Design Sensitivity Analysis for Identification the Optimum Shape and Geometry Optimization by Using Finite Element Method (original) (raw)

JOURNAL OF MECHANICS OF CONTINUA AND MATHEMATICAL SCIENCES

Among many analysis methods, sensitivity analysis is one of a significant method used for many engineering solutions in many applications like the major uncertainties, model validation, model refinement and decisions making. There are different challenges in optimization and improvement of engineering products, like products life, esthetical shape, weight and durability. The main objectives of this work are to optimize the shape geometry and increase the service life of the product by determining and then minimizing the stresses concentration through predicting the influence of any change in geometry to recommend the optimum design. Sensitivities measurement is normally calculated based on computational technique conjunction with direct differentiation method. In this work, Finite element software under ABAQUS/CAE code has been adopted for analysis and simulation. In ABAQUS, and by default; appropriate perturbation can be determine automatically depend on a heuristic algorithm by using central differencing method. In this work; rubber brace are used for analysis, and the main design parameters used to specify the product sensitivity of the final geometry are: product thickness, small fillets and modules of elasticity. A reasonable result has been estimated in terms of stresses and product dimensions. Due to nonlinearity behavior; the reduction in stresses concentration is about 9%, and the product fillet yields to new values with small increment due the variable mass scaling used in boundary conditions. As results of this analysis, the zones of high stress values are specified, and the most effecting parameters on this stresses are determined. It's concluded that this technique is useful for many features like contacts, viscoelasticity and also in nonlinear analyses. Even more, sensitivity analysis can used to develop and improve the design before any further analysis.