A method defining product dependability requirements and specifying input data to facilitate simulation of dependability

摘要

The dissertation focuses on two areas in the early stages of the product design process. First, a definition of dependability requirements that will guide the product design process. Second, a method for assessing that a proposed design solution will fulfill the numeric requirements set for its dependability. In the research the concepts and mathematics to derive a product's dependability requirements from demanded quality are developed. The allocation of dependability requirements to the product's functional entities is based on the function's complexity from the technical standpoint and importance from the customer's perspective. The dissertation also presents a method to refine input data from the dependability data available in order to simulate the dependability of a proposed design solution. The failure rate and the repair time of the object under design review are to be seen as a random quantity; their precise value cannot be known in advance. To the part failure rate and repair time data obtained from the various data sources (plant-specific, generic, and expert judgment) are applied distribution functions, which are combined according to the relative reliability of the data. The parameters of the probability distributions applied in simulation are determined by the (combined) part failure rate and repair time distribution functions. The applicability of the methods developed here to product design has been tested in numerous industry product development and engineering departments. In addition, the accuracy of allocation has been tested through extensive numerical experiments. The results confirm that the methods can be used to verify at an early stage of the design process whether the proposed design solution meets the dependability goals set for it. In addition the methods make it easy to demonstrate the relationship between needs set for the product and known technical solutions. This connection is important in order to avoid promising something that can not be achieved or whose achievement will come too expensive. The results also show that investment in the education of engineering designers is absolutely necessary for the future in order that product reliability and maintainability can be incorporated into the design by the initial designer. The companies should also develop failure related data collection methods in order to serve the product design process.