FACULTY:

Kurt S. Anderson
Professor

4006 Jonsson Engineering Center
Department of Mechanical, Aerospace & Nuclear Engineering
110 - 8th Street
Troy, New York 12180-3590

Tel: (518) 276-2339
Fax: (518) 276-2623

E-Mail: anderk5@rpi.edu
Website: www.rpi.edu/~anderk5

Background:

After receiving his BS degree in mechanical engineering from the University of California at Berkeley in 1982, Dr. Anderson went on to earn a MS in the area of dynamic systems and control from the same institution in 1984. He then spent the next two years working in the areas of dynamics, structural dynamics, and controls for TRW Space and Technology (Now Northrop-Grumman) in Redondo Beach, California. After this period, he entered the Ph.D. program in Applied and Computational Mechanics at Stanford University, earning his degree in 1990. Dr. Anderson then accepted a position as researcher and principal dynamics engineering at TRW where he was associated with various spacecraft and research programs. In late 1991 Dr. Anderson was invited to Germany for a two-year period as a visiting scholar, lecturer, and Alexander von Humboldt Research Fellow at the Technische Hochscule - Darmstadt. In 1993 he joined the faculty of the Department of Aerospace Engineering, Applied Mechanics, and Aviation at The Ohio State University, in Columbus where he remained until coming to Rensselaer as faculty member in August 1995.

Research Interests and Activities:

Professors Anderson's primary research goals are associated with development of advanced algorithms for modeling, simulating, and analyzing the behavior of complex dynamic systems. Examples of such systems include, but are not limited to molecular systems, spacecraft, robotic systems, automotive applications, the human body, and manufacturing operations. These analysis and simulation tools emphasize the use of algorithms which obtain the desired accuracy, while requiring far fewer computational operations than their more traditional counterparts. This results in simulations which run much more quickly, or equally important, allows a level modeling and analysis which would otherwise be prohibitively expensive. This is a accomplished through the use of special low operational order algorithms, multirate temporal integration methods, and intelligent exploitation of parallel computing.

STUDENTS:

Kishor Bhalerao
PhD Student

5303 Jonsson Engineering Center
Department of Mechanical, Aerospace & Nuclear Engineering
110 - 8th Street
Troy, New York 12180-3590

Tel: (518) 276-2009

E-Mail: bhalek@rpi.edu
Website: www.rpi.edu/~bhalek

Research Motivation:

Many engineering systems involve unilateral constraints. In the past people have tried to treat such systems using methods developed for bilateral constraints. With this type of approach the system jumps from one set of active bilateral constraints to another as the unlateral constraints become active (or inactive). The difficulty with this approach is that to be sure one has the correct set of active constraints one should perform a 2C possibility combinatorial search (where C is the number of potentially active constraints. Worse still, it is possible that none of these may realizable solutions. Such a problem can quickly become intractable for large C.

Complementarity Theory provides a powerful means for more efficiently dealing with these types of systems. Most of the work to date in using Complementarity Theory in a multibody dynamics context has involved the use of underlying Newton-Euler type formulations. Our work here is in researching how to better meld the use of complementarity method, with efficient multibody formulations, so that the best of each of these approaches may be brought to bare on difficult unilateral constraint problems. Applications being considered for this method include robot manipulator problems, certain assembly operations, and molecular dynamics of biopolymers.

Projects:

  1. Derivation, verification and validation of a Hybrid Divide-and-Conguer/Complementarity Algorithm in a general context of multibody systems.
  2. Development of an operational C++ parallel code to test the performance of the algorithm and compare it to the performance of competing methods.
  3. Investigate the uniqueness of the solutions returns and resolve how to guarantee that the solution returned is phyically correct
  4. Apply this approach to specific Robotic and Molecular dynamic test cases.


Mohammad Poursina
PhD Student

5303 Jonsson Engineering Center
Department of Mechanical, Aerospace & Nuclear Engineering
110 - 8th Street
Troy, New York 12180-3590

Tel: (518) 276-2009

E-Mail: poursm@rpi.edu
Website: www.rpi.edu/~poursm

Research Motivation:

In the world of computation, time is the enemy. Scientists wage a battle of wits to squeeze as many computations as they can into the shortest possible span of time. The larger and more complex the problem, the more cunning their techniques must be.1 Given limitations on available time and computing resources, and the desire/need for the accuracy in the calculated results, considerable attention must be paid to putting the computational effort only when and where it is needed.

The above statement best summarizes the objective of my research work in large-scale dynamic systems with direct applications to molecular dynamics (MD) simulations. In MD, the number of bodies involved could potentially be in millions while on the temporal scale, the desired simulation runs can amount to order of 1015 time steps. This computational complexity of MD simulations makes it a rich area of research for exploiting the computational multibody dynamics algorithms and advanced temporal integration schemes for increasing the celerity of computations.

1 Reiter C. 1995. Forecast: clear weather ahead. Northwest. Perspect. Winter 23–24

Projects:

  1. Implementation of multibody dynamics algorithms for MD simulations.
  2. Design of efficient algorithm(s) for flexible body dynamics.
  3. Development of Object-Oriented Multibody Simulation software.
  4. Exploration of substructuring strategies for reduced order MD simulations.


Caitlin Thomson
Undergraduate Student

5303 Jonsson Engineering Center
Department of Mechanical, Aerospace & Nuclear Engineering
110 - 8th Street
Troy, New York 12180-3590

Tel: (518) 276-2009

E-Mail: thomaa9@rpi.edu
Website: www.rpi.edu/~thomaa9

Research Motivation:

A faster and more efficient method to solving a problem is of no use if it provides inaccurate results. Thus, it is of utmost importance to first test the accuracy of a new method with simple problems before moving on to more complex scenarios. It is much easier to determine the flaws of a method using a problem with only a few parameters rather than one with several variables and constraints.

At the Computational Dynamics Laboratory, we are developing a new Adaptive resolution algorithm formulation for dynamic systems simulation. Such a formulation is better able to put computing effort when and where we need it.

Projects:

  1. Validation and verification of the novel state-time dynamics formulation.
  2. Modeling and simulation of a laser-powered lightcraft utilizing the Autolev.
  3. Gain understanding of basic theory underlying Multibody Methods such and the projection approches emphasized in this lab.
  4. Aid in the development and testing of new modules for POEMS.

Matt McDaniel
Undergraduate Student

5303 Jonsson Engineering Center
Department of Mechanical, Aerospace & Nuclear Engineering
110 - 8th Street
Troy, New York 12180-3590

Tel: (518) 276-2009

E-Mail: mcdanm@rpi.edu
Website: www.rpi.edu/~mcdanm

Research Motivation:

Dynamics is the language of motion, and with it one may computationally model many complex systems. However, current computational methods in dynamics have limitations and need to be enhanced in order to be used in more general and useful applications. At the Computational Dynamics Laboratory at Rensselaer, multibody simulators are being developed that greatly reduce the cost of analysis of a system. One such simulator is the Parallel Open-source Efficient Multibody Simulator (POEMS), an outgrowth of the YAMS (Yet Another Multibody Simulator)project started by Dr. James Critchley. I have the desire to enhance the development of POEMS by adding Kane's method and other useful tools, such as Verlet integration, which is very useful for molecular dynamics applications.

Projects:

  1. Analysis and documentation of the POEMS code.
  2. Adding more diverse capabilities to POEMS using Kane's method.
  3. Implementation and testing of the Verlet Method for numerical integration.
  4. Add the Divide ad Conquer Algorithm (DCA) and Flexible Divide and Conquer Algorithm (FDCA) to POEMS.

Ben nagler
Undergraduate Student

5303 Jonsson Engineering Center
Department of Mechanical, Aerospace & Nuclear Engineering
110 - 8th Street
Troy, New York 12180-3590

Tel: (518) 276-2009

E-Mail: nagleb@rpi.edu
Website: www.rpi.edu/~nagleb

Research Motivation:

Working both with Professor Anderson and Professor Myrabo on understanding and predicting the flight characteristics of a Laser Propelled Lightcraft. Such technology offers the potential of ecomonic launch of small spacecraft into low earth orbits.

At the Computational Dynamics Laboratory, we are developing a detailled simulation model with incorporates experimentally determined aerodynamics, separate engine and beam models, as well the full nonlinear equaions of motion for a seven degree-of-freedom (7 DOF) [General despun payload, plus reaction-wheel type rotor for attitude stabilization of craft).

Projects:

  1. Validation of full system simulation model through comparison with experimental flight data.
  2. Inclusion of Magnus Force model.
  3. Development of visualization tools associated with this system.
  4. Extension of the model for control puroses.

Last modified on July 12, 2004 by J. Evans; Created by J. Evans
©2004 Rensselaer - Department of Mechanical, Aerospace and Nuclear Engineering