Date created: | Last Updated:

Category: Project

Description: Engineering designers are constantly seeking ways to be more innovative, decisive, and informed of emerging technologies in the design of consumer products. Design tools, such as functional decomposition, morphology, and Pugh charts help stimulate the design process. However, many early-design-phase design tools require designers to have experiential or empirical design knowledge; many of these approaches are intractable for use by novice designers or designers with little experience designing for certain new objectives. In contrast to these current tools, using repositories to store product design information can provide additional and extensive design knowledge to the global design community. Using repository data—and resultant data-driven design approaches—in the design of new products can be especially impactful for DfX design objectives such as product sustainability, about which many engineering designers have limited knowledge. In this paper, we discuss the creation of a sustainable design repository – a collection of product data that includes environmental impact information. Through the initialization of a 47-product repository case study, we seek to create data-driven design processes that can influence designers to consider environmental sustainability. We found, for example, that in the first year of a product’s life, 29-64% of the environmental impact occurs during the product’s use phase, and that uncertainty in input data (such as component manufacturing location and disposal method) can significantly contribute to environmental impact variation. The creation of this sustainable design repository highlights the need for the consideration of input uncertainties when conducting environmental impact analysis. Additionally, the repository has also been used in tandem with machine learning to understand design decisions that lead to more sustainable products. This sustainable design repository enables subsequent data-driven design research in that it provides a large dataset on which machine learning approaches can operate.