The use of different types of simulations in the initial stages of product development has become an everyday and consistent practice. A step forward of virtual assessment with computer technology in product development is non-parametric optimization, which not only evaluate the existing, but also propose new and better solution.
The global market today requires new products to be lighter, cheaper, robust, resistant to fatigue and to be launched on the market in time. The software companies have in the last few years launched besides the simulation tool also topology and other non-parametric optimization technologies. In this article, various options that the user has when optimizing with a computer are briefly presented.
- PROPERTIES OF NON-PARAMETRIC OPTIMIZATION
The term non-parametric optimization is used for the type of optimization, which optimize the coordinates of the nodes or the values of the state variables (usually density) in the elements integration points directly. Optimization thus works on the of finite elements model. In this text, the focus is on software simulation and optimization tools provided by Dassault Systemes in its SIMULIA brand.
- NON-PARAMETRIC OPTIMIZATION
3.1 Topology optimization
Topology optimization begins with the definition of design space, in which the geometry itself can be located. The optimization algorithm iteratively redistributes the density of the material within the design space in such a way, that objective function is minimized or maximized. At the same time, the defined constraints must be respected.
Figure 1: Topology optimization of the fork
An example of topology optimization of the fork is shown in Figure 1. The goal of optimization is to increase stiffness as much as possible while respecting the constraint that the initial design volume must be reduced by 57%. Also, manufacturing constraint is added, which prescribe, that the forks are manufactured by casting technology.
On the basis of topology optimization results, we can build a parametrized CAD model, which captures surfaces, obtained by smoothing operation in topology optimization software.
3.2 Shape optimization
The next type of non-parametric optimization is shape optimization. Its task is to optimize the surface of the finite element model with optimization of the nodal coordinates on that surface. Usually the objective of shape optimization is to minimize the stress concentration. The example of the connecting rod shape optimization is shown in Figure 2.
Figure 2: Connecting rod shape optimization
Similar as in the case of topology optimization, it is necessary to build a parametrized CAD model based on the FEM model, which accurately captures node-based surfaces.
3.3 Sizing optimization
Similar to shape optimization, sizing optimization is used at the end of the design process when the general layout of a component is fixed, and only minor changes are allowed. With sizing optimization shell thickness in selected regions can be optimized. An example of this type of optimization is shown in Figure 3, where optimized sheet thickness on the vehicle body model is shown. This case is the feasibility study on FEM model, developed at the NCAC Centre of George Washington University.
Figure 3: Sizing optimization of vehicle body
The result of the optimization was reduced body weight by 15% (49 kg) with preserved torsional, bending and axial stiffness.
3.4 Bead optimization
Bead optimization adds stiffening beads to a shell structure based on theory of bending lines and is an excellent tool for increasing the stiffness and eigenfrequencies of the structure. Consequently, beads reduce the acoustic emission of thin sheet metal products. Optimization itself requires 2 to 3 simulations and determines the shape and height of the beads. It can greatly improve product development and shorten the development cycle. An example of the bead optimization is shown in Figure 4.
Figure 4: Bead optimization
The computer optimization technologies, presented in this article, cannot replace engineering knowledge and long-term experiences of the user, but they can be of great help in development, especially in cases of combined loads and various boundary conditions.
Author: Dr Marko Vrh, CADCAM Lab d.o.o.