A knowledge-based product development system in the chemical industry

Abstract

Because of the large search space involved in ingredient formulation for chemical product development, time spent on the determination of appropriate ingredients constitutes a significant portion of the new product development (NPD) time. Case-based reasoning (CBR) is effective in solving ingredient formulation problems by referring to how similar products were formulated. For some chemical products, sensorial properties, such as smoothness and greasiness, are important attributes. Decision makers tend to use fuzzy terms such as “very smooth” and “slightly greasy” to describe those attributes. Solely using CBR is not robust enough to specify their preferences on those attributes and thus the case retrieval results might not be satisfactory. This paper proposes a knowledge-based product development system (KPDS), hybridizing CBR with fuzzy-based analytic hierarchy process (fuzzy-AHP), to support chemical product development. Chemical product attributes are classified into functional product attributes (FPAs) and sensorial product attributes (SPAs). The desired FPAs are firstly used to filter and retrieve similar past NPD cases in the CBR. When calculating the similarity of the cases retrieved, the SPAs are considered and their weights are derived by fuzzy-AHP so as to identify the most suitable case(s) for problem solving. This paper provides a detailed step-by-step procedure to formulate chemical products according to the desired product properties with the use of the KPDS. It will be of value to other researchers and industrial practitioners who are responsible for chemical product development.