Simulation of system models containing zero-order causal paths. II: Numerical implications of class 1 zero-order causal paths (Q1179831): Difference between revisions
From MaRDI portal
Changed an Item |
Changed an Item |
||
Property / describes a project that uses | |||
Property / describes a project that uses: DASSL / rank | |||
Normal rank |
Revision as of 17:53, 29 February 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Simulation of system models containing zero-order causal paths. II: Numerical implications of class 1 zero-order causal paths |
scientific article |
Statements
Simulation of system models containing zero-order causal paths. II: Numerical implications of class 1 zero-order causal paths (English)
0 references
27 June 1992
0 references
This paper continues the work developed in part 1 [ibid. 328, No. 516, 959-979 (1991; reviewed above)] on the classification of zero-order causal paths (ZOCPs) and discusses different numerical techniques for dealing with class 1 ZOCPs with respect to computational efficiency and accuracy. (A class 1 ZOCP is a path between a port with derivative causality and a port with integral causality and such problems often arise in constrained mechanical systems of the form \(x'=f_ 1(x,y',u,t)\), \(y=f_ 2(x,y,t)\)). The approaches examined include specific techniques (such as the parasitic elements approach or the Lagrange multiplier approach) that eliminate the class 1 paths by transferring the dependent storage elements to the independent ones. The ensuing problems are then solved using standard stiff ordinary differential equation techniques but requires changing the causal bond graph model. These specific techniques are compared with that of solving the system of differential-algebraic equations directly both by Runge-Kutta methods and the backward differentiation formula based code DASSL on a constrained mechanical system and a planar pendulum problem. Since these class 1 problems are nearly always index 1 problems this direct approach works well. It has the advantage of being more accurate than specific techniques and requires no additional model modification, although it can be more computationally expensive.
0 references
Runge-Kutta methods
0 references
backward differentiation formula
0 references
zero-order causal paths
0 references
computational efficiency
0 references
derivative causality
0 references
integral causality
0 references
parasitic elements
0 references
Lagrange multiplier
0 references
causal bond graph model
0 references
differential-algebraic equations
0 references
code DASSL
0 references
index 1 problems
0 references