What exactly is FMEA, and what does this RPN score tell me?
🎓
FMEA stands for Failure Mode and Effects Analysis. It's a systematic method to find and prioritize potential failures in a product or process. The RPN—Risk Priority Number—is a single score that quantifies the risk level of each failure. Basically, it's the product of three ratings: how severe the failure is (S), how often it happens (O), and how hard it is to detect before it causes harm (D). In this simulator, you can see how changing any of these three scores directly changes the RPN.
🙋
Wait, really? So a high RPN means we should fix that failure first? How do I decide what's "high"?
🎓
Exactly! A higher RPN means a higher risk priority. In practice, teams set a threshold to focus their efforts. For instance, they might decide all failures with an RPN above 100 need immediate action. That's what the "RPN Threshold Filter" slider in this tool is for. Try moving it to 150 and see how the Pareto chart highlights only the most critical failures. This helps you avoid getting overwhelmed by minor issues.
🙋
That makes sense. But how do I get the numbers for Severity, Occurrence, and Detection in the first place? It seems subjective.
🎓
Great question. It's based on standardized rating tables, like the AIAG 5th edition this tool follows. For example, a "Severity" of 10 is for failures that are hazardous without warning. "Occurrence" can be estimated from historical data or simulation. This is where CAE comes in! A finite element analysis (FEA) can predict how often a part might fail under stress, giving you a data-driven "O" score. After you design a fix, you can re-calculate the RPN here to prove your countermeasure reduced the risk.
Physical Model & Key Equations
The core of FMEA risk quantification is the Risk Priority Number (RPN) equation. It multiplies three independent risk factors to produce a score ranging from 1 to 1000.
$$\text{RPN}= S \times O \times D$$
RPN: Risk Priority Number (1 to 1000). S: Severity of the failure effect (1=No effect to 10=Hazardous). O: Probability of Occurrence of the failure cause (1=Extremely unlikely to 10=Almost certain). D: Ability to Detect the failure before it reaches the customer (1=Almost certain detection to 10=Absolute uncertainty).
The effectiveness of a corrective action is measured by comparing the RPN before and after its implementation. The goal is to reduce the RPN, primarily by lowering the Occurrence (O) or improving Detection (D).
This comparison is a key output of the simulator. A significant drop in RPN demonstrates that the proposed design change or process control effectively mitigates the identified risk.
Frequently Asked Questions
No. RPN is calculated as S × O × D, and each evaluation value is an integer from 1 to 10. Therefore, the maximum value is 10 × 10 × 10 = 1000, and the minimum value is 1 × 1 × 1 = 1. The calculation result will never fall outside this range.
This simulator complies with the AIAG FMEA 5th Edition evaluation criteria table. Please select values from 1 to 10 for Severity (S) based on customer impact, Occurrence (O) based on historical data or similar cases, and Detection (D) based on the effectiveness of current controls. Details of each criterion can be checked via the help button within the tool.
The Pareto chart identifies the main failure modes until the cumulative RPN percentage reaches 80%. First, focus on developing countermeasures for the top 20% of failure modes, and after implementing countermeasures, recalculate to check if the RPN has been reduced. If the effect is insufficient, proceed to the next priority item.
The risk matrix color-codes risk areas based on the combination of S and O, but if D (Detection) is extremely low (e.g., D=10), the RPN becomes high. However, even for undetectable failures, if S is low and O is rare, it may be acceptable from a cost-benefit perspective. The tool recommends using both the matrix and RPN for comprehensive judgment.
Real-World Applications
Automotive Safety Systems: Used extensively in developing components like brake systems or airbag controllers. Engineers perform FMEA on electronic control unit (ECU) software and hardware, using RPN to prioritize fixes for potential malfunctions that could lead to injury, directly supporting ISO 26262 functional safety standards.
Aerospace Component Design: Applied to critical flight components like turbine blades or landing gear. Stress analysis from CAE simulations feeds into the Occurrence (O) score. A high RPN for a crack propagation failure mode would trigger a material change or a new inspection protocol before the part is ever manufactured.
Medical Device Manufacturing: Crucial for ensuring the reliability of devices like insulin pumps or pacemakers. FMEA analyzes potential failure modes in both the device itself and the assembly process. A high RPN related to a sterile packaging defect would lead to improved detection methods (lowering D) on the production line.
Consumer Electronics Durability: Used to improve product lifespan and user experience. For a smartphone, an FMEA might identify repeated connector flexing as a failure mode. CAE fatigue analysis sets the O score, and a high RPN justifies a design change to a more robust connector type, which is then validated by a new, lower RPN calculation.
Common Misconceptions and Points to Note
When you start using an RPN simulator, there are a few pitfalls that beginners often stumble into. The first one is "don't be fooled by the number magic". Since RPN is S×O×D, for example, an RPN of 125 with S=5, O=5, D=5 and an RPN of 135 with S=9, O=3, D=5 are numerically close, right? However, the latter is a Severity 9 event, close to "danger without warning". Even if the RPN is slightly lower, the golden rule is to prioritize events with high Severity (S) that involve human life or regulatory violations above all else. The second is "variation in subjective assessment". One engineer might rate "about once a year" as O=4, while another might rate it as O=3. To prevent this, you should concretize the evaluation criteria within your company. For instance, define "O=3: one report in the past 3 years" or "O=4: one report in the past year", and try to base your evaluations as much as possible on objective data (failure reports, probability from CAE analysis results, etc.). The third is "don't misinterpret the meaning of Detection D". D is the "difficulty of detection". If inspection is perfect and failures are found with certainty, then D=1; if detection is completely impossible, then D=10. Be careful not to move the slider based on "ease of detection" by mistake!