Failure classifications

Device failures result in a specific failure mode, e.g., a transmitter could fail with the signal stuck within the acceptable range. Experience and knowledge of how the device functions is necessary to identify the failure modes. These modes can then be classified as either safe, where the failure causes the device to go toward its safe state, or dangerous, where the failure causes the device to fail to function. In the case of a transmitter, a stuck signal would be considered a dangerous failure. 

Device failures can sometimes be detected by online, automatic diagnostics that notify the plant operator that the device has failed so that compensating measures can be implemented. These failures are classified as detected, leading to the identification of dangerous detected (DD) or safe detected (SD) failures. If online diagnostics are not available, the failure may remain undetected until a process demand occurs or the device is proof tested. These undetected failures may be dangerous undetected (DU) or safe undetected (SU). 

Copyright International Society of Automation Provided by IHS under license with ISA 

No reproduction or networking permitted without license frcm IHS 

Other terms have been used through the years for these failure classifications  

---

The project began with an extensive evaluation of 900 reported incidents involving failures of fixed pipework on chemical and major hazard plant. As part of the analysis a failure classification scheme was developed which considered the chief causes of failures, the possible prevention or recovery mechanism that could have prevented the failure and the underlying cause. The classification scheme is summarized in Figure 2.13. A typical event classification would be... 

In order to develop an actual tool for safety management it does not suffice to stop at the analysis stage of failure classification mentioned above. These classification results have to be translated into proposals for effective preventive and corrective action (see module 6 of the NMMS framework in Chapter 4). To fulfil this purpose a socalled Preliminary Classil icalion/Action Matrix is proposed below (see figure 5.3.),... 

In the previous chapters the purposes of near miss reporting have been outlined and a framework of designing such a safety management tool has been presented. The importance of human behaviour as a dominant factor in incident sequences was stressed by developing a system failure classification scheme largely based on a theoretical model of operator behaviour. Also an overview was given of the organisational factors necessary for a successful implementation of a NMMS. 

It should be noted that lightning is considered a random event. Many failure classification schemes use the term random failure because stress events are generally random. Some failure classification schemes use the term physical failure for the same thing. 

To achieve data of high quaUty, as much as possible shall be collected automatically. Automatic collection is consistent—not depending on human errors— and complete, as far as it is specified and implemented. However, automatic collection is not generally applicable, since some measurements include judgements, for example, failure classification. Manual data collection is based on templates and forms, either on paper or electfonically. 

Failure classification No safety effect Minor Major Hazardous Catastrophic... 

Figure G.1 illustrates the primary and secondary concerns of each failure classification and how they affect the plant operation. The Probability to Fail on Demand Dangerous Undetected (PFDDU) is related to the Safety Integrity Level, which is defined by the H RA findings and documented in the safety requirements specification.

The probability of failure associated with each of the failure classifications shown in Figure G.2 can be defined by equations using simplifying assumptions. The most important assumption is that failures are random (i.e., effectively occurring at equal time intervals over the life of the device or subsystem). First, a... 

Figure G.3 — Venn Diagram showing the probability of failure on demand for each failure classification...

Safe and dangerous failure classifications and rates of the device... 

Fig. 4.56 Pie charts (Baumkuchen-Diagram) for the failure classification (Source ISO 26262,...

Step 1 Review the description of the failure and verify the initial failure classification (if necessary correct it). 

Vehicle Failure Classifications and Defects in Vehicle Parts... 

Chapter 5 is devoted to transportation systems failures. Some of the topics covered in the chapter are mechanical failure-related aviation accidents, vehicle failure classifications, rail defects and weld failures, rail and road tanker failure modes and failure consequences, ship failures and their consequences, and failures in marine environments and microanalysis techniques for failure investigation. Chapter 6 presents a total of 11 mathematical models for performing various types of reliability analysis of transportation systems. 

---