Abstract
This article presents a metric for measuring and improving the robustness of engineering systems. Robustness is the ability of the system to avoid failure modes, even in the presence of realistic noises. The metric is the operating window (OW). The OW is the range in some input noise that produces a fixed failure rate in the failure modes. It can be thought of as an inverse method of measuring the ratio of output variance to input variance. The OW is determined while keeping the magnitude of some aggravating noise—the stressing noise—large and controlled. This allows the OW to be measured quickly. This OW approach has proved in practice to be very superior to the traditional ways of trying to develop reliability.