A Short History of Failure Analysis

The Evolution of Failure Analysis

From 1949 to today: which method was born when, what unique mechanism each one added, and why you would choose it. QRM (2005) is the framework that chooses between all the methods below — and it appears on the timeline in its chronological place, not ahead of them.

Timeline: who came first

Ordered by the year organizations actually started working with each method. Under each method — the original diagram and explanation, in the blue frame.

1949

Generic · Military & industry

FMEA

US military standard MIL-P-1629; later adopted by NASA in the Apollo program in the 1960s.

When to use it: you have a defined, finite component or process and want to map every possible failure mode systematically, bottom-up. The base method from which every other method on this timeline was born.

From the original page · FMEA parallels

Every industry that adopted FMEA adapted it to its own constraints, but the core loop stayed the same in all of them — and the 2019 update (further down the timeline) touched only one step of it:

Core FMEA 1 Failure mode 2 Effect 3 Root cause S×O×D Sev·Occ·Det 5 RPN→AP Prioritize 6 Corrective action

The loop all methods share: identify what could break, what would happen as a result and why, rate the risk, and act — then repeat it throughout the product's life.

1959

Food

HACCP

Developed by Pillsbury together with NASA and the US Army, to guarantee completely safe food for astronauts.

When to use it: you have a continuous production line, not a single product — define critical control points in advance and check them in real time, pass/fail, without waiting for a risk score.

From the original page · FMEA parallels

HACCP Codex Alimentarius

Instead of rating every possible failure — set specific stop points on the production line, and measure there in real time.

CCP CCP Receiving Cooling Packing
At a CCP, any deviation from the critical limit is an immediate stop — pass/fail, not a risk score.

The difference from FMEA: binary control at critical points in real time, not an up-front numeric rating of every scenario.

1963

Industrial safety · Chemical processes

HAZOP

Developed at the British company ICI to analyze the safety of chemical plants.

When to use it: you are analyzing a continuous, complex process — piping, pressures, temperatures — and want to check deviations from the design intent systematically, not just list what might break.

From the original page · FMEA parallels

HAZOP · Bow-Tie CCPS · IEC 61882

Causes and barriers on one side, consequences and barriers on the other — all around a single central event.

Causes Consequences
HAZOP builds the causes side with guide words (No / More / Less / Reverse) applied to every process parameter.

The difference from FMEA: not only "what can break" — also which barriers already exist, on both sides of the event.

1993

Automotive

DFMEA · PFMEA

Standardized by AIAG in the US as part of the QS-9000 requirements for automotive suppliers.

When to use them: you are developing a product and its production process in parallel, and need auditable risk documentation between the customer (OEM) and the supplier.

From the original page · FMEA parallels

DFMEA · PFMEA AIAG-VDA · ISO 26262

The automotive industry runs two separate FMEA tracks throughout development — with DRBFM on top, looking only at what changed.

DDesign FMEA — failure modes in the product design itself PProcess FMEA — failure modes in the manufacturing and assembly process
Baseline design After change DRBFM: what exactly changed here?
ASIL (ISO 26262) is derived from Severity × Exposure × Controllability, not from RPN.

The difference from FMEA: D and P are two separate analyses (product vs. process) running side by side; DRBFM adds a delta-focused layer around a single change. Prioritization itself also changed here in 2019: RPN was replaced by Action Priority (AP) — details below ↓.

1996

Aviation

FHA + FMECA

Standardized together in SAE ARP4761 for civil aviation.

When to use them: you need to show quantitatively that system-level safety requirements converge with component failure analysis — a precondition for flight certification.

From the original page · FMEA parallels

FHA + FMECA SAE ARP4761

Two analyses meet in the middle: one goes down from aircraft level, the other goes up from component level.

PSSA FHA — system risk ↓ FMECA — component failure ↑
The severity scale is locked to probability: a catastrophic failure must be less likely than 1 in 10⁹ per flight hour.

The difference from FMEA: not one one-way flow — two analyses in opposite directions that must converge.

2005

Pharmaceuticals

QRM

Published as the ICH Q9 guideline, shared by regulators and industry.

When to use it: it is not another tool competing on this list — it is the framework that decides which tool from the timeline to use, when and why. Its question is not "what might break" but "which tool fits the question I am asking right now": new product development calls for one tool, process validation for another, a market complaint investigation for a third.

From the original page · FMEA parallels

QRM ICH Q9

Not one method — an umbrella framework that picks the right tool for the question being asked, from a toolbox.

QRM FMEA HACCP HAZOP FTA
The choice of tool follows from the life-cycle stage and the question: development, validation, cleaning or a market complaint.

The difference from FMEA: FMEA is one of the tools in the box — QRM is the framework that decides when to use it.

2019

Automotive · Methodology update

AP

The joint AIAG-VDA FMEA handbook, which replaced RPN.

When to use it: instead of RPN, when you need consistent prioritization that is always led by severity — not a product score that can hide a safety risk behind a low number.

From the original page · FMEA parallels

AIAG (USA) and VDA (Germany) published a joint FMEA handbook that replaced both of their previous versions. The main change: dropping RPN as the prioritization measure, and replacing it with a decision matrix that sets a single priority level.

Until 2019RPN — Risk Priority Number

S9 × O3 × D5 = RPN 135 → H High severity → safety risk S5 × O9 × D3 = RPN 135 → M High occurrence → frequent, less severe failure
The same RPN (135) for both failures — but AP ranks them completely differently: H vs. M. That is exactly the masking the change set out to fix.
Mathematical masking — multiplication can give the same result for very different situations
Wrong focus — chasing a high number, not necessarily high severity
No standardization — every organization set its own RPN action threshold
2019 AIAG-VDA

From 2019AP — Action Priority

S O D High S — overrides all H M L
S is always checked first: high severity jumps straight to H, regardless of occurrence or detection.
H — High immediate action required
M — Medium action recommended
L — Low action may not be needed

Adoption in practice: mandatory for automakers and Tier 1/2 suppliers working with Ford, GM, Stellantis, Volkswagen, BMW and Mercedes — as a precondition in PPAP approvals. Gradually spreading to industrial components, electronics and medical devices, but the traditional RPN is still common wherever there is no explicit customer requirement to adopt the new handbook.

Summary table — by birth year

Year Method Industry Framework / standard What's unique
1949 FMEAFailure Mode and Effects Analysis Generic MIL-STD-1629A The reference point: failure → effect → cause → S·O·D → RPN
1959 HACCPHazard Analysis and Critical Control Points Food Codex Alimentarius Binary critical control points along the process line, not a rating
1963 HAZOPHazard and Operability Study Industrial safety CCPS · IEC 61882 Guide words on process deviations, and barriers on both sides of the event (Bow-Tie)
1993 DFMEA · PFMEADesign FMEA · Process FMEA Automotive AIAG-VDA · ISO 26262 Separate D (design) and P (process) tracks; change-focused DRBFM; ASIL from S·E·C
1996 FHA + FMECAFunctional Hazard Assessment + Failure Mode, Effects & Criticality Analysis Aviation SAE ARP4761 Two-way analysis that converges; severity tied by formula to allowed probability
2005 QRMQuality Risk Management Pharmaceuticals ICH Q9 Umbrella framework that picks a tool (including FMEA itself) by the question and life-cycle stage
2019 APAction Priority Automotive AIAG-VDA H/M/L prioritization led by severity, instead of the RPN product

Written by Nitsan Halevi · Practical Quality · Videos · Terms & privacy