ergonomics development, as well as in the standardized application of CAE tools during the overall design engineering process, for which he won several awards, including the President's level Customer-Driven Quality award. He is a past recipient of the Forest R McFarland Award from SAE, holds two patents, and has co-authored over 40 technical papers. He received the Army Materiel Command (AMC) Systems Analysis awards in 2010 and 2012 for pioneering application of modelling and simulation methodologies in underbody blasts. He is also a member of the Army Acquisition Corps, and is a Certified Acquisition Professional in Systems Engineering (SE), Program Management (PM) and Science & Technology (S&T) Management.

E-mail: [email protected]

Lalita Udpa

Lalita Udpa is professor of Electrical and Computer Engineering at Michigan State University where she leads the Nondestructive Evaluation Laboratory. She holds BS and MS degrees in Physics from the University of Poona and MS and PhD. degrees in Electrical Engineering from Colorado State University. Her research interests include various aspects of NDE, such as development of computational models for the forward problem in NDE, signal and image processing, pattern recognition and neural networks, and development of solution techniques for inverse problems. Her current projects include finite-element modelling of electromagnetic NDE methods applied to aircraft geometries, application of neural network and signal processing algorithms to NDE data, and development of image processing techniques for automated analysis of NDE and biomedical images.

E-mail: [email protected]

S.R.H. Hoole

S. Ratnajeevan H. Hoole is a professor of Electrical and Computer Engineering at Michigan State University. He has previously served as a member of the University Grants Commission of Sri Lanka where with six others he regulated the administration of all 15 universities in that country. He also briefly was vice chancellor, University of Jaffna. A fellow of the IEEE and chartered engineer, he holds a doctorate from Carnegie Mellon University and a higher doctorate, the DSc (Eng.) degree, from London. Besides Engineering, he has contributed much to the learned literature in the humanities and social sciences.

E-mail: [email protected]