Estimation of the reliability of space nuclear power systems by probabilistic risk assessment techniques

Abstract

A successful space mission depends on the reliable operation of the spacecraft's electrical power system. For payloads requiring high power levels, various designs of space nuclear power systems (SNPS) are available. Designers have conducted limited spacecraft component reliability analysis and full-scale testing of SNPS is impractical. Therefore, a properly-designed reliability analysis, systematically applied, may provide an effective means for making judgments about the relative reliability of competing SNPSS. This work examines the applicability of probabilistic risk assessment (PRA) techniques for estimating SNPS reliability from design studies. The nuclear electric power industry has used PRA techniques to accurately analyze the reliability of complex systems. However, these PRA techniques for nuclear power plants require modifications for SNPS reliability assessment. This study validates these modified PRA techniques by examining the reliability of the SP-I 00 and the Small Ex-core Heat Pipe Thermionic Reactor (SEHPTR). The present analysis focuses on the SNPS failure to produce nominal electrical power. Typical events threatening the reliability of the SNPS will consist of hardware failures, external events, and command errors or software deficiencies. This work involves the following systematic steps for each SNPS: e System familiarization ³Performance of a "failure modes and effects analysis" to deten-nine how the failures of components might cause a system failure ³Construction of system and component fault trees ³Reliability data estimation³Fault tree quantification (using CAFTA'O and UNCERT'O) 'Me reliability data estimation relies on occurrence probabilities for each component failure mode. Various methods for estimating failure rates from existing reliability databases or from engineering approximations were investigated. This work employs the Monte Carlo sampling technique to associate numerical uncertainty levels with the quantitative reliability estimates produced for each SNPS. The quantitative results estimate the reliability of the systems studied as 0.9494 for the SP-100 and 0.9453 for the SEHPTR. The associated error factor is approximately 2.0, corresponding to the system modeling and reliability data uncertainties. Importance measures and sensitivity analyses indicate that the fuel damage, sensor, electrical component, mechanical component, drive, and power conditioning, control, and distribution subsystem failures can be critical to the system's reliability.