Open Access

The quality of a repairable system can be described using its availability. Typically, a high degree of availability is demanded by the customer. To analyze the availability of a repairable system, the specification of reliability and maintainability are needed. Usually, they are demonstrated based on limited sample sizes, e.g., by analyzing the failure times of a life time test. The evaluation of the test results yields a mean-value distribution of reliability and maintainability as well as its confidence interval. Consequently, the calculated availability based on these inputs also need to be expressed including a confidence interval. In this paper, firstly several approaches to calculate the confidence interval of steady-state availability based on reliability and maintainability are presented. Afterwards, a procedure to investigate and evaluate the quality and accuracy of the confidence intervals calculated with the presented methods is shown. Therefore, the coverage probability as the most common indicator is used. Based on an exemplary parameter study which is performed, the accuracy of the confidence intervals determined with the methods is investigated and evaluated in the case of exponentially distributed failure and repair times. Finally, several hints for an effective availability calculation with confidence intervals are given.