Industrial Technology - Linked-in Industrial Technology - Twitter Industrial Technology - News Feed
Latest Issue
Diary and Events

Advanced Engineering 2021

NEC Birmingham(B40 1NT)

03/11/2021 - 04/11/2021

Join us in our 12th and most important edition to date, as we invite engineers and management from all (more)

Evaluating the performance of safety control systems

In recent years a culture has arisen in which probabilistic techniques are used to evaluate safety-related performance of control devices. Peter Still from Schneider Electric looks at two of the probabilistic measures used - Mean Time To Failure and Mean Time To dangerous Failure - and explains the difference.

The reliability of a control system component can be broken into three phases - infant, constant and wear out. The infant phase has an initial high failure rate and is typically caused by latent manufacturing defects. This rapidly declines to a lower and reasonably stable failure rate, a phase that can last many years. Finally the failure rate increases as the devices considered reach the end of their useful life - the 'wear-out' phase. 

Most reputable manufacturers will test their products sufficiently to overcome most of the infant failure phase before devices are shipped, and will take a responsible approach to specifying their products' usable lifetime so, provided a suitable maintenance regime is in place to ensure that components are replaced before they wear out, it is the constant failure rate that is of most interest.

Mean Time To Failure (MTTF) is sometimes incorrectly interpreted to mean the time before a device reaches its wear-out phase. However, it is in fact the reciprocal of the average failure rate during the constant failure rate phase.  As this measurement is based on the failure rate across the entire device population, it is of little use when predicting the life of an individual product.

For example, if a device has an MTTF of 100 years, this does not mean that the device can be expected to operate without failure for 100 years; it is a statistical measure relating to the total number of such devices in service. For common failure distributions approximately two-thirds of the population can be expected to have failed after a time equivalent to the MTTF. 

Measure of quality
MTTF is simply a measure that allows device manufacturers to model the quality of their products and is only valid when the device is used in the right application as per the manufacturer's specifications. It is useful as a 'figure of merit', but is of little use to a system designer wishing to estimate how long a single sample of that device will perform in service. In addition variation of the operating conditions can result in disparity of the failure rate by several orders of magnitude. Until the conditions for estimating MTTF are standardised there can be wide differences between the figures published by different manufacturers.

The Mean Time To dangerous Failure (MTTFd) is derived from the MTTF by multiplying the MTTF by a factor that represents the proportion of failures that are not dangerous, so the MTTFd will be much greater than the MTTF. One problem with this is that the component manufacturer does not know which failures are dangerous unless the application is known, so it has to either quote MTTF for more than one failure mode, or make an assumption about which failure mode(s) is (are) dangerous, or state an overall failure mode and leave the system designer to estimate the proportion of failures that are dangerous in a given application. 

A dangerous failure is one that can prevent a safety function from being performed. Again, as with MTTF, it is important to remember that if the component's MTTFd is 100 years, it does not mean each of those components will last this long without a dangerous fault occurring.  

The complexity of today's machine control devices means that it can be impossible to predict the consequences of some failures, so every effort must be made to limit the likelihood of failures occurring.  Therefore, using probabilistic techniques is essential when evaluating the performance of components. In addition, system designers can also help to ensure devices are of the upmost reliability and quality by choosing to specify from a reputable manufacturer that is committed to producing products to the highest specification, exceeding standards.
Download pdf

Other News from Schneider Electric

Schneider Electric boosts next generation TeSys Giga offering with digital innovation

Latest news about Machine Building

Additional Information
Text styles