Increasing Reliability of Metal Forming Processes in Early Design Stages

The identification of an appropriate design for engineering structures and processes is a development task over a long period of time. Finally, the developed design should be primarily reliable. Focusing on reliability purposes just in an advanced design stage may necessitates modifications which are cumbersome to realize and expensive additionally. Hence, the assessment of reliability demands has to be set in at the very beginning of the design process. Thereby, the needed information to assess the reliability in a customary way is missing in general. Thus, a first measure requires to model uncertainty on the basis of available information, which originates in early design stages from experience and expert knowledge, to determine reliability statements. This is done reasonably with the aid of the uncertainty characteristic fuzziness. Thus, a reliability analysis is performed by means of a fuzzy structural analysis. In order to keep those procedures applicable for computational expensive applications, which is always a demanding task in reliability analysis, the fuzzy results are determined approximatively. The results of the reliability analysis provide not only the basis for comparative evaluations, but rather should be utilized to infer design specifications for the further design process. On account of intended reliability requirements the resulting point sets can be subdivided into permissible and non-permissible points. The aim is to abandon non-permissible regions from the design space and determine alternative design spaces appropriately. On account of the fact, that interaction between input parameters should be avoided, those alternative design spaces are described by means of hypercuboids. The functional dependencies between input and result parameters can not be assumed to be one-to-one in advance. Thus, the alternative input spaces can be non-connected. This is taken into account by solving an inverse problem. In this paper, a cluster analysis approach is applied. On the basis of the cluster results, the announced alternative design spaces are specified by means of hypercuboids. Those alternative design spaces provide the basis to update the design in view of reliability requirements in an early design stage. In general, the hypercuboids should be as large as possible to provide a latitude for the further design process, but should be also well-founded on available permissible points. The presented approach, which combines multiple innovative solution statements, provides a tool to affect the design process in early stages in view of reliability requirements while leaving scope for development, which is of prime importance for multidisciplinary design tasks. Thereby, the applicability of the presented approach is demonstrated be means of an industry-relevant example.

application/pdf C-II-03.pdf — 1.4 MB