A Truncated Sequential Test for Constant Failure Rate

This paper presents a truncated sequential procedure for testing if the failure rate of a device is constant versus a linearly increasing failure rate. A number of items, N, are placed on test and their failure times are recorded sequentially. After each observation one decides either to accept or reject a constant failure rate or to wait for another observation. Simulation under various parametric conditions yields information on the initial choice of N, average sample sizes, and points on the OC curve. Selection of the various parametric conditions is discussed. An example illustrates the use of the test.