CAE-Companion-2018-2019

Engineering WISSEN CAE

Further, non-parametric sizing optimization allows for an easy and straightforward realization of manufacturing and design restrictions. Even large scale industrial applications with up to millions of design variables are optimized highly efficiently for an overall increase in eco-efficiency. Bead Optimization Bead optimization supports the engineer to find the layout of bead stiffeners for shell-like structures, as, e.g. sheet metal components. Beads improve the static stiffness and dynamic behavior of the component due to an increase in moment of inertia. Manual selection of a proper bead pattern is often very difficult and requires a time-consuming trial- and error process. Compendia like “Steife Blech- und Kunststoffkon- struktionen” by Oehler andWeber, Springer-Verlag GmbH (1972), support the engineer in this complex task. However, standard bead pattern are only available for simple geome- tries e.g. rectangular plates. For more complex parts, non-parametric optimization approaches are able to automatically generate a bead layout (i.e. location, orientation and bead pattern) based on the FEA results. During the optimization (in general in 2-5 iterations) the nodes of the analysis model are moved normal to the ini- tial sheet surface. The resulting bead layout can be controlled by specification of a maximum bead width, height and other form parameters. Objective and constraint values for bead optimization can be chosen e.g. from: „ „ volume, mass „ „ center of gravity, moments of inertia „ „ stress, strain „ „ compliance, reaction forces, reaction moments, internal forces „ „ displacements, velocity, acceleration „ „ natural frequencies „ „ results from frequency response analysis To ensure the producibility of the part, manufacturing re- strictions have to be considered. The result is a clear, directly producible component design with significantly better perfor- mance compared to bead patterns derived from catalogues.

The improved design can be transferred into any CAD-system. Sizing Optimization Sizing is a tool to optimize dimensions of a structure (design). These could be geometric dimensions or properties like cross section parameters (e.g. radii) and thicknesses of finite elements. Sizing optimization is mostly applied for sheet metal structures at a later stage of the development process when the general layout of the component (i.e. the topology) is more or less fixed. Starting with the design area representing the part of the structure to be modified the optimization system determines a design proposal with new sizing dimensions (e.g. new shell thickness distribution) with an optimum relation between „ „ parametric optimization (modifying geometric parameters) using general optimization systems „ „ non parametric optimization (modifying shell parameters) as integrated into an FE-simulation code or provided by structural optimization software like, e.g. SIMULIA Tosca Structure Whereas parametric approaches modify a certain number of geometric parameters or assign constant shell thicknesses to some predefined areas non-parametric optimization applies changes for each single finite (shell) element in the model. Only this free sizing allows for the full improvement potential. Based on the resulting distribution of dimensions, e.g. thick- nesses, elements can afterwards be clustered to combined areas of constant thickness. weight, stiffness and dynamic behavior. Sizing optimization can be realized by

Figure 5: Sizing optimization of a car body – minimization of mass under consideration of stiffness requirements (model courtesy by The National Crash Analysis Center (NCAC)) Objective and constraint values for sizing can be chosen from „ „ volume, mass, center of gravity, moments of inertia „ „ compliance, nodal displacements, rotations, reaction forces, reaction moments, internal forces „ „ natural frequencies „ „ results from frequency response analysis, acoustic measures „ „ thermal values „ „ and any combination of those

Figure 6: Bead optimization of a muffler (images courtesy of Tenneco)

Continued on page 42

40