Case Study

A closer look
at difficult and
confusing aspects
of failure mode
and effects analysis

8 |
February 2019   LEAN & SIX SIGMA REVIEW
Solve Your
FMEA
Frustrations
T
Topic he concept of failure mode and effects value. The six areas worth discussing are:
Failure mode and effects analysis analysis (FMEA) is rather simple and widely 1. Types of failure modes (FM).
known. In theory, everyone should be able 2. Cause or effect.
Author
Gary G. Jing to do it with some basic training. Yet, in practice, 3. Risk assessment.
a huge variation in quality and competency exists. 4. Types of control.
Email Confusion and various opinions about how to 5. Process vs. design.
[email protected] handle details exist. This article sheds some light 6. Interaction between structures.
on common confusion and disputes.
Types of FM
AIAG and VDA FM refers to the ways failures can occur. It’s the
Because ISO 9001:2015 raised the need for risk- most critical part of FMEA and can be the most
based thinking and management, FMEA gained difficult. The success of FMEA largely hinges on
renewed attention and broader applications. As whether the most critical FMs are appropriately
part of the management system upgrade, the identified. FMs also are the sources to potentially
Automotive Industry Action Group (AIAG) is flood the system and bury the most critical FMs
partnering with the German Association of the with trivial things.
Automotive Industry (VDA) to upgrade and har- In many cases, FMs are identified ad hoc. Some
monize their FMEA practices.1 Dramatic changes structures can help avoid missing important oppor-
are proposed. tunities. ASQ offers FMEA training, which adopted
Other industries are likely to quickly follow a so-called “5+1” structure to help systematically
suit. At this transformational time, revisiting some identify FMs. The “5+1” structure refers to the
typical struggles FMEA endures can add significant numbered paragraphs one and four that follow.
Personally, I think the “5+1” structure is insufficient
and prefer a more expanded list, as follows:
Figure 1  FAULT TREE ANALYSIS EXAMPLE
1. Failing to meet the specification.
++ Complete failure.
++ Too little (partial, uneven or incomplete).
++ Too much (over).
++ Intermittent.
++ Failure over time.
2. Incorrect or inappropriate requirements.
++ Wrong, missing, hidden, unstated or assumed.
3. Unintended use, application or environment.
4. Doing harm to others.
The first three sub-bullets under numbered para-
graph one are obvious and not worth mentioning.
The sub-bullets “Intermittent” and “Failure over time,”

|
LEAN & SIX SIGMA REVIEW   asq.org/pub/sixsigma 9
Case Study > Solve Your FMEA Frustrations

Table 1  NEWLY PROPOSED ACTION PRIORITY LOGIC BY AIAG-VDA*

S O D AP DFMEA ACTION PRIORITY LOGIC S O D AP DFMEA ACTION PRIORITY LOGIC

9-10 6-10 1-10 H High priority due to safety and/or 5-8 4-5 7-10 H High priority due to the loss or
regulatory effects that have a high degradation of an essential or con-
or very high occurrence rating. venience vehicle function that has
a moderate occurrence rating and
9-10 4-5 7-10 H High priority due to safety and/or
high detection rating.
regulatory effects that have a mod-
erate occurrence rating and high 5-8 4-5 5-6 H High priority due to the loss or
detection rating. degradation of an essential or con-
venience vehicle function that has
9-10 4-5 5-6 H High priority due to safety and/
a moderate occurrence rating and
or regulatory effects that have a
moderate detection rating.
moderate occurrence rating and
moderate detection rating. 5-8 4-5 1-4 M Medium priority due to the loss or
degradation of an essential or con-
9-10 4-5 1-4 M Medium priority due to safety and/
venience vehicle function that has
or regulatory effects that have a
a moderate occurrence rating and
moderate occurrence rating and
low detection rating.
low detection rating.
5-8 1-3 7-10 M Medium priority due to the loss or
9-10 1-3 7-10 H High priority due to safety and/or
degradation of an essential or con-
regulatory effects that have a low
venience vehicle function that has
occurrence rating and very high
a low occurrence rating and high
detection rating.
detection rating.
9-10 1-3 5-6 M Medium priority due to safety and/
5-8 1-3 5-6 M Medium priority due to the loss or
or regulatory effects that have a low
degradation of an essential or con-
occurrence rating and moderate
venience vehicle function that has a
detection rating.
low occurrence rating and moder-
9-10 1-3 1-4 L Low priority due to safety and/or ate detection rating.
regulatory effects that have a low
5-8 1-3 1-4 L Low priority due to the loss or
occurrence rating and low detection
degradation of an essential or
rating.
convenience vehicle function that
5-8 8-10 2-10 H High priority due to the loss or has a low occurrence rating and low
degradation of an essential or con- detection rating.
venience vehicle function that has a
2-4 8-10 1-10 H High priority due to perceived
very high occurrence rating.
quality (appearance, sound, haptics)
5-8 6-7 7-10 H High priority due to the loss or with a very high occurrence rating.
degradation of an essential or con-
venience vehicle function that has
2-4 6-7 7-10 H High priority due to perceived
a high occurrence rating and high
quality (appearance, sound, haptics)
detection rating.
with a high occurrence rating and
5-8 6-7 5-6 H High priority due to the loss or high detection rating.
degradation of an essential or
convenience vehicle function that 2-4 6-7 5-6 H High priority due to perceived
has a high occurrence rating and quality (appearance, sound, haptics)
moderate detection rating. with a high occurrence rating and
moderate detection rating.
5-8 6-7 1-4 M Medium priority due to the loss or
degradation of an essential or con- 2-4 6-7 1-4 M Medium priority due to perceived
venience vehicle function that has quality (appearance, sound, haptics)
a high occurrence rating and low with a high occurrence rating and
detection rating. low detection rating.

10 |
February 2019   LEAN & SIX SIGMA REVIEW
 S O D AP DFMEA ACTION PRIORITY LOGIC S O D AP DFMEA ACTION PRIORITY LOGIC

2-4 4-5 7-10 H High priority due to perceived 2-4 1-3 5-6 L Low priority due to perceived
quality (appearance, sound, haptics) quality (appearance, sound, haptics)
with a moderate occurrence rating with a low occurrence rating and
and high detection rating. moderate detection rating.

2-4 4-5 5-6 M Medium priority due to perceived 2-4 1-3 1-4 L Low priority due to perceived
quality (appearance, sound, haptics) quality (appearance, sound, haptics)
with a moderate occurrence rating with a low occurrence rating and
and moderate detection rating. low detection rating.

2-4 4-5 1-4 L Low priority due to perceived 1 1-10 1-10 L Low priority due to no discernible
quality (appearance, sound, haptics) effect.
with a moderate occurrence rating
and low detection rating. *Automotive Industry Action Group (AIAG) and German Association of the
Automotive Industry (VDA), AIAG-VDA FMEA Handbook, first edition, 2018.
2-4 1-3 7-10 M Medium priority due to perceived
AP = action priority
quality (appearance, sound, haptics)
D = detection
with a low occurrence rating and
DFMEA = design failure mode and effects analysis
high detection rating.
O = occurrence
S = severity

as well as numbered paragraph four, “Doing harm Risk assessment

to others,” however, are worth more attention. Also Could the same FM have multiple effects with differ-
worth concern are inappropriate requirements and ent levels of severity (S), occurrence (O) or detection
unintended use, which is the opposite side of “5+1.” (D)? Yes—it’s possible to have any combination of
the three. To simplify the analysis, a popular practice A frequent debate
Cause or effect is to focus on the most severe effect. There is a small is whether an
A frequent debate is whether an identified probability of missing an effect with lower severity identified problem
problem belongs to FM, effect or cause. Indeed, but a higher risk priority number (RPN), but that’s belongs to FM,
the role may change depending on the situa- a risk taken when trading for efficiency. effect or cause.
tion. To better understand the dynamic, some Whether recommended actions can lower the
knowledge of the chain of causation is needed.2 severity of an FM is frequently debated. There are
In short, causes and effects form infinite chains two schools of thought with opposite views, but
of causation. In that chain of causation, an event ultimately what’s affected is Criticality = S x O. To
is an effect if it is upstream and a cause if it keep things simple and avoid confusion, a more
is downstream. common practice is to keep the severity unchanged
The cause and effect role may shift depending and only change the occurrence.
on what link in the chain is being studied. In fact, The new AIAG-VDA harmonizing effort proposes
an FM may have multiple causes and effects. The dramatic changes in risk assessment practice
opposite is also true: A cause or effect may have that will affect the industry significantly. The
multiple associated FMs. Because the FMEA
format isn’t designed to handle complicated
cause and effect relationships, it’s not a primary Figure 2 PREVENTION VS. DETECTION CONTROLS
tool used for diagnosing, finding the root cause
or problem solving.
Failure Failure Failure
Using Figure 1 (p. 9) as an example, is electrical cause mode effects
fault a cause, FM or effect? The answer is it
depends on what level you are looking at:
++ At the car level, it’s a cause.
++ At the subassembly level, it’s an FM. Prevention Detection
(reduces the cause) (counters the mode)
++ At the component level, it’s an effect.

|
LEAN & SIX SIGMA REVIEW   asq.org/pub/sixsigma 11
Case Study > Solve Your FMEA Frustrations

Table 2  PREVENTION VS. DETECTION CONTROLS

DFMEA PFMEA

Prevention controls Typically occur prior to final- Occur prior to manufactur-

izing the design and reduce ing the product
the risk of failures during + Equipment setup
product validation testing requirements
+ Engineering analysis and + Poka-yoke
evaluation: anti-overstress
feature
+ Tolerance stack-up or
statistical analysis
+ Finite element analysis
The goal
Detection controls Detect the failure mode,
typically during product
Occur after manufactur-
ing the product but prior
of FMEA is
validation testing, prior
to releasing design for
to shipping product to
customers
to prevent
customer orders
+ Reliability testing and
+ Product testing
+ Inspection
potential
performance testing con- future
ducted as part of product
validation testing problems.
DFMEA = design failure mode and effects analysis
PFMEA = process failure mode and effects analysis

organizations are eliminating RPN and replacing it with sophisticated

action priority (AP), which employs a comprehensive table detailing
priorities for various combinations of the three elements, as shown in
Table 1 (pp. 10–11). It certainly has pros and cons, and—like any major
change—already has incited heated debates.

Types of control
One of the most confusing struggles is how to appropriately identify
whether existing controls are prevention or detection. As shown in
Figure 2 (p. 11), prevention controls reduce the cause, and detection
controls counter the mode.
There are different items for design FMEA (DFMEA) and process
FMEA (PFMEA), as well as prevention and detection—there is no over-
lap, as shown in Table 2. For example, inspection in production can’t
be used as control for DFMEA. In general:
++ Prevention focuses on preventing failures from occurring and
thus affects occurrence.
++ Detection focuses on preventing effects from occurring and
thus affects detection.

Process vs. design

Another common struggle is determining whether a problem (FM) is
a process issue or design issue. This is related to the scope of FMEA.
Different views on this matter exist. For example, ASQ training adopted
a critical assumption for PFMEA: All incoming components are defect

12 |
February 2019   LEAN & SIX SIGMA REVIEW
free. Therefore, defective incoming parts are out Figure 3  NEWLY PROPOSED FMEA STEPS BY AIAG-VDA
of scope. The thought behind this assumption is
to contain FMEA workload. Yet in my view, if it
will cause problems in your process, it should be AIAG-VDA FMEA proposed first edition
in scope.
The goal of FMEA is to prevent potential future
1 Scope definition
problems, so anything we do should build around 2 Structure analysis
that. Any problems that may jeopardize this goal 3 Function analysis
are in scope. Under this notion, failures and solu-
tions are not limited to the current design—they
4 Failure analysis
may lead to new designs, structures, components 5 Risk analysis A common struggle
or processes. is determining
6 Optimization
Some FMs may belong to both process and whether a problem
design. Many process issues can be addressed Design/process FMEA (FM) is a process
through changing the design, which may be harder issue or design issue.
to do after product release but quite desirable pre- AIAG = Automotive Industry Action Group
release. It’s my opinion that if DFMEA doesn’t lead FMEA = failure mode and effects analysis
VDA = German Association of the Automotive Industry
to design change during the development stage,
it’s not as effective as it can be.

Interaction between (sub) Not all columns on an FMEA form are equally
structures (IBS) important. The most important one is FM, followed REFERENCES
IBS-induced issues are hard to capture. The new by actions, then RPN. Everything else on the form is 1. Automotive Industry Action Group
AIAG-VDA harmonizing effort puts more emphasis secondary to help make better-informed decisions. (AIAG) and the German Association
on structural analysis, as shown in Figure 3, which Focusing resources on the most important of the Automotive Industry (VDA),
AIAG-VDA FMEA Handbook, first
should help. Still, the table format for FMEA isn’t items—new, unique and difficult items, FM and
edition, 2018.
designed to handle a sophisticated relationship action—can significantly improve the efficiency 2. Gary G. Jing, “Flip the Switch,” Quality
between substructures. System FMEA may capture and effectiveness of FMEA. Progress, October 2008, pp. 50-55.
interactions better, but not DFMEA. FMEA is not
a primary tool for investigation.

Key takeaways GARY G. JING is a Master Black Belt and lean Six Sigma
Although FMEA appears to be simple, constant deployment leader, and has most recently worked as a
struggles exist regarding its efficiency and effec- continuous improvement director at CommScope in Shakopee,
tiveness. The problem is worsened when the cost MN. He is on the U.S. delegation to ISO TC 176 and participated
of FMEA actions isn’t considered. in the development of ISO 9000:2015 as the secretary of SC 1/
It’ll be more effective and efficient to systemati- WG 1, which is responsible for the ISO 9000 standard. He earned
cally integrate FMEA with related activities than to a doctorate in industrial engineering from the University of Cincinnati. A
perform FMEA by itself. fellow of ASQ, Jing is an ASQ-certified quality manager and quality engineer.

DILBERT

|
LEAN & SIX SIGMA REVIEW   asq.org/pub/sixsigma 13