Using quality function deployment for collaborative product
design and optimal selection of module mix
Chih-Hsuan Wang
⇑, Jiun-Nan Chen
Department of Industrial Engineering & Management, National Chiao Tung University, Taiwan
a r t i c l e
i n f o
Article history: Received 12 July 2011
Received in revised form 1 May 2012 Accepted 29 June 2012
Available online 16 July 2012 Keywords:
QFD NPD Fuzzy Delphi Fuzzy DEMATEL
Linear integer programming
a b s t r a c t
In response to fast-growing and rapidly-changing markets, launching new products faster than compet-itors cannot only assist firms in acquiring larger market share but also reducing development lead time, significantly. However, owing to its intrinsically uncertain properties of managing NPD (new product development), manufacturing companies often struggle with the dilemma of increasing product variety or controlling manufacturing complexity. In this study, a fuzzy MCDM (multi-criteria decision making) based QFD (quality function deployment) which integrates fuzzy Delphi, fuzzy DEMATEL (decision mak-ing trial and evaluation laboratory), with LIP (linear integer programmmak-ing) is proposed to assist an enter-prise in fulfilling collaborative product design and optimal selection of module mix when aiming at multi-segments. In particular, Fuzzy Delphi is adopted to gather marketing information from invited cus-tomers and their assessments of marketing requirements are pooled to reach a consensus; fuzzy DEMA-TEL is utilized to derive the priorities of technical attributes in a market-oriented manner; and LIP is used to maximize product capability with consideration of supplier’s budget constraints of manufacturing resources. Furthermore, a real case study on developing various types of sport and water digital cameras is demonstrated to validate the proposed approach.
Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction
In an era of customer-oriented global economy, dominating the majority market with a single product line becomes very challeng-ing and almost infeasible for most companies (Hsiao & Liu, 2005). Traditionally, to satisfy market majorities, companies considered providing products with high quality, low cost, fast delivery and courteous after-sales service at most. Nowadays, owing to fiercely competitive environments and rapidly changing demand, the capa-bility and the speed of developing niche products and launching them into the niche segments gradually dominate the competition paradigm, particularly when a transition has been shifting from ‘‘supply push’’ to ‘‘demand pull’’ (Jiao, Ma, & Tseng, 2003). To put it another way, ‘‘mass customization’’ embarks a new paradigm for modern manufacturing industries since it treats each customer as an individual and attempts to provide ‘‘tailor-made’’ featured products that was only offered in the pre-industrial ‘‘craft’’ era.
Over the past two decades, numerous publications originated from different disciplines have witnessed in the field of customer requirement management (Jiao & Chen, 2006). For example, various
fields such as marketing research, consumer behavior, collaborative design, and concurrent engineering, attempt to contribute to different stages for new product development (NPD). Among them, marketing research and consumer behavior emphasize the front issues relevant to collecting the information of customer preference via specific channels. In contrast, collaborative design and concur-rent engineering focus on utilizing a systematic and parallel approach for integrating a wide spectrum of product design and re-lated manufacturing processes (Lin, Wang, Chen, & Chang, 2008). Although high product variety does stimulate product sales, compa-nies still inevitably face the trade-offs between the diversity of cus-tomer needs and numerous adverse effects, such as larger inventory cost, longer cycle time and expensive research investment.
As a result, it is very imperative for companies to keep high flex-ibility while incurring limited manufacturing cost, concurrently. In practice, two common techniques have been proposed to tackle the above-mentioned issue, including product family architecture (Jiao & Tseng, 1999; Moon, Simpson, & Kumara, 2010) and modular product or product family design (Hsiao & Liu, 2005; Kreng & Lee, 2004). Modular product design offers a feasible way by devel-oping a product architecture, in which physical relationships across modules are limited while functional relationships among components within a module are coherent. Furthermore, product family design based on a standard platform usually provides a cost-effective way to develop highly related but differentiated
0360-8352/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.cie.2012.06.014
⇑Corresponding author. Address: 1001 University Road, Hsinchu 30013, Taiwan. Tel.: +886 3 5712121; fax: +886 3 5722392.
E-mail addresses: chihwang@mail.nctu.edu.tw, chihswang@gmail.com (C.-H. Wang).
Contents lists available atSciVerse ScienceDirect
Computers & Industrial Engineering
products. By sharing/reusing physical manufacturing resources and intangible human capitals, companies can efficiently balance the benefit and cost for NPD.
Based on previous studies, most of them are deficient in con-structing a systematic approach to assist companies in achieving mass customization while keeping reasonable manufacturing cost. In this study, a fuzzy MCDM based QFD (quality function deploy-ment) is proposed to fulfill collaborative product design and opti-mal selection of module mixes when aiming at multi-segments. Moreover, this paper contributes to this domain by presenting the following merits:
QFD provides a communication platform to gather different opinions between industrial experts and even among customer individuals.
QFD is capable to transform intangible marketing requirements (MRs) into measurable technical attributes (TAs) and to accom-modate the dependences between MRs and TAs and the corre-lations among themselves.
In additional to deriving the weights of MRs and TAs, the pro-posed fuzzy MCDM based QFD could further identify the opti-mal module mix (product variety) for a specific market segment.
The remaining of this paper is organized as follows. Section 2 overviews the related works and Section 3 introduces the proposed framework which integrates fuzzy Delphi, fuzzy DEMATEL with LIP (linear integer programming). A real example regarding collabora-tive design for various sport & water digital cameras is illustrated in Section 4. Conclusions are drawn in Section 5.
2. Related works
Quality function deployment (Akao, 1990) originated in Japan in the 1970s has been widely applied to various industries for prod-uct development, concept evaluation, service design, and compet-itor benchmarking. Basically, customers’ desires on a specific product or service can be represented by a set of intangible mar-keting requirements (MRs). Thereafter, a series of technical attri-butes (TAs) that impact on MRs need to be determined and realized for product development or service design. Typically, the
conventional QFD consists of the following four phases (Chan,
Kao, Ng, & Wu, 1999; Lin, Cheng, Tseng, & Tsai, 2010): phase one translates marketing requirements into technical attributes; phase two translates technical attributes into part characteristics; phase three translates part characteristics into manufacturing operation, and phase four translates manufacturing operations into produc-tion requirements. Specifically, at phase one of QFD, the so-called HoQ (house of quality) provides a communication platform to fuse diverse opinions among cross-functional team members (see Fig. 1).
To fast understand the research trend regarding QFD, represen-tative publications are reviewed and listed below. First, to deter-mine the importance degrees of MRs, AHP (analytical hierarchy process)/fuzzy AHP (Kwong & Bai, 2002, 2003), fuzzy Delphi (Chen & Ko, 2008; Karsak, 2004), and fuzzy group decision (Büyüközkan, Feyziog˘lu, & Ruan, 2007; Sein, Ho, Lai, & Chang, 1999) have been suggested, respectively. Second, to improve the weakness of AHP/ fuzzy AHP, numerous papers adopt ANP (analytical network pro-cess)/fuzzy ANP to consider the dependences between MRs and
TAs and the correlations among themselves, such asKarsak, Sozer,
and Alptekin (2002), Büyüközkan, Ertay, Kahraman, and Ruan (2004), Kahraman, Ertay, and Büyüközkan (2006), Lin et al. (2010), andLee, Kang, Yang, and Lin (2010). Recently, various opti-mization schemes with consideration of budget cost or resource
constraints have been incorporated into the QFD. For example, zero-one goal programming or fuzzy goal programming is formu-lated to determine the level or a mix of design requirements (Chen & Weng, 2006; Karsak, 2004; Karsak et al., 2002). A two-phase QFD which combines ANP/fuzzy ANP with goal programming is utilized to determine the optimal varieties of product attributes for distinct market segments (Lee et al., 2010; Liu & Hsiao, 2006; Park, Shin, In-sun, & Hyemi, 2008).
After reviewing the above-mentioned studies, several critical shortcomings are found and listed below:
A systematic approach to efficiently identify the causal impacts of MRs on TAs and the correlations among themselves is imper-ative, yet, rarely addressed and incorporated into the entire decision-making process.
AHP/fuzzy AHP (Saaty, 1980) are capable to determine the
weights of ‘‘independent’’ criteria, but they are limited to han-dle a scenario in which the interdependences exist among crite-ria or the number of critecrite-ria is over a reasonable threshold.
ANP/fuzzy ANP (Saaty, 1996) are commonly adopted to
accom-modate the complicated interdependences among criteria, but they might be infeasible in processing a scenario in which numerous criteria appear on a hierarchy.
Suppose that there are n mutually interdependent criteria (associated with an n-order matrix), to completely describe their interrelationships among all criteria, we might need to conduct up to n2ðn 1Þ=2ðn Cn
2Þ times of pair-comparisons for obtaining
Technical assessment among competitors
Marketing assessment
among competitors
Interrelationship (dependence) Matrix
between MRs and TAs
Marketing Requirements ( MR s) Correlations among TAs
Technical Attributes (TAs)
Correlations among
MR
s
Fig. 1. A general framework for the conventional HoQ (house of quality).
Table 1
An overall comparison among AHP, ANP, DEMATEL and proposed method. AHP ANP DEMATEL Proposed Handling an independent
hierarchy structure
Yes Yes Yes Yes
Handling an interdependent network structure
No Yes Yes Yes
Conducting pairwise comparisons among criteria
Limited Tedious Not necessary
Not necessary Deriving the importance
weights of criteria
Yes Yes No Yes
Handling numerous criteria within the same decision level
Limited Limited Yes Yes
Identifying causal relationships among criteria
the importance weights of all elements. Obviously, when n be-comes larger than a reasonable threshold, its running time of ANP might be quite tedious and result in a serious problem of effi-ciency or consistence for an evaluator. To overcome the difficulties incurred by the ANP, fuzzy Delphi is combined with fuzzy DEMA-TEL to construct a fuzzy MCDM based QFD. For convenience, an overall comparison among the aforementioned techniques is de-scribed inTable 1.
3. The proposed techniques
As indicated byFig. 2, a hybrid research framework consisting
of fuzzy Delphi, fuzzy DEMATEL, and LIP (linear integer program-ming) is presented to realize two critical issues: collaborative product design and optimal selection of module mixes. In particu-lar, the concept of fuzzy set theory is incorporated into the conven-tional QFD to accommodate linguistic properties of human judgment (seeTable 2). For clarity, their details are simply outlined as follows:
Fuzzy Delphi is used to gather customers’ assessments of initial weights of MRs and to fuse experts’ opinions on the initial val-ues of dependences and correlations.
Fuzzy DEMATEL is utilized to incorporate the dependences between MRs and TAs and the correlations among themselves and the priorities of TAs could be further derived.
With consideration of cost constraints for multi-segments, LIP is employed to determine the optimal combinations of module mix for the purpose of target marketing.
Based on previous studies (Chen & Weng, 2003; Delice &
Güngör, 2009; Wasserman, 1993), the final weights of MRs and TAs are derived as follows:
WtMRi¼ Wt0MRiþ 1 m 1 Xm i–k kik Wt0MRk; ð1Þ R0 ij¼ Pn k¼1Rik
c
kj Pn j¼1 Pn k¼1Rikc
kj ; ð2Þ WtTAj¼X m i¼1 WtMRi R0ij; ð3Þwhere WtMRi/WtTAjrepresent the weight of MRi and TAj, respec-tively. Meanwhile, we assume that m marketing requirements and n technical attributes exist in the QFD, R0
ijis the normalized depen-dences between MRiand TAj, and kik=
c
kj denote the correlationsamong MRs/TAs, respectively. In view of Eqs.(1)–(3), the
depen-dences between MRs and TAs and the correlations among them-selves have been appropriately incorporated to generate their importance degrees.
3.1. Use of fuzzy Delphi method to determine the initial weights of MRs Delphi method has been widely applied to various industries as a group-decision based forecasting technique. It normally requires a group of partially or completely anonymous experts responding their opinions via the preset questionnaires and involves several rounds of iterations to reach a consensus. In the first round, all ex-perts respond to the questionnaire and the results are evaluated and returned to experts through a feedback process. In reality, Del-phi method often suffers from low convergence among experts, high execution cost and tedious operating process. Besides, be-cause linguistic human judgments are usually imprecise, evalua-tion terms expressed in fuzzy sense might be more feasible in practice. Instead of using crisp values in the conventional Delphi, Murry, Pipino, and Gigch (1985)proposed a hybrid scheme which intends to incorporate the concept of fuzzy set into the
conven-tional Delphi to overcome the above-mentioned flaws.Cheng and
Lin (2002)present fuzzy Delphi method to achieve a consensus of experts’ opinions which are denoted by fuzzy numbers.
Fol-lowed byKarsak (2004), fuzzy Delphi method is slightly modified
to generate the weights of MRs. The whole process is operated as follows:
The invited evaluators, who act as domain experts, are respon-sible to rate the importance degrees of marketing requirements. In particular, the rating scale is measured in terms of a triangu-lar fuzzy number as:
~
wij¼ ðwija;wijb;wijcÞ; 1 6 i 6 m; 1 6 j 6 p; ð4Þ
where m denotes the number of MRs, p represents the number of invited evaluators, and ~wijis the importance rating of MRiassigned by evaluator j.
Aggregate the importance rating to attain its mean value of MRi:
~ wim¼1 p Xp j¼1 wija;X p j¼1 wijb;X p j¼1 wijc !
¼ ðwima;wimb;wimcÞ; ð5Þ
Then, the differences between ~wijand ~wimare calculated and sent back to the evaluators for reconsideration.
For later rounds, all evaluators are required to revise their fuzzy rating and the process is repeated in a similar way until the gaps between successive mean values are reasonably converged. To calculate the distance between two fuzzy numbers, the follow-ing is adopted (Geng, Chu, Xue, & Zhang, 2010):
dð ~wt im; ~wtþ1imÞ ¼ 1ffiffiffi 3 p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðwt ima wtþ1imaÞ 2 þ ðwt imb wtþ1imbÞ 2 þ ðwt imc wtþ1imcÞ 2 q ; ð6Þ where ~wt
im= ~wtþ1im represent fuzzy mean value at iteration t/t + 1.
LIP Fuzzy DEMATEL
Identification of marketing requirements (MRs) and technical attributes (TAs)
Generating the initial weights of MRs and their interdependences and correlations
Deriving the final weights of MRs and TAs
Optimizing product varieties with respect to distinct market segments
Fuzzy Delphi
Fig. 2. The proposed research framework.
Table 2
Linguistic rating scale used in fuzzy MCDM methods.
Triangular fuzzy number Delphi/DEMATEL ~ 1 (1, 1, 3) VL (very low) ~ 3 (1, 3, 5) L (low) ~ 5 (3, 5, 7) M (medium) ~ 7 (5, 7, 9) H (high) ~ 9 (7, 9, 9) VH (very high)
Based on ‘‘the center of area’’ approach (Yager, 1978), the pro-cess of defuzzification can be applied to the fuzzy rating of each marketing requirement and its defuzzified crisp value is repre-sented as:
wim¼wimaþ wimbþ wimc
3 ; ð7Þ
Generating the initial weight for each marketing requirement via the normalization process:
Wt0 MRi¼ wim Pm i¼1wim ; ð8Þ
where Wt0MRidenotes the initial weight of MRi.
3.2. Use of fuzzy DEMATEL to derive the final weights of MRs and TAs DEMATEL (decision making trial and evaluation laboratory), developed by the science and human affairs program of the Battelle Memorial Institute of Geneva Research Centre (Fontela & Garbus, 1976), is able to visualize the complex interdependent relationship among all evaluation criteria through converting the causal rela-tionship into a visible structure of the whole system. Suppose that p experts are invited to assess m marketing requirements (MRs)
and n technical attributes (TAs). Followed byLin and Wu (2008),
fuzzy DEMATEL is applied to our problem and its details are de-scribed as follows:
Assigning a fuzzy rating scale to measure the direct-relation matrix:
As seen inFig. 3, a (m + n) (m + n) fuzzy matrix eX with an ele-ment of ~xk
ij¼ ðl k
ij;mkij;ukijÞ is evaluated by expert k, which represents the impact of MRion TAjand all the diagonal elements of matrix eX will be set as zero ð~xk
ii¼ ð0; 0; 0ÞÞ. By averaging all experts’ scores, the direct-relation matrix eA can be characterized with an element of ~aij: ~ aij¼1 S XS k¼1 ~ xk
ij¼ ðalij;amij;auijÞ; ð9Þ
Normalizing the direct-relation matrix:
The normalized matrix eB can be obtained by normalizing the
matrix eA: ~
bij¼1
X
~aij¼ ðblij;bmij;buijÞ; where ð10ÞX
¼ Max max 16i6n Xn j¼1 uij;max 16j6n Xn i¼1 uij ! ; ð11ÞDeriving the total-relation matrix:
Once the normalized matrix eB has been obtained, the total-rela-tion matrix eT can be derived based on Eqs.(12)–(15):
eT ¼ eB þ eB2þ eB3þ ¼ eBðI eBÞ1
; ð12Þ
where ~tij¼ ðtlij;tmij;tuijÞ and the amount of three matrix elements are list below:
matrix½tlij ¼ BlðI BlÞ1
; ð13Þ
matrix½tmij ¼ BmðI BmÞ1;
ð14Þ
matrix½tuij ¼ BuðI BuÞ1;
ð15Þ where I denotes an identity matrix and Bl/Bm/Burepresents the crisp matrix composed of lower/medium/upper values of the normalized matrix, respectively.
Defuzzifying the total-relation matrix eT (Yager, 1978) and com-puting a causal diagram through the dispatcher group D and the receiver group R, where D is the sum of rows in crisp matrix T and R is the sum of columns:
Tij¼tlijþ tmijþ tuij
3 ; ð16Þ Di¼X n j¼1 Tij; ð17Þ Rj¼X n i¼1 Tij: ð18Þ
After a crisp matrix T is obtained via Eq.(16), the dependences
be-tween MRs and TAs ðR0ijÞ and the correlations among themselves
ðkik=
c
kjÞ will be automatically extracted from the matrix T for deriv-ing the weights of TAs (see Eqs.(1)–(3)). A causal diagram can be visualized by displaying (D + R, D R): the horizontal axis ‘‘D + R’’ named ‘‘prominence’’ reveals how much important the criterion is and the vertical axis ‘‘D R’’ named ‘‘influence’’ categorizes the cri-terion into either the cause group or the effect group.3.3. Use of linear integer programming for optimal selection of module mixes
As indicated byFig. 4, based on different affordable prices, the entire market is partitioned into three segments, like the low-end, middle-low-end, and high-end. Here, the company attempts to configure various types of digital cameras for multi-segments through considering the optimal product varieties and the constraints of manufacturing cost (relevant to pricing policies), simultaneously. Hence, the primary goal of a company is to maxi-mize its product capability, or state this equivalently, to maximaxi-mize its customer satisfaction because the latter is proportional to the former, positively. For convenience, we describe the objective, decision variables, and constraints as follows (see Eqs.(19)–(22)).
MR1 . . . . MRm TA1 . . . . TAn MR1
. MRm
m
m× correlation matrix m×n dependence matrix
TA1 . TAn
m
n× zero matrix n×n correlation matrix
Fig. 3. Input of the direct-relation matrix for fuzzy DEMATEL.
Characterizing a sport & water digital camera by using various modules
Lens A3, A4, A5 Panel A13, A14 CCD A1, A2 Low-end Segment High-end Segment Chip
A6, A7, A8, A9
Meddle-end Segment
Frame
A10, A11, A12
Fig. 4. Product configuration and module decomposition for three market segments.
MaxX i X j PijXij s:t:X j Xij¼ 1; for
8
i; X i X j CijXij6Bs; for8
s; Xij2 f0; 1g; ð19Þ Pij¼X k WtTAðkÞ ~yijðkÞ; ð20Þ ~yijðkÞ ¼ yijðkÞ minjyijðkÞ
maxjyijðkÞ minjyijðkÞ; ð21Þ
~
yijðkÞ ¼ maxjyijðkÞ yijðkÞ
maxjyijðkÞ minjyijðkÞ; ð22Þ
where i is the module index of the product, j is the alternative index for the module, Xijdenoted by a zero–one integer means the selec-tion of alternative j for module i, yij(k) represents the kth attribute of alternative j for module i, Pij and Cij denote its functional perfor-mance and manufacturing cost, respectively. For simplification, we assume that the product is characterized by various functional modules and each module which is composed of specific attributes has limited alternatives. Meanwhile, the whole market is divided into several segments and each segment has a cost constraint of
Bs for pricing the target product. In particular, functional
performance Pij(see Eq.(20)) is modeled by a weighted sum of var-ious TAs. They are normalized by one of the following types: either the-larger-the-better (see Eq.(21)for the benefit set) or the-smal-ler-the-better (see Eq.(22)for the cost set).
4. A real case study
The market growth of digital camera has become much more saturated than before and Kodak’s bankruptcy in 2011 has strongly impacted on the entire industry since several Taiwanese OEM/ ODM camera manufacturers are now incurring huge loss of ‘‘ac-counts receivable’’. To take advantage of its great cost-controlling capability and large manufacturing capacity, a Taiwanese elec-tronic company is planning to gradually change herself from an OEM/ODM firm into a brand company. To avoid fierce price compe-tition, this company is now thinking the way to differentiate them-selves from the other suppliers.
For fast acquiring market share, this company also focuses its core resources on aiming at recently emerging regions like East Europe, Middle-South America, and South-East Asia. In view of
Table 3
A sample denoting the correlations among MRs.
MRs R1 R2 R3 R4 R5 R1 Photo quality L R2 Video capability L R3 Electronic function H R4 Robust function H L R5 User interface L Table 4
A sample denoting the correlations among TAs.
TAs A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11 A12 A13 A14
A1 CCD mega pixels H
A2 CCD sensor area (inch) H VH
A3 Optical zoom H H
A4 Wide angle (mm) H
A5 Max aperture H M
A6 Dynamic video resolution H
A7 High-speed shutter (Y/N) H
A8 GPS receiver (Y/N) L
A9 Wireless LAN (Y/N) L
A10 Water proof (m) H M
A11 Shock proof (m) H M
A12 Freeze proof (°C) M M
A13 LCD size (inch) M
A14 Screen resolution M
Table 5
A sample denoting the dependences between MRs and TAs.
A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11 A12 A13 A14
R1 H M VH M H R2 H VH R3 H H R4 VH H H R5 M H Table 6
Derived weights (priorities) of MRs and TAs.
MRs Weights Rank TAs Weights Rank
R1 0.348 1 A1 0.090 4 R2 0.165 3 A2 0.093 3 R3 0.104 4 A3 0.117 2 R4 0.288 2 A4 0.076 7 R5 0.095 5 A5 0.118 1 A6 0.049 12 A7 0.054 11 A8 0.038 13 A9 0.037 14 A10 0.079 6 A11 0.066 8 A12 0.080 5 A13 0.052 9 A14 0.052 9
Fig. 4again, five basic modules that characterize a digital camera include CCD (charge couple device), lens, chip, frame, and panel and each module is composed of specific components and limited alternatives. Without loss of generality, the whole market is assumed to be divided into three segments that are based on customers’ affordable price (relevant to supplier’s manufacturing cost).
4.1. Marketing planning to derive the priorities of MRs and TAs In the beginning, a common technique named ‘‘majority-vot-ing’’ is employed to assist invited experts in determining represen-tative MRs and TAs. Thereafter, fuzzy Delphi method is used to generate the initial weights of MRs (see Eq.(8)), the correlations
among MRs/TAs (see Tables 3 and 4), and the dependences
be-tween MRs and TAs (seeTable 5). Here, both the ‘‘majority-voting’’ scheme and fuzzy Delphi method are adopted to help evaluators efficiently reach a consensus during a group decision process.
Secondly, we input initial values of dependences and correla-tions into the direct-relation matrix eX. After processed by fuzzy
DEMATEL, final values of dependences (Rij) and correlations
ðkik=
c
kjÞ could be extracted through the direct-relation matrix eT . Based on Eq.(1), the weights of MRs can be obtained and their pri-orities are presented as R1 R4 R2 R3 R5, as indicated by Ta-ble 6. Obviously, photo quality (R1), robust function (R4), and video capability (R2) are listed as the top three priorities for the sport & water digital camera.Based on Eqs.(2), (3), the weights of TAs are derived and the top six priorities demonstrate an order of A5 A3 A2 A1
A12 A10 (seeTable 6 again). Apparently, significant TAs result
from their strong dependences on dominant MRs. For example, ‘‘max aperture’’ (A5), ‘‘optical zoom’’ (A3), ‘‘CCD sensor area’’ (A2), and ‘‘CCD pixels’’ (A1) are closely related to ‘‘photo quality’’ (R1). Similarly, ‘‘freeze proof’’ (A12) and ‘‘water proof’’ (A10) are mainly affected by ‘‘robust function’’ (R4).
4.2. Visualization of their causal interrelationships between MRs and TAs
To identify the complicated relationships among all relevant factors inherent in the QFD,Table 7depicts various types of scoring through using Eqs.(16)–(18). In brief, the active score ‘‘D’’ denotes the ‘‘dispatched’’ impact of the corresponding factor on the others while the passive ‘‘R’’ score represents the sum of ‘‘received’’ influ-ence of the corresponding factor from the others. Moreover, the prominence score ‘‘D + R’’ defined by ‘‘adding the active score to the passive score’’ is regarded as its general degree of ‘‘impor-tance’’. By contrast, the influence score ‘‘D R’’ defined by ‘‘sub-tracting the active score from the passive score’’ indicates the intensity of ‘‘causality’’.
Based on Table 7, their relative influence (referred to the
vertical axis) and absolute prominence (referred to the horizontal axis) are displayed inFig. 5. Owing to ‘‘positive’’ influence, MRs (denoted by the symbol of ‘‘square’’) are classified into the ‘‘cause’’
Table 7
Various types of scoring for visualizing a causal diagram of the QFD.
Active score Passive score Prominence score Influence score
D R D + R D R R1 1.516 0.052 1.567 1.464 R2 0.691 0.052 0.743 0.640 R3 0.917 0.271 1.187 0.646 R4 1.349 0.314 1.662 1.035 R5 0.529 0.065 0.594 0.465 A1 0.342 0.610 0.953 0.268 A2 0.663 0.913 1.576 0.250 A3 0.598 0.880 1.478 0.281 A4 0.329 0.542 0.871 0.213 A5 0.577 1.072 1.650 0.495 A6 0.259 0.541 0.800 0.281 A7 0.259 0.575 0.834 0.316 A8 0.052 0.327 0.378 0.275 A9 0.052 0.327 0.378 0.275 A10 0.414 0.322 0.736 0.092 A11 0.414 0.271 0.684 0.143 A12 0.345 0.679 1.024 0.334 A13 0.172 0.953 1.126 0.781 A14 0.172 0.888 1.060 0.715
The dependences between MRs and TAs
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 Prominence Influence MRs TAs R4 R1 R5 R2 R3 A3 A2 A5 T1 A4 A1 A12 A14 A13 A6 A7 A6 A11 A10 A7 -1 -0.5 0 0.5 1 1.5 2
Fig. 5. A structural diagram to visual causal impacts inherent in the QFD.
Table 8
Combinations of module mixes for configuring the sport & water digital camera.
Module 1: CCD Module 2: Lens Module 3: Chip Module 4: Frame Module 5: Panel
M11 M12 M13 M21 M22 M23 M31 M32 M33 M41 M42 M43 M51 M52 M53 A1 1200 1400 1000 A2 1/2.3 1/2.3 1/1.8 A3 3 4 5 A4 28 28 25 A5 3.9 3.5 3.1 A6 640 480 720 480 1080 720 A7 0 0 1 A8 0 1 1 A9 0 0 1 A10 3 5 10 A11 1.5 2 3 A12 10 10 20 A13 2.7 3.0 3.3 A14 23K 46K 92K Cost ($TWD) 1050 1400 2100 1400 2000 2800 2100 3000 4200 1750 2500 3500 700 1000 1400
(dispatcher) group. Conversely, TAs (denoted by the symbol of ‘‘diamond’’) are categorized into the ‘‘effect’’ (receiver) group due to ‘‘negative’’ influence. As a result, by virtue of fuzzy DEMATEL, product planners or project managers are capable to generate the priorities of TAs in a market-oriented manner and to visualize their complex interrelationships, concurrently.
4.3. Optimizing module mixes for distinct market segments
Theoretically, the principle of target marketing assumes that di-verse customers should have different requirements for product capability and affordable price. Normally, higher purchasing price (manufacturing cost) will sustain better product specifications. According to its marketing survey, the marketing department of the consulted company offers an industrial dataset, as shown in Ta-ble 8. In this study, a sport and water digital camera is character-ized by five functional modules and each module is composed of specific attributes which work collaboratively. Besides, each mod-ule has three alternatives that are associated with different manu-facturing costs.
By virtue of linear integer programming (see Eqs.(19)–(22)), the optimal selection of various module mixes for multi-segments are described inTable 9. Here, an asterisk denotes the ‘‘selected’’ ule for the corresponding segment. For instance, five specific mod-ules are suggested (i.e. M11 for CCD, M23 for lens, M32 for chip, M41 for frame, M53 for panel) to be fabricated together when acquiring the middle-end segment. Similar explanations could be applied to the low-end and the high-end segments, respectively. Rather than using the roof of the QFD to indicate positive/negative correlations among TAs, specific alternatives of various functional modules are used to accommodate their complex correlations, including the ex-tremely exclusive constraints among TAs.
5. Conclusions
In the era of global customization, to survive in a wide range of market segments, companies need to balance the trade-offs be-tween enhancing product varieties and controlling manufacturing complexity. Consequently, numerous paradigms have received much attention, including product family architecture, platform-based development, and modular product design. In this paper, a fuzzy MCDM based QFD which integrates fuzzy Delphi, fuzzy DEM-ATEL, with LIP is presented to accomplish two fundamental tasks of NPD: collaborative product design and optimal selection of module mix with respect to distinct multi-segments. More importantly, this paper demonstrates the following merits:
To reduce the gap between customer needs and product devel-opment, this study is capable to gather opinions between indi-vidual customers and industrial experts and then fuse their assessments to reach a consensus.
To understand the causal impacts of marketing requirements on technical attributes, this study could visualize their complicated interrelationships and derive the priorities of technical attri-butes in a market-oriented manner.
To assist an enterprise in optimizing product varieties with respect to multi-segments, this study utilizes linear integer pro-gramming to maximize product capability with consideration of budget constraints on manufacturing costs, concurrently. For simplification, this study assumes that the entire market is partitioned into three segments which are based on the pricing policy of an enterprise. Future study might extend our current framework to a more general scenario in which market segmenta-tion is based on customer preference. In addisegmenta-tion, other classical techniques like conjoint analysis (Luce & Tukey, 1964) or Kano model (1984)might be further incorporated into the proposed framework to fulfill sufficient customer involvement.
References
Akao, Y. (1990). Quality function deployment: Integrating customer requirements into product design. Cambridge, MA: Springer.
Büyüközkan, G., Ertay, T., Kahraman, C., & Ruan, D. (2004). Determining the importance weights in the house of quality using the fuzzy analytic approach. International Journal of Intelligent Systems, 19, 443–461.
Büyüközkan, G., Feyziog˘lu, O., & Ruan, D. (2007). Fuzzy group decision-making to multiple preference formats in quality function deployment. Computers in Industry, 58, 392–402.
Chan, L. K., Kao, H. P., Ng, A., & Wu, M. L. (1999). Rating the importance of customer needs in quality function deployment by fuzzy and entropy methods. International Journal of Production Research, 37(11), 2499–2518.
Chen, L. H., & Ko, W. C. (2008). A fuzzy nonlinear model for quality function deployment considering Kano’s concept. Mathematical and Computer Modeling, 48, 581–593.
Chen, L. H., & Weng, M. C. (2003). A fuzzy model for exploiting quality function deployment. Mathematical and Computer Modeling, 38, 559–570.
Chen, L. H., & Weng, M. C. (2006). An evaluation approach to engineering design in QFD processes using fuzzy goal programming models. European Journal of Operational Research, 172, 230–248.
Cheng, C. H., & Lin, Y. (2002). Evaluating the best main battle tank using fuzzy decision theory with linguistic criteria evaluation. European Journal of Operational Research, 142, 174–186.
Delice, E. K., & Güngör, Z. (2009). A new mixed integer linear programming model for product development using quality function deployment. Computers and Industrial Engineering, 57, 906–912.
Fontela, E., & Garbus, A. (1976). The DEMATEL observer. Battelle Institute. Geneva Research Centre.
Geng, X., Chu, X., Xue, D., & Zhang, Z. (2010). An integrated approach for rating engineering characteristics’ final importance in product-service system development. Computers and Industrial Engineering, 59, 585–594.
Hsiao, S. W., & Liu, E. (2005). A structural component-based approach for designing product family. Computers in Industry, 56, 13–28.
Jiao, J., & Chen, C. H. (2006). Customer requirement management in product development: A review of research issues. Concurrent Engineering, 14(3), 173–185.
Jiao, J., Ma, Q., & Tseng, M. M. (2003). Towards high value-added products and services: Mass customization and beyond. Technovation, 23, 809–821. Jiao, J., & Tseng, M. M. (1999). A methodology of developing product family
architecture for mass customization. Journal of Intelligent Manufacturing, 10, 3–20.
Kahraman, C., Ertay, T., & Büyüközkan, G. (2006). A fuzzy optimization model for QFD planning process using analytic network approach. European Journal of Operational Research, 171, 390–411.
Kano, N. (1984). Attractive quality and must-be quality. The Journal of Japanese Society for Quality Control, 14(2), 39–48.
Karsak, E. E. (2004). Fuzzy multiple objective decision making approach to prioritize design requirements in quality function deployment. International Journal of Production Research, 42(18), 3957–3974.
Karsak, E. E., Sozer, S., & Alptekin, S. E. (2002). Product planning in quality function deployment using a combined analytical network process and goal programming approach. Computers and Industrial Engineering, 44, 171–190. Table 9
Results of selected module mix for three market segments (asterisk symbols correspond to selected modules).
Segment (cost budget) Module 1: CCD Module 2: Lens Module 3: Chip Module 4: Frame Module 5: Panel
M11 M12 M13 M21 M22 M23 M31 M32 M33 M41 M42 M43 M51 M52 M53
Low-end ($7000)
Middle-end ($10,000)
Kreng, V. B., & Lee, T. P. (2004). QFD-based modular product design with linear integer programming – A case study. Journal of Engineering Design, 15(3), 261–284.
Kwong, C. K., & Bai, H. (2002). A fuzzy AHP approach to the determination of importance weights of customer requirements in quality function deployment. Journal of Intelligent Manufacturing, 13, 367–377.
Kwong, C. K., & Bai, H. (2003). Determining the importance weights for the customer requirements in QFD using a fuzzy AHP with an extent analysis approach. IIE Transactions, 35, 619–626.
Lee, H. I., Kang, H. Y., Yang, C. Y., & Lin, C. Y. (2010). An evaluation framework for product planning using FANP, QFD, and multi-choice goal programming. International Journal of Production Research, 48(13), 3977–3997.
Lin, Y. H., Cheng, H. P., Tseng, M. L., & Tsai, C. C. (2010). Using QFD and ANP to analyze the environmental production requirements in linguistic preferences. Expert Systems with Applications, 37, 2186–2196.
Lin, M. C., Wang, C. C., Chen, M. S., & Chang, C. A. (2008). Using AHP and TOPSIS approaches in customer-driven product design process. Computers in Industry, 59, 17–31.
Lin, C. J., & Wu, W. W. (2008). A causal analytical method for group decision-making under fuzzy environment. Expert Systems with Applications, 34, 205–213.
Liu, E., & Hsiao, S. W. (2006). ANP–GP approach for product variety design. International Journal of Advanced Manufacturing Technology, 29, 216–225. Luce, R. D., & Tukey, J. W. (1964). Simultaneous conjoint measurement: A new scale
type of fundamental measurement. Journal of Mathematical Psychology, 1, 1–27. Moon, S. K., Simpson, T. W., & Kumara, S. R. T. (2010). A methodology for knowledge discovery to support product family design. Annals of Operations Research, 174, 201–218.
Murry, T. J., Pipino, L. L., & Gigch, J. P. (1985). A pilot study of fuzzy set modification of Delphi. Human Systems Management, 5(1), 76–80.
Park, J., Shin, D., Insun, P., & Hyemi, H. (2008). A product platform concept development method. Journal of Engineering Design, 19(6), 515–532. Saaty, T. L. (1980). The analytic hierarchy process. New York: McGraw-Hill. Saaty, T. L. (1996). Decision making with dependence and feedback: The analytic
network process. Pittsburgh: RWS publications.
Sein, E. S., Ho, S. A., Lai, Y. J., & Chang, S. I. (1999). An integrated group decision-making approach to quality function deployment. IIE Transactions, 31, 553–567. Wasserman, G. S. (1993). On how to prioritize design requirements during the QFD
planning process. IIE Transactions, 25(3), 59–65.
Yager, R. (1978). On a general class of fuzzy connective. Fuzzy Sets and Systems, 4(3), 235–242.
