Design Space Exploration for Fulfilling Product Property Requirements in Additive Manufacturing

Additive manufacturing (AM) provides the ability to manufacture complex parts with desired shape and functionalities directly from digital models. AM has also aided in efficient development of new products by reducing lead time for prototype manufacturing. However, despite such advantages, AM has limitations. One such example is the lack of information regarding the properties of products manufactured using AM. These properties of AM products such as tensile strength, toughness, or surface roughness are typically determined experimentally which is time-consuming and costly. Thus, with the evolution and adoption of industry 4.0, there is a need for computational design tools and methods that can determine end product properties by design space exploration of AM process chain. We present an inverse design method based on exploration of process–structure–property–performance (p–s–p–p) relationships in AM. This method can be utilized by designers to determine different parameters in the manufacturing process chain for achieving AM product properties. It is a multi-stage method where end goals of products are first identified. Design decisions are made at the last step of the process chain to meet these goals, and these decisions are inversely passed to previous stages of the process to meet the end goals. The compromise decision support problem (cDSP) forms the primary mathematical construct of this method. The efficacy of the method is demonstrated by performing design space exploration of processing parameters to meet fused deposition modeling (FDM) product property goals. In the first stage of this method, we establish the forward information flow in the process chain. Next, the property goals are identified. Finally, the cDSP is used for solution space exploration to determine satisficing processing parameters to meet these goals. The primary focus of this work is to demonstrate the utility of this method and associated design constructs in the AM domain. In the future, we plan to extend this method for p–s–p–p design space exploration of different AM processes.