Soft Computing (2021) 25:15221–15241

https://doi.org/10.1007/s00500-021-06359-z (0123456789().,-volV)(0123456789().,-volV)

SOFT COMPUTING IN DECISION MAKING AND IN MODELING IN ECONOMICS

An improving failure mode and effect analysis method for pallet

exchange rack risk analysis
Chuanxi Jin1 • Yan Ran1 • Genbao Zhang1,2

Accepted: 24 September 2021 / Published online: 12 October 2021

 The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2021

Abstract
In order to enhance quality and reliability of mechanical and electrical products, the methods of taking corresponding
corrective measures to eliminate or alleviate product failure in advance have been widely concerned. Failure mode and
effects analysis (FMEA) is a typical prevention reliability analysis method. However, there are some drawbacks in
traditional FMEA method. To overcome these drawbacks, we propose a hybrid risk evaluation method, which combines
picture fuzzy sets (PFSs), the PF-linear programming model (PF-LPM) method and the PF-weighted aggregated sum
product assessment (PF-WASPAS) method. We adopt PFSs to evaluate risks of products. In order to overcome drawback
of the traditional distance between PFSs, some new distance measures between PFSs based on the Dice similarity and the
Jaccard similarity are proposed by us. The PF-LPM method which considers the subjective weights of risk factors and
calculates synthetical deviation with the Dice similarity-based distance is utilized to calculate the weights of risk factors.
Moreover, the PFWA operator and the PFWG operator are used by us to fuse experts’ evaluation information. Then, the
PF-WASPAS method is utilized to rank failure modes. Finally, an illustrative example with respect to pallet exchange rack
is introduced, and the rationality, effectiveness and applicability of the proposed method are verified by a discussion and
comparison.

Keywords Failure mode and effects analysis  Picture fuzzy sets  PFWA operator  PFWG operator  WASPAS

1 Introduction improvement techniques have been proposed. For instance,

failure mode and effects analysis (FMEA) (Boral et al.
As one of the six key quality characteristics (reliability, 2020a; Fang et al. 2020), fault tree analysis (FTA) (Hu
precision, precision life, maintainability, availability and et al. 2020a), event tree analysis (ETA) (Purba et al. 2020),
precision stability) of CNC machine tools, reliability plays hazard and operability studies (Marhavilas et al. 2020) and
an important role in the quality improvement of CNC root cause analysis (RCA) (Velasquez 2020). Among these
machine tools (Chen et al. 2021). Reliability refers to the methods, the FMEA is the most representative one,
ability or possibility of a component, product or system to because it’s main feature is taking actions to prevent the
perform a specified function without failure under a certain occurrence of failure in advance, rather than improving
period of time and certain conditions (Jin et al. 2020). In reliability through post-mortem testing. The implementa-
order to reduce the failure probability of products and tion process of traditional FMEA method mainly includes
improve reliability of products, many reliability six steps: identifying all potential failure modes of products
according to failure data, analyzing the causes and effect of
each failure mode based on experience, inviting experts to
& Yan Ran
rate risks by utilizing crisp value, calculating risk priority
ranyan@cqu.edu.cn
numbers (RPNs) of each failure modes, ranking failure
1
College of Mechanical Engineering and State Key modes based on RPN values and taking corresponding
Laboratory Mechanical Transmission, Chongqing University, corrective measures to eliminate or alleviate them (Li et al.
Chongqing 400044, China 2019a; Lo et al. 2019; Liu et al. 2019a).
2
College of Mechanical Engineering, Chongqing University of
Arts and Sciences, Chongqing 402160, China

123
15222 C. Jin et al.

The ease of use of FMEA technique has let to the neutral degree of membership, negative degree of mem-
extension of its application in different fields, for example, bership and refusal degree of membership, which can
risk assessment of maritime autonomous surface ships express the hesitation and uncertainty of experts’ evalua-
(Chang et al. 2021), failure analysis of diesel engine piston tion information; (2) the traditional distance measure
in transport utility vehicles (Deulgaonkar et al. 2021), risk between PFSs fails to satisfy the axiomatic requirements of
evaluation and mitigation of sustainable road freight being PF distance measures, and may produce counterin-
transport operation (Dadsena et al. 2019) and risk analysis tuitive results in calculating the distance between different
of sequential processes in food industry (Rezaee et al. PFSs. Therefore, it is necessary for us to develop new
2018) and so on. However, the traditional FMEA method distance measures between PFSs to solve this problem. The
does not have absolute advantages in the efficiency and distance measure based on Dice similarity and Jaccard
effectiveness of risk evaluation. The reasons for this similarity can solve this problem well; (3) because experts
problem mainly include the following aspects (Zheng and come from different fields and have different knowledge
Tang 2020; Boral et al. 2020b): (1) Rating risks by using background, the experts’ evaluation information may be
crisp value, which ignores the hesitation and uncertainty of inconsistent. In order to reach a consensus, we need to fuse
experts’ evaluation information; (2) The weights of the the experts’ evaluation information under PFSs environ-
three risk factors (S, O and D) are equal. It is not irrational, ment. PFSs provide a parameterized family of aggregation
for different risk analysis cases, their weights should not be operators such as the PFWA operator and the PFWG
equal; (3) The mathematical form (i.e., multiplication) operator; (4) in the traditional FMEA method, the risk
adopted for calculating the RPN value is irrational, because factors are equally weighted. However, for different risk
it is strongly sensitive to variations in criticality factor evaluation cases, the weights of risk factors should not be
evaluations; (4) The same RPN value may indicate totally equal. Moreover, the subjective weights of risk factors
different risk implications. should be taken into account. The PF-LPM method can
In order to overcome the above drawbacks and improve solve these problems well, which calculates synthetical
the effectiveness and efficiency of the traditional FMEA deviation with the Dice similarity-based distance, and
method, in this paper, we propose a new risk evaluation calculates the weights of risk factors by establishing a
method based on the PF-LPM and the PF-WASPAS. linear programming model; (5) the ranking mechanism of
Moreover, in order to extend the traditional distance traditional FMEA method is irrational. However, the PF-
measures between two PFSs, some new distance measures WASPAS method is simple and direct, and can yield rea-
based on Dice similarity and Jaccard similarity are pro- sonable, acceptable and relatively accurate results in
posed by us. The proposed FMEA method mainly includes ranking failure modes.
three parts: risk evaluation, calculating the weights of risk Therefore, the major contributions of this paper are as
factors with PF-LPM and prioritizing risks with the PF- follows: (1) rating risks by using PFSs, which overcomes
WASPAS method. In risk revaluation, taking the hesitation the drawback that the traditional FMEA method rates risks
and uncertainty of experts’ evaluation information into by crisp values and cannot express the hesitation and
account, PFSs are utilized to rate risks. In calculating the uncertainty of experts’ evaluation information; (2) devel-
weights of risk factors with PF-LPM, we consider the oping some new distance measures based on Dice simi-
subjective weight of risk factors, and adopt the Dice sim- larity and Jaccard similarity, which extends traditional
ilarity-based distance to calculate synthetical deviation. In distance measure between PFSs, and overcomes the
order to obtain more consistent expert evaluation infor- drawback that the traditional distance measure fails to
mation, the PFWA operator and the PFWG operator are satisfy the axiomatic requirements of being PF distance
utilized to fuse experts’ evaluation information. Finally, the measures, and may produce counterintuitive results in
weights of risk factors are calculated with PF-LPM. For calculating the distance between different PFSs; (3) fusing
prioritization of risk factors, the PF-WASPAS method is experts’ evaluation information by using the PFWA oper-
utilized to rank failure modes. ator and the PFWG operator, which can make experts’
Based on the above analysis, we can sum up the main evaluation information more consistent; (4) calculating the
motivation of this paper as follows: (1) in risk revaluation, weights of risk factors with the PF-LPM, which can
under time pressure and limited knowledge and data, for overcome the drawback that the risk factors are weighted
some complex products or systems, experts may not be so equally in the traditional FMEA method; (5) ranking fail-
focused, and different experts have different information ure modes with the PF-WASPAS method, which can
processing capabilities. Therefore, the obtained experts’ overcome the drawback that the ranking mechanism of the
evaluation information may be uncertain and hesitant. traditional FMEA method is irrational.
However, the PFSs is of great help in solving this problem. The organization of the rest paper is as follows: The
Its parameters include positive degree of membership, literature review is synthesized in Sect. 2. In Sect. 3, some

123
An improving failure mode and effect analysis method for pallet exchange rack risk analysis 15223

basic knowledge of the PFSs, the PFWA operator and the between PFSs to determinate the flood disaster risk in the
PFWG operator is briefly reviewed. Some new distance South region of India. Son et al. (2017) introduced some
measures between PFSs are introduced in Sect. 4. Section 5 new distance measures between PFSs and adopted them in
presents the proposed FMEA method, which includes risk the clustering analysis. Khan et al. (2020) proposed a Bi-
evaluation, calculating the weights of risk factors with PF- parametric PF distance measure and demonstrated the
LPM and prioritizing risks with the PF-WASPAS method. application of the PF distance measure in the pattern
In Sect. 6, an illustrative example with respect to the pallet recognition and medical diagnosis. Ganie et al. (2021)
exchange rack is introduced. A discussion and comparison proposed the concepts of the PF correlation measures and
process are reported in Sect. 7, and conclusions are finally adopted it in pattern recognition, medical diagnosis, and
drawn in Sect. 8. clustering analysis. Liu et al. (2019b) presented a picture
fuzzy ordered weighted distance measure and adopted it in
a practical application of investment alternatives selection.
2 Literature review Duong et al. (2021) proposed a novel dissimilarity measure
on PFSs and adopted it in multi-criteria decision-making
2.1 The applications of PFSs problem. Although these distance measure methods are
widely used, these distance measures cannot satisfy the
The concept of PFSs was first proposed by Coung in 2014. axiomatic requirements of being distance measures
It emerged from IFSs, but compared with IFSs, it is more between PFSs, and may produce counterintuitive results in
powerful and suitable to deal with cases requiring human calculating the distance between different PFSs. To this
opinion involving more answers of types: yes, abstain, no end, we propose some new distance measures between
and refusal (Xu et al. 2019). The ease of use of the PFSs PFSs based on Dice similarity and Jaccard similarity. To
has let to the extension of its application in different fields. the best of our knowledge, there is no study that does these
For example, Luo et al. (2020) adopted PFSs to indicate the works.
subjective evaluation information in evaluating the thermal
comfort of underground mines. In order to select the 2.3 Weighting methods and ranking methods
optimal electric vehicle charging station (EVCS) site from
all the feasible sites in Beijing, the PFSs was utilized by Ju In traditional FMEA method, the risk factors are equally
et al. (2019) to rate EVCS sites. In the same vein, to select weighted. It is irrational, because for different risk cases,
an optimal emergency alternative from four feasible the weights of risk factors should not be equal. In order to
emergency alternatives, Li et al. (2019b) adopted PFSs to deal with this problem, many weighting methods have been
rate emergency alternatives. Wang et al. (2018a) adopted proposed, For instance, in order to consider the experts’
PFSs to determine whether the risks are prioritized to psychological behavior in the risk evaluation in FMEA
guarantee the quality, safety, and timely completion of the approach, Wang et al. (2018b) adopt the prospect theory to
construction project. Zhang et al. (2020) adopted PFSs to calculate the weights of risk factors. A weighting method
rate suppliers in a practical supplier selection of a pump based on fuzzy analytical hierarchy process (FAHP) which
enterprise. In addition, the PFSs were also utilized in has capability to incorporate inherent inconsistencies of a
medical diagnosis and pattern recognition (Zeng et al. decision-making process was developed by Boral et al.
2019), risk evaluation of flood disaster (Singh et al. 2018), (2020a) to calculate the weights of risk factors. Gul et al.
multiple attribute decision making problems (Van Dinh (2020) built a fuzzy best–worst method (FBWM) which
and Xuan Thao 2018), building materials recognition (Wei determines the importance weights of criteria using two
and Gao 2018), and so forth. However, to the best of our comparison matrices to weight risk factors. Mohsen et al.
knowledge, there is no study that rates risks by using PFSs (2017) utilized the concept of Shannon entropy to deploy
in FMEA. objective weights, and took the subjective weights of risk
factors assigned by experts into account. A consensus-
2.2 Distance measures of PFSs driven methodology was proposed by Zhang et al. (2019)
to generate the weights of risk factors. Additionally, data
In order to describe difference between two PFSs, many envelopment analysis was also utilized to calculate the
distance measure methods have been proposed. For objective weights of risk factors. In this paper, we combine
instance, Dinh et al. (2018) proposed some new distance PFSs and LPM method to calculate the weights of risk
measures between PFSs and adopted it in the Dinah pattern factors, and take the subjective weights of risk factors into
recognition problem. Some new PF distance measures were account. For LPM method, it is also widely used in dif-
proposed by Dutta et al. (2018) to solve medical diagnosis ferent fields. For example, to highlight the overall differ-
problem. Singh et al. (2018) used the distance measures ence among the performance values of alternatives, Yi

123
15224 C. Jin et al.

et al. (2019) adopted LPM method to calculate indicator 3 Preliminaries

weights. Under condition that the weights may be partially
known or even completely unknown, Wu et al. (2020) In this section, some basic knowledge about PFSs that will
adopted LPM method to determine the weights of customer be utilized in the subsequent research is briefly reviewed.
requirements. Li et al. (2019c) built the LPM method in
dynamic situation to calculate indicator weights with the 3.1 Picture fuzzy sets
purpose of widening the overall difference of the devel-
opment of the cities. As an extension of FSs and IFSs, PFSs can express the
In traditional FMEA method, failure modes are ranked evaluation information more accurately and comprehen-
based on RPN values. According to Introduction, we can sively than FSs and IFSs when experts face opinions
obtain that this ranking mechanism is irrational. To this involving four types of answers. It is defined as following.
end, Luo et al. (2019) proposed the multi-attributive border
approximation area comparison (MABAC) method based Definition 1 Tian et al. (2019) A picture fuzzy set T over
on likelihood to rank roadway support schemes. In Nie the universe of discourse X is interpreted as
et al. (2018), an alternative FMEA model which combines T ¼ fhaT ðxÞ; bT ðxÞ; cT ðxÞijx 2 Xg, where positive degree
best–worst, maximizing derivation and COPRAS methods of membership aT ðxÞ : X ! ½0; 1, neutral degree of
was utilized to evaluate the risk of a supercritical water membership bT ðxÞ : X ! ½0; 1 and negative degree of
gasification system. Chen et al. (2018) proposed a new membership cT ðxÞ : X ! ½0; 1 in a fuzzy set T such that
FMEA model which combined gray relational projection for every x 2 X. Also, the degree of refusal is defined for x
(GRP) method and dempster-shafer theory method to as pT ðxÞ ¼ 1  aT ðxÞ  bT ðxÞ  cT ðxÞ. In particular, when
evaluate the risks of aircraft turbine rotor blades. In order only one element exists in X, the PFS can be degraded to a
to improve the capabilities of the traditional FMEA picture fuzzy number (PFN). For the sake of convenience,
method, Zhao et al. (2016) proposed a modified MULTI- we call T ¼ ða; b; cÞ a PFN.
MOORA approach based on interval-valued intuitionistic Then, some operational rules with respect to PFSs are
fuzzy sets and continuous weighted entropy. Shin et al. introduced.
adopted the DEMATEL approach to identify and prioritize
failure modes of a software development process. Wang Definition 2 (Tian et al. 2019) Let T1 ¼ ða1 ; b1 ; c1 Þ, T2 ¼
et al. (2018b) proposed a risk evaluation method based on ða2 ; b2 ; c2 Þ and T ¼ ða; b; cÞ be three PFNs, their opera-
prospect theory and Choquet integral, which considered tional rules contain
decision-maker’s psychological behavior and risk factors’ (1) T ¼ ðc; b; aÞ,
interaction relationships simultaneously. Hu et al. (2020b) (2) T1  T2 ¼ ða1 þ a2  a1 a2 ; b1 b2 ; c1 c2 Þ,
proposed a FMEA model based on multi-granularity lin- (3) T1  T2 ¼ ða1 a2 ; b1 þ b2  b1 b2 ; c1 þ c2  c1 c2 Þ,
guistic terms and the dempster-shafer evidence theory, An
(4) T k ¼ ðak ; 1  ð1  bÞk ; 1  ð1  cÞk Þ,
integrated robust DEA-FMEA approach was proposed by
Yousefi et al. (2018) to evaluate and prioritize safety and (5) kT ¼ ð1  ð1  aÞk ; bk ; ck Þ.
environment risks, which considered two extra risk factors: In order to compare two PFNs, Jana et al. introduced the
cost and duration of treatment. In this paper, the PF- concepts of the score function and the accuracy function of
WASPAS method is utilized by us to rank failure modes, a PFN.
which combine PFSs and the WASPAS method. For the
WASPAS method, Mardani et al. (2020) adopted the Definition 3 Jana et al. (2019) For a PFN T ¼ ða; b; cÞ, its
WASPAS method to rank digital technologies (DTs) under score function and accuracy function can be defined as
hesitant fuzzy sets. A hybrid fuzzy multi-criteria decision- SðTÞ ¼ a  c ð1Þ
making method based on level based weight assessment-
AðTÞ ¼ a þ b þ c ð2Þ
WASPAS-Heronian (LBWA-WASPAS-H) model was
proposed by Pamucar et al. (2020) to select ground access
mode. To the best of our knowledge, there is no study that
Definition 4 Jana et al. (2019) Suppose T1 = (a1, b1, c1)
weights risk factors with the PF-LPM method and ranks
and T2 = (a2, b2, c2) are two arbitrary PFNs. The prefer-
failure modes with the PF-WASPAS method.
ence relations between them can be determined by the
following principles:
(1) When SðT1 Þ\SðT2 Þ, then T1  T2 ;
(2) When SðT1 Þ ¼ SðT2 Þ, if AðT1 Þ\AðT2 Þ, then T1  T2 ,
if AðT1 Þ ¼ AðT2 Þ, then T1  T2 .

123
An improving failure mode and effect analysis method for pallet exchange rack risk analysis 15225

To measure the deviation or difference between two 4.1 The Dice similarity-based distance measure
PFNs, the normalized Hamming distance of PFSs was
proposed by Zhang et al. (2020). The concept of the Dice similarity was first proposed by
Dice (1945), it is defined as follows:
Definition 5 Suppose T1 = (a1, b1, c1) and T2 = (a2, b2, c2)
are two arbitrary PFNs, then the normalized Hamming Definition 7 Let T1 ¼ ha1 ðxi Þ; b1 ðxi Þ; c1 ðxi Þi and T2 ¼
distance between T1and T2 can be obtained as follows: ha2 ðxi Þ; b2 ðxi Þ; c2 ðxi Þi be two PFSs in a given finite set
1 X ¼ fx1 ; x2 ;    xn g. Then, the Dice similarity between T1
dðT1 ; T2 Þ ¼ ðja1  a2 j þ jb1  b2 j þ jc1  c2 jÞ ð3Þ and T2 can be obtained as follows:
3
SD ðT1 ; T2 Þ
n  
1X 2ða1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi ÞÞ
¼
3.2 PFWA operator and PFWG operator n i¼1 a21 ðxi Þ þ b21 ðxi Þ þ c21 ðxi Þ þ a22 ðxi Þ þ b22 ðxi Þ þ c22 ðxi Þ
ð6Þ
To fuse experts’ evaluation information, the picture fuzzy
weighted averaging (PFWA) operator and the picture fuzzy
weighted geometric mean (PFWG) operator of PFSs are Then, the Dice similarity-based distance measure
defined. between PFSs is proposed by us to extend traditional dis-
Definition 6 Wei (2017) Let Tj = (aj, bj cj) (j = 1,2,…,n) tance measure, it is defined as follows:
be a collection of PFNs. The PFWA operator is a mapping Definition 8 Let T1 ¼ ha1 ðxi Þ; b1 ðxi Þ; c1 ðxi Þi and T2 ¼
Pn ? P such that ha2 ðxi Þ; b2 ðxi Þ; c2 ðxi Þi be two PFSs in a given finite set
n X ¼ fx1 ; x2 ;    xn g. Then, the Dice similarity-based dis-
PFWAðT1 ; T2 ; . . .; Tn Þ ¼  ðxj Tj Þ
j¼1 tance measure between T1 and T2 can be obtained as
! ð4Þ
Y
n  x Y n   Y
n   follows:
x x
¼ 1 1  aj j ; bj j ; cj j
dDS ðT1 ; T2 Þ ¼ 1  SD ðT1 ; T2 Þ ¼ 1
j¼1 j¼1 j¼1
n  
1X 2ða1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi ÞÞ
where x = (x1,x2,…,xn)T is the weight vector of Tj 
n i¼1 a21 ðxi Þ þ b21 ðxi Þ þ c21 ðxi Þ þ a22 ðxi Þ þ b22 ðxi Þ þ c22 ðxi Þ
(j = 1,2,…,n), it can be calculated by a weighted genera-
ð7Þ
tion method based on the normal distribution (Chen et al.
P
2020), and xj [ 0, nj¼1 xj ¼ 1.

Definition 7 Wei (2017) Let Tj = (aj, bj cj)(j = 1,2,…,n) Property 1 Based on the above definition, we can obtain
be a collection of PFNs. The PFWG operator is a mapping that:
Pn ? P such that
(1) 0 dDS ðT1 ; T2 Þ\1;
n
PFWGðT1 ; T2 ; . . .; Tn Þ ¼  ðTj Þ xj (2) dDS ðT1 ; T2 Þ ¼ dDS ðT2 ; T1 Þ;
j¼1
! (3) dDS ðT1 ; T2 Þ ¼ 0 if and only if T1 ¼ T2 .
Y
n  xj Y
n  xj Y
n  xj
¼ aj ;1  1  bj ;1  1  cj Proof
j¼1 j¼1 j¼1

ð5Þ (1) According to the definition of PFSs, aðxÞ 2 ½0; 1,

bðxÞ 2 ½0; 1 and cðxÞ 2 ½0; 1, then, dDS ðT1 ; T2 Þ
0.
where x = (x1,x2,…,xn)T is the weight vector of Tj Because a2 þ b2
2ab, we have 2ða1 ðxi Þa2 ðxi Þþ
(j = 1,2,…,n), it can be calculated by a weighted genera- b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi ÞÞ a21 ðxi Þ þ b21 ðxi Þ þ c21
tion method based on the normal distribution (Chen et al.
P ðxi Þ þ a22 ðxi Þ þ b22 ðxi Þþ c22 ðxi Þ, then, 0
2020), and xj [ 0, nj¼1 xj ¼ 1.
dDS ðT1 ; T2 Þ\1.
(2) According to the definition, we can obtain that the
Dice similarity-based distance measure satisfies the
4 New distance measures between PFSs second condition.
(3) If T1 ¼ T2 , then a1 ðxi Þ ¼ a2 ðxi Þ, b1 ðxi Þ ¼ b2 ðxi Þ and
In this section, some new distance measures between PFSs c1 ðxi Þ ¼ c2 ðxi Þ. Therefore, we can obtain that
based on the Dice similarity and Jaccard similarity are dDS ðT1 ; T2 Þ ¼ 0. If dDS ðT1 ; T2 Þ ¼ 0, there is
proposed to overcome drawback that information loss may 2ða1 ðxi Þ a2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi ÞÞ ¼
be caused in traditional distance measure.

123
15226 C. Jin et al.

a21 ðxi Þ þ b21 ðxi Þ þ c21 ðxi Þþ a22 ðxi Þ þ b22 ðxi Þ þ c22 ðxi Þ. Definition 10 Let T1 ¼ ha1 ðxi Þ; b1 ðxi Þ; c1 ðxi Þi and T2 ¼
then, ða1 ðxi Þ  a2 ðxi ÞÞ2 þðb1 ðxi Þ  b2 ðxi ÞÞ2 þ ha2 ðxi Þ; b2 ðxi Þ; c2 ðxi Þi be two PFSs in a given finite set
X ¼ fx1 ; x2 ;    xn g. Then, the Jaccard similarity between
ðc1 ðxi Þ  c2 ðxi ÞÞ2 ¼ 0. Thus, a1 ðxi Þ ¼ a2 ðxi Þ,
T1 and T2 can be obtained as follows:
b1 ðxi Þ ¼ b2 ðxi Þ and c1 ðxi Þ ¼ c2 ðxi Þ, that is, T1 ¼ T2 .
This completes the proof.

!
1X n
a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ
SJ ðT1 ; T2 Þ ¼  2 
n i¼1 a1 ðxi Þ þ b21 ðxi Þ þ c21 ðxi Þ þ a22 ðxi Þ þ b22 ðxi Þ þ c22 ðxi Þ  a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ
ð9Þ

Because the weights of PFNs have great influence on the Then, the Jaccard similarity-based distance measure
distance measure between PFNs, we proposed the concept between PFSs is proposed by us to extend traditional dis-
of the Dice similarity-based weighted distance measure, it tance measure, it is defined as follows.
is defined as follows.
Definition 11 Let T1 ¼ ha1 ðxi Þ; b1 ðxi Þ; c1 ðxi Þi and T2 ¼
Definition 9 Let T1 ¼ ha1 ðxi Þ; b1 ðxi Þ; c1 ðxi Þi and T2 ¼ ha2 ðxi Þ; b2 ðxi Þ; c2 ðxi Þi be two PFSs in a given finite set
ha2 ðxi Þ; b2 ðxi Þ; c2 ðxi Þi be two PFSs in a given finite set X ¼ fx1 ; x2 ;    xn g. Then, the Jaccard similarity-based
X ¼ fx1 ; x2 ;    xn g. Then, the Dice similarity-based distance measure between T1 and T2 can be obtained as
weighted distance measure between T1 and T2 can be follows:
obtained as follows: dJS ðT1 ; T2 Þ ¼ 1  SJ ðT1 ; T2 Þ ¼ 1
x 0 1
dDS ðT1 ; T2 Þ ¼ 1  SD ðT1 ; T2 Þ ¼ 1
  n B C
1X n
2ða1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi ÞÞ 1X B
B a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ C
C
 xi 2  
n i¼1 a1 ðxi Þ þ b21 ðxi Þ þ c21 ðxi Þ þ a22 ðxi Þ þ b22 ðxi Þ þ c22 ðxi Þ n i¼1 @ a1 ðxi Þ þ b1 ðxi Þ þ c1 ðxi Þ þ a2 ðxi Þ þ b2 ðxi Þ þ c2 ðxi Þ C
B 2 2 2 2 2 2
A
ð8Þ a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ
ð10Þ

In the same vein, the Dice similarity-based weighted

distance measure between PFSs also satisfies: (1) Property 2 Similarly, the Jaccard similarity-based dis-
x x x
0 dDS ðT1 ; T2 Þ\1; (2) dDS ðT1 ; T2 Þ ¼ dDS ðT2 ; T1 Þ; (3) tance measure between PFSs satisfies:
x
dDS ðT1 ; T2 Þ ¼ 0 if and only if T1 ¼ T2 .
(1) 0 dJS ðT1 ; T2 Þ\1;
4.2 The Jaccard similarity-based distance (2) dJS ðT1 ; T2 Þ ¼ dJS ðT2 ; T1 Þ;
measures (3) dJS ðT1 ; T2 Þ ¼ 0 if and only if T1 ¼ T2 .

The concept of the Jaccard similarity was first proposed by Proof

Jaccard in 1901 (Jaccard 1901), it is defined as follows. (1) Since a2 þ b2
2ab, we can obtain that

2ða1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi ÞÞ a21 ðxi Þ þ b21 ðxi Þ þ c21 ðxi Þ þ a22 ðxi Þ þ b22 ðxi Þ þ c22 ðxi Þ
 
) a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ a21 ðxi Þ þ b21 ðxi Þ þ c21 ðxi Þ þ a22 ðxi Þ þ b22 ðxi Þ þ c22 ðxi Þ
 a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ
a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ
)   1
a21 ðxi Þ þ b21 ðxi Þ þ c21 ðxi Þ þ a22 ðxi Þ þ b22 ðxi Þ þ c22 ðxi Þ
a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ

123
An improving failure mode and effect analysis method for pallet exchange rack risk analysis 15227

and prioritizing risks with PF-WASPAS method. In risk

evaluation, PFSs is utilized to evaluate risks. In calculating
Thus, 0 dJS ðT1 ; T2 Þ\1. the weights of risk factors with the PF-LPM, the proposed
Dice similarity-based distance of PFSs is used to calculate
(2) It is obvious that dJS ðT1 ; T2 Þ ¼ dJS ðT2 ; T1 Þ.
synthetical deviation. The PFWA operator and PFWG
(3) If T1 ¼ T2 , then dDS ðT1 ; T2 Þ ¼ 0. If dDS ðT1 ; T2 Þ ¼ 0,
operator are utilized to fuse experts’ evaluation informa-
we can obtain that
tion, and the weights of risk factors are calculated with PF-
LPM method. In prioritizing risks with the PF-WASPAS
a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ method, the failure modes are ranked by using the PF-
  ¼1 WASPAS method. Next, we will illustrate each step of the
a21 ðxi Þ þ b21 ðxi Þ þ c21 ðxi Þ þ a22 ðxi Þ þ b22 ðxi Þ þ c22 ðxi Þ
proposed FMEA method in detail according to Fig. 1.
 a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ
) a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ 5.1 Risk evaluation
 
¼ a21 ðxi Þ þ b21 ðxi Þ þ c21 ðxi Þ þ a22 ðxi Þ þ b22 ðxi Þ þ c22 ðxi Þ
 a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ In traditional FMEA method, we determine all potential
2 2 2 failure modes of products based on experts’ experience and
) ða1 ðxi Þ  a2 ðxi ÞÞ þðb1 ðxi Þ  b2 ðxi ÞÞ þðc1 ðxi Þ  c2 ðxi ÞÞ
evaluate risks by using crisp values. However, on one hand,
¼0 because experts’ evaluation information is subjective, some
key failure modes may be lost in risk evaluation process.
Thus, T1 ¼ T2 . This completes the proof. To solve this problem, the failure tree of product is con-
Then, we introduce the weights of PFSs into the Jaccard structed to determine potential failure modes based on
similarity-based distance measure, and define the concept failure data collected. On other hand, in real life, it is
of the Jaccard similarity-based weighted distance measure impossible to describe the exact risk level with specific
between PFSs. numbers owing to time pressure and limited knowledge or
data, and various hesitation and uncertainty will appear in
Definition 12 Let T1 ¼ ha1 ðxi Þ; b1 ðxi Þ; c1 ðxi Þi and T2 ¼ the experts’ subjective evaluation. In this case, given that
ha2 ðxi Þ; b2 ðxi Þ; c2 ðxi Þi be two PFSs in a given finite set PFSs have an inherent superiority in solving this problem,
X ¼ fx1 ; x2 ;    xn g. Then, the Jaccard similarity-based we can adopt PFSs to rate risks. The concrete risk evalu-
weighted distance measure between T1 and T2 can be ation process includes the following three steps.
obtained as follows:
Step 1 Determine the failure modes and define risk
dJSx ðT1 ; T2 Þ ¼ 1  SJ ðT1 ; T2 Þ ¼ 1
0 1 factors.
B C Before conducting a risk evaluation, it is necessary for
1X n B a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ C
 xi B   C
n i¼1 B a
@ 1 i
2
ðx Þ þ b 2
ðx
1 i Þ þ c 2
ðx
1 i Þ þ a 2
ðx
2 i Þ þ b 2
ðx
2 i Þ þ c 2
ðx
2 i Þ C
A us to form a technical team and an expert team. The
technical team is composed of fifteen basic technical staffs
a1 ðxi Þa2 ðxi Þ þ b1 ðxi Þb2 ðxi Þ þ c1 ðxi Þc2 ðxi Þ
in the company, who come from five different departments:
ð11Þ design, manufacturing, technical service, business and
production management, each with 3–5 years of experi-
ence in CNC machine tool manufacturing industry. The
Similarly, the Jaccard similarity-based weighted dis- expert team is composed of seven senior technical staffs in
tance measure between PFSs also satisfies: (1) the company, who come from five different team too, but
x x x
0 dJS ðT1 ; T2 Þ\1; (2) dJS ðT1 ; T2 Þ ¼ dJS ðT2 ; T1 Þ; (3) each with more than 15 years of experience.
x
dJS ðT1 ; T2 Þ ¼ 0 if and only if T1 ¼ T2 . In order to determine potential failure modes of product,
we invite staffs in the technical team to collect product
data, which includes design drawings, design specification
5 The proposed FMEA method and fault data, and to establish failure tree based on product
data collected. Then, based on failure tree constructed, we
In this paper, we proposed a new risk evaluation method invite l experts ek ðk ¼ 1; 2;    lÞ; l 2 ½0; 7 to determine
based on the PF-LPM and the PF-WASPAS. The concrete potential failure modes of product that cause system fail-
steps of the proposed FMEA method are shown in Fig. 1. ure, denoted by FM ¼ fFM1 ; FM2    FMm g, where
According to Fig. 1, we can obtain that the proposed FMi ði ¼ 1; 2;    mÞ indicates the ith failure mode. Based
FMEA method mainly includes three steps: risk evaluation, on traditional RPN method, we can obtain that risk factors
calculating the weights of risk factors with the PF-LPM include S, O and D,

123
15228 C. Jin et al.

Fig. 1 The framework of the

proposed FMEA method

Table 1 Linguistic terms and corresponding picture fuzzy numbers for rating failure modes are expressed as linguistic terms.
Linguistic terms and corresponding picture fuzzy numbers
Linguistic terms Picture fuzzy numbers
are shown in Table 1. In order to rate risks, three of seven
Very very low (VVL) (0,0,1) experts in expert team are invited to provide their assess-
Very low (VL) (0,0.05,0.9) ments over failure modes using linguistic terms to establish
Low (L) (0.2,0.45,0.25) linguistic terms decision matrices, denoted by T k ¼
Fair (F) (0.45,0.4,0.1) 
High (H) (0.8,0.1,0.05) Tiyk ; k ¼ 1; 2;    l; i ¼ 1; 2;    m; y ¼ 1; 2;    p;
mp
Very high (VH) (0.9,0.05,0)
where Tiyk indicates a linguistic term information of failure
Very very high (VVH) (1,0,0)
mode FMi with respect to risk factor RFj given by expert
ek . After converting linguistic terms information into cor-
responding PFNs based on Table 1, the PF evaluation
  information of the ith failure mode with respect to the jth
RF ¼ RF1 ; RF2 ;    RFp ¼ fS; O; Dg; p ¼ 3, where risk factor provided by expert ek can be denoted as xkiy ¼
RFy ðy ¼ 1; 2;    pÞ means the yth risk factor.
ðaiy ; biy ; ciy Þ; k ¼ 1; 2;    l; i ¼ 1; 2;    m; y ¼ 1; 2;    p,
Step 2 Evaluate risks by linguistic terms and construct and PFNs decision matrices can be established as

PFNs decision matrices. X k ¼ Xiyk .
mp
In the real life, due to time pressure and lack of
knowledge or data, there may be hesitation or uncertainty Step 3 Standardize PFNs decision matrices.
about experts’ evaluation information. In this paper, we
define that seven grades of experts’ evaluation information

123
An improving failure mode and effect analysis method for pallet exchange rack risk analysis 15229

In order to make experts’ evaluation information more • Step 4 Calculate syncretic decision matrix.
realistic, we need to standardize PFNs decision matrices as
In order to obtain more consistent evaluation results, the
follows:
( k PFWG operator is utilized to fuse standardized PFNs
  xiy S and O decision matrices. So syncretic decision matrix X ¼
Xiy ¼ aiy ; biy ; ciy ¼  k c
k
ð12Þ
xiy D ðxiy Þmp can be calculated as follows:
3
xiy ¼ PFWGðx1iy ; x2iy ; x3iy Þ ¼  ðxkiy Þwk
k¼1
5.2 Calculating the weights of risk factors !
3 
Y wk 3 
Y w k 3 
Y wk
with PF-LPM ¼ akiy ;1  1 bkiy ;1  1 ckiy
k¼1 k¼1 k¼1

In order to overcome drawback that three risk factors (S, O ð15Þ

and D) are equally weighted in the traditional FMEA
where xkiy ¼ ðakiy ; bkiy ; ckiy Þ.
method, we propose a new method of weighting risk fac-
tors, called PF-LPM, which takes into account subjective • Step 5 Establish linear programming model.
weights of risk factors and constructs maximizing devia-
For a risk evaluation process, if the evaluation values of
tion by using the Dice similarity-based distance. The
all failure modes under a risk factor are more different,
concrete process includes the following seven steps.
which indicates this risk factor has a significant contribu-
• Step 1 Construct subjective weight matrices of risk tion to the final ranking of failure modes. Conversely, if the
factors. evaluation values of all failure modes under a risk factor
have few differences, which indicates that this risk factor
Similar to Step 2 in Sect. 5.1, the three of seven experts
has a lower contribution to the final ranking of failure
in expert team are invited to rate weight of risk factors, and
modes. Therefore, for the risk factor that has large differ-
construct subjective weight matrices of risk factors
ences in evaluation values, it should be assigned a greater
xSk ¼ ðxSk Sk
y Þ1p ; k ¼ 1; 2;    l, where xy indicates sub- weight. Otherwise, it should be assigned a smaller weight.
jective weight of yth risk factors given by expert ek . In order to describe these differences, the concept of syn-
• Step 2 Calculate syncretic subjective weight matrix. thetical deviation is introduced based on the Dice simi-
larity-based distance. It is defined as follows:
In order to obtain more consistent evaluation results, the
1 XX X
p m m
PFWA operator is utilized to fuse experts’ evaluation    
D xO ¼ dDS xiy ; xhy xO
y ð16Þ
information. So syncretic subjective weight matrix xS ¼ m  1 y¼1 i¼1 h¼1;h6¼i
ðay Þ1p can be calculated as follows:
where xO y ; y ¼ 1;2; . . .; p is objective weight of yth risk
3
ay ¼ PFWAðxS1 S2 S3 Sk factors, and dDS xiy ; xhy ði; h ¼ 1; 2; . . .; mÞ is the Dice
y ; xy ; xy Þ ¼  ðwk xy Þ
k¼1 ! similarity-based distance between the evaluation values on
3 
Y wk 3 
Y wk Y3  wk
¼ 1 1 aSk ; bSk ; cSk failure mode FMi and FMh with respect to risk factor RFj .
y y y
k¼1 k¼1 k¼1 Based on the fact that the larger the synthetical deviation
ð13Þ is, the more reasonable the weights of risk factors are, a
linear programming model is established by us to calculate
T
where xSk
y ¼ ðaSk Sk Sk
y ; by ; c y Þ
and w ¼ ðw1 ; w2 ; w3 Þ is objective weights of risk factors:
weight vector of experts, it can be calculated by a weighted 8
< MaxDðxO Þ ¼ 1 Xp Xm Xm  
generation method based on the normal distribution. d x ; x xO
h¼1;h6¼i DS iy hy y
P m
Pp  1 y¼1 i¼1
Moreover, wk [ 0 and 3k¼1 wk ¼ 1. : O O
Subjcect to x
y¼1 y ¼ 1; x y
0; y ¼ 1; 2; . . .; p

• Step 3 Determine subjective weight vector of risk ð17Þ

factors.
Based on score function of PFNs, we can determine Step 6 Determine objective weight vector of risk factors.
subjective weight vector of risk factors as follows:


The solution of the above linear programming model is

Sðay Þ

S the objective weight vector of risk factors. Therefore, by

xy ¼ Pp

ð14Þ
y¼1Sðay Þ
solving the above model, we can get the objective weight

where xSy indicates the subjective weight of yth risk factor. vector of risk factors xO ¼ xOy :
1p

123
15230 C. Jin et al.

Step 7 Determine synthetical weight vector of risk Step 4 Calculate the linear combination value of WSM
factors. and WPM.
In the end, the synthetical weight vector of risk factors In order to obtain joint optimal solution, we need to
x ¼ ðxy Þ1p can be calculated as follows: calculate the linear combination value of WSM and WPM.
xy ¼ bxSy þ ð1  bÞxO ð18Þ QAi ¼ kWSMi þ ð1  kÞWPMi ð21Þ
y

where b 2 ½0; 1 is the importance coefficient of subjective where QAi indicates the obtained linear combination value
weights. of ith failure mode, and k 2 ½0; 1 is importance coefficient
of the weighted sum value.
5.3 Prioritizing risks with PF-WASPAS method Step 5 Rank failure modes.

In this section, we propose a new extension to WASPAS In order to rank failure modes, it is necessary for us to
method under PFSs environment to rank failure modes. It calculate score function value Qi of the linear combination
overcomes drawback that ranking mechanism is irrational value of each failure mode.
in the traditional FMEA method. Similar to the VIKOR Qi ¼ SðQAi Þ ð22Þ
ranking method, the WASPAS method is a compromise
If any failure modes have the same Qi values, we need to
solution method that is based on the idea of weighted sum
calculate accuracy function value, and based on Definition
method (WSM) and weighted product method (WPM).
4, the values are assigned in descending order, that is, the
Combined with PFSs, the PF-WASPAS method have an
failure mode with maximum value is ranked first. For
innate superiority in ranking failure modes, because it has
critical failure modes, it is necessary for us take the cor-
many advantages, for instance, it has simple and straight-
responding corrective measures to eliminate them.
forward process, and follows joint optimality concept for
obtaining the final rank of failure modes. Moreover, it
determines the final ranking order based on the linear
6 An illustrative example
combination of WSM and WPM and also uses PFSs to
describe uncertain preference information. The concrete
6.1 Background
process includes the following five steps.
Step 1 Determine syncretic decision matrix and weights In order to explain the implementation process of the
of risk factors. proposed FMEA method, in this paper, we take the pallet
exchange rack of a horizontal machining center produced
According to Sect. 5.2, we can obtain syncretic decision
by a large CNC machine tool manufacturing company in
matrix X ¼ ðxiy Þmp and weight vector of risk factors
China as research object. The pallet exchange rack is an
x ¼ ðxy Þ1p . important functional part of the horizontal machining
Step 2 Calculate the weighted sum value for each failure center. Its main function is to realize the position exchange
modes. between the processed workpiece and the blank to be
processed, so as to send the blank to be processed to the
The weighted sum value of ith failure mode can be processing position, and realize the automatic positioning
calculated as follows: of the workpiece. At the same time, the processed work-
p piece is sent out of the processing position. The structure of
WSMi ¼  xy xiy ð19Þ
y¼1 the pallet exchange rack is presented in Fig. 2, which
mainly includes pallet lifting mechanism driven by the
hydraulic lifting cylinder and pallet rotating mechanism
Step 3 Calculate the weighted product value for each driven by the hydraulic rotating cylinder. When working,
failure modes. the bracket 4 is lifted by the hydraulic lifting cylinder first.
Then, the rack is driven by the rotating hydraulic cylinder,
The weighted product value of ith failure mode can be
and is meshed with gear shaft 5 to realize the rotation of the
calculated as follows:
bracket 4. There are two pallets, the included angle
p
WPMi ¼  ðxiy Þxy ð20Þ between the two pallets is 180 degrees, and they can rotate
y¼1 around the mandrel 3. The two pallets are symmetrically
arranged with the mandrel 3 as the center, and when
working, the blank and the processed workpiece are,

123
An improving failure mode and effect analysis method for pallet exchange rack risk analysis 15231

company in China, so the members of the technical team

and the expert team are selected from the employees of this
company. For the technical team, it is composed of fifteen
basic technical staffs in the company, who come from five
different departments: design, manufacturing, technical
service, business and production management, each with
3–5 years of experience in CNC machine tool manufac-
turing industry. For the expert team, it is composed of
seven senior technical staffs in the company, who come
from five different teams too, but each with more than
15 years of experience.
After establishing the technical team and the expert
team, the staffs in the technical team are invited to collect
product data, which includes design drawings, design
specification and fault data of pallet exchange rack, and to
establish failure tree based on product data collected, as
shown in Fig. 3. After that, the traditional FMEA of pallet
exchange rack is carried out by the technical team, the
results are presented as Table 2. Based on Fig. 3 and
Fig. 2 Structure diagram of pallet exchange rack. 1. exchange rack, 2.
piston, 3. mandrel, 4. bracket, 5. gear shaft
Table 2, three of seven experts ek ðk ¼ 1; 2;    lÞ; l ¼ 3 are
invited to determine potential failure modes that cause
respectively, placed on these two pallets. When the pallet system failure and risk factors of pallet exchange rack. The
rotates 180 degrees, the position of the blank and the pallet exchange rack has seven potential failure modes,
processed workpiece can be exchanged. The rotating denoted by FM ¼ fFM1 ; FM2    FMm g; m ¼ 7, and three
hydraulic cylinder is a hydraulic cylinder with buffer risk factors, denoted by
 
device, and when the bracket 4 quickly rotates to the cor- RF ¼ RF1 ; RF2 ;    RFp ¼ fS; O; Dg; p ¼ 3.
rect position, the buffer device can make the rotating In the same vein, we invite three of seven experts from
bracket 4 decelerate and buffer automatically. expert team to rate risks based on Table 1, and establish
As core part of horizontal machining center, the relia- linguistic terms decision matrices

bility of pallet exchange rack has a great impact on the k k
T ¼ Tiy ; k ¼ 1; 2; 3; i ¼ 1; 2;    7; y ¼ 1; 2; 3. The
73
reliability of the horizontal machining center. Therefore, in
evaluation results are shown in Table 3. For example, the
order to improve quality and reliability of horizontal
evaluation information of failure mode FM1 with respect to
machining center, it is necessary and interesting for us to
risk factor S given by expert e1 is VVH. Then, according to
conduct a FMEA for pallet exchange rack, find out the key
Table 1, all linguistic terms will be translated into corre-
failure modes of pallet exchange rack and take corre-
sponding PFNs, and we can obtain PFNs decision matrices.
sponding corrective measures to eliminate them.
For instance, VVH ? (1,0,0). Because risk factors have
two types, benefit type (D) and cost type (S and D), we
6.2 Implementation
need to standardize PFNs decision matrices. Based on
Eq. (12), we can obtain standardized PFNs decision
In this section, we conduct a risk evaluation process based 
on the proposed FMEA method to find out key risks of matrices X k ¼ Xiyk ; k ¼ 1; 2; 3, the results are pre-
73
pallet exchange rack and take corresponding corrective sented in Table 4. For example,
measures to eliminate them, thus improve quality and H  ð0:8; 0:1; 0:05Þ ! H  ð0:05; 0:1; 0:8Þ.
reliability of pallet exchange rack. According to Fig. 1, we
can obtain that the proposed FMEA method includes three
main processes: risk evaluation, calculating the weights of Step 2 Calculating the weights of risk factors with PF-
risk factors, and prioritizing risks. Then, we will introduce LPM.
the whole implementation process in detail. In this section, we adopt PF-LPM to calculate weights of
Step 1 Risk evaluation. risk factors. It mainly includes calculating subjective
weights and calculating objective weights. In the calcula-
Considering that this horizontal machining center is tion process of subjective weights, similar to Step 1, we
produced by a large CNC machine tool manufacturing invite three of seven experts to rate risk factors, the

123
15232 C. Jin et al.

Fig. 3 Failure tree of pallet

exchange rack

Table 2 The FMEA of the pallet exchange rack

NO Failure modes Failure causes S O D RPN

FM1 The pallet was not exchanged to the correct X1: The pallet did not lift to the correct position; X2: The pallet did not 8 6 7 336
position rotate to the correct position
FM2 The speed of pallet exchange was too fast FM5; FM6 4 7 6 168
or too slow
FM3 The pallet collided with magazine door X3: Control program error 8 3 3 72
during the exchange
FM4 Pallet exchange stagnation X4: Control program error; FM7; 6 4 6 144
X5: Detection switch fault
FM5 The hydraulic flow was too large X6: The adjustment of throttle valve is unreasonable; 4 5 6 120
X7: throttle valve failure
FM6 The hydraulic flow was too small X8: The adjustment of throttle valve is unreasonable; 4 6 6 144
X9: throttle valve failure;
X10: The solenoid valve is blocked;
X11: The pipeline is blocked;
X12: The solenoid valve fault
FM7 Hydraulic failure X13: The solenoid valve is blocked; 6 7 4 168
X14: The pipeline is blocked;
X15: The solenoid valve fault

evaluation results are shown in Table 5. For example, the matrix of risk factors xS ¼ ðay Þ3 , the fused results are
evaluation information of risk factor S given by experts e1 presented in Table 5. For instance,
is H. Then, all linguistic terms will be translated into cor- a1 ¼ PFWAðxS1 S2 S3
1 ; x1 ; x1 Þ ¼ PFWAðð0:8; 0:1; 0:05Þ;
responding PFNs, so the subjective weight matrices xSk ¼ ð0:9; 0:05; 0Þ; ð0:9; 0:05; 0ÞÞ ¼ ð0:8851; 0:0574; 0Þ.
ðxSk
y Þ13 ; k ¼ 1; 2; 3 can be obtained.
According to Eq. (14), the subjective weights vector of
Based on Eq. 13, the experts’ evaluation information is risk factors can be obtained, that is,
S S
fused by PFWA operator. The weights of experts are x ¼ ð0:5458; 0:0978; 0:3564Þ. For example, x1 ¼
w¼ ð0:2; 0:45; 0:35Þ, which can be calculated by a P3jSða1 Þj ¼ j0:88510jþj0:32410:1655
j0:88510j
jþj0:65110:0732j ¼ 0:5458.
weighted generation method based on the normal distri- y¼1
j Sða y Þ j
bution. Then, we can obtain syncretic subjective weight In the calculation process of objective weights, firstly,
we need to fuse experts’ evaluation information based on

123
An improving failure mode and effect analysis method for pallet exchange rack risk analysis 15233

Table 3 Linguistic terms decision matrices we can obtain synthetical deviation about objective weights
e Risk factors Failure modes
of risk factors as follows:
  1
FM1 FM2 FM3 FM4 FM5 FM6 FM7 D xO ¼ ð13:4492xO O
1 þ 18:2079x2 þ 3:8432x3 Þ
O
6
e1 S VVH L VVH VH L L VH
where deviation
O F L VVL VL L H VH  
D H H F H H H L dDS xiy ; xhy ; i; h ¼ 1; 2;    7and i 6¼ h; y ¼ 1; 2; 3
e2 S VH F VVH H L L H
can be calculated by using the Dice similarity-based dis-
O H F VL L L H H
tance, for instance, dDS ðx11 ; x21 Þ ¼ 1
D VH H F F H H F 2ð0:91920:3826þ0:04020:4104þ00:1322Þ
e3 S VH F VVH VH L L VH ¼ 0:3753.
0:91922 þ0:04022 þ02 þ0:38262 þ0:41042 þ0:13222
O F F VL L L F H After that, we can construct a linear programming model
D VH H H H VH VH F based on Eq. (17) as follows:
8
< Max DðxO Þ ¼ 1 ð13:4492xO þ 18:2079xO þ 3:8432xO Þ
1 2 3
6
: Subjcect to P3 xO ¼ 1; xO
0; y ¼ 1; 2; 3
y¼1 y y
Eq. (15) to obtain the syncretic decision matrix
X ¼ ðxiy Þ73 . The fused results are presented in Table 8 of By solving the above linear programming model, we can
Appendix A. For example, obtain objective weight vector of risk factors.
x11 ¼ PFWGðx111 ; x211 ; x311 Þ xO ¼ ð0:3788; 0:5129; 0:1083Þ
¼ PFWGðð0; 0; 1Þ; ð0:9; 0:05; 0Þ; ð0:9; 0:05; 0ÞÞ
Finally, based on Eq. (18), we can assume that b = 0.5, and
¼ ð0:9192; 0:0402; 0Þ
obtain the synthetical weight vector of risk factors
Then, we need to calculate synthetical deviation and
xS ¼ ð0:4623; 0:3054; 0:2323Þ
establish linear programming model. Based on Eq. (16),

Table 4 Standardized PFNs decision matrices

e Risk Failure modes
factors
FM1 FM2 FM3 FM4 FM5 FM6 FM7

e1 S (0, 0, 1) (0.2, 0.45, 0.25) (1, 0, 0) (0.9, 0.05, 0) (0.2, 0.45, 0.25) (0.2, 0.45, 0.25) (0.9, 0.05, 0)
O (0.45, 0.4, 0.1) (0.2, 0.45, 0.25) (0, 0, 1) (0, 0.05, 0.9) (0.2, 0.45, 0.25) (0.8, 0.1, 0.05) (0.9, 0.05, 0)
D (0.05, 0.1, 0.8) (0.05, 0.1, 0.8) (0.1, 0.4, 0.45) (0.05, 0.1, 0.8) (0.05, 0.1, 0.8) (0.05, 0.1, 0.8) (0.25, 0.45, 0.2)
e2 S (0.9, 0.05, 0) (0.45, 0.4, 0.1) (1, 0, 0) (0.8, 0.1, 0.05) (0.2, 0.45, 0.25) (0.2, 0.45, 0.25) (0.8, 0.1, 0.05)
O (0.8, 0.1, 0.05) (0.45, 0.4, 0.1) (0, 0.05, 0.9) (0.2, 0.45, 0.25) (0.2, 0.45, 0.25) (0.8, 0.1, 0.05) (0.8, 0.1, 0.05)
D (0, 0.05, 0.9) (0.05, 0.1, 0.8) (0.1, 0.4, 0.45) (0.1, 0.4, 0.45) (0.05, 0.1, 0.8) (0.05, 0.1, 0.8) (0.1, 0.4, 0.45)
e3 S (0.9, 0.05, 0) (0.45, 0.4, 0.1) (1, 0, 0) (0.9,0.05, 0) (0.2, 0.45, 0.25) (0.2, 0.45, 0.25) (0.9, 0.05, 0)
O (0.45, 0.4, 0.1) (0.45, 0.4, 0.1) (0, 0.05, 0.9) (0.2, 0.45, 0.25) (0.2, 0.45, 0.25) (0.45, 0.4, 0.1) (0.8, 0.1, 0.05)
D (0, 0.05, 0.9) (0.05, 0.1, 0.8) (0.05, 0.1, 0.8) (0.05, 0.1, 0.8) (0, 0.05, 0.9) (0, 0.05, 0.9) (0.1, 0.4, 0.45)

Table 5 The subjective weight

Experts Risk factors
matrices and syncretic
subjective weight matrix of risk S O D S O D
factors
e1 H L F (0.8, 0.1, 0.05) (0.2, 0.45, 0.25) (0.45, 0.4, 0.1)
e2 VH F H (0.9, 0.05, 0) (0.45, 0.4, 0.1) (0.8, 0.1, 0.05)
e3 VH L F (0.9, 0.05, 0) (0.2, 0.45, 0.25) (0.45, 0.4, 0.1)
ay (0.8851, 0.0574, 0) (0.3241, 0.4268, 0.1655) (0.6511, 0.2144, 0.0732)
xSy 0.5458 0.0978 0.3564

123
15234 C. Jin et al.

the exchange rack are controlled by the hydraulic system

and the limit switch, therefore, we can take the following
Step 3 Prioritizing risks with PF-WASPAS method. measures to eliminate it: (1) Checking the limit switch
In this section, the PF-WASPAS method will be utilized regularly; (2) Checking the solenoid valve regularly and
to ranking failure modes of pallet exchange rack. pipeline, and timely replace the defective pipeline; (3)
According to the above calculation, we have obtain the Debugging the program before machining; (4) Ensuring
syncretic decision matrix X ¼ ðxiy Þ73 and the synthetical accurate engagement of gear and rack.
weight vector of risk factors
xS ¼ ð0:4623; 0:3054; 0:2323Þ:
Then, based on Eqs. (19) and (20), we can calculate the
7 Discussion and comparison
weighted sum value and the weighted product value of
In this section, a discussion and comparison process is
each failure mode, for instance,
3
carried out by us to verify the rationality and effectiveness
WSM1 ¼  xy x1y ¼ 0:4623ð0:9192; 0:0402; 0Þ of the proposed method, it mainly includes three parts:
y¼1
sensitivity analysis of different distance measures, com-
0:3054ð0:583; 0:2799; 0:0778Þ 0:2323ð0; 0:0602;
parison analysis of different ranking methods, and man-
0:8851Þ ¼ ð0:7607; 0:0799; 0Þ,
3
agerial implications.
WPM1 ¼  ðx1y Þxy ¼ ð0:9192; 0:0402; 0Þ0:4623
y¼1
7.1 Sensitivity analysis of different distance
ð0:583; 0:2799; 0:0778Þ0:3054 ð0; 0:0602; measures
0:2323
0:8851Þ ¼ ð0; 0:1251; 0:4099Þ ,The calculated
results are shown in Table 9 of Appendix A. Based on According to Sect. 2, we know that many distance measure
Eq. (21), we can assume that k = 0.5, and obtain the linear methods have been proposed to describe the differences
combination value of WSM and WPM. For instance, between two PFSs. When different distance measures are
QA1 ¼ 0:5ð0:7607; 0:0799; 0Þ þ ð1 utilized to calculate the synthetical deviation, we can
 0:5Þ ð0; 0:1251; 0:4099Þ obtain different weights of risk factors, which may lead to
¼ ð0:5108; 0:1; 0Þ different ranking results. Therefore, it is necessary and
interesting for us to explore the influence of different dis-
The calculated results of other failure modes are shown tance measures on failure modes’ ranking.
in Table 9 of Appendix A. In order to conduct sensitivity analysis of different dis-
At last, the score function value of the linear combina- tance measures, the traditional distance, the Ganie’s dis-
tion value of each failure mode can be calculated based on tance, the Singh’s distance and the proposed Dice
Eq. (22), and based on these values, the failure modes are similarity-based distance are used by us to calculate
assigned in descending order, that is, the failure mode with weights of risk factors, and rank failure modes based on the
maximum value is ranked first. Therefore, the final ranking above case. The concept of the Ganie’s distance was first
results of all failure modes are proposed by Ganie in (2021), it is defined as follows:
FM1 [ FM7 [ FM6 [ FM5 [ FM2 [ FM4 [ FM3.
According to the ranking results, we can obtain that Definition 13 Let T1 ¼ ha1 ðxi Þ; b1 ðxi Þ; c1 ðxi Þi and T2 ¼
failure mode FM1 should be given the highest priority for ha2 ðxi Þ; b2 ðxi Þ; c2 ðxi Þi be two PFSs in a given finite set
risk mitigation and safety improvement. Based on Table 2, X ¼ fx1 ; x2 ;    xn g. Then, the Ganie’s distance measure
the reasons that cause failure mode FM1 mainly include the between T1 and T2 can be obtained as follows:
pallet did not lift to the correct position and the pallet did
not rotate to the correct position. The rotation and lifting of

2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3

3 a1 ðxi Þa2 ðxi Þ þ 3 b1 ðxi Þb2 ðxi Þ þ 3 c1 ðxi Þc2 ðxi Þ þ 3 p1 ðxi Þp2 ðxi Þ
6 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 7
n 6 þ ð1  a1 ðxi Þ  b1 ðxi ÞÞð1  a2 ðxi Þ  b2 ðxi ÞÞ 7
1 X 6 7
DG ðT1 ; T2 Þ ¼ 1  6 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 7 ð23Þ
4n i¼1 6 þ ð1  a ðx Þ  c ðx ÞÞð1  a ðx Þ  c ðx ÞÞ 7
4 1 i 1 i 2 i 2 i 5
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
þ ð1  b1 ðxi Þ  c1 ðxi ÞÞð1  b2 ðxi Þ  c2 ðxi ÞÞ

123
An improving failure mode and effect analysis method for pallet exchange rack risk analysis 15235

The concept of the Singh’s distance was proposed by that the proposed Dice similarity-based distance is better
Singh in 2018 (Singh et al. 2018), it is defined as follows: than the Ganie’s distance and the Singh’s distance, and
these three distance measures are better than the traditional
Definition 14 Let T1 ¼ ha1 ðxi Þ; b1 ðxi Þ; c1 ðxi Þi and T2 ¼
distance.
ha2 ðxi Þ; b2 ðxi Þ; c2 ðxi Þi be two PFSs in a given finite set
X ¼ fx1 ; x2 ;    xn g. Then, the Singh’s distance measure
7.2 Comparison analysis of different ranking
between T1 and T2 can be obtained as follows:
" methods
1 Xn 
DS ðT1 ; T2 Þ ¼ 1  max ja1 ðxi Þ  a2 ðxi Þj2 ;
4n i¼1 When different ranking methods are utilized to rank failure
i12 modes of the pallet exchange rack, we can obtain different
jb1 ðxi Þ  b2 ðxi Þj2 ; jc1 ðxi Þ  c2 ðxi Þj2 ; jp1 ðxi Þ  p2 ðxi Þj2 ranking results. Some of these ranking results are quite
different, and some are almost consistent. In order to verify
ð24Þ
the rationality and effectiveness of the proposed method, in
this section, a comparison analysis with respect to the
traditional RPN, the intuitionistic fuzzy-linear program-
The final ranking results of these four distance measures ming model-multi-attributive border approximation area
are shown in Table 6. In order to provide a more visual comparison (IF-LPM-MABAC) method which calculates
comparison, a line chart is utilized by us to describe the weights of risk factors with LPM, and ranks failure modes
final ranking results, as shown in Fig. 4. with MABAC method (Liu et al. 2019c), the spherical
According to Table 6 and Fig. 4, we can obtain that the fuzzy-WASPAS (SF-WASPAS) method (Aydogdu and
ranking results of the proposed Dice similarity-based dis- Gul 2020) which calculates weights of risk factors with
tance are highly related with those of the Ganie’s distance entropy measure, and ranks failure modes with MABAC
and the Singh’s distance, but have great difference with method and the proposed FMEA method is conducted.
those of the traditional distance. For example, except for Based on the above case, the ranking results of these four
FM3 and FM5, the ranking results of the Ganie’s distance methods are shown in Table 7. Similar to Sect. 7.1, a line
and the proposed Dice similarity-based distance are iden- chart is utilized by us to describe the final ranking results,
tical. In the same vein, except for FM1 and FM7, the as presented in Fig. 5.
ranking results of the Singh’s distance and the proposed The comparison of the ranking results mainly includes
Dice similarity-based distance are identical. However, for three aspects. Firstly, in order to find the statistical sig-
the traditional distance, except for FM5 and FM7, the nificance of the difference between the ranking results
ranking results of the traditional distance and the proposed obtained using the proposed FMEA method and the ones
Dice similarity-based distance are completely different. obtained from the traditional RPN method, the IF-LPM-
Moreover, the first-ranked and the last-ranked failure MABAC method and the SF-WASPAS method, the
modes of the proposed Dice similarity-based distance are Spearman rank correlation coefficient (SRCC) is utilized
consistent with traditional RPN method, that is FM1 and by us. It was first proposed by Raju in 1999, and was
FM3. Therefore, based on above analysis, we can obtain defined as follows:

Table 6 Sensitivity analysis results

Failure modes The traditional distance The Ganie’s distance The Singh’s distance The Dice similarity-based distance
Qi Ranking Qi Ranking Qi Ranking Qi Ranking

FM1 0.0231 3 0.6352 1 0.6237 2 1 1

FM2 - 0.5581 7 - 0.0024 5 0.2206 5 0.0018 5
FM3 0.6328 1 0.1524 4 - 0.0926 7 - 0.3162 7
FM4 - 0.2891 5 - 0.2475 6 0.0045 6 - 0.1215 6
FM5 - 0.1287 4 - 0.3377 7 0.3521 4 0.1403 4
FM6 - 0.3872 6 0.2647 3 0.4375 3 0.5108 3
FM7 0.3219 2 0.4926 2 0.8457 1 0.5834 2

123
15236 C. Jin et al.

Fig. 4 Sensitivity analysis results

Fig. 5 Comparison analysis results
P
m
6 D2i method, which is mainly reflected in the following two
i¼1
R¼1 ð25Þ aspects: (1) In the traditional RPN method, the ranking of
mðm2  1Þ
failure modes FM2 and FM7 are identical, because they
where Di is the difference of ith failure mode’s ranking in have the same RPN value 168. However, in real life, FM7
these two methods and R indicates the Spearman rank should be given higher priority than FM2, that is because
correlation coefficient value. the main power of pallet exchange rack comes from
The calculated results are presented in Table 7. Based on hydraulic system, the hydraulic system failure will paralyze
Table 7, we can obtain that the proposed FMEA method is the whole pallet exchange device. While the speed of pallet
highly related with the IF-LPM-MABAC method (0.897) exchange was too fast or too slow have little influence on
and the SF-WASPAS method (0.856), but have great dif- the lifting and rotating motion of the exchange rack. This
ference with the traditional RPN method (0.427). More- problem is solved by the proposed FMEA method per-
over, in these four ranking methods, the first-ranked and fectly. Similarly, FM6 should be given higher priority than
the last-ranked failure modes are same, that is, FM1 and FM4; (2) In the traditional FMEA method, the failure mode
FM3. Therefore, based on the above rough comparison, we FM5 ranks the sixth, the failure mode FM2 ranks the sec-
can obtain that the proposed FMEA method is rational. ond. However, in the proposed FMEA method, the failure
Secondly, by comparing ranking results of the proposed mode FM5 ranks the fourth, the failure mode FM2 ranks the
FMEA method and the traditional RPN method, we can fifth, which indicates that FM5 should be given higher
obtain the proposed FMEA method have innate superiority priority than FM2, and this ranking result is consistent with
in overcoming the drawbacks of the traditional RPN those of the other two methods.

Table 7 Comparison analysis results

Failure modes The traditional RPN IF-LPM-MABAC SF-WASPAS The proposed method
S O D RPNi Ranking RPVi Ranking Qi Ranking Qi Ranking

FM1 8 6 7 336 1 0.798 1 0.7665 1 1 1

FM2 4 7 6 168 2 - 0.015 5 0.0017 6 0.0018 5
FM3 8 3 3 72 7 - 0.337 7 - 0.3276 7 - 0.3162 7
FM4 6 4 6 144 4 - 0.200 6 0.4912 3 - 0.1215 6
FM5 4 5 6 120 6 0.516 3 0.1107 4 0.1403 4
FM6 4 6 6 144 4 0.359 4 0.0328 5 0.5108 3
FM7 6 7 4 168 2 0.687 2 0.5487 2 0.5834 2
K 0.427 0.897 0.856 1

123
An improving failure mode and effect analysis method for pallet exchange rack risk analysis 15237

Finally, compared with the IF-LPM-MABAC method employees of the company, we give the following five
and the SF-WASPAS method, by comparing the IF-LPM- suggestions:
MABAC method and the proposed FMEA method, based
(1) Conducting a training in the company to improve the
on Table 7 and Fig. 5, we can obtain that the ranking
quality and reliability awareness of employees. For
results of the proposed FMEA method are consistent with
an enterprise, only when the quality and reliability
the IF-LPM-MABAC method, except for FM5 and FM6.
awareness of each employee is improved, the quality
The reasons that lead to this phenomenon mainly include
and reliability of products can be improved. There-
the following three aspects: (1) IFSs and PFSs are utilized
fore, it is necessary for managers to conduct a
to rate risks in the IF-LPM-MABAC method and the pro-
training with respect to the quality and reliability of
posed FMEA method, and PFSs is an extension of FSs and
products.
IFSs, which can overcome the drawback that the risks are
(2) Establishing a professional quality and reliability
rated with crisp values, which cannot express the uncer-
team in the company. In the proposed FMEA
tainty and hesitation of experts’ evaluation information in
method, we need to form a technical team and an
traditional RPN method; (2) The LPM method is utilized to
expert team to rate risks. To facilitate implementa-
calculate the weights of risk factors in these two methods,
tion, the members of the technical team and expert
which can overcome the drawback that three risk factors
team can be selected from the quality and reliability
are equally weighted in the traditional RPN method; (3)
team.
The ranking mechanism of these two methods are different,
(3) Arranging the staff to formulate corresponding
one is MABAC method, the other is WASPAS method,
corrective measures to eliminate or alleviate failure
both theses two ranking methods can overcome drawback
of products based on the ranking results and the
that the ranking mechanism is irrational in the traditional
actual situation of the company, and invite profes-
RPN method. In the same vein, by comparing the SF-
sional technicians to carry out these formulated
WASPAS method and the proposed FMEA method, based
corrective measures.
on Table 7 and Fig. 5, we can obtain that the ranking
(4) For the important links in the production process, the
results of these two methods have slight differences, for
quality and reliability warning or early warning
instance, FM2, FM4 and FM6. The reasons that lead to this
should be set up. The establishment of quality and
phenomenon mainly include the following three aspects:
reliability warning or early warning can remind
(1) SFSs and PFSs are utilized to rate risks in the SF-
employees to pay attention to important links in the
WASPAS method and the proposed FMEA method; (2)
production process.
The LPM method is utilized to calculate the weights of risk
(5) Regularly carrying out quality and reliability com-
factors in the proposed FMEA method, while the SF-
munication activities in the company. The improve-
WASPAS method calculates weights of risk factors with
ment of the quality and reliability of a product is the
entropy measure; (3) The WASPAS method is utilized to
result of the cooperation of various departments, so
rank failure modes in these two methods. The above
carrying out such activities is conducive to the
mentioned aspects can overcome the drawbacks of the
communication between various departments.
traditional RPN method, too. Therefore, based on the above
comparison analysis, we can obtain the proposed FMEA Therefore, the proposed FMEA method is not only
method is rational and effective in ranking failure modes rational and effective, but also applicable. As long as
and overcome the drawbacks of the traditional RPN managers manage properly, this method has innate
method. advantages in improving the quality and reality of products
and enterprise competitiveness.
7.3 Managerial implications

In order to let the proposed FMEA method be better

implemented in the enterprise, for the managers and

123
15238 C. Jin et al.

8 Conclusions

(0.8535, 0.0728, 0.0228)

(0.8191, 0.0902, 0.0402)
(0.1201, 0.4104, 0.4072)
In this paper, we propose a new risk evaluation method
based on PF-LPM and PF-WASPAS. Moreover, in order to
calculate the difference between two PFSs, some new
distance measures based on the Dice similarity and Jaccard

FM7
similarity are proposed. In the proposed FMEA method,
firstly, PFSs are utilized to rate risks. Secondly, the PFWA

(0.6541, 0.2191, 0.0678)

operator and PFWG operator are utilized to fuse experts’
evaluation information. The synthetical deviation of eval-

(0, 0.0828, 0.831)

(0.2, 0.45, 0.25)
uation information is calculated by using the Dice simi-
larity-based distance, and the weights of risk factors are
calculated with the PF-LPM method. Then, the failure
modes are ranked with the PF-WASPAS method. At last,

FM6
we take the pallet exchange rack of a horizontal machining

(0, 0.0828, 0.8431)

center produced by a large CNC machine tool manufac-

(0.2, 0.45, 0.25)

(0.2, 0.45, 0.25)
turing company in China as research object to explain the
implementation process of the proposed FMEA method.
In order to verify the rationality and effectiveness of the
proposed method, a discussion with respect to sensitivity

FM5
analysis of different distance measures, comparison anal-

(0.8535, 0.0728, 0.0228)

(0.0683, 0.2501, 0.6847)

ysis of different ranking methods and managerial impli-
cations is conducted. The results of sensitivity analysis

(0, 0.3865, 0.4988)

indicate that the proposed Dice similarity-based distance is
better than the Ganie’s distance and the Singh’s distance,
and these three distance measures are better than the tra-
ditional distance. For comparison analysis of the traditional

FM4
RPN method, the IF-LPM-MABAC method, the SF-
WASPAS method and the proposed FMEA method, the

(0.0785, 0.3085, 0.614)

comparison results indicate that the proposed FMEA
method is rational and valid in ranking failure modes. In
order to let the proposed FMEA method be better imple-
(0, 0.0402, 1)

mented in the enterprise, we give a managerial implications

(1, 0, 0)

process. In the future, we will explore interval valued form

FM3

of PFSs and use them to rate risks. Moreover, we will

explore the combination of interval value form of PFS with
(0.3826, 0.4104, 0.1322)
(0.3826, 0.4104, 0.1322)

other weight calculation methods and ranking methods.

(0.05, 0.1, 0.8)

Appendix A

See Tables 8 and 9.

FM2
Table 8 The syncretic decision matrix

(0.583, 0.2799, 0.0778)

(0.9192, 0.0402, 0)

(0, 0.0602, 0.8851)

Failure modes

FM1
factors
Risk

O
D
S

123
An improving failure mode and effect analysis method for pallet exchange rack risk analysis 15239

Acknowledgements This work was financially supported by the

(0.6679, 0.1406, 0..0845)

(0.5345, 0.1702, 0.1347)
(0.7631, 0.1162, 0.053)
National Natural Science Foundation, China (No. 51835001;
51705048); the National Major Scientific and Technological Special
Project for ‘‘High-grade CNC and Basic Manufacturing Equipment’’,
China (2018ZX04032-001; 2019ZX04005-001).

Declarations
0.5834
FM7

2
Conflict of interest The authors declare that they have no known
(0.7607, 0.0799, 0)
(0, 0.1251, 0.4099)

competing financial interests or personal relationships that could have

appeared to influence the work reported in this paper.
(0.5108, 0.1, 0)

Ethical approval This article does not contain any studies with human
participants or animals performed by any of the authors.
0.5108
FM6

Informed consent Informed consent was obtained from all individual

3

participants included in the study.

(0.3638, 0.1893, 0.2235)
(0.5953, 0.1615, 0.129)
(0, 0.222, 0.3872)

References
Aydogdu A, Gul S (2020) A novel entropy proposition for spherical
0.1403

fuzzy sets and its application in multiple attribute decision-

FM5

making. Int J Intell Syst 35(9):1354–1374. https://doi.org/10.

4

1002/int.22256
Boral S, Howard I, Chaturvedi SK, McKee K, Naikan VNA (2020a)
(0.3478, 0.2438, 0.2226)

(0.1924, 0.2752, 0.3139)

An integrated approach for fuzzy failure modes and effects

(0, 0.3106, 0.4427)

analysis using fuzzy AHP and fuzzy MAIRCA. Eng Fail Anal
108:16. https://doi.org/10.1016/j.engfailanal.2019.104195
Boral S, Howard I, Chaturvedi SK, McKee K, Naikan VNA (2020b)
A novel hybrid multi-criteria group decision making approach
- 0.1215

for failure mode and effect analysis: an essential requirement for

sustainable manufacturing. Sust Prod Consum 21:14–32. https://
FM4

doi.org/10.1016/j.spc.2019.10.005
6

Chang CH, Kontovas C, Yu Q, Yang ZL (2021) Risk assessment of

(0.1574, 0.3037, 0.3316)

the operations of maritime autonomous surface ships. Reliab Eng

(0.0821, 0.34, 0.3983)

Syst Saf 207:11. https://doi.org/10.1016/j.ress.2020.107324

(0, 0.3806, 0.4785)

Chen LY, Deng Y (2018) A new failure mode and effects analysis
model using Dempster-Shafer evidence theory and grey rela-
tional projection method. Eng Appl Artif Intell 76:13–20. https://
- 0.3162

doi.org/10.1016/j.engappai.2018.08.010
Chen YF, Ran Y, Wang ZC, Li XL, Yang X, Zhang GB (2020) An
FM3

extended MULTIMOORA method based on OWGA operator

7

and Choquet integral for risk prioritization identification of

(0.3176, 0.2956, 0.2009)
(0.2385, 0.3495, 0.3829)
(0.2791, 0.3214, 0.2773)

failure modes. Eng Appl Artif Intell 91:12. https://doi.org/10.

1016/j.engappai.2020.103605
Chen YF, Ran Y, Wang ZC, Zhang GB (2021) Research on global
coupling analysis and approximate decoupling of quality char-
acteristics in the early design stage. Proc Inst Mech Eng Part C J
Mech Eng Scis. https://doi.org/10.1177/0954406220934844
Dadsena KK, Sarmah SP, Naikan VNA (2019) Risk evaluation and
0.0018
FM2

mitigation of sustainable road freight transport operation: a case

of trucking industry. Int J Prod Res 57(19):6223–6245. https://
5

doi.org/10.1080/00207543.2019.1578429
Failure modes

(0, 0.0935, 1)

Deulgaonkar VR, Ingolikar N, Borkar A, Ghute S, Awate N (2021)

Failure analysis of diesel engine piston in transport utility
(1, 0, 0)

(1, 0, 0)
Table 9 Ranking results

vehicles. Eng Fail Anal 120:10. https://doi.org/10.1016/j.engfai

FM1

lanal.2020.105008
1
1

Dice LR (1945) Measures of the amount of ecologic association

Ranking results
Ranking values

between species. Ecology 26(3):297–302. https://doi.org/10.

2307/1932409
Duong TTT, Thao NX (2021) 1 A novel dissimilarity measure on
picture fuzzy sets and its application in multi-criteria decision
WPMi
WSMi

QAi
Qi

123
15240 C. Jin et al.

making. Soft Comput 25(1):15–25. https://doi.org/10.1007/ Liu HC, You JX, Duan CY (2019c) An integrated approach for failure
s00500-020-05405-6 mode and effect analysis under interval-valued intuitionistic
Dutta P (2018) Medical diagnosis based on distance measures fuzzy environment. Int J Prod Econ 207:163–172. https://doi.org/
between picture fuzzy sets. Int J Fuzzy Syst Appl 7:15–36. 10.1016/j.ijpe.2017.03.008
https://doi.org/10.4018/IJFSA.2018100102 Lo H-W, Liou JJH, Huang C-N, Chuang Y-C (2019) A novel failure
Fang H, Li J, Song WY (2020) Failure mode and effects analysis: an mode and effect analysis model for machine tool risk analysis.
integrated approach based on rough set theory and prospect Reliab Eng Syst Saf 183:173–183. https://doi.org/10.1016/j.ress.
theory. Soft Comput 24(9):6673–6685. https://doi.org/10.1007/ 2018.11.018
s00500-019-04305-8 Luo SZ, Liang WZ (2019) Optimization of roadway support schemes
Ganie AH, Singh S (2021) An innovative picture fuzzy distance with likelihood-based MABAC method. Appl Soft Comput
measure and novel multi-attribute decision-making method. 80:80–92. https://doi.org/10.1016/j.asoc.2019.03.020
Complex Intell Syst. https://doi.org/10.1007/s40747-020-00235- Luo SZ, Liang WZ, Zhao GY (2020) Likelihood-based hybrid
3 ORESTE method for evaluating the thermal comfort in under-
Gul M, Yucesan M, Celik E (2020) A manufacturing failure mode ground mines. Appl Soft Comput. https://doi.org/10.1016/j.asoc.
and effect analysis based on fuzzy and probabilistic risk analysis. 2019.105983
Appl Soft Comput 96:12. https://doi.org/10.1016/j.asoc.2020. Mardani A, Saraji MK, Mishra AR, Rani P (2020) A novel extended
106689 approach under hesitant fuzzy sets to design a framework for
Hu LH, Kang R, Pan X, Zuo DJ (2020a) Risk assessment of uncertain assessing the key challenges of digital health interventions
random system-Level-1 and level-2 joint propagation of uncer- adoption during the COVID-19 outbreak. Appl Soft Comput
tainty and probability in fault tree analysis. Reliab Eng Syst Saf 96:14. https://doi.org/10.1016/j.asoc.2020.106613
198:10. https://doi.org/10.1016/j.ress.2020.106874 Marhavilas PK, Filippidis M, Koulinas GK, Koulouriotis DE (2020)
Hu Y, Gou L, Deng X, Jiang W (2020b) Failure mode and effect An expanded HAZOP-study with fuzzy-AHP (XPA-HAZOP
analysis using multi-linguistic terms and Dempster-Shafer technique): application in a sour crude-oil processing plant. Saf
evidence theory. Qual Reliab Eng Int. https://doi.org/10.1002/ Sci 124:14. https://doi.org/10.1016/j.ssci.2019.104590
qre.2773 Mohsen O, Fereshteh N (2017) An extended VIKOR method based on
Jaccard P (1901) Distribution de la ore alpine dans la Bassin de entropy measure for the failure modes risk assessment: a case
Dranses et dans quelques regions voisines. Bulletin de la Société study of the geothermal power plant (GPP). Saf Sci 92:160–172.
Vaudoise des. Sci Nat 7(140):241–272. https://doi.org/10.5169/ https://doi.org/10.1016/j.ssci.2016.10.006
seals-266440 Nie RX, Tian ZP, Wang XK, Wang JQ, Wang TL (2018) Risk
Jana C, Senapati T, Pal M, Yager RR (2019) Picture fuzzy Dombi evaluation by FMEA of supercritical water gasification system
aggregation operators: application to MADM process. Appl Soft using multi-granular linguistic distribution assessment. Knowl
Comput 74:99–109. https://doi.org/10.1016/j.asoc.2018.10.021 Based Syst 162:185–201. https://doi.org/10.1016/j.knosys.2018.
Jin CX, Ran Y, Wang ZC, Huang GQ, Xiao LM, Zhang GB (2020) 05.030
Reliability analysis of gear rotation meta-action unit based on Pamucar D, Deveci M, Canitez F, Lukovac V (2020) Selecting an
Weibull and inverse Gaussian competing failure process. Eng airport ground access mode using novel fuzzy LBWA-WAS-
Fail Anal 117:16. https://doi.org/10.1016/j.engfailanal.2020. PAS-H decision making model. Eng Appl Artif Intell 93:20.
104953 https://doi.org/10.1016/j.engappai.2020.103703
Ju YB, Ju DW, Gonzalez E, Giannakis M, Wang AH (2019) Study of Purba JH, Tjahyani DTS, Widodo S, Ekariansyah AS (2020) Fuzzy
site selection of electric vehicle charging station based on probability based event tree analysis for calculating core damage
extended GRP method under picture fuzzy environment. Comput frequency in nuclear power plant probabilistic safety assessment.
Ind Eng 135:1271–1285. https://doi.org/10.1016/j.cie.2018.07. Prog Nucl Energy 125:9. https://doi.org/10.1016/j.pnucene.2020.
048 103376
Khan MJ, Kumam P, Deebani W, Kumam W, Shah Z (2020) Bi- Rezaee MJ, Yousefi S, Valipour M, Dehdar MM (2018) Risk analysis
parametric distance and similarity measures of picture fuzzy sets of sequential processes in food industry integrating multi-stage
and their applications in medical diagnosis. Egypt Inform J. fuzzy cognitive map and process failure mode and effects
https://doi.org/10.1016/j.eij.2020.08.002 analysis. Comput Ind Eng 123:325–337. https://doi.org/10.1016/
Li J, Fang H, Song W (2019a) Failure mode and effects analysis using j.cie.2018.07.012
variable precision rough set theory and TODIM method. IEEE Singh P, Mishra NK, Kumar M, Saxena S, Singh V (2018) Risk
Trans Reliab 68(4):1242–1256. https://doi.org/10.1109/tr.2019. analysis of flood disaster based on similarity measures in picture
2927654 fuzzy environment. Afr Mat 29(7–8):1019–1038. https://doi.org/
Li X, Ju YB, Ju DW, Zhang WK, Dong PW, Wang AH (2019b) 10.1007/s13370-018-0597-x
Multi-attribute group decision making method based on EDAS Son LH (2017) Measuring analogousness in picture fuzzy sets: from
under picture fuzzy environment. IEEE Access picture distance measures to picture association measures. Fuzzy
7:141179–141192. https://doi.org/10.1109/access.2019.2943348 Optim Decis Making 16(3):359–378. https://doi.org/10.1007/
Li YL, Wang R, Chin KS (2019c) New failure mode and effect s10700-016-9249-5
analysis approach considering consensus under interval-valued Tian C, Peng JJ, Zhang S, Zhang WY, Wang JQ (2019) Weighted
intuitionistic fuzzy environment. Soft Comput picture fuzzy aggregation operators and their applications to
23(22):11611–11626. https://doi.org/10.1007/s00500-018- multi-criteria decision-making problems. Comput Ind Eng
03706-5 137:12. https://doi.org/10.1016/j.cie.2019.106037
Liu H-C, Hu Y-P, Wang J-J, Sun M (2019a) Failure mode and effects Van Dinh N, Xuan Thao N (2018) Some measures of picture fuzzy
analysis using two-dimensional uncertain linguistic variables and sets and their application in multi-attribute decision making. Int J
alternative queuing method. IEEE Trans Reliab 68(2):554–565. Math Sci Comput 4(3):23–41. https://doi.org/10.5815/ijmsc.
https://doi.org/10.1109/tr.2018.2866029 2018.03.03
Liu M, Zeng S, Balezentis T, Streimikiene D (2019b) Picture fuzzy Velasquez RMA (2020) Root cause analysis for inverters in solar
weighted distance measures and their application to investment photo-voltaic plants. Eng Fail Anal 118:18. https://doi.org/10.
selection. Amfiteatru Econ 21(52):682–695 1016/j.engfailanal.2020.104856

123
An improving failure mode and effect analysis method for pallet exchange rack risk analysis 15241

Wang L, Zhang HY, Wang JQ, Li L (2018a) Picture fuzzy normalized Zeng S, Asharf S, Arif M, Abdullah S (2019) Application of
projection-based VIKOR method for the risk evaluation of exponential jensen picture fuzzy divergence measure in multi-
construction project. Appl Soft Comput 64:216–226. https://doi. criteria group decision making. Mathematics 7(2):191. https://
org/10.1016/j.asoc.2017.12.014 doi.org/10.3390/math7020191
Wang WZ, Liu XW, Qin Y, Fu Y (2018b) A risk evaluation and Zhang HJ, Dong YC, Palomares-Carrascosa I, Zhou HW (2019)
prioritization method for FMEA with prospect theory and Failure mode and effect analysis in a linguistic context: a
Choquet integral. Saf Sci 110:152–163. https://doi.org/10.1016/ consensus-based multiattribute group decision-making approach.
j.ssci.2018.08.009 IEEE Trans Reliab 68(2):566–582. https://doi.org/10.1109/tr.
Wei G (2017) Picture fuzzy aggregation operators and their appli- 2018.2869787
cation to multiple attribute decision making. J Intell Fuzzy Syst Zhang PW, Tao ZF, Liu JP, Jin FF, Zhang JT (2020) An ELECTRE
33(2):713–724. https://doi.org/10.3233/jifs-161798 TRI-based outranking approach for multi-attribute group deci-
Wei G, Gao H (2018) The generalized dice similarity measures for sion making with picture fuzzy sets. J Intell Fuzzy Syst
picture fuzzy sets and their applications. Informatica 38(4):4855–4868. https://doi.org/10.3233/jifs-191540
29(1):107–124. https://doi.org/10.15388/Informatica.2018.160 Zhao H, You J-X, Liu H-C (2016) Failure mode and effect analysis
Wu SM, You XY, Liu HC, Wang LE (2020) Improving quality using MULTIMOORA method with continuous weighted
function deployment analysis with the cloud MULTIMOORA entropy under interval-valued intuitionistic fuzzy environment.
method. Int Trans Oper Res 27(3):1600–1621. https://doi.org/10. Soft Comput 21(18):5355–5367. https://doi.org/10.1007/s00500-
1111/itor.12484 016-2118-x
Xu Y, Shang XP, Wang J, Zhang RT, Li WZ, Xing YP (2019) A Zheng H, Tang Y (2020) A novel failure mode and effects analysis
method to multi-attribute decision making with picture fuzzy model using triangular distribution-based basic probability
information based on Muirhead mean. J Intell Fuzzy Syst assignment in the evidence theory. IEEE Access
36(4):3833–3849. https://doi.org/10.3233/jifs-172130 8:66813–66827. https://doi.org/10.1109/access.2020.2986807
Yi PT, Dong QK, Li WW (2019) Evaluation of city sustainability
using the deviation maximization method. Sustain Cities Soc Publisher’s Note Springer Nature remains neutral with regard to
50:8. https://doi.org/10.1016/j.scs.2019.101529 jurisdictional claims in published maps and institutional affiliations.
Yousefi S, Alizadeh A, Hayati J, Baghery M (2018) HSE risk
prioritization using robust DEA-FMEA approach with undesir-
able outputs: a study of automotive parts industry in Iran. Saf Sci
102:144–158. https://doi.org/10.1016/j.ssci.2017.10.015

123