Limitations of Failure Modes and Effects Analysis
Although Failure Modes and Effects Analysis (FMEA) is highly effective in analyzing various system failure modes, this analysis technique has limitations.
Examination of human error is limited. A traditional FMEA uses potential equipment failures as the basis for analysis. All of the questions focus on how equipment functional failures can occur. A typical FMEA addresses potential human errors only to the extent that human errors produce equipment failures of interest. Human operations that do not cause equipment failures are often overlooked.
Focus is on single-event initiators of problems. A traditional FMEA tries to predict the potential effects of specific equipment failures. These equipment failures are generally analyzed one by one, which means that important combinations of equipment failures may be overlooked.
Examination of external influences is limited. A typical FMEA addresses potential external influences (environmental conditions, system contamination, external impacts, etc.) only to the extent that these events produce equipment failures of interest. External influences that directly affect safety are often overlooked in an FMEA if they do not cause equipment failures.
Results depend on the mode of operation. The effects of certain equipment failure modes often vary widely, depending on the mode of system operation. For example, a fuel transfer mechanism is of little importance while the fueling station is not operating. A single FMEA generally accounts for possible effects of equipment failures only during one mode of operation or a few closely related modes of operation. More than one FMEA may, therefore, be necessary for a system with multiple modes of operation.
The process prioritizes or ranks risk but does not address it and is not fully quantitative. This action alone does not correct the failure mode. Additional actions are required to mitigate identified risk.
Outputs are only as good as the inputs. Identifying failure modes is a brainstorming activity. Any information that is omitted can cause important failure modes to go unnoticed.
Methods of rating failure modes are not consistent. Assigning numbers to severity and occurrence to failure modes is completely subjective and depends on the user. Therefore, an analysis from different sources may not reflect the same risk priority numbers for the same process.