You are on page 1of 18

IJOPM

15,7 Continuous quality


improvement through
Taguchi’s online quality
60
Received June 1994
control methods
Revised September 1994 P. Aravindan and S.R. Devadasan
Mechanical Engineering Department, PSG College of Technology,
Coimbatore, India
B.V. Dharmendra
Mechanical Engineering Department, UBDT College of Engineering,
Davangere, India, and
V. Selladurai
Mechanical Engineering Department, Coimbatore Institute of
Technology, Coimbatore, India

A global perspective view of total quality management


Ever since the industrial world began to experience tough global competition,
the generic term “total quality management” (TQM) has become the centre of
discussions. Though globally TQM is approved as a ladder for achieving
excellence in quality to reach a superior level in the highly competitive market,
there still exist differing views on the approaches involved in achieving it. Even
as the industrial world witnesses strident campaigning by quality engineering
experts and academicians over implementing successful TQM, many sceptical
observations are made by the manufacturers who attempted to implement TQM
in their firms at the expense of money and time. A brief but careful study made
by the authors reveals that even though considerable efforts are made in modern
manufacturing firms to facilitate employees’ participation in quality
improvement projects through quality circle programmes, only subtle attempts
are made to review the existing statistical quality and process control practices
which basically embrace the principle of compromising quality by justifying the
deviations from the target parameters. This is a gross violation of the principles,
objectives and philosophy of TQM. As a matter of fact, as long as these
conventional statistical practices dominate manufacturing firms, there is little
chance of effecting TQM, the philosophy of which lies on two main pillars,
namely, continuous improvement of quality, and achieving quality through the
backing of all in the concern. Already plenty of successful case studies from the

International Journal of Operations The authors are grateful to the anonymous referees for their extremely useful and helpful
& Production Management, Vol. 15
No. 7, 1995, pp. 60-77. © MCB comments on the earlier version of this manuscript which helped considerbly to improve the
University Press, 0144-3577 presentation of the article.
industrial world have been reported with regard to making people support Quality
quality through quality circle programmes[1]. However, success stories leading improvement
to the achievement of continuous quality improvement, which is not only a through TOLQC
means but also the ultimate goal of TQM, is reported in literature to a very slight
extent for it is misinterpreted as an ideal goal not normally possible to achieve.

Essence of research 61
Since statistical techniques do not aid continuous quality improvement, the need
has arisen to find new techniques and tools to replace the present-day quality
systems infected with statistical techniques. In this context, major portions of the
methods proclaimed by Japanese-born quality engineering expert Dr Genuchi
Taguchi, called “on-line quality control methods”, are found to offer meaningful
solutions in the process of ensuring continuous quality improvement in
manufacturing firms. Though a large number of case studies have been
reported[2] of the implementation of Taguchi’s offline quality control methods
describing the conduct of a minimum number of experiments to decide the
optimum parameters, there is virtually no successful case study in the literature
about the implementation of Taguchi’s online quality control methods (hereafter
shortly referred to as TOLQC methods). However, the theoretical parts of TOLQC
methods applied to hypothetical case studies are described in[3]. This clearly leads
to the categorical conclusion that for TOLQC methods, there exists a large gap
between theory and practice. This article explains a part of the research that has
been carried out to fill this gap by making efforts to move TOLQC methods from
mere theoretical form to manufacturing practice through their implementation in
a typical manufacturing firm.

Research methodology
The first phase of the work constituted a thorough literature survey to study and
review the implementation strategies of Taguchi methods in general and TOLQC
methods in particular. This phase culminated in the evolution of a well-structured
breakthrough action plan, not only to try to study the implementation feasibilities,
but also to replace the quality system encompassing statistical practices with one
that is based on TOLQC methods. The second phase, called the implementation
phase, constituted the selection of a typical manufacturing firm which faces the
usual quality-related tasks of all sizes of manufacturing firms, followed by the
practical implementation of TOLQC methods in it. The article elaborates on the
background and features of Taguchi methods and experiences overcome during
the implementation phase. It concludes by discussing the inferences drawn from
these experiences of this implementation and insists on intensified training and
awareness programmes on the implementation strategies to attain the ultimate
goals of TQM.

Overview of Taguchi methods


Though Taguchi started to publicize his methods from the early 1960s onwards,
the industrial world began to feel them and introduce them collectively as an
IJOPM evolutionary approach only from the early 1980s. Conferences and such other
15,7 deliberations on Taguchi methods began to emerge only from the late 1980s[2].
A thorough literature review on Taguchi methods [2,3] suggests the following
as their important features:
● Infusing quality in the design stage itself, since even stringent inspection
procedures can never fully overcome bad design.
62
● Increased concentration on the quality of design rather than inspection.
● Focusing on “robust design” which designs products so that they are not
affected by disruption, either during production or service.
● Using statistical procedures in the design phase only, at appropriate
stages, and completely removing them from the production phase.
● Relying on the models developed based on the philosophy “loss to the
society” rather than statistical techniques during production phase.
In a nutshell, Taguchi’s approach is based on his way of describing an
intuitively agreed philosophy which he puts under the term quality loss,
defined by him as “the loss a product costs society from the time it is released
for shipment”[3].
Careful study of the above points indicates that quality is required to be
infused during two stages, that is before and during production. The quality
engineering methods proposed by Taguchi for application before production
are known as offline quality control methods and those proposed for
implementing during production are called on-line quality control methods.

Offline quality control methods


The activities employed during design, product planning and development
stages to infuse quality are referred to as offline quality control[3]. The methods
proposed by Taguchi for this basically envisage the process known as “design
of experiments” in which matrices known as orthogonal arrays are suggested
that aid in conducting the minimum number of prototype experiments to
determine the optimum parameters before going for actual production. Today,
Taguchi’s offline quality control methods are popular in various parts of the
world, especially in Japan, Europe, the USA and some parts of developing
countries like India[2,4,5]. Reports of case histories on these methods are
numerous. As a matter of fact, the manufacturing fraternity today views the
application of Taguchi’s offline quality control methods as a necessary step
towards the installation of a prevention-based quality system which is a
prerequisite for successful TQM.

Online quality control methods


The quality engineering activities required during production are referred to as
online quality control. The methods suggested by Taguchi[6] are totally
different from the existing statistical techniques used in traditional
manufacturing firms for they rest solely on the concept that quality loss arises
as a result of deviations from the target. In other words, this is unlike traditional Quality
interpretations where it is assumed that the product, process and even service improvement
attain their full value as long as they fall between certain deviations from the through TOLQC
target. This contrasts with the principles and philosophies of TQM, which
fundamentally emphasize the never-ending cycle of continuous quality
improvement to reach the target. While setting apart the statistical techniques
and tools, Taguchi has contributed a number of models which encompass the 63
intuitively agreed concept that the products, processes and services lose their
quality value as soon as they start to deviate from their targets. These
distinguishing features are shown in Figures 1 and 2.

Online versus offline quality control methods


Though Taguchi’s offline control methods pave the way for the installation of a
prevention-based quality system, their scope in attaining continuous quality

Upper limit
Quality
parameter Target

Lower limit

Product/process/service
Frequency of observation Figure 1.
Traditional
interpretation of
100 per cent loss No loss
quality loss

Upper limit
Quality
parameter Target

Lower limit

Product/process/service
Frequency of observation Figure 2.
Taguchi’s
interpretation of
100 per cent loss Increasing loss from the target quality loss
IJOPM improvement is limited. However, TOLQC methods strongly aid in attaining
15,7 continuous quality improvement which will extend an impetus towards
sustained TQM. A major survey, which ranged from literature available in
international journals and proceedings of national and international
conferences to the interviewing of manufacturers, revealed that the industrial
world has still to accept the benefits of TOLQC methods as meaningful criteria
64 to effect full-fledged TQM. Needless to say, it is high time to initiate the
implementation of TOLQC methods in manufacturing firms.

Taguchi’s models for online quality control


Taguchi has contributed a number of mathematical models on online quality
control with a view to replacing the existing statistical equations and
calculations. Incidentally, all of them find their fundamental base from his
simplest model named “quality loss function” (QLF). Basically, QLF is a
mathematical model that correlates the quality loss in money value to the
deviation from the target specification.
The QLF model is highly meaningful in the total quality organization where
the priority is to improve quality continuously with reduced investment and
with the ultimate objective of producing zero defect products.
Taguchi depicts QLF as follows [3,7]:
O
L( y ) = ( y − T )2
δ 2

or,
O 2 1 n =n 
L( y ) = τ ; τ 2 =  ∑ ( y − T )2 
δ2  n n = 1 
where
L(y) = quality loss in money value;
O = the loss in money value caused due to the out-of-specification range
of quality parameters;
y = measured quality parameter;
T = target quality parameter;
δ = tolerance specification from the target quality parameter;
τ 2 = mean squared deviation.
Fundamentally, QLF attaches a fraction of loss in money value by fitting a
quadratic curve. This is highly intuitively acceptable since it creates the
realization that departure from the target itself costs the product, process and
service quality. The essential features of this model can be understood from
Figure 3. The derivation and rationale behind the development of QLF can be
had from[3]. Figure 3 shows Taguchi’s quality loss curve with the quality level
measurement characteristic of the product denoted as “P” whose fraction of
Quality value loss

Quality
in percentage

improvement
through TOLQC

100
65

P
L (y)

T–R T (y – T ) T+R

Quality parameter value (y )


Key :
T = The target quality level
R = Allowable range
Figure 3.
T+R = Maximum allowable range Taguchi’s quadratic
T–R = Minimum allowable range quality loss curve

quality loss is given by the QLF model. This approach is highly revolutionary
since in the traditional quality evaluation system, the same product “P” is
considered to be 100 per cent acceptable though it has deviated considerably
from the target. That is, it is presumed that, in spite of the deviation from the
target specification, the product suffers no quality loss. This approach is
outdated and meaningless now when the industrial world has realized the need
for moving towards target as the milestone for effecting TQM.
Based on QLF and considering various manufacturing parameters such as
sensitivity analysis, tool maintenance, diagnosis interval, process recovery, etc.,
Taguchi has contributed a number of models[3] to facilitate the evolution of
optimum conditions that will pave the way for continuous quality improvement
at economical investments. Though this research covered five models, due to
lack of space, the scope of this article is limited to describing the manufacturing
application of the model for online quality control of variable characteristics
and its supporting models. The set of models referred to here is applicable to the
typical manufacturing situations in which quality systems encompass the
manual feedback quality control procedures that envisage the adjustment of the
system at an interval of more than one unit of production.
IJOPM TOLQC model for variable characteristics
15,7 The TOLQC model for variable characteristics with feedback control system
which the adjustment interval is more than one production unit considers the
following costs and losses as the yardsticks for the evaluation of quality costs
of the production system:
● Diagnosis cost. This is the investment and expenditure required per
66 product to inspect and diagnose the defects, if any, as and when the
production process is going on.
● Adjustment cost. This is the investment and expenditure required per
product towards correcting the process by making periodic adjustments.
For example, in a production environment the expenditure required for
tool changing, change in tool set-ups, etc., collectively form an
adjustment cost.
● Loss due to deviation. This is the sum of the losses that arise under the
following circumstances:
– deviations observed from the target specifications though the
manufacturing conditions are found to be within control limits;
– deviations observed from the target specifications when the
manufacturing conditions are found to be out of control limits; and
– when the system under study is prone to measurement errors.
The summation of diagnosis cost, adjustment cost and loss due to deviation
constitute the TOLQC model for variable characteristics with adjustment
interval more than one unit of production. This particular TOLQC model is
depicted below.

Present system
 
M O  D p2  n p + 1  D p2
A 2
Lcp = + +  + + s + σm 
np up δ 2  3  2  u
  p 
 
Optimum parameters
1/ 4
 
2u p M δ  3 A D p2  Do 2
no = × ; Do =  × ×δ2 ; uo = u p ×
O Dp O up  D p2
 
Optimized system
 
M O  Do2  no + 1  Do2
A 
Lco = + +  + + s + σm 
2
no uo δ 2  3  2  uo 
 
where, Quality
L =
quality cost expended in the present system; improvement
Lco =
quality cost expended in the optimized system; through TOLQC
M =
expenditure incurred for each quality characteristic measurement;
np =
measurement interval in number of production units for the present
system; 67
no = measurement interval in number of production units for the
optimized system;
A = expenditure incurred for each adjustment;
up = adjustment interval in number of production units for the present
system;
uo = adjustment interval in number of production units for the optimized
system;
O = the loss of money value caused due to the out-of-specification range
of quality parameters;
δ = tolerance specification from the target quality parameter;
Dp = adjustment control limit for the present system;
Do = adjustment control limit for the optimized system;
s = number of production units surpassed or stopped during each
measurement;
σm = measurement error.
The measurement error σm can be calculated using the following equation:

1 
( ) ( ) ( )
2 2 2
σm
2
=  m2 − m1 + m3 − m2 + L mn − mn − 1 
n − 1  
where
n = number of units measured in an experiment;
m1, m2, m3, . . . , m4 and mn – 1 refer to the quality characteristic values
observed for first, second, third,…, (n – 1)th and nth production units
respectively.
Readers are advised to refer to [3] if they are interested in having more details
such as derivation, assumptions made, theories used, etc. regarding the above
model and relationships.

Breakthrough action plan


After analysing the basic nature of TOLQC methods and the present attitude
prevailing in manufacturing firms with regard to TQM, it was realized that the
design of a well-structured and focused “breakthrough action plan” was
IJOPM required. The authors made thorough studies of the different procedures and
15,7 sequences proposed by various quality engineering experts and
academicians[7-12] for carrying out quality improvement projects successfully.
The theoretical knowledge acquired during these reviews was presented to a
number of manufacturers and their views were collected. Based on this
collective comparative study, the authors evolved a “breakthrough action plan”.
68 This formed an essential part of this research since, besides calling for technical
computations, TOLQC methods emphasize more management commitment
and employee involvement for their successful implementation. The
breakthrough action plan designed for successfully implementing the TOLQC
methods during this research is shown in Figure 4.
The following sections describe the essential features and background of the
breakthrough action plan.

Initiation
The breakthrough action plan begins by obtaining management commitment
and support. These are vital since the process of applying TOLQC methods has
to pass through various critical activities such as major change in quality
policies, data collection, study of the present system, etc. After this, the tedious
task of identifying production and product problems with the active
consultation of managerial personnel and operating personnel is carried out.
The problems identified must be subjected to either intuitive or systematic
Pareto analysis with respect to their importance in quality improvement and
financial viability. This is important in the sense that it reduces the chances of
considering non-economical problems which will not improve profitability.

Experimentation
After the identification of the problems, several brainstorming sessions are
conducted at different levels of the organization to identify the critical
combination of factors which play a significant role in quality improvement and
deserve attention. After this, the experiment to be conducted is planned
considering various interacting practical factors such as non-disturbance to
production, co-operation of employees, etc. Then, the experiment is conducted
and the results are carefully collected and recorded. During this process, care
has to be taken to see that the experiment is conducted in the same condition as
the actual. After this, using the recorded data, the quality losses and costs in
terms of money value for all production and/or product problems identified are
computed using QLF and TOLQC models for assessing the quality levels and
investments respectively. Using TOLQC mathematical relationships (shown in
the model), the combination of factors is identified which will reduce the quality
costs to the least extent without moving further away from the target
specification. This can also be supplemented by brainstorming sessions to
unearth more optimum parameters which will reduce quality cost. Now the
optimized factors and parameters are incorporated into the system and the
confirmation experiments are run. The results of the confirmation experiments
Obtain management commitment and support
Quality
improvement
through TOLQC
Listing production/product problems

Brainstorming sessions to identify 69


critical factors

Design experiment

Run experiments; record results

Compute quality loss using


Taguchi's quality loss function (QLF)

Compute quality cost using TOLQC model

Identify combination of factors giving least


quality cost closest to target specification

Modified factors set to least


quality loss levels

Run confirmation experiment

Check for missed Compute the quality cost using


factors (S) TOLQC model

Compute quality loss using QLF

Not significantly Quality cost and No


loss are decreased

Yes

Get management commitment

Figure 4.
Install improved quality system Breakthrough action
plan
IJOPM are recorded and the quality losses are calculated using the QLF model. The
15,7 quality losses before and after the incorporation of optimum parameters and
factors are compared.

Management presentation
If the computed quality cost and loss have decreased compared to the previous
70 computation, then the findings are reported to the management. After getting
the approval of the management, the quality system encompassing modified
quality policy and procedures is evolved. Since continued pressure is placed on
industry towards ISO 9000 compatible quality systems, it is to be ensured that
the modified system is at least nearly compatible to any one of the ISO 9000
series quality system standards.
However, if the quality loss is not reduced, then the factors which might have
been missed are incorporated and the confirmation experiment is repeated. The
process is repeated until the quality loss is reduced to a significant level.
Moreover, if the quality loss is decreased by only an insignificant amount, it is
advisable again to conduct brainstorming sessions to identify the factors and
levels which are economically important for consideration and the whole
procedure is repeated.

Case study
As previously stated, the second phase of the research constituted the actual
implementation of TOLQC methods. This phase can be considered as the heart
of this research for the objective of the whole work is to aim at transferring the
theoretically available knowledge on TOLQC methods to practice which shows
the path for continuous quality improvement. Extreme care was taken to select
a manufacturing firm for it should reasonably represent the global managerial
attitude towards TQM. Out of 55 manufacturing firms with various product
ranges, a pump manufacturing firm with the following quality managerial
attitudes was selected:
● The management is in the pursuit of producing world-class quality
products to face tough global competition.
● The management shows keen interest in implementing TQM and has
made considerable financial investment in this regard.
● Management had initiated a quality circle movement ten years previously.
After initial minor setbacks, the quality circle movement has gained
momentum and employees, especially workers, feel more recognized.
● As part of TQM, from the early 1980s, the management had been
spending a lot towards training employees of all levels. However, even
today no visible benefits are observed.
● Management feels dejected since no significant breakthrough with regard
to continuous quality improvement is observed despite the remarkable
investment made for TQM and quality circles programmes.
● Throughout the firm statistical techniques are used for checking quality Quality
parameters. improvement
● Management slowly starts to consider whether there is any need for through TOLQC
further investments towards TQM programmes, for they see no visible
quality improvement or financial benefits.
● Management is trying hard to get ISO 9001 quality system certification. 71
Since the quality management related atmosphere mentioned above strongly
coincides with most of today’s manufacturing firms, especially in developing and
developed countries, the authors decided that it would be prudent to analyse the
implementation strategies of TOLQC methods in the pump manufacturing firm
selected, for the experiences and inferences drawn out would represent highly
worthwhile samples. The following sections briefly narrate their experiences
when they attempted to implement the TOLQC methods in the firm selected
during different stages.

Obtaining management commitment and support


Since the management has been showing keen interest to effect TQM, there was no
initial reluctance to implement TOLQC methods. However, when top management
was informed of the need for collecting large amounts of data, studying the entire
manufacturing system and modifying the present quality policy, there was some
hesitation. At this juncture, it became necessary to conduct a series of
presentations on TOLQC methods to highlight their essence. Moreover, it was
assured that strategic quality planning procedures would be drawn such that the
implementation study did not affect normal working of the firm. In this way, the
authors succeeded in getting management commitment and support with few
terms and conditions.

Listing production/product problems


During this stage, the authors stepped on to the shopfloors and had informal talks
with supervisors and operators in order to collect the production and/or product
problems. Since some of the operators did not want to reveal their identity, the
information was collected anonymously and treated as confidential. In addition,
the proposals from various quality circles were also collected to prepare the list of
problems. The collected problems were referred to the middle and top
management to check their authenticity. Those that were not connected with
quality and those whose solving strategies were not within the reach of the
management were dropped. Finally, 156 problems which deserved immediate
attention were listed.

Brainstorming sessions to identify critical factors


The listed problems were analysed thoroughly by conducting a number of
brainstorming sessions, during which tools like Pareto analysis and cause- and-
effect diagrams were used to identify the product that not only is susceptible to
frequent quality problems, but also affects the profit margin to a considerable
IJOPM extent. At the end of this stage, it was realized that at that time the most pressing
15,7 need was to show interest in continuous quality improvement of the direct coupled
centrifugal pump, not only for quality-related reasons but also from the
profitability point of view since it is the fastest-moving major product of the firm.
Again, a few rounds of brainstorming sessions were conducted to identify the
vital few out of 29 parts of this pump. At the end, it was decided to concentrate the
72 study on four parts, namely, the coupled pump shaft, the impeller, supporting
bracket and supporting frame side half casing. In all brainstorming sessions,
apart from the ideas of the authors and managerial personnel, the suggestions of
experienced operators were also pooled, which ensured the tapping of knowledge
from experience.

Design of experiments
For the four parts selected for the study, the experiments were designed to involve
the study of quality from the viewpoint of product, process and service
considerations. During this stage, extreme care had to be taken to ensure that the
experiments were not disturbing the production schedule to any great extent. All
designed experiments were subjected to either oral or written approval of
concerned managerial personnel not only to ensure smooth conduct but also to get
their continued commitment and support. A total of 55 experiments were designed
at the end of this stage.

Running the experiments and recording the results


During this stage, the experiments were run and the results were recorded without
affecting the production to any great extent. As far as possible, the experiments
were made to coincide with the routine practices which ensured not only the non-
disturbance to production, but also considerable time savings in experimental set-
ups. An important problem that was encountered during this stage was that, since
the experiments relied mainly on measurements, it was necessary to compel the
inspection personnel to remove various unused measuring instruments from
storage. Hence, it can be inferred that TOLQC methods relied on the increased
application of measurement technology rather than mere comparisons by gauges.

Computing quality losses and costs


The recorded results were used to calculate the quality losses and costs using QLF
and appropriate TOLQC models respectively. The computation of quality losses
was necessary to judge the quality level whereas the computation of quality costs
was necessary to estimate the investment for quality. Also, the computation of
quality losses and costs resulted in a realization of the seriousness of quality
problems in terms of money value.

Identifying combination of factors giving least quality loss closest to target


specifications
While some of the combinations of factors giving least quality loss closest to
target specifications, such as change in process parameters, frequency of tool
changes and set-ups, etc. were identified by using TOLQC models themselves, Quality
some were identified by conducting a few brainstorming sessions. The improvement
brainstorming sessions were found to be very helpful during this stage for they through TOLQC
not only aided the identification of optimum parameters but also in checking the
practical feasibility of applying the combination of factors identified by the
TOLQC models.
73
Running confirmation experiments
The combination of factors identified during the previous stages were
incorporated into the systems and the confirmation experiments were run.
Though some problems which are common to any quality improvement
programmes, such as resistance to change, scepticism regarding new suggestions,
hindrance to the routine work, etc. arose from different quarters, they were all
tackled efficiently by maintaining good human relationships. As in the previous
experiments, the results were recorded systematically.

Computing quality losses and costs


The recorded results were used to again compute the quality losses and costs
using QLF and appropriate TOLQC models respectively to check for the practical
validity of the optimized systems.

Quality analysis by comparison of quality losses and costs


The computed results were analysed for the improvement in quality by comparing
the quality losses and costs of the existing and optimized systems. If the quality
losses and costs had decreased considerably in the optimized systems compared
to the previous ones, then the optimized parameters were approved and the
optimized experiments were declared successful. Out of 55 experiments
conducted, 40 were declared successful in this manner. However, in ten
experiments, though the quality costs were reduced, no considerable decrease in
quality losses was observed. Those experiments were again subjected to
brainstorming discussions to identify missed factors, which were then
incorporated and subjected to the confirmation experiments which were proved
successful. In five confirmation experiments, the quality losses were not reduced
at all. Hence, major lapses in the design and conduct of experiments were
suspected and the whole procedure was repeated from the beginning. All these
experiments turned out to be successful except two, in which case even after three
repetitions of the whole processes the missed combination of factors could not be
traced. The problems connected with these experiments were referred to the co-
ordinators of quality circles, who, after conducting a number of quality circle
meetings, unearthed the missed combination of factors. The confirmation
experiments with the incorporation of these missed combination of factors
revealed a considerable decrease in quality loss which led to the declaration that
the experiments ended successfully.
Thus, all the designed experiments were successfully conducted apart from a
few hurdles that are common to any quality improvement programme[13], which
IJOPM were solved amicably from time to time. In order to give an overall view of the
15,7 computation methodology adopted during the experiments, sample computations
carried out during the experiments for coupled pump shaft are given in the
Appendix.

Getting management approval


74 The quality losses and costs computed after all confirmation experiments revealed
that a considerable degree of improvements in quality was possible in products,
processes and service with reduced investment of Rupees 2,205,000 per annum
which is about 22 per cent of the total turnover of the firm. Before preparing the
final report, the outcomes of all experiments were presented to middle managers.
This presentation was very helpful to check whether the proposed optimum
combination of factors and parameters was feasible from a management point of
view. Though some of the factors and parameters had to be dropped on the advice
of middle managers, the savings in investment for quality still worked out at
Rupees 2,134,000, the magnitude of which was considered to be significant.
Thus, after getting the concurrence and support of all, the final report was
prepared. While preparing the report, extreme care was taken to avoid
unnecessary jargon and lengthy descriptions that would only induce boredom and
fatigue. Lucid and clear language and pictorial representations were used which
would not only facilitate easy reading but also quick grasping of information. The
report was submitted to top management who, after going through it, called for a
brief presentation. After incorporating a few changes suggested during this
presentation, the proposals of the report were approved.
Today, in a progressive manner, the firm is implementing the proposals, which
have started to deliver fruitful results. As a matter of fact, the management, which
initially was sceptical of TOLQC methods, today feels overwhelmed on seeing the
benefits accruing out of them. Especially, the fact that TOLQC methods indirectly
stress the increased and efficient use of measurement technology is being
appreciated from all quarters.

System installation
Since the ultimate goal of TOLQC methods is itself continuous quality
improvement, it was realized that, in order to ensure perpetual application of these
methods, a quality system encompassing quality policy, procedures, manuals,
records and work instructions was imperative. It was realized that this research
would become meaningful and useful only when the quality system pertaining to
TOLQC methods was developed and implemented. One important point that
required special emphasis was that the management was interested in developing
the quality system compatible to ISO 9001 standard. However, ISO 9001
emphasizes statistical techniques in its clause 20 whereas the TOLQC approach
calls for the total elimination of statistical techniques during production. Hence,
barring clause 20 of ISO 9001, all other clauses were considered and a nearly ISO
9001 compatible quality system was proposed which is being implemented
stage by stage.
Conclusions Quality
This article explained a part of research which was carried out to study improvement
feasibilities of TOLQC methods as a means to effect TQM. The first phase of the through TOLQC
work constituted the detailed literature study of all Taguchi methods and their
implementation strategies, particularly with reference to on-line quality control
methods followed by evolving a feasible breakthrough action plan to implement
them in typical manufacturing systems. The second phase constituted the 75
actual implementation of TOLQC methods, attempted in a pump
manufacturing firm. The most important observation made during this phase
was that the initial scepticism of management regarding TOLQC methods
began to vanish when the implementation study started to yield fruitful results.
It was proved that the quality level could further be improved with less
investment on applying TOLQC methods. Moreover, the experience in
implementing TOLQC methods reveals that human relationship and
management are to be given priority over computation, since there is an indirect
emphasis on total cultural change. The research proceeded further by
developing a quality system exclusively for the perpetual implementation of
TOLQC methods. Also, it is emphasized that, in order to derive further benefits,
training and awareness programmes are to be initiated on a large scale. Finally,
it is concluded that in this tough global competitive era, in which the industrial
world is at the threshold of anxiety to exercise continuous quality improvement
in manufacturing firms, the work carried out in this research imbibes a sense of
confidence over the successful application of TOLQC methods as the means to
effect continuous quality improvement, which is the backbone of TQM.

References
1. Dhillon, B.S., “Quality circle: bibliography”, International Journal of Reliability & Quality
Management, Vol. 5 No. 1, 1988, pp. 53-75.
2. Bendell, T. (Ed.), Taguchi Methods – Proceedings of the 1988 European Conference,
Elsevier Applied Science, London, 1988.
3. Taguchi, G., Elsayed, E.A. and Hsiang, T., Quality Engineering in Production Systems,
McGraw-Hill, New York, NY, 1989.
4. Nagaraj, N. and Mathew, T., “A case study on designing optimum process parameters
using Taguchi technique”, Proceedings of the 35th Annual Convention of Indian
Institution of Industrial Engineering, Tata McGraw-Hill, India, 1993, pp. 251-7.
5. Reddy, P.B.S. et al., “Application of Taguchi’s robust design methodology to copper coating
process – a case study”, Proceedings of the 35th Annual Convention of Indian Institution
of Industrial Engineering, Tata McGraw-Hill, India, 1993, pp. 295-304.
6. Cheng, T.C.E., “Quality control: changing with the times”, International Journal of Quality
and Reliability Management, Vol. 7 No. 6, 1990, pp. 37-46.
7. Ashley, S., “Applying Taguchi’s quality engineering to technology development”,
Mechanical Engineering, July 1992, pp. 58-60.
8. Goh, T.N., “Taguchi methods: some technical, cultural and pedagogical perspectives”,
Quality and Reliability Engineering International, Vol. 9, 1993, pp. 185-202.
9. Juran, J.M., Quality Planning and Analysis, 3rd ed., McGraw-Hill, New York, NY, 1980.
10. Crosby, P.B., Quality is Free, McGraw-Hill, New York, NY, 1980.
IJOPM 11. Deming, W.E., Out of the Crisis, MIT Press, Cambridge, MA, 1982.
12. Ross, P.J., Taguchi Techniques for Quality Engineering, McGraw-Hill, New York, NY, 1988.
15,7
13. Modarress, B. and Ansari, A.A., “Two new dimensions in the cost of quality”, International
Journal of Reliability & Quality Management, Vol. 4 No. 4, 1987, pp. 9-20.

Appendix
76 In order to illustrate the computation methodology adopted during the experiments, sample
computations carried out during the experiments for coupled pump shaft (referred to as shaft
hereafter) are given here. In this shaft, one portion is considered as a critical area where the
bearing is to seat. The diameter of this portion is 25mm with tolerance of + 0.01mm. After
turning, the final operation is done in a cylindrical grinding machine. In spite of long experience
and improved skills, the operators and supervisors had not been able to identify the combination
of factors which would lead to the least occurrence of quality problems. The relevant information
collected during different stages of the experiment is:
Expenditure incurred for each quality characteristics measurement (M) = 17 paise = Rupee
0.17;
Expenditure incurred for each adjustment (A) = Rupee 0.34.
The above two expenditures were estimated based on the fact that a cylindrical operator is paid
a wage of ten Rupees per hour. Therefore, when he spends one minute for quality characteristic
measurement, it is inferred that the firm spends Rupee 0.17. Likewise, when he spends two
minutes for each adjustment it is inferred that Rupee 0.34 is expended for that purpose each time.
Measurement interval in number of production units for the present system (np) = 1;
Adjustment interval in number of production units for the present system (up) = 30;
Adjustment control limit for the present system (Dp) = 0.01mm;
Number of production units surpassed or stopped during each measurement (s) = 2.
The experiment was run for the present system and the consecutive measurement of 35
components were made. The measurement error (σm2 ) as calculated using the relationship shown
in the TOLQC model which worked out to 0.000036mm2. With this information, the quality cost
was calculated as given below:

0.16 0.36 110  0.01 2  1 + 1  0.01 2 


Lcp = + +  + + 2 + 0.000036 
( 0.01 )  3  2  30 
2
1 30
= Rupees 87.45.

The above result reveals the fact that a sum of Rupees 87.45 is spent as quality cost for each shaft
in the present system. It is considered to be alarming considering that the market value of the
shaft itself is only Rupees 110. That is, quality cost forms around 80 per cent of its price. Hence,
it was decided to find the combination of optimum factors and parameters that will reduce the
quality cost significantly. The details are given below:
Measurement interval in number of production units for the optimized system:

2 × 30 × 0.16 0.01
no = × = 0.2945 = 1 unit (approximately)
110 0.01

Adjustment control limit for the optimized system:


 3 × 0.32 0.01 2 
1/4
Quality
Do =  × 0.01 
2
× = 0.013mm improvement
 110 30  through TOLQC
Adjustment interval in number of production units for the optimized system:
2
0.0013
u o = 30 × = 50.7 = 50 (rounding off to minimum).
2 77
0.0013
Followed by the computation of the above optimum parameters, strategies were drawn to
incorporate them in the present system. Moreover, the decision was taken to incorporate some
more features such as change in cutting speed, feed rate, etc. in the present system to achieve a
higher quality level.
Before incorporating these optimum parameters and features, however, it was necessary to
check whether the quality level was increased in spite of savings in quality cost. Hence, the
quality level of the coupled pump shaft manufactured in the present system was gauged by
computing quality loss using Taguchi’s QLF model. Out of the 35 shafts measured, the mean
squared deviation (τ2 ) was found to be 0.0000294mm2. Therefore, quality loss of each shaft under
the present system is,
110
Lp = × 0.0000294 = Rupees 32.35.
2
0.01
The quality cost expended after the incorporation of the optimum parameters is computed as
follows:

0.16 0.32 110  0.000132  1 + 1  0.00132 


Lco = + +  + + 2 + 0.000036 
( 0.01 )   2  50 
2
1 50 3
= Rupees 40.50
This means that, to manufacture a shaft in the optimized system with optimum combination of
factors, the quality cost to be expended is Rupees 40.50. Also, it reveals that a saving of Rupees
46.95 (that is Lcp – Lco = 87.45 – 40.50 = 46.95) per shaft in quality cost on installation of the
optimized system. This saving is realized to be enormous on considering the fact that the
company manufactures 15,000 coupled pump shaft per annum and hence the total saving per
annum will be Rupees 704,250 (that is 15,000 × 46.95).
To check whether the quality level has improved in spite of reduced quality cost, a confirmation
experiment with the incorporation of the optimum combination of factors and parameters was
run. The consecutive measurement of 50 coupled shafts revealed a mean squared drift of 0.0000202
mm2. Hence, quality loss for each coupled pump shaft manufactured in the optimized system is,
110
Lo = × 0.0000202 = Rupees 22.20.
2
0.01
This computation reveals the decrease in quality loss in the optimized system compared to the
present system, thereby proving that the quality level is improved in the optimized system.
Therefore, the confirmation experiment was declared successful and the optimized combination
of parameters and features was approved. This was followed by report preparation.
Meanwhile, though the proposed optimized system shows enormous savings in investment
with an increase in quality level, in order to explore further the possibilities of decreasing the
quality cost without retarding the quality level, the situation was referred to the quality circle co-
ordinators. This was done to ensure non-stop continuous quality improvement even after the
installation of the optimized system.

You might also like