You are on page 1of 34

Volume 29 Number 2 I Fall 2006

contents: inside this iSSLIC::

Increasing
Editor's Comments
Dr Roy M. Woodheacl, CVS 1 Value in
Editor Value World
Product
Integrating Value Methodology 2
into Product Development and
Project Management Processes
Design
at Pratt Ei Whitney Canada
Alain LeBlanc, CD, PEng, MSc, MEng

Methodological and Practical 8


Aspects of Data Mining in the
Product Development Process
Prof Tai Shen Huang
-

Value Driven Design 18


Paul D. Collopy, PhD
Joseph E Otero, Jr, CVS-Life

Managing Value Engineering 26


In New Product Development
Don J. Gerhardt, CVS, PhD, PE
Extending Our Reach

This edition of Value World has four papers that provide We plan to address this situation by republishing
insights into ways value management can be used in the conference papers from around the world—not just SAVE
context of increasing value in product design. This is International conferences—to reliably send out three
appropriate not only to members of SAVE International,
editions per year. Our hope here is that by increasing
but also to our own publication. If we view Value World as a
publication reliability we will become a more trusted place
product, then we can learn from these papers and question
for potential authors. We also plan to make Value World an
whether our proposed strategies are maximising value. My
electronic publication so that we can dramatically increase
analysis is that we are not, and so need to do something
different. our circulation numbers and extend our reach without
This realisation was triggered by a need to find a new creating cost burdens for our society. We hope this will
editor. We owe a lot to the hard work of Derek Thomson also make it more appealing to new authors and that, in
and Kirsty Hunter from Glasgow Caledonian University in time, we can move away from republishing conference
Scotland. Particular thanks go to Derek for the excellent papers and get back to our goal of only publishing original
work he has done in modernising the operational processes material.
used to review, publish, and disseminate Value World.
In order to make this even more beneficial to SAVE
His efforts have made Value World a more professional
International, we will also carry advertisements and hope
publication and one we can be proud of. This has been
that we can build a revenue stream that will later open the
done whilst he has started a new career and the time
way to other options such as a ValueWorld discussion forum
constraints that have brought have caused a heavy burden
for him. Due to his current workloads, meeting deadlines and blogs.
has become difficult for him; but that does not diminish the So, we are reinventing our plan to grow Value World and
excellent contribution he has made to our community. He hope you the readers will enjoy the benefits of a wider reach
has delivered some good editions and will be remembered to other practitioners, managers, clients and academics
for his achievements. We thank him and wish him well in from around the world.
his future career. We also acknowledge the excellent work
Kirsty has done to build links with other societies and the
Best wishes
potential to access their conference papers.
The big problem facing Value World right now is the
low number of papers submitted to our review process.
Given we only circulate about 2,000 copies, the marketing
potential for authors is obviously not high enough and so Roy Woodhead, Ph.D., CVS, TVM, PVM
we need to extend our reach to attract more authors and Vice Presidents - Education
spread awareness of our capabilities as a community of SAVE International
value managers.

VALUE ORLD Volume 29, Number 2, Fall 2006 1


Integrating Value Methodologies into
Product Development and
Project Management Processes
at Pratt & Whitney Canada
Alain LeBlanc, CD, P.Eng., MSc., MEng.

This article is reprinted from the SAVE International 2004 Confer- the design evolution process.
ence Proceedings. For the purpose of this paper, these methodologies are:
• value engineering
Abstract • function performance specification
Value methodologies have been integrated into the • quality function deployment
product development and project management processes • parametric paired comparison
for the past eight years at Pratt & Whitney Canada. Val- • direct value measurement
ue engineering, function performance specification, qual- • risk assessment.
ity function development, parametric paired comparison,
direct value measurement and risk assessment are among
some of these different methodologies that have found their Value Methodologies at P&WC
respective niche in the different phases of the product de- Value Engineering (VE). In 1996 a VE program was initi-
velopment. This has led to a significant increase in prod- ated at P&WC to tackle the pressing need to preserve the
uct value creation and project management effectiveness. cost margin on existing products and new products un-
This paper will describe how these methodologies were der development. This led to a solid cost reduction plan
integrated into the design organisation and its processes, in which VE played a key role in providing the means to
and that they are more effective when applied upon project develop alternative solutions (LeBlanc, 2000). This first
management's buy-in. wave of VE deployment led to the development of internal
expertise and the application of VE at the front end of the
new program introduction process, integrating the func-
Introduction tion analysis within the target costing process. The VE
The engineering organisation at Pratt & Whitney Can- methodology is defined under an internal Work Instruc-
ada is divided into three major groups, each responsible for tion Manual (WIM 448).
playing a key role in the introduction of new value prod- Function P erformance Specification (FPS). Function analysis
ucts. The design group is responsible for the conceptual has been applied to the development of a product perfor-
and definition activities associated with new products. The mance specification tool. Based on the environment method
development group is responsible for the testing and sub- (Brun, 1992), a function specification table was developed
stantiation of new products. The project engineering group in 1998 for the development of the engine control specifi-
manages the entire product development process with re- cation. Although similar to the technical FAST diagram, it
spect to interface with the customer, budget management, provides for a rigorous definition of the customer and stake-
product certification and ensuring an integrated design holders' needs. The FPS model (Report VE-1998-044) was
and development cycle. The value management expertise used on two new programs to challenge the relationships
resides primarily within the project engineering group, al- and the trade-offs between the engine control function and
though some experts are resident outside the Engineering other related product functions.
organisation due to personnel rotation and application of 9uality Function Deployment (ED). QFD uses a series of
the value methodologies in other areas of the business. Dif- matrices which focus on gathering, understanding and de-
ferent value methodologies have been adapted to the busi- ploying the "voice of the customer" throughout an organisa-
ness environment and their applications are a function of tion, ensuring that this "voice" is reflected throughout the

2 Volume 29, Number 2, Fall 2006 VALUE ORLD


development, production, installation, service and disposi- Value Methodologies Integration
tion of a product. The most common arrangement of these
matrices is known as the House of Quality, for its obvious
into the Concept Phase
The concept phase, also called the study phase, is the
shape (ASI, 1997).
first very important step. It defines the market we are aiming
QFD was introduced, in 1998, as a tool to prioritise the
for and is used to develop a good understanding of the cus-
product portfolio and technologies. Subsequently it was
tomer drivers, an analysis of the potential competition, the
used to validate the contractual scope for a new industrial
possible solution architectures, and, of course, the business
engine product in collaboration with the power plant in-
case drivers. The main deliverables of the study phase are:
tegrator and a risk-sharing partner. The next application
was for the recently launched PW610 engine program in • market definition
the general aviation market area. Its application has been • competitive analysis
limited to the first level of deployment, i.e. the House of • product specification
Quality. QFD was also used extensively for the purpose of • technology readiness
strategic sourcing and core competency analysis. • architecture study
Parametric Paired Comparison(PPC). PPC is a systematic • product target cost
tool used to provide weighted priorities by comparing mul- • project risk exposure
tiple attributes against each other (in pairs). It is a com- • business case study.
plementary process to the QFD and VE methodologies. A Direct value measurement (DVM), parametric paired
developed Pratt & Whitney tool (Otero, 1997) was used comparison (PPC), and quality function deployment (QFD)
extensively from 1997 to 1998. Lessons learned and an inno- are very effective at measuring and prioritising the customer
vative approach led to an internal Work Instruction Manu- values and transferring these values into the product char-
al (WIM 451) called "Parametric Paired Comparison". acteristics (Modafferi, 2003). DVM allowed us to quantify
Direct Value Measurement (DVM). DVM was applied in- specific product values through market surveys such as the
ternally to quantify market values through interventions at example described in Figure 1. In this case, the take-off dis-
key aerospace gatherings. It defines an intrinsic value for tance was assessed among others where potential custom-
a product or service independent of its cost (Cook, 1999). ers were asked to quantify the price increase they would
The market survey data allowed marketing and advanced pay for a substantial decrease in the aircraft take-off dis-
design organisations to validate perceived values against tance. This was instrumental for the market definition pro-
the customers' feedback and reinforce the QFD analysis. cess, which lead to the creation of several value curves in-
Risk Assessment. A risk assessment methodology was dicating the threshold points at which the customer would
adapted from a process developed at the United Tecimol- switch to the baseline proposition.
ogy Research Centre called "Risk-Reduced Development Understanding the relative importance of these values
Planning" (Zeidner, 1997) and available as an internal helped us in prioritising these values using the paramet-
Work Instruction Manual (WIM 447) called "Risk Assess- ric paired comparison (PPC). This methodology helped
ment Methodology". Risk assessment identifies the risk as- us determining value proposition for which the customer
sociated with each value proposition. One of the resulting would be differentiating our product from another. It is
impacts is the quantification of a more reliable product cost
risk exposure during the concept phase, its inclusion into
the cost targets and a more accurate product cost expecta-
tion curve as a function of the development cycle (Routhier ecalvallm -
PC4,:t1.7+7, 0'17 •ht° ,Iir1 ,17 , 77., 1 ,-17 1 .
and Delisle, 2001). .14 .e.,t UltrUtt
Finally, although not discussed in this paper, other
methodologies such as value stream mapping and Kaizen
have been used in support of achieving competitive excel-
lence in other business processes.
The paper will now describe how these methodologies 1+7C43 rt
have been integrated at P&WC as part of the new produc- 111111111111.1
tion introduction process. An overview of the product defi-
nition process and examples of value methodology applica-
tions are discussed.
Figure 1: Direct value measurement applied to take-off distance.

VALUE ORLD Volume 29, Number 2, Fall 2006 3


Attribute Minimum Significant Maximum Absolute Relative
Description Acceptable Improvement Differentiation A B C D E F G H Score Score
A Reliability al a2 al el fl gl a3 7 6
B Life b2 d3 e2 fl g2 b2 4 3
Installation
C . d3 e3 f2 g2 cl 1 1
Time
Development
D el dl gl d2 9 8
Time
Direct
E Operating el el e3 12 10
Cost
Acquisition
F fl f3 8 7
Cost
Maintenance
G g3 9 8
Time
Maintenance
H 0 1
Frequency

Figure 2: Parametric paired comparison methodology.


(1 = "Slightly More Important", 2 = "More Important, 3 = "Much More Important')
important to quantify the minimum acceptable, significant into the QFD first House of Quality. The advantage of the
improvement and maximum differentiation value levels for QFD was that it brought into a two dimensional framework
each of the key market attributes (Routhier, 2001). PPC is a vast amount of highly organised material to facilitate the
also playing a key role in getting the consensus within the direction of the concepts and the trade-offs. These results
Study team. The essence of the methodology is to compare were very effective in the preparation of the product defini-
attributes, two by two. This is done using the matrix in the tion requirement specification.
right hand of the spreadsheet, as illustrated in Figure 2. Once the former methodologies have been applied and
For example, considering attributes A (Reliability) and B concepts were derived, risk assessment came into play in as-
(Life or durability), the participants were asked to quantify sessing the risk exposure associated with the value proposi-
between a significant im.provement over the minimum ac- tions. When first introduced, quantifying the risk exposure
ceptable of A and a significant imp' rovement over the mini- was first seen as an obstacle to sell a project. It eventually
mum acceptable of B. A scale between one (slightly more became instrumental in selling the projects as the passport
process forced a more stringent discipline in the business
important) to 3 (most important) is used to score the dif-
case accuracy and confidence level of the cost targets. At
ferential between each attribute.
this stage, project scope and risk management were shown
Once completed, the PPC results can be incorporated
to be very critical for planning the right
Table 1: Risk Assessment Methodology— projects prior to offering to customers
Process Input and Output (Halpin, 2003; Faucher, 2003). The in-
put and output of the risk assessment
Process Input Process Output process are summarised in Table 1.
• Product (weight, performance, ...) • Risk summary known as "Risk The Risk assessment has been
• Key project characteristics (sched- Cube"
ule, budget)
forcing the study team lead by proj-
• Qualification of risks in terms of
• Design cross-section relative risk importance ect engineering to quantify the impact
• Engineering cost estimate • Top risks list with quantified im- of the value proposition (Plan A) and
• Technology level assessment pact on project and product values its probability to revert a lower value
• Risk historical checklist • Overall program risk exposure proposition (Plan B). The relationship
• Agreed severity matrix • Others
• Applicable likelihood dimensions
between the likelihood of failure and
• Trade off risk factors the consequence of not delivering the
• Contractual penalties value proposition is called the risk ex-
• Others posure. Figure 3 shows an extract of

4 Volume 29, Number 2, Fall 2006 VALUE ORLD


an application of the methodology. The interesting element • Entry-into-Service
of the methodology is that for each uncertainty the team The value methodologies are applied to support the
must derive the success vision and the alternative design
product design and project management teams for the
solution or fallback vision. In some instances reverting to
product optimisation and program decision-making. The
the fall back vision may turn out to be a more valuable solu-
function performance specification tool for engine controls
tion when considering all the impacts on the project.
was used at this stage to define and do the trade-offs be-
The main deliverables of this phase are:
tween the operability definition team and the other engine
• Component and System Specifications modules. Important trade-offs have been reported internal-
• Partnership Specifications for risk sharing venture ly for the recent new product definition programs. It was
• Supplier Selection demonstrated that it is critical to provide the design team
• Component Level Cost Targets with a mechanism that focus not only on the product design
• Technology Risk Mitigation Plan but also on the functions and the best way to achieve them
• Business Case Validation from a total system viewpoint. As the renowned French pi-
At this point, budgets are validated; partners and sup- lot said, "A designer knows he has achieved perfection not
pliers are confirmed. Successful completion of this review when there is nothing to add, but when there is nothing left
launches the detail design phase and effectively a 30-month to take away."
countdown to certification. The Full Design machine is en- PPC was shown to be a valuable decision-making tool
gaged with project management and design resources ap- for design alternatives and other project issues. When
plying the company's best practices, knowledge and skills conducting VE events it is imperative to build a team con-
in optimising the study concept and the subsequent pre- sensus at the beginning of the event. PPC has shown to be
detail design. very effective to achieve this as the team members are faced
The ultimate deliverables of this phase are: to challenge each other on their perceived values of the
• Bill of Materials Definition product. VE was very effective at understanding the cost
• Bill of Materials Characterisation structure of the product and generating alternative ideas.
• Product Certification Moreover, the VE methodology was also very effective at
• Production Plan bringing the suppliers onboard in discrete events where
Likelihood Matrix Severity Matri

Sever ity of Fa i lure


>,
Req u iremene ts

Ris k Leve l
Know le dg e of
Req u iremen ts

Li ke li hoo d of

CL) 73
KNow le dg e /

U
Comm itte d

Exp e r ience
Sta bi lity of
Resou rces

CU r ,cfY
Fa i lu re

-(13) 5 2' fn
L
'el) . ,t7)
Uncertainty Issue Success Vision L.) c' Y-i L' 0

togippl implymoup
Ability to meet Aerosmoothness Red e sign
Imp ro ba ble

Imp ro ba b le

Imp ro ba ble

Imp ro ba ble

Marg inal

Marg ina l
aerosmoothness requirements are tion fo man u factur-
requirements met

Accoustics External noise


Imp ro ba b ly

Very Li ke ly

Marg ina l
VerYLi ke ly

levels are met


with current .c7)
- - .-
design
2 0 0
I
Sig n i fican t I

Diameter of air Adequate cross- Modificat i on


S ig n i fican t

starter duct start capability cowl inner loft lines -irs


Me dium
Ala

c
with 2 in. versus to avoidin er e —
2.5 in. duct ()cation o
FT
(El
ge nerator 2

Figure 3: Risk Assessment Methodology—Extract from a Risk Assessment Event

VALUE ORLD Volume 29, Number 2, Fall 2006 5


Component Name Before VE After VE moves further, risk as-
sessment is applied to

D is tr ibu t ion

D is tr ibu t ion
production readiness,
Fu nct ion

Actuator

Funct ion

Funct ion
Cost pe r
Cos t per
Hous ing

Therma l
Num ber

customer support and

Bun dle

Cover
S lider
Tu be
other issues that form

Cost

Cost
Function Name part of the total pro-
1 Transfer Heat 313 X X X X 313 30% 256 31% gram deliverables.
2 Contain Fluid 110 100 X X 0 210 20% 188 23%
3 Modulate Fuel Temperature X X 82 83 X 165 16% 165 20% Conclusion
4 Locate/Position 33 69 56 7 0 165 This paper has
16% 110 13%
5 Respect Envelope 17 33 X X X 50 5% 17 2% described the extent
6 Resist Environment 0 33 X 10 X 43 4% 33 4%
of integration of the
7. Permit Maintainability
value methodologies
0 0 15 X 28 43 4% 18 2%
that has occurred at
8 Mix hot and cold fuel X X X 18 X 18 2% 13 2%
P&WC over the past
9 Interface 0 18 X X X 18 2% 11 1% eight years. Originally
10 Prevents Leaks/Seal 10 5 X X 0 15 1% 15 2% launched as part of a
11 Achieve Service Life 0 0 0 0 X 0 0% 0 0% cost reduction pro-
Total Cost per Component —4 483 258 153 118 28 826 gram, value engineer-
Before VE (%) 46% 25% 15% 11% 3% ing and the subsequent
After VE (%) 52% 19% 19% 10% 0.4%
other methodologies
were adapted to the
Figure 4: VE application at component design level. business and their ap-
plication moved from
both the design and manufacturing requirements can be cost reduction to tar-
challenged in a controlled manner. Major cost reductions get costing to cost and value management.
and product enhancement were achieved over the eight The value methodologies have been used extensively
years influenced by a value improvement embedded culture during the concept, pre-detail and detail design phases
within project engineering, design and other groups such were most suited to help the development teams to plan
as manufacturing. Figure 4 shows a typical component cost and define the most competitive value products. Signifi-
breakdown against its functions before and after the VE cant shifts in P&WC product value curves were achieved
event, conducted with a supplier. in part due to the effectiveness of these methodologies and
Risk assessment, as a pro-
cess, takes more importance
as the product development
moves forward being extend-
ed to Risk Management, as
risk response must be dem- 100
Uncertainty #1
RISK EXPOSU RE

onstrated. Risk mitigation 90 Eliminated


plans are defined for the top 60
Uncertainty #2
Eliminated — Baseline
product risk items, which are
monitored through risk wa- 40
• •, +3 Months
terfall charts, as shown in Fig- 20
ure 5 (Bharath, 2003). As the Uncertainty #3, 4, 5, 6, 7
product evolves further risk 0 Eliminated

is re-assessed on a regular ba- 2002 2003 2004 v


. 2005 2006 2007 2008
sis with a new perspective, as Bundled Risk
more knowledge is available Reduction Tasks TIMELINE
to validate the assumptions Figure 5:Risk waterfall chart example.
made at the previous stages. (Risk Exposure andN on Recurring Cost (NRE) with Time)
As the development schedule

6 Volume 29, Number 2, Fall 2006 VALUE ORLD


their integration in the new product development process. Halpin, John, Achieving Project Management Excellence—
The key lessons learned to successfully implement value Key to Value Management, Pratt & Whitney Canada,
methodologies are as follows: CSVA Annual Conference, Longueuil, 2003
• Their full effectiveness requires their integration with LeBlanc, Alain, Four Years of Value Engineering Experience
design and project management processes. at Pratt & Whitney Canada, Using Value Methodolo-
• The methodologies must be adapted to the organisa- gies from Cost Reduction to Cost Targeting to Cost &
tion culture and must continue to evolve as the require- Value Management, CSVA Annual Conference, Mon-
treal, 2000
ments develop.
• Their continued effectiveness over time depends upon Modafferi, Mario, Creating Product Value - Process, Meth-
the accountability of the process owner, the buy-in of odology and Results, Pratt & Whitney Canada, CSVA
Annual Conference, Longueuil, 2003
the process users and the support from management.
Otero, Joseph F., Real-time Integrated Computer tools for Value
Engineering Events (Value Management software tool-
Acknowledgement set). SAVE International conference proceedings, 1997
The author wants to recognise all those who contribut-
Routhier, Pierre, Parametric Paired Comparison: a New
ed to the implementation of the value methodologies over
Methods for Rapid and Accurate Customer Needs As-
the last eight years at Pratt & Whitney Canada. It includes
sessment, QFD Institute 13th Symposium on QFD, 2001
the value methodology champions, the VE team members,
the management, and the Valorex consulting teams, which Routhier, Pierre and Eric Delisle, Risk Management—Over-
was onboard during the first three years of the VE pro- view of Risk Theory and its Applications in VE, CSVA
gram. Annual Conference, Toronto, 2001
Zeidner L.E., Risk-Reduced Development Planning, United
Technology Research Center, 1997
Quoted References and
Communications Other References and Communications
Anonymous, Work Instruction Manual 447, Risk Assessment Cooper, Robin and Regine Slagmulder, Target Costing and
Methodology, Pratt & Whitney Canada, 1999, 13 pp. Value Engineering, Institute of Management Accoun-
Anonymous, Work Instruction Manual 448, Value Engineering tants, Productivity Press, Portland, ON, 1997, 387 pp.
Methodology, Pratt & Whitney Canada, 1999, 11 pp. Kaufman, Jerry J., Managing Value Management, SAVE In-
Anonymous, Work Instruction Manual 451, Parametric Paired ternational Accredited CVS Module 2 Training Mate-
Comparison, Pratt & Whitney Canada, 2003, 6 pp. rial, Montreal, 1999
Anonymous, VE-1998-044, Function Performance Specifica- LeBlanc, Alain, Value Methodologies and Target Costing,
tion, Pratt & Whitney Canada, 1998 CSVA Annual Conference, Toronto, 2001
Anonymous, quality Function Deployment for Products, Ameri- Prichard, Carl L., Risk Management, Concept and Guidance, second
can Supply Institute, Livonia, MI, 1997 edition, ESI International, Arlington, VA, 2001, 342 pp.
Bharath, Keppel, Risk Identification and Mitigation in Proj- Tahmazian, Berge, Quest for Value, American Value Jour-
ect Management Excellence, McGill University 5th Sup- nal, 1997
ply Chain Management Conference, Montreal, 2003 Thiry, Michel, Value Management Practice, Project Manage-
•Brun, Guy. Analyse fonctionnelle et le cahier des charges ment Institute, Sylva, NC, 1997, 112 pp.
fonctionnel. Training Material, Valorex, 1992 Transport Canada, "Railway Safety Management System
Cook, Harry E., Product Value and Target Costing, Depart- Guide" (TP13548, 2001).
ment of General Engineering, University of Illinois at Wideman, R. Max, 1992, Project and ProgramRisk Management,
Urbana-Champaign, CSVA Annual Conference, Mon- The PMBOK Handbook Series, Vol.6, Project manage-
treal, 1999 ment Institute
Faucher, Christian, Process Integrity and its Application Zeidner L.E., The Collaborative Innovative (CI) Process,
to Project Management Excellence, Pratt & Whitney United Technology Research Centre, CSVA Annual
Canada, CSVA Annual Conference, Longueuil, 2003 Conference, Montreal, 1999

VALUE ORLD Volume 29, Number 2, Fall 2006 7


Methodological and Practical Aspects of
Data Mining in the Product Development
Process
Prof Tai-Shen Huang

Abstract Keywords: Data mining, product design process, Industrial


It is well recognized in industry that product design Design
and development significantly impacts a company's busi-
ness success in competitive markets. Increasingly, business
enterprises are realizing how important it is to "know what 1. Introduction
they know" and to be able to make use of the vast amounts This research describes the structure of data mining as
of knowledge accumulated in recent years. Understanding it applies to the management of new product innovation
technologies of knowledge management, such as data ware- (see figure 1). One of the most important question that prod-
housing, data mining, as it applies to product innovation uct innovation is concerned with is in trying to determine
can give companies a competitive advantage. In particular, the sort of innovation that could result in a new product.
it is through data mining that the extraction of hidden pre- In order to answer that question, it is necessary to define
dictive information from large databases can identify valu- "newness" into commonly accepted categories as this will
able customers, predict future markets, enhance product influence the outcome of the product innovation inquiry.
innovation efficiency, and enable firms to
make knowledge-driven decisions.
Research into this area provides a data Defining problems
mining structure for discovering the in-
flow of resources that can be used to opti-
mize product innovation. Such structures Building databases
include various hi.dden knowledge. For
example, when the life-cycle of a product
Analyzing and searching
shortens, manufacturers and designers Inflow of resources
need to realize that a reduction of cost rules
is necessary to maintain a competitive
advantages. The purpose of data mining Strategic planning Building models
is to excavate information that will help
companies arrive at such conclusions. It
is imperative that effective methods are Applying data mining to
developed to further product design in- Concept generation
product innovation
novation.
This research focuses on building a
data mining structure that fits new prod- Strategic evaluation Evaluating models
uct innovation and design development
and adopts the decision-tree model to fa-
cilitate the prediction of trends. The ap- Technical development Data mining process
plication of the proposed structure and
system to a horticulture tools company
for the design of various tools is presented
Commercialization
to elaborate benefits of data mining for
product innovation.
Figure 1: Datamining process vs. product innovation process

8 Volume 29, Number 2, Fall 2006 VALUE ORLD


We can define "newness" into five accepted categories: builds a data mining structure that supports this step.
• New-to-the-world products: products that constitute • Concept generation: This step functions as the prob-
inventions. lem-solving phase that follows strategic planning. This
• New category entries: products that launch firms into is vital to product innovation as the most fruitful ide-
new markets. These products are not new to the world, ation involves identifying problems and suggesting so-
merely products from a market that was not previously lutions to the strategic planning.
available to the firm. • Strategic evaluation: Strategic evaluation is the step
• Additions to product lines: products that operate as where the ideas derived from concept generation are
line extensions, flankers, etc. evaluated. During this step, firms actually propose solu-
• Product improvement: products that have been im- tions to the competitive problem, followed by concept
proved. This is the most prevalent form of "newness" as tests to determine its effects on potential consumers.
virtually every product on the market today has been Strategic evaluation uses a scoring model of some type
improved, often many times. to aid corporations in deciding whether to undertake
• Repositioning: products that are retargeted for a new development.
user or application [3-4]. • Technical development: This step consists of invento-
The product innovation process consists of six steps: rying the firm's operations (manufacturing and other)
• Inflow of resources: This paper focuses on uncovering skills.
data regarding the inflow of new product resources. • Commercialization: Tradition211y, the term commer-
This includes customer, product, and market resources ciali7ation has described either the decision to market
(see figure 2). Data about product resources come from or the point when the firm decides to market a prod-
competitive activities, comparisons with competitors, uct [4] .
and investigations into competitive sources (custom- We propose that the data associated with product
ers' need) to find the core competitive opportunities. process innovations will become relatively more explicit
• Strategic planning: This step focuses on special op- through data mining. Data mining in the practice of a new
portunity analysis. The process of creatively recogniz- design method can enable the implementation of better in-
ing opportunities is called opportunity identification. novation ideas and concepts.
Product innovation opportunities that are ongoing, This application of data mining adopts a decision-tree
standing, and strategic must be checked. This research as its predictive model[17]. Each branch of the tree is a clas-
sified category with the representing the partitions of the
1---
Customer data within their classifications. Because decision-trees
score so highly on so many of the critical features of data
Competitive advantage source mining, they can be used in to explore and predict a wide
variety of business problems.

2. Data mining overview


"Mining" means to find something that already exists.
Product
Therefore, data mining can be defined as a process of identi-
ore competitive activity fying hidden patterns and relationships, and trends within
data. For our purposes, data mining is described function-
ally with the specific goal of uncovering relevant data for
business professionals [1]. The objective of data mining,
whether for business or otherwise, is to extract new infor-
C I : product-oriented database mation from large quantities of data. The benefit of data
C2: customer-oriented database
mining is that it can turn data into actionable results. In
particular, data mining makes possible the extraction of
C3: market-oriented database hidden predictive information from large database orga-
C : core competitive opportunity nizations for identifying valuable customers, predicting
future behaviors, and enabling firms to make knowledge-
Figure 2: The core competitive opportunity resources. driven decisions [16].

VALUE ORLD Volume 29, Number 2, Fall 2006 9


Data mining is a process that attempts to identify and using data mining, deciding the input and output for-
interpret patterns in information that are important for the mat, deciding cost effectiveness, and so on. In product
performance of certain tasks. This research integrates the innovation, there are three defined categories: custom-
inflow of data about new product process with new prod- er, product, and market.
uct innovation by using data mining technology. Under-
2) Building database: building database means the discov-
standing the characteristics of data mining associated with
ery of data dependencies. In the relational data model,
a specific transformation system will provide more strate-
the definition of the relations is about the relationship
gic choices for new product innovation. For example, suc-
among their attributes. These attributes come from four
cess in an increasingly competitive marketplace depends
steps: defining the problem, selecting the data, building
critically on the quality of organized knowledge which de-
the model, and selecting the models (See figure 4).
pends on the successful integration of data mining technol-
ogy with product innovation. 3) Analyzing and searching rules: data expressed in logi-
In industry, there are clear reasons for investing in data cal languages are distinguished by two main features:
mining technology. If data mining can provide actionable (I) They are readable and understandable. (2) They are
results that improve the product design process, then data excellent for representing crisp boxes and data element
mining becomes a competitive tool. Presently, data mining groupings.
is used to both increase revenues and to reduce costs. In- 4) Building models: include the development of models
novative organizations are using data mining to reconfigure (or the extraction of knowledge patterns). These are
product offerings for the purposes of increasing sales and to the major result of data mining. Equally important is
minimizing losses due to error or fraud. their presentation and visualization.
2.1. The data mining process 5) Applying data mining to product innovation: searching
Data mining is a process that uses a variety of data product data from database which consists of inflow
analysis tools to discover patterns and relationships in data of resources (the first step of the product innovation
in order to make valid predictions. The resources derived process).
from the data result from setting up a large data warehouse 6) Evaluating the model: estimating how well a particular
that integrates data from multiple sources, resolves data pattern meets the criteria of the data mining process.
problems, and loads the data into a database. Therefore, evaluating the model reflects whether the
Having examined the overall infrastructure of data strategic planning is in place. The cycle of evaluating
mining, we are now ready to take a closer look at data min- the model supplies feedback that support product in-
ing as a process. This novation.
research delineates Defining problems 4
six major stages in 2.2. Building database
the data mining pro- Data mining begins at the bottom of knowledge with
cess: defining prob- Building databases flow data. Some of the flow data must define the scope of
lems, building data our design problems. We select for data obtained through
bases, analyzing and queries or filters becomes they are applicable to the task.
searching for rules, Analyzing and
The selected data becomes the base for building an inte-
building models, ap- searching rules grated database. After building the database, the selected
plying data mining data must be analyzed to build models. In product design,
to product innova- the database needs many formal models in order to apply
tion, and evaluating Building models
data mining to new product innovation. Selecting models
models (see figure is the final step of the database process. If hidden data can
3):
Applying data mining be made explicit through visualization, it can be used to ef-
1) Define the prob- fectively improve product innovation.
to product innovation
lem: includes The four steps to building a database are as follows:
accurately de- A defining problems, selecting data, building models, and se-
scribing the Evaluating models lecting models (see figure 4).
problem to de-
1) Defining problems: includes accurately describing de-
termine the ap-
sign problems obtained from searching the inflow of
propriateness of Figure 3: Dating mining process.
product innovation data.

10 Volume 29, Number 2, Fall 2006 VALUE ORLD


Inflow of data
41--
Feedback

1. Defining problem
Purpose :
2. Selecting data Pattern
1. Decision making
-

3. Building models mined


2. Prediction
4. Selecting models

Building database

Figure 4: The relation of database to data mining.

2) Selecting data: determines the appropriateness of using ues of other variables of interest. The aim of prediction
data that will be integrated into the database. is to predict certain data values based on similar groups
3) Building models: includes researching rules and devel- of data.
oping models.
4) Selecting models: involves selecting the best model for 3. Decision tree
solving the problems. Selecting models is key to suc- The reality is that data mining should always be con-
cessful product innovation. sidered a process, and data interpretation is the heart of
Why is building a database so important? Because the this process. How should we interpret data? Because of
database is the base of data mining. Data mining is used to the complexity of data mining, many algorithms and tech-
uncover hidden data and to predict, and to support deci- niques have been developed. These algorithms include neu-
sion-making. We describe the relationship of a database to ral networks, fuzzy theory, and decision trees, etc [16].
data mining. This research chose the decision tree to organize prod-
uct data resources, because of its commonality and long-
1) Building a database: data mining may involve data from
standing utility in describing and classifying data sets in
multiple data sources, which may reside in a distribut-
the field of computer science. Decision-tree approaches are
ed database system, or in a tighter form, a multi-data-
good for handling classification problems. Classification is
base. The complexity involved in distributed database
the process of using historical data to build a model for the
systems has stimulated organization to find alternative
purpose of understanding and prediction. Many programs
ways to achieve decision support. In effect, the database
have been developed that perform automatic induction of
is an emerging approach for effective decision support.
a decision tree. The greatest benefit to decision-tree ap-
2) Visualization: data visualization graphically represents proaches is understandability.
the structure that exists among data sets.
3) Pattern mined: extracted patterns are the major result 3.1. The meaning of a decision tree
of data mining Equally important are their presenta- Decision-tree is a common knowledge representation
tion and visualization. used for classification. In classification, one is given data
from a specific instance, and the decision tree predicts,
4) Users: data mining should benefit human users. Al- based on the data, into which of two or more classes the in-
though data mining could be an automated process, stance belongs. Each instance contains data from multiple
human users can play a very important role in guiding attributes. Instances are collections of previously acquired
the mining process. However, data mining focuses on data which are sorted into class labels. It does so by deter-
finding human-interpretable patterns describing the mining which tests best divide the instances into separate
data. classes, forming a tree. This procedure can be conceived as
5) Prediction: it is very important to determine the major a greedy search through the space of all possible decision
factors that influence the prediction and the develop- trees by scanning through the instances in a given node to
ment of trends. Prediction involves using variables or determine the gain from each split and picking the single
fields in the database to predict unknown or future val- split that provides the greatest gain. Then the instances are

VALUE ORLD Volume 29, Number 2, Fall 2006 11


partitioned based on the split, and this procedure is applied prioritization tool (See figure 6).
recursively until all the instances in a node are of the same
The prediction of a decision tree is based on the data
class [16-20] .
derived from classified instances. Each instance contains
3.2. The two types of decision tree data from multiple attributes. The decision tree struc-
The two types of decision tree include: top-down deci- ture consists of a number of branches, one root, a number
sion tree and bottom-up decision tree. Top-down decision of nodes and a number of leaves. One branch is a chain of
tree starts at the abstract and general levels of the ontol- nodes from root to a leaf; and each node involves one attri-
ogy. One benefit of top-down decision tree is that general bute. An attribute in a tree provides the information about
objects can be defined to cover "what if" scenarios, even if the importance of the associated attributes.
they are not totally implemented at first. Bottom-up deci-
sion tree begins with more concrete objects and then moves
up to complete the design.
4. Product innovation decision-tree
In this paper, we propose the structure for combin-
ing a top-down and bottom-up decision-tree. In the first
step of the data-mining process, three resources are em-
ployed to define problems: customer, product, and market

4. Root
direction(see figures 2 and 3).
1) The direction of product indicates core competitive ac-
tivity. It divides into four departments: function, mate-
op Node
rial, technique, and form
0 Leaf 2) The direction of customer indicates sources of compet-
itive advantages. It includes three departments: cus-
XI X2 and X3 Attributes tomer needs, wants, and cycles.
CI C2 C3 and C4 Classes 3) The direction of market indicates competitors. It includes
five departments: market size and marketing, etc.
Figure 5: Top-down structure of decision. The second step of the data-mining process, building
the database, includes three categories: customer-oriented,
1) Top-down : strategic buckets method: this begins with product-oriented, and market-oriented databases.
the business's strategy and then moves to setting aside 1) Customer-oriented database: the combination of cus-
funds-envelopes or buckets of money destined for dif- tomer and product directions. The core problem is
ferent types of projects (See figure 5). whether customers are satisfied with the product.
2) Bottom-up : building strategic criteria into project se- 2) Product-oriented database: the combination of product
lection tools: here strategic fit is achieved simply by in- and market directions. The core problem is whether the
cluding numerous strategic criteria into the go/kill and
product fits the capability of market.

Root

Node

0 Leaf

X I X2 ... X4 Attributes

CI C2 C3 ... C5 Classes

Figure 6: Bottom-up structure of decision tree.

12 Volume 29, Number 2, Fall 2006 VALUE ORLD


Design purpose
eed

Customer Level
Cl
Fit

Product Level
F2
Needed place

Market Level ,Stop Stop/


T1
Functionl Function2 I Function3 Function4 I Function5
/
Extracting data
M1 1 M22
\ /
Figure 8: Detailed fimction decision-tree.
Figure 7: Product target decision-tree.
gestions (see figure 8).
3) Market-oriented database: the combination of market When developing product innovation, the research
and customer directions. The core problem is whether used integrated decision tree systems, which include bot-
the market could discover and achieve customer needs tom-up and top-down decision trees.
(see figure2). Making an integrated decision adopts a top-down de-
cision tree because it can hold the targets that support the
The third step of data mining process, analyzing and
development of new product innovation. Making detailed
searching rules, adopts a decision-tree. The fourth step of the
designs adopt a bottom-up decision tree because it can have
data mining process, building models, take uses top-down
potential activities that explore new product design.
and bottom-up decision tree models. The structure then, has
a top-down product targeting design (see figure 7) and a de-
tailed bottom-up design structures (see figure 8). 5. Case study
The product target decision-tree (see figure 7) needs We have implemented these data-mining techniques
tangible targets and so determining design purpose is the into an innovation design case and will illustrate their test
initiating step of the design project. The term is classified results. The case study comes from design companies using
into four levels: customer, product, market, and extracting WWW techniques. The identities of their clients are with-
data. Arriving at a decision for the design purpose begins at held at the client's request. These companies had conduct-
the customer level. If a customer level concern could fit the ed extensive market studies of customer experiences with
design purpose, then the project will go query product level various functionalities of related hand tools and electronic
concerns. The last level, extracting data, is to find whether a products.
search data is appropriate to supporting the design purpose.
The detailed function decision-tree creates the plat- 5.1 Case study: horticulture tools
form for collecting various data. Choosing different models, This shape design system provides comprehensive, en-
data-mining through visualization, will show different sug- terprise-level, design solutions and professional services.

HTML
HTTP (HTML) Case-Based
Designer ITS Server SQL Server Database

HTTP (ASP)
*.ASP, ASP.DLL

Figure 9: The structure of communication between designer and the system.

VALUE ORLD Volume 29, Number 2, Fall 2006 13


-

,VTO 12 - B 174
EB_I
-..11tia#.9.i ;Z-.1.1.11140:0519:Fit
Ilortictilture cutti, H.ndhold torrnnfrvo xiIiazy design sy,tr•rn
[
A , B C D ' E
1 AA DI @Sketch D1 @Sketch D1 @PRI D1 @I'M
2 al 1 60' 30i 30 120
3 o2 1 50! 251 251 100
5 :43
, 40 I 201 20 80
301 15 ! 15 1 60
6 a5 I 20i 101 10 40
7 i i
Figure 10: The system home page.
Figure 11: Feature data for product shape design.

mm

feEkld.Fig

1;Sig■lhOF

r- r

It)

•:1.7_,1.1.4101111016440;W:

Figure 12: Design options. Figure 13: Fundamental shapes.

Figure 14: Case-based shape database. Figure 15: Design variable parameters.

The decision series suite of knowledge tools solves design to less-than-optimal product demand forecasting. The de-
challenges in a variety of markets, including product trend, signer was perplexed with product innovation design and
shape design, and future development. The decision series development, and could not advance either design or devel-
suite includes pattern discovery tools based on the grey opment because of a lack of design knowledge for finalizing
prediction theory, clustering, and association rules. The a design strategy.
communication structure between designer and shape de-
sign system is shown in Figure 9. 5.3 The solution
A solution, that combined elements of clustering and
5.2 The problem grey prediction theory, which enabled the designer to au-
A designer found that product value was suffering due tomatically review a product's history and equate feature

14 Volume 29, Number 2, Fall 2006


VALUE ORLD
data groupings to design patterns, has been designed and 6.1. Tools
implemented in the shape design system. Management is The tool used in this research is the logical approach,
now able to explore the lowest level of detail and forecast which can be understood as a set of rules, whereas a neu-
development requirements for creating an innovative de- ral network is obscure, with weights that have no intuitive
sign. In addition, the designer receives exacted data which meaning. The model is easy to explain to a design user be-
enables them to continuously fine-tune replenishment sys- cause it applies background domain knowledge.
tems by combining feature design data with fundamental
6.2. Userability
shape basis.
The performance of the logical approach will be faster
5.4 The results than the neural network; the risk of missing code is lower
The shape design system is able to forecast product de- than neural network too. It is a very easy method for creat-
ing a models.
sign trends at the beginning design stage. Meanwhile, the
system can cluster similar items into actionable groups to 6.3. Properties of model
streamline the design process. Design company can now Although neural networks provide more accurate mod-
predict demand for product development trends and more els, it has problems with over-fitting and finding sub-opti-
effectively create a good product shape.. System operation mal solutions. The Logical approach needs more data for a
steps, including user interface, feature data, case-based pre- model to ensure that there is critical mass in the leaf nodes
diction database, shape database, and are shown in Figure. following the many branches. Neural networks require that
10 to Figure. 15. the data be transformed into a binary format.
For the construction of a model, the logical approach
is quicker and easier than neural networks because neural
6. Logical (Decision Trees) vs. networks have many parameters that must be set and re-
Equational (Neural Nets) quire more skilled manipulation to ensure over-fitting does
In earlier approaches to data mining, the data set was not occur. So, the cost of building a neural network model
maintained for future pattern matching. Approaches based is higher than the logical approach, and it requires more
on pattern distillation fall into three categories: logical, time and building skills.
crosstabulation, and equational. These technologies ex-
tract patterns from a data set and then use the patterns for 7. Conclusion
various purposes. The logical approach deals with both nu- Supporting design innovation processes with tecimol-
meric and non-numeric data. Equations require all data to ogy and methods from the field of knowledge management
be numeric, while crosstabulations work only on non-nu- can have a beneficial effect both on product and financial
meric data. Table 1 summarizes the pros and cons of these development. The new knowledge made available by data
categories. mining can lead to more competitive products. To process
Approach Pros Cons
Logical Work well with multidimensional and OLAP Unable to work with smooth surfaces that
data typically occur in nature
Able to deal with numeric and nonnumeric
data in a uniform manner
Cross-tabulation Simple to use with small number of non- Not scalable
numeric values
Equational Works well on large sets of data Ability to handle numeric values
Works well with complex multi-dimen- Ability to handle conjunctions
sional models
Ability to approximate smooth surfaces Require all data to be numeric (nonnu-
meric must be coded)
System can quickly become a "blackbox"

Table 1: Pros and cons to data mining approaches.

VALUE ORLD Volume 29, Number 2, Fall 2006 15


this information we found that the integrated decision tree ference on engineering design, Glasgow, UK, pp.13-20
systems, which include bottom-up and top-down deci-
[10]V Hubka and W E Eder, 2001, functions revisited,
sion tree are most applicable to product innovation. New
13th international conference on engineering design,
product innovative development adopts a top-down deci-
Glasgow, UK, pp.6 9-76
sion tree because it can hold the target. Making detailed
designs adopts a bottom-up decision tree because it can [11] C T Hansen, 2001, verification of a new model of de-
analyze new product design. Finally, tools and experience cision-making in design Decision-making in design,
from knowledge management and best business practices 13th international conference on engineering design,
are vital in the assessment of designs in relation to business Glasgow, UK, pp.101-108
goals. [12]Motokazu Orihata and Chihiro Watanabe, 1999, the
Knowledge management in design will support deci- interaction between product concept and institutional
sion-making with broader, more transparent and acces- inducement: a new driver of product innovation, Tech-
sible knowledge bases, and organize data in generally rec- novation Vol. 20, pp.11-23
ognized and widely used decision-making models such as
the Integrated Decision Tree model. It will also provide an [13]Raghavan Parthasarthy and Jan Hammond, 2001, Prod-
objective and quantifiable foundation for trend-analysis uct innovation input and outcome: moderating effects
and market predictions. Such technology, constructs, and of the innovation process, Journal of Engineering and
models will greatly improve design, product, and design- Technology Management, Vol.19, pp.75-91
related business decisions. [14] John H. Friar, 1995, Competitive advantage through
product performance innovation in a competitive mar-
ket, the Journal of product innovation management,
Acknowledgements Vol. 12, pp.33-42
This research was partially supported by Chaoyang
University of Technology in Taiwan under the project [15]Udo-Ernst Haner, 2002, Innovation quality-a concep-
number CYUT93-D-003 tual framework, International journal of product eco-
nomics, Vol.80, pp.31-37
[16]Chris Rygielski, Jyun-Cheng Wang, David C. Yen, 2002,
References
Data mining techniques for customer relationship man-
[1] Robert Groth, 2000, Data Mining, Hall PTR, New Jer-
sey agemet, Technology in Society, Vol. 24, pp.483-502
[17]Chris Clifton, Bhavani Thuraisingham, 2001, Emerging
Zhengxin Chen, 2001, Data Mining and Uncertain Rea-
standards for data mining, Computer standards and in-
soning, Wiley Inter-science, Canada
terfaces, Vol. 23, pp.187-193
Robert G. Cooper, 2001, Winning at New Product, Per-
seus publishing, New York [18]Jules J. Berman, 2002, Confidentiality issues for medi-
cal data miners, Artficial Intelligence in Medicine, Vol.
C. Merle Crawford, 1996, New Products Management, 26, pp.25-36
Mc Graw Hill, America
[19]Helen M. Moshkovich, Alexander I. Mechitove, and
Paul Belliveau, Abble Griffin, and Stephen Somermeyer, David L. Olson, 2002, Rule induction in data mining:
2002, the PDMA Toolbook for New Product Develop- effect of ordinal scales, Expert Systems with Applica-
ment, Wiley Inter-science, Canada tions, Vo122, pp.303-311
[6] Vijay Atluri and John Hale, 2000, Research Advances
[20]Nianyi Chen, Dongping Daniel Zhu, and Wenhua
in Database and Information Systems Security, Kluwer Wang, 2000, Intelligent materials processing by hyper-
Academic, America
space data mining, Vol. 13, pp.527-532
[7] Ranjit K. Roy, 2001, Design of Experiments Using the
[21]Jeffrey P. Bradford and Jose A. B. Fortes, 2001, Charac-
TAGUCHI Approach,Wiley Inter-science, Canada terization and Parallefization of decision-tree induc-
[8] John M. Usher, Utpal Roy, and Hamid R. Parsaei, 1998, tion, Journal of Parallel and Distributed Computing ,
Integrated Product and Process Development, Methods Vol. 61, pp.322-349
tools and Technologies, Wiley Inter-science, Canada
[22]Zhou Yuanhui, Lu Yuchang, and Shi Chunyi, 1997,
[9] Horvdth, 2001, A contemporary survey of scientific re- Mining classification rules in multistrategy learning
search into engineering design, 13th international con- approach, Intelligent Data Analysis, Vol.2 pp.165-185

16 Volume 29, Number 2, Fall 2006


VALUE ORLD

[23]S. Lavington, N. Dewhurst, E. Wilkins, and A. Freitas, W.L. Ip, 2002, An innovative scheme for product and
1999, Interfacing knowledge discovery algorithms to process design, Journal of Materials Processing Tech-
large database management systems, Information and
nology, Vol.123, pp.85-92
Software Technology, Vol. 41, pp. 605-617
[24]M. A. Friedl and C. E. Brodley, 1997, Decision tree clas- [36] Leslie Monplaisir, 1999, An integrated CSCW architec-
sification of land cover from remotely sensed data, Re- ture for integrated product/ process design and devel-
mote Sensing of Environment, Vol. 61, pp. 399-409 opment, Robotics and Computer-Integrated Manufac-
[25]Cristina Olaru and Louis Wehenkel, 2003, A complete turing, Vol.15, pp.145-153
fuzzy decision tree technique, Fuzzy Sets and Systems, [37]Ting-Kuo Peng and Amy J.C. Trappey, 1997, A step to-
Vol. 138, pp.221-254
ward STEP-compatible engineering data management:
[26]Richard Nock and Pascal Jappy, 1999, Decision tree the data models of product structure and engineering
based induction of decision lists, Intelligent Data Anal-
changes, Robotics and Computer-Integrated Manufac-
ysis, Vol. 3, pp. 227-240
turing, Vol.14, pp.89-109
[27]L. M. Bartlett and J. D. Andrews, 2000, An ordering
heuristic to develop the binary decision diagram based [38]X.F. Zha and H. Du, 2002, A PDES/STEP-based model
on structural importance, Reliability Engineering and and system for concurrent integrated design and as-
System Safety, Vol.72, pp. 31-38 sembly planning, Computer-Aided Design, Vol.34,
[28]Han-Pang Huang and Chao-Chiun Liang, 2002, Strat- pp.1087-1110
egy-based decision making of a soccer robot system us-
[39]Philippe Girard and Guy Doumeingts, 2003, Modeling
ing a real-time self-organizing fuzzy decision tree, Vol.
127,pp. 49-64 the engineering design system to improve performance,
Computers and Industrial Engineering, Vol.46, pp.43-
[29]Pi-Sheng Deng, 1996, Using case-based reasoning ap-
67
proach to the support of ill-structured decisions, Euo-
pean Journal of Operational Research, Vol. 93, pp. 511- [40]Roger J. Calantone, Shawnee K. Vickery and Cornelia
521 Droge, 1995, Business performance and strategic new
[30]Kweku-Muata and Osei-Bryson, 2004, Evaluation of product development activities: an empirical investiga-
decision trees: a multi-criteria approach, Computers tion, The Journal of Product Innovation Management,
and Operations Research, Vol. 31, pp.1933-1945
Vol.12, pp.214-223
[31]Rattikorn Hewett and John Leuchner, 2002, Restruc-
turing decision tables for elucidation of knowledge, [41]David D. Ardayfio, 2000, Principles and practices of de-
Data and Knowledge Engineering, Vol. 46, pp. 271-290 sign innovation, Technological Forecasting and Social
Change, Vol.64, pp.155-169
[32]Pedro Domingos, 1997, Knowledge discovery via mul-
tiple models, Intelligent Data Analysis, Vol.2, pp.187- [42] Information Discovery, Inc. A characterization of
202 data mining technologies and processes: an Infor-
[33]Arun Sen, 2002, Metadata management: past, present mation Discovery, Inc. White Paper. http://www.
and future, Decision Support System, Vol.37, pp.151-173 dmreview.com/portal —ros.cfm?NavID-92&White-
[34]Fujun Wang, John J. Mills and Venkat Devarajan, 2001, Paper-6 0 &PortalID-9
A conceptual approach managing design resource,
Computers in Industry, Vol.47, pp.169-183 Professor Tai Shen Huang teaches at the Guaduate Institute of De-
-

[35] Henry C.W. Lau, Bing Jiang, Felix T.S. Chan and Ralph sign, Chaoyang University Technology in Taiwan.

It's not a REAL value study without a CERTIFIED VALUE SPECIALIST.


• For more information on finding a Certified Value Specialist or becoming certified, contact SAVE International. SAVE
• International is the only organization renowned throughout the world as the value industry standard for certification.
• 136 South Keowee Street, Dayton, OH 45402 USA • T 937.224.7283 • info@value-eng.org • www.value-eng.org

VALUE WORLD Volume 29, Number 2, Fall 2006 17


Value Driven Design
Paul D. Collopy, Ph.D. and Joseph F. Otero, Jr., CVS-Life

Abstract sible airplane? Optiinization literally means choosing the


In October of 2005, the Value-Driven Design Program best. When optimization is available, all other design pro-
Committee held a one-day workshop in Hartford, Con- cesses are inferior.
necticut, sponsored by Pratt & Whitney, to explore the Optimization has three essential parts:
application of Value-Driven Design to a commercial aero- 1) A definition of better;
space product. The trial application was a supersonic busi- 2) Generation of alternatives; and,
ness jet.
Value-Driven Design is a tool to enable design optimi- 3) Evaluation of pairs of alternatives to determine which
is better.
zation of large systems, such as aircraft, by the formulation
of system and component value models. In the workshop, a By successive comparison of alternatives, we can find
team of engineers primarily from the American Institute ofwhich is best. At its core, optimization is that simple.
Aeronautics and Astronautics and SAVE International for- Optimization is a very refined techni.que that can in-
mulated a value model for the aircraft and derived a value clude far more than these essentials. For example, in engi-
model for a critical component, the supersonic jet engine.neering design, better is defined in terms of attributes such
This paper describes the workshop and what we learned as energy efficiency, reliability, and manufacturing cost.
from the experience. It introduces the Value-Driven DesignAlternatives are created through imagination, skill and ex-
(VDD) process and explains the benefits of VDD. In re- perience. The design optimization process is shown in Fig-
ure 1 (below left). Creation of alternatives is the upper right
counting the workshop, the authors describe what worked
arc. Initially, alternatives are rough descriptions in terms
well, and what lessons were learned of value to future users
of a few parameters. For example, a wing design might be
of VDD.
described by the span, chord, taper, sweep and thickness.
In the lower left arc, this description is elaborated into, for
Optimization example, a computer-aided design model of the wing, with
Why would we ever design an airplane that simply the entire outer surface and inner structure modeled. The
meets requirements, rather than designing the best pos- lower right arc assesses this detailed configuration to es-
timate the attributes of the wing:
Value lift, drag, weight, cost, and so on.
Formal training in engineering
Search school is almost entirely focused
on how to do the analyses in the
lower left arc.
Finally, the upper left arc
evaluates the attributes so that
Attributes Design Cycle Parametem this particular design can be com-
pared to others. Making such
comparisons is equivalent to plac-
ing the designs in order, or, math-
ematically, mapping them into an
Analysis Definition t`.< ordered set. Real numbers are the
Including Mass complete ordered field, so evalu-
Performance, and ation, in practice, always entails
Cost Estimation mapping a set of attributes onto
Configuration
the real numbers. In Value-Driven
F igure 1: Design optimization process. Design, we call the map from at-
tributes to real numbers a value

18 Volume 29, Number 2, Fall 2006 VALUE ORLD


model. You can think of a value model as a scoring process the shortfall, or the program is canceled.
where better designs receive higher scores. Value-Driven Design (VDD) aims to achieve the best
Optimal design has been seriously studied by a large design by moving away from requirements and working in-
academic community for decades. The discipline has made stead from a value model for every system, subsystem, and
great strides forward, and increases in computer capability component, to enable universal application of optimal de-
have enabled practical application of powerful optimization sign. The key to VDD is the value model, in particular, value
techniques. However, most design tasks do not use optimi- models for subsystems and components. The mathematical
zation. Why do engineering teams neglect the opportunity derivation of component value models is presented in Col-
to design the best possible system or component? lopy (2004' To understand the issues, it is worthwhile to
Mainly, it is because today's systems engineering pro- review the basis for requirements-based design.
cess does not provide a value model. When optimization Complex systems cannot be designed in a single act
is done, it uses a single attribute as the score, yielding the by a single person—they are too complicated. Therefore,
lowest weight or lowest cost design, invariably not what is a strategy of divide and conquer is employed (Dykstra,
desired. Within the standard requirements-based process 1965).2 The system in partitioned into a hierarchy of sub-
there is no effective way to distinguish which of two de- systems and components. Given a set of requirements for
signs is better. Requirements give only a crude measure: if the system, requirements are chosen for the components
Design A meets the requirements, and Design B does not, A such that, if every subsystem and component meets its re-
is better. However, requirements provide no guidance for quirements, the entire system, when integrated, will meet
choosing among designs when all meet the requirements. its requirements.
If an engineer can sacrifice reliability to reduce weight or Analogously, Value-Driven Design works from a sys-
increase durability at a higher weight, requirements pro- tem value model to create component value models, such
vide no guidance. Also, it is not unusual for an engineer to that, if each component is the very best design possible, ac-
determine that there is no way for a component to meet all cording to its value model, then the system will be the very
its requirements. Perhaps it can meet the efficiency require- best design according to its own value model.
ment, but exceed the weight limit; or meet the weight limit The basic steps in VDD are:
and miss the required efficiency. Requirements provide no
indication as to which alternative is better. 1) Development of a system value model.
Cost as an Independent Variable (CAIV) is a newer 2) Derivation of subsystem and component value models
process that gives more guidance. CAIV says the best de- from the system value model.
sign is one that meets all requirements at the lowest pos- 3) Optimal design of all components and all subsystems.
sible cost. As long as at least one design alternative meets
requirements, CAIV can lead to selection of a best design, 4) Integration of the parts into the full system.
where best is all about cost. The Hartford workshop was held to work through
Both traditional requirements-based design and CAIV steps (1) and (2), and to get a feel for their difficulty and
can lead to absurd choices. Imagine an aircraft with a re- practicality. We also learned useful lessons about how to
quirement of 2500 mile range and 5000 hours mean time apply VDD to actual projects.
between failure. Consider two alternatives: Design A has
7000 miles of range and 4999 hours between failures. De-
sign B has 2501 miles of range and 5001 hours between fail- Details Of How A Value Model Is
ures. In all other measures, the designs are identical. Re- Created
quirements-based design and CAIV will both pick B over At the October, 2005, meeting, the team developed a
A, although A is obviously the better design. value model for a supersonic business jet and derived from
The way we execute such designs in practice is to let the value model an objective function to optimize engine
the design proceed for a time until the system is perceived design for the hypothetical aircraft. The point of this exer-
to be too heavy, too unreliable, or whatever. At that point, cise was to show that, using Value-Driven Design, once the
great consternation is expressed over weight or whichever value model for the top level system is established, consis-
attribute exceeded requirements, and the design is redone tent objective functions can be derived for all components
with everyone admonished to do better about weight or of the system. These objective functions largely replace
other unsatisfactory measure. The whole process can be re- requirement specifications as guidance for component de-
peated again for another attribute. Eventually, the system signers. The trial application was chosen to avoid inference
meets requirements, or the customer becomes resigned to with any potential products then under consideration by

ORLD Volume 29, Number 2, Fall 2006 19


Pratt & Whitney, a designer and manufac- U.S. CMI and Jcint Use Airperts
.uoliMJrpera
with Highest Frequencyof GA Aircraft
turer of turbine engines, rocket engines, and Maximum WeigFrt Re5trictiora
10 OIL .44rpor Is
turbine engine services. 350000

The following steps were taken. Each AIRPOIrrs


AVAILABLE
will be discussed shortly in some detail. ap0000
to ihe OSILI
1) The workshop began with a review of
background material on supersonic
business jets (SSBJ5). 4
* •
—41-4 1.4*
2) Then, the basic structure of the value ,4 4 4, 4-

model was decided on. 1 4.ren,


144

3) Next, stakeholders were identified.


4) That made it possible to select the key
attributes of the SSBJ design.
5) Then the team completed the SSBJ value
model.
6) Then key attributes of the engine design 2E00 4000 soto ono 1 00G0 120E0 14000 16030
Langti crt Longa it Runwa y (113 et)
were next identified.
7) Relationships between engine at- Figure 2: Gross weight and takeoff field length limitations.
tributes and aircraft attributes were
worked out.
Total Mission Range with Subsonic Portion
8) A spreadsheet was put together High Boom, 6 pax, Full Fuel, ISA, No Wind
5500
with all this information, and the
objective function for engine de- 5250
sign was derived.
55 % Survey Respondents Require 5000 nm
5000
Tota l Miss ion Rang e ( nm)

Step 1. Review Back- 4750


ground On The Super-
4500
sonic Business Jet 38 % Survey Respondents:
We reviewed a Raytheon Study, Require 4500 rim,
4250
sponsored by NASA, on the feasibil- 1

ity of a supersonic business jet (SSBJ). 4000 - - - - - - - - - -


16 % Survey Respondents "."-*-4
Two of the most important elements of I
Require 4000 nm
-4...

3750 1
the study are shown below. The first 1
shows how gross weight and take-
3500
off field length can limit the number 0 500 1000 1500 2000 2500 3000 3500 4000
of airports available to a business jet. Subsonic Range of Mission (nm)
(See Figure 2, above.) The second chart
Figure 3: Results of surveys concerning importance of aircraft range.
shows results of surveys of potential
SSBJ customers concerning
the importance of aircraft
Goodness = NPV (Production Profit) + Societal Good - Developmental Cost
range. (See Figure 3, midle.)
Production Profit = Market Share X Market Size X (Price - Manufacturing Cost)

Price + NPVC (Customer Profit) + Charm - Competition

Customer Profit + Utilization X (Value per Flight - Cost per Flight)

Figure 4: Value model based on profit to the manufacturer.

20 Volume 29, Number 2, Fall 2006 VALUE ORLD


Step 2; Determine Val- the design: O Passengers
O Customer company
ue Model Structure • Flight crew
• Customer corporations • Utilization
A straightforward value model
• Passengers O Customer company
might be the following, where value
• Leasing company O Passengers
is based on profit to the manufactur-
• Key acquisition decision makers O Leasing company
er, adjusted for social considerations
• Insurer O Fuel Suppliers
external to the market. (See Figure 4,
• Key suppliers O FAA Rules
below, previous page.)
• Fuel suppliers O Maintainers
Surplus Value theory provides a
• ATC system O Airport Operations
simpler formulation that will place
O Defense Department Opera-
the same ordering on designs and was • Airport Ops
tions
selected to keep the exercise within • Regulatory Agencies • Cost per Flight
the two-day time constraint of the O FAA O Fuel suppliers
workshop. O ICAO O Airport Operations
Surplus Value = Disc o X • Local Restrictions O Flight Crew
Market Size X (Disc c X Uti- • Congress O Maintainers
lization X [Value per Flight • Maintainence O Customer company
- Cost per Flight] + Charm • Defense Department Applica- O Insurer
+ Societal Good per Unit - tions
Manufacturing Cost) - De-
• Societal Good
• ITAR O Defense Department Opera-
velopment Cost
tions
Two of these variables are param- O Regulatory Agencies
eters which we treated as constants:
Step 4. Select The Key O Congress
Disc = Disc = 8. Attributes Of The SSBJ O Customer company
This leaves eight variables, which Design. With this, we were ready to iden-
act as collectors for the relevant ef- Next, we listed the stakeholders tify the SSBJ design attributes. We
fects of the aircraft attributes: under the collectors from the value started with a long list inspired by the
• Market Size model structure. This, we hoped, categorization of stakeholders above:
• Utilization would help us find the important SSBJ
• • Market Size
• Value per Flight attributes.
O Size
• Cost per Flight • Market Size Payload
• Charm O Aircraft manufacturing com- - Total Weight
• Societal Good pany - Cabin Diameter
• Manufacturing Cost O Customer company - Vehicle Length
• Development Cost O Leasing company - Cabin Length
We anticipated that the last two, O Passengers - Amenities
Manufacturing Cost and Develop- O ITAR - Number of Passen-
ment Cost, would be SSBJ attributes. O Defense Department Applica- gers
The other six would be calculated tions O Gross Weight
from SSBJ attributes. O FAA O Power
O Airport Operations O Cooling
O ATC Systems O Range
Step 3. Identify Stake- O Financiers O Total Op. Cost
holders In The SSBJ • Value per Flight O Block Time
Design O Passengers O Aircraft survivability equip-
ment
Preparatory to identifying the O Customer company
O Total Operating Cost
most important attributes of the SSBJ O Defense Department Applica-
• Flight Crew
design, we brainstormed a list of the tions
• Fuel
stakeholders who would care about • Charm

VALUE ORLD Volume 29, Number 2, Fall 2006 21


• Maintenance the list was to expedite the process, Cost per Flight = (O&M
• Insurance thereby keeping it within the time of Cost X Block Time + $3 X
• Landing Fees the workshop): Gross Weight) ~ $1,000
• Storage
• Manufacturing Cost Charm = $5 million -
O Support Footprint
• Development Cost Airport Noise
• Cost per Flight
O Fuel Burn • Gross Weight
Societal Good = $8 million
O Block Time • Field Length
- Airport Noise
O Flight Crew Size • Time to Climb
O Maintenance cost • Dispatch Reliability • Under the baseline assumptions,
O Insurance Rate • Noise the SSBJ was a big money loser, which
O Gross Weight • Range was not good because then the opti-
O Landing Fees • O&M Cost mal design was an aircraft that flew as
O Storage Cost • Speed little as possible. Therefore, we arbi-
• Societal Good • Cabin Diameter trarily added $25 thousand per flight
O Emissions • Payload Weight to Value per Flight to move the opera-
O Noise We later split Noise into Boom tion into the black. This would not be
• Utilization Noise and Airport Noise. necessary in the design of a real, plau-
O Dispatch Reliability sible system.
O Maintainability
• MTBF/MTTR (Mean Step 5. Complete The
Time Between Failures / SSI3J Value Model Step 6. Identify Key
Mean Time Between Re- Building the value model is now Attributes Of The SSI3J
placements) a matter of calculating the eight vari-
O Payload Utilization Engine Design
ables in the value model structure as The purpose of this step was to il-
O Turn around time
a function of the thirteen attributes. lustrate that the value model of a com-
O Peculiar support equipment
Two of the calculations are trivial, be- plex subsystem could be derived from
O All Weather Capacity
cause Manufacturing Cost and Devel- the information in the overall value
• Value per Flight
O Climb rate opment Cost are both attributes. We model. The intent was to demonstrate
O Block Time worked out two intermediate vari- the viability and simplicity of develop-
O Payload Utilization ables which were useful in calculating ing a value model for each component
• Number of passengers the rest. of a complex system.
• Weights Block Time = (Range ~ The significant engine design at-
O Range 8,000 nm) 2 + (1 - Gross tributes were chosen to be:
O Gross weight Weight ~ 180,000 lbs.) + • Takeoff Thrust Perf
O Take off field length (1 - Field Length/10,000 • Climb Thrust Perf
• Charm Feet)
• Weight (engine)
O Speed
O Range The six collectors are: • Specific Fuel Consumption (fuel
O Amenities consumption rate of the engines)
Market Size - 250 Aircraft • Reliability (engine)
O Cabin Diameter
per Year X Access Factor • Landing / TakeoffNoise
O Luxury Outfitting
O Survivability Utilization = 200 Flights • Manufacturing Cost (engine)
O Cabin Noise per Year X (1 - 2 X (1 - • Maintenance Cost (engine)
O Safety Dispatch Reliability)) • Development Cost (engine)
• Manufacturing Cost Emissions did not make the list
Value per Flight = Access
• Development Cost because emissions was not in the top
Factor X (1 e - e -
From this list, the team picked 10 payload Weight 0.25 thousand lbs.) 13 SSBJ attributes, so there was no
to 12 attributes that seemed most im- $12,000 per hour X (6 way for emissions to impact the value
portant (this artificial shortening of - Block Time) model.

22 Volume 29, Number 2, Fall 2006 VALUE ORLD


Step 7. Work Out Relations Of En- products of each partial derivative with the corresponding
attribute, plus an offset. This is a first degree (linear) Tay-
gine Attributes To Aircraft Attri-
lor series approximation to the overall functional relation-
butes ship between the engine attributes and the ultimate score,
The next step was to express the SSBJ attributes in surplus value. When such linear objective functions are de-
terms of the engine attributes. rived for all components of a large system, each component
Manufacturing Cost = $20 million + 2 X can be optimized with the assurance that, when the com-
Manufacturing Cost (engine) ponents are integrated, the overall system will be optimal.

Development Cost + $3,000 million + Potential Uses of Value-Driven


Development Cost (engine)
Design
Gross Weight = 18,000 lbs. ~ (0.55 - 0.44 X Value-Driven Design (VDD) was originally developed
SFC ~ 0.55) as a design tool, to enable the widespread application of
optimization to the design of complex systems. However,
Field Length = 2,250 ft. X 160,000 lbs. ~ (2 X
once design engineers begin to think in terms of value,
Takeoff Thrust Performance)
much wider opportunities appear.
Time to Climb = 20 minutes X (20,000 lbf
(2 X Climb Thrust Performance)) Trade Studies
The most obvious extension of VDD, if it is an exten-
Dispatch Reliability = 99.5% X Reliability sion at all, is to use the system value model or derivative
(engine) subsystem models to evaluate alternatives to trade studies.
Boom Noise = 0.4 This is very similar to the way most trade studies are done
today except that the value model provides evaluation cri-
Airport Noise = -2 + Landing/Takeoff Noise teria that are consistent across the project and (if the sys-
tem value model is well done) resonant with the needs of
Range = 5,000 nm X 0.55 ~ SFC
the customer and the business.
O&M Cost = #3,600 X SFC + Maintenance In another sense, VDD extends system engineering
Cost (engine) + $400 trade study methodology into design engineering: every de-
sign decision is treated as a mini-trade study with a design
Speed = 1.8 Mach Number
space full of alternatives.
Cabin Diameter = 70 inches
Parametric Studies
Payload Weight = 18,000 lbs. - 2 X Weight Closely related is the utility of VDD value models for
(engine) parametric analysis and response surface analysis. Plot val-
ue, determined from the value model, as a potential surface
over a multidimensional design space. Such studies not only
Step 8. Derive The Engine Design locate good designs, they inform why the chosen design is a
Objective Function good one and they illustrate the robustness of the design to
Next, we put all the attributes and equations into a variations in the parameters.
spreadsheet. (See Tables 1 & 2, opposite page.) The equa-
tions make surplus value a function of aircraft attributes, Technology Evaluation
and aircraft attributes a function of engine attributes. One of the widest application areas for complex system
Therefore, surplus value is indirectly a function of engine value models in recent years has been technology evalua-
attributes. Perturbing each engine attribute and dividing tion (Collopy and Horton, 2002). 3 System and subsystem
the resulting surplus value perturbation by the attribute value models tend to be denominated in monetary units for
perturbation derived a simple linear version of this relation- convenience and elegance. Thus, the value calculated by
ship. This ratio approximates the partial derivative of sur- the model really is the dollar value of the system, relative to
plus value versus each attribute, and we refer to as a value some baseline. The model reflects a change in the system as
gradient. The greater the value gradient of an attribute, the a change in the monetary value. To evaluate a technology in
greater is its potential impact on product value. The linear the context of a system application, use the value model to
engine value model, or objective function, is the sum of the evaluate the system with the technology and without the

VALUE ORLD Volume 29, Number 2, Fall 2006 23


technology. The difference is the value of the technology Summary And Conclusions
when it is applied in the system. This is a gross value. To Value-Driven Design and its associated value model
make an informed decision to invest in a technology, one present a method for optimizing the design of complex sys-
must: tems for greatest value, instead of for a single attribute.
The case study of the workshop demonstrated that a
1) Evaluate the gross value of each of the system applica- value model could be developed for a complex system and
tions of the technology. its complex components. The model took into consider-
ation multiple customers and many attributes. It evolved
2) Adjust for risk: multiply each by the probability that value gradients for each of the engine attributes and could
the technology will succeed in that application (that is, have evolved value gradients for each of the overall system
complete development, successfully transition to ap- attributes.
plication, and successfully yield the benefits, combined The value gradients can be used in simple trade stud-
ies and other manual optimization exercises. Both the value
with the probability that the system will be success-
gradients and the model can be used in fully automated de-
fully fielded). sign optimization.
3) Adjust for time: discount the value from application
based on the time until the technology will be fielded, References
using an appropriate annual discount rate. 1 CoRopy, Paul D. "Economic-Based Distributed Optimal
Design." AIAA Paper 2001-4675, American Institute of
4) Subtract the cost of the technology development pro-
Aeronautics and Astronautics, Reston, VA, 2001.
gram.
2 Dijkstra, Edsger. "Programming Considered as a Human
The result after these steps is the expected net present Activity." Pages 213-217 in Proceedings of the 1965 IFIP
value of the technology. More sophisticated analyses, such Congress, North-Holland, Amsterdam, 1965.
as real options, can provided a more informed and often Collopy, Paul D. and Horton, Randolph R. "Value Mod-
more attractive value. The VDD value model is essential in eling for Technology Evaluation." AIAA Paper 2002-
3622, American Institute of Aeronautics and Astronau-
these analyses also.
tics, Reston, VA, 2002.
Schedule Management 4 Lippitz, Michael J.; O'Keefe, Sean; and White, John P. "Ad-
Days until design completion can be included along vancing the Revolution in Business Affairs." Pages 165
- 202 in Keeping the Edge: Managing Defense for the
with the other component attributes in the component Future. Edited by Ashton B. Carter and John P. White,
VDD chart. Determine the cost per day of delay in project Preventive Defense Project, Ashton B. Carter and Wil-
completion (do not be surprised if this factor is on the or- liam J. Perry, co-directors, Cambridge MA and Stanford
der of millions of dollars per day). Through a PERT analysis CA, 2000.
or equivalent, determine the probability that the particular
design team's work is on the critical path. The cost per day
of delays in the design team's delivery is (roughly) the prob-
ability of being on critical path times the cost per day of
total project delay.

Value Based Acquisition


-

Complex systems are seldom or never developed by a


single company—they require a supply chain of companies.
Value-Based Acquisition (Lippitz et. al., 2000) 4 ties com-
ponent price to component value and naturally encourages
the whole supply chain to engage in distributed optimal
design.

24 Volume 29, Number 2, Fall 2006 VALUE ORLD


Tables 1 & 2: Surplus Value Model

Surplus Value Model Aircraft Attributes


Surplus Value $ millions $5,846 Mfg. Cost $ million/aircraft $30.00
Market Size aircraft/year 145.38 Dev. Cost $ millions $5,000
Utilization flights/year 194.61 Gross Weight thousand lbs. 163.64
Value per Flight $ thousands/ft. 31.795 Field Length feet 9,000
Cost per Flight $ thousands/flt. 9.748 Time to Climb minutes 20
Charm $ millions/aircraft 0.00 Dispatch Reliability % 99.10%
Societal Good $ millions/aircraft -4.00 Boom Noise dP/P 0.4
Mfg. Cost $ millions/aircraft 30 Airport Noise EPNL dB -4
Dev. Cost $ millions 5,000 Range nautical miles 5,000
O&M Cost $/flight hour $2,880
DiscC none 8.00 Speed Mach 1.8
DiscP none 8.00 Cabin Diameter inches 70
Payload Weight thousand lbs. 8.00
Block Time hours 3.214286
Access Factor none 0.581534
Fudge Factor* $ thousands/flt. 25

Sensitivity Analysis ot Determine


Engine Objective Function
Engine Objective Function value

Engine Attributes Status Gradient Value nominal perturbed nominal perturbed

Takeoff Thrust
lbf 20,000 1.795 35,891 20,000 19,800 5,846.23 5,487.33
Perf
Clime Thrust Perf lbf 10,000 0.324 3,243 10,000 9,900 5,846.23 5,813.80
Weight (engine) thousand lbs. 5.00 -4,251 -21,256 5.00 4.95 5,846.23 6,058.79
SFC pph/lbf 0.55 -191,256 0.55 0.549 5,846.23 6,194.32
% successful
Reliability (engine) 99.60% 122,469 121,979 99.60% 99.40% 5,846.23 5,601.30
departures
Lndg Tkof Noise
dB -2.00 -1,163 2,326 -2.00 -1.98 5,846.23 5,822.97
(Obs)
$ millions/
Mfg Cost (engine) $5.00 -2,326 -11,631 $5.00 $4.95 5,846.23 5,962.54
engine
Maint Cost $/engine Flight
$500 -5.820 -2,910 $500 $495 5,846.23 5,875.34
(engine) hour
Dev Cost (engine) $ millions $2,000 -1.000 -2,000 $2,000 $1,980 5,846.23 5,866.23
offset 71,649.45
Surplus Value $ millions 5,846.23 5,846.23

VALUE ORLD Volume 29, Number 2, Fall 2006 25


Managing Value Engineering in New
Product Development
Don J. Gerhardt, CVS, PhD, PE

Abstract SAVE International, The Value Society, has published


Intense worldwide competition in the global market a Value Methodology Standard which describes the VA/VE
place is providing a demanding environment for companies. process'''. A process known as the value methodology job
Organizations that cannot provide high value products and plan is used for VA workshops. Pre-study and post-study
services to their customers will cease to exist. Fortunately activity is part of the plan.
there is a set of productivity tools centered on the value The plan consists of six sequential steps:
engineering (VE) process that can provide for continued 1) Information Phase
growth and prosperity if applied properly. This paper pres- 2) Function Analysis Phase
ents a process on how to apply VE tools to New Product 3) Creative Phase
Development (NPD). The use of VE in NPD can provide 4) Evaluation Phase
high value products and services to customers while simul- 5) Development Phase
taneously providing profits to manufacturers and suppli- 6) Presentation & Implementation Phase
ers.
Ingersoll Rand has developed a process for VE in NPD
based on best practices in the industry and internal best
Introduction practices. Many of the external best practices are from Ja-
The organization's ability to consistently develop high pan.
value products for its customers is paramount for survival.
The use of value engineering tools in the New Product De-
velopment (NPD) process can help insure that new prod-
New Product Development
The percentage of successful new products that meet
ucts are developed that customers want and provide excel-
lent return on the investment. customer desires and organization's operating income re-
quirements has traditionally been low. Robert Cooper in
The term value analysis (VA) is used when the value
his 3-part series on "Benchmarking Best NPD Practices" in-
process is applied to existing products. The term value en-
dicates that approximately one in 10 product concepts suc-
gineering (VE) is used when the process is applied to new
ceed commercially". The proper application of tools associ-
products. Many businesses in many countries have been
ated with VE can improve the success rate. The VE tools
successful applying VA to existing products. The applica-
tion of VE to new product development (NPD) has been that are used in the VE process at Ingersoll Rand include
more cli fficult. Japanese companies have been most success- the following:
ful in applying VE to NPD'4. Many Japanese companies be- 1) Product-Technology Roadmap
gan using VA in the 1960's and moved to VE in NPD during 2) QFD/VOC Quality Function Deployment/Voice of the
the 1970's. Japanese companies have continued to develop Customer
the process for VE in NPD. Isuzu as an example introduced
3) Competitive Benchmarking & Tear-down Analysis
VA in 1959, and VE in NPD in 1979. Hitachi Construction
Equipment introduced VA in 1960 and VE in NPD in the 4) Target Costing
1970's. 5) Part Cost Models
VANE evolved since it's beginnings in the 1940's as
an excellent productivity tool. VANE was originated by 6) Value Engineering (Zero, First and Second Look)
Lawrence Miles at the General Electric Co. 34. The heart of 7) DFMA Design for Manufacture and Assembly
VA/VE is analyzing the function of a product, process or 8) DFMEA & PFMEA Design & Process Failure Mode
service. Value is maximized by optimizing the equation: and Effects Analysis.
Value = Function/Cost

26 Volume 29, Number 2, Fall 2006 VALUE ORLD


Value Engineering in the NPD where VE tools should be applied.
Some Western companies have started to incorporate
Process
VE in the NPD process. An example is Pratt & Whitney
When Larry Miles developed the value principles they
Canada29. Pratt & Whitney includes VE, Function Perfor-
were first applied to existing products. Relatively few com-
mance Specification, QFD, Parametric Paired Comparison,
panies have progressed from VA on existing products to VE
Direct Value Measurement and Risk Assessment in their
on new products. There are several reasons for this.
NPD process. Minimizing risk and eliminating quality prob-
1) VA is relatively easy to apply and the results can be lems provide high value to customers. Ingersoll Rand in-
very significant. Organizations become satisfied with corporates Design Failure Mode Effects Analysis (DFMEA)
the results and do not move to the next level. and Process Failure Mode Effects Analysis (PFMEA) into
2) Most NPD processes are relatively complex and there the VA/VE process2.
is a reluctance to add more steps to the process
3) Many managers, engineers, financial professionals, Product-technology Roadmap
marketing professionals and sourcing professionals are Product-Technology Roadmapping is a business plan-
not aware of the tremendous impact VE can have on ning tool that helps in developing the strategy to provide
maximizing customer value and operating income high value products that customers want. The process is
4) There are not many publications available that describe illustrated in Figure 2 (top, opposite page). A forward cost
the process for successful application of VE in NPD model described by Albright 4 can be used for future target
costs for the product. The first step of the roadmap pro-
Abbie Griffin published an article on new product de-
cess is to identify customer drivers, define the market and
velopment practices including a summary chart of prior
develop a competitive strategy from the competitive land-
best practice studies 22. VE in NPD is only recently show-
scape. This information is then used in the second step to
ing up in NPD in the United States 32. It has been part of
map to the product drivers and a product roadmap is de-
the Japanese NPD process for over 30 years. New product
veloped. The technology roadmap is developed in the third
development processes typically have six stages. Booz, Al-
step with forward cost models. The technology roadmap is
len and Hamilton were one of the first to document the six
then mapped to the summary and action plan.
stages in 19689. Cooper added Review Gates to the Stages
The forward cost model helps to set target costs over
to develop the Stage-Gate model of new product develop-
the time horizon of the roadmap plan. The cost targets are
mentn. Figure 1 (below) shows the six NPD stages and
developed in the context of expected price ranges of future
competitors' products. An industry experience curve is one
NPD Stage VE Tools tool to help set cost targets. Planning is focused on technol-
1 Business Opportunity & • Technology Road-
ogy elements that have the greatest impact on achieving the
Project Planning mapping target costs.
• QFD/VOC
• Tear-Down
2 Concept Feasibility & • Cost Models Voice Of The Customer
Planning • Target Costs It is essential to understand the qualitative and quan-
• Function Analysis titative wants of the customer in order to provide value to
• Zero Look VE the customer. There are a number of ways to obtain the
3 Product Design • First Look VE VOC. These methods include:
• DFA
• DFM Process Selec- 1) Interviews (Face to face and phone)
tion 2) Surveys (Mail, phone, internet)
• DFMEA 3) Observations
4 Product Testing & • Second Look VE 4) Customer feedback
Process Design • PFMEA 5) Focus groups
5 Process Implementation 6) Trade shows
& Product Verification
Typical market surveys do not have the necessary de-
6 Product Launch &
Production
tail on the value perceived by the customer for various
functions. Conjoint analysis, alternative solutions matrices
Figure 1: NPD Process with value engineering. and value mapping are techniques used to help understand

ORLD Volume 29, Number 2, Fall 2006 27


Product-Technology Roadmap
A Business Planning Tool
Feature Fvolufion Plan
I I I I I
Product Architecture Roadma9
tiM in MIMI
Experience Curves Forward Cost Model Man to inveshmnts
Compeve Strategy
ftwoomi.

ralko
Product Driverframets Man to Elements Action Sum
Cometitive Landscape

MantoProductDrivers
Customer Drivers
Calwas Dirt

Product Roadmap
Market Definition
.411
Summary and Action
Plan
Technology
IMMO Roadmap
!PA1
.. ..
Product
Market and Roadmap
Competitive Strategy

Figure 2 Product-Technology Roadmap. Printed with permission from the Albright Strategy Group.

what customers value 3357. H. E. Cook of the University of of Mitsubishi Heavy Industries around 1972. Yoji Akao 3 in-
Illinois at Urbana-Champaign has been developing more troduced QFD to the USA in 1983. The "Function Deploy-
sophisticated methods to obtain VOC for VE in NPD. He ment" in QFD has its origin in the functional analysis of
co-authored a number of technical papers on VOC and VE VE.
that are published in SAE Publication SP-126647. SAE paper Larry Shillito was an expert on both QFD and VE. His
970763 in the series describes a Direct Value (DV) survey book Advanced Q.FD, Linking Technology to Market and Company
method that combines elements of conjoint analysis, con- Needs explains the relations between QFD & VE". He uses
tingent valuation, prospect theory and choice theory. SAE the term customer-oriented product concepting (COPC)
paper 970764 discusses evaluating mail survey techniques which combines several QFD matrices with VE. His later
for determining the value of vehicle options. Coo-
per indicates in his recent work on Benchmarking QFD
4
best NPD practices-III that Voice of the Customer House of Quality Correlations
and market input is one of the strongest discrimi-
nators between best and worst performers 13. You
2 How (Technical Response)
have to know what is of value to customers in or-
der to provide high value products. Karlsson and
3 5
Ryan26 describe an Analytic Hierarchy Process
(AHP) where customers compare requirements What Planning
pairwise according to their relative value and cost. (Customer Wants) Relationships Matrix
The results can be used to make VE decisions that
provide the highest value to customers.

QFD — Quality Function Deployment Technical Matrix


6
QFD is a tool that takes VOC data and pres-
ents it in a matrix format. QFD originated in Ja-
pan, It was first introduced at the Kobe shipyards Figure 3 FD — House of uality

28 Volume 29, Number 2, Fall 2006 VALUE ORLD


book" discusses value measurement techniques to quantify 2) Form a list of competitive or related products
structured VOC. Jim Rains presented a recent paper on the 3) Conduct an information search
relationship between QFD and VE44. 4) Tear down multiple products in class
QFD analysis can contain up to 30 matrices. The first 5) Benchmark by function
matrix is referred to as the House of Quality and is depicted 6) Establish best-in-class competitors by function
in Figure 3 (previous page, bottom). Room 1 contains the 7) Plot industry trends.
customer wants. The Planning matrix in Room 5 contains Ingersoll Rand started incorporated tear-down analysis
data from customers on the perceived value of our product into the VANE process in 1998. Current practice includes
and the competitor products. The Technical Matrix Room incorporating tear-down data in the Boothroyd Dewhurst
6 contains competitive data from tests and tear-down anal- DFMA software and links from part numbers to video clips
ysis. of the disassembly25.

Competitive Benchmarking And Target Costing


Teardown Analysis Target Costing originated in Japan at Toyota in the
Competitive benchmarking not only involves bench- 1959'4 following the introduction of value engineering in Ja-
marking products, but also materials and manufacturing pan". Target costing is the allowable cost of a product that
processes. The NPD-VE team needs to have knowledge of yields the required rate of return. It is a simple concept, but
the best worldwide technology, materials and processes in multifaceted and difficult to implement for Western com-
order to provide products with the highest value to the cus- panies. Target costing involves understanding the value
tomer. that customers are willing to pay for functions and the of-
The US auto industry was one of the first implementers ferings of competitors in the market place. VANE is used
of product tear-down in the 1960s 21. Competitive vehicles, to meet the target costs. Products are not put into produc-
were reviewed for initial defects, tested and operated for tion if the target costs are not met because required profit
thousands of miles over various roads. Failures that oc- margins will not be obtained.
curred during testing and operation were reviewed. The Traditionally many Western companies have estab-
vehicles were then torn down and the parts were displayed lished product price by adding profit margin to their cost
on tables in a large building. Engineers, manufacturing ex- of the product in the "Cost Plus" model.
perts and marketing personnel were invited to review and
analyze the parts for ideas and best practice concepts. Cost Plus Model: Cost + Profit margin = Price
General Motors introduced the static tear-down meth- Traditional "Cost Plus" models no longer work in the
od to Isuzu in the early 1970s. Isuzu further refined the tear- competitive global economy. The selling price is now driven
down process and incorporated it into their VA/VE pro- by the market place. Companies must now use the "Target
cess. The process is documented in the book Value Analysis Costing" model if they want adequate profit margins.
Tear-Down by Yoshihiko Sato and Jerry Kaufman". Sato and
Kaufmann present the following data on the percentage of Target Costing Model: Price - Profit Margin = Cost
VANE ideas generated during their five steps:
Yoshikawa et al" published a book in 1993, describ-
Step Percent Contribution ing target costing, functional analysis and cost tables. The
1.Select competitor product Consortium for Advanced Manufacturing International
2. Disassembly 30% (CAMI) has been active in promoting target costing. CAMI
3. Analysis 40% along with the Institute of Management Accountants and
4. Display 10% the Society of Management Accountants of Canada pub-
5. Examination 15% lished Management Accounting Guideline 28 on Implementing Tar-
get Costing in 199410. Ansari and the CAM-I Target Cost Core
Otto and Wood43 have a chapter in their product de- group' published additional information on target costing
sign book on product teardown. They also discuss other in 1997. Cooper and Slagmulder" studied target costing and
VANE tools used in NPD including customer needs, prod- value engineering at 25 Japanese companies and reported
uct function and DFMA. They present a seven-step process the results in 1997. The companies included Isuzu, Kom-
for product benchmarking. atsu, Nissan, Olympus, Sony and Toyota.
1) Form a list of design issues Target costs are established for parts manufactured in

VALUE ORLD Volume 29, Number 2, Fall 2006 29


house and by suppliers. Value engineering is used to meet another39. There are many small suppliers who have cost es-
the target costs. Jeffrey Liker 3° discusses an example at timating software for individual manufacturing processes.
Toyota in his book the "Toyota Way". Toyota established a
joint venture, Trim Masters, Inc. (TMI), in 1994, to supply
seats to Toyota's Georgetown, Kentucky plant. The joint Value Engineering
venture partners include Toyota Tsusho, Johnson Controls Larry Miles developed the initial principles of value
and Araco, Toyota's long-term interior parts supplier in Ja- analysis at GE in the 1940s. The principle is based on maxi-
pan. A target cost of 30% improvement was established for mizing value by analyzing functions of a product, process
the 2004 model. Toyota worked jointly with TMI to meet or service. Value is maximized by providing function and
the target cost by applying value engineering in the prod- quality at the minimum cost. VA is the term used when
uct design stage. An eight hour seminar on target costing is applied to existing products and VE when applier to new
available from the Advanced Value Group'. products. VE in NPD has evolved into three phases. The
three phases are sometimes referred to as product develop-
ment VE in Japan.
Cost Tables and Part Cost Models
1) Zero Look VE
Japanese companies have been developing and using
2) First Look VE
cost tables for many decades. Cost tables are detailed da-
3) Second Look VE
tabases of cost information. The cost tables were originally
developed for material costs, but now include all the manu- Zero Look VE
facturing processes and design variables that affect cost.
Originally the data was kept manually in tables. Computer The name "Zero-Look VE" evolved as a consequence of
costs models are now used. Most American companies do the principles of VE being applied earlier and earlier in the
not have the detailed historical costs like the Japanese. NPD process. At one time First Look VE was the earliest
Yasuhiro Monden" presents the following ways the that VE principles were applied in NPD. It became appar-
Japanese use cost tables. ent that additional benefits could be obtained by moving
VE further forward in the NPD process. Zero Look VE is the
• Estimating costs for planned products or planned
application of VE principles at the concept proposal stage.
parts
One of its objectives is to introduce new forms of function-
• Estimating costs for new technologies ality that did not previously exist'''. Sometimes this phase
• Predicting profitability at the manufacturing stage is referred to as "Product Planning VE." Sawaguchi 49 from
the Japanese SANNO Institute describes the application of
• Setting price targets for parts to be purchased
the "Combination of Patterns of Evolution of Technologi-
• Checking market-competitiveness of purchased parts cal Systems" from TRIZ with Zero Look VE and First Look
• Presenting VE-related concepts to suppliers VE. The Japanese refer to Zero Look VE and First Look VE
as Product Development VE. During the Zero Look VE cre-
• Negotiating reductions in purchased part prices at the
ativity techniques such as brainstorming and TRIZ 6'23.24 are
manufacturing stage.
used to establish possible solutions to meet the function
Japanese companies typically have cost planning de- and objectives
partments that maintain the cost tables. The cost plan-
ning department typically works in the product planning First Look VE
activity which reports to high level administration. The First-look VE focuses on the elements of product de-
cost planning department has important functions which sign once the overall concept has been established during
can include insuring that business plans are profitable, es- Zero-look VE. First-look VE is used to meet the target costs
tablishing target costs with Engineering and managing the which were established during Stage 2 of the NPD stage
VA/VE activity. process. Suppliers are asked to participate in first look VE
Some American companies have developed cost models to meet the target costs. Sometimes several cycles of First-
for certain commodities. Ford2° has them for electrical wir- Look VE is required to meet the target costs.
ing harnesses and heat exchangers. Part costing software
is available from various suppliers. DFM part costing from Second Look VE
Boothroyd Dewhurst 7 is one of the most well know. The Second-look VE is applied to selected subsystems and
original data was developed with grants from the NSF in parts where target costs are not being met. It can be ap-
the early 1980's. Costimator software from MTI Systems is plied during the last half of the planning stage and during

30 Volume 29, Number 2, Fall 2006 VALUE ORLD


the development stage. The objective is to improve the Frontier in Strategic Cost Management. Irwin. 1997.
value, increase the functionality and lower the cost of the 6 Bolton, James. Utilization of TRIZ as a Value Engineering
proposed components in order to meet the target cost and Tool. SJVE Conference. 2005.
functionality objectives. Second-look VE activity in Japan
closely resembles VA activity in the USA. Nakashima 41 from Boothroyd Dewhurst DFMA www.dfma.com
Toshiba indicated that Second-look VE is not as profitable 8 Boothroyd, Geoffrey, Peter Dewhurst and Winston Knight.
when compared to Zero-look and First-Look VE in the Product Design for Manufacture and Assembly. Marcel
product planning and development stage. Decker. 1994.
Booz, Allen and Hamilton. Management of New Products.
Design For Manufacture and New York. 1968.

Assembly CMA/IMA/CAMI. Management Accounting Guideline


The principles of DFMA are excellent to apply with 28, Implementing Target Costing. CMA. 1994.
VE in NPD. Pioneering research in DFMA was done by " Cook, H. E. Value and the Domain of Value Engineering.
Geoffrey Boothroyd, Peter Dewhurst and Winston Knight. Value World. SAVE. Volume 22, Number 3.1999.
Geoffrey Boothroyd received grants from the National Sci- 12 Cooper, R. G. Winning at New Products. Addison Wes-
ence Foundation, SME and industry for research on DFMA. ley. New York. 1993.
Material handling and part assembly times are summarized
in charts in Boothroyd's book'. DFMA helps to reduce as- 13 Cooper, R. G., S.J. Edgett and E.J. Kleinschmidt. Bench-
marking Best NPD Practices-I, II & III". Research-
sembly time and combine functions to provide higher value
Technology Management. Industrial Research Insti-
products".
tute. 2004.
DFMA can be applied with the manual technique by
looking up data in the charts. The data is also available with 14 Cooper, Robin, and Regine Slagmulder. Target Costing
software. DFMA software is available from Boothroyd De- and Value Engineering. Portland: Productivity Press,
whurst Inc. 7 and from Munro & Associates 40. Selecting the 1997.
optimum manufacturing process is important to provide " Fukuoka, Nobuyuki. A New supporting Method for New
the highest value products to customers. Swift and Booker Product Concept Making Based on Customer Function
from the University of Hull provide information on Process Evaluations. SJVE Conference. 2003.
selection".
" Gerhardt, D. J., W. R. Hutchinson and D. K. Mistry. De-
Summary
sign for Manufacture and Assembly: Case Studies in
The application of VE in the NPD process can help in-
its Implementation. International Journal of Advanced
sure that products are developed that provide high value to
Manufacturing Technology. 1991.
customers and excellent returns to the company. The VE
tools are proven and work when properly applied. VE tools '7 Gerhardt, Don J. Meeting with Hitachi. 2003.
are becoming necessary for survival in the highly competi- " Gerhardt, Don J. Meeting with Isuzu. 2003.
tive global economy. " Gerhardt, Don J. Meetings with Denso, Omron and Toy-
ota. 2005.
References 20 Gerhardt, Don J. Personal observation at Ford Motor
Advanced Value Group. Jim Rains. Target Costing Semi- Company, 1974.
nar. www.avgconsulting.com 2003.
21 Gerhardt, Don J. Personal observation at the General Mo-
2 AIAG. Potential Failure Mode and Effects Analysis. www. tors Milford Proving Grounds, 1965.
AIAG.com 2001. 22 Griffin, Abbie. PDMA Research on New Product Devel-
3 Akao, Yoji. Quality Function Deployment, Integrating opment Practices: Updating Trends and Benchmarking
Customer Requirements into Product Design. Produc- Best Practices. J. Prod. Innov. Manag. 1997.
tivity Press. 1988. 23 Hamamoto, Yuichi. Applying VE+TRIZ Method to the
4 Albright, Richard E. How to Use Roadmapping for Global Planning Phase of a Development Project. SAVE Con-
Platform Products. PDMA VISIONS. Vol. XXVI. Octo- ference. 2005.
ber 2002. www.albrightstrategy.com 24 Hanik,
Peter and J. J. Kaufman. VE/TRIZ: A Technology
Ansari, Shahid and Jan E. Bell. Target Costing, The Next Partnership. SAVE Conference. 2005.

ORLD Volume 29, Number 2, Fall 2006 31


" Hynds, Jeff and Don Gerhardt. Ingersoll-Rand Tear- 43 Otto, Kevin N. and Kristin L. Wood. Product Design,
Down Procedure. 2005. Techniques in Reverse Engineering and New Product
26 Karlsson, Joachim and Kevin Ryan. A Cost Value Ap- Development. Prentice Hall. 2001.
proach for Prioritizing Requirements. IEEE software. Rains, James A. The Intersection of Quality Function
Sept.-Oct. 1997. Deployment and Value Engineering. SJVE Conference.
27 Kigawa, Michijiro. Improvement on Customer Satis- 2005.
faction and Efforts on Value for Customers. Value Im- 45 Rains, James A. and Mark Sullivan. Unlocking the Power
provement Activities with the Use of IT Tools. Hitachi of Target Costing: A Platform for the Strategic Use of
Construction Machinery Co. SJVE Conference. 2002. VM. Value World. SAVE. Volume 22, Number 1. 1999.
28 King,
Thomas R. Value Engineering, Theory & Practice 46 Ried, Axel. Setting Corporate Guidelines for VE in New
in Industry. 2000. Product Development. SAVE Conference. 2004.
29 LeBlanc, Alain. Integrating Value Methodologies into 47 SAE, Value-Based Decisions for Automotive Engineering.
Product Development and Project Management Pro- SP-1266. 1977.
cesses at Pratt & Whitney Canada. SAVE Conference.
48 SAVE. Value Methodology Standard. 1998. www.value-
2004.
eng.org
Liker, Jeffrey K. The Toyota Way. McGraw Hill. 2004.
49 Sawaguchi, Manabu. Study of Effective New Product
" McConville, G. P. and H. E. Cook. Evaluating Mail Survey Development activities through Combination of Pat-
Techniques for Determining the Value of Vehicle Op- terns of Evolution of technological systems and VE.
tions. SAE #970764. 1997. SANNO Institute of Management. TRIZCON 2001.
32 McMahon, Edward S. Value Engineering and Design. 5 ° Sato, Yoshihiko and J. Jerry Kaufman. VA Tear-Down: A
SAVE Conference. 2005. New Value Analysis Process. SAVE Conference. 2004.
" Mello, Sheila. Customer-centric Product Definition. PDC 51 Sato, Yoshihiko and J. Jerry Kaufman. Value Analysis
Professional Publishing. 2002. Tear-Down: A New Process for Product Development
34 Miles, Lawrence. Techniques of Value Analysis and Engi- and Innovation. Industrial Press and Society of Manu-
neering. Lawrence D. Miles Value Foundation. 1989. facturing Engineers. 2005.
35 Mizutani, Takanori. An Effective VE Approach at Devel- 52 SAVE. Value Methodology Standard. SAVE Internation-
opment and Design Stages for Automotive Parts Manu- al. 1998.
factures. Denso. SAVE Conference. 2003. 53 Shillito, M. Larry. Advanced QFD, Linking Technology
" Monden, Yasuhiro. Cost Reduction Systems, Target Cost- to Market and Company Needs. John Wiley & Sons.
ing and Kaizen Costing. Productivity Press. 1995. 1994.
37 Monden, Yasuhiro. Toyota Management System. Produc- 54 Shillito, M. Larry. Voice of the Customer. CRC Press.
tivity Press. 2004. 2001.
" Monroe, E. M. and H. E. Cook. Determining the Value " Shingo, Shigeo. Study of Toyota Production System. Ja-
of Vehicle Attributes Using a PC Based Tool. SAE pan Management Association. Tokyo. 1981.
#970763. 1977. 56 Swift, K. G. and J. D. Booker. Process Selection from De-
39 MTI Systems Costimator software www.mtisystems.com sign to Manufacture. Arnold. 1997.
40 Munro Associates. Sandy Munro. www.munroassoc. " Takaki, Kenji. The Effectiveness of Conjoint Analysis in
COM VE Functional Evaluation. SAVE Conference. 2005
Nakashima, J. and T. Hattori. Effective Propulsive tech- " Takemura, Masaya and Manabu Sawaguchi. The Possi-
nique on 2nd Look Value Engineering. SAVE Confer- bathes of VE Activities as New Product Planning by
ence. 1992. Utilizing TRIZ Techniques. SANNO Institute of Man-
42 Okuhara, Masao, Masayasu Tanaka, and Hiroo Hirose. agement. SAVE Conference. 2004.
A New supporting Method for New Product Concept- 59 Yoshikawa, Takeo, John Innes, Falconer Mitchell and
making based on the Customers' Function Evaluations. Masayasu Tanaka. Contemporary Cost Management.
SAVE Conference. 2004. Chapman & Hall. 1993.

32 Volume 29, Number 2, Fall 2006 VALUE ORLD


Editorial Poli cy SAVE International
136 S. Keowee Street
Value World is published by SAVE International and is distributed internationally.
Dayton, OH 45402, USA
Value World welcomes original articles on value engineering and related disciplines.
937-224-7283
Reprints or abstracts from other journals or periodicals are acceptable, provided that
937-222-5794 (FAX)
prior permission is obtained from the copyright holder(s). Value World's policy is to
info@value-eng.org
provide a medium for contributors to express themselves professionally on
www.value-eng.org
advances in the state of the art. The views expressed in Value World are neither
approved nor disapproved by SAVE International.

Editorial Staff Subscriptions: A yearly subscription for SAVE Interntional emb


included in their annual dues. The yearly rate for nonme bers in the U
Editor-in-Chief: Dr. Roy M. Woodhead, CVS
States is $75; international is $100 including airmail postaia.
Editorial Board: Dr. Jose Albors; Dr. Muhammad A. Al-Ghamdi; Dr. Tatiana
Bachkirova; Henry A. Ball, CVS; Dr. Herve Christofol; Vaughan W. Coffey; Howard
Ellegant, AIA, CVS-Life; Prof. Antonio Fernandes; Theodore C. Fowler, CVS-Life; Change of Address: Send address changes to Value
Jean-Pierre Grandhaye; Wilhelm Hahn, CVS; Kirsty Hunter; J. Jerry Kaufman, 136 S. Keowee Street, Dayton, OH 45402, U.S.A.
CVS; Prof. John R. Kelly; Dr. Stephen Kirk, CVS-Life; Dr. Mei-yung Leung, AVS;
Mary Ann Lewis; Prof. Steven Male; Prof. Jean Michel; Gary Myers, PE, CVS; Dr.
Ferenc Nadasdi, CVS; Dr. Masao Okuhara; Joseph F. Otero, Jr., CVS; Lucie Parrot,
PE, CVS; Prof. Dan Seni; Dr. Jong won Seo; Dr. Eugene Sweeny; Prof. Vincent
Thomson; Prof. Qui Wanhua; Tony Wilson; Dr. Roy M. Woodhead, CVS, 02006 SAVE International, All rights reserved

Board of Directors
President: R. Terry Hays, CVS-Life, FSAVE
Executive Vice President: David C. Wilson, P.Eng., CVS
Vice President—Communications: Rodney Curtis, PE, CVS
Vice President—Construction: Earl C. Wilson, PE, CVS
Vice President—Conferences: Renee L. Hoekstra, CVS
Vice President—Education: Roy M. Woodhead, PhD, HONS, CVS
Vice President—Finance& Administration: Howard B. Greenfield,
PE, CVS-Life, FSAVE
Vice President—Government: Katherine E Bethany, CVS
Vice President—Manufacturing: Bijay Nayak, PhD, PEng, CPEng,
CMfgE, CME, FIEAust, AVS
Vice President—Global Affairs: Donald Hannan, CVS-Life, FSAVE, FNMA,
FAIM, MHKIVM (VMF), MBE
Vice President—Membership: Daniel Fanning, PE, AVS

SAVE International

V.
PRESORTED
"The Value Society" STANDARD
U.S. POSTAGE
136 S. Keowee Street
SAVE PAID
INTERNATIONAL Dayton, OH 45402 DAYTON, OH
U.S.A. PERMIT NO. 1519
Change Service Requested

AUTO**MIXED ADC 450


J. Kaufman CVS-Lifa, FSAVE
J. J. Kaufman Associates, inc.
20606 Sunset Bena Ln
Spring TX 77379-7935

You might also like