←

FMEA in Manufacturing: Method, RPN, PFMEA and MES Data

By Christian Fieg · Last updated: March 2026

What Is FMEA (Failure Mode and Effects Analysis)?

FMEA (Failure Mode and Effects Analysis) is a systematic methodology for identifying potential failure modes in a product or process, assessing the severity of their effects, estimating how often they are likely to occur, and evaluating how well current controls can detect them before they reach the customer. The goal is to prioritize risks and implement preventive actions before failures happen, not after.

FMEA originated in the U.S. military in the late 1940s (MIL-P-1629) and was adopted by NASA and the aerospace industry in the 1960s. The automotive industry made FMEA a mandatory quality tool in the 1990s through the AIAG (Automotive Industry Action Group) standard. Today, FMEA is required by IATF 16949 (automotive quality management), ISO 13485 (medical devices), and widely used across aerospace, electronics, food processing, and general manufacturing.

In manufacturing, FMEA is not a one-time exercise. It is a living document that must be updated whenever production processes change, new failure modes are discovered, or real-world data shows that the original risk assessment was wrong. This is where the connection between FMEA and real-time production data becomes critical: without actual failure frequency and detection data from the shop floor, FMEA ratings remain educated guesses.

Types of FMEA

FMEA type	What it analyzes	When it is performed	Who leads it
DFMEA (Design FMEA)	Potential failure modes in the product design. Focuses on: Will the design function as intended? What happens if a component fails?	During product development, before design release. Updated when design changes occur.	Design engineering team. Inputs from test, manufacturing, and quality engineers.
PFMEA (Process FMEA)	Potential failure modes in the manufacturing process. Focuses on: What can go wrong during production? What process parameters can drift? What if a machine malfunctions?	During process planning, before start of production (SOP). Updated continuously based on production experience.	Manufacturing engineering team. Inputs from production, maintenance, quality, and operators.
MFMEA (Machine FMEA)	Potential failure modes of production equipment. Focuses on: What machine components can fail? What is the impact on product quality and production output?	During machine design or machine qualification. Updated based on maintenance history.	Machine builder or maintenance engineering. Inputs from production and quality.

For manufacturing companies, the PFMEA is the most operationally relevant type because it directly addresses what can go wrong during daily production. The remainder of this article focuses primarily on PFMEA, though the methodology applies to all types.

The 7-Step FMEA Process (AIAG-VDA Standard)

In 2019, AIAG and VDA (Verband der Automobilindustrie) published a harmonized FMEA handbook that replaced the previous separate AIAG and VDA approaches. This standard defines a 7-step process:

Step	Name	What happens	Output
1	Planning and preparation	Define the scope of the FMEA. Identify the product or process to be analyzed. Assemble the cross-functional team. Define boundaries, assumptions, and constraints.	FMEA project plan. Team composition. Scope definition.
2	Structure analysis	Break down the system into its components (DFMEA) or the process into its steps (PFMEA). Create a structure tree or process flow diagram.	Process flow diagram. Structure tree. Clear hierarchy of system/process elements.
3	Function analysis	Define the function of each element and the requirements it must fulfill. Link functions to requirements (what does "good" look like?).	Function tree. Requirements linked to each process step or component.
4	Failure analysis	Identify potential failure modes (what can go wrong), failure effects (what happens if it goes wrong), and failure causes (why does it go wrong). Build failure chains: cause, mode, effect.	Failure network. Cause-mode-effect chains for each process step.
5	Risk analysis	Rate each failure chain on three criteria: Severity (S), Occurrence (O), and Detection (D). Each rated on a scale of 1 to 10. Calculate RPN or determine Action Priority (AP).	Risk ratings for every failure chain. Prioritized list of risks.
6	Optimization	Define preventive and detection actions for high-priority risks. Assign responsibility and target dates. Implement actions.	Action plan with owners, deadlines, and expected impact on S, O, or D ratings.
7	Results documentation	Re-rate risks after actions are implemented. Document the updated FMEA. Communicate results to all stakeholders. Archive as a living document.	Updated FMEA document. Evidence of risk reduction. Lessons learned.

Severity, Occurrence, Detection: The Three Rating Scales

The core of FMEA risk assessment is rating each failure chain on three dimensions. Each dimension uses a scale from 1 (best) to 10 (worst).

Rating dimension	Question it answers	Scale (1 to 10)	Manufacturing example
Severity (S)	How serious is the effect of the failure if it reaches the customer?	1 = no effect. 5 = moderate effect (partial loss of function). 9 = safety issue with warning. 10 = safety issue without warning.	A dimensional defect on a non-visible surface = S3. A weld defect on a structural automotive part = S9.
Occurrence (O)	How often does the failure cause occur?	1 = extremely unlikely (<1 per million). 4 = occasional (1 per 10,000). 7 = frequent (1 per 100). 10 = almost certain (>1 per 10).	A tool breakage that happens once per year = O2. A cycle time deviation that happens every shift = O8.
Detection (D)	How likely is it that current controls will detect the failure before it reaches the customer?	1 = almost certain detection (automatic 100% inspection). 5 = moderate detection (periodic sampling). 10 = no detection (no control exists).	Automatic inline camera inspection = D2. Visual inspection by operator = D6. No inspection for this defect = D10.

RPN vs. Action Priority (AP)

The traditional approach calculates a Risk Priority Number (RPN) = S x O x D. RPN ranges from 1 to 1,000. Higher RPN = higher priority for action. While widely used, RPN has a known weakness: it treats all three dimensions as equally important and does not distinguish between, for example, S=10 x O=1 x D=1 (RPN=10, safety-critical) and S=1 x O=10 x D=1 (RPN=10, cosmetic issue).

The AIAG-VDA 2019 standard introduced Action Priority (AP) as an alternative. AP uses a lookup table that considers the combination of S, O, and D values to assign a priority of High (H), Medium (M), or Low (L). AP ensures that high-severity failure modes always receive higher priority, regardless of occurrence and detection ratings.

PFMEA Example: Injection Molding Process

Process step	Failure mode	Failure effect	Failure cause	S	O	D	RPN
Injection	Short shot (incomplete filling of mold cavity).	Part is dimensionally incomplete. Customer receives non-functional part.	Injection pressure too low. Nozzle blockage. Material viscosity out of spec.	7	4	3	84
Cooling	Warpage (part deformation after ejection).	Part does not meet dimensional tolerances. Assembly issues at customer.	Cooling time too short. Mold temperature uneven. Cooling channel blockage.	6	5	5	150
Ejection	Ejector pin marks on visible surface.	Cosmetic defect. Customer complaint. Sorting action required.	Ejector pin wear. Insufficient ejection force. Part sticking to mold.	4	6	6	144
Material drying	Moisture content above specification.	Splay marks, bubbles, reduced mechanical properties. Field failure risk.	Dryer malfunction. Drying time not monitored. Hopper lid left open.	8	3	7	168

In this example, "Material drying: moisture content above specification" has the highest RPN (168) despite moderate occurrence, because the detection rating is high (D=7, difficult to detect visually) and severity is significant (S=8, reduced mechanical properties can cause field failures). This failure mode would receive priority for improved detection, for example, automatic moisture content measurement or process data monitoring that tracks dryer parameters in real time.

How MES Data Improves FMEA

The biggest weakness of most FMEAs is that the Occurrence and Detection ratings are based on team estimates rather than actual production data. An MES provides the real-world data that transforms FMEA from an opinion-based exercise into an evidence-based risk management tool.

FMEA dimension	Without MES data	With MES data
Occurrence (O)	Team estimates based on experience and memory. "I think this happens about once a month." Subjective, inconsistent across team members, often underestimates true frequency.	Actual failure frequency from production data. Downtime Pareto shows exact count: "Machine alarm #47 occurred 23 times in the last 30 days." Objective, verifiable, traceable.
Detection (D)	Team assumes that controls work as designed. "Our operator checks every 50th part, so detection is good." No data on actual detection effectiveness.	Scrap and rework data correlated with inspection points. If scrap at station N+3 includes defects that should have been caught at station N, the detection control at station N is not working as assumed.
Severity (S)	Based on design specifications and DFMEA. Typically stable. Less affected by production data.	Severity ratings remain primarily design-driven, but MES traceability data can reveal when a "moderate" defect actually causes downstream assembly failures, justifying a higher S rating.
Action verification	Actions are implemented and the team assumes they work. Re-rating is done by consensus without measuring actual impact.	MES data shows whether the action actually reduced failure frequency. Before action: alarm #47 occurred 23 times/month. After action: alarm #47 occurred 4 times/month. Re-rating based on evidence.
FMEA review triggers	FMEA is reviewed on a calendar basis (annually) or when a customer complaint forces a review. Many failure modes are never re-evaluated.	MES alarm trends, scrap rate changes, and new downtime patterns trigger FMEA reviews automatically. If a previously rare failure mode becomes frequent, the data signals the need for a review.

At Neoperl, SYMESTIC's automatic alarm detection and correlation with quality data enabled a structured root cause analysis workflow. The correlation between SPS alarms and quality defects is exactly the type of data that transforms PFMEA reviews from subjective discussions into evidence-based decisions. At Meleghy Automotive, downtime Pareto analysis across 6 plants provides the Occurrence data needed to re-rate PFMEA entries with actual frequency data rather than estimates.

FMEA and Related Quality Methods

Method	Purpose	Relationship to FMEA
Six Sigma (DMAIC)	Data-driven problem-solving methodology for reducing variation and defects in existing processes.	FMEA is used in the Analyze phase of DMAIC to identify potential failure modes contributing to the problem. After improvements, FMEA is updated in the Control phase.
Control Plan	Document that specifies which process parameters and product characteristics must be controlled, how they are measured, and what to do when they are out of specification.	The Control Plan is derived directly from the PFMEA. Every significant failure mode in the PFMEA should have a corresponding control in the Control Plan. They are companion documents.
8D Problem Solving	Structured corrective action process for resolving customer complaints and internal quality issues.	When an 8D investigation reveals a root cause, the corresponding PFMEA should be updated: was this failure mode already listed? Was the Occurrence or Detection rating accurate? If not, the FMEA needs correction.
SPC (Statistical Process Control)	Monitoring process parameters using control charts to detect process drift before defects occur.	SPC is a detection control referenced in the PFMEA. The Detection rating for a failure mode depends on whether SPC is applied to the relevant parameter and whether the control chart is monitored in real time.
OEE	Measures equipment effectiveness: Availability x Performance x Quality.	The Quality factor of OEE captures scrap and rework, which are direct consequences of PFMEA failure modes occurring. Low Quality OEE indicates that high-severity or high-occurrence PFMEA failure modes are active.

Frequently Asked Questions About FMEA

What is the difference between DFMEA and PFMEA?

DFMEA (Design FMEA) analyzes potential failures in the product design itself: Will the material choice withstand the load? Will the geometry allow proper assembly? PFMEA (Process FMEA) analyzes potential failures in the manufacturing process: What if the welding current drifts? What if the operator installs the wrong component? DFMEA is led by design engineering during product development. PFMEA is led by manufacturing engineering during process planning and updated throughout production life.

What is a good RPN threshold for action?

There is no universal RPN threshold. The common practice of "take action if RPN > 100" is not supported by the AIAG-VDA standard. The problem with a fixed threshold is that it ignores severity: an RPN of 90 with S=9, O=2, D=5 (safety-critical) is more important than an RPN of 120 with S=2, O=6, D=10 (cosmetic issue). The AIAG-VDA Action Priority (AP) approach addresses this by using a lookup table that prioritizes based on the combination of S, O, and D, ensuring that high-severity failure modes always receive attention regardless of total RPN.

How often should a PFMEA be reviewed?

A PFMEA should be reviewed whenever: a process change is implemented, a new failure mode is discovered in production, a customer complaint reveals a defect not covered in the FMEA, actual production data shows that Occurrence or Detection ratings are significantly wrong, or a scheduled periodic review is due (typically annually). With MES data providing continuous visibility into failure frequencies and detection effectiveness, PFMEA reviews can shift from calendar-based to event-based, making them more timely and relevant.

Is FMEA mandatory?

For automotive suppliers certified to IATF 16949, FMEA is mandatory. The standard requires both DFMEA and PFMEA as part of the APQP (Advanced Product Quality Planning) process. For medical device manufacturers under ISO 13485, risk analysis (often FMEA-based) is required by ISO 14971. In other industries, FMEA is not legally mandated but is widely adopted as a best practice for risk management.

How does an MES help with FMEA?

An MES provides the production data that makes FMEA ratings evidence-based instead of opinion-based. Specifically: downtime Pareto and alarm frequency data calibrate Occurrence ratings. Scrap and rework data correlated with inspection stations calibrate Detection ratings. Before/after comparisons after corrective actions verify whether the action actually reduced the risk. And trend monitoring triggers FMEA reviews when previously rare failure modes become frequent.