Towards cost-effective reliability through visualization of the reliability option space

In planning a complex system's development there can be many options to improve its reliability. Typically their sum total cost exceeds the budget available, so it is necessary to select judiciously from among them. Reliability models can be employed to calculate the cost and reliability implications of a candidate selection. However, there will typically be many such candidate selections, so we employ heuristic search techniques to explore reliability cost-benefit spaces, and visualization to present the results of these machine-conducted searches back to the expert designers to assist them in their decision making. This helps them understand (1) the overall cost-benefit trade space (how much reliability can be attained for a given level of expenditure, and how much additional reliability can be had for a modest increase in expenditure), and (2) the contribution of individual options (when an individual option (e.g., a test) contributes cost-effectively to the overall reliability). The net result is that expert human designers can make cost and benefit informed decisions on attainment of reliability. This approach is demonstrated on data drawn from planning the developments of advanced technologies for spacecraft.