Bayesian Radiation Modeling for Spacecraft Reliability Prediction

In this case study, a proposed Bayesian reliability model is applied to Total Ionizing Dose (TID) field data. TID damage results when high energy particles interact with onboard semiconductor devices and accumulate charge within the silicon and silicon oxide. TID shifts the semiconductor device parameters away from their desired values. The result is a degradation in device performance, circuit performance, device functionality, and an increased risk of non-compliant performance and device functional failure. Additionally, variability in the space weather environment and the nondeterministic nature of the radiation damage make accurate predictions of on-orbit device response a challenge. These and other sources of uncertainty are currently addressed by applying conservative but industry familiar risk mitigation techniques including Radiation Design Margins (RDM), radiation shielding measures, and worst-case circuit analyses. However, these approaches do not propagate and quantify the uncertainty, which leaves them prone to overestimating or underestimating what the sufficient amount of radiation protection is. By leveraging Bayesian methods, the proposed model methodically accounts for this uncertainty in the reliability estimates and reduces it as new data is incorporated.