A New Insight Into Recursive Forward Dynamics Algorithm And Simulation Studies Of Closed Loop Systems

Abstract

Rigid multibody systems have been studied extensivley due to its direct application in design and analysis of various mechanical systems such as robots and spacecraft structures. The dynamics of multibody system is governed by its equations of motion and various terms associated with it, such as the mass matrix, the generalized force vector, are well known..Forward dynamics algorithms play an important role in the simulation of multibody systems and the recursive forward dynamics algorithm for branched multibody systems is very popular. The recursive forward dynamic algorithm is highly efficient algorithm with O(n) computational complexity and scores over other algorithms when number of rigid bodies n in the system is very large. The algorithm involves finding an important mass matrix, which has been popularly termed as articulated body inertia (AB inertia). To find ijth term of any general mass matrix, we separately give virtual change to ith and jth generalized coordinates. At each point of the multibody system, the dot product of the resulting virtual displacements are taken with each other and eventually integrated over the entire multibody system, weighted by the mass. This quantity divided by the virtual changes in ith and jth coordinates gives the ijth element of the mass matrix. This is one of the fundamental ways of looking at the mass matrix. However, in literature, the AB inertia is obtained as a result of mathematical manipulation and its physical or geometrical significance from the above view point is not clear. In this thesis we present a more geometric and physical explanation for the AB inertia. The main step is to obtain a new set of generalized coordinates which relate directly to the AB inertia. We have also shown the equivalence of our method with existing methods. A comprehensive treatement on change of generalized coordinates and its effect on equations of motion has also been presented as preliminaries. The second part of the thesis deals with closed loop multibody systems.A few years ago an iterative algorithm called the sequential regularization method (SRM) was proposed for simulation of closed loop multibody systems with attractive claims on its efficiency. In literature we find that this algorithm has been implemented and studied only for planar multibody systems. As a part of the thesis work, we have developed a C-programming language code which can simulate 3-dimensional spatial multibody systems using the SRM algorithm. The programme can also perform simulation using a relatively efficient Conventional algorithm having O(n+m3) complexity, where m denotes number of closed loop constraints. Simulation studies have been carried out on a few multibody systems using the two algorithms. Some of the results have been also been validated using the commercial simulation package -ADAMS. As a result of our simulation studies, we have detected certain points, after which the solution from SRM loses it convergence. More study is required to understand this lack of convergence.