Ananya Nandy, Andy Dong, Kosa Goucher-Lambert

J. Mech. Des. Mar 2022, 144(3): 031401, https://doi.org/10.1115/1.4052302

This research empirically questions the meaning of design similarity, a central concept in design activities relevant to several applications, including competitive benchmarking, design-by-analogy, and patent infringement. Specifically, the work focuses on the similarity of functional implementations across products (as opposed to form), using structured functional models that are expressed mathematically. The paper investigates the differences between several representative design similarity measures when applied to real engineered products – a set of energy harvesting devices. The devices apply technologies such as wind, solar, and wave energy harvesting with different functional implementations. The analysis revealed the use of network-based similarity measures, for example, can result in unintuitive “functionally similar” devices that cross the boundaries of human-determined technological categories. While such outcomes may not be ideal for directly transferring functional implementations to new designs, based on the findings, we propose that these types of measures could be useful for discovering “surprising but useful” functional implementations. On the other hand, more widely adopted measures, such as cosine similarity, can be applied when the desired function implementation is clearer. Despite the availability and adoption of many types of similarity measures, this work demonstrates the need for design researchers and practitioners to carefully assess how their chosen measure operationalizes functional similarity, depending on the stage of the design process.