Issue 50, April 2005

Standards Based Reliability Prediction: Applicability and Usage to Augment RBDs
Part I: Introduction to Standards Based Reliability Prediction and Lambda Predict

Note: This is the first part of a three-part series about standards based reliability prediction and ReliaSoft's newly introduced Lambda Predict software. This part serves as an introduction and discusses the appropriateness of using standards based reliability prediction as well as when it should be avoided. Part II will discuss the major reliability prediction standards. Part III will give an example of how standards based reliability prediction and Lambda Predict can be used to complete a reliability block diagram (RBD) when data are not available.

There are many situations in which you need to make a prediction about a products performance before the product is actually in production, by which point test data or warranty data should become available for analysis. Many companies have development programs that require design engineers to produce designs that would meet a certain reliability goal before the project is allowed to move on to the following phases (building prototypes, pre-manufacturing and full manufacturing), thus avoiding committing employees and big budgets to a product that has not shown satisfactory reliability in the design stage. The design can involve components or subsystems that have not been tested before, have no history of being used in the field by customers, and are not redesigned components or subsystems of existing components with prior histories.

In other cases, companies do not have the capabilities, resources or time to test certain (non-crucial) components/subsystems of a system and need to use some estimates of the failure rate of these components to complete their system reliability analysis.

Lastly, manufacturers are often required to submit reliability predictions based on a specific standard with their bid for a project.

In these situations and other possible scenarios in which failure data are not available, using a standards based reliability prediction tool becomes the best available approach.

The following are a few advantages and disadvantages related to standards based reliability prediction.

Advantages of using standards based reliability prediction:

• It can help you complete your system reliability block diagrams (RBDs) or fault tree analyses (FTAs) when data for certain components/subsystems within the system are not available.
• It is sometimes accepted and/or required by government and/or industry contracts for bidding purposes.

Disadvantages of using standards based reliability prediction:

• Reliance on standards that may not reflect your products actual performance.
• Although standards based reliability prediction addresses prediction under different usage levels and environmental conditions, these conditions may not accurately reflect your products actual application.
• Some of the standards are old and have not followed latest advances in various technologies.
• The end result from such predictions is a constant failure rate estimation that can only be used within the context of an exponential reliability model (i.e. no wearouts, no early failures). This is not necessarily accurate for all components and certainly not for most mechanical components. In addition, certain aspects of reliability analysis, such as preventive maintenance analysis and burn-in analysis, cannot be performed on components/subsystems that follow the exponential distribution.

In conclusion, it is wise to keep in mind that performing reliability assessments based on the results from in-house testing and data pertaining to the performance of a product in the field yields more accurate results than standards based reliability prediction. One should also recognize, however, that standards based reliability prediction plays an important role in many reliability programs across government and industry and can be used effectively under certain scenarios, as discussed in this article. Engineers and analysts should exercise judgment in selecting the most appropriate analysis approach given the task at hand instead of relying completely on either standards based reliability prediction or life data analysis to address their organizations reliability analysis needs.             

ReliaSoft's Lambda Predict standards based reliability prediction software supports all of the major reliability prediction standards (MIL-217, Bellcore/Telcordia, NSWC, RDF 2000 and China 299B) and comes with an extensive collection of pre-defined component libraries. The prediction standards supported by Lambda Predict will be discussed in the next issue of the Reliability HotWire.
 

Copyright 2005 ReliaSoft Corporation, ALL RIGHTS RESERVED