The Aerospace Corporation1, UCLA Department of Mathematics2,
This paper describes the methodology and results of an integrated hardware and software model of a space observation satellite. The purpose of the analysis was to assess the current status and predict whether readiness for launch could be achieved. The satellite computing architecture is based on a dual redundant spacecraft processor and two sets of dual redundant processors for the payload. The availability model was implemented as a Stochastic Analysis Network using the Mobius tool (University of Illinois at Urbana Champaign). The model includes both hardware and software elements and addresses three categories of recoverable failures and a fourth category of mission terminating failures. Parameters for hardware were taken directly from reliability predictions provided by the contractor. Failure rates for software were empirically derived using the CASRE software reliability growth prediction tool (NASA/JPL). The baseline data were discrepancy and failure reports produced by the contractor during various stages of integration testing. The results showed that the failure rate during this period was stable and that no significant reliability growth for software was achieved. Where data were not available, parametric analyses of the resultant integrated model were used to assess uncertainty limits and confidence intervals.