See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/257119220

A framework for assessing poka-yoke devices

Article  in  Journal of Manufacturing Systems · July 2012

DOI: 10.1016/j.jmsy.2012.04.001

CITATIONS READS

57 3,476

3 authors:

Tarcisio Abreu Saurin Jose Luis Duarte Ribeiro

Universidade Federal do Rio Grande do Sul Universidade Federal do Rio Grande do Sul
222 PUBLICATIONS   4,042 CITATIONS    316 PUBLICATIONS   5,140 CITATIONS   

SEE PROFILE SEE PROFILE

Gabriel Vidor
Universidade de Caxias do Sul (UCS)
55 PUBLICATIONS   322 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Environmentally sustainable innovation: production and consumption View project

Lean Industry 4.0 View project

All content following this page was uploaded by Tarcisio Abreu Saurin on 11 April 2016.

The user has requested enhancement of the downloaded file.

 Journal of Manufacturing Systems 31 (2012) 358–366

Contents lists available at SciVerse ScienceDirect

Journal of Manufacturing Systems

journal homepage: www.elsevier.com/locate/jmansys

Technical paper

A framework for assessing poka-yoke devices

Tarcisio Abreu Saurin ∗ , José Luis Duarte Ribeiro, Gabriel Vidor
DEPROT/UFRGS (Industrial Engineering and Transportation Department, Federal University of Rio Grande do Sul, Brazil), Av. Osvaldo Aranha 99, 5 andar, Porto Alegre, RS 90.035-190,
Brazil

a r t i c l e i n f o a b s t r a c t

Article history: This study introduces a framework for assessing poka-yoke devices (PD), encompassing both those
Received 17 October 2010 designed for quality control (referred to as quality PD) and those designed to control hazards to health and
Received in revised form safety at work (referred to as safety PD). The framework assesses the processes of the design, operation
22 December 2011
and maintenance of PD, rather than the outcomes of these processes. The development of the framework
Accepted 26 April 2012
involved three stages: (a) defining 15 attributes of PD, identifying those that provide fail-safe charac-
Available online 22 May 2012
teristics and those that inform best practices in design, implementation and maintenance; (b) defining
what the set of evidence and what the sources of evidence should be for assessing the existence of each
Keywords:
Poka-yoke
attribute; the sources include documents, interviews, observations and a meeting to discuss the results
Performance measurement of the assessment with company representatives; and (c) defining a scoring system to express the results
Lean production of the assessment. The application of the framework is illustrated by means of assessing four PD; two of
Quality control them being concerned with quality and two with safety.
Health and safety at work © 2012 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.

1. Introduction
The principles and practices of lean production (LP) have been
increasingly used by a number of industries [1,2]. This study
emphasizes the use of poka-yoke devices (PD), a LP practice
especially tailored to eliminate the production of defective parts,
complementing statistical process control techniques [3,4]. Since
this term started being disseminated in the West in the context of
spreading the Toyota Production System [5], PD has been a topic of
interest, mostly among practitioners, due to the apparent simplicity
of how they function and their intuitive design features [6]. PD have
been used in a variety of contexts, such as construction, health care
and information technology, but not necessarily associated with LP
implementations [7–9].
However, there seems to be a scant academic interest in this
topic, conveying the misleading message that the design and oper-
ation of PD can be explained by common sense [6]. In fact, similarly
with other important LP practices (e.g., kaizen and visual man-
agement), the literature on PD is, to a great extent, directed to
practitioners [10]. As a result, the attributes of PD are presented in a
fragmented way over a number of studies [5,7,9,11], without being
integrated into scientifically validated frameworks for designing
and assessing them. In this study, the focus is on the development
∗ Corresponding author. Tel.: +55 51 3308 4299; fax: +55 51 3308 4007.
E-mail addresses: tasaurin@yahoo.com.br, saurin@ufrgs.br,
tasaurin@terra.com.br (T.A. Saurin), ribeiro@producao.ufrgs.br (J.L.D. Ribeiro),
gvidor@producao.ufrgs.br (G. Vidor).
of a framework for assessing the use of PD. No reference was found
on methods, which had been tested and validated in real-world
settings, which could be used for assessing the use of PD.
From both a practical and theoretical perspective, the need to
develop a framework for assessing PD is part of the need to develop
methods for assessing the use of lean practices. In fact, compared
with the efforts made to address “how to become leaner”, the
issue of “how lean the system is” has received less attention [12].
Although a number of methods have been drawn up to assess the
extent to which a company adopts lean principles [12–14], fewer
methods [15,16] aim at assessing the use of lean practices, the
implementation of which takes place directly and visibly on the
shop-floor (e.g., PD). Survey has been the preferred strategy for
assessing the use of lean practices, aiming at investigating large
samples of companies [16–20]. While these surveys have merits,
they tend to be of little use as tools to companies that aim at assess-
ing how their particular lean practices are performing.
Thus, similarly with previous studies on the assessment of lean
practices [15], this study adopts the method of auditing as an alter-
native. This method is well-known in most industries, combining
both qualitative and quantitative assessments [21], which can
provide more pragmatic insights for individual companies, in com-
parison with those provided by surveys. An audit is a systematic,
independent and documented process to obtain evidence and to
assess them objectively, measuring the extent to which audit cri-
teria are met. Auditing allows the gap between desired and actual
performance to be identified, as well as to enable the assessment
of information that can be used to improve performance [22].

0278-6125/$ – see front matter © 2012 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jmsy.2012.04.001
 T.A. Saurin et al. / Journal of Manufacturing Systems 31 (2012) 358–366
The attributes encompassed by the framework proposed by this
article are applicable both to PD designed for quality control and
to those designed to control hazards to health and safety at work
(H&S). This integrative perspective between quality and H&S might
be considered as a distinctive feature of the proposed framework, as
both companies and the academic literature usually deal with each
type of PD separately. According to Stewart and Grout [6], this is
due to the fact that, typically, researchers and practitioners devoted
to quality control have little technical knowledge of H&S, and vice
versa. The application of the framework is exemplified by means
of applying it so as to analyze four PD: two of which are concerned
with quality and the other two with safety.
2. Definition of poka-yoke
The literature presents a number of similar definitions of poka-
yoke, even though they use different key terms and are often far
from precise. According to Shingo [5] a poka-yoke is a mecha-
nism for detecting errors and defects, which inspects 100% of the
pieces, working independently on the operator’s attention span.
To Grout [7] a poka-yoke is the use of process or design features
to prevent errors or the negative impact of errors. Middleton [23]
defines poka-yoke as the systematic practice of eradicating errors,
by locating their root cause. Plonka [24] considers that a poka-yoke
is a mechanism for detecting, eliminating, and correcting errors at
their source, before they reach the customer. Other studies sim-
ply define a poka-yoke by means of examples, either by simply
substituting this label with others, such as sensors and jigs, or by
translations such as mistake-proofing [25] or error-proofing [1].
Such synonyms draw on concepts that are intuitively meaningful
in the sense that everyone associates something with them, so they
feel they understand them [26].
In this study, based on the above mentioned definitions, a
poka-yoke is defined as a device that either prevents or detects
abnormalities, which might be detrimental either to product qual-
ity or to employees’ H&S. A poka-yoke might be considered as
three types of devices [27]: (a) physical, if they block the flow of
mass, energy or information, and do not depend on users inter-
preting them (e.g. a wall); (b) functional, if they might be turned
on or turned off due to an event (e.g. a lock or a password), with-
out depending on user interpretation; (c) symbolic, if they require
interpretation, yet are physically present at the moment they are
necessary (e.g. a safety sign). Immaterial devices [27] do not con-
stitute PD, since, regardless of their demanding user interpretation,
they are not physically present at the moment they are necessary
(for example regulations and organizational culture).
It is worth emphasizing two implications of the definition this
article adopts: (a) PD might demand interpretation from operators,
if they comprise symbolic devices; (b) PD can be either proactive,
if they prevent abnormalities, or reactive, if they detect abnormal-
ities. From a quality perspective, the difference between proactive
and reactive lies in the type of inspection performed by the PD
[5]: (a) source inspection, which avoids the occurrence of a defect
(proactive PD); (b) self-inspection, which detects a defect in the
very operation in which it is generated; (c) successive inspection,
which detects a defect in the operation following the operation in
which it was generated; (d) judgment inspection, which detects a
defect two or more operations ahead of the one it was generated.
Inspection types (b), (c) and (d) are reactive ones, since they detect
defects, rather than prevent them. While a PD might be integrated
within any operation (e.g., transportation, processing, inspection,
etc.), all PD perform at least one out of the four types of inspection
previously mentioned [5]. From a safety perspective, proactive PD
are those that avoid a human-machine interaction that may cause
359
an accident, while reactive PD are those that contain the energy
released by this interaction after it has happened.
Another well-known classification of PD, proposed by Shingo
[5], concerns the one that differentiates a control function from
a warning function. The control function requires there to be one
or more of the following features: (a) it turns off the machine, or
physically blocks the continuity of a manual process, once it detects
the abnormality; (b) it does not allow the operator to choose how
to carry out the task, but obligates him to perform it correctly;
(c) it automatically excludes defective parts from the production
flow. The warning function means that the poka-yoke signals, by
means of symbolic devices (e.g. lights and/or audible alarm), the
occurrence of the abnormality [5].
3. Research method
3.1. Overview of the research method
The development of the framework followed three stages: (a)
defining the attributes of PD; (b) defining evidence and sources of
evidence to assess the existence of each attribute; (c) developing
a scoring system to express the results of the assessment. After
being designed, the framework was tested by assessing four PD.
Two of them, deemed quality poka-yoke 1 (PQ1) and safety poka-
yoke 1 (PS1) were assessed in company A. Another quality poka-
yoke (PQ2) was assessed in company B. Both companies A and B
are suppliers of auto-parts for major car manufacturers, and they
had adopted LP as a formal corporative policy. In order to check the
applicability of the framework in a non-manufacturing and non-
lean environment, a second safety poka-yoke (PS2) was assessed
in a construction company (company C).
3.2. Definition of attributes of PD
The definition of the attributes was based on five criteria:
(a) the attributes should be, as far as possible, generalizable both
for safety and quality PD;
(b) the attributes should take into account the whole life-cycle of
the PD, ranging from its design process to its replacement;
(c) the attributes should be concerned with PD used in the manu-
facturing of products, rather than being concerned with PD used
in other phases of the life-cycle of a product, such as its design
and use by end customers. It is worth noting that this criterion
is not contradictory with the previous one. While criterion (b)
states that the whole life-cycle of the PD should be taken into
account by the attributes, criterion (c) states that the attributes
should consider a specific stage of the life-cycle of the products
inspected by PD (i.e., the stage of manufacturing);
(d) the attributes should be consistent with the concept of poka-
yoke presented in section 2 of this article (definition of poka-
yoke), while recognizing that they are not necessarily fail-safe,
even though this is the ideal situation;
(e) the attributes should be grounded on a sound theoretical basis,
mostly drawn from the fields of human factors and LP.
Based on these criteria, as well as on previous studies
[5–7,9,11,23,24], 15 attributes of PD were defined. They were
grouped in two categories (see Tables 1 and 2): (a) eight attributes
that assess if fail-safe characteristics exist; (b) seven attributes that
assess the existence of best practices for the design, implementa-
tion and maintenance of PD.
It should be noted that, even though all attributes listed in
Tables 1 and 2 are complied with, labeling a poka-yoke as 100% fail-
safe is an over-simplification. The attributes are mostly concerned
360 T.A. Saurin et al. / Journal of Manufacturing Systems 31 (2012) 358–366

Table 1
Attributes that provide PD with fail-safe features.

Fail-safe attributes Sources of evidence Principles

FS1 The poka-yoke has both control and a highly visible warning function (Score 4) Design of the poka-yoke (drawings HF1, HF6, LP2, LP4
The poka-yoke has only a control function (Score 3) and technical specifications);
The poka-yoke has only a warning function, which is highly visible to the operator (Score observation of its functioning
2)
The poka-yoke has only a warning function, with low visibility to the operator (Score 1)
There are neither control nor warning functions (Score 0)
FS2 The poka-yoke is used as a means of conducting source inspection (Score 4) Design of the poka-yoke; HF2, HF7, LP2
The poka-yoke is used as means of conducting self-inspection (Score 3) observation of its functioning
The poka-yoke is used as a means of conducting successive inspection (Score 2)
The poka-yoke is used as a means of conducting inspection by judgment (Score 1)
No inspection is carried out (score 0)
FS3 All product models are inspected by the poka-yoke and, for each of them, 100% of the Design of the poka-yoke; HF3
pieces are inspected (i.e., inspection is not based on sampling) (Score 4) observation of its functioning
Less than 100% of the product models are inspected by the poka-yoke, even though, for
each of them, 100% of the pieces are inspected (Score 2)
The poka-yoke does not ensure that 100% of the pieces are inspected, whatever the
product model is (Score 0)
FS4 The functioning of the poka-yoke does not require any action from the operator Design of the poka-yoke; HF3
specifically for this purpose (e.g. the operator does not need to press buttons or use observation of its functioning
equipment, such as barcode readers, specifically to start the poka-yoke) (Score 4)
The functioning of the poka-yoke requires one action from the operator specifically for
this purpose (Score 2)
The functioning of the poka-yoke requires two or more actions from the operator
specifically for this purpose (Score 0)
FS5 The poka-yoke does not create any new opportunities for unintentional errors by operators Design of the poka-yoke; HF4
* Errors that impact on H&S should be considered only in attribute FS6 (Score 4) observation of its functioning;
The poka-yoke creates opportunities for new errors, but they have minor impacts for interviews with workers; standard
product quality and production performance (Score 2) work of the operator
The poka-yoke creates opportunities for new errors, which might have major impacts for
production quality and production performance (Score 0)
FS6 The poka-yoke does not put the health and safety of workers at risk (Score 4) Design of the poka-yoke; HF5
The poka-yoke is in conflict with H&S regulations, even though it does not create major observation of its functioning;
risks (Score 2) hazard analysis of operation;
The poka-yoke is in conflict with H&S regulations, creating major risks to the operators interviews; standard work; safety
(Score 0) regulations
FS7 The maintenance plan for the poka-yoke provides that, if necessary, it is calibrated or Maintenance plan of poka-yoke, or LP5
replaced regularly – typically, calibration or replacement is required for PD that suffer of the machine on which it is
wear and tear which affects their functionality (Score 4) installed; interview with those in
Even if both calibration and replacement are necessary, the maintenance plan provides charge of maintenance
that only one out of the two activities are performed (Score 2)
Even if both calibration and replacement are necessary, the maintenance plan neglects
both activities (Score 0)
FS8 The maintenance plan for the poka-yoke provides, at the beginning of each shift, a test to Maintenance plan of poka-yoke, or LP5
verify its effectiveness – similarly to attribute FS7, this attribute is normally necessary only of the machine on which it is
for PD that are subject to a wear and tear that affects its functionality (Score 4) installed; interview with those in
The maintenance plan for the poka-yoke provides tests to verify its effectiveness, but it charge of maintenance
does not specify how often the tests should be carried out (Score 2)
The maintenance plan for the poka-yoke does not provide systematic tests to verify its
effectiveness (Score 0)

Note: the initials HF and LP refer to the principles of human factors and lean production associated with the attributes – these principles are detailed in Table 3.

with the processes of the design, operation and maintenance of
PD, rather than with the outcomes of these processes. Moreover, as
happens with any human endeavor, these processes have vulner-
abilities that cannot be fully anticipated or controlled, so there is
always a residual risk due to unanticipated variability [27].
Table 3 shows the eight principles of human factors (HF) and the
six principles of lean production (LP) associated with the attributes
listed in Tables 1 and 2. It is worth noting that two principles classi-
fied as human factors, could also be interpreted as principles of lean
production: (a) HF2, insofar as controlling problems at the source
is equivalent to the lean ideal of inspection at source; (b) HF6, to
the extent that the principle of making the problems visible is also
widely known as part of the lean philosophy of visual management,
as indicated by principle LP4.
The association with HF and LP indicates that the design of a
poka-yoke should follow a socio-technical approach. This means
that the design of a poka-yoke should harmonize the human
requirements with the technical requirements, instead of optimiz-
ing the technical dimension and adapting the human dimension to
this [28].
3.3. Evidence and sources of evidence for assessing the use of PD
3.3.1. Characterization of the company and the operation
assessed
Before assessing the PD, it is important to know the organiza-
tional context in which they are used, since this may well contribute
to explaining the results of the assessment. In particular, it is pro-
posed that the following features of context be investigated, by
means of interviews with company representatives and analysis of
documents: (a) history of the use of LP and PD in the company; (b)
the main features of the management systems linked to PD (quality
or H&S) in order to identify dependency relationships with other
elements of management. For example, it can be verified how the
PD make a contribution to ensuring that the company meets the
legal requirements, in the case of H&S.
Similarly, there is a need to be thoroughly familiar with the
operation in which the poka-yoke is used. It is recommended the
following be identified: (a) the position of the operation that uses
the poka-yoke in the value stream, which helps to understand its
systemic impacts; (b) the profile of the operators who work in the
 T.A. Saurin et al. / Journal of Manufacturing Systems 31 (2012) 358–366 361

Table 2
Best practices attributes for the design, implementation and maintenance of PD.

Best practices Sources of evidence Principles

BP1 The company has formal and objective criteria to choose the operations in which a poka-yoke should be Interview with representative of LP6
used (Score 4) the design team; database with
The company has general and informal guidelines that support the choice of operations in which a records of abnormalities; FMEA
poka-yoke should be used (Score 2) spreadsheets
The company defines where a poka-yoke should be used based on gut feelings from staff (Score 0)
BP2 During the design phase, a formal quantitative assessment was made of the cost-benefit of the poka-yoke Calculation of the cost-benefit LP1
and, during full-scale operation, its performance matched the expected benefits (Score 4) analysis; performance indicators
Just a qualitative cost-benefit assessment was carried out, after the implementation of the poka-yoke before and after the poka-yoke,
(Score 2) interviews with operators and
No cost-benefit assessment was carried out (Score 0) managers
BP3 The poka-yoke was designed by a multi-functional team (e.g., representatives from the areas of quality, Interview with representative of HF8, LP6
H&S and front-line operators), which held formal design meetings (Score 4) the design team; interviews with
Although the poka-yoke was designed by a team which held formal design meetings, the team was formed operators; minutes of meetings of
by people from the department concerned with its focus (i.e. quality or safety department) (Score 2) the design team
The poka-yoke was designed by a single individual, from the department concerned with its focus (i.e.
quality or safety department) (Score 0)
BP4 There is a document that details how the poka-yoke works, available at the workstation where it is used Observation of the workstation LP3, LP4
(Score 4) – The document requested may be interpreted as a “user manual”, since it is not intended for the that uses the poka-yoke; interview
person who makes the poka-yoke, but for its end user with operators; document which
There is a document that details how the poka-yoke works, but it is not available at the workstation where details the operation of the
it is used (Score 2) poka-yoke
There is no document detailing how the poka-yoke works (Score 0)
BP5 There is an area, visible to the supervisors and operators of the workstation that uses the poka-yoke, Observation of the workstation; HF6, LP4
dedicated to storing defective parts identified by the poka-yoke (Score 4) interviews with operators
There is an area dedicated to storing defective parts identified by the poka-yoke, but it is not easily visible
to the supervisors and operators (Score 2)
The defective parts identified by the poka-yoke are stored in an improvised area, which is not visible to
the supervisors and operators (Score 0)
BP6 There are visual aids (e.g. posters), standardized (e.g., colors, characters and drawings are standardized), Observation of the workstation; LP4
identifying the existence of a poka-yoke in the operation (Score 4) observation and comparison of
There are visual aids identifying the existence of a poka-yoke in the operation, but they are not visual devices that identify the
standardized (Score 4) poka-yoke
There are no visual aids identifying the existence of a poka-yoke in the operation (Score 0)
BP7 There are historical records of corrective and preventive maintenance of the poka-yoke (Score 4) Records of corrective and LP5
There are historical records of either preventive or corrective maintenance of the poka-yoke, but not of preventive maintenance; interview
both (Score 2) with those in charge of
There are no historical records of maintenance of the poka-yoke (Score 0) maintenance

Note: attribute BP5 is not applicable to safety PD.

Table 3
Principles of human factors (HF) and lean production (LP) that underpin the poka-
yoke devices’ attributes.
HF1 Workers’ degrees of freedom should be reduced by design,
thus leaving no room for unnecessary adaptations of
procedures [29]
HF2 Problems should be controlled at source [28]
HF3 During routine tasks, which are conducted at a low level of
awareness, humans have their attention focused on
anywhere else but the task [30]
HF4 Changing any element of a system provokes impacts on
other system elements. In particular, automation frequently
implies changing the modes of human error, rather than
eliminating them [30]
HF5 H&S requirements should be an integral part of product and
process design, in any industry [28]
HF6 Systems should be simple in design and make problems
visible [28]
HF7 Reaction to failures is an acceptable counter-measure when
the hazards could not have been eliminated at source [27]
HF8 It is employees who know best what problems there are
with their jobs and which alternatives will most satisfy
them [31]
LP1 Base your management decisions on a long-term
philosophy, even at the expense of short-term financial
goals [1]
LP2 Build a culture of stopping to fix problem so as to get quality
right at the first time [1]
LP3 Standardized tasks are the foundation for continuous
improvement and employee empowerment [1]
LP4 Use visual controls so no problems are hidden [1]
LP5 Use only reliable, thoroughly tested technology that serves
your people and processes [1]
LP6 Make decisions slowly by consensus, thoroughly
considering all options; implement decisions rapidly [1]
operation which has the poka-yoke (e.g., length of experience); (c)
the number of products that share the operation that has the poka-
yoke; and (d) the number of replications of the poka-yoke in the
plant, as this number has an impact on its management structure.
3.3.2. Assessment of the existence of the attributes
For each of the 15 attributes, the sources of evidence were estab-
lished (see Tables 1 and 2), thus seeking, to the extent possible, to
establish at least two sources per attribute. As to the sources that
involve interviews, the questions to be posed are related to the
actual wording of the attribute. For example, to assess attribute
BP1 (see Table 2), a representative of the design team should be
questioned about the reasons that led to setting up the poka-yoke
in that specific operation. From Tables 1 and 2, it is also possible to
draw up lists that compile all the sources of evidence of one and the
same type. For example, a specific list could be developed stating
all observations that should be carried out and another list could
state all documents that should be consulted.
3.4. Assignment of scores to the PD
The assessment of each attribute should be conducted by assign-
ing a score over a scale, in which the score of zero means the
attribute is not met, while the score of four means it is fully met.
It is also possible that the attribute is not applicable to the poka-
yoke, in which case no score is given. It is worth stressing that the
auditors can assign any score (not restricted to integers) between
zero and four. This accounts for the variety of means to deal with
362 T.A. Saurin et al. / Journal of Manufacturing Systems 31 (2012) 358–366
the attributes, allowing degrees of freedom to assign the scores in
situations that were not fully anticipated by the framework.
In order to provide guidance to the auditors, at least three
performance levels were defined for each attribute, establishing
references for scoring four, two or zero. For attributes FS1 and FS2,
which were based on a well-known classification of PD proposed
by Shingo [5], it was possible to establish five levels, since there
was a clear conceptual hierarchy between them.
The result of the assessment of each poka-yoke can be expressed
by two scores: the average of the scores of the attributes which
confer fail-safe characteristics; the average of the scores of the
attributes associated with best practices. The two scores are inde-
pendent and it is not recommended that an average between
the two be calculated, since the number and importance of the
attributes of each group are different. It is recommended that the
assessment of attributes be performed by two auditors, who should
reach a consensus on the final score and prepare a report on the
results of the assessment, in which they should include justifica-
tions for each score assigned. In particular, justifications should be
presented for the scores that are not equal to any of the performance
levels pre-defined by the framework.
3.5. Application of the framework to the industrial cases
3.5.1. Preparatory stage
This step included three activities: (a) introducing the assess-
ment procedures to the representatives of the companies; (b)
choosing the PD to be assessed; and (c) negotiating the schedule
for the assessment. For each poka-yoke chosen, it was specified
what abnormality it was that the poka-yoke sought to control. In
the description of the abnormality, the unwanted action, condition
or decision was noted, as were the respective consequences.
In companies A and B, in which three PD (PQ1, PQ2 and PS1)
were assessed, there was a documented definition of poka-yoke,
similar to the one adopted in this article, but restricted to quality
control. Based on this definition, companies A and B had a record,
respectively, of 70 and 43 quality PD.
In these companies, what initially motivated the use of PD arose
from the need to meet regulatory requirements. The use of quality
PD is a requirement of ISO/TS 16949:2009 [32], the certification
of which is mandatory by imposition of the automobile assem-
bly companies. According to item 8.5.2.2 of this standard, “the
organization should use error-proof methods in their processes
of corrective action.” However, since the standard does not pro-
vide any details of the requirements, beyond the statement quoted,
some clients set requirements that the PD should meet, all of which
are included in the list of attributes put forward in Tables 1 and 2.
The safety PD was also used in company A in order to meet legal
requirements. In this case, the legislation referred to H&S regula-
tions, especially with regard to the safety of presses. The company
had been emphasizing the need to use safety PD in order to react
to and prevent interdictions being imposed by auditors from the
Ministry of Work and Employment.
In company C, a safety poka-yoke (PS2) was assessed, which had
also been set up as required by H&S regulations. In this company,
the employees were not familiar with either the term poka-yoke or
LP, so that there had been no systematic initiative to set them up.
3.5.2. Data collection
In the four cases studied, the characterization of the company
and the operation that owned the poka-yoke was based on both
observations and an interview (with the quality manager in com-
panies A and B, with the civil engineer responsible for the site in
company C). As for the interviews required as sources of evidence
for assessing the attributes, these were conducted in the three com-
panies with the operators of the work stations which had the PD.
In companies A and B, it was also possible to interview a member
of the design team.
In the three companies, the assessment of attributes was carried
out jointly by two members of the research team, one of whom had
a formal degree in industrial engineering and training in LP systems,
while the other, in addition to these, also had formal training in H&S
and ergonomics – this was the leading auditor. Thus, as occurred
in this study, it is recommended that the audit team have formal
educational qualifications in the areas mentioned, whatever the
type of poka-yoke assessed, be it quality or safety.
As the final stage of applying the framework, in each company, a
feedback meeting involved a member of the research team, a man-
ager and the operators of the work stations that had the PD. The
discussions in each meeting focused on: the results of the assess-
ment; the reasons that led to some attributes being met and others
not; and on opportunities for performance improvement.
It is worth noting that, while external auditors applied the
framework in the case studies, it also could and should be applied
by internal auditors. In particular, this would be useful to prepare
the company to the mandatory external audits of PD conducted
by clients, such as the auto-assemblers. Moreover, both for inter-
nal and external audits, the operational recommendations to apply
the framework ideally should follow the ISO 19011 standard [21],
which provides detailed guidelines on how to conduct an audit (e.g.,
defining who the leading examiner will be and holding a meeting
to reach consensus on the final scores).
4. Characteristics of the companies and PD investigated
4.1. Company and the operations in which PQ1 and PS1 were
used
Company A manufactures components and transmission sys-
tems for light and heavy vehicles. At the time of data collection, the
plant visited had about 600 employees and there were three man-
agers, directly linked to the plant director, who coordinated a LP
program at the plant. Efforts to set up LP practices began in 2004.
PQ1 (Fig. 1) was located in a preparatory operation for polishing
a component of automobile axles. In this operation, a compo-
nent called end yoke, produced in a previous operation, should be
screwed into a clamp, which was part of the tool mounting com-
ponent within the polishing machine. In the operation of screwing
it down, there is a need for the end yoke to be slotted into a tem-
plate, using holes in its base. All end yokes models can be slotted
onto the same template, without setups. If the holes have not been
made correctly in the previous operation, the end yoke will not fit
onto the template and the screwdriver would not switch on. On the
basis of the template, there was a sensor used to detect the pres-
ence of the holes. At the end of the polishing operation, the end
yoke should be unscrewed from the clamp.
The PS1 consisted of a presence sensor installed on a press. In
this machine, a greasy liquid should be injected into the four ends
of the part illustrated in Fig. 2. If the operator placed any part of
his body in a hazardous area, the press would turn off automati-
cally. In addition, a lamp placed on the workbench would indicate,
by switching off the red light and switching on a green light, the
moment that the greasing operation had been completed, freeing
the operator to remove the piece.
4.2. Company and the operation in which PQ2 was used
The assessment of PQ2 was conducted in a large manufacturer
of brake pads for automobile assembly plants. As the result of
a corporate policy, this company has adopted LP principles and
 T.A. Saurin et al. / Journal of Manufacturing Systems 31 (2012) 358–366 363

Fig. 1. From left to right: end yoke and clamp; operator positioning the end yoke on the template; operator screwing the clamp onto the end yoke.

practices for about ten years, with the emphasis at present being
on maintaining and deepening the lean system.
The PQ2 was set up about three years ago and is located in the
operation that checks the dimensions of the brake pads, the last
step of the manufacturing process before packing and shipment to
the client. This inspection consists of verifying that the angles and
positions of the components of the brake pads are as per the spec-
ifications. There are ten models of brake pads, with the possibility
of there being up to three changes of models per shift.
The poka-yoke consists of a machine with three principal com-
ponents (Fig. 3): (a) a conveyor belt, which carries the parts one
by one, on the machine; (b) a video camera, inside the machine,
which captures images of the brake pads in order to compare the
dimensions of the part with the dimensions of the standard one; (c)
a monitor, which shows comparative images between the standard
and the brake pad. When an abnormality is detected, the monitor
displays a visual alert for the operator to remove the defective part.
This operator is dedicated to this operation, and stands in front of
the monitor, only leaving this place in order to remove the defective
parts from the production line.
The initiative to set up the poka-yoke was due to the large
number of complaints from clients about defects in the dimen-
sions of the brake pads. As a result, the inspection machine was
designed in-house, and involved members of the quality control
circles.
4.3. Company and the operation in which PS2 was used
The assessment of PS2 took place on a building site, under the
responsibility of a small construction company that undertakes
residential works. This company does not apply LP principles and
practices. H&S management lays the emphasis on meeting the legal
requirements and is coordinated by professionals from an out-
sourced company.
PS2 is about access gates to the freight elevator, which are on
all floors (Fig. 4). Such gates are mandatory according to H&S reg-
ulations, and have to be designed so that they can only be opened
when the elevator platform is stopped at the floor level. Moreover,
the elevator will not move if a gate is open on any floor. The ele-
vator operator has a control panel, by which he can both program
the elevator to stop at any given floor and locate its position. The
choice of this poka-yoke for assessment was due to the fact that it
is a legal requirement and is widely used by builders throughout
the country. The abnormality to be controlled, in this case, refers
to the possibility of an employee placing part of his body inside
the elevator shaft (for example, to check if the elevator is coming,

Fig. 2. From left to right: presence sensor; part in the shape of a cross into which the grease was injected; lamp which signaled the end of the cycle.
364 T.A. Saurin et al. / Journal of Manufacturing Systems 31 (2012) 358–366

Fig. 3. From left to right: brake pads entering the inspection operation; external view of the inspection machine and monitor in the top right-hand corner; the camera which
captures images to compare with the standard.

Fig. 4. On the left: access gate to the freight elevator. On the right: platform of the elevator.

or to communicate with the winch operator) and being hit by the poka-yoke was specified in great detail by H&S regulations,
moving platform. leaving no room for changes.

Even though the other scores have been equal to the pre-
5. Results of the assessment
defined levels, it is worth stressing best practices and improvement
opportunities detected during the assessment, which illustrate the
Table 4 presents the results of the assessment of the four PD.
usefulness of the framework and can be insightful for other compa-
It can be noted that, for all of them, the mean for the attributes
nies. Concerning best practices, two examples can be emphasized:
that are linked to fail-safe features is higher than the mean for the
attributes that are associated with best practices.
From the sixty possible scores to be assigned, six were not equal (a) PQ1 was the only PD to be in full compliance with the attribute
to the predefined performance levels established in Tables 1 and 2, FS8. At the beginning of each shift, the operator should test the
and three were not applicable. Thus, especially in these cases, a functioning of the poka-yoke using two standard pieces (one of
justification is necessary, as presented below: them being a good component and the other being a defective
one), referred to as master parts by the company. The results of
these tests were documented on a form, signed by the worker;
(a) the score of 3.5 assigned to PQ1 (attribute FS1), is explained by
(b) PQ1 and PQ2 were the only PD to be in full compliance with
the low visibility of the warning function. Even though there
BP6. The plants had an effective strategy for visual manage-
was a light that would come on to warn that the sensor had
ment, which included the identification of all workstations that
detected a problem, it was too small and not very visible to the
had quality PD (Fig. 5). According to the company representa-
operator;
tives who were interviewed, the identification boards made it
(b) PS2 scored 3.5 (attribute FS6), since, regardless of not being in
conflict with any specific H&S regulation, a minor risk of getting
the hands stuck in the entrance door of the lift was created by
the poka-yoke;
(c) PS1 and PS2 scored 1.0 (attribute FS8) because, even though
their maintenance plans did not provide systematic tests to ver-
ify their effectiveness, their functioning could be tested at any
time by workers, even involuntarily, without requiring specific
technical resources to conduct the test, such as master parts;
(d) PQ1 and PQ2 scored 1.0 (attribute BP3) because their design
process did not involve a team that held formal design meet-
ings, which was the minimum requirement to score 2.0.
However, the design was not fully conducted by a single person,
since the workers interviewed reported that they were infor-
mally consulted during the design phase. For PS2, attribute BP3
was considered to be not applicable, since the design of the Fig. 5. Identification board for PQ1.
 T.A. Saurin et al. / Journal of Manufacturing Systems 31 (2012) 358–366
Table 4
Results of the assessment of the PD.
FS1 FS2 FS3 FS4 FS5 FS6 FS7 FS8
PQ1 3.5 1 4 4 4 4 0 4 3.06
PQ2 1 2 4 4 4 2 0 0
PS1 4 4 4 4 4 4 0 1
PS2 3 4 4 4 4 3.5 0 1
Notes: NA, not applicable; M, mean.
easier to track the existing PD (e.g. for auditing) and prevented
them from being inadvertently turned off.
Concerning the improvement opportunities, four of them were
selected to illustrate the usefulness of the framework, as presented
below:
(a) The plants that adopted the best practices above cited, con-
cerning attributes FS8 and BP6, should extend those practices
to their safety PD;
(b) Concerning the attribute FS6, a non-compliance with H&S reg-
ulations was detected in PQ2, since the workstation design
did not allow work to be performed in the seated position.
Moreover, there was no seat near the machine, so the worker
could rest during pauses. The height of the computer screen
(1.75 m above the floor) was also noteworthy, which makes it
difficult for shorter workers to see. It is worth noting that the
workstations in which PQ1 and PS1 were located, were not in
compliance with regulations either (i.e. there were no seats).
However, this fault was not associated with the PD, but rather
with the workstations. Indeed, PQ2 was the workstation, since
the station was fully dedicated to quality control;
(c) all of the four PD were in non-compliance with the attribute FS7
(score 0.00), since there were no specifications either on the cal-
ibration or substitution procedures. This would be necessary,
for example, in PQ2. Although the adjustment of the camera
at the right position for each product model was made auto-
matically, there was no periodical calibration of the mechanical
systems that would ensure the right adjustment.
An opportunity for improvement concerning calibration was
also detected during the assessment of PQ1. It was realized
that the screwdriver was not calibrated periodically, which
leaves room for connections of components without the spec-
ified torque. However, insufficient torque was an opportunity
for error that was completely independent of PQ1;
(d) The low scores (2) assigned to PQ1 and PQ2, concerning
attribute BP1, were due to the largely reactive criteria for taking
the decision to use PD. In both cases, the company representa-
tives emphasized that the PD were designed at the request of
clients, in order to reduce a high rate of defects and complaints.
By contrast, PS1 and PS2 (score 4) were in place since the respec-
tive workstations had been set up, in order to be in compliance
with H&S regulations.
6. Discussion and conclusions
This study introduced a framework for assessing the processes
of design, operation and maintenance of PD, which has three dis-
tinctive characteristics: (a) definition of attributes applicable both
to quality and safety PD; (b) definition of attributes that pro-
vide fail-safe features and attributes that are simply best practices
of design, operation and maintenance; and (c) a scoring system,
which supports the comparison among PD. Also, criteria for defin-
ing the attributes were stated in order to support adaptations of
the framework to the evolution of PD practice and theory. These
365
M BP1 BP2 BP3 BP4 BP5 BP6 BP7 M
2 2 1 0 2 4 2 1.86
2.12 2 2 1 0 2 4 2 1.86
3.12 4 2 2 0 NA 0 2 1.67
2.94 4 2 NA 2 NA 0 4 2.40
criteria add conceptual precision and generalizability to the frame-
work, since possible adaptations will be guided by properly defined
principles.
The framework might be used at least during two stages of the
life-cycle of a poka-yoke: (a) during the design of a new poka-yoke,
as a check-list and source of insights for designers; (b) as a tool to
be applied on a regular basis to all PD of a company, thus provid-
ing benchmark measures. In this respect, the frequency of internal
audits could be based on the frequency of external audits, which
are usually mandatory by clients and have well defined timeframes
(e.g., annual).
Moreover, as it was pointed out by the assessment of PS2 in the
construction company, the framework is also useful for companies
that do not use lean production. Even though it is a single case
study, it has also shown that the results of the assessment can be
better in a non-lean environment. For example, PS2 performed bet-
ter than PQ2, which was used in a major transnational that formally
adopted lean production. As an advantage, PS2 was implemented
as a requirement of regulations, having been used for many years in
a fairly uniform way by construction companies all over the coun-
try. By contrast, PQ2 was designed as a prototype. Of course, an
application of the framework in a larger number of PD and compa-
nies is necessary to identify trends concerning lean versus non-lean
environments. A hypothesis to be tested, based on the prelimi-
nary piece of evidence provided by this study, is that, for a safety
poka-yoke, the implementation within a lean environment is less
important, since they are often demanded by regulations, rather
than due to the specific production philosophy adopted by the
company.
The comparison between quality and safety PD also indicated
that the companies were not aware of their similarities and, as
a result, they were not taking advantage of the synergistic rela-
tionships. In company A, it was clear that quality and safety PD
were managed by completely different mechanisms and staff. In
this company, no reasonable justification was identified for not
extending some practices used by quality PD to the safety ones.
As another example of the impact of this fragmented management
structure, PQ2 displayed H&S hazards, a problem originated from
the lack of a design team involving both H&S and quality control
staff.
The case studies have also shown that the auditors should
be acquainted with the concepts underlying the attributes of PD,
which require background in both lean and human factors. This
required knowledge base is a result of the socio-technical nature
of the poka-yoke design process, which emphasizes the need to
go beyond the traditionally fragmented way that, companies, and
often operations management research, deal with human and tech-
nical issues in the work environment.
In addition to the expertise required, another possible drawback
to applying the framework is concerned with the unavailability, in
the companies investigated, of the recommended sources of evi-
dence. For example, in the assessment of PQ1 and PQ2, there was
no preliminary hazard analysis of the operations, making it difficult
to assess the extent of the H&S hazards. Nevertheless, the unavail-
ability of the sources of evidence should be interpreted as a result
366 T.A. Saurin et al. / Journal of Manufacturing Systems 31 (2012) 358–366
of practical implications for the company investigated, since this
indicates the opportunity for providing sources.
Another important insight arising from the case studies con-
cerns the risk of being too reductionist when applying the
framework. The PD is frequently a small component of a broader
workstation. Therefore, sometimes it is not straightforward to
define the boundaries of the poka-yoke (i.e. where the poka-yoke
ends and where the rest of the workstation begins). This diffi-
culty was encountered, for example, during the assessment of PQ1.
The lack of calibration of the screwdriver, while being a problem,
was not considered to be a failure of the poka-yoke, since the
screwdriver, by itself, had not any inspection role concerning the
abnormality investigated. Nevertheless, a holistic view during the
application process made it possible to identify this shortcoming,
even though it did not alter the score of the poka-yoke. By contrast,
in the assessment of PQ2, the whole workstation was considered
to be the poka-yoke. On the one hand, this eliminated doubts on
the border between the poka-yoke and the workstation. On the
other hand, this was intrinsically bad, since, if the workstation was
the poka-yoke, this meant that it was fully dedicated to reactive
inspection.
As regards the scoring system, it proved to have discriminatory
power, according to the company representatives who received
feedback of the results. However, two important recommendations
should be stressed to reduce the subjectivity of the assignment of
scores: (a) at least two auditors should reach a consensus on the
final score, as happened in the case studies; (b) a report should
justify the scores assigned, especially those that were not equal to
the pre-defined performance levels proposed by the framework. In
other words, this means that the auditors should leave a trail, to
show how they reached their scores and conclusions from the data
available.
As a consequence of the limitations of this study, a number
of opportunities to continue this research in future projects were
detected, such as: (a) adapting the framework to other sectors,
such as services (e.g. hospitals) and transportation (e.g. aviation
and railways), since changes in the attributes might be necessary;
(b) adapting the framework so it can be used in other stages of the
life-cycle of a product, rather than in manufacturing (e.g. design
and maintenance); (c) investigating how the framework could be
used in the design stage of new PD; (d) assessing if there are
statistical correlations between the level of compliance with the
attributes and other factors, such as the technological complex-
ity of poka-yoke; the sector in which they have been used; and
the extent to which LP has been used in the organization; and (e)
using the framework to identify, among the existing publications
that catalog examples of PD, those which are in compliance with
the attributes that provide fail-safe characteristics and, as a result,
should be of primary interest with regard to disseminating them.
Two additional needs to be addressed by future studies are partic-
ularly important since they have to do with methodological issues
associated with the framework: (a) assessing its reliability (i.e.,
checking if similar results will be found by different teams of audi-
tors); and (b) assessing its ability to predict the effectiveness of PD
(i.e. the predictive validity of the framework), in terms of measures
such as scrap rates, productivity, client satisfaction and incident
rates.
View publication stats
References
[1] Liker J. The Toyota Way: 14 management principles from the world’s greatest
manufacturer. McGraw-Hill; 2004.
[2] Smalley A. Creating level pull. Lean Enterprise Institute; 2004.
[3] Montgomery DC. Introduction to statistical quality control. Wiley; 2009.
[4] Nada O, ElMaraghy H, ElMaraghy W. Quality prediction in manufacturing sys-
tem design. Journal of Manufacturing Systems 2006;25:153–71.
[5] Shingo S. Zero quality control: source inspection and the poka-yoke system.
Productivity Press; 1988.
[6] Stewart M, Grout R. The human side of mistake-proofing. Production and Oper-
ations Management 2001;10(4):440–59.
[7] Grout J. Mistake-proofing the design of health care processes. Agency for Health
Care Research and Quality; 2007.
[8] Formoso CT, Santos A, Powell J. An exploratory study on the applicability of
process transparency in construction sites. Journal of Construction Research
2002;3(1):35–54.
[9] Shimbun NK. Poka-yoke: improving product quality by preventing defects.
Productivity Press; 1988.
[10] Farris A, Van Aken E, Doolen T, Worley J. Critical success factors for human
resource outcomes in kaizen events: an empirical study. International Journal
of Production Economics 2009;117(1):42–65.
[11] Hinckley CM. Combining mistake-proofing and Jidoka to achieve world
class quality in clinical chemistry. Accreditation and Quality Assurance
2007;12:223–30.
[12] Wan H, Chen F. A leanness measure of manufacturing systems for quantify-
ing impacts of lean initiatives. International Journal of Production Research
2008;46(23):6567–84.
[13] Gurumurthy A, Kodali R. Application of benchmarking for assessing the
lean manufacturing implementation. Benchmarking: An International Journal
2009;16(2):274–308.
[14] Shingo Prize. Recognizing business excellence in the United States, Canada and
Mexico: application guidelines; 1998, http://www.shingoprize.org/Accessed in
22/08/2005.
[15] Saurin TA, Marodin G, Ribeiro JLD. A framework for assessing the use of lean pro-
duction practices in manufacturing cells. International Journal of Production
Research 2011;49(11):3211–30.
[16] Doolen T, Hacker M. A review of lean assessment in organizations: an
exploratory study of lean practices by electronic manufacturers. Journal of
Manufacturing Systems 2005;24(1):55–67.
[17] Shah R, Ward P. Lean manufacturing: context, practice bundles, and perfor-
mance. Journal of Operations Management 2003;21(2):129–49.
[18] White R, Pearson J, Wilson J. JIT manufacturing: a survey of implementation in
small and large US manufacturers. Management Science 1999;45:1–15.
[19] Vinodh S, Chinta SK. Leanness assessment using multi-grade fuzzy approach.
International Journal of Production Research 2011;49(2):431–45.
[20] Matson JE, Matson JO. Just-in-time implementation issues among automotive
suppliers in the southern USA. Supply Chain Management: An International
Journal 2007;12(6):432–43.
[21] ISO 19011. Guidelines for quality and/or environmental management systems
auditing. International Organization for Standardization; 2002.
[22] Chiesa V, Coughlan P, Voss C. Development of a technical innovation audit.
Journal of Production and Innovation Management 1996;13:105–36.
[23] Middleton P. Lean software development: two case studies. Software Quality
Journal 2001;9:241–52.
[24] Plonka FE. Developing a lean and agile work force. Journal of Human Factors
and Ergonomics in Manufacturing 1997;7(1):11–27.
[25] Chen J, Black J. A fuzzy logic based approach for pokayoke stoplight con-
trol in unmanned manufacturing cells. Journal of Manufacturing Systems
1996;15:33–42.
[26] Dekker S, Hollangel E. Human factors and folk models. Cognition, Technology
and Work 2004;6:79–86.
[27] Hollnagel E. Barrier analysis and accident prevention. Ashgate; 2004.
[28] Clegg C. Sociotechnical principles for system design. Applied Ergonomics
2000;31:463–77.
[29] Rasmussen J, Petersen A, Goodstein L. Cognitive systems engineering. Wiley;
1994.
[30] Reason J. Managing the risks of organizational accidents. Ashgate; 1997.
[31] Hendrick H. Applying ergonomics to systems: some documented lessons
learned. Applied Ergonomics 2008;39:418–26.
[32] ISO/TS 16949: 2009. Quality management systems – particular requirements
for the application of ISO 9001:2008 for automotive production and rele-
vant service part organizations. International Organization for Standardization;
2009.