Part one of this two-part blog addresses process FMEA, not design FMEA.
Both the way FMEA component analyses are prepared and how they are used will affect their usefulness. FMEA is associated with mechanical equipment; it’s a step-by-step approach aimed at identifying all possible failures in equipment used in a manufacturing process—and “all” is a big word.
Is Failure Mode and Effects Analysis (FMEA) useful? Will it reduce process interruptions? Perhaps, but it’s not the panacea that practitioners think it can be. Things usually break and fail for a reason; it’s rare that they give ample warning and just wear out.
Unfortunately, FMEA does not use “deductive reasoning” where correct and well understood premises lead to truth. Conversely, FMEA is an “inductive reasoning” methodology, meaning some evidence, based on observations and experience, may support the outcome but is probable rather than truth. It’s like going to WebMD to ask why your knee hurts, and it tells you it could be muscle strain, meniscus damage, ligament injury, tendon damage, or the result of an impact—not enough information to be conclusive. Only a very clear understanding of the exact failure mechanism can estimate or improve such determinations of the failure probability. Initial speculation on cause brings uncertainty when FMEA estimates an outcome — it does not always work.
How Is FMEA Related to FMECA?
FMECA, or Failure Modes, Effect and Criticality Analysis, is a common extension to FMEA, which includes an additional estimation of the criticality or risk of the failure mode. Many installations use an alternative in reliability probability numbers (RPN) as an attempt to estimate criticality along with FMEA.
How Is FMEA Used?
FMEA is a common maintenance management process for developing asset maintenance strategies that currently, mostly comprise preventive maintenance inspections, tasks and schedules. An FMEA tool can be as simple as a spreadsheet recording FMEA analyses. However, a library or database of information often contains FMEA analyses focused on the types of failures with learned maintenance resolution activities. Each FMEA library entry typically includes mean-time-between-failure (MTBF) information that is used to define when inspections or maintenance tasks should be conducted in order to prevent asset failures. Some attempt to declare remaining useful life, but such estimates can be specious because they neglect to consider degradation from random process errors. FMEA systems are great when the cause is known precisely and the resolution activity matches that precision.
With the emergence of improved condition-based monitoring, especially IIoT-enabled devices, some organizations attempt to extend the usage of their FMEA libraries to estimate the probable failure mode or cause of a detected anomaly. It’s important to note that anomalies are always deviations from an estimated normal behavior, which is a major issue, and the anomaly alerts always require human intervention and can be fraught with errors.
Are There Limitations to FMEA?
As mentioned earlier, FMEA is a technique to estimate the likelihood of a failure mode based on judgments made by the practitioners; it is a probability, not a truth. If applied rigorously, FMEA can work well for wear-based failures that are known explicitly and can be detected and addressed through inspections and preventive maintenance tasks. A functional limitation of FMEA is that only about 20 percent of all failure modes are wear-based, as reported by ARCweb and others. The remainder are not random but are most likely caused by errant process conditions that can actually damage equipment by taking them outside safety and design parameters.
The deviations can be minuscule and barely measurable, but they can still cause degradation. Typically, definitive maintenance inspection techniques cannot assess process issues that can induce degradation. As a result, many abnormal conditions that deserve attention are often undetected or ignored as random process events, and assets often run to failure.
The consequence is that at best, FMEA libraries may help in addressing a portion of the 20 percent or so wear-n-tear issues, but not damage causing process conditions. Additionally, and often overlooked, the FMEA library technique usually only recognizes issues too late, when damage is already done. At this juncture, an inspection and analysis may determine the resulting wear-based failure mode and corrective action required, but fails to recognize the reason why it happened at all—the root cause. When you can detect and understand the root cause, earlier intervention and potential process adjustments can result in less or no damage, or for inevitable failures, time to plan, either way resulting in fewer production interruptions.
See FMEA in action with our APM solutions.