Design and Analysis of Experiments
Design and Analysis of Experiments
| Hi-index | 0.00 |
A simple methodology to evaluate a variety of architecture designs for safety-critical systems is presented. The approach involves including all the appropriate components in a representative fault tree. Evaluating the fault tree results in the cutsets, the system availability, and the system reliability. Next, a Monte Carlo simulation that includes parameters such as mean time to repair and repair time distributions is applied to the fault tree. The parameters included affect the total downtime of the system and impact the system usability. Design of experiments (DOE) is used to track the parameters and the associated levels. The inputs for the Monte Carlo simulation can be adjusted from the elements in the DOE matrix and the results for those variables can be evaluated. Using the DOE methodology allows ranking the results according to their impact on the responses (number of failures, downtime, relative costs, and so on). Using these responses, a figure of merit is assigned to each design. The figure of merit is used to compare the alternative designs and determine the optimal one to implement. This paper uses an example of a simple sensor--controller--actuator system designed with multiple levels of redundancy to illustrate the methodology and the application of the figure of merit to compare the considered architectures.