scope:

This Section of the Standard sets out the prescribed method for estimating a lower single sided confidence limit to the reliability of a series system when the available data comprises the attribute sampling test results obtained on each of a set of sub-systems forming the series system.

The lower single sided confidence limit may be regarded as the lowest value that the system reliability could reasonably take on the evidence of the sub-system test data. Conclusions formed in this way require a high chance (measured by the confidence level) of being valid. A confidence level of 95% has been adopted; this confoms to normal statistical practice and ensures a common basis for the quotation of reliability confidence limits.

The method caters for values of the confidence limit down to 0.5 and has been designed to have an accuracy of ±0.005.

The method applies only to the case of a series system as defined in clause 2c.

The reliability of the series system is assumed to be the product of the reliabilities of the sub-systems, ie any interfaces are deemed 100% reliable.

The method is based on the proviso that the sizes of the sub-system samples are equal. If this cannot be met the sampling results obtained in respect of unequal sample sizes can be modified in the manner described in clause 5 at the expense of giving a conservative answer.

The method is based on sub-system sampling by attributes.