How reliable are reliability figures?

How reliable are the Reliability Availability Maintainability (RAM) figures that are touted by gas turbine OEMs — as proofs of how well their machines are performing?

RAM statistics are key performance indicators for operators and equipment manufacturers alike. These calculations follow standard methods, such as GADS and IEEE, which are designed to provide meaningful comparison and benchmarking. No doubt, the involvement of third-party data collection has improved the reliability of these figures.

But, extremely good numbers should be considered with caution, especially when a particular fleet is known to exhibit issues that can induce frequent or catastrophic failures. Users should be skeptical about short (not continuous) reporting periods or small and non-representative fleet populations. They should ask: Is the data collection uninterrupted? Is the population statistically representative? Have units been removed from the original population?

There’s room for mischief in how one displays the data. An OEM could potentially cherry pick figures from specific periods and show worthy figures, but the small print has to be read closely to see if the data is not representative.

An angry OEM representative complains about a rival OEM: “Impressive and unrealistic numbers are shown hidden behind advertised ‘third party unbiassed’ data, while showing in very small numbers that the date represents only two quarters. Do they think we are stupid?”

He may have a point.