Professional Documents
Culture Documents
OATAO is an open access repository that collects the work of Toulouse researchers and
makes it freely available over the web where possible.
Vareilles, Elise and Coudert, Thierry and Aldanondo, Michel and Geneste,
Laurent and Abeille, Joël System design and project planning: Model and
rules to manage their interactions. (2015) Integrated Computer-Aided
Engineering, vol. 22 (n°4). pp. 327-342. ISSN 1069-2509
Abstract. This article proposes a model and rules dealing with the management of the interaction between system design processes
and project planning ones. An industrial benchmark analysis has reinforced our belief that the interaction between the two
processes has to be supported by models, processes and relevant tools. Firstly, after presenting the results of the analysis, the
different entities are defined and the one-to-one relationship or bijection between the structure of the system and the structure
of the project is made. Then, a model, taking into account design activities and planning activities as well as management of
interactions, is proposed in compliance with existing project and design standards. A process of interaction is presented to carry
out design and project management. Two interaction modes have been proposed. On the one hand, the structural interaction
establishes links between entities of the two domains. On the other hand, the behavioral interaction (subject of this paper) is based
on the definition of states for each entity following feasibility and verification criteria, and can thus manage the changes between
states. Some rules are defined (precedence and synchronous rules) to forbid certain changes when they are inconsistent and to
synchronize them.
Keywords: System Design, Project Planning, Aiding Decisions, Knowledge based Systems
In this section, the background of our work is de- 3.2. Project Planning Process
fined and the structural coupling is synthesized with
regard to existing approaches. In Subsection 3.1, after The project planning domain of our work concerns
reviewing design process definitions, we define what the Project Time Management (PTM) process as de-
we mean by system design in this paper. In Subsection fined by the Project Management Institute [25]. In the
3.2, after reviewing planning process definitions, we proposed approach, the project planning definition is a
define what we mean by project planning. Then in Sub- top-down approach, where some kind of global plan-
section 3.3, after a literature review of interactions be- ning is achieved at a high level and is progressively
tween design and project processes, structural coupling detailed at lower levels by means of sub-projects. This
is synthesized. multi-level and multi-project approach makes it pos-
sible to perform adequate multi-level planning by si- – Simultaneously, axiomatic design identifies var-
multaneously considering, at all planning levels, differ- ious domains (Customer Needs, Functional Re-
ent objectives, constraints, degrees of aggregation, and quirements, Design Parameters and Process Vari-
capacity flexibility (see for instance [14] for a study ables) and sees them as interacting [36]. An ex-
on hierarchical multi-project planning). In order to de- ample of implementation is presented in [13]. The
fine a design project, we consider that project planning interactions between domains are clearly defined:
involves: not only design towards planning but also planning
towards design.
1. Project activities definition,
– Another approach, introduced by Gero [10], pro-
2. Resource and duration identification,
poses models based on three domains: Function,
3. Scheduling activities and resources, and
Behavior and Structure (FBS). The aim of this
4. If needed recursive decomposition at the lower
study is to take into account product behavior (ex-
level of some activities.
pected and effective) and to inventory in a formal
Scheduling of activities and resources is based on way eight sub-processes of design. However, tools
several techniques (see, for instance, [15] or [19]) that for interactions between processes are not explic-
are not detailed in this article. itly considered.
– A study that is very close to the problem addressed
3.3. Design and Planning Processes Interactions was undertaken by Stewart and Tate [34] who were
interested in the coupling of axiomatic design with
In this subsection, firstly, a literature review of inter- project planning in the case of software engineer-
actions between design and project processes is carried ing. Their idea was to associate design variables
out. Secondly, structural coupling, the basis of behav- with the tasks of the development process. This
ioral interaction, is synthesized. approach was implemented with an ad hoc devel-
opment coupled with the Microsoft Project® soft-
3.3.1. State of the art
ware package and tested in a software engineering
The axiomatic design and the above-mentioned stan-
context.
dards allow four interacting domains to be identified:
– The work of R. Lu [28] describes an approach cou-
1. The requirements or specifications, pling task management and design. The structure
2. The design solutions, of projects is represented by means of a Working
3. The tasks or activities, and Breakdown Structure (WBS) and is related to a
4. The resources. Product Breakdown Structure (PBS). A matrix rep-
resents the relationships between both domains.
The first two domains relate to the system design
– In [31] and [32], the authors proposed a Project
process and the last two domains to the project planning
Product Lifecycle Management approach (PPLM).
process. Of the few studies that address this coupling
The aim of their work is to develop a methodology
problem, one can mention:
and a software environment for integrating the
– The study of [20] explicated a link between PBS product that is being developed with the project as
(Product Breakdown Structure) and WBS (Work undertaken by the company.
Breakdown Structure). The author stated that the – In [5], the authors proposed a model to enable
WBS tree (or structure) is derived from that of the concurrent product design and assembly sequence
PBS. The isomorphism of PBS and WBS trees is planning. The aim of their work is to manage vi-
suggested. Other trees are introduced at the end tal yet complex and inherent product relationship
of the paper, such as a specification tree and a information to allow such concurrent product de-
drawing tree. sign and assembly sequence planning. This paper
– The studies initialized at M.I.T. [9] on the use of gives the detailed description of the background
methods and techniques used in product design and models which highlight the need for a more
in order to facilitate project design. These studies efficient PLM approach.
are the source of scientific developments around – In [11], the authors consider that a design project,
DSM (Design Structure Matrix), such as those of in mechanical system engineering, is a network of
Lindemann [22] where the interactions between various interacting design domains such as project,
the four identified domains are defined . product and process design. The paper presents
an object-oriented design methodology integrated 0,2..*
0..1
into a human-based and co-operative design life System 1 1
System
1 1..* Alternative
Requirements Solution
cycle. This methodology was implemented in an 1 1 1
educational web-based environment and applied 1 1
Requirement 1 1 1..* Alternative
to student design projects. Definition
1
Project
1
Development
Task 0,2..* Task
All these studies indeed confirm the four domains
1..*
(requirements, solutions, tasks and resources) and the
existence of causal links that engender interactions be- Fig. 1. Integrated Model of Structural Coupling
tween these four domains. However, except in [34],
[11], [28] and [32], no tools are provided to support or (Bill of Materials) of a system and the WBS (Work
aid interactions between both the design and planning Breakdown Structure) of a project appears to be the
processes. only comprehensive solution.
3.3.2. Structural Coupling Synthesis This bijective link is fully compliant with the Ax-
Structural coupling is based on a structural and hi- iomatic Design of Suh [35] and its zigzagging method-
erarchical decomposition of systems into sub-systems ology, with the work of R. Lu [28] which links PBS
and/or projects into sub-projects (guided by their intrin- and WBS and their relationship with a matrix, and with
sic complexity). Two coupled entities are then defined: the work of [31] and [32] who have proposed a Project
Product Lifecycle Management approach (PPLM). In
– A system entity is composed of one set of require-
our proposal and as shown in Fig. 1, we link also in a
ments and one or more solutions (or system alter-
bijective way and at any level in the hierarchy:
natives), as shown in the upper part of Fig. 1.
– A project entity is composed of one task of re- – A system S and a project P,
quirement definition and one or more tasks of so- – The set of system requirements SR and the require-
lution design (or an alternative development task), ment definition task PR,
as shown in the lower part of Fig. 1. – A system alternative SA and an alternative devel-
It is important to preserve the semantic of the enti- opment task PA.
ties of both domains without mixing the information. These entities are the foundations of structural and
Indeed, designers and managers have to keep their re- behavioral couplings (see [1], [3], [39] and [40] for
spective autonomy of decision-making. Once a deci- an overview of the meta-model supporting structural
sion is made on one side, they have to be informed coupling).
of any negative impacts (or potential problems) on the Regarding industrial benchmark results, the coupling
other side. In such a case, the current situation should between system design and project planning processes
be resolved in a style of mediation: engineers and man- does not only draw bijective links between system and
agers have to share their requirements, constraints and project entities. Structural coupling ensures an isomor-
goals in order to react appropriately and ensure the suc- phism between system and project structures, but this is
cess of the project. This mediation can be supported not sufficient to manage all their interactions. Managers
by the AHP method [29] in order to evaluate different also need to synchronize the two processes in order to
solutions and select the one that best suits system and plan and monitor the complete system design project.
project requirements at the same time. This synchronization is the main goal of the paper. Our
Furthermore, in this article we favor a progressive proposal is quite original:
top-down design approach where sub-projects and sub-
systems are defined only when required and when their – Firstly, the monitoring is done through two spe-
context is well known. For instance, the decision to cific attributes which qualify the feasibility and
subcontract the design of a particular sub-system can the verification of system and project entities,
be taken only after a suitable requirement analysis and – Secondly, these two attributes foster the control
when these requirements have been compared with of both processes as well as their synchronization
available resources. At this time, the choice of a sub- thanks to precedence and coupling rules, and
contractor and the definition of the required sub-project, – Thirdly, the synchronization is recursive: it is un-
as well as its characteristics and constraints, can be es- dertaken at each level and between two consecu-
tablished. Therefore, a bijective link between the BOM tive levels of the system and project hierarchies.
Our proposal is fully compliant with the Systems En- We have to precise that the attribute of feasibility or
gineering V-cycle. On one side, the feasibility attribute that of verification for the whole system S or for the
characterizes the left side of the V-cycle. It corresponds whole project P levels are not considered in our pro-
to the ability of being able a priori to reach the goals posals. Indeed, these attributes aggregate the informa-
regarding the current context (requirements, constraints, tion on the feasibility and verification and are closely
risks and uncertainty). On the other side, the verifica- dependent on those of SR and SA for the system S, and
tion attribute characterizes the right side of the V-cycle. on those of PR and PA tasks for the project P.
It corresponds to the achievement of a solution lead-
ing to the success of the project (in terms of quality, 4.1. Feasibility Attribute Definition
schedule and cost). Each time a system is decomposed
into sub-systems and a project into sub-projects, a new In preliminary design, the feasibility characterizes
V-cycle has to be conducted for each of them: the fea- the ability of a system, a product or a service to be
sibility and verification attributes have to be evaluated developed technically and economically. During the
for each sub-system and sub-project in order to have an feasibility evaluation phase, the design manager and the
overview of the progress of the whole project. planning manager have to determine if they can reach a
feasible solution with respect to their requirements, for
instance performance and technical characteristics for
4. Synchronous Coupling Attributes a system or cost and duration for a project.
In order to qualify the feasibility of each system and
Synchronous coupling attributes characterize the re- each project, we propose adding a specific attribute to
quirements SR and alternatives SA for the system S, each entity, noted as Fa. This attribute has three states
and on those of the requirement definition PR and alter- with different semantics:
native development PA tasks for the project P. These – UD, for undetermined and default value, mean-
coupling attributes help the program manager to better ing that the design manager or the planning man-
identify potential problems, as soon as possible and ager has not yet determined if, with respect to the
respond appropriately to deviations and changes. requirements, a solution seems achievable or not,
In the following sub-sections, we start with the defi- – OK, for feasible, meaning that it seems possible to
nition of two specific attributes which qualify the feasi- achieve a solution with respect to the requirements
bility and the verification of systems and projects. We il- for the design or for the planning,
lustrate their meaning for the system and for the project – KO, for unfeasible, meaning that it seems impos-
entities at any level of the decomposition. Secondly, we sible to achieve a solution with respect to the re-
set up the precedence relation that links them, before quirements for the design or for the planning.
proposing a synthesis.
In the rest of the paper, we note XhlxRi_xAj
where: In order to value the feasibility attribute xRni .F a, a
design manager or a planning manager firstly have to
– X can be any entity: SR, SA, PR or PA, collect information on their own context and secondly
– hl corresponds to the level of the entity in the to analyze it considering the degree of uncertainty and
hierarchy of systems or projects, the risk of failure. It is important to note that the evalua-
– xRi_xAj corresponds: tion of the feasibility attribute relies on the skill and ex-
pertise of the managers and their own subjective point
* for xRi to the id of requirement entities, such of view on the whole situation. For the design side,
as SR or PR,
when the needs and requirements have been gathered,
* for xAj to the id of alternative entities, such as the design manager has to analyze them, pre-design a
SA or PA.
solution and evaluate the risk of reaching an unfeasi-
For example, let us consider a landing gear system ble solution. For the project side, when the plan has
S11 composed of two sub-systems noted as S2111 for the been scheduled and the resources allocated, the plan-
wheel and S2112 for the brake, and the associated design ning manager has to analyze the plan and evaluate the
project P11 composed of two sub-projects noted as P2111 risk of reaching an unfeasible project. It is only after
for the wheel project and P2112 for the brake project. this phase that the managers can switch the value of
The structure of the system S11 and of the project P11 their attribute to feasible, xRni .F a = OK or unfeasible,
are illustrated in Fig. 2. xRni .F a = KO.
Landing gear System Landing gear Project
SR11 S11 PR11 P11
SA1_1
1 PA1_1
1
SR111
2 sub-sys S2111 SR112
2 sub-sys S2112 PR111
2 sub-pr P2111 PR112
2 sub-pr P2112
SA111_1
2 SA112_1
2 PA111_1
2 PA112_1
2
Fig. 2. Example of the Decomposition of a System into two Sub-systems and of a Project into two Sub-projects
For the design side, when it seems that a feasible mined ∀k ∈ [2, l], SRijk
n+1 .F a = UD. In the same way,
solution is reachable for the requirements, the designer for the project side, as long as the planning manager
can start working on the solution itself SAi_j n . While has not identified each sub-project requirement (esti-
(s)he has not described the functioning principles and mated resource workload, duration, budget, . . . ), has
possibly decomposed his/her system into sub-systems not scheduled it and is not convinced of the feasibil-
and identified their requirements, (s)he cannot deter- ity of his/her decomposed alternative development task
mine the feasibility of the solution. For the project side, PAi_j
n .F a 6= OK, the feasibility of each sub-project
when it seems that a feasible project is reachable for the requirement definition task cannot be determined: the
plan and resource allocation, the planning manager can feasibility attribute of each of the l sub-project re-
start working on the alternative development task PAi_j n . quirement definition tasks is therefore undetermined:
When (s)he has scheduled and possibly decomposed ∀k ∈ [2, l], PRijk
n+1 .F a = UD.
it into sub-projects, (s)he can switch the value of this We can see that a relation of precedence exists be-
attribute to feasible, P Ai_j
n .F a = OK or unfeasible, tween the attributes of feasibility of a decomposed solu-
P Ai_j
n .F a = KO. tion xAi_j
n and of the requirements of each sub-system
ijk
We can see that a relation of precedence exists be- xRn+1 . Therefore, we set up the rule dr1:
tween the attributes of feasibility of xRni and of each
potential solution xAi_j
n . Therefore, we set up the rule
r1: dr1 : xAi_j
n .F a 6= OK
ijk
⇒ ∀k ∈ [2, l], xRn+1 .F a = UD (2)
r1 : xRni .F a 6= OK
⇒ ∀j ∈ [1, m], xAi_j
n .F a = UD (1) In Fig. 3 and 4, the relation described by the rule dr1
(eq. 2) between the feasibility attributes of a decom-
In Fig. 3 and 4, the relation described by the rule posed entity and two sub-entities is synthesized.
r1 (eq. 1) between the feasibility attributes is synthe- With regard to the association of these attributes with
sized. With regard to the association of these attributes, the decomposed entity xAi_1 1 and its two sub-entities
and without considering any rules or relations, there X2i11 and X2i12 , and without considering any rules or
are 3 ∗ 3 = 9 possible combinations for each pair of relations, there are 3 ∗ 3 ∗ 3 = 27 possible combinations.
xRni and xAi_j n . Considering the rule r1 (eq. 1), only 5 Considering the rule dr1 (eq. 2), only 11 combinations
combinations are permissible. are permissible.
Let us consider now the case of a decomposed solu-
tion into l sub-systems and l sub-projects. For the de- 4.2. Verification Attribute Definition
sign side, while the designer has not identified each sub-
system requirement and is not convinced of the feasi- The definition of the verification comes from the ISO
bility of his/her decomposed solution SAi_jn .F a 6= OK, 9000:2000 standard [17]. After a complete design, the
the feasibility of each sub-system requirement can- verification confirms through the provision of objective
not be determined: the feasibility attribute of each of evidence that the specified requirements have been ful-
the l sub-systems requirements is therefore undeter- filled. During the verification phase, the design man-
ager and the planning manager have to justify that the
solution matches their requirements to the letter. We can remark that a task PRin that has not been
In order to qualify the verification of each system properly completed with respect to its requirements
and each project, we propose adding a specific attribute PRin .V e = KO does not compromise the rest of the
to each entity, noted as Ve. This attribute has three states complete project: the project can still continue but its
with different semantics: requirements and objectives (estimated resource work-
load, duration, budget, . . . ) have to be reassessed in
– UD, for undetermined and default value, mean-
order to achieve the design.
ing that the design manager or the planning man-
In Fig. 4, the relation described by the rule r2 (eq. 3)
ager has not yet proven that the solution exactly
between the verification attributes is synthesized. With
matches the requirements to the letter or not,
regard to the association of these attributes, and without
– OK, for verified, meaning that the solution fulfills
considering any rules or relations, there are 3 ∗ 3 = 9
the requirements to the letter for the design or for
possible combinations for each pair of PRin and PAi_j n .
the planning,
Considering the rule r2 (eq. 3), only 4 combinations
– KO, for unverified, meaning that the solution does
not match at least one requirement for the design are permissible.
or for the planning. Let us consider now the case of a decomposed so-
lution into l sub-systems and l sub-projects. For the
The system requirements entity is composed of needs design side, when the design of an alternative has been
(expression of the stakeholders’requirements or the completed, the design manager has to verify its validity
specifications stemming from the upper level if it exists) according to the requirements, SAi_j n .V e. As long as
formalized by means of text or expressed as specifica- (s)he has not tested the functioning principles and po-
tions (both functional and technical) that are declined tentially integrated all the sub-systems into the system,
into technical requirements by the designers. As regards (s)he cannot determine the verification of the solution.
the definition of the verification, we do not associate The design manager has to wait for the verification of
a verification attribute with the system requirements. each sub-system and for its integration in order to deter-
Indeed, in this case, it implies verifying that the re- mine the verification of the decomposed solution. If one
quirements match the needs expressed by the client, the of the sub-systems is not verified ∃SAijk_p
n+1 .V e = KO,
stakeholders or the upper level. We assume that this is de facto the complete solution cannot be either. But the
the case and that a large majority of the specifications fact that all the sub-systems have been verified ∀k ∈
have already been formalized and tested against the [2, l], ∀p ∈ [1, m], SAijk_p
n+1 .V e = OK, is not enough
upper-level needs. to determine if the decomposed solution matches the
For the project side, as regards verification, until a requirements to the letter or not because a problem can
task ends, a planning manager cannot verify that it has still occur during the integrations and tests.
run well: the verification attribute of all the tasks is In the same way, for a decomposed alternative de-
therefore undetermined: PRin .V e = UD ∧ PAi_j n .V e = velopment task, the project manager has to wait for
UD. After a task has been carried out, the planning man- the verification of each sub-project and for the in-
ager has to verify its validity compared to the project tegration of its information, such as resource work-
requirements: resource workload, real-time consump- load, real time-consumption, or budget in order to de-
tion, budget, . . . When (s)he has analyzed and if nec- termine the verification of the decomposed alterna-
essary, integrated all the information coming from the tive development task PAi_j n .V e. If one of the sub-
sub-projects, (s)he can verify the task and switch the projects is not verified ∃PAijk_p
n+1 .V e = KO, de facto
value of its attribute to verified or unverified. A relation the complete alternative development task cannot be
of precedence exists between the attributes of verifica- either. But the fact that all the sub-projects are veri-
tion of these two tasks: we cannot verify an alternative fied ∀k ∈ [2, l], ∀p ∈ [1, m], PAijk_p
n+1 .V e = OK, is not
development task before a requirement definition task. enough to determine if the decomposed alternative de-
We set up the rule r2: velopment task matches its requirements to the letter or
not because a problem can still occur before its end.
r2 : PRin .V e 6= OK We can see that a logical and precedence relation ex-
ists between the attributes of verification of the l lower-
⇒ ∀j ∈ [1, m], PAi_j
n .V e = UD (3) level entities xAijk_p
n+1 and of the decomposed entity
itself xAi_j
n . Therefore, the rule dr2 can be applied: 4.4. Feasibility and Verification Attributes Synthesis
dr2 : ∀k ∈ [2, l], ∀p ∈ [1, m], Seven synchronous attributes SRin .F a, SAi_j n .F a,
SAi_j i i_j i i_j
n .V e, PRn .F a, PAn .F a, PRn .V e and PAn .V e
ijk_p
i_j
∃xAn+1 .V e = KO ⇒ xAn .V e = KO
ensure consistency in the design process by imposing
or precedence relations, through three precedence rules r1
∀xAijk_p i_j (eq. 1), r2 (eq. 3) and r3 (eq. 5), one top-down prece-
n+1 .V e = OK ⇒ xAn .V e 6= UD
(4) dence rule dr1 (eq. 2) and one bottom-up logical and
precedence rule dr2 (eq. 4), as synthesized in Fig. 3
and 4.
In Fig. 3 and 4, the relation described by rule dr2 (eq.
For the design side, without considering any rules or
4) between the verification attributes of a decomposed
relations, there are 3 ∗ 3 ∗ 3 = 27 possible combinations.
entity and of two lower-level entities is synthesized.
Considering the precedence relations r1 and r3, and
With regard to the association of these attributes with
placing ourselves at a single decomposition level, there
the decomposed entity xAi_1 1 and its two sub-entities
X2i11 and X2i12 , and without considering any rules or are 7 possible combinations.
relations, there are 3 ∗ 3 ∗ 3 = 27 possible combinations. For the project side, without considering any rules
Considering rule dr2 (eq. 4), only 11 combinations are or relations, there are 3 ∗ 3 ∗ 3 ∗ 3 = 81 possible com-
permissible. binations. Considering the precedence relations r1, r2
and r3, and placing ourselves at a single decomposition
4.3. Feasibility and Verification Links level, there are 15 possible combinations.
Taking into account both sides and without consider-
The possible states of this attribute V e depends on ing any coupling rules, there are 27 ∗ 81 = 2187 pos-
the previous value F a: an entity X cannot be verified sible combinations. If only the precedence relations r1
if it has not been previously feasible. Therefore, we set (eq. 1), r2 (eq. 3) and r3 (eq. 5) are considered, there
up the rule r3: are 7 ∗ 15 = 105 possible combinations.
dr1 dr2
r1 r3
dr1 SRi11
2 .F a SAi11_1
2 .F a SAi11_1
2 .V e dr2
r1 r3
SRi12
2 .F a SAi12_1
2 .F a SAi12_1
2 .V e
r2
r1
r3 r3
PRi1 .Fa PRi1 .Ve PAi_1
1 .F a PAi_1
1 .V e
dr1 dr2
r3 r3
dr1 PRi11
2 .F a PRi11
2 .V e PAi11_1
2 .F a PAi11_1
2 .V e dr2
r1
r2
r3 r3
PRi12
2 .F a PRi12
2 .V e PAi12_1
2 .F a PAi12_1
2 .V e
r1
r2
and cO2 for the alternative development task and the cation of the PRin .V e task. This relation is formalized
system alternative entity as:
Secondly, synchronization of the two processes is Let us consider the pair (system alternative SAni_j ,
needed when one of them has reached a solution. Let alternative development task PAi_j n ), starting with the
us consider the pair (system requirements SRin , require- system design side. When the whole solution has been
ment definition task PRin ). Let us start with the system completely designed, the design manager has achieved
design side. When all the system requirements have his/her job and can verify the consistency of the solution
been gathered and analyzed, the design manager has with regard to the requirements. If the solution matches
achieved his/her job and can deliver his/her conclusions the requirements SAi_jn .V e = OK, the corresponding
on the risk of reaching an unfeasible solution. If the risk task PAi_j i_j
n has to end PAn .V e 6= UD. On the other
is weak SRin .F a = OK, the corresponding task PRin i_j
hand, if SAn .V e = KO, it will be necessary to modify
has to end PRin .V e 6= UD. On the other hand, where the requirements in order to finish the design. In this
SRin .F a = KO, the requirements have to be modified case, the system design comes back to its initial state:
in order to continue the design. In this case, the system SRi_j i_j
n .F a = UD∧SAn .F a = UD∧SAn .V e = UD
i_j
design comes back to its initial state: SRin .F a = UD and the project side is necessarily impacted by this
but the project side is not necessarily impacted by this modification: tasks PRin and PAi_jn have to be restarted
modification. A renegotiation of the requirements with and lengthened. The project keeps its time consumption
regard to the work already done can be supported by but the complete process has to be carried out once
the AHP method [29]. Let us now look at the planning more: PRin .F a = UD ∧ PRin .V e = UD ∧ PAi_j n .F a =
project side. When the task PRin is nearing its end, all UD ∧ PAi_jn .V e = UD. Let us now start with the plan-
the time has been consumed (PRin .V e 6= UD), so a ning project side. When task PAi_j n is nearing its end
decision has to be made on the feasibility of the system and all the time has been consumed (PAi_j n .V e 6= UD),
regarding the requirements. These two analyses allow the system alternative has to be verified with regard to
us to deduce that a relation exists between the feasibil- the requirements and then delivered. If a solution has
ity of the system requirements SRin .F a and the verifi- not been designed, the complete project can start once
more. These two analyses allow us to deduce that a rela- – The coupling process between entities on the same
tion exists between the verification of the system alter- decomposition level (paragraphs [A], [B], [C] and
native SAi_j i_j
n .V e and of the verification of the PAn .V e [D]);
task. This relation is formalized as: – The coupling process between entities between
two successive levels (paragraphs [E] and [F]);
– The fact that a project entity which has not been
properly completed does not compromise the de-
c2 : SAi_j i_j
n .V e 6= UD ⇔ PAn .V e 6= UD (9) sign project (paragraph [G]);
– That this is not the case for a system entity (para-
The goal of ensuring a better consistency between graph [H]).
system design and project planning has now been
reached through two specific attributes and nine In the proposed scenarios, we follow the succession of
rules. With regard to the seven entities (SRin .F a, states 1, 2, 8, 12, 19 and finally 24 for the first scenario
SAi_j i_j i i i_j
n .F a, SAn .V e, PRn .F a, PRn .V e, PAn .F a and
and 27 for the second one in Fig. 6.
i_j [A] First of all, a program manager is appointed.
PAn .V e), there are 2187 possible combinations (cf.
4.4). Considering now only the precedence rules r1 (eq. (S)he starts creating a project entity and via the struc-
1), r2 (eq. 3) and r3 (eq. 5), there are 105 out of 2187 tural coupling, six entities are automatically created
possible combinations (cf. 4.4). Considering now all P11 , PR11 and PA1_11 for the project side and S11 , SR11
1_1
the precedence rules, and the coupling rules cO1 (eq. and SA1 for the system side, and matched {(S11 , P11 ),
6), cO2 (eq. 7), c1 (eq. 8) and c2 (eq. 9), the number (SR11 , PR11 ), (SA1_1 1_1
1 , PA1 )}. All of these entities have
of possible combinations is reduced from 105 down to their attributes of feasibility and of verification to UD,
27, as shown in Fig. 6. Only these 27 combinations are state 1 in Fig. 6.
allowed by the proposed coupling in order to monitor Then the program manager appoints the design man-
and synchronize these two processes. ager and the planning manager. Once this is done, (s)he
In all the dead-end states (states 3, 6, 7, 9, 11, 13,14, gives them their orientations (objectives and require-
15, 17 and 18), a problem that can jeopardize the suc- ments) and defines their decision frames; for instance,
cess of the project, had occurred. The design manager the global budgets (for design and project), the time
and the project manager have to find a common solu- allotted to conduct the global project and the quantity
tion that best suits system and project requirements at of resources available. In our example, a 50-kilogram
the same time to solve it. When the solution has been landing gear system has to be designed in 22 time units.
found, the process comes back to the initial state (state [B] The project side must be considered first. After
1) in Fig. 6, while keeping time, budget and resource collecting the information relevant to his/her context
consumption, for the project planning side, in order to and exchanging views with all the people involved in
consolidate these information at the project end. his/her project (design and planning sides), the planning
manager starts estimating the duration, allocating the
resources and scheduling the requirement definition
task PR11 . In our example, (s)he allocates 2 time units
6. Coupling Process Experimentation
to task PR11 and 16 units to task PA1_1 1 . Two units are
needed to verify project P11 , and in order to reduce the
This section highlights our proposals for coupling
risk of unfeasibility, a margin of 2 units is taken.
(structural and synchronous) with the simplified but re-
When the plan has been scheduled and the resources
alistic example of a landing gear system (cf. section 4).
allocated, it appears to the planning manager that this
This landing gear system is composed of one potential
task PR11 is viable: (s)he switches its feasibility attribute
solution for level 1 decomposed into two sub-systems
to OK, PR11 .F a = OK, state 2 in Fig. 6. During state
(a wheel and a brake) and two sub-projects (a wheel
2:
project and a brake project) for level 2, as shown in Fig.
2. In order to value the feasibility and the verification – Rule r3 (eq. 5) is applied and the requirement
attributes, only two requirements are used: the weight definition task can be now analyzed and verified.
for the system side and the duration for the project side. – Rule r1 (eq. 1) is applied and in our case, the
Four typical coupling situations, belonging to a sce- planning manager concentrates on the alternative
nario, are described and show: development task PA1_1 1 . (S)he does exactly the
r1 r3
SRin .F a SAi_j
n .F a SAi_j
n .V e
c1
cO1 cO2 c2
PRin .F a PRin .V e PAi_j
n .F a PAi_j
n .V e
r3 r3
r1
r2
1 legend
UD UD UD SR.F a SA.F a SA.V e
UD UD UD UD PR.F a PR.V e PA.F a PA.V e
2 3
UD UD UD UD UD UD
OK UD UD UD KO UD UD UD
6 5 4 7 8 9
KO UD UD OK UD UD OK UD UD KO UD UD UD UD UD UD UD UD
OK OK UD UD OK OK UD UD OK KO UD UD OK KO UD UD OK UD OK UD OK UD KO UD
11 10 13 12 14 15
OK UD UD OK UD UD OK UD UD OK UD UD KO UD UD KO UD UD
OK OK KO UD OK OK OK UD OK KO KO UD OK KO OK UD OK OK OK UD OK KO OK UD
16 17 18 19
OK OK UD OK KO UD OK KO UD OK OK UD
OK OK OK UD OK OK OK UD OK KO OK UD OK KO OK UD
20 24
OK OK OK OK OK OK
OK OK OK OK OK KO OK OK
21 25
OK OK KO OK OK KO
OK OK OK KO OK KO OK KO
23 22 26 27
OK OK KO OK OK OK OK OK OK OK OK KO
OK OK OK OK OK OK OK KO OK KO OK KO OK KO OK OK
same job: duration estimation, resource allocation state 8 in Fig. 6. At this moment, this means that the
and schedulling. whole project planning (requirement definition task
– Rule cO1 (eq. 6) is applied, so the designers are PR11 and alternative development task PA1_11 ) are both
now allowed to start working and collecting the feasible with no conclusion on the feasibility of the
system needs and requirements. system requirements SR11 . During state 8:
[C] It appears to the planning manager that the al- – Rule r3 (eq. 5) is applied and only after the verifi-
ternative development task PA1_1 1 is also viable: (s)he cation of the requirement definition task PR11 (rule
switches its feasible attribute to OK, PA1_1
1 .F a = OK, r1 (eq. 1)), can task PA1_1
1 be verified.
xRi xRi
– The coupling rule cO2 (eq. 7) is applied. But as for each pair of (sub-system Shl , sub-project Phl :
111 111 112 112
the design manager has not yet determined the fea- {(S2 , P2 ), (S2 , P2 }).
sibility of the system requirements, the designers [F] When the design team has described the function-
cannot start working on a solution (rule r1, eq. 1 ing principles of solution SA1_1 1 and given the require-
has not yet been applied to the system side). ments of each of the sub-systems SR111 2 and SR112
2 , the
feasibility of the solution can be determined. The de-
[D] When the needs and requirements have been col-
sign manager specifies the weight of the sub-systems:
lected and analyzed (a 50-kilogram landing gear sys-
35 kg for the axle beam assembly (not considered in our
tem designed in 22 time units), it seems to the design
example), 10 kg for the wheel and 5 kg for the brake.
manager that a solution is reachable: (s)he switches the
In our scenario, we consider that, regarding the require-
feasibility attribute of the system requirements entity to
ments and the decomposition into two sub-systems, a
feasible SR11 .F a = OK, state 12 in Fig. 6. Simultane-
solution is reachable: the feasibility attribute of the al-
ously, the coupling rule c1 (eq. 8) is applied and forces
ternative is switched to feasible SA1_1 1 .F a = OK, state
the corresponding task to end, PR11 .V e 6= UD. In our
19 in Fig. 6. Rule r3 (eq. 5) is applied but the design
scenario, as this task does not respect its requirements
manager has to wait for the verification of each of the
(for instance, the design team has spent too much time
sub-systems in order to verify the system alternative
collecting and analyzing the system requirements, 3
validity.
units instead of 2), this attribute is switched to unveri-
[G] When the design of each sub-system , sub − sys
fied, PR11 .V e = KO, by the planning manager, state 12
S2111 and sub − sys S2112 , has been completed, the de-
in Fig. 6. But the fact that this task has consumed more
sign manager has to integrate all the sub-systems, test
than expected is not critical for the whole project. The
the functioning of the system alternative SA1_1 1 , and
planning manager has to negotiate new objectives and
switch its verification attribute to OK or KO depend-
requirements with the program manager. This negoti-
ing on the results. In our first scenario, all sub-systems
ation can involve the design manager in order to take
are verified SA111_1
1 .V e = OK and SA112_1 1 .V e = OK
into account design problems and constraints. If the
and their integration matches the requirements to the let-
negotiation is a success, the project continues, as in our
ter (9.9 kg for the wheel and 5 kg for the brake): the veri-
scenario ; otherwise, it is stopped. In our example, the
fication attribute of the system alternative is switched by
negotiation has led to splitting the margin as follows:
the design manager to verified SA1 .V e = OK, state 24
1 unit for task PR11 and 1 unit for task PA1_1 1 . During in Fig. 6. The coupling rule c2 (eq. 9) is then applied and
state 12:
forces the corresponding task to end PA1_1 1 .V e 6= UD.
– Rule r1 (eq. 1) can be applied: as rule cO2 (eq. 7) In our scenario, the task respects its requirements to
has already been applied, the design team can now the letter (17 units); its verification attribute is switched
start designing the system alternative SA1_1
1 . to verified PA1_1
1 .V e = OK, state 24 in Fig. 6. At this
– Rule r2 (eq. 3) is applied and the alternative de- stage, all the feasibility and verification attributes on
velopment task PA1_11 can now be verified. both sides have been determined, and the system design
project is a success.
[E] In our first scenario, we consider that the system
[H] In our second scenario, the verification attribute
alternative SA1_1
1 is too complex to be designed in a sim- of the system alternative is switched to unverified
ple way, so the design manager has to split it into several
SA1 .V e = KO, state 27 in Fig. 6. This switch can
sub-systems. In our example, the landing gear system
come:
is decomposed into a wheel and a brake. In this case,
firstly, the design manager decomposes his/her system – From an integration problem of verified sub-
alternative into x sub-systems (in our case, SA1_1 1 is systems SA111_11 .V e = OK and SA112_1
1 .V e =
split into two sub-systems sub−sys S2111 and sub−sys OK leading to a functioning problem. For instance,
S2112 ). Secondly, via the structural coupling, the project the wheel weighs more than required: its weight
is decomposed in the same way into x sub-projects (in has not been verified properly by the sub-system
our case, PA1_11 is split into two sub-projects sub − pr design manager in charge of its design and it has
P2111 and sub−pr P2112 ) that the planning manager has been qualified as verified SA111_1
1 .V e = OK. In
to analyze with regard to her/his project requirements this case, it is the system design manager who has
(time and availability of resources). A complete cou- to switch this attribute to unverified SA1 .V e =
pling process has to be restarted for this new level and KO ;
– Or from a design problem, when it is impossi- tities (discussed in section 3) is the base of the pro-
ble for the sub-system design manager to reach posed structural coupling. The model that supports cou-
a feasible solution regarding the requirements. pling groups together project entities (PRin and PAi_jn ),
For instance, it is not possible to design a wheel systems entities (SRin and SAi_jn ) and coupling rules.
with such a low weight. The sub-system design In section 4, we have proposed and defined specific
manager has switched the verification attribute attributes which are used to develop synchronous cou-
SA1111_1 .V e to KO. In such a case, the switch to pling: feasibility and verification attributes. These at-
unverified is done automatically through rule dr21 tributes allow states to be defined as entities during
(eq. 4). their life cycle and make it possible to control their
consistency throughout the design of the whole system.
The coupling rule c2 (eq. 9) is then applied and forces
Three precedence rules r1 (eq. 1), r2 (eq. 3) and r3 (eq.
the corresponding task to end PA1_1 1 .V e 6= UD. The 5) on a single level of decomposition, as well as two
task respects its requirements to the letter (less than 17
hierarchic rules dr1 (eq. 2) and dr2 (eq. 4) have been
units), so its verification attribute is switched to verified
set up between system attributes or project attributes. In
PA1_1
1 .V e = OK, state 27 in Fig. 6. At this stage, all section 5, synchronous coupling is described by iden-
the feasibility and verification attributes on both sides
tifying different states and transitions between these
have been determined. The project is interrupted be-
states for entities, and by defining some rules which
cause there is a problem on the system design side. Two
guarantee the consistency of the transitions. In other
options are possible:
words, the proposed coupling synchronizes the project
– Either the complete project ends; planning process with the system design process while
– Or a discussion, involving the design manager, the preserving consistency between the changing of states.
project manager, the program manager and if nec- It authorizes certain attribute changes and forbids oth-
essary, all the stakeholders, has to be conducted in ers, leading to better planned and controlled design
order to redefine the objectives, requirements and tasks. Firstly, two precedence coupling rules cO1 (eq.
constraints. This discussion can lead, for instance, 6) and cO2 (eq. 7) order the two processes: the feasi-
to enlarging the budget of the project, to fixing bility of project entities must be qualified before those
a new deadline, to allocating more resources, or of the system. Secondly, two synchronized coupling
to changing some of the requirements, etc. If the rules c1 (eq. 8) and c2 (eq. 9) synchronize them: none
discussion comes to a compromise, the process of the processes prevails on the other but a change in
of feasibility and verification must be carried out one or the other environments has a strong impact on
again: the process comes back to the initial state the other (for instance, the delivery of a system implies
(state 1 in Fig. 6 where all the attributes are val- the end of the associated task and vice-versa). Finally,
uated to UD). But in this case, the time, budget in section 6, a coupling scenario is described in order
and resource consumption has to be kept (and not to illustrate the proposal.
reset) in order to consolidate the relevant infor- The proposed synchronous coupling is quite simple
mation at the end of the project. For instance, in but very powerful. The formalization of the behavioral
our example, a new weight distribution between interaction, which no previous scientific paper has made
sub-systems can be decided on. explicit, is unquestionably original. The nine identified
rules have been set up thanks to the fifteen company in-
terviews. In some specific situations, some of them can
7. Conclusion be needless. Of course, the less the coupling rules are
applied, the more the resulting synchronous coupling
The aim of this article is to propose a model and state diagram has states compared to the one presented
rules capable of supporting a coupling between the sys- in Fig. 6.
tem design process and the project planning process. These rules have been integrated in the ATLAS
The context and the background of the study has been IT platform which has been developed in order to
presented in section 2 and 3. The multi-level design test and validate our proposals. Some tutorials and
process and planning process have been described in the link to the prototype can be found on the follow-
accordance with academic standards and with standards ing webpage http://perso.mines-albi.fr/
used in companies. The natural and logical bijective ~vareille/, heading "Projects and industrial part-
relationships between system entities and project en- ners" and project "ATLAS".
It must be clear that according to the industrial devel- [9] S. D. Eppinger, D. E. Whitney, R. P. Smith D. A. Gebala, Orga-
opment context, other rules could be added. This opens nizing the tasks in complex design projects., MIT Workshop on
CAOPD, Springer-Verlag New York, Inc. 1991.
interesting issues leading to a kind of customization
[10] J. S. Gero, Design prototypes: a knowledge representation
of the coupling process. Structural and synchronous schema for design, Artificial Intelligence magazine. Vol. 11 No.
couplings can be added, without any difficulty, to PLM 4, 1990.
software. In this case, engineers and managers would [11] S. Gomes and J-C. Sagot, A concurrent engineering experience
be forced to discuss together and to find a compromise based on a cooperative and object oriented design methodology,
Best papers book, 3rd International Conference on Integrated
acceptable to both sides, before a project failure occurs. Design and Manufacturing Engineering, Kluwer Academics Pub-
Structural and synchronous couplings could help them lisher, pp. 11-18, 2002.
to identify problems as early as possible and thus make [12] E. Gómez de Silva Garza, M. L. Maher, Design by interactive
better decisions by considering the overall situation. exploration using memory-based techniques, Knowledge-Based
Perspectives of this work concern the integration of System Journal, Vol. 9, No. 3, pp. 151-161, http://dx.doi.
org/10.1016/0950-7051(95)01016-5, 1996.
local decoupling mechanisms between both domains [13] A. Goncalves-Coelho, Axiomatic Design and the Concurrent
when required at lower levels of decomposition. In Engineering Paradigm, Proceedings of COSME, Brasov, Roma-
this case, the coupling could be carried out manually nia, 2004.
(the program manager explicitly constructs the links [14] H. Hans, W. Herroelenb, R. Leusb, G. Wullink, A hierarchical
approach to multi-project planning under uncertainty, The Inter-
between entities) and all the identified couplings con-
national Journal of Management Science, Vol. 35, pp. 563 - 577,
tinue to work. Alternatively, no coupling is performed 2005.
and is thus no longer supported by the integrated model. [15] W. Herroelen, B. De Reyck, E. Demeulemeester, Resource-
Clearly, this method of operation needs to be checked constrained project scheduling: a survey of recent developments,
and reserved for specific situations where designers of Computers and Operations Research, Vol. 25, No. 4, pp. 279-302,
1998.
low levels want to decompose a very simple system to [16] J. Huysentruyt, D. Chen, Contribution to the development of
develop it without a framework given by the planning a general theory of design, Proceedings of the 8th International
environment. Conference of Modeling and Simulation - MOSIM’10 - May
10-12, 2010 - Hammamet - Tunisia, 2010.
[17] ISO 9000:2000, Quality management systems - Funda-
mentals and vocabulary, International Organisation of Stan-
References dardization, http://www.iso.org/iso/catalogue_
detail?csnumber=29280, 2005.
[1] J. Abeille, T. Coudert, E. Vareilles, L. Geneste, M. Aldanondo, [18] A. Karnie and Y. Reich, Managing the Synamic of New Product
T. Roux, Formalization of an integrated system/project design Development Processes. A new Product Lifecycle Management
framework : first models and processes, SPRINGER , pp.207- Paradigm, Springer, P. 13, 2012.
217, ISBN 978-3-642-15653-3, 1st International Conference on [19] T. Kis, Project scheduling: a review of recent books, Operations
Complex Systems Design and Management CSDM2010, Paris, Research Letters, Vol. 33, Issue 1, pp. 105-110, 2005.
FRANCE, 2010. [20] M. Lamers, Do you manage a project, or what? A reply to
[2] L. Albano, N. Suh, Axiomatic approach to structural design, "Do you manage work, deliverables or resources", International
Research in Engineering Design, Vol. 4, pp. 171-183, 1992. Journal of Project Management, Vol. 20, Issue 4, pp. 325-329,
[3] M. Aldanondo, E. Vareilles, M. Djefel, Towards an association 2002.
of product configuration with production planning, International [21] P. Lawrence and J. Scanlan, Planning in the Dark: Why Major
Journal of Mass Customisation 2010 - Vol. 3, No. 4 pp. 316 - Engineering Projects Fail to Achieve Key Goals, Journal of Tech-
332, 2010. nology Assessment and Strategic Management, Vol. 19, Issue 4,
[4] L. T. M. Blessing, Comparison of Design Models Proposed in pp. 509-525, 2012.
Prescriptive Literature, Proceedings of the COST A3 / COST [22] U. Lindemann, A vision to overcome "chaotic" design for X pro-
A4 International Research Workshop on the role of design in the cesses in early phases, Int. Proc. of Conference on Engineering
shaping of technology, Lyon, 1996. Design (ICED), Paris, France, 2007.
[5] F. Demoly, X.-T. Yan, B. Eynard, S. Gomes and D. Kiritsis, [23] J. N. Martin, Evolution of EIA-632 from an interim standard
Integrated product relationships management: a model to enable to a full standard, Proceedings of INCOSE 1998 Symposium,
concurrent product design and assembly sequence planning, In 1998.
Journal of Engineering Design, Vol. 23, pp. 544-561, 2012. [24] G. Pahl, W. Beitz, Engineering Design: a Systematic Approach,
[6] G. E. Dieter, Engineering design - A materials and processing Springer-Verlag, 1996.
approach, Mc Graw-Hill International Editions, 2000. [25] PMBOK Guiden A Guide to the Project Management Body of
[7] I. Dörfler and O. Baumann, Changing Organizational Routines in Knowledge, Fifth Edition, Project Management Institute, 2013.
Response to a Drastic Failure: The Case of Airbus A380 Program, [26] G. Pol, C. Merlo, J. Legardeur, G. Jared, Analysing collabo-
Available at SSRN 2123297, 2012. rative practices in design to support project managers, Interna-
[8] EIA Standard, Processes for Engineering a System, Electronic tional Journal of Computer Integrated Manufacturing, Taylor &
Industries Alliance, 1999. Francis, Vol. 20, No. 7, pp. 654-668, November 2007.
[27] H. Li, Y. Fan, C. Dunne, P. Pedrazzoli, Integration of business [36] N. Suh, Axiomatic Design: Advances and Applications, Oxford
processes in Web-based collaborative product development, Inter- Series, 2001.
national Journal of Computer Integrated Manufacturing, Taylor [37] M. X. Tang, A knowledge-based architecture for Intelligent
& Francis, Vol. 18, No. 6, pp. 452-462, September 2005. Design Support, International Journal of Knowledge Engineering
[28] R. Lu, W. Peng, C. Wang, Integration of Product Design Pro- Review, vol. 12 (4), pp.387-460, 1997.
cess and Task Management for Product Data Management Sys- [38] D. G. Ullman, The mechanical design process, third edition,
tems, in proceedings of IFIP International Federation for Infor- McGraw-Hill, Higher Education, New York, 2003.
mation Processing, Volume 254, Research and Practical Issues [39] É. Vareilles, M. Aldanondo, M. Djefel, P. Gaborit, Coupling
of Enterprise Information Systems II Volume 1, eds. L. Xu, Tjoa interactively Product and Project Configuration: a Proposal us-
A., Chaudhry S. (Boston: Springer), pp. 207-218, 2007. ing Constraints Programming, International Mass Customization
[29] T. L. Saati, Engineering Design: a Systematic Approach, RWS Meeting (IMCM) and International Conference on Economic,
Publications, 3rd Revised edition, 2012. Technical and Organizational Aspects of Product Configuration
[30] H. A. Simon, The science of the artificial, MIT Press, 1969. Systems (PETO), 2008.
[31] A. Sharon, V. Perelman, D. Dori, A Project-Product Lifecy- [40] É. Vareilles, T. Coudert, M. Aldanondo, L. Geneste, J. Abeille,
cle Management Approach for Improved Systems Engineering Coupling system design and project planning: discussion on a
Practices, Proc. Eighteenth Annual International Symposium of bijective link between system and project structures, forthteenth
the International Council on Systems Engineering (INCOSE), edition of IFAC’s triennal symposium on Information Control
Utrecht, the Netherlands, 2008. Problems in Manufacturing (INCOM), 2012.
[32] A. Sharon, O. de Weck, D. Dori, Is there a Complete Project [41] J. X. Wang, M. X. Tang, B. Gabrys, An Agent-Based System
Plan? A Model-Based Project Planning Approach, Proceedings Supporting Collaborative Product Design, R.J. Howlett, L.C. Jain
of the Nineteenth Annual International Symposium of the Inter- (Eds.): KES 2006, Part II, LNAI 4252, pp. 670 - 677, Springer-
national Council on Systems Engineering (INCOSE), Singapore, Verlag Berlin Heidelberg, 2006.
2009. [42] H. Yoshikawa, Design philosophy: the state of the art, CIRP
[33] B. Shore, Systematic biases and culture in project failures, annals manufacturing technology, 38(2) : pages 579-586, 1989.
Project Management Journal, Vol. 39, Issue 4, pp. 5-16, 2008. [43] X. Zhang, Y. Li, S. Zhang, C. M. Schlick, Modelling and simu-
[34] D. Stewart, D. Tate, Integration of Axiomatic design and project lation of the task scheduling behavior in collaborative product
planning, Proc. of first Int Conference on Axiomatic Design, development process, Integrated Computer-Aided Engineering,
Cambridge USA, 2000. vol. 20 (1), pp. 31-44, 2013.
[35] N. Suh, The principles of Design. Oxford series on Advance
Manufacturing, Oxford University Press, New York, 1990.