PSI - Issue 19

Xavier Hermite et al. / Procedia Structural Integrity 19 (2019) 130–139 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

131

2

1. Introduction

Tests are usually performed to demonstrate reliability target. Depending on the knowledge of failure mechanisms, which could occur in service, and the survival probability to reach at a given confidence level, the number of tests can considerably vary. In addition, wear-out tests may take a lot of time to lead to the failure modes and accelerated testing methods might be used to reduce the whole test time. Combination of failure modes and their effects analysis (FMEA) and high accelerated life testing and stress screening (HALT – HASS) can lead to the main failure mechanisms, whose survival probability may be improved to reach the reliability target. In parallel, experience feedback analysis can highlight the weaknesses of a similar technical solution to consider in the development of a new product. Physical or experimental methods applied to a specific failure mechanism can then allow the identification of controllable service influence factors to be used to reduce the tests duration through high accelerated stress testing methods (HAST). The rise of data science benefits also to the test reduction since data analysis tools are more and more efficient to give added value to the whole test process. Defining the data to collect during the tests and analysing them in real time may give precious information to define the most efficient test strategy minimizing the sample number. All these methods will be presented in this paper. However, no real industrial cases are shown to illustrate the purpose, as the aim of this document is to present a guideline, which should be adapted to each case. It is important to note that parts of this methodology could be applied for reliability characterization, but test reduction is more effective for reliability demonstration, because reaching the target (or in the worst case, to be under the target) is the end point of the test process. In the frame of Reliability, Availability, Maintainability and Safety (RAMS) analysis, FMEA consists in a critical analysis of each component of a system in terms of failure modes and their effects, directly related to the dreaded events previously highlighted in the preliminary risk analysis (PRA). Criticality of a failure mode can be computed considering its detectability, its probability and the gravity of the related dreaded event. The only way to reduce the criticality of a failure mode is to improve its detectability or to reduce its occurrence, improving its availability [1] . Depending on the gravity of the failure, availability improvement will consider either the reliability only (higher level of gravity, where the failure probability should be as lower as possible) or the best combination between reliability and maintainability. This quotation, performed through preliminary assessment in the early steps of the product development, allows a classification of the critical parts of the design. Risk management plan is usually built considering the gravity at first order, and tests prioritization is therefore managed by high reliability demonstration needs. Although FMEA involves components failure modes, tests specification will focus on failure mechanisms leading to them. For instance, a breakage can occur by one-time overload, wear, fatigue, creep, etc. Occurrence probability of a failure mode is also a computation of its probability for each of these mechanisms through an “OR” gate (product of the survival probabilities). At the early steps of the product development, the reliability is unknown and might be characterized to assess the need of redesign. In case of redesign, allocation of reliability target for each failure mechanism is performed to reach the dreaded event criticality target, and new test specifications should be run for reliability demonstration (see Erreur ! Source du renvoi introuvable. ). Number of tests, without considering their required number of samples, can therefore be very important, although FMEA can considerably reduce the test number by prioritization through criticality analysis. 2. Failure modes and their effect analysis (FMEA): experiment design and experience feedback analysis 2.1. Tests prioritization through FMEA