Introduction

Note: Nothing in this standard supercedes applicable laws and regulations.

Note: In the event of conflict between the English and domestic language, the English language shall take precedence.

Purpose. This standard provides instructions and guidelines for implementing Degradation Analysis, as part of a Weibull analysis, to determine product reliability. In Degradation Analysis, the performance or condition of each part is measured repeatedly throughout the test, with the time to failure (defined as the performance/condition degrading to some specified allowable limit of acceptability) being predicted for any parts that have not failed by the end of testing. These predicted failure times are included along with any actual failure times in the Weibull analysis. Using these predicted times to failure provides more information to the Weibull analysis than does calling these unfailed parts right censored (or suspended) at the point they were taken off test.

Degradation Analysis may be used for both Development and Validation of a product. It is primarily used in a Test-to-Failure validation plan, where, in effect, it increases the number of failure points, and thus improves the Weibull analysis. Both time and cost can be minimized while still implementing a Test-to-Failure plan, because prediction of future failure times reduces the number of actual failures needed to get a good Weibull analysis and reliability assessment. Degradation Analysis may also be used for a Success Test plan, as the measured performance and predicted lives add valuable information about the product and its capabilities, as well as giving an early warning of impending failures.

Applicability. This standard can be applied to performance metrics that are used as part of a definition of failure in a product reliability requirement. Such failures are often called "soft failures" because there is still some functionality, even though it is at a level that is not considered acceptable. The identification and setting of customer-focused performance metrics is one of the steps in the Design for Six Sigma (DFSS) methodology.

It can also be used with measurements of a part's condition, such as the thickness of the brake lining, which is not a measure of performance directly, but which affects performance when the condition changes enough. For simplicity in the remainder of this specification, both direct measurement of performance, and measurement of a condition that affects performance, will be referred to as "performance."

The following conditions determine its applicability:

a. The performance metric must indicate some aspect of the product's operation, with a definable point separating acceptable from unacceptable performance.

b. The performance must change in a continuous manner during the test, as opposed to a long period with no change followed by a rapid change at the end. (Note that "symptoms" such as noise typically do not meet this condition, so they do not usually make good candidates for Degradation Analysis. Besides not having a set of performance data to extrapolation from, there is also little time savings because the failure occurs soon after the start of the symptoms.)

c. The test content must be able to elicit this performance change.

d. The performance must be objectively measureable during the test. (Note that a subjective rating is not an appropriate measure for Degradation Analysis – besides not being objective, the rating scale is generally not linear.)

Degradation Analysis is primarily applied to components and subsystems being tested in a lab durability test, although it may also be applied to a vehicle-level test (e.g., tire tread depth on a vehicle durability test).

Degradation Analysis can be used for both "normal stress" testing, and for Calibrated Accelerated Life Testing (CALT) where different parts are tested at one of at least three different overstress levels, with the lives at the "normal" stress being predicted using a life vs. stress curve.