Failure mode identification through clustering analysis

Research has shown that nearly 80% of the costs and problems associated with product design are created during product development, and cost and quality are essentially designed into products during the conceptual design stage. Failure identification procedures (such as failure modes and effects analysis (FMEA), failure modes, effects and criticality analysis (FMECA) and fault tree analysis (FTA)) and design of experiments are currently being used for both quality control and for the detection of potential failure modes during the design stage or post-product launch. Although all of these methods have their own advantages, they do not provide the designer with an indication of the predominant failures that should receive considerable attention while the product is being designed. The work presented here proposes a statistical clustering procedure to identify potential failures in the conceptual design stage. A functional approach, which hypothesizes that similarities exist between different failure modes based on the functionality of the product/component, is employed to identify failure modes. The various steps of the methodology are illustrated using a hypothetical design example. Copyright (C) 2004 John Wiley Sons, Ltd.