A reliability modeling framework for the hard disk drive development process

Abstract

Motivated by the fact that the major causes of catastrophic failure in micro hard disk drives are mostly induced by the presence of particles, a new particle-induced failure susceptibility metric, called the Cumulative Particle Counts (CPC), is proposed for managing reliability risk in a fast-paced hard disk drive product development process. This work is thought to represent the first successful attempt to predict particle-induced failure through an accelerated testing framework which leverages on existing streams of research for both particle-injection-based and inherent-particle-generation laboratory experiments to produce a practical reliability prediction framework. In particular, a new testing technique that injects particles into hard disk drives so as to increase the susceptibility of failure is introduced. The experimental results are then analyzed through a proposed framework which comprises the modeling of a CPC-to-failure distribution. The framework also requires the estimation of the growth curve for the CPC in a prime hard disk drive under normal operating conditions without particle injection. Both parametric and non-parametric inferences are presented for the estimation of the CPC growth curve. Statistical inferential procedures are developed in relation to a proposed non-linear CPC growth curve with a change-point. Finally, two applications of the framework to design selection during an actual hard disk drive development project and the subsequent assessment of reliability growth are discussed.

Keywords:

• Reliability improvement
• accelerated testing
• generalized non-linear models
• change-point
• monotone LOWESS
• hard disk drives

Acknowledgement

The authors would like to thank the three anonymous referees and Professor David Coit for their valuable suggestions and inputs on this work.

Notes

¹ This transition function adheres to the three conditions defined in Bacon and Watts (1971). The emphasis here is on the analysis framework and, thus, the particular form is not of great importance. Goodness-of-fit tests based on the likelihood ratio of the proposed models (see Table 4), fitted plot (see Fig. 7) and residual plots (see Fig. 8) can be used to assess the suitability of the model.

²The linearized model to evaluate ₁ ⁰ and ₂ ⁰ is ln (f(t)) − ln (trn( _τ ⁰ − t)) = β₁ + β₂ ln t. This model is fitted with data before the change-point. The linear model for evaluating ₄ ⁰ is f(t) = e ^β₃ + e ^β₄ (t − _τ ⁰ ₄ ⁰ is obtained by taking the natural logarithm of the slope of this function.