A Comparison of Pearl String and Horizon Method in Terms of Reliability Demonstration Testing

In the last decade much progress was made using the probability of test success $(P_{ts})$ metric in order to find an optimal test strategy for single stage loads considering Success Run and End of Life tests, see [1]–[3]. In this paper a study is conducted comparing different strategies in terms of $P_{ts}$ values using the Pearl String and Horizon methods for reliability test planning while applying different operating load spectra. The results reveal the Horizon method generally achieves higher $P_{ts}$ values, indicating a superior performance. However, the importance of considering the cost-effectiveness of achieving higher $P_{ts}$ leads to further information to optimize test planning. Thus, the study suggests that the optimal approach to reliability test planning is dependent on the operating load spectra, the test strategy, number of specimen and the associated values for $P_{ts}$ and cost.

1 Introduction and Motivation

Reliability demonstration testing plays a vital role in ensuring the quality and dependability of products, particularly in industries where failure can have severe consequences. Although many simulations can be done during the design development process, in the end a test is needed. However, planning such tests poses challenges due to the unpredictable test effort and the scattering failure times. Additionally, determining the optimal test load levels, particularly for accelerated test conditions in EoL testing, is a complex task. The identification of the optimal test configuration is influenced by different test strategies and parameters, as depicted in Figure 1. A test planning strategy for SRT and non-accelerated EoL tests based on prior knowledge using the Probability of Test Success as an additional optimization criterion was developed by Dazer et. al [1]. Here, refers to the probability of a test configuration to successfully demonstrate a lifetime requirement with confidence under defined boundary conditions [1], [2]. In a reliability DOE, the different test strategies with different sets of parameters were investigated with respect to test cost and . Herzig et. al. [3] extended the approach for accelerated test conditions when using EoL tests with a single-stage field load. The approach was extended to include operating load spectra that replace the single-stage field load, by Benz et al. [10], [11]. With this extension it is possible to evaluate the test strategy in terms of operating load spectra and the resulting accumulated damage. Figure 1:

Introducing an Operating Load Spectrum Generation and Damage Accumulation in the Procedure for Optimal Planning of End of Life Tests