Efficient Simulation of Physical System Models
Using Inlined Implicit Runge-Kutta Algorithms
Abstract
Stiff systems commonly occur in science and engineering, and the use of an implicit
integration algorithm is typically needed to simulate them. As model complexity
increases, the need for efficient ways to solve these types of systems is becoming
of increasing importance. Using a technique called inline–integration with implicit
Runge–Kutta (IRK) algorithms and tearing may lead to a more efficient simulation.
To further increase the efficiency of the simulation, the step–size of the integration
algorithm can be controlled. By using larger integration steps when allowable, the
simulation can progress with fewer computations while still maintaining the desired
accuracy.
In this thesis, for the purpose of step–size control, two new embedding methods
will be proposed. The first will be for HW-SDIRK(3)4, a singly diagonally implicit
Runge–Kutta (SDIRK) algorithm and the second for Lobatto IIIC(6). These two
embedding methods will then be compared with those previously found for the Radau
IIA family and Lobatto IIIC(4), all fully–implicit Runge–Kutta algorithms.
Interested in reading the
entire thesis?
(118 pages, 1,506,190 bytes, pdf)
Interested in viewing the
Powerpoint presentation?
Homepage
Last modified: July 18, 2005 -- © François Cellier