Uncertainty analysis, anomaly screening and performance margin for spacecraft thermal design and testing

To ensure success after launch, the thermal behavior of spacecraft is simulated with large distributed-element models, and elaborate ground testing is conducted to emulate the space environment. The nodalized models can be cumbersome and unsuited to estimating statistical uncertainty, without which the engineer must depend on legacy design principles and their own experience to estimate performance risk, set pass/fail criteria, and assign margin. We model the end-to-end sequence of thermal testing, model correlation, and in-flight operation using a combination of lumped-element models, covariance analysis and Monte Carlo simulations. In addition to predicting the range of possible outcomes and key sensitivities, we also show the extent to which a set of thermal balance tests reduces the uncertainty, and how effective they are at screening out unexpected anomalies. The breadth, speed and straightforward implementation of the model is complementary to both the traditional nodalized models, and more recent developments in the field of Uncertainty Quantification. We apply the analysis to NASA’s Near-Earth Object Surveyor mission which has a test campaign comprising three thermal balance tests at different levels of integration, and predict that while testing will provide only a modest reduction in the uncertainty of flight performance, it should be effective for screening out anomalies, such as design errors and workmanship issues.