Reliability prediction at early functional product development stages

Sammanfattning: A trend among industrial companies is to change business strategies from hardware-oriented to more service-oriented solutions, e.g. functional product business models. Functional products are typically constituted by hardware together with a service support system. For functional product businesses, availability is one critical property upon which the customer and provider must agree. Hence, during functional product development and operation, it is important to enable simulations of functional product availability, which is a function of reliability and maintainability. To more rapidly converge on optimal solutions, simulation-driven design strategies have further been proposed by several researchers. In these strategies, the simulations are used to drive the development rather than simply verify suggested solutions. Measured data or estimated data are often used as input to reliability prediction methods such as fault tree analysis and failure mode and effect analysis. However, when designing new systems, reliability input data may not exist and, hence, prototypes are often manufactured and tested, which requires a significant amount of time.The objective of the work presented in this thesis is to develop a simulation-driven methodology for how to predict hardware reliability, as a part of functional product availability.This methodology shall be applied at early concept stages of the functional product development process, where limited component reliability information exists.The conducted research is based on theories regarding product development methodologies, reliability prediction methods and deterministic simulation methods (e.g. rigid body dynamics). The research presented in this thesis followed a 5-step procedure including as-is study, to-be scenario development (where a future functional business situation is described), method development, method verification (through case studies) and method validation.During the as-is study, existing reliability methods were evaluated according tosuitability in different development stages. Fault tree analysis and probabilistic variation mode and effect analysis were found to give accurate results (since objective input data are used). However, those methods would need further development in order to be used for reliability prediction at early concept stages. Traditional probabilistic variation mode and effect analysis did not result in reliability in terms of a probabilistic quantity. Therefore, the probabilistic variation mode and effect analysis method was further developed and verified through a case study which can be used for probabilistic measures.A method based on deterministic simulations to derive component reliabilityinformation has been further developed. This method takes different variations into account and through a series of simulations, input data for system reliability (such as fault tree analysis and probabilistic variation mode and effect analysis) can be derived.Hence, by combining deterministic and probabilistic simulations, hardware system reliability can be predicted, even when limited component reliability information exists. This hardware reliability prediction method is a critical part of a simulation-driven methodology to be used at early stages of functional product development