Contents of Journal of Mechanical Engineering 59, 1 (2008) 



AZIM, G. A., ATTIA, H. A.: Neural networks construction 
application to the kinematical analysis of the five-point 
suspension                                                                1

SEENI KANNAN, P., PERIASAMY, V. M., NAGARAJ, P.: Design 
strategy for six-cylinder stationary diesel engine exhaust 
systems                                                                  13

MURČINKOVÁ, Z., KOMPIŠ, V.: Advanced modelling of short-fibre 
reinforced composites                                                    27

RAO, A. CH., SRINATH, A.: Allocation of joint backlashes 
in manipulator linkages with emphasis on platform-type robot             41

Letter to the Editor
SUDHAKAR, A. V., RAJU, V. R. CH., MURTHY, V. B. K., RAO, 
K. M., KUMAR, M. S. R. N.: Study of thermoelastic behaviour 
of skew laminated composite plate with circular cutout using 
finite element method                                                    53



Abstracts



Neural networks construction application to the kinematical analysis of the five-point suspension

GAMIL A. AZIM, HAZEM A. ATTIA


In this paper, a General Neural Networks Regression (GNNR) is presented. The position analysis of multi-link five-point suspension system is solved using GNNR. The mechanism of wheel suspension is a multi-body system (MBS) which is a system of bodies (in this case rigid links of given length) whose mutual position is bounded by geometrical constraints (joints) and by active kinematical driving constraints in the form of prescribed motion related to the degrees-of-freedom (DOF) of the given mechanism. GNNR is first trained with coordinates of the defined points and input driving variables. After training, performance is measured by having the network generate the coordinates of the defined points in terms of driving input variables. It is found that GNNR provides a simple and effective way to model the spatial kinematics of the five point's suspension and eliminate the convergence problems associated with algorithmic solution methods.

Design strategy for six-cylinder stationary diesel engine exhaust systems

PAULDURAY SEENI KANNAN, VELAYUTHAMPALAYAM MUNIAPPAN PERIASAMY, PERIA SAMY NAGARAJ

Overall engine performance of an engine can be obtained from the proper optimized design of engine inlet and exhaust systems. This paper presents the influence of an exhaust manifold pressure on the performance improvement of a reciprocating four-stroke engine. The application of computational methods for the development of high performance of a four-stroke stationary engine has been evaluated. A one-dimensional fluid dynamic code has been employed to simulate engine performance at full load, and data predicted from computer simulation have been compared with experimental results. To do this aforementioned validation process, computer simulation techniques were applied to develop a new exhaust system so as to optimize volumetric efficiency over a wider range of speed. These techniques proved to be powerful and effective in identification of the modifications required to obtain the engine performance targets. Findings from these studies are used to derive exhaust system design guidelines that define the optimum exhaust system geometry to give effective scavenging and fine cylinder charging.


Advanced modelling of short-fibre reinforced composites

ZUZANA MURČINKOVÁ, VLADIMÍR KOMPIŠ

In this paper, a Method of External Finite Element Approximation\linebreak (MEFEA) to model such problems like composites reinforced by short fibres will be presented. MEFEA is an enhanced classic FEM with idea of external approximations. There are shape functions in the discrete solution space that do not belong to the infinite dimensional solution space. The domain is split in subdomains (cells) and the approximation is built on each of these subdomains independently of each other. The method is similar to Hybrid Trefftz Finite Element Method, where Trefftz functions are used inside each element (subdomain). The displacement and force boundary conditions are met only approximately whereas the governing equations are fulfilled exactly in the volume for linear elasticity, making it possible to assess accuracy in terms of error in boundary conditions. The main benefit is that the discretization can be done directly on a 3D CAD geometry with all details (features) for the analysis.


Allocation of joint backlashes in manipulator linkages with emphasis on platform-type robot

ATLURI CHAKRADHARA RAO, ADUSUMILLI SRINATH

Joint clearances in linkages and backlashes in gear trains lead to "position errors". Accuracy suffers if clearances and backlashes are allocated liberally while tighter allocation leads to higher manufacturing cost. Hence acceptable procedure is to allocate joint clearances or backlashes in such a way that the maximum position error is limited to a specified value. A simple method based on the concepts of parallelism and optimization is proposed. This method is extended to platform-type spatial robots.