Modelling uncertainty in reliability growth planning for continuous-use systems utilising disparate source data

ABSTRACT This paper presents an improved approach to reliability growth planning for continuous-use systems through all test and evaluation phases that considers the three sources of uncertainty: the initial reliability of the system at the commencement of testing, the management strategy and the effectiveness of corrective actions implemented to address failure modes. The presented model is a Bayesian extension of the current reliability growth planning model used for continuous-use system reliability growth planning during developmental and acceptance testing. The developed approach shows how the estimated initial system reliability may be combined with management planning parameters to form a prior distribution of uncertainty at reliability growth test entry. The prior distribution is then used to inform individual test phase posterior distributions that reflect expected reliability test outcomes within uncertainty bounds. The demonstrated model is flexible enough to consider disparate data derived from differences in system, environmental and human factor test inputs..