You are on page 1of 12

Int J Adv Manuf Technol (2008) 36:419–430

DOI 10.1007/s00170-006-0853-3

ORIGINAL ARTICLE

Quality function deployment planning for platform design


Fereydoon Jariri & Seyed Hesameddin Zegordi

Received: 7 June 2006 / Accepted: 20 October 2006 / Published online: 22 November 2006
# Springer-Verlag London Limited 2006

Abstract Platform design is an important design activity in We introduce a mathematical model which incorporates
the automotive industry. This task is usually performed by these two important approaches into a single model.
highly qualified engineers. In this paper we introduce the Following the introduction and literature review, the case
mathematical programming model which uses QFD data of platform design in one of the biggest automakers in the
and provide the starting solution for the design team. The Middle East is studied and a discussion will show the
model essentially optimizes customer satisfaction subject to difference between the company solution and the mathe-
target cost. The empirical results for one of the biggest matical model solution. The huge savings is achieved
automakers in the Middle East are provided. The compar- through a mathematical modelling solution. Although the
ison has been made between company solution and solution of the mathematical model is not the end of design
mathematical model solution and shows the quality of the process, it does provide the robust starting solution and will
mathematical model solution. The second mathematical be completed with some modifications by the design team.
programming went through the design process and provid-
ed more insight for the design team. Introducing an index
for customer satisfaction is the secondary result of our 2 QFD and target costing
solution.
2.1 Brief theory review for QFD and target costing
Keywords Quality function deployment . Target costing .
Mathematical programming . Automobile design . The basic rationale of QFD is to systematically take the
Cost management customers’ desires down to the level of detailed operations.
The two QFD processes, the American Supplier Institute’s
(ASI) four-phase approach and GOAL/QPC matrix of
1 Introduction matrices approach, are widely accepted as effective pro-
cesses to implement [2].The ASI’s four-phase approach
Quality function deployment (QFD) and mathematical translates the customers’ needs into technical requirements,
programming together provide efficient solutions for new and subsequently component characteristics, process steps
product development teams. The history of QFD methods and operational steps (Fig. 1). Each of the translations uses
with OR/MS methodologies is rich [1]. Here we shortly a matrix, called a house of quality (HOQ). HOQ is a
review QFD and target costing methods and their applica- complex matrix [4] that provides means for interfunctional
tions in different functional areas. Then the case of platform planning and communications. The fundamental rationale
design in the automotive industry will be discussed. of HOQ is introduced in several publications [4, 5].
The target costing process is demonstrated in Fig. 2. We
F. Jariri (*) : S. H. Zegordi follow the course depicted in Fig. 2 whereby the target
Department of Industrial Engineering,
costing process is being used for a new product or service.
Tarbiat Modares University,
Tehran, Iran The target costing process begins with developing an
e-mail: fjariri@gmail.com understanding of unmet needs in the marketplace and then
420 Int J Adv Manuf Technol (2008) 36:419–430

Fig. 1 The four-phase approach 1.Planning Matrix


of QFD [3]

Technical
2.Design Matrix
Requirement

Customer
Attributes Component 3.Operational
Matrix
Characteristics

Technical
Requirement 4. Control Matrix
Process Steps

Component
Characteristics Operational
Steps

Process
Steps

determining what the customer would pay to have their Lee and Monden [12] performed an international
needs met, i.e., the “target price”. Internal goals and comparison of cost management systems which have been
pressures determine the profit margin desired. Calculating claimed as manufacturing-friendly in the U.S. and Japan.
the allowable target cost is the next step, where: Specifically they compared activity based costing (ABC) to
target costing and Kaizen costing, two of the most popular
Target Cost ¼ estimated selling price  desired profit:
methods on the international horizon, regarding their
relative merits in strategic cost management and operational
During the next step, the target cost is apportioned improvement and control. A field study of a world-class
among the key cost elements, and further broken down to Japanese automaker, combined with an analysis of cost
the materials or components level. This is the level at which management literature, is relied upon to examine desirable
the organization actually tries to manage the cost elements. characteristics of cost management systems that can avoid
Based on the component level goals or targets, the next step the problems many competitive manufacturers have en-
involves undertaking various cost management activities to countered in satisfying management accounting information
achieve the target component/materials prices. needs under rapidly changing market conditions in the
global context.
2.2 Literature review for QFD & target costing Dekker and Smidt [10] reported the results of a survey
among Dutch firms listed at the Amsterdam stock exchange
Both approaches provide successful results in the area of on the adoption and use of costing practices that resemble
cost management. QFD planning with mathematical the Japanese target costing concept. They reported that
programming has been studied by various researchers [7– nineteen out of thirty-two manufacturing firms claimed to
9]. Target costing is another approach for design cost use these practices, although they used different names for
management [10, 11]. As discussed in several articles, them. The study suggests that these practices were
target costing is a totally different approach in the area of developed independently of the Japanese practice. Adop-
cost management; it starts from the market and estimation tion is highest among assembling firms and is related to a
of product price in the market. Subtracting company profit competitive and unpredictable environment. Cost reduction
from the market price gives the prices for different is the main objective and benefit of these practices. The
components which the company should produce. Usually product development and design departments are leading in
value engineering should be recruited here to meet the the target cost management process, which is mainly
capabilities of the production firm with the market expect- performed in team structures.
ations in terms of price. The target costing discussion is Bayou and Reinstein [13] applied the new method of
originally motivated by cost management activities and associative costing. They believe their method does not
attracts significant attention in supplier management in- contain the three main drawbacks of the two currently used
volvement [6], but little attention in R&D activities. Here cost management methods, namely, ABC and target
we emphasize this approach for the design activities and costing. The first drawback they mentioned is that the
trying to combine it with QFD data. Related literature in product “user” and “consumer” are not identical as often
target costing methodology is as follows. assumed in target costing projects. Second, each of the two
Int J Adv Manuf Technol (2008) 36:419–430 421

Fig. 2 Target costing process


Product/
[6] Product/Service Service Customer
Marketing Input Characteristics Input
Desired

Competitive Target Customer


Market Selling Input
Condition Price

Competitive Desired Management


Market Profit Input
Condition Margin

Target Cost = Target Price – Desired Profit

Supply Cost Breakdown to Engineering/


Management/ materials/ R&D Input
Supplier Input Component level

Supply R&D
Management Design
Cost Management Activities
* Supplier Development
* Design Change
*Material Change
* Specification Change
* Cost Tradeoffs
Suppliers/ Manufacturing
Marketing

Overall Target
Cost Achieved

Continuous
Improvement

techniques has a limited costing focus: ABC focuses on make trade-off decisions on a continuum beginning with
indirect costs and target costing on unit-level costs. Third, the design-to-cost point and ending at the cost-to-design
neither technique accounts for resource interactions and extreme, e.g., when the best perceived design and the
cost associations. They believe their method—associative acceptable cost level of this design are incongruent.
costing—does not contain these limitations. This method Applications of QFD in different functional areas are
uses the well known statistical techniques of clustering, full reported in many articles. Product development and product
factorial design and analysis of variance. It concludes that design are among these fields. Wei et al. [14] proposed how
in product design programs, the design team may need to a design automation system that integrates a traditional
422 Int J Adv Manuf Technol (2008) 36:419–430

segmental design process into a solid unified system design teams in understanding and analysing the system
composed of need assessment, computer aided design, interrelationships and obtaining optimal target engineering
computer aided engineering and rapid prototyping would be characteristic values.
performed. That system utilizes the computer network in The application of QFD in decision making DM is also
conjunction with the QFD technique coupled with a well- vast. In the evaluation phase of DM, QFD shows a good
designed expert system to precisely transform customers’ application. Lowe et al. [20] present a tool developed from
needs into producible product specifications. The product is the QFD technique. This tool allows a rapid evaluation of
then displayed using a three-dimensional representation and the feasibility of using the thixoforming process to
simulated through numerical analysis using a finite element manufacture products.
method. The resulting feasible design alternative is finally
linked to the rapid prototyper to produce the sample object.
These integrations are in the same path that we follow in 3 Product family strategy and platform design
our mathematical model. Schmidt [15] discussed that the
traditional QFD procedure takes neither the development of Platform design problems in the automotive industry are the
market orientated nor the constructional product concept, matter of challenging research studies. In this section we
nor the coordination of both, into account. In order to briefly review a few papers in this regard and in the next
overcome this and other deficiencies of the traditional QFD section we applied the QFD-based mathematical model to
method, the author developed the process model of tackle the platform design problem.
“integrated concept development”. It is proposed to fill Numerous manufacturing industries are seeking ways to
the gap between marketing science and engineering, reduce manufacturing costs and development time of
consequently relating market orientated concept develop- product families through implementation of platform
ment and testing to the house of quality concept of QFD. strategies. Deploying several variants from a single product
In the automotive world, Thomas discussed how the platform can yield considerable savings in time, cost and
leaders in that industry face challenges in different categories maximize profit. Currently, a number of methodologies to
such as global competition, increased labour costs, rising determine optimum product platform architecture have been
customer expectations, shorter product life cycles and developed. These methodologies are helpful in determining
government regulations and clearly illustrated how concur- the components that should be common between products
rent engineering in meeting these demands embraces and therefore represent the constituent elements of a
supporting subsystems that include computer aided drafting product platform.
and design, QFD, and design for manufacture and assembly. Several publications are available in the area of product
In auto part systems, Ngai et al. [16] have demonstrated the platform design. In this section we review five articles
use of expert system technology and the technique of QFD which are closely related to this topic.
in supporting car audio design planning. QFD is applied as Simpson et al. [21] proposed the product platform
a knowledge acquisition method in this study to support car concept exploration method. In this paper, the authors state
audio design teams in the development of products in a that it is a “formal method that facilitates the synthesis and
structured way that relates market demand via engineering exploration of a common product platform concept that can
specifications to parts specifications. be scaled into an appropriate family of products”. The
The history of OR/MS techniques and QFD applications in method applies to scalable product platforms and families,
decision making is also rich. Lee and Kusiak [17] illustrated and consists of five steps: (1) market segmentation grid
how OR/MS techniques (integer programming, a modified creation, (2) factor and range classification, (3) meta-model
maximum spanning tree and genetic algorithm) can provide creation and validation, (4) product platform specifications
a solution which determines a priority order of design rules, aggregation, and (5) product platform and family develop-
when design rules affect each other and when they may ment. The method is demonstrated through a universal
conflict with the interests of other rules. The article discussed motor case study, where families of ten motors are designed
the relationship of the design rules and the design require- by varying the stack length. The compromise decision
ments and the interaction between any pair of design rules in support problem formulation is used for the optimization
the house of quality. Chaung [18] combines the analytic problem formulation. This work presents a systematic way
hierarchy process (AHP) and QFD techniques to support a to determine the product platforms and families.
facility location decision from the requirement perspective. Martin and Ishii [22] proposed another platform design
Moskowitz and Kim [19] developed a decision support method, called design for variety method, to develop a
system prototype for QFD based upon an integrated modularized product platform. The authors used genera-
mathematical programming formulation. The model helps tional variety index (GVI) and coupling index (CI) to
a design team build a house of quality chart and supports design platform architecture that can be easily changed in
Int J Adv Manuf Technol (2008) 36:419–430 423

the future. In the paper, GVI is defined as an “indicator and less cost. A mathematical model is a proper candidate
of the amount of redesign required for a component to for resolving this trade-off by putting one factor in the
meet the future market requirements”. The CI “indicates the objective function and the other in the constraint sets.
strength of coupling between the components in a product. From the other side, as we discussed earlier, many
The stranger the coupling between components, the more researchers used mathematical programming tools in their
likely a change in one will require a change in other”. The QFD method investigations. Especially Pearson et al. [27]
method is demonstrated through a water cooler example, used QFD data for manpower planning. They didn’t
where GVI and CI for seven major components are consider cost in their problem; however, they provided the
calculated. Then, for components with high GVI and CI, sound basis for merging the QFD data matrix with a
flexible designs were generated to reduce GVI and CI, thus mathematical programming matrix.
lowering future redesign (switch) cost. Here we use our previously developed model [28] to
Li and Azarm [23] developed a design process for tackle the platform design problem. The optimization
product line (family) design under uncertainty and compe- problem is maximizing customer satisfaction subject to
tition. The design process is divided into the design target cost. Target costing is the subject of vast discussion
alternative generation stage and the design evaluation stage. in cost management literature but has gained limited
During the design alternative generation stage, each design attention in the OR/MS community. The topic was
alternative is optimized through multi-objective optimiza- described by Fortune magazine as “Japanese smart secret
tion. In the design evaluation stage, each design alternative weapon”.
is evaluated using multi-objective genetic algorithm, due to We propose the following definition to build the model.
the combinatorics nature of the formulated optimization Our approach is the same as Pearson et al. [27] which was
problem. In the end, the best product line (family) is chosen developed for manpower planning and modify it to
using the selection rule, which takes into account designer’s consider the target cost. Thus, we have:
utility of the product line balance. The proposed process is
i : ith customer requirements i ¼ 1; . . . ; n
demonstrated through a case study, where a cordless
screwdriver family is designed. Of three major components k : kth system k ¼ 1; . . . ; m
(motor, gear, battery), the motor was designated as the LkL : number of level for kth system L ¼ 1; . . . ; Lk
platform component a priori. Through optimization of three uikL : the intensity that the Lth level of the kth sytem
has on ith customer requirements:
components, the author identified best designs for several
different uncertainty scenarios. uikL ; are the elements of HOQÞ:
Finally, Gonzalez-Zugasti et al. [24, 25] introduced a
quantitative method for architect product platforms, and a These different levels (HOQ levels) are simply the
framework to assess the value of product platform based different alternatives which exist to satisfy the desired
families using the real options approach. “function” in the second table of QFD.
E.S. Suh [26] studied the flexible product platform in his
research and surveyed the topic comprehensively. wi : Weight fori-th customer requirements
Although the focus of our research is in the automotive xkL : Decision variable—the percent of alternativeL for
industry where huge saving is expected, other products are systemk
also good candidates for our approach. One example for CkL : The cost for performing in levelL fork-th system
this is the Walkman platform whereby Sony introduced yi : The summation of effects of systems fori-th
some 250 different Walkmans during the 1980s. customer requirement (Notice for its computation in
Along with the problem of platform design which was the formula)
surveyed in the above mentioned articles, we looked through γik : The relation between systems (roof of HOQ)
this problem by QFD analysis and tried to incorporate QFD The point we should make here is now we have cubic of
data through the decision making process for platform design. quality instead of house of quality. It means for each customer
requirement and for each level of system, we have the value of
uikL (uikL can have the value 1-3-9 as the regular HOQ). Now
4 Mathematical model we have the following mathematical programming:
X n

The motivation for developing the mathematical program- Max Z ¼ w i yi ð1Þ


i¼1
ming for the problem initialized with the fact that there is
an inherent trade-off in the problem. This trade-off exists X
LK
between the customer satisfaction and the cost spent for Subject to : xkL ¼ 1 k ¼ 1; 2; :::; m ð2Þ
that. It means the customer is looking for more satisfaction L¼1
424 Int J Adv Manuf Technol (2008) 36:419–430

m X
X LK m1 X
X m X
Lk X
Lj Table 1 Platform systems
yi ¼ uikL xkL þ γkj xkL xjγ
k¼1 L¼1 k¼1 j¼kþ1 L¼1 v¼1 Code Systems / subsystems
ð3Þ
P Powertrain & cooling
i ¼ 1; 2; :::; n P.1 Engine
P.2 Transmission
P.3 Intake
P.4 Exhaust
X
LK P.5 Fuel system
CkL xkL  TCk k ¼ 1; :::; m ð4Þ P.6 Cooling
L¼1 P.7 HVAC
V VDS
V.1 Frt axle
X
m
V.2 Rr axle
TCi  T ð5Þ V.3 Wheel & tire
i¼1
V.4 Steering
V.5 Rr & frt brakes
B Body
xkL  0 ð6Þ B.1 Frt end
where T is the target cost calculated for the problem. B.2 Rr end
B.3 Body side & roof
Equation 1, the objective function, maximizes the
B.4 Closures
customer satisfaction. Thewi are weights for thei-th cus- B.5 Underframe
tomer requirement andyi would be computed by Eq. 3. B.6 Paint & sealants
Equation 2 guarantees that each system consisting of T Interior & exterior
different levels (percentage) would be equal to one T.1 Fascia
(100%). Equation 3 simply gathers the effect of each T.2 Sound deadening & int equipment
selected level of different alternatives on the i-th customer T.3 Seatings
requirement. Equation 4 determines the upper boundTCk for T.4 Restraints
T.5 Exterior equipment
the cost of system k.TCk themselves are variables and the
T.6 Mechanisms
model allows them to be computed in an optimum way. T.7 Lightening & signaling
Finally Eq. 5 determines the upper bound of the target cost T.8 Bare glasses
for the entire system. Target cost is the input of the model T.9 Sealing
and would be provided by the decision maker. E Electrical
E.1 Audible warning
E.2 Body electronic & electrical control
E.3 Electrical cable - fixing and supply
5 Case study—platform design
E.4 Driver information
E.5 Communication and entertainment
One of the biggest automakers in the Middle East decided E.6 Platform electronic & electrical control
to design its own platform. Having their own platform has E.7 Mechanism electrical equipment
several advantages which were mentioned before. The O Other
automobile platform consists of 34 different systems which
are shown in Table 1. These systems are categorized into
five main systems which consist of: altogether there exist five different choices which we call
level to select among them. By that we mean for each
1. Powertrain & cooling
individual system of platform, we can select a percentage
2. Vehicle dynamic systems (VDS)
from each alternative. For instance for engine, it is possible
3. Body
to choose 93% of alternative 3 and 7% of the new
4. Interior and exterior
alternative. It means we choose alternative 3 as the basis
5. Electrical
for engine and 7% of parts would be new in this scenario.
The reader can appreciate the huge amount of different
For our company these systems can be chosen from three scenarios which are possible to form each system and
commercially available alternatives which we will call: therefore the complexity of the problem. This phase is
alternative 1, alternative 2 and alternative 3. Two other called “concept selection” in automobile design terminolo-
options also exist which we call: exist and new. Therefore, gy. Now the problem is how to select among these
Int J Adv Manuf Technol (2008) 36:419–430 425

alternatives (levels) to maximize customer satisfaction— the original system is engine). Engine consists of three
market share—due to the target cost. As was previously subsystems:
mentioned, target cost has a different accounting philoso-
1. Base engine
phy for cost management. Maximizing customer satisfac-
2. Engine mounts
tion—as the philosophy of QFD—directly affects sales and
3. Exhaust manifold
therefore maintains the current market and provides strong
basis for increasing market share. The mathematical model is
a powerful tool in the “concept selection” phase of platform Table 3 shows how the designer selected the subsystems
design and can tackle the complexity of the problem. and how the 93.3% and 6.7% are computed. These
procedures are followed by a design team for all 34
5.1 Factory solution different systems and the results are summarized in Table 2.
Therefore each cell in Table 2 represents the percentage of
Qualified engineers chose the solution which is given in that level which was chosen for a specified system.
Table 2 for the platform design problem. For instance, for Obviously the combination of more than two levels is
system1—engine—93.3% of level 3 (alternative 3) will be possible. For instance, system 20 in Table 2 which is
chosen and the remaining 6.7% of parts would be new ‘sound deadening and interior equipment’ would be formed
(level 5). This decision essentially was based on the by 25.4% of alternative 1, 7.7% of existing parts and 66.9%
subsystems which exist for the original system (in our case of newly developed parts.

Table 2 Company solution


Systems / subsystems Alternative 1 Alternative 2 Alternative 3 Exist New

1 Engine 93.3 6.7


2 Transmission 88.6 11.4
3 Intake 74 26
4 Exhaust 57.5 42.5
5 Fuel system 22.2 16.7 16.7 44.4
6 Cooling 92.9 7.1
7 HVAC 75.6 24.4
8 FRT axle 63.8 36.3
9 RR axle 51.3 48.7
10 Wheel & tyre 100
11 Steering 60 40
12 Rear & front brakes 90 10
13 Front end 100
14 Rear end 40 60
15 Body side & roof 100
16 Closures 100
17 Underframe 100
18 Paint & sealants 80 20
19 FASCIA 15 85
20 Sound deadening & int equip 25.4 7.7 66.9
21 Seatings 38.5 61.5
22 Restraints 42.9 57.1
23 Exterior equipment 12.5 87.5
24 Mechanisms 5 47.5 47.5
25 Lighting & signaling 20 80
26 Bare glasses 100
27 Sealings 100
28 Audible warning 75 25
29 Body electronic & control 32.5 47.5 20
30 Electrical cable fixing & supply 36 20 44
31 Driver information 100
32 Communication & entertainment 26.7 66.7 6.7
33 Platform electronic 23.3 63.3 13.3
34 Mechanism electrical equip. 21.7 41.7 8.3 28.3
426 Int J Adv Manuf Technol (2008) 36:419–430

Table 3 Subsystems for engine The model tries to identify the optimum amount or
percentage that should be chosen from the specified
Alt 1 Alt 2 Alt 3 Exist New
alternative to maximize customer satisfaction due to target
Base engine 0 0 100 0 0 cost. The problem is formulated and solved by GAMS
Engine mounts 0 0 80 0 20 (general algebraic modeling system) software and the
Exhaust manifold 0 0 100 0 0 solution provided in Table 4. It should be noted that the
Engine 0 0 93.3 0 6.7
solution by mathematical programming needs some mod-
ifications to be applicable in terms of physical condition of
the problem. By that we mean it needs some postprocessing
5.2 Mathematical programming solution activities by the design team to be an applicable solution.
For instance, the design team should make a decision for
The mathematical programming essentially deals with about 61%, which is the solution of the mathematical model
decision making. It tries to model the problem properly for alternative 1 in the engine system. They may come up
and achieve the solution in an optimum way. The decision with the solution that 65% from alternative 1 and 35% from
making concern in our problem is the amount or the alternative 2 is best fitted to their design goals.
percentage that should be chosen from the specified The target cost of 72,000,000 local units is used for the
alternatives to form the desired system. problem for all systems. QFD and subsequently the HOQ
The systems of Eqs. 1–6 described the behaviour of the matrix play important roles in our formulation. The uijk as the
model by mixed integer zero-one nonlinear programming. basis for our mathematical programming matrix identify the

Table 4 Mathematical model


solution Systems / subsystems Alternative 1 Alternative 2 Alternative 3 Exist New

1 Engine 61 39
2 Transmission 100
3 Intake 60 40
4 Exhaust 80 20
5 Fuel system 80 20
6 Cooling 80 20
7 HVAC 80 20
8 FRT axle 80 20
9 RR axle 100
10 Wheel & tyre 100
11 Steering 80 20
12 Rear & front brakes 100
13 Front end 100
14 Rear end 70 30
15 Body side & roof 100
16 Closures 100
17 Underframe 100
18 Paint & sealants 100
19 FASCIA 100
20 Sound deadening & int equip 100
21 Seatings 100
22 Restraints 80 20
23 Exterior equipment 100
24 Mechanisms 70 30
25 Lighting & signaling 100
26 Bare glasses 100
27 Sealings 100
28 Audible warning 100
29 Body electronic & control 73 27
30 Electrical cable fixing & supply 60 20 20
31 Driver information 100
32 Communication & entertainment 100
33 Platform electronic 50 50
34 Mechanism electrical equip 100
Table 5 Subsytems and percentage for platform systems

Subsystem Subsystem Subsystem Subsystem Subsystem Subsystem Subsystem Subsystem Subsystem Subsys Subsys Subsys
1 2 3 4 5 6 7 8 9 10 11 12

Engine Base Eng mounts Exhaust


engine manifold
Percent 70 20 10
Transmission Gearbox Differential Clutch/torque Transmission Gear shift Drive shaft Pedals
convertor mounts mechanism
Percent 35 20 10 10 10 10 5
Intake Air filter Clean side Dirty side pipes Retainers/ Resonator
pipes fixing
Percent 5 30 30 20 15
Int J Adv Manuf Technol (2008) 36:419–430

Exhaust Silencer Retainers/ Pipes Catalytic


fixing convertor
Percent 15 20 35 30
Fuel system Fuel tank Filler piper Filler cap Carbon Electrovalve Pipes Retainers Filter Fuel
canister regulator
Percent 25 5 5 5 10 15 10 10 15
Cooling Water Oil cooling Intercooler Fans Pumps Circuit Holding
cooling tanks
Percent 20 20 20 5 15 10 10
HVAC Pipes HVAC unit Thermostat Retainer Sealings Valves Condenser Fans Evaporator
Percent 10 20 10 10 10 10 5 5 10
FRT axle FRT FRT FRT stabilizer FRT spindles FRT knuckles FRT control FRT FRT bump
spring damper arm bearings stop
Percent 20 25 15 10 10 10 5 5
RR axle RR spring RR damper RR stabilizer RR spindles RR knuckles RR control RR RR bump
arm bearings stop
Percent 20 25 15 10 10 10 5 5
Wheel & Wheel Tyre
tyre
Percent 70 30
427
428 Int J Adv Manuf Technol (2008) 36:419–430

Table 6 Mathematical model


solution with details Systems / subsystems Alternative 1 Alternative 2 Alternative 3 Exist New

1 Engine 70 30
2 Transmission 95 5
3 Intake 100
4 Exhaust 100
5 Fuel system 100
6 Cooling 10 90
7 HVAC 10 90
8 FRT axle 100
9 RR axle 100
10 Wheel & tyre 100
11 Steering 100
12 Rear & front brakes 100
13 Front end 100
14 Rear end 100
15 Body side & roof 100
16 Closures 100
17 Underframe 100
18 Paint & sealants 100
19 FASCIA 100
20 Sound deadening & int equip 100
21 Seatings 100
22 Restraints 100
23 Exterior equipment 100
24 Mechanisms 100
25 Lighting & signaling 100
26 Bare glasses 100
27 Sealings 100
28 Audible warning 7 7 86
29 Body electronic & control 100
30 Electrical cable fixing & supply 100
31 Driver information 100
32 Communication & entertainment 100
33 Platform electronic 20 50 30
Target cost=72,000,000 Rials; 34 Mechanism electrical equip 100
Z=1093

importance of k-th level of j-th systems for the i-th customer 6 More mathematics
requirements (they accept 1-3-9 as regular HOQ). These are
the inputs for the model and were provided by the technical In the last section we received the solution that can help the
team. Customer requirements are another important factor in design team in their decision making process. However we
QFD and therefore our formulation. Thirteen customer can go one step further and help decision makers in their
requirements were considered for the platform problem but task with the aid of a mathematical model. We modify our
for the sake of privacy are not listed here. Another advantage model here with two more constraint sets to investigate the
of our model is to look to the QFD and target costing with a details of the problem more accurately.
single model. Companies usually perform these two tech- First we identify the subsystems for each 34 platform
niques sequentially and therefore the result affect each other systems. For instance for the engine we have the following
and thus a non-optimal result would be achieved. Our model subsystems:
overcomes this drawback and looks through the problem in a
(a) Base engine 70%
systematic way to achieve optimality.
(b) Engine mounts 20%
Table 7 Comparison of results (c) Exhaust manifold 10%
Company Mathematical model This means the engine system consists of three indepen-
Customer satisfaction 1,037 1,093
dent subsystems and the first subsystem (base engine) has
Cost (local currency) 78,073,000 72,000,000 70% of the value of the original system (engine). The 70%
comes from the price analysis performed by the financial
Int J Adv Manuf Technol (2008) 36:419–430 429

1,200 The system of Eqs. 1–8 was applied to the platform


1,000 Customer design problem and the solution is summarized in Table 6.
800 Satisfaction The model is a mid-sized model with almost 1,000
600 integer variables and 800 constraints.
Cost * 10000
400 The main purpose of Eqs. 7 and 8 is to take care of the
200 physical feasibility of the problem. Physical feasibility is an
0 important subject here and these two equations enforce the
Company Math Model decision variables to accept only the percentage that is
Solution Solution feasible for the problem and the other percentage would not
Fig. 3 Comparison of results
be considered at all.
To discuss further the topic of physical feasibility, we
mention a point here that our model would be applied in the
department. The list of subsystems for some 34 main “concept selection” phase of the design process and any
platform systems is summarized in Table 5. The rows under inconsistency or infeasibility would be solved in the “detail
each subsystems show the percent that this subsystem design” phase by the designer.
owned from the original system.
In terms of mathematics, now we have the zero-one
variables xijk which would be defined as follows: 7 Conclusion
xijk 1 if the kth subsystem of the jth level has
been chosen for the ith system As we can see in Table 6 the solution by modified
0 otherwise mathematical model has superior results in terms of customer
satisfaction and cost. The model went through the details of
Now we have two sets of constraints which will be
the design process and tried to identify the proper subsystem
added to Eqs. 1–6.
for each platform system. As it was shown in Table 7, the
The first equations show one subsystem should be
model increased customer satisfaction by 5.4% and reduced
chosen from a specified level, such that:
Lj
cost by 7.7%. Figure 3 presents the results in histogram
X k ¼ a; b; :::: format. These statistics for a plant with production of
xijk ¼ 1 ð7Þ
j¼1
i ¼ 1; 2; . . . L 600,000 vehicles per year would show the huge amount of
savings for the company. The calculations show the amount
where L is the number of systems.
of $300 savings for each car which results in 180 million
Now if aik are the percentage of the k subsystem from
dollars savings for the factory each year. Our experience
system i, they are input and uncontrolled parameters and
confirmed the previous result of combining QFD with
would be identified by an expert.
mathematical programming. In our case we went one step
In our engine example:
further and bring the target cost concept to this combination.
a1a ¼ 70% a1b ¼ 20% a1c ¼ 10%
The model also provided an index in its objective
respectively. function which reflects the customer satisfaction indices
Now we have the following set of constraints: for the selected scenario. This index is a powerful tool for
X j ¼ 1; 2; . . . ; Lj evaluating different scenarios and choosing the best one.
xij ¼ aik xijk ð8Þ
k¼a;b...
i ¼ 1; 2; ::; L This index indirectly helps to show that market share would
increase by providing this concept to the market. This can
All these 34 systems have distinct and defined “func- be recognized as productivity tools which reflect the effects
tions” which are uncontrolled parameters. There are several of choosing different scenarios in satisfying customer
alternatives or solutions which are controllable parameters demands. In the absence of a mathematical model there is
that are candidate to satisfy those “functions”. Our model no index provided to be used as the comparison tool for
tries to choose the best possible set of alternatives to measuring customer satisfaction criteria.
perform the desired system performance. The model is
general enough to cover all systems and closely related with
the topic “modularity”.
References
xij as described before is the percent of alternative j for
system i and according to decision variables xijk can be
1. Chan L, Wu M (2002) Quality function deployment: a literature
computed by constraints (Eq. 8).
review. Eur J Oper Res 143:463–497
Equations 7 and 8 can be added to the model and xij 2. Revelle JB, Moran JW, Cox CA (1998) The QFD handbook.
represent positive variables. Wiley, New York
430 Int J Adv Manuf Technol (2008) 36:419–430

3. Govers CPM (2000) QFD is not only a tool but also a way of 17. Lee GH, Kusiak A (2001) The house of quality for design rule
quality management. Int J Prod Econ 7(1):151–159 priority. Int J Adv Manuf Technol 17(4):288–296
4. Cohen L (1995) Quality function deployment—how to make 18. Chuang PT (2001) Combining the analytic hierarchy process and
QFD work for you. Addison-Wesley, Reading, MA quality function deployment for a location decision from a
5. Hauser JR, Clausing D (1988) The house of quality. Harvard requirement perspective. Int J Adv Manuf Technol 18(11):842–
Business Review, May–June, pp 63–73 849
6. Ellram LM (2002) Supply management’s involvement in the 19. Moskowitz H, Kim KJ (1997) QFD optimizer: a novice friendly
target costing process. Eur J Purch Supply Manag 8:235–244 quality function deployment decision support system for optimiz-
7. Bode J, Fung RYK (1998) Cost engineering with quality function ing product designs. Comput Ind Eng 32(3):641–655
deployment. Comput Ind Eng 35(3/4):587–590 20. Lowe A, Ridgway K, Atkinson H (2000) QFD in new production
8. Wasserman GS (1993) On how to prioritize design requirements technology evaluation. Int J Prod Econ 67(2):103–112
during the QFD planning phase. IIE Trans 25(3):59–65 21. Simpson T, Maier J, Mistree F (2001) Product platform design:
9. Fung RYK, Tang J, Tu Y, Wang D (2002) Product design method and application. Res Eng Des 13(1):2–22
resources optimisation using a non-linear fuzzy quality function 22. Martin M, Ishii K (2002) Design for variety: developing
employment model. Int J Prod Res 40(3):585–599 standardized and modularized product platform architecture. Res
10. Dekker H, Smidt P (2003) A survey of the adoption and use of Eng Des 13(4):213–235
target costing in Dutch firms. Int J Prod Econ 84(3):293–305 23. Li H, Azarm S (2002) An approach for product line design
11. Cooper R, Slagmulder R (1997) Target costing and value selection under uncertainty and competition. J Mech Des 124
engineering. Productivity Press, Portland, OR (3):385–392
12. Lee J, Monden Y (1996) An international comparison of 24. Gonzalez-Zugasti JP, Otto K, Baker J (2000) A method for
manufacturing-friendly cost management systems. Int J Account architecting product platforms. Res Eng Des 12(2):61–72
31(2):197–212 25. Gonzalez-Zugasti JP, Otto K, Baker J (2001) Assessing value in
13. Bayou M, Reinstein A (2004) Accounting for cost interactions in platformed product family design. Res Eng Des 13(1):30–41
designing products. In: Lee J, Epstein M (eds) Advances in 26. Suh ES (2005) Flexible product platform. PhD Dissertation,
management accounting, vol 12. JAI Press, Greenwich, CT, pp Massachusetts Institute of Technology, Cambridge, MA, USA
151–170 27. Pearson K, Alidaee B, Rego C, Kochenberger G (2003) Using
14. Wei CC, Liu PH, Chen CB (2000) An automated system for quality function deployment in NAVY manpower planning, a
product specification and design. Assem Autom 20(3):225–233 supply chain management approach. Working paper. The Univer-
15. Schmidt R (1997) The implementation of simultaneous engineer- sity of Mississippi, USA
ing in the stage of product concept development: a process 28. Jariri F, Zegordi H (2005) Quality function deployment, value
orientated improvement of quality function deployment. Eur J engineering and target costing, an integrated framework in design
Oper Res 100(2):293–314 cost management; a mathematical programming approach. Work-
16. Ngai EWT, Chow DYH (1999) ICADS: intelligent car audio ing paper. Department of Industrial Engineering, Tarbiat Modares
design systems for product planning. Exp Syst 16(1):19–32 University, Iran

You might also like