J Fail. Anal. and Preven.
(2007) 7:321–328
DOI 10.1007/s11668-007-9060-2
C Æ A Æ S Æ E H Æ I Æ S Æ T Æ O
Utility Priority Number Evaluation for FMEA
Jih Kuang Chen
Submitted: 18 January 2007 / in revised form: 10 August 2007 / Published online: 11 September 2007

ASM International 2007
Abstract Traditionally, decisions on how to improve an
operation are based on risk priority number (RPN) in the
failure mode and effects analysis (FMEA). Many scholars
questioned the RPN method and proposed some new
methods to improve the decision process, but these meth-
ods are only measuring from the risks viewpoint while
ignoring the importance of corrective actions. The correc-
tive actions may be interdependent; hence, if the
implementation of corrective actions is in proper order,
selection may maximize the improvement effect, bring
favorable results in the shortest times, and provide the
lowest cost. This study aims to evaluate the structure of
hierarchy and interdependence of corrective action by
interpretive structural model (ISM), then to calculate the
weight of a corrective action through the analytic network
process (ANP), then to combine the utility of corrective
actions and make a decision on improvement priority order
of FMEA by utility priority number (UPN). Finally, it
verifies the feasibility and effectiveness of this method by
application to a case study.
Keywords RPN
FMEA
ISM
ANP
Introduction
Failure mode and effects analysis (FMEA) is a design
and analysis technology that is directly applicable to
failure and prevention. The technology provides a
J. K. Chen (&)
Institute of Technology Management, Chung Hua University,
No. 707 Sec. 2 Wu Fu Rd, Hsin-Chu City 300, Taiwan, Republic
of China
e-mail: pony@qualinet.tw
Æ R Æ Y
structured systematic identification of the potential failure
modes in design or manufacturing, then, by studying the
impact of failure to the system, provides a qualitative
evaluation of the necessary corrective actions by focusing
on the problems affecting systematic reliability. FMEA
has been widely adopted and has become standard prac-
tice in Japanese, American, and European manufacturing
companies [1]. Stamatis [2] pointed out that FMEA is
used in the electromechanical, semiconductor, and medi-
cal device industries and for computer hardware and
software. Onodera [3] investigated about 100 FMEA
applications in various industries in Japan and found that
the FMEA is successfully used in the areas of electronics,
automobiles, consumer products, electrical generating
power plant, building and road construction, telecommu-
nications, and so forth. In addition, it is widely used in
service industry; for example, Linton [4] applied FMEA
to E-Commerce; Reiling et al. [5] applied it to medical
errors; Shahin [6] integrated the Kano mode and then
applied it to travel. Even some special fields, such as
preventing medical accidents [7], environment manage-
ment systems [8], and supplier development [9], have
used FMEA.
Traditionally, using FMEA to improve a system/
technology makes the improvements by addressing prob-
lems in order from bigger risk priority number (RPN) to
smaller ones. However, Gilchrist [10] and many other
scholars questioned the RPN method and proposed new
methods to prioritize applications such as criticality score
evaluate, level of risk, critical analysis and matrix
method, and so forth. Several new methods were pro-
posed in recent years; however, these methods still
measure from a risk viewpoint, while ignoring the
implementation of corrective actions. Corrective actions
may be not independent and may have the relationship of
123
322
hierarchy or even interdependence. For example, to carry
out the corrective action A, one may be wise to start by
implementing action B and then action C before A is
done. The implementation of action B may affect the
previous setting of action A; thus B should be accom-
plished first. Further, the implementation of the corrective
actions in proper order not only maximizes the
improvement effect but also brings a favorable result in
the shortest time at the lowest costs. Therefore, when
performing a FMEA, in addition to the measurement of
risks, it is important to involve the utility of potential
corrective actions.
Against this background, this study aims to evaluate
the hierarchy and interdependence actions by applying the
interpretive structural model (ISM), then calculating the
weight of corrective action through analytic network
process (ANP) and then determining the improvement
priority order by utility priority number (UPN). Finally,
the study verifies the feasibility and effectiveness of the
method through the case of surface mounting technology
(SMT) process improvement in an electronic assembly
plant.
Evaluation of Risk Priority Number
Conventional Priority Evaluation
Traditionally in FMEA, the RPN is used to conduct the risk
assessment. This process is widely adopted. For example,
the Potential Failure Mode and Effects Analysis Manual
(AIAG, 1995) section QS-9000 [11] shows the risk factors
as:
• Severity (S): Result generated from failure
• Occurrence (O): Opportunity or probability of a failure
• Detection (D): Opportunity for an unidentified failure
because of the difficulty of defect detection
The three factors are to be scored from 1 to 10 on the
basis of degree. Risk priority number (RPN) is the product
of occurrence, detection, and severity, which can be
expressed as:
RPN ¼ S  O  D
RPN is the selection foundation for improvement; the
larger RPN represents more potential for improvement.
Corrective action is taken by the relevant department
beginning with the department having the greatest
contribution to risk, as shown by the RPN. RPN should
be recalculated after corrections are made to see if the risks
have gone down and to check the efficiency of the
corrective action for each failure contributor.
123
J Fail. Anal. and Preven. (2007) 7:321–328
Argument of RPN
Ben-Daya & Raouf [12] advocated giving the occurrence
factor the most weight in a risk priority number formula.
They argued that the occurrence factor is the most impor-
tant because it affects the likelihood of a fault reaching the
customer. Kara-Zaitri & Fleming [13] argued that the risk
priority number method of FMEA is intrinsically sub-
jective because guidelines for rating severity, occurrence,
and detection vary from one institution to another. Addi-
tionally, the same risk priority number can be obtained
using a number of different combinations of severity,
occurrence, and detection factors.
Bowles [14] pointed out that the FMEA scales for
severity and detection are basically only qualitative. For
example, a rank 8 severity may not be twice as severe as a
rank 4 severity. Bowles stated that when the severity,
detection, and occurrence are multiplied together to form
the risk priority number, the ratings are treated as if they
have been quantified objectively. Sankar and Prabhu [15]
noted that the distortion is compounded by the nonlinear
nature of the individual ranking scales. Subjectivity in the
ranking scales adds inherent inconsistency. Bowles [16]
compared four types of measurement scales–nominal
measurements, ordinal measurements, equal interval mea-
surements, and ratio measurements—and suggested the
RPN should be dropped and an entirely different prioriti-
zation technique should be used.
Methodology
Interpretive structural model (ISM) is a complicated
system methodology proposed by Warfield [17]. ISM is a
framework of hierarchy of a group of factors in a known
order, which are adopted to FMEA by using the following
steps:
S1 S2
S3 S4
Fig. 1 Binary relation
 Table 1 FMEA report; failure mode effect, broken circuit or poor reliability
Failure Mode Severity Cause Occurrence Current Control Method Detection RPN

Open joint 8 Damaged or bent lead in welding flux 2 Material incoming inspection 2 32
Excessive intermetallic mixture 5 Material incoming inspection 4 160
Unclear or incorrect upper and bottom temperature mark 7 Set according to SOP parameter 5 280
J Fail. Anal. and Preven. (2007) 7:321–328

Unusual reflow oven temperature 2 Set according to SOP parameter 2 32

Decomposition flux of incomplete air discharge 6 Set according to SOP parameter 4 192
or chemical characterization of adhesives
Air sinks phenomenon 5 Operator minimizes the air access according to SOP 6 240
Insufficient welding flux 8 Operator controls the welding flux amount 7 448
according to the size of accessory pad
Improper control of cooling speed 2 Operator controls cooling speed according to SOP 2 32
Suction of welding flux into via hole 2 Design according to the experience of engineering 3 48
Over small welding spot length 3 Design according to the experience of engineering 3 72
Welding spot distortion 6 Twisted or bent board 3 Material incoming inspection 2 36
Over large wave soldering height 5 Set according to SOP parameter 5 150
Uneven wave soldering 2 Set according to SOP parameter 3 36
Unusual welding temperature 6 Set according to SOP parameter 6 216
Uneven distribution of weight 2 Set according to SOP parameter 2 24
Repeated welding 3 Operator controls according to SOP 3 54
Wrong board position 4 Operator controls the position according to SOP 7 168
Over low speed of conveyor 3 Operator controls the speed according to SOP 3 54
Over large plane around the components 6 Design according to the experience of engineering 5 180
Poor fixers or tray design 3 Design according to the experience of engineering 3 54
Wrong proportion of hole position and lead bound 2 Design according to the experience of engineering 3 36
Over large unfilled hole 3 Design according to the experience of engineering 3 54
323

123
324 J Fail. Anal. and Preven. (2007) 7:321–328

Table 2 Priority and corrective actions

RPN Priority Corrective Action No.

160 8 Choose high-flow alloy compound and control the formation of intermetallic S1
280 2 Clearly mark the upper and bottom temperature limit of the welding flux S2
(lead: 180 to *230 C; lead-free 180 to *210 C)
192 5 Optimize the heating curve to enable welding flux; S3
adhesive and flux produce chemical reaction and form
240 3 Allow enough time for the existence of melting and dissolving states S4
448 1 Clearly mark the size of welding spot and required amount of welding flux S5
150 9 Maintain correct wave soldering height (check the quantity S6
of stannum, add stannum when necessary)
216 4 Set the reflow oven temperature limit according to the property S7
of welding flux (lead: 180 to *230 C; lead-free 180 to *210 C)
168 7 Place equipment board according to its heat-resistance capacity S8
180 6 Avoid designing large-scale plane around components (surrounding clearing 3 mm) S9

Fig. 2 The relation of orders of the corrective action
1. Analyze the binary relation of corrective actions:
If there are n corrective actions, its set is expressed by S;
therefore,
• S = {s1, s2, …, sn}.
• (si,sj) is the ordered pair of si and sj.
• Cartesian product is: S · S = {(si,sj) | si,sj 2 S; 8, i,j}.
• The relationship of factors in S is defined as binary
relation.
• R refers to the set consists of the ordered pair (si,sj)
meeting the binary relation in product set S · S; that
is, R is a partial set of S · S; R  S · S.
• If the ordered pair (si,sj) 2 R then si is related to sj; si
R sj, 8si,sj 2 S. If the ordered pair (si,sj) 62 R then the
 j, 8si,sj 2 S.
factor si is irrelevant to sj; siRs
• Each factor in the set S can be regarded as a node, for
which a solution can be found by graph theory. If two
nodes are related, they can be linked by direct line as
shown in Fig. 1.
2. Adjacency matrix: After comparison of binary relation R
in Set S with all ordered pairs, the binary matrix: A of
n · n may be evaluated. The matrix A is also called as the
adjacency
  matrix of the binary relation R, where:
A ¼ aij nn
 j.
where: aij = 1, if siRs j; aij = 0, if siRs
3. Reachability matrix:
• Since each node can reach its own node, add
adjacency matrix A to identity matrix (I): N = A + I.
• N is called as element connection matrix, the path length
0 or 1 can be used to express its reachability. Then
evaluate the power matrix of N by Boolean operation,
until it meets N = N2 = N3 = … = Nr–1 = Nr = M.
• M is called as reachability matrix, which is of
transitive relation with element connection matrix N.
If M(si,sj) = 1, it means there is path existing from
node si to sj; If M(si,sj) = 0, it means there is no path
existing from node si to sj.
4. Expression of hierarchy: In all the figures got from the
reachability matrix M, all the nodes can be concluded as:
• Adjacency reachability set (R): R {sj|sj 2 S,M
(si,sj) = 1}.
• Adjacency antecedent set (A): A {sj|sj 2 S,M
(sj,si) = 1}.
• If n factors in Set S meets the condition: R(si) \
A(si) = R(si), 8 si 2 S. If so, draw out and list si in the
same hierarchy, eliminate it from the reachable
matrix M, and repeatedly evaluate adjacency reach-
ability set R and adjacency antecedent set A until all
the factors are drawn out. Finally, establish the
hierarchy structure according to it.
Based on the hierarchy structure, one can thus calculate
the weights of corrective action by ANP method. ANP was
put forward by Saaty [18], which differs from AHP [19] in
that it allows the inner dependence within cluster and outer
dependence among clusters. It provides a complete

123
J Fail. Anal. and Preven. (2007) 7:321–328 325

Fig. 3 Obtain reachability

matrix M with Boolean
operation

Fig. 4 Repeatedly evaluate

adjacency reachability set R and
adjacency antecedent set A until
all the factors are drawn out

123
326 J Fail. Anal. and Preven. (2007) 7:321–328

• The first part is control hierarchy, which refers to the

Fig. 5 Hierarchy structure
structure; thus, researchers may find the interactions
between elements and clusters from problems, then deduce
the priority and proportion of each scheme. The ANP
method includes two parts:
network relationship of guideline and subguideline,
influencing the internal relationship of systems.
• The second part refers to the network relationship
between elements and clusters.
Hence, one first establishes pairwise comparison matrix
of corrective action, then the eigenvector of each pairwise
comparison matrix shall be calculated, usually by NGM
method (normalization of the geometric mean of the rows).
Then the eigenvectors of each pair comparison matrix are
grouped into a supermatrix, but it may be the unweighted
supermatrix. It should be transformed into weighted su-
permatrix; that is, the sum of any column shall be 1.
Multiplication is performed several times until the extreme
value is fixed, that is, limiting supermatrix, then stop.
Finally, all the weights would be obtained, and they stand

Fig. 6 Supermatrix structure

Fig. 7 Weighted supermatrix

123
J Fail. Anal. and Preven. (2007) 7:321–328 327

Table 3 Comparison of priority order

RPN Old No. of Weight UPN New
Priority Corrective Priority
Action

160 8 S1 0.6 96 1
280 2 S2 0.11 30.8 7
192 5 S3 0.2 38.4 6
240 3 S4 0.2 48 4
448 1 S5 0.18 80.64 2
150 9 S6 0.05 7.5 9
216 4 S7 0.2 43.2 5
168 7 S8 0.1 16.8 8
180 6 S9 0.4 72 3
Fig. 8 Constringent supermatrix after multiplying the weighted
supermatrix until its constringency
further calculate the adjacency reachability (R) and adja-
cency antecedent set (A) of M, and draw out si if they
for the utility of corrective actions. Further, the corrective meet R(si) \ A(si) = R(si), and list them in the same layer.
actions can be used to make decisions on improvement Then, one eliminates the drawn factor from the reachable
priority order of FMEA by UPN, which is calculated by: matrix M, repeatedly evaluates adjacency reachability set
R and adjacency antecedent set A until all the factors are
UPNi ¼ RPNi  Ci ¼ Si  Oi  Di  Ci ðEq1Þ drawn out (Fig. 4).
Finally, one establishes the hierarchy structure as Fig. 5.
where Ci is the weight of the utility of corrective actions.
From the hierarchy structure, S1 and S9 are control level,
whose weights are given as 0.6 and 0.4, respectively. The
lower layer is regarded as network level, by which one can
Case Study evaluate the pairwise comparison matrix, and then evaluate
its eigenvectors to group into supermatrix. Its structure is
In electronic assembly industry, the main operation of shown in Fig. 6. To group the supermatrix, first transform
SMT manufacturing is the placements of parts of the them into weighted supermatrix as shown in Fig. 7. The
main board on the printed circuit board (PCB) by a part constringent supermatrix is obtained after multiplying the
mounter machine. The parts are then fixed on PCB by weighted supermatrix until its constringency; it can show
exposure to a reflow oven and flow soldering machine. the weight of each corrective action (Fig. 8).
However, open joints and distorted welding spots are The weights of corrective actions are summarized in
common problems in SMT manufacturing process. Table 3. When these weights are used in Eq. 1, the UPN
Therefore, the research team tries to improve manufac- shown in Table 3 are obtained. The difference of priority
turing quality and implement FMEA by examining the order between traditional RPN with UPN are also shown in
data summarized in Table 1. Table 3.
Upon analysis, there are 10 possible causes of open
joints and 12 causes of a distorted welding spot. Assess-
ment of the RPN provides the priority order shown in Conclusion
Table 2. The research team is to choose actions with a RPN
higher than 100 as the improvement priority items. There The priority order determined by RPN and UPN evaluation
were nine such items, and proposed corrective actions for is slightly difficult: the eighth finding in priority by RPN
each item are shown in Table 2. An assessment of the ranks first by UPN; the second in priority by RPN ranks
ordered pairs in the binary relation provided for the cor- seventh by UPN; the sixth in priority by RPN ranks third
rective action in Item 9 is shown as Fig. 2. by UPN. There is no big difference in the other items.
The adjacency matrix A can be obtained from the However, by following the new priority order, the utility of
binary relation in Fig. 2, then it can be transformed into corrective actions is improved and the improvement effect
element connection matrix N. Finally, the reachability brings the favorable result in the shortest time at lowest
matrix M can be obtained by Boolean operation (Fig. 3). cost. The method is indeed feasible as demonstrated
In this way, one can get reachability matrix M = N4, then through this case study.

123
328
References
1. Hung, G.Q., Nie, M., Mark, K.L.: Web-based failure mode and
effect analysis, Comput. Ind. Eng., 37, 177–180 (1999).
2. Stamatis D.H.: Failure Mode and Effect Analysis: FMEA from
Theory to Execution, ASQC Quality Press, Milwaukee, WI
(1995).
3. Onodera, K.: Effective techniques of FMEA at each life-cycle
stage, Proceeding Annual Reliability and Maintainability Sym-
posium, 50–56 (1997).
4. Linton, J.D.: Facing the challenges of service automation: An
enabler for e-commerce and productivity gain in traditional ser-
vices, IEEE Trans. Eng. Manage., 50(4), 478–484 (2003).
5. Reiling, J.G., Knutzen, B.L., Stoecklein, M.: FMEA—the cure
for medical errors, Qual. Progr., 36(8), 67–71 (2003).
6. Shahin, A.: Integration of FMEA and the Kano model: An
exploratory examination, Int. J. Qual. Reliab. Manage., 21(6/7),
731–746 (2005).
7. Reliey, T.T.: FMEA in preventing medical accidents, ASQ
Annual Quality Congress Proceedings, 657–664 (2002).
8. Vandenbrande, W.: The FMEA method in environment man-
agement systems, Stand. Kach., 2, 98–101 (2003).
9. Sawhney, R., Padiyar, A. Li, Y.: FMEA based approach for
supplier development, IIE Annual Conference and Exhibition,
7–16 (2004).
123
J Fail. Anal. and Preven. (2007) 7:321–328
10. Gilchrist, W.: Modeling failure modes and effects analysis, Int. J.
Qual. Reliab. Manage., 10(5), 16–23 (1993).
11. Potential Failure Mode and Effects Analysis, 2nd ed., Automo-
tive Industry Action Group, Chrysler Corporation, Ford Motor
Company, General Motors Corporation (1995).
12. Ben-Daya, M., Raouf, A.: A revised failure mode and effects
analysis model, Int. J. Qual. Reliabil. Manage., 13, 43–47 (1993).
13. Kara-Zaitri, C., Fleming, P.V.: Applications of fizzy inference
methods to failure modes effects and criticality analysis IFME-
CA, International Conference on Safety and Reliability, 2403–
2414 (1997).
14. Bowles, J.B.: The new SAE FMECA standard, Proceedings Annual
Reliability and Maintainability Symposium, 48–53 (1998).
15. Sankar, N.R., Prabhu, B.S.: Modified approach for prioritization
of failures in a system failure mode and effects analysis, Int. J.
Qual. Reliabil. Manage., 18(3), 324–335 (2001).
16. Bowles, J.B.: An assessment of RPN prioritization in a failure
modes effects and criticality analysis, Proceeding Annual Reli-
ability and Maintainability Symposium, 380–386 (2003).
17. Warfield, J.N.: On arranging elements of a hierarchy in graphic
form, IEEE Trans. Systems, Man, Cybernet., 2, 121–132 (1973).
18. Saaty T.L.: Decision Making with Dependence and Feedback:
The Analytic Network Process, RWS, Pittsburgh, PA (1996).
19. Saaty, T.L.: The Analytic Hierarchical Process, McGraw-Hill,
New York (1980).