Design for quality inagile manufacturing

environment through modifiedorthogonal array-based

experimentationS.R. Devadasan

Department of Production Engineering, PSG College of Technology,Coimbatore, India

S. GoshteeswaranDepartment of Production Engineering, Amrita Institute of Technology,

Coimbatore, India, and

J. GokulachandranDepartment of Mechanical Engineering, Amrita Institute of Technology,

Coimbatore, India

Abstract

Purpose – To provide a modified orthogonal array-based model for enabling the researchers andpractitioners to exploit the technique, “design of experiments” in an agile manufacturing environment.

Design/methodology/approach – The characteristics of Taguchi’s off-line models and agilemanufacturing were studied. A theoretical model of modified orthogonal array-based experimentationwas designed. This model was subjected to implementation study in an Indian pump-manufacturingcompany.

Findings – The model contributed in this paper has shown its feasibility in achieving quality in agilemanufacturing environment.

Research limitations/implications – The authors are residing in an Indian city where themajority of the companies have not adopted agile manufacturing criteria. Hence, it was not possible tocarry out implementation study in an agile manufacturing company. Future researchers shouldexamine the practical validity of the proposed model in agile manufacturing companies.

Practical implications – Since the manufacturing organizations are fast becoming agile, due to thecustomers’ dynamic demands coupled with competition, the traditional quality improvementtechniques are becoming obsolete. The model contributed in this paper is found to be useful inachieving continuous quality improvement in AM environment. Hence the model would be a usefultechnique for today’s practitioners whose activities are increasingly focused towards achieving agilityin manufacturing.

Originality/value – The literature survey covering articles on agile manufacturing indicates that noresearcher or practitioner has contributed a model that would exploit the technique, “design ofexperiments” in an agile manufacturing environment. Hence the proposed model is expected to be ofhigh value for researchers and practitioners to explore the way of achieving continuous qualityimprovement in agile manufacturing environment.

Keywords Agile production, Experimentation, Quality improvement, Delphi method, Taguchi methods

Paper type Research paper

The Emerald Research Register for this journal is available at The current issue and full text archive of this journal is available at

www.emeraldinsight.com/researchregister www.emeraldinsight.com/1741-038X.htm

JMTM16,6

576

Received August 2003Revised January 2004Accepted March 2004

Journal of Manufacturing TechnologyManagementVol. 16 No. 6, 2005pp. 576-597q Emerald Group Publishing Limited1741-038XDOI 10.1108/17410380510609456

IntroductionDuring the 1970s and 1980s, manufacturing organizations were compelled to focus onenhancing productivity and quality due to the emanation of competitive businesssituation. Of late, it has become so fierce that manufacturing organizations are lookingfor a strategy to tame the situation (Rooks, 2000). It has been widely acknowledged thatthis predicament is due to customers’ requirements, which change very often.Recognizing and imparting them in practice are very much vital for survival of anyorganization. Hence, the manufacturing community has been forced to respond in sucha manner that their production should be low in volumes and high in varieties(Maskell, 2001). Awareness of this state of affairs made some researchers to contributetheir might in this regard and they brought out the paradigm called “agilemanufacturing” (AM) (Hooper et al., 2001). In a nutshell AM is a combination of flexiblemanufacturing system concepts and lean manufacturing principles (Sarkis, 2001) andcalls for quick response to the customers’ demands (Parkinson, 1999). On realizing itsimportance in attaining competence in modern scenario, various forums have beenstarted exclusively for promoting AM principles (Struebing, 1995). As AM paradigm isstill at its nascent stage, significant works are being carried out to make it operationalin today’s organizations (Sharifi and Zhang, 2001). Particularly, the task of researchersis focused towards modifying the existing manufacturing strategies so as to makethem compatible in an AM environment (van Assen et al., 2000, Meredith and Francis,2000). Among all manufacturing strategies, continuous quality improvement comes tothe forefront. At this juncture, the contribution of Taguchi methods towards achievingcontinuous quality improvement is noted with interest. The large volume ofapplications of Taguchi methods reported in literature (Bendell, 1988; Antony andAntony, 2001) signifies their power. Hence it was thought that the exploitation ofTaguchi methods in AM environment would contribute considerably towards theintegration of quality improvement approaches in an AM environment. Taguchi, aJapanese born quality guru, has contributed a number of models, which he hasclassified into off-line and on-line models (Aravindan et al., 1995). As the name implies,off-line models are applied before the actual production starts, whereas on-line modelsare applied during production. This paper reports the research work in whichTaguchi’s off-line model was considered and its role in bringing out agility inmanufacturing companies was examined through subjective and non-subjectiveexperimentation.

Distinguishing features of agile manufacturingA literature overview indicates that the term AM was coined in the early 1990s(Ramasesh et al., 2001). According to one definition:

AM is the science of a business system that integrates management, technology andworkforce making the system flexible enough for a manufacturer to switch over from onecomponent that is being produced to another component that is desired to be produced in acost-effective manner, in a short time within the framework of the system (Chowdiah, 1996).

Fundamentally AM emphasizes upon the design and development of manufacturingsystems which would enable the enterprises to quickly respond to the dynamicdemands of the customers (Hormozi, 2001) without sacrificing the profit margin (Poweret al., 2001). Although various definitions and interpretations of AM are present in

Agilemanufacturing

environment

577

literature, it is very rare to find research papers reporting the distinguishingcharacteristics of AM.

One of the present research requirements is the identification of distinguishingfeatures of AM from that of traditional manufacturing principles. A literature overviewcombined with appropriate interpretations enabled us to identify 20 distinguishedfeatures (van Assen et al., 2000; Bajgoric, 2001; Crocitto and Youssef, 2003; Goldmanet al., 1995; Montgomery and Levine, 1996; Narain et al., 2000; Onuh and Hon, 2001;Pawar and Shaiffi, 2002; Power et al., 2001; Spring and Dalrymple, 2000). The detailsare presented in Table I. As shown, first, AM calls for organizational restructuring byincorporating team-based architectures (Ebrahimpur and Jacob, 2001). Someorganizations have already started progressing in this direction. For example, mostof the software developing companies adopt team management principles to work withthe focused approach for attaining goals specified to teams. This process is facilitatedby instituting closely packed human and information networks, which significantlyimprove both quality and the productivity. Another important criterion is the aspect ofconverting static organization into learning organizations. In traditional organizations,employees after getting initial training become specialists in their area of work. In anAM environment, as processes and products are required to be frequently varied,employees are required to learn new knowledge and skills continuously. For thispurpose the manufacturer should be willing to promote and invest on relevantactivities that would facilitate experimentation and innovations (Lee et al., 2000).Appropriate measures should be taken to tune the employees to become learners ratherthan mere job executors. Another salient point insisted is about the supply chainmanagement (SCM) (Power et al., 2001). Traditional organizations adopt onlysubcontracting procedure whereas in an AM environment, depending upon thecustomer’s dynamic demands, new technologies and methods would have to beincorporated as developing the same within the organization would be quiteuneconomical. Hence, the need arises for employing suppliers as partners. As suitablesuppliers are to be selected amongst many and they have to be managed, SCM assumesspecial significance in an AM environment.

An important point of concern is about the time required for an organization toattain agility (van Assen, 2000). This will vary depending upon various factors. If bothmanagement and employees are enthusiastic, the transformation time needed will bevery less. Likewise if the management has already adopted team managementstructure, the time required will be very much less. Another aspect that concerns themanufacturing community more is the strategies to be adopted in an AM environment.The major manufacturing strategies are quality improvement, productivityenhancement, time management and cost reduction (Murugesh et al., 1997). Of all,quality assumes special importance, as quality has been a major factor in attainingcompetence and sustenance in global market. Meanwhile there is a fear amongmanufacturing community about the increased investment and changing approachesbeing called for in attaining higher level of quality in the emerging AM paradigm. Inorder to overcome this state of affair, it is suggested that the existing quality strategiesmay be adopted in an AM environment and the techniques and tools for attaining themmay be suitably refined (Crocitto and Youssef, 2003). The contribution in this direction,by either theorists or practitioners, most preferably by both will allay the fear ofmanufacturers over the implementation of quality strategies in an AM environment.

JMTM16,6

578

Sn

o.C

rite

ria

Tra

dit

ion

alm

anu

fact

uri

ng

com

pan

yA

gil

em

anu

fact

uri

ng

com

pan

y

1O

rgan

izat

ion

alst

ruct

ure

Ver

tica

l,tr

adit

ion

alan

dli

ne

Fla

tten

edan

dte

am-m

anag

ed2

Dev

olu

tion

ofau

thor

ity

Sca

tter

edan

dn

on-i

nte

gra

ted

Sel

f-au

ton

omou

san

din

teg

rate

d3

Man

ufa

ctu

rin

gse

t-u

ps

Rig

id,

lon

g-l

asti

ng

and

into

lera

ble

toch

ang

esF

lex

ible

,ea

sily

coll

apsi

ble

,an

dq

uic

kre

spon

seto

chan

ges

4S

tatu

sof

qu

alit

yC

ust

omer

s’sa

tisf

acti

onC

ust

omer

s’d

elig

ht

5S

tatu

sof

pro

du

ctiv

ity

Un

ifor

mw

ith

no

reas

onab

leev

alu

atio

nan

dim

pro

vem

ent

Rap

idin

crea

sew

ith

pra

ctic

ally

feas

ible

eval

uat

ion

and

inte

gra

tion

wit

hq

ual

ity

6E

mp

loy

ees’

stat

us

Ex

iste

nce

ofsp

ecia

list

s.N

oex

pos

ure

toot

her

fun

ctio

ns

and

skil

ls.

Infl

exib

lean

dig

nor

ant

toch

ang

es

Lea

rnin

gem

plo

yee

s.M

ult

i-sk

ille

d,

mu

lti-

fun

ctio

nal

and

self

-com

mit

ted

emp

loy

ees

7E

mp

loy

eein

vol

vem

ent

Ver

yfe

eble

,id

eas

and

kn

owle

dg

ear

ese

ldom

shar

edor

uti

lize

dF

ull

emp

ower

men

t,id

eas

and

kn

owle

dg

ear

efu

lly

uti

lize

d8

Nat

ure

ofm

anag

emen

tA

uto

crat

ican

dst

agn

ant

Lea

rnin

gor

gan

izat

ion

and

man

agem

ent

9C

ust

omer

resp

onse

adop

tion

Ver

ysl

owly

,ta

kes

pla

ced

ue

tob

ure

aucr

acy

Ver

yfa

stan

d10

0p

erce

nt

resp

onse

env

isag

ed10

Pro

du

ctli

fecy

cle

Lon

gan

din

effe

ctiv

eS

hor

tan

def

fect

ive

11P

rod

uct

serv

ice

life

Lon

gan

din

flex

ible

Sh

ort

and

flex

ible

12D

esig

nim

pro

vem

ent

Ver

yra

re,

gen

eral

lyon

lym

odifi

cati

ons

are

don

eV

ery

freq

uen

tly

and

syst

emat

ical

lyex

erci

sed

by

con

du

ctin

gex

per

imen

ts13

Pro

du

ctio

nm

eth

odol

ogy

Dom

inat

edb

yin

tern

alm

anu

fact

uri

ng

Dom

inat

edb

ym

ain

asse

mb

lyan

dex

tern

alm

anu

fact

uri

ng

and

pro

cure

men

t14

Man

ufa

ctu

rin

gp

lan

nin

gL

ong

and

cost

-in

effe

ctiv

enes

sS

hor

t,ju

st-i

n-t

ime

pu

rch

ase

and

leas

tid

lein

ves

tmen

t15

Cos

tm

anag

emen

tT

rad

itio

nal

typ

eA

ctiv

ity

-,st

rate

gy

-,q

ual

ity

-an

dp

rod

uct

ivit

y-b

ased

cost

ing

syst

ems

16A

uto

mat

ion

typ

eD

irec

tan

dri

gid

auto

mat

ion

Fle

xib

lean

dad

apta

ble

toau

tom

atio

n17

Info

rmat

ion

tech

nol

ogy

(IT

)in

teg

rati

onIn

teg

rati

onof

exis

tin

glo

ng

pro

ced

ure

sw

ith

info

rmat

ion

tech

nol

ogy

(IT

)(t

his

isre

ferr

edto

asd

irec

tin

teg

rati

on)

Pro

ced

ure

sar

esh

orte

ned

and

sim

pli

fied

bef

ore

inte

gra

tin

gw

ith

IT;

max

imu

mu

tili

ties

ofIT

dev

ices

are

inco

rpor

ated

(th

isis

refe

rred

toas

“sim

pli

fied

ITin

teg

rati

on”)

18C

han

ge

inb

usi

nes

san

dte

chn

ical

pro

cess

Ver

yd

iffi

cult

toin

corp

orat

e,al

mos

tim

pos

sib

leE

con

omic

al.

Pre

vio

us

set-

up

enab

les

econ

omic

alch

ang

ein

set-

up

19T

ime

man

agem

ent

Ver

yin

effi

cien

tE

ffici

ent

20S

up

ply

chai

nm

anag

emen

tS

ub

con

trac

tin

gty

pe

Par

tner

ship

wit

hsu

pp

lier

s

Table I.Distinguishing featuresof traditional and agile

manufacturingenvironments

Agilemanufacturing

environment

579

Hence, a small piece of research work was carried out for investigating the suitabilityof applying Taguchi’s off-line model in an AM environment. Before discussing thedetails of this work, the salient features of Taguchi’s off-line model are presented in thenext section.

Taguchi’s off-line modelIn recent years, significant efforts are being made in manufacturing arena to reduce thelead-time of design and production (Ebrahimpur and Jacob, 2001). For example,currently it takes more than a year to float an automobile vehicle into market even afterthe successful completion of its design. The main reason attributed to this situation isthat, production cannot be started before the design is subjected to exhaustive testing.In order to test the validity of design, prototype models are constructed and a largenumber of experiments are required to be conducted. This process consumessignificant amount of time and money and requires patience. Taguchi has offeredsolution to overcome this situation (Taguchi, 1989; Ross, 1996). The entire procedureinvolved in conducting experiments (Antony and Antony, 2001) using Taguchi’soff-line model is depicted in Figure 1. As shown, the process starts with the precisedefinition of the problem. Based on the problem definition, the factors that areresponsible for the existence of the problem are identified. Against each factor, thelevels at which they act are to be identified. For a single problem, a number of factorswith at least two levels can be identified. A large number of experiments are requiredto be conducted to identify the best combination of factors and their correspondinglevels. However, it is not viable to conduct all those experiments.

In order to reduce the number of experiments without losing the chance ofidentifying the optimal combination of factors and their corresponding levels, Taguchihas advocated the exploitation of Latin square along with a graphical tool named aslinear graph. The Latin square with its corresponding linear graph is called asorthogonal array (OA). The elements of OAs were developed based on equally likelyprobability of occurring the sample of events from a lot. OAs are denoted by Taguchias L4, L8, L16 etc. Presumably “L” refers to Latin square. The suffixes indicate thenumber of rows of the corresponding OA. The number of rows indicates the number ofexperiments to be conducted. Figure 2 shows L4 orthogonal array. The factors areassigned to the columns according to the specification of linear graph. Columns 1 and 2can exist separately and the column 3 would contain the elements, which would reflectthe combination of the elements of columns 1 and 2. The numbers in the matrixindicate the levels of the corresponding factors.

The design of experiments using OAs is illustrated by considering the L4 OAshown in Figure 2. Experiment 1 is designed by considering level 1 of factor I, level 1 offactor II and level 1 of combination of factors I and II. In cases, where the combinationeffect of factors I and II is not significant, column 3 is ignored. After obtaining theapproval from the management, the experiments are conducted and the results arerecorded. Followed by this, the results are plotted on a graph for initial interpretationand further analysis is made possible by statistical tool namely ANOVA by whichsignificance of factors are ascertained. Further, calculation of “percentagecontribution” is used to verify whether any significant factors are left out. If so, thenthey are traced down and included in a new OA, and further experiments areconducted. If any one of the experiments has shown optimum results, then the

JMTM16,6

580

Figure 1.Implementation

methodology of Taguchi’soff-line quality control

model

Agilemanufacturing

environment

581

corresponding factors and their levels can be implemented in real production. Howeverthis will occur very rarely because OA considered initially, always a low resolution oneand there may be skepticism about the optimality of factors. To overcome this, a highresolution OA has to be considered and experiments have to be conducted according tothis OA. The results of analysis should be utilized to design a modified experiment,which should be conducted with the aim to attain optimality. This is known asconfirmation experiment whose results must further be interpreted and analyzed todetermine whether optimality has been achieved in practice. If it is felt that optimalityis yet to be achieved, then this process should be repeated. Soon after the optimality isachieved in an experiment, then the setup should be implemented in real situation.

A careful analysis would indicate that Taguchi’s off-line model is useful in twoways. First, the time taken in experimentation is greatly reduced. Second, the quality ofresults obtained are good since the model adopts the OAs for conducting experimentswhich represent the accurate sample of the entire domain of experimentation.Interestingly, these two benefits are found to be coinciding with the imperatives of AM.

Taguchi’s off-line model in AMAn overview on the literature indicates that all the research works involving Taguchi’soff-line models have been conducted by considering quantitative features (Antony,2002) such as cutting speed, feed, depth of cut, etc. In the case of an AM environment,experiments are required to be conducted not only in areas involving quantitativefeatures but also in subjective areas. In other words, in an AM environment, the qualityof not only the products but also of the system should be ascertained. This is due to thereason that, the AM system is not static and is dynamic in nature. Each time a newproduct or variant of existing product is to be manufactured according to customers’requirements, the quality of the system to be employed for the purpose is required to beplanned in advance.

A critical study indicates that Taguchi’s off-line model can also be exploited forensuring the quality of the system employed in an AM environment (Starkey et al.,1997). Whenever a new product is to be introduced experiments need to be designedbased on few or all criteria described in Table I. For example if organizational structureis considered as a factor, then semi-vertical structure and full vertical structure arechosen as its levels. An important observation in this regard is the difficulty ofconducting experimentation. Experimentation consisting of organizational structure as

Figure 2.Features of L4 orthogonalarray

JMTM16,6

582

a factor cannot be practically and feasibly conducted since installing two structuressimultaneously or in succession is not possible in practice. Delphi approach can beexploited to overcome this situation. Delphi approach calls for interviewing the expertsusing carefully designed questionnaires and deriving their unbiased opinions. Theresponses are analyzed to determine the optimum results (McCarthy and Atthirawong,2003). Hence, according to the Delphi approach, after designing experiments forenhancing agility by considering critical factors, the executives and employees may beinterviewed to evaluate the success rate of each experiment. In case, the opinionderived from different personnel varied widely, then an exposure program may beorganized and again interviews may be conducted to assess the success rate of thecorresponding experiment. In fact, this approach is a diluted form of brainstormingsessions as insisted by the researchers for conducting experiments based on Taguchi’soff-line method (Antony, 1999, 2001).

In the case of assessing quality of products to be introduced in the AM environment,real time experimentation may be conducted by employing Taguchi’s off-line model.However, even in this case, exploiting new technologies and methods would not be aprudent proposition and hence, real time experimentation is not encouraged. Thereason is that, installing experimental setup requires enormous time and money, bothare critical factors in the AM environment. Hence, in the case of optimizing quality ofproducts in the AM environment too, it is suggested that the Delphi approach may beemployed. Further, the Delphi approach for experimentation should be so structuredthat the analysis of results demand the use of simple statistical tools to avoid confusionand consumption of excessive time and money while identifying the optimal set-ups.On the whole, it is appraised that Taguchi’s off-line model would be a suitabletechnique to be employed in the AM environment, provided its scope should beenlarged from product quality to system quality aspects. Moreover, as narrated in thissection, the conduct of experiments in the AM environment should be modified to theconduct of interviews by adopting the Delphi approach. The exact methodology ofexploiting Taguchi’s off-line model in the AM environment can be understood byreferring to the case study presented in the next section.

Case studyIn order to testify the theoretical confabulations of this research project, interest wasevoked to carry out a case study. For this purpose, the status of AM in variousindustrial sectors of manufacturing was overviewed. Even though the manufacturingcommunity has just begun to hear much about agility, it is interesting to note thatsome industries have already started incorporating certain AM criteria. For example incomputer industry, new models are quite frequently introduced to meet the needs ofcustomers of different nature. However this trend is yet to infiltrate into other areas likethe field of mechanical engineering, where in spite of tremendous technologicaldevelopments (CAD, CAM, Robotics, etc.) having taken place, they are not exploiteddecisively to achieve agility. This has resulted in declined growth of companies,producing traditional products like pumps, compressors etc. One exception to thisstatement is the automobile industry in which new models are brought out in relativelylesser time during the recent times. In this context, it was felt necessary to apply thetheoretical developments to a mechanical engineering industry, which is not yetfamiliar with AM principles. Accordingly, a small-sized company involved in

Agilemanufacturing

environment

583

manufacturing pumps for domestic and agricultural purposes was identified. Thecompany is located in the city of Coimbatore, situated in the southern part of India.Coimbatore city is known for textile and mechanical engineering industries. Besidesfew electronic-based industries also have come up recently. Although a good industrialculture is prevailing in this city, the annual financial turnover of the industry as awhole is only moderate since the majority of the industries are small-sized and veryfew are medium-sized. As the turnover is not very good, these companies delay theirdecision to invest on the latest manufacturing technologies and philosophies. In spite ofthis discouraging situation, of late, few companies have started showing keen interestin adopting new manufacturing technologies and philosophies. The company chosenwas one among them.

The company has three plants and one of them was selected for our study. Thatplant has 75 employees including executives and produces three models of its productnamely submersible, jet, and centrifugal pumps. Out of these, the submersible pump isthe most important one because it contributes maximum profit to the company.Moreover, customer base is fast depleting for the other two types of pumps, whereas itis on increasing trend for the submersible pumps. Hence, it was decided that anysuggestion for improving its manufacturing would be highly useful to the company forits growth. Due to this reason, it was decided to restrict the study to the manufacturingof the submersible pumps. Even though description of the operational features of thesubmersible pump is beyond the scope of this paper, in order to have a clearperspective of the submersible models, the parts and specification of them arepresented in Table II.

In order to infuse agility using Taguchi’s off-line model, two phases of activitieswere required to be carried out. In the first phase, agile product design was considered.The most customer sensitive and important quality related features were given prioritywhile collecting data. These data were used to identify the factors, which would behelpful in imparting agility if a new model were to be designed. The initial analysisindicated that most of the factors are static in nature. This means, the company cannotchange values of these factors, as it is not within its control. For example the operatingvoltage cannot be varied since electric power is supplied at standardized voltage by thegovernment run Power Corporation. Hence, only the variables within the discretion ofthe design engineer were considered as factors for imparting agility in product design.Accordingly four factors, namely: impeller diameter, impeller thickness, speed, andimpeller material were chosen. Apart from this, three more customer sensitive factors,which are not incorporated in existing design, were considered. They were hardness ofimpeller shaft, incorporation of remote control, and provision for sump level controller.After identifying the factors, levels of each factor had to be determined. It was foundout that the existing levels were acceptable to the company since there was noconsiderable quality problem experienced in the existing design. Hence, it was decidedto have level 1 as the existing parameters. However this could be done only in the caseof existing factors and could not be done in the case of newly added factors. The factorsand levels for agile product experimentation are shown in Table III.

On considering the available facilities and time, it was decided to choose L8 OA(Ross, 1996). Seven factors were directly assigned against each column and Table IVwas developed by allowing the factors to take their levels. As mentioned earlier, in anAM environment, long lead-time cannot be tolerated. It was decided to look for other

JMTM16,6

584

Det

ails

Mod

el1

Mod

el2

Mod

el3

Mod

el4

Mod

el5

Mod

el6

Mod

el7

Mod

el8

Cod

eR

F1/

10R

F7/

6R

F4/

8R

F2/

12R

F1/

15R

F7/

9R

F4/

12R

F2/

16P

ower

inh

orse

pow

er(H

P)

0.5

11

11

1.5

1.5

1.5

Sin

gle

or3

ph

ase

11

11

11

11

Vol

tag

e(V

)24

024

024

0,24

024

024

024

024

024

0F

req

uen

cy(H

z)50

5050

5050

5050

50C

urr

ent

inam

per

es6

99

99

1212

12S

pee

din

rev

olu

tion

sp

erm

inu

te(r

pm

)2,

800

2,80

02,

800

2,80

02,

800

2,80

02,

800

2,80

0T

otal

hea

din

met

res

(m)

4827

3250

7240

4867

Dis

char

ge

m3/h

our

16.

63.

62

16.

63.

62

Imp

elle

rty

pe

Rad

ial

Rad

ial

Rad

ial

Rad

ial

Rad

ial

Rad

ial

Rad

ial

Rad

ial

Imp

elle

rm

ater

ial

Nor

yl

Nor

yl

Nor

yl

Nor

yl

Nor

yl

Nor

yl

Nor

yl

Nor

yl

No.

ofst

ages

106

812

159

1216

Effi

cien

cyin

per

cen

tag

e16

2826

2319

2927

20P

rice

inIn

dia

nru

pee

s(R

s)8,

600

10,0

0010

,000

10,0

0010

,000

12,0

0012

,000

12,0

00

Table II.Parts and specification

details of currentlymanufactured models of

submersible pumps

Agilemanufacturing

environment

585

alternatives in place of conducting practical experiments. Hence, initially, there was atendency to prefer computer-based animation and simulation packages. However itwas noted that many simulation packages were developed based on unrealisticassumptions and hence they fail to represent the reality. Their validity in practicalsituations is uncertain. Hence, the Delphi approach of interviewing experts forassessing the results of the experiments was adopted. Since the Delphi approach wasused, randomizing and repeating of experiments were not carried out. Little difficultywas experienced in adopting the Delphi approach, as it was found out that employeesof the company were not trained on latest developments. Even managerial levelemployees did not have sufficient opportunity to keep themselves abreast with thelatest developments. But when informed, the management showed keen interesttowards adopting them. An attempt was made to expose the details of AM to all the tenmanagerial level employees. Out of these, two executives assimilated and showedenthusiasm in associating with this project. Hence these two executives were requestedto respond according to the requirements of the Delphi approach. But, instead ofpreparing a questionnaire, due to its simplicity, the Table IV itself was shown to theexecutives to obtain the responses.

The opinions expressed by the two executives are presented in Table V. One of theauthors personally sat with the executives to derive their opinions and filled the lastthree columns of Table V. This personal appraisal was useful in preventing theexecutives from misinterpreting the details. As shown in Table V, the first experimentcorresponded to the existing manufacturing practice. This experiment was consideredas the basis and both the executives assessed the efficiency improvement or decrementas the result of proposed experiments. Quite interestingly, both anticipated theefficiency improvement in all of the remaining experiments. The difference inefficiency improvement, anticipated by both executives, ranged between only 5 percentand 10 percent. Hence the mean values of anticipated efficiencies proposed by bothexecutives against each experiment were used for subsequent analysis.

The conventional Taguchi’s approach envisages the use of statistical techniques foranalyzing the results of experiments. Since it is a tedious process, it was decided toanalyze the results using a simple two-axis graph to extract optimum parameters byeither extrapolation or interpolation. Accordingly, the graph shown in Figure 3 wasdeveloped. As shown, the efficiencies against each experiment were plotted.Efficiencies anticipated by the executives showed uniform increasing or decreasingtrends. The anticipated efficiencies indicated increasing trend from experiment 1 to 3.Between the experiments 3 and 6, it indicated uniformly intermittent increasing and

LevelsSl. no. Factors Level 1 Level 2

1 Impeller diameter in mm 75 802 Impeller thickness in mm 25 303 Speed in rpm 2,800 3,0004 Material of impeller Noryl Gunmetal5 Hardness in HRC 22 256 Remote control Wire control Cordless7 Sump level controller Wire Ccontrol Cordless

Table III.Factors and levels chosenfor agile productexperimentation

JMTM16,6

586

12

34

56

7F

acto

rs/

exp

t.n

o.Im

pel

ler

dia

.in

mm

Imp

elle

rth

ick

nes

sm

mS

pee

dR

PM

Imp

elle

rm

ater

ial

Har

dn

ess

(HR

C)

Rem

ote

con

trol

Su

mp

lev

elco

ntr

olle

rR

esu

lt(e

ffici

ency

)In

crea

sein

effi

cien

cyR

eact

ion

s

175

252,

800

Nor

yl

22W

ire

con

trol

Wir

eco

ntr

olM

oder

ate

0(r

efer

ence

)–

275

252,

800

Gu

nm

etal

25C

ord

less

Cor

dle

ssL

ow

375

303,

000

Nor

yl

22C

ord

less

Cor

dle

ssH

igh

Mor

ed

isch

arg

e4

7530

3,00

0G

un

met

al25

Wir

eco

ntr

olW

ire

con

trol

Mod

erat

eM

ore

dis

char

ge

580

253,

000

Nor

yl

25W

ire

con

trol

Cor

dle

ssH

igh

Mor

ed

isch

arg

e/h

igh

hea

d/h

igh

spee

d6

8025

3,00

0G

un

met

al22

Cor

dle

ssW

ire

con

trol

Mod

erat

e

780

302,

800

Nor

yl

25C

ord

less

Wir

eco

ntr

olV

ery

hig

hM

ore

dis

char

ge/

hig

hh

ead

880

302,

800

Gu

nm

etal

22W

ire

con

trol

Cor

dle

ssM

oder

ate

Table IV.Design of experiments for

product agility

Agilemanufacturing

environment

587

Fac

tors

/ex

pt.

no.

Imp

elle

rd

ia.

inm

m

Imp

elle

rth

ick

nes

sm

mS

pee

dR

PM

Imp

elle

rm

ater

ial

Har

dn

ess

HR

CR

emot

eco

ntr

olS

um

ple

vel

con

trol

ler

Res

ult

(effi

cien

cy)

Incr

ease

(þ)/

dec

reas

e(2

)in

effi

cien

cyR

eact

ion

s

175

252,

800

Nor

yl

22W

ire

con

trol

Wir

eco

ntr

olM

oder

ate

0(r

efer

ence

)–

275

252,

800

Gu

nm

etal

25C

ord

less

Cor

dle

ssL

owþ

1025

%

375

303,

000

Nor

yl

22C

ord

less

Cor

dle

ssH

igh

þ25

30%

mor

ed

isch

arg

e4

7530

3,00

0G

un

met

al25

Wir

eco

ntr

olW

ire

con

trol

Mod

erat

eþ

1520

%m

ore

dis

char

ge

580

253,

000

Nor

yl

25W

ire

con

trol

Cor

dle

ssH

igh

þ25

50%

mor

ed

isch

arg

e/h

igh

hea

d/h

igh

spee

d6

8025

3,00

0G

un

met

al22

Cor

dle

ssW

ire

con

trol

Mod

erat

eþ

1530

%

780

302,

800

Nor

yl

25C

ord

less

Wir

eco

ntr

olV

ery

hig

hþ

3060

%m

ore

dis

char

ge/

hig

hh

ead

880

302,

800

Gu

nm

etal

22W

ire

con

trol

Cor

dle

ssM

oder

ate

þ15

30%

Table V.Design of experiments forproduct agility(consolidated)

JMTM16,6

588

decreasing trends. From experiment number 6, it increased to a peak towardsexperiment number 7 and then showed decreasing trend towards the eighthexperiment. On the whole, it revealed that experiment number 7 would offer maximumefficiency. However there was a chance that if an experiment is designed byincorporating intermediate levels of non-sensitive factors of the seventh and eighthexperiments, it might still yield better efficiency. During the interviews, the executivesindicated the factors, which are very sensitive from manufacturing point of view.According to them, the levels of these factors cannot be so easily altered. Hence it wasdecided to conduct one confirmation experiment (that is, by interviewing the sameexecutives again) by incorporating revised and modified levels of the non-sensitivefactors. The details of this experiment are shown in Table VI. Both executives wereinterviewed by informing about the details of the ninth experiment. The meananticipated efficiency increase was 35 percent, which is the highest of all theexperiments conducted so far. This confirmed with the graphical extrapolation thatwas carried out theoretically. Moreover the executives anticipated that onincorporation of the factors and levels of confirmation experiment, there will be anincrease of 75 percent increase in discharge volume and delivery head in comparison tothe present factors and levels adopted. Hence, it was inferred that agile product designhas been achieved.

Agile process designThe first phase of the experimentation revealed the possibilities of imparting agility inproducts by refining the existing design by conducting limited number of experimentsusing the Delphi approach. As mentioned in the previous section, it was anticipatedthat adoption of a suitable OA for the purpose of interviewing executives wouldprovide an optimum solution for imparting system agility. However the systembecomes agile only if the agility criteria are implemented in a feasible form. In order tocarry out this task, the second phase of experimentation became a necessity. Hence,theoretically drawn 20 agility criteria (shown in Table I) were examined for thepossibilities of adoption in the company. Out of all, it was found that 11 criteria werefound to be critical and feasible for adoption in the company. The fundamental reasonbehind in selecting these criteria was that, the company has not yet adopted thesecriteria, but it was ready to implement them in stages. However implementation instages would consume considerable amount of time. Hence, it was discernible that a

Figure 3.Graphical analysis of theresults of experiments on

agile product design

Agilemanufacturing

environment

589

12

34

56

7

Fac

tors

exp

t.n

o.

Imp

elle

rd

ia.

inm

m

Imp

elle

rth

ick

nes

sm

mS

pee

dR

PM

Imp

elle

rm

ater

ial

Har

dn

ess

(HR

C)

Rem

ote

con

trol

Su

mp

lev

elco

ntr

olle

rR

esu

lt(e

ffici

ency

)In

crea

sein

effi

cien

cyR

eact

ion

s(l

evel

ofac

cep

tan

ce)

780

302,

800

Nor

yl

25C

ord

less

Wir

eco

ntr

olV

ery

hig

hþ

3060

%m

ore

dis

char

ge/

hig

hh

ead

880

302,

800

Gu

nm

etal

22W

ire

con

trol

Cor

dle

ssM

oder

ate

þ15

30%

9C

onfi

rmat

ion

exp

erim

ent

8030

2,80

0N

ory

l23

Sem

i-p

ower

con

trol

Sem

i-p

ower

con

trol

Bes

tþ

3575

%m

ore

dis

char

ge/

hig

hh

ead

Table VI.Design of experiments forproduct agilityconfirmationexperimentation

JMTM16,6

590

most feasible combination of traditional and agility criteria would serve the immediatepurpose of attaining agility in the system. This will enhance the implementationfeasibility of agile product design achieved during the first stage. The 11 criteria wereconsidered as factors and subsequently their levels were decided. The levels werechosen in such a way that traditional approach was considered as level 1 and agileapproach as level 2 and were denoted by the symbol followed by numerals 1 or 2 as thecase may be. L12 OA was chosen and accordingly. Table VII was formed by assigningthe factors and their levels.

Unlike in the case of agile product design, Table VII was not shown to theexecutives since it was realized that it would be difficult for the executives to imaginethe holistic view of each experiment. In order to overcome this, a questionnairecontaining 12 questions was prepared for each experiment. Both executives were giventhe questionnaire pertaining to experiment 1. The executives were guided whileresponding to the questions. The remaining experiments were conducted byinterviewing the executives by modifying the questionnaire suitability. But the overallformat was not changed. In addition to the questionnaire, each executive was given aresponse sheet and requested to encircle against each question to denote the successrate. Due to lack of space, the questionnaire and the response sheet are not included inthis paper. The encircled numbers were multiplied by a factor 10 to convert them into apercentage of success rate. The mean percentage success rates against all questionswere identified as agility success rate corresponding to that experiment. Apart fromthis, the executives were requested to assess their acceptance rating of eachexperiment. In some cases, it was noted that, in spite of higher efficiencies, executivesexpressed different rates of acceptance for implementing the factors and levels for theexperiments in practice due to various reasons. The efficiencies and acceptance ratingsalong with the mean values are shown in Table VIII.

The results of agile system design experiments are shown in Figure 4 as atwo-dimensional graph. Table VII reveals that experiments 2, 8, and 12 result inmaximum possible efficiencies in obtaining system agility. Hence, the levels ofdifferent factors in these three experiments were analyzed further and the levels withmaximum success efficiency rating were extracted to create a level for confirmationexperiment. The details are presented in Table IX. In this, experiment 13 is theconfirmation experiment. The assignments of levels are described here for factor 1. Inthe case of factor 1, the level 2 was given a high rating by the executives. Hence, in theconfirmation experiment level 2 was assigned under factor 1. Likewise all levels wereallotted and a corresponding questionnaire was prepared. Again each executive wasinterviewed and the final test results were obtained. These results are shown inTable X. The mean agility efficiency of 67.6 percent was anticipated by the executives.

Followed by this, acceptance level was also revealed by them whose mean was 58percent. As these two values were highest of all other experiments, it was inferred thatif the company implements the agility criteria with the levels shown in confirmationexperiment (number 13), then the journey for attaining agility would be smooth andquick.

Concluding remarksAM is also viewed as a set of bridges which are meant for integrating technologies andmanagement philosophies (Sarkis, 2001). Most of the other methods and philosophies

Agilemanufacturing

environment

591

Fac

tors

12

34

56

78

910

11

Ex

pt.

no.

Org

ani-

zati

onst

ruct

ure

Man

ufa

c-tu

rin

gse

t-u

pS

tatu

sof

qu

alit

y

Em

plo

yee

inv

olv

e-m

ent

Cu

stom

erre

spon

se

Des

ign

imp

rov

e-m

ent

Man

ufa

c-tu

rin

gp

lan

nin

g

Su

pp

lych

ain

man

agem

ent

Em

plo

yee

stat

us

Au

tom

atio

nty

pe

ITin

teg

rati

on

1O

1M

1S

1E

1C

1D

1M

P1

SC

M1

ES

1A

1IT

12

O1

M1

S1

E1

C1

D2

MP

2S

CM

2E

S2

A2

IT2

3O

1M

1S

2E

2C

2D

1M

P1

SC

M1

ES

2A

2IT

24

O1

M2

S1

E2

C2

D1

MP

2S

CM

2E

S1

A1

IT2

5O

1M

2S

2E

1C

2D

2M

P1

SC

M2

ES

1A

2IT

16

O1

M2

S2

E2

C1

D2

MP

2S

CM

1E

S2

A1

IT1

7O

2M

1S

2E

2C

1D

1M

P2

SC

M2

ES

1A

2IT

18

O2

M1

S2

E1

C2

D2

MP

2S

CM

1E

S1

A1

IT2

9O

2M

1S

1E

2C

2D

2M

P1

SC

M2

ES

2A

1IT

110

O2

M2

S2

E1

C1

D1

MP

1S

CM

2E

S2

A1

IT2

11O

2M

2S

1E

2C

2D

2M

P1

SC

M1

ES

1A

2IT

212

O2

M2

S1

E1

C1

D1

MP

2S

CM

1E

S2

A2

IT1

Table VII.Design of experiments forsystem agility

JMTM16,6

592

emerged during recent years emphasize either partial or no integration of technologyand management. As the trend of AM spreads very fast and has already captured quitea number of manufacturing companies and research laboratories, it is high time thatthe conventional techniques applied in traditional manufacturing environments arerequired to be examined for their compatibility in AM environments. In this context,the piece of research work reported in this paper was taken up by consideringTaguchi’s off line quality control method in AM environment. Since the authors areresiding in a city with companies wherein the recent research findings take a long timeto penetrate, it was not possible to identify an AM company. Hence a manufacturingcompany, which is following traditional approach, was chosen. Fortunately the

Experiment

Executive1 SR

%

Executive2 SR

%

SRaverage

%

Executive1 AL

%

Executive2 AL

%

ALaverage

% Remarks

1 59.00 55.00 57 50 40 45

2 60.00 59.00 60 60 70 65 High value

3 56.40 52.70 54 60 60 60

4 56.40 54.50 55 50 55 53

5 58.20 55.50 57 60 50 55

6 57.30 52.70 55 40 50 45

7 56.36 52.70 54 40 50 45

8 60.90 55.50 58 50 40 45

9 58.20 55.50 57 50 50 50

10 55.50 51.80 54 60 50 55

11 57.20 57.30 57 50 50 50

12 59.10 56.40 58 60 50 55

Note: SR ¼ success rate; AL ¼ acceptance level

Table VIII.Percentage of efficiencies

and acceptance values

Figure 4.Graphical analysis of theresults of experiments on

system

Agilemanufacturing

environment

593

Fac

tors

12

34

56

78

910

11

Ex

pt.

no.

Org

ani-

zati

onst

ruct

ure

Man

ufa

c-tu

rin

gse

t-u

pS

tatu

sof

qu

alit

yE

mp

loy

eein

vol

vem

ent

Cu

stom

erre

spon

seD

esig

nim

pro

vem

ent

Man

ufa

c-tu

rin

gp

lan

nin

gS

up

ply

chai

nm

anag

emen

tE

mp

loy

eest

atu

sA

uto

mat

ion

typ

eIT

inte

gra

tion

2O

1M

1S

1E

1C

1D

2M

P2

SC

M2

ES

2A

2IT

28

O2

M1

S2

E1

C2

D2

MP

2S

CM

1E

S1

A1

IT2

12O

2M

2S

1E

1C

2D

1M

P2

SC

M1

ES

2A

2IT

113

O2

M1

S1

E1

C2

D2

MP

2S

CM

2E

S1

A2

IT2

Table IX.Confirmation experimentfor organizational agility

JMTM16,6

594

management showed keen interest on AM and the work enjoyed full support of bothmanagement and employees. In spite of this positive factor, the results evolved out ofthis research project are yet to be implemented, as the management is required to makea major decision in this direction. In future it is proposed to insist the management toimplement the results of this project, which will be monitored continuously, andchanges or modifications will be suitably carried out in the levels of factors. Furtherdue to shortage of time, Taguchi’s online quality control methods were not examinedduring this study which is left for future work. On the whole, this work creates animpression that Taguchi methods with suitable refinements will be useful in impartingagility in traditional organizations.

References

Antony, J. (1999), “Ten useful and practical tips for making your industrial experimentssuccessful”, The TQM Magazine, Vol. 11 No. 4, pp. 252-6.

Antony, J. (2001), “Improving the manufacturing process quality using design of experiments:a case study”, International Journal of Operations & Production Management, Vol. 21Nos 5/6, pp. 812-22.

Antony, J. (2002), “Robust design in new product development process: a neglected methodologyin manufacturing organisations”, Work Study, Vol. 51 No. 2, pp. 81-4.

Antony, J. and Antony, F.J. (2001), “Teaching the Taguchi method to industrial engineers”, WorkStudy, Vol. 50 No. 4, pp. 141-9.

Aravindan, P., Devadasan, S.R., Dharmendra, B.V. and Selladurai, V. (1995), “Continuous qualityimprovement through Taguchi’s online quality control methods”, International Journal ofOperations & Production Management, Vol. 15 No. 7, pp. 60-77.

Bajgoric, N. (2001), “Web-based information access for agile management”, International Journalof Agile Management Systems, Vol. 2 No. 2, pp. 121-9.

Bendell, T. (1988), “Taguchi methods”, Proceedings of 1988 European Conference, ElsevierApplied Science, Amsterdam.

Chowdiah, M.P. (1996), “Agile manufacturing for global competitiveness”, Souvenir ofInternational Conference on Agile Manufacturing (ICAM 96), Bangalore, pp. 37-46.

Crocitto, M. and Youssef, M. (2003), “The human side of organizational agility”, IndustrialManagement & Data Systems, Vol. 103 No. 6, pp. 388-97.

Ebrahimpur, G. and Jacob, M. (2001), “Restructuring for agility at Volvo Car Technical Service(VCTS)”, European Journal of Innovation Management, Vol. 4 No. 2, pp. 64-72.

Experiment

Executive1 SR

%

Executive2 SR

%

SRaverage

%Executive1 AL %

Executive2 AL

%

ALaverage

% Remarks

2 60 59 60 60 70 65

8 60.9 55.5 58 50 40 45

12 59.1 56.4 58 60 50 55

13 67.2 68.1 67.6 60 55 58 Highest SRTable X.

Final test results

Agilemanufacturing

environment

595

Goldman, S.L., Nagel, R.N. and Preiss, K. (1995), Agile Competitors and Virtual Organizations –Strategies for Enriching the Customer, Van Nostrand Reinhold, New York, NY.

Hooper, M.J., Steeple, D. and Winters, C.N. (2001), “Costing customer value: an approach for theagile enterprise”, International Journal of Operations & Production Management, Vol. 2No. 56, pp. 630-44.

Hormozi, A.M. (2001), “Agile manufacturing: the next logical step”, Benchmarking:An International Journal, Vol. 8 No. 2, pp. 132-43.

Lee, G., Bennett, D. and Oakes, I. (2000), “Technological and organizational change in small tomedium-sized manufacturing companies”, International Journal of Operations &Production Management, Vol. 20 No. 5, pp. 549-72.

McCarthy, B.L. and Atthirawong, W. (2003), “Factors affecting location decisions in internationaloperations: a Delphi study”, International Journal of Operations & ProductionManagement, Vol. 23 No. 7, pp. 794-818.

Maskell, B. (2001), “The age of agile manufacturing”, Supply Chain Management:An International Journal, Vol. 6 No. 1, pp. 5-11.

Meredith, S. and Francis, D. (2000), “Journey towards agility: the agile wheel explored”,The TQM Magazine, Vol. 12 No. 2, pp. 137-43.

Montgomery, J.C. and Levine, L.O. (1996), The Transition to Agile Manufacturing: StayingFlexible for Competitive Advantage, ASQC Quality Press, Milwaukee, WI.

Murugesh, R., Devadasan, S.R., Aravindan, P. and Natarajan, R. (1997), “An exploration on theadoption and modeling of strategic productivity management approach in manufacturingsystems”, International Journal of Operations & Production Management, Vol. 17 No. 3,pp. 239-55.

Narain, R., Yadav, R.C., Sarkis, J. and Cordeiro, J.J. (2000), “The strategic implications offlexibility in manufacturing systems”, International Journal of Agile ManagementSystems, Vol. 2 No. 3, pp. 202-13.

Onuh, S.O. and Hon, K.K.B. (2001), “Integration of rapid prototyping technology into FMS foragile manufacturing”, Integrated Manufacturing Systems, Vol. 12 No. 3, pp. 179-86.

Parkinson, S. (1999), “Agile manufacturing”, Work Study, Vol. 48 No. 4, pp. 134-7.

Pawar, K.S. and Shaiffi, S. (2002), “Managing the product design process: exchanging knowledgeand experience”, Integrated Manufacturing Systems, Vol. 13 No. 2, pp. 91-6.

Power, D.J., Sohal, A.S. and Rahman, S.U. (2001), “Critical success factors in agile supply chainmanagement – an empirical study”, International Journal of Physical Distribution &Logistics Management, Vol. 31 No. 4, pp. 247-65.

Ramasesh, R., Kulkarni, S. and Jayakumar, M. (2001), “Agility in manufacturing systems:an exploratory modeling framework and simulation”, Integrated Manufacturing Systems,Vol. 12 No. 7, pp. 534-48.

Rooks, B. (2000), “Winning ways for manufacturing”, Assembly Automation, Vol. 20 No. 1,pp. 35-9.

Ross, J.P. (1996), Taguchi Techniques for Quality Engineering, McGraw-Hill Book Company,New York, NY.

Sarkis, J. (2001), “Benchmarking for agility”, Benchmarking: An International Journal, Vol. 8No. 8, pp. 88-107.

Sharifi, H. and Zhang, Z. (2001), “Agile manufacturing in practice application of a methodology”,International Journal of Operations & Production Management, Vol. 21 Nos 5/6, pp. 772-94.

JMTM16,6

596

Spring, M. and Dalrymple, J.F. (2000), “Product customization and manufacturing strategy”,International Journal of Operations & Productions Management, Vol. 20 No. 4, pp. 441-67.

Starkey, M., Aughton, J. and Brewin, R. (1997), “Extending process thinking: design ofexperiments in sales and marketing”, The TQM Magazine, Vol. 9 No. 6, pp. 434-9.

Struebing, L. (1995), “New approach to agile manufacturing”, Quality Progress USA, Vol. 28No. 12, pp. 18-19.

Taguchi, G. (1989), Quality Engineering in Production Systems, McGrawHill Book Company,New York, NY.

van Assen, M.F. (2000), “Agile-based competence management: the relation between agilemanufacturing and time-based competence management”, International Journal of AgileManagement Systems, Vol. 2 No. 2, pp. 142-55.

van Assen, M.F., Hans, E.W. and van de Valde, S.L. (2000), “An agile planning and controlframework for customer-order-driven discrete parts manufacturing environment”,International Journal of Agile Management Systems, Vol. 2 No. 1, pp. 16-23.

Agilemanufacturing

environment

597