Interval estimation of availability of a series system

For a system that consists of n independent subsystems connected in series, based on a sample of complete periods for each subsystem, an approximate interval estimate of the steady-state availability of the system is presented for three cases: when each lifetime and repair-time has an exponential distribution; when each lifetime has an exponential distribution and each repair-time has a log-normal distribution; and the case where each lifetime and repair-time has a log-normal distribution. For a large system it often costs less to test subsystems than to test the system. Consequently, it is more economical to estimate the steady-state availability of the system by using the data from each subsystem. Moreover, it is time-saving since subsystems can be tested simultaneously prior to assembly of the system. Numerical examples show good confidence bounds for the steady-state availability