DESIGN METHODS
639
INTERNATIONAL DESIGN CONFERENCE - DESIGN 2010
Dubrovnik - Croatia, May 17 - 20, 2010.
FROM REQUIREMENTS TO DESIGN
SPECIFICATIONS- A FORMAL APPROACH
W. Brace and K. Thramboulidis
Keywords: requirements, requirements checklist, design specification,
requirements formalization, model-centric requirements engineering
1. Introduction
The traditional development process for mechatronic systems is criticized as inappropriate for the
development of systems characterized by complexity, dynamics and uncertainty as is the case with
today’s products. Software is playing an always increasing role in the development of these systems
and it has become the evolving driver for innovations. It does not only implement a significant part of
the functionality of today’s mechatronic systems, but it is also used to realize many of their
competitive advantages [Thramboulidis 2009]. However, the current process is traditionally divided
into software, electronics and mechanics, with every discipline to emphasize on its own approaches,
methodologies and tools. Moreover, vocabularies used in processes and methodologies are different
making the collaboration between the disciplines a great challenge.
System development activities such as requirements and design specification, implementation and
verification are well defined in software engineering. In terms of software engineering the term
“design specification” is used to refer to the various models that are produced during the design
process and describe the various models of the proposed solutions. Therefore, “design specifications”
are descriptions of solution space while “analysis specifications” are descriptions of problem space.
Analysis specifications include both requirements specifications but also the problem space structuring
that is represented by analysis models such as class diagrams that capture the key concepts of the
problem domain and in this sense they provide a specific structuring of the problem space. Dorst and
Cross (2001) emphasize on the co-evolution of problem and solution spaces up to the point of time
that a match will be found, and consider artefacts of both spaces very important. However, it is clear
that the most important are the ones of the problem space since it is not possible to have an acceptable
system even with the best solution space if this is based on an incorrect problem space formulation. On
the other hand it is possible to have an acceptable system, even without a very good solution space
formulation assuming that this is based on a correct problem space model. It is clear that both
activities i.e. those involved in problem space formulation (problem space structuring) as well as those
involved in solution space formulation include design decisions. Requirements engineering along with
domain analysis improve the knowledge on the problem space and make the reasoning steps during
the subsequent design process more effective. A well defined model-driven process in software
engineering might result at the end to the automatic synthesis of the executable code [Douglas 2006].
Such type of formal processes has not been obtained in engineering design. This is due to the fact that
knowledge in engineering design is still more empirical and concepts are not defined with the
precision and uniformity as in the software engineering domain. In addition, the role of knowledge
management in the early development phases has to be intensified due to increased systems
complexity, and more demanding productivity requirements and competitiveness.
640 DESIGN METHODS
Raw requirements, i.e. requirements that have not been analyzed and formulated, are usually
expressed in narrative format. These narrative requirements (NR) provide the foundation for the
design efforts but do not necessarily provide the complete knowledge required for the subsequent
design process. Therefore it is important to analyze and formulate them to become abstract,
unambiguous, traceable, and validatable, i.e. well-formed. Several approaches are proposed to refine
and extend requirements [Otto et al. 2001, Ulman 2002, Pahl et al. 2007]. According to these,
information processing during the initial development phases translates NR to a formalized
requirement specification, which should be complete, unambiguous, precise, and verifiable. But these
proposals are document centric and very often, labour intensive. Moreover, requirement engineering
activities are information-driven and knowledge oriented. There are many characteristics associated
with NR information. These have to be analyzed to find relationship between the vast information in
order to be structured to facilitate the design activity [Pahl et al. 2007]. Knowledge acquisition,
sharing, and integration are among the activities that enable the designer to effectively address
complexity in product development. The required information and knowledge is barely explicitly
accounted for in the early development phases, and the designer has to collect the required knowledge
from various sources.
The approach presented in this paper aims to formalize requirements which are expressed in narrative
format, as is usually the case for customer requirements. It is a model-centric approach on refining and
extending NR, and is based on an integrated framework of logical reasoning and incidence matrix
operations described in [Tomiyama et al. 2002, Kusiack 1992]. The approach is composed of two
major sub-processes: that the first models the relevant to the requirements checklist (RC) information
and knowledge into a single knowledge framework; and the second formalizes the narrative system
requirements using the knowledge framework obtained from the first sub-process. It is not an intention
of this work to elaborate on the first stage of requirements elicitation.
The whole approach is considered as an extension of the SysML way of capturing requirements.
SysML is exploited in [Thramboulidis 2009] to propose an architectural framework to address the
challenge of synergistic integration in engineering design. The 3+1 SysML view model architecture is
proposed to define a framework for the synergistic integration of the various disciplines involved in
the development of Mechatronic Systems. SysML requirement diagrams are used in this architecture
as a means for capturing system requirements. However, such a specification does not take into
account the work that has been done during the last years in mechanical engineering design. The
approach presented in this paper exploits the existing knowledge on requirements engineering in the
mechanical engineering domain and integrates it with the SysML formalism to provide a solid
requirements specification for the 3+1 SysML view model architecture, to be used in the subsequent
development phases.
The remainder of this paper is structured as follows: In section 2, background and related work are
given. The proposed approach for the requirements formalization is presented in section 3. A case
study is presented in section 4 to demonstrate the applicability of the proposed approach. Finally
conclusions and future work are given in the last section.
2. Theoretical background
2.1 State of the art in problem structuring
There is no single universally acclaimed sequence of steps in engineering design. Researchers have
outlined the design process in as few as 4 steps to as many as 25 steps. A complete design process
includes, as the first phase, conceptual design which is the process by which the design is initiated and
carried to the point of creating a number of possible solutions. The discrete activities considered under
conceptual design are, identification of customer needs (IoC), problem definition (PD), information
gathering (IG), conceptualization, concept selection, refinement, and design review. This paper
focuses on PD and IG and uses the term task clarification or problem structuring (PS). The first three
activities are used to refer to activities that are known in software engineering as requirements
elicitation, requirements refinement and formalization, while the latter activities are equivalent to
architectural design in the software engineering domain. PS is based on three elements: 1) designers
DESIGN METHODS
641
strategy in PS, 2) the design requirements generated, and 3) the information access [Restrepo et al.
2004]. All three elements are crucial and need a better formalism. Dorst and Cross (2001) in extensive
creative design practice studies have come out with two notional design “spaces” – problem space and
solution space. These spaces co-evolve with interchange of information between them. PS is a process
of drawing upon knowledge (or external information) to compensate for missing information and
using it to construct the problem space. Requirements are used to specify the design assignment
(defining the problem space) and to constraint the desired solution (exploring the solution space) and
are therefore an important aspect of PS [Restrepo et al. 2004]. They are often given in narrative format
as design problems (narrative requirements). The design problem can be characterized as not being
subjected to systematization, incomplete, vague and there is a lack of information in each of the three
components stated. NR is dynamically generated during the design process from a design requirement
into a design specification (DS). The requirement formalism is triggered by prior knowledge or by
knowledge through interaction with design object or with external sources of information.
Requirements are discriminated to functional requirements and non-functional requirements or
constraints.
Several approaches that use decomposition to enhance dynamic generation of requirements include the
use of RC and function models (scenario creation) [Cross, 2000, Ulman 2002, Otto et al. 2001, Pahl et
al. 2007]. Both methods are effective for design specifications and analysis, i.e. problem space
structuring. Engineering design is an activity that requires both logic and creativity. Systematic
methods and tools have continuously been created to better conduct the logical analysis in order to
unleash the designer to engage in the creative aspects of problem solving. Research in different design
disciplines has produced models concentrated on different aspects of design process [Cross, 2000,
Restrepo et al. 2004]. For instance, architecture models propose that solution concepts go before
problem structuring. Models from software design propose the designer negotiating the structure of the
design problem. Engineering design provide models, with the basis that problem analysis precedes
synthesis of solution [Restrepo et al. 2004, Pahl et al. 2007]. In a complex system design, the design
strategy is not discipline specific and a combination of these strategies is crucial. The RC is generic
based on design goals, constraints and all relevant system parameters and information needed for a
successful system design. RC and function models are used not only by Pahl and Beitz (2007); various
engineering textbooks make use of the idea of the checklist with different names and categories. RC
such as the one shown in table 1 provides a decomposition strategy useful for logical analysis and
intelligent system application for a combination of the different model strategies in requirement
formalism. The RC is used to facilitate the shift from a document-centric to a model-centric problem
structuring. Rational design methods unlike creative methods encourage a systematic approach to
design. Checklist is the simplest kind of rational design method. It externalizes the requirement
process so that important design issues are not overlooked, and formalizes the process by making a
record of items which can be analysed and checked-off as they are completed. It also allows sub-
division of task such as allocation task to different team members [Cross, 2000].
Table 1. Part of a requirements checklist [Pahl et al., 2007]
Category Example
Geometry Size, height, length, diameter, space, connection, arrangement
Forces
Direction, magnitude, frequencies, Weight, load, deformation, stiffness, elasticity,
inertia, resonance
Energy
Output, efficiency, loss, friction, ventilation, state, pressure, temperature, heating,
cooling, supply,
Material
Flow and transport, physical and chemical properties, design for manufacturing
(DFM)
Costs Maximum permissible manufacturing cost, cost of tooling, material cost, time,
Schedules
Time constraints, end date of development, project planning and control, delivery
date
642 DESIGN METHODS
2.2 Related work
Many approaches exist for NR processing, refinement and formalization. A graph based language has
been used to describe the behavioural specifications of a design as well as the behaviour of the
available physical components.
An application of the concept of Quality Function Deployment (QFD) to improve product quality
based on customer requirements has been described [Cross 2000, Otto et al. 2001, Ulman 2002].
Kusiack [Kusiack 1992] utilizes logic and matrix to describe requirements and transform them into
function specifications. SysML, a flexible modelling language derived from UML, supports
specification, analysis, design, verification, and validation of a broad range of complex systems.
Requirements are defined in SysML in the form of diagrams. The requirements diagram can depict the
requirements in graphical, tabular, or tree structure format. Tomiyama et al. (2002) have identified the
following knowledge/information operations in a design phase: knowledge/information acquisition,
reorganizing, confirmation, revision, conflict resolution, solution synthesis, and object analysis. The
approach of Kusiack and Tomiyama seems promising as it also manipulates information using basic
logical procedures. Since requirement activities are information driven and the requirement checklist
is knowledge oriented we built on these approaches.
In summary, most of the requirements are modelled based on small parts of the information needed in
the product development process. For instance the QFD method is best for collecting and refining
functional requirements hence the “F” in its name. In the QFD the “how” step consists of the design
specification and this must be modelled beforehand.
3. The proposed approach
In this section, we describe our approach for requirements formalization. The approach is motivated
by the need to better organize the knowledge necessary to enhance the ability of the designer to
express the proposed solution. By logically formalising the requirement process, a computable
requirement formulation model can be arrived.
3.1 The basis
The transformation of NR to DS is modelled by logical formalism. Logic basically has two types of
reasoning which is by deduction and reduction. A mathematical logical reasoning scheme is as shown
in equation (1) [Tomiyama et al. 2002]:
⊢
(1)
Equation (1) shows the simplest set-up of reasoning where deduction is a reasoning process to derive
axioms (A) from given theorems (T
h
) using inference rule “σ” (modus ponens). The sets A and T
h
consist of logical formulae. The symbol “
” denotes “deducible”. Axioms A is decomposed as
follows:
⨁
(2)
∈
(3)
Where K is knowledge, F
d
is a set of “definition facts”, and T
d
is semantic distinction of definitions.
Observed facts F
o
can be considered as part of T
h
such that:
⊃
∈
(4)
Where T
o
is statement about observations in extra-logic world. Therefore equation (1) can be written
as:
⋃
⊢
(5)
Deduction is to obtain T
h
from K and F
d
, and abduction could be performed in the following three
reasoning mode:
1. Obtain K from Fd from partially given Th
2. Obtain K from Fd and partially given Th
3. Obtain Fd from K and partially given Th
DESIGN METHODS
643
In order to clarify the relationship between these logical operations and reasoning in the requirement
formulation process, we adopt the model-based abduction method. Abduction in design is a process to
derive design solution (entity) from required properties [Tomiyama et al. 2002]. To formalize
abduction in design, design knowledge (i.e., axioms) is classified by:
Knowledge that describes an entity´s functions: e→f
Knowledge that describes an entity´s property: e→p
Where e denotes an entity, p denotes an attribute, and f denotes a function. Both these types of models
can be used bidirectional.
The algorithm of model-based abduction is outlined as follows:
The process adopted is to derive solution candidates from axiom K
i
:
1. Set requirements that can be treated by axioms Ki as theory Thi (equation 5)
→
,
→
…
→
..
⋃
⊢
,
(6)
2. Derive solution candidates by abduction with assumptions. Incidence matrix is used to
establish the relationship between Fdi and Ki.
∶
;
1
0
(7)
The matrix includes both the simple one-to-one relationship (equation 7) and a more complex
interaction. Each nonzero entry in the matrix has a value ɑij ≥ 1. For ɑij > 1, Fdi is satisfied by
multiple categories from Ki. Thus a Fdi can appear in more than one Ki-axiom.
3. Analyze Fdi in the attribute and function space by deduction using Ki . This will enrich Thi.
In the next sections, we formalize the algorithm of model-based abduction as operations to the
formalization of the RC and the narrative system requirement.
3.2 Requirements checklist modelling
An application of the logical formalism in requirement formulation is shown in figure 1. Information
access is one of the foundation of PS. Choices of the designers during problem structuring depend
partly on obtaining proper information and having easy access to information. Relevance is a product
of the interaction between the designer and the information source. Improving the relevance of the
information retrieved by a system will improve its accessibility [Restrepo et al. 2004].
Figure 1. Logical approach to requirement checklist model and requirement formalization
To deal with the knowledge and relevant information required to synthesise the NR, we propose to
organize them based on categories as the ones defined in existing RCs, for instance, as the one shown
in Table 1. Since knowledge base is domain specific, the RC is first analyzed to identify additional
categories. These categories are further expanded based on specific domain (for instance the car
industries have its own specific knowledge/information apart from the general knowledge for
requirement formulation). In the RC modelling, various design relevant knowledge/information are
integrated and stored in a single work space.
Requirement
checklist (RC)
Knowledge/
Information
sources
Requirement
checklist
model
(C)
Logical
operation
Narrative
Requirement
(NR)
Logical/matrix
operations
Design
specification
(DS)
Functional/
Physical
models
Behaviour
models
(analysis)
Abduction
Deduction
644 DESIGN METHODS
Figure 2. Logical interpretation of requirement category model
The algorithm of model-based abduction is formalized as operations to the RC modelling as shown in
figure 2. The steps included are the following:
Step-1: Set up sources of knowledge/information for requirement formulation in the working
space (T
h
)
Step-2: Select identified categories (RC
i
) to derive abstract entity concepts from T
h
(relevant
reasoning)
Step-3: build aspect-specific RC model (C
i
) using RC
i
. Transport information from T
h
and
other modellers, if necessary. This corresponds to enriching information about RC and
obtaining a specialized RC model as the one shown in Figure 3.
Figure 3. Part of a requirement checklist model
3.3 Requirements formalization
The algorithm of model-based abduction is formalized as operations to the requirement formulation
framework outlined in figure 1. In the procedure, first from a given NR, an initial requirement (R
i
) is
set up in the theorem (T
h
) working space. Verbs, adverbs (f
i
, - function concept) nouns, and adjectives
(p
i
, - attribute concepts) are then identified in R
i
. This is to help identify the key list in R which
corresponds to entity concepts (e
i
) (Figure 3) of the modelled checklist (C). C is selected in the
knowledge base (K
i
) domain. Entities e
i
from C are utilized to derive abstract concepts (f
i
, p
i
in R)
from T
h
. If there is no more e
i
to be used, then the operation is terminated and e
i
is mapped onto f
i
and
p
i
. The interaction between R and C is modelled with incidence matrix. A design specification list
table (DS) is then developed for the selected categories of C and the identified requirements.
The next stage involves analysing the identified initial requirements. This is done using knowledge
and information from C. First requirements under the functional performance category in the DS-table
are analyzed using the inference “if then”, i.e. If sub function A is present, Then sub function B,
relationship to arrange sub functions in a logical order. Physical entities in the sub functions are
decomposed; in this way, the complexity of the sub-function is reduced. The function requirement is
rearranged according to the action necessary to attain that particular function. Next, the physical
parameters and subsequent dimensional quantities of the various physical entities are identified. The
DS-table is then updated with the newly identified requirements.
DESIGN METHODS
645
In the next step, the non-functional parameters in the DS-table are analyzed. Most requirements under
this category are often unbounded, i.e. making relative statements that cannot be verified, as for
example “simple maintenance”. These requirements have to be restated defining specific bounds in
order to transform them to be validatable. In this case, the requirement is first developed by identifying
essential information from C. This information is further refined also using information from C, thus
in this way the requirements are defined systematically. The DS-table is then updated. This process is
iterative and the procedure described above is applied with other sets of requirements in the table to
make DS information well-formed.
The validity of the established requirements is confirmed first by creating specific function models
based on the flow of energy, material and signal expressed with the use of a block diagram. The
function models which express solution-neutral relationships between inputs and outputs of the system
can be numerous. But the above synthesis process allows the narrowing of the range of models. Next
the physical relationship of the function model is established using both the physical and dimensional
parameters. The behaviours of the function and physical models are then modelled, simulated and
compared. The established requirements are arranged in a clear order and in the case of a complex
system; they are further assigned into identifiable subsystems, functions, or assemblies. The proposed
formalism can be applied for structuring of different types of design problems independent of
engineering discipline. Design problem may apply to different design modules namely, original
(novelty), adaptive (using established solution principles) or variant (modular) design. In this article
the approach is applied to a case study of an ongoing project in an adaptive design of a mobile work
machine.
4. Case study: The case of an underground work machine
To understand the approach and formalism presented in the previous section, we consider as a case
study the development of a new version of an underground work machine used for loading and
carrying soil in a forward motion uphill a certain distance to dump and reverse downhill for more soil.
Table 2. a) Example of a narrative requirement and b) Formulated initial requirements
Underground work machine
Required: a machine to load and transfer soil uphill a
total distance of 300m to dump soil with following
characteristics:
Efficient energy consumption
Easy to maintain
Maximum machine length 7417mm
Maximum machine width 1473mm
Maximum machine height 2143mm
Maximum operating load 4000kg
Attention to be paid to safety and environmental
issues. Cost should not exceed 50,000 euro. Finished
product should be marketed within 2 years.
abbr. Requirements
R1
load , transfer and dump soil
uphill
f1,f2,f3,P1,P2
R2
Efficient energy consumption
P3,P4
R3
Simple maintenance
P
5
R4
soil transfer distance ≥ 300m
P
6
R5
machine length ≤ 7417mm
P
7
R6
machine width ≤ 1473mm
P
8
R7
machine height ≤ 2143mm
P
9
R8
operating weight ≤ 4000kg
P
10
R9
safety
P
11
R10
environment issues
P
12
R11
Cost < 50,000 euro
P
13
R12
Project time < 2 years
P
14
The requirements for this new version are to reduce energy consumption and harmful emission as
shown in table 2a. Formalization of the narrative customer requirements using our approach is as
follows.
Step-1: initial requirement setup
From the narrative requirement, an initial requirement is formulated as shown in table 2b. Then,
function and attribute concepts are identified (Underlined in the initial requirements list)
Step‐2:mapping entity concepts onto the abstract concepts
Entity concepts (e
i
) from C (Figure 3) are mapped onto the function and attribute concepts defined in
step 1. As an example we take the initial requirement R1 (table 2b) and referring to e
i
, we have the
646 DESIGN METHODS
following mapping. {e
1
→f
1
, e
1
→f
2
, e
6
→P
1
, e
14
→P
2
} (transfer → transfer, load → load, material →
soil, operation → uphill) according to Figure 3 and Table 2b.
Step-3: creating interaction matrix
An incidence matrix is constructed to identify the interactions between R and C as shown in table 3.
Table 3. Part of incidence matrix table representing formal interaction between initial
requirement and requirement checklist
R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11 R12
Function C1 e
1
Geometry C2 e
2
e
2
e
2
Force C3 e
3
Energy C4 e
4
Kinematics C5 e
5
Material C6 e
6
Safety C8 e
8
Ergonomics C9
Production C10 e
10
e
10
e
10
e
10
Transport C13 e
13
e
13
e
13
e
13
e
13
Operations C14 e
14
e
14
e
14
Maintenance C15 e
15
Recycling C16
Economics C17 e
17
schedule C18 e
18
The DS- table is then created for the identified requirements under the various C-categories. Due to
space limitation not all identified requirements in our example are shown in table 4.
Table 4. Part of design specification table for underground working machine
Date Design Specification sheet
Project: Underground loading machine
Responsible
Functional performance
Load , transfer and dump soil
Constraints
Geometry
Loader length ≤ 7417mm
Loader width ≤ 1473mm
Loader height ≤ 2143mm
Force
Operating weight ≤ 4000kg
Energy
Efficient energy consumption
Kinematics
Distance moved with load ≥ 300m
Material
Soil
loader
Safety
safety
Maintenance
Simplify maintenance
Economics
Cost < 50,000euro
Schedule
Project time > 2 years
DESIGN METHODS
647
Step-4: functional performance requirements analysis
The requirement statement “Load, transfer and dump soil” (table 4) is arranged in a logical order
based on the inference rule using block diagrams (figure 4a). Entities in individual blocks are then
decomposed, for instance as shown in figure 4b the “load soil” sub function will require two geometric
entities and the loader may also require 2 entities.
The sub functions are then rearranged in a logical order to include the actions and entities involved in
a particular function. Figure 4c shows the basic procedure for loading including a human actor.
Figure 4. a) Logical order of initial requirement b) Entities required for load function c) Basic
procedure for loading
The entities in figure 4c (human, loader, soil) and the functions (move, position, connect) are analyzed
using knowledge and information from C. This allows identifying various functions and entities
required for the specific process. Examples of “human” and “move” analysis are shown in figure 5.
Figure 5. Analysis of entities and functions in loading machine
Step-5: Identification of physical parameters and dimensional quantities
The physical parameters and subsequent dimensional quantities of the various entities are then
identified and the DS-table is updated.
Step-6: non-functional requirements analysis
In this step, we take as an example the initial requirement “simplify maintenance” under maintenance
category in the DS-table (table 4). This is analyzed using knowledge/information from C15-
maintenance (figure 3). The procedure is as shown in figure 6. This procedure is continued to obtain
all relevant and identifiable requirements and DS-table is then updated.
The validity of the established requirements is confirmed by first generating function and physical
models. The behaviours of these models are simulated and compared. Work on the behaviour
simulation and comparison has been done in a previous work [Brace et al. 2009] and interested readers
may refer to it. If the evaluation and comparison are acceptable, the DS is asserted and updated. This
process is also iterative and continues until a well-formed DS is established.
648 DESIGN METHODS
Figure 6. Non-functional requirement analysis
5. Conclusion
A well formed requirements specification provides a solid basis for the subsequent design phases.
Design specifications are the most important artefacts in the product development process given that
they are based on a correctly formulated requirements specification. An approach has been described
in this paper to formalize the raw requirements that are usually expressed in narrative format. A formal
basis for the proposed approach was given taking into account the existing knowledge on requirements
management from the mechanical engineering discipline which were combined with current software
engineering practices, such as SysML, workflow diagrams and knowledge management. The use of a
knowledge rich requirement checklist makes requirement formulation more model-centric and helps
the designer to eliminate personal preferences improving this way his effectiveness and productivity.
The checklist based requirement specifications are also useful in improving discussion and
collaboration in a design team. In systems engineering, it will enhance the system developer to
identify and allocate responsibilities based on the checklist categories. The systematic gradual analysis
also helps to identify and emit requirements that have no direct bearing on the system functionality
and its essential characteristics. This formalized description is our first step towards a semi automated
requirements refinement process that will exploit semantic web technologies such as ontologies and
knowledge management. Since a first justification of our formal requirements procedure has been
established, future work will complete this analysis by focusing on the exploitation of machine
readable knowledge representation that will favour the partial automation of the requirements
formulization process.
Acknowledgement
This work is part of an ongoing research project “Hybridization of mobile work machines” (Hyblab) funded by
the Multidisciplinary Institute of Digitalization and Energy (MIDE) of Aalto University School of Science and
Technology.
References
Brace, W., Coatanéa, E., Kauranne, H., Heiska, M., “Early Design Modeling and Simulation of Behaviours:
Case study of mobile work machine “, Proceedings of ASME- IDETC 2009, San Diego, USA, 2009.
Cross, N., “Engineering Design Methods Strategies for Product Design” 3rd edition, WileyBlackwell, 2000.
Dorst, K.,Cross, N., “Creativity in the design process: co-evolution of problem–solution”. Design Studies, 22(5),
pp. 425–437, 2001.
Douglas, S., “Model-Driven Engineering”, IEEE Computer, vol. 39,no. 2, pp. 25-31, 2006.
DESIGN METHODS
649
Kusiak, A., ed., “Intelligent Design and Manufacturing”, Wiley, New York, 1992.
Otto, K., Wood K., “Product Design: Techniques in Reverse Engineering and New Product Development”,
Prentice Hall, Upper Saddle River, NJ, 2001.
Pahl, G., Beitz, W., “Engineering Designs: a Systematic Approach. Third Edition”, Springer-Verlag, London,
2007.
Restrepo, J., Christiaans, H., “Problem Structuring and Information Access in Design” J. of Design Research,
vol. 4, no.2, 2004.
Thramboulidis, K., “The 3+1 SysML View-Model in Model Integrated Mechatronics”, Journal of Software
Engineering and Applications (JSEA), vol.3, no.2, pp.109-118.
Tomiyama, T., Yoshioka, M., Tsumaya, A., "A Knowledge Operation model of Synthesis", in A. Chakrabarti
(ed.): Engineering Design Synthesis - Understanding, Approaches, and Tools, Springer-Verlag, London, 2002,
pp. 67-90.
Ullman, D., “The Mechanical Design Process”, 3
rd
edition McGraw Hill, New York, NY. 2002.
William Brace, MSc. Tech.
Research Scientist
Department of Engineering Design and Production
Faculty of Engineering and Architecture, Aalto University School of Science and Technology
Otakaari 4, 00076 AALTO, Finland
Telephone: +358 (0) 9 470 23451
Email: [email protected]
650 DESIGN METHODS
