Estimation of failure rate of consumer products based on components failure rate and its product safety assessment

Abstract

We have proposed a method of product safety assessment by using hazard based FTA in our previous study. In the study, a risk level for a product was calculated by using FTA, and introduced into a manufacturer management system which set criteria for a design and shipment stage in manufacturing. A risk level for a product has a certain distribution in the process of manufacturing due to components quality and their reliability distribution. It is useful to find the probability distribution form of a risk level for a product so that a risk level can be used as criteria of design quality target and shipment condition in a manufacturing management system. In this paper we focus on a failure rate for a component of a product which is the dominant cause of unsafe incidents of the product. The NITE database of incident reports was analyzed for an electric fan to calculate a component failure rate. The failure rate is composed of an accidental and wear-out failure modes. Assuming that the probability distribution is identical at the crossing point of two modes, we estimate a failure rate at the crossing point. We compare the distribution with the normal, Weibull, and log-normal distribution by correlation coefficient and choose the distribution with the maximum correlation. Furthermore we apply the probability distribution to the failure rate calculated by IEC/TR62380 and find that the probability of failure rate for a standard environment is 50% and severe environment failure rate becomes 70% of probability of the distribution that is obtained by the incident database analysis. The knowledge of probability distribution can help to set a risk level standard as criteria for a management system according to an environment of product usage.