Product classification to support approximate life-cycle assessment of design concepts

Abstract Many companies are beginning to change the way they develop products due to increasing awareness of sustainable development. Designers, who play a key role in product development, are being asked to incorporate environmental criteria into the design process. The need for analytically based conceptual design methods for integrated life-cycle assessment (LCA) has motivated the development of an approximate life-cycle assessment concept based upon learning algorithms. Although preliminary tests on general approximate models showed promise, it was observed that grouping products to create specialized learning surrogate LCA models for different classes of products might further improve results. This paper presents work to develop an automated classification system to support the specialization of surrogate LCA models for different groups of products. Hierarchical clustering is used to guide a systematic identification of product groups based upon environmental categories. These groupings are then used to create automated classification schemes using the C4.5 decision tree algorithm. Although further data are needed to induce good generalization performance, resulting product classification systems are considered to be a viable approach to support specialized learning surrogate LCA models for different classes of products.