Session: 06-04-01 Design for Additive Manufacturing I

Paper Number: 68912

Start Time: Friday, 11:25 AM

68912 - Structural Optimization for Segment-based Design of Part Candidates For Incremental Manufacturing 

Companies face growing market-specific requirements and highly individualized consumer demands, leading to a diversification of the product range, such as in the automotive industry. Here, high demand for individualized equipment and growing technical diversification, e.g., in drive systems, leads to a rising number of product variants while the batch sizes decrease. Consequently, variant-specific investments and batch-specific setup costs rise, which can cause traditional manufacturing processes (for instance, injection molding or die-casting) to reach their economic limits.

One contribution to enable the economical production of variants while keeping sufficient production volumes is Incremental Manufacturing (IM). This hybrid approach is based on the finalization of pre-produced parts using additive (AM) and subtractive manufacturing (SM) to add variant-specific design features. Furthermore, assembly and measuring steps can be carried out in the fully robot-based manufacturing cell. Overcoming the limitations of conventional serial-linked manufacturing strategies enhances the complexity in product design and production planning. Therefore, new methods and tools to support product development in the early design stages are investigated to exploit the AM-based production concept's full potential.

In the context of part re-design for IM, a graph-based design approach exists that supports the designer in generating part concepts based on automated CAD model segmentation of a reference product. Here, the extracted part segments are concretized based on algorithms, for example, by assigning predestined manufacturing processes, suitable pre-produced parts, materials, and joining technologies. For a concretization and early potential evaluation concerning the part candidate's functional properties (mechanical stiffness), a simulation and optimization framework is proposed in this publication. This framework enables an integrated optimization of the pre-produced parts (sizing) and the AM or SM part segment's design by topology optimization. The output of the framework is a ranking of IM part candidates, which, in addition to fulfilling mechanical constraints, are characterized by minimal part mass.

This paper reviews approaches aiming at structural optimization of parts or assemblies based on a combination of structural optimization methods, specifically topology with sizing optimization. Based on the literature review, a discussion of the framework's requirements, such as the robust automated generation of the simulation models based on the design graphs, is carried out. Subsequently, the framework for combined structural optimization of pre-produced parts and AM/SM features for the novel manufacturing approach is presented. The proposed approach relies on embedding a density-based topology optimization (SIMP) method into a sequential design strategy for global optimization (Bayesian optimization). Thus, optimized IM part designs with a minimal mass can be determined with a minimum of evaluations of the computationally intensive objective function. Relevant constraints are taken into account to bound the solution space, for instance, compliance constraint and parameter bounds for the pre-produced parts. Thereby, the simulation models for each iteration are generated script-driven by using parametric CAD software and FEA preprocessor. The proposed optimization framework is demonstrated and discussed by means of a case study.

In summary, this publication presents an optimization framework that enables automated derivation of mass-optimized designs for part candidates for Incremental Manufacturing. The framework involves sizing the proposed pre-produced parts and topology optimization of the additive or subtractive manufactured part segments. In the context of mass customization through (hybrid) AM, the presented method represents an essential building block for automated part candidate detection and concept development.

 

Presenting Author: Julian Redeker Technische Universitat Braunschweig

Authors:

Julian Redeker Technische Universität Braunschweig
Thomas Vietor Technische Universität Braunschweig