Reliability assessment of high-Quality new products with data scarcity

This study concerns the reliability assessment in high-quality new product development in which there is scarcity of data resulting from few or zero failures or the unavailability of failure time information. In such circumstances, traditional reliability assessment methods tend to be inadequate and ineffective. This paper describes a pragmatic approach adopted to address this practical issue. A Bayesian method using reparameterization of the Weibull distribution is proposed, which elicits priors in a meaningful way from technical experts and based on historical data. Unlike existing procedures found in the literature, the method here is developed from the perspective of availability of failure time data. Through a case study from the hard disk drive industry, it is demonstrated that the proposed method can provide an effective and practical solution to the challenging real-life problem. Furthermore, it is shown that failure time information has a significant effect on the inference about the Weibull shape parameter.