If you think your plant has complex reliability problems, what if you had to track them all the way to Mars? That’s what NASA faces, among other outer space reliability challenges that require performance and risk analyses of multiphase missions.
NASA’s extremely complex missions are prone to sudden catastrophic failure if equipment falters or unforeseen events occur. For these situations, NASA trains to expect the unexpected and tests its equipment and systems in extreme conditions as well as develops risk analysis tests to foresee potential problems.
Working with an industry partner, GoldSim Technology Group LLC of Issaquah, Wash., NASA recently developed reliability analysis software that can model complex, highly dynamic systems, taking variations in input parameters and the evolution of the system into account over the course of a mission. The partnerships’ research focused on the insertion of the Mars Reconnaissance Orbiter; reliability analyses of systems with redundant and/or repairable components; optimization analysis of system configuration with respect to cost and reliability; and sensitivity analyses to optimally identify areas of uncertainty reduction or performance enhancement, according to NASA.
GoldSim’s work with NASA was funded through a Small Business Innovation Research (SBIR) grant with the Jet Propulsion Laboratory to develop and demonstrate a new module for the software. An integrated, total-system risk and reliability analysis approach based on dynamic, probabilistic simulation is the result. The Reliability Module added features to the software to facilitate simulation of space missions.
The GoldSim simulation software was originally developed for long-term safety analyses of nuclear waste disposal, and is currently used by the U.S. Department of Energy and the U.S. Nuclear Regulatory Commission, among others. It is used for performance assessment in the nuclear arena, most notably at the Yucca Mountain Project in Nevada, the nation’s first long-term geologic repository for spent nuclear fuel and high-level radioactive waste.
Most prior NASA studies of mission risk and reliability used fault tree analysis, NASA says. Instead of fault trees, the software actually simulates the behavior of mission components over the duration of the mission, explicitly representing how the components evolve over time and the probabilities involved. The intensive analysis computes the possible ways that the mission may fail and their probabilities.
The new approach complements NASA’s systems analysis approaches and is particularly useful for impact studies, risk/cost/benefit analyses, risk/uncertainty analyses, model integration, and model-based design/model-based engineering. Development of the new module of the software marked the successful completion of the NASA-funded research project to demonstrate the suitability of the software for evaluating the risk and reliability of future NASA missions.
The software is highly graphical and extensible, able to quantitatively represent the uncertainty inherent in complex systems, and allows users to create compelling models that are easy to communicate and explain to diverse audiences. Users build a model in an intuitive manner by literally drawing an influence diagram of their system. In a sense, it is like a “visual spreadsheet” that allows users to graphically create and manipulate data and equations. It moves beyond spreadsheets, however, by making it much easier for users to evaluate how systems evolve over time and predict their future behavior.
Engineered systems modeling, environmental systems modeling, and business modeling are the primary applications for the software. The Reliability Module developed through the NASA SBIR grant supports the software’s ability to perform engineering systems analysis, as it consists of predicted reliability metrics for the overall system and for individual components within it. The module also can be used to compute the probability of specific consequences, such as catastrophic failure of the system, to support risk analysis. The software catalogs and analyzes failure scenarios, allowing identification of key sources of unreliability and risk.
“We also think it has a lot of potential in other fields where risk and reliability are important — areas such as defense, automotive, and electronics,” says Ian Miller, president of GoldSim. “It isn’t just a matter of doing risk calculations; it’s important to really understand the sources of risk or unreliability in order to be able to design optimal solutions.”
The software is used to create computer models that simulate the performance of complex systems in business, engineering, and science. In general, the purpose of such models is to simulate the future behavior of a system and then determine what can be done to improve its performance. The Reliability Module was made available to the public in September 2005.
E-mail Managing Editor Ken Schnepf at [email protected]