Failure Mode and Effects Analysis, or FMEA, is a structured engineering method for finding possible failures before they cause harm, downtime, or expensive redesigns. It is used in products, machines, processes, and software controlled systems to make risk visible and comparable. By listing failure modes, their effects, and their causes, engineers can decide which problems deserve attention first.
FMEA matters because it turns vague concerns into a documented action plan.
A typical FMEA rates each failure mode using severity, occurrence, and detection scores, often on a 1 to 10 scale. These ratings are combined into a Risk Priority Number using RPN = S × O × D, where higher values signal higher priority for action. Corrective actions may reduce the chance of failure, improve detection, or limit the effect of the failure.
After actions are taken, engineers recalculate the ratings to check whether the risk has been reduced.
Understanding Engineering: FMEA
A useful FMEA starts with a clear boundary. A team chooses one product function or one step in a process, then describes what normal operation should achieve. For a bicycle brake, the function is to slow the wheel when the rider pulls the lever.
The team then follows a chain from failure mode to local effect, user effect, possible cause, and existing control. A worn cable may create too much slack. The brake pads then move too little.
The rider needs a longer stopping distance. Possible causes include poor material choice, water entering the cable housing, or missed maintenance. This level of detail prevents teams from writing vague entries such as brake problem.
The ratings in an FMEA are estimates, not exact measurements. Different people can give different scores unless the team uses written rating rules. For example, a high severity score should be tied to a defined outcome such as injury, legal noncompliance, or complete loss of a required function.
Occurrence should be based on test data, factory records, field returns, or a justified engineering estimate. Detection needs careful thought. It does not mean that the failure is easy to notice after a customer reports it.
It means that a current check can reliably find the cause or failure before the product reaches the user. A final visual inspection may find a missing screw, but it may not reveal a weak internal electrical connection.
Controls fall into two important groups. Prevention controls make a cause less likely. Examples include selecting a stronger material, adding a guide that stops incorrect assembly, setting a process temperature limit, or training workers with a standard procedure.
Detection controls find trouble after it begins, such as a pressure test, software check, measurement gauge, or automated camera. Prevention is usually better because it removes the source of trouble instead of relying on someone to catch it later. Some effects can be limited even when the cause remains possible.
A fuse can cut power during an electrical fault. A backup sensor can keep a machine in a safe state when the main sensor gives an implausible reading.
A large risk number does not automatically decide every action. A failure with extreme severity may need attention even if it is rare and easy to detect. Many organizations set rules that require action for safety risks, regardless of the combined score.
Teams should write actions that are specific, owned by one person, and linked to a due date. After the change, they need evidence that it works. This may come from repeated tests, production data, inspection results, or a trial run.
Students meet the same reasoning in lab work and everyday design. A bridge made from craft sticks can fail because joints loosen, loads are uneven, or moisture weakens glue.
Listing the causes separately leads to better fixes than simply adding more glue everywhere. Good FMEA work is careful thinking about evidence, assumptions, and the limits of each control.
Key Facts
- FMEA stands for Failure Mode and Effects Analysis.
- A failure mode is the specific way a part, process, or system can fail.
- RPN = S × O × D, where S is severity, O is occurrence, and D is detection.
- Severity measures how serious the effect is if the failure happens.
- Occurrence measures how likely the cause of the failure is to happen.
- Detection measures how likely current controls are to catch the failure before it reaches the user.
Vocabulary
- Failure mode
- A failure mode is the specific way a component, process step, or system function can stop working as intended.
- Effect
- An effect is the result or consequence of a failure mode on the system, user, process, or environment.
- Cause
- A cause is the reason a failure mode may occur, such as wear, contamination, incorrect assembly, or software error.
- Risk Priority Number
- The Risk Priority Number is a score calculated from severity, occurrence, and detection ratings to help rank risks.
- Corrective action
- A corrective action is a design, process, inspection, or control change intended to reduce risk.
Common Mistakes to Avoid
- Treating RPN as the only decision rule is wrong because a low RPN can hide a failure with very high severity that still needs urgent attention.
- Confusing occurrence and detection is wrong because occurrence describes how often the cause happens, while detection describes how likely the controls are to catch it.
- Writing vague failure modes is wrong because phrases like machine fails do not identify the exact part, function, or condition that must be controlled.
- Skipping the updated ratings after corrective action is wrong because the team must verify whether the action actually reduced severity, occurrence, detection, or overall risk.
Practice Questions
- 1 A conveyor motor failure mode has severity S = 8, occurrence O = 4, and detection D = 5. Calculate the RPN.
- 2 A pump seal leak initially has S = 7, O = 6, and D = 4. After adding a better seal material and leak sensor, the ratings become S = 7, O = 3, and D = 2. Calculate the original RPN, the new RPN, and the percent reduction in RPN.
- 3 Two failure modes have the same RPN. One has very high severity but low occurrence, while the other has moderate severity and high occurrence. Explain how an engineering team should decide which one to address first.