Sources of Variation in Failure Rates

A failure rate generated from collecting data on a system will be dependent upon all the circumstances under which the system operates. Consequently, the failure rate data should only be used for predictions on a system in which the circumstances are identical. Otherwise, the failure rate applicable to the second system will need to be adjusted. 

Unfortunately, the circumstances of a data collection exercise are rarely adequately described and therefore, any data will be based on some explicit assumptions, some implicit assumptions, and some assumptions that are completely ignored. 

It is important to appreciate that a failure rate is not an intrinsic and immutable property of a piece of equipment, and an engineer involved either in collecting or using data must fully understand the factors that influence failure rate derivation and use. This section discusses many of the circumstances that can create variations in failure rates. 

The various data cells in a taxonomy include a written description of the equipment and a boundary diagram to identify exactly what equipment is included within the cell. Any change in the boundary diagram or deviation from it in failure attribution during data processing will influence the failure rate and its comparability with others. 

The various levels of the taxonomy represent factors that have an impact on failure rate. For example, lined pipe (CCPS taxonomy number 3.2.2) has a level that groups pipe into 0-6 size and over 6 . Unless the pipe size is specified, there is no way of knowing whether a given failure rate came from the 0-6 or the over 6 range. 

---