Value driven tradespace exploration: A new approach to optimize reliability specification and allocation

摘要

The Value Driven Tradespace Exploration (VDTSE) framework developed in this paper is a new and sophisticated approach to optimizing reliability throughout design, and in so doing dynamically apportion reliability goals to all system elements. The VDTSE framework is an extension of existing approaches that have been successfully used to optimize many design dimensions. This allows (for example) reliability, cost and other design characteristics (such as weight, volume and speed) to be automatically and continually optimized throughout the design process. This represents a substantial improvement on `conventional' approaches to reliability goal setting that involve `fixed' reliability requirements that reflect a single scenario of `satisfactory' performance. This results in a short-sited `binary' approach to reliability value - the system is `satisfactory' if it exceeds the requirement. No value is placed on exceeding requirements, nor is value assigned systems that do not meet strict requirements but may offer more `business' value through other benefits such as reduced cost, weight or volume. For `conventional' approaches that involve specifying reliability requirements, to result in optimal systems the customer needs to have exhaustively analyzed all plausible design configurations, accurately modeled all trends in emerging technology, and put forth a demonstrable requirement that aligns with their analysis. In short, the customer needs to enact the design process before the producer does - a process which is impractical and inefficient. The VDTSE framework outlined herein avoids all these issues. It uses component design characteristics (that include cost and reliability) to establish a tradespace of potential system design solutions. By establishing the concept of `value' to be a function of design characteristics, a Pareto frontier can be identified which contains the set of al locally optimized design solutions. The VDTSE involves an algorithm that rapidly identifies the Pareto frontier from a large number of candidate designs. Finally, this allows the optimum design to be determined (in terms of organizational value) that also involves reliability goals apportioned to individual components and sub-systems.