Root-cause analysis

Failure analysis

Root-cause failure analysis

Typically, an incidence may be attributed to several sources, such as:

Extreme conditions, such as the fault of a hardware component or an operator's slip
A design mistake or slip that enables the undesired behavior
An operator's action that does not match the designer's expectations
Management negligence, such as insufficient training or enforcing safety assurance procedures.

Failure analysis is typically about the sequence of events from a trigger to an incidence, and is used primarily for reactive resilience assurance, namely, for incidence investigation.

Common practices of failure analysis

Common practice for root-cause analysis include FTA, ETA, FMEA, and HAZOP. A comparison of these methods is available in an article by Silvianita et al. (2011). Theoretically, such methods can reveal many kinds of failure modes.

Failure analysis can be top-down (effect => cause) or bottom-up (cause => effect). For example, FTA is top-down, and FMEA is bottom-up.

These techniques are useful in incidence investigation, to understanding the way the system has failed.

Visualizing the failure analysis

The guide proposes to use Scenario-Activity-Situation-Fault (SASF) charts for reactive failure analysis. The use of this chart is demonstrated with the description of the TMI accident analysis ...

The complexity of root-cause analysis

The number of all possible sequences grows exponentially with the number of hazards. In practice, it is huge and the work required to go through all them makes it impractical. This is illustrated in the following chart ...

Therefore, this method can be used only in reactive assurance, for reactive analysis,

Human-factors methods of failure analysis

Failures are also described based on a model of human performance during interaction with the machine, in fuzzy context .

The failure model

The guide is based on a Universal failure model ..., describing common ways of incidence generation ....