13TH INTERNATIONAL DEPENDENCY AND STRUCTURE MODELLING CONFERENCE, DSM’11
CAMBRIDGE, MASSACHUSETTS, USA, SEPTEMBER 14 – 15, 2011
USING THE FF-DMM MATRIX TO REPRESENT
FUNCTIONAL FLOW IN PRODUCT ARCHITECTURE
Vincent Holley1,2, Bernard Yannou1 and Marija Jankovic1
1
Ecole Centrale Paris, Laboratoire de Génie Industriel, Paris, France
Schlumberger Riboud Product Centre – SRPC, Clamart, France
2
ABSTRACT
In the context of new product development, highly constraint multi-disciplinary systems are difficult
to design and generally lead to a non optimal but acceptable solution (Seepersad et al., 2008). The
design of such product implies to collaborate soon in the choice of concepts. Our industrial analysis
shows that concept choice is leading to collaborative problems when a design department implies a
stronger influence than others. This attitude to favor one design department decreases product
performance interest. As concept evaluation is a key point in product designs, this design stage must
take into account design department’s point-of-views. This article describe our approach, based on
enrich representation of functional flow in DMM, for the mapping of collaboration contribution to
functions.
Keywords: Multi-physics, collaborative design, DMM
1
INTRODUCTION
In this article, we propose to semantically enrich conventional representation model of product
complexity. We use a Domain Mapping Matrix (DMM) to link functions and product architecture. A
first contribution is the enrichment of this representation. We enrich DMM representations with
functional flow sequencing along the architectural modules; this ontological enrichment of design data
makes it easier to envision and manage design challenges for multi-physics systems. This article goes
further into Holley et al.’s (2010) publication.
2
PROBLEM STATEMENT
This research study is conducted in collaboration with Schlumberger, worldwide leader in petroleum
services. The recent developments of onboard electronic cards are an example of this multi-physics
problem. Electronic card has to be integrated in a box attached to a main mechanical component. The
whole assembly is going in a tube (with a diameter limited by the drill). In order to develop this
product the expertise of three design departments is needed (mechanical, electrical and packaging).
Every department is optimizing their design to maximize functionalities; for example the compactness
evaluated through the number of electronic cards by product foot length. Eighteen months after the
concept choice, the project failed due to incapability to manage the impacts of the capacitor size.
Capacitor size is a key design parameter influencing the number of electronic cards by product foot
length and so the function compactness.
Current approach doesn’t include the identification of collaboration contribution to function
achievement. The actual proposition does not permit to explore the limits of the complex problem. As
key design parameters between design departments influence product functions, as it happens in our
industrial example, the collaboration on capacitor size was over constraint and no dimensioning set
can be found to meet expected performances.
3
LITERATURE REVIEW
The research literature is mainly addressing previous issues with the usage of Design Mapping Matrix
(DMM). The aim of crossing functions and product architecture is to capture design rules. For instance
Gorbea et al. (2008) propose a quite interesting method using MDM (mix of DSM and DMM) to map
115
dependencies in architectures. Rule extraction and generation in this approach is based upon
components and functions analysis. The proposed MDM is composed of three matrices: a functionsfunctions DSM, a components DSM and a Components-functions DMM. A set of these three matrices
is generated for each architecture. Basic matrix operation, as sum and subtraction, are used to compare
several matrices and therefore extract rules for each of them. Sum of MDMs enables the determination
of which component is compatible with all architectures. Subtraction MDMs called “delta MDMs” is
useful in comparing differences amongst two architectures. Danilovic et al. (2007) propose another
approach called “Periodic Table”, mixing Design Structure Matrix and Domain Mapping Matrix, in
which a DMM is used to map correlation between components and functions. This information
available to characterize these data describes only the existence or not.
The main lack of the literature is that concept analysis is not related to functional analysis. Gorbea et
al. (2008) propose a simple analysis of functional path in which we cannot see/identify the degree of
participation of a component to the overall product performances. Moreover, the evaluation of the
design concepts is proposed embracing only one global performance. Therefore this does not permit
the integration of different advantages or disadvantages in possible design solutions. Moreover, this
literature review is not managing that architecture can have different structure, or that functional path
can be achieved in different manners.
4
OUR GLOBAL APPROACH: THE MPDS METHOD
The goal of our global approach is to map design department point-of-views, architecture alternatives,
functional needs and expected performances. With this process, our approach aims at helping
designers to model their collaborations with other design departments and to assess their impacts on
the final product. The proposed MPDS method (Multi-Physics Design Scorecards) matrix based
method that is organized in the three steps describes in Figure 1.
Figure 1. A1 SADT of the MPDS method
The “fill matrices” step objective is to gather project data based on “functional analysis” and
“concepts brainstorming” into three matrices: Functional Flow – Domain Mapping Matrix (FF-DMM),
Physical Connection – Design Structure Matrix (PC-DSM), and Voice of Design Department (VoDD),
which will be used to generate six design assessment cards based on connectivity maps. The
capitalization of MPDS results in the Collaborative-FMEA has for objective to quickly highly
collaborative design risk about the project. Therefore, the six design assessment cards extracted from
connectivity maps are used as an input.
This article will focus on the mapping the connections between components and functions; and the
identification of collaboration contribution to functions through the use of DMM. The proposed
116
functionnal flow aimss at identifyy collaboratioons between
n design departments to functions, th
he obtain
matrix is
i called FF-DMM (Funnctional Flow – DMM)). Thus, ourr research w
wants to imp
prove the
definitioon of the collaboration obbjectives throough the mod
del of functioonal path in cconcepts.
5
FF
F-DMM’S ONTOLOG
GY
FF-DMM
M is a crosss functional flow and arrchitecture mapping
m
mattrix using a DMM form
mat that is
enrichedd by the integgration of a functional
f
floow. The dataa employed in
i the use off the FF-DMM
M matrix
is definee as the follow
wing:
Design Depaartment reprresents the department
D
d
in
i charge off the designn of a modu
ule of the
s
system.
The design deparrtment is ideentified by itss name and that
t of its enggineers.
F
Function
deffines both what
w
the prodduct must do to meet clieent requiremeents (main functions)
fu
a
and
what itt must do to stay in working co
ondition (seervice functiions). Functtions are
c
characterized
d by a name and a utillity. The utility correspoonds to the goal of the function
(
(whatever
thhe client needds or the funcction needs to
t keep the syystem in worrking conditiion).
F
Functional
F
Flow
represeents functional flows thro
ough the archhitecture of tthe product. Function
c
chains
are sensitive
s
to the order off deploymen
nt of functioons from one module to
o another
m
module.
M
Module
desiignates a parrt of the sysstem that mu
ust exist in order
o
to perfform a functtion. The
M
Module
has a name.
T
Technical
Soolution repreesents a poteential solutio
on to the dessign of a moodule. Each technical
s
solution
is asssigned a nam
me.
6
FF
F-DMM: AN
A EXPER
RIMENTAL
L EXAMPL
LE
This appproach has been
b
experim
mented on an
a industrial applicationn who aims to develop onboard
electroniic cards undeer the scope of a project. The previou
us card devellopments were not able to
o achieve
environm
mental constrains: high pressure,
p
higgh temperaturre under shoock and vibraations as req
quired for
our sub-pproject.
Any teaam member can construuct the FF-D
DMM matrix
x after the functional
f
annalysis and concepts
brainstorrming. The main results of the funnctional anallysis are funnctions definned with naames and
functionnal flows. Thhe principall results of the conceptts brainstorm
ming are cooncepts defin
ned with
moduless and techniccal solutions.. The data coollection pro
ocess for the FF-DMM m
matrix, repreesented in
Figure 2, is as follow
ws (an examplle, extracted from
fr
Holley ett al., 2010, is given in Figu
ure 3):
1. A
Add functionn names to thhe matrix (“11” in Figure 2).
2
2. List
L technicaal solutions (e.g., “I”, “Dellta”, “Pivot”) below their modules
m
(e.gg., “chassis”) (“2”).
(
3. Assign
A
a color to designn departmennts, and use it to shade in the namee of the mod
dule they
d
design
(in Fiigure 3, eachh design depaartment is asssigned a coloor).
4. Fill
F in the boody of the matrix
m
(“3”) with
w the resu
ults of the funnctional anallysis (rules for
f filling
i the body of
in
o the matrixx are explaineed after Figu
ure 3).
Figure 2.
2 FF-DMM formalism
fo
Functionns are expresssed in rows (“1” in Figuure 2); modulles and their technical soolutions are expressed
e
in colum
mns (“2”). Thhe data contaained in the matrix
m
(“3”) presents the correlation bbetween the potential
117
contribuution of the technical
t
sollution, which representss the architeccture, to thee effectiveneess of the
functionn. Functional flows are described
d
by horizontally
y filling in booxes with a number. Thee number
designattes the order of deploym
ment of the fuunction throu
ugh each of the technicaal solutions. The
T rulebased foormalism used to describee function floow is as follo
ows:
Each row loccated in “1” (see Figure 2)
E
2 representss a function,
P
Parallel
rowss in “3” (see Figure 2) beehind a functtion correspoonds to function flow alteernatives.
F
Function
proopagation thhrough the prroduct archittecture, so called
c
functioon flow, is described
d
u
using
numbeer from 1 to n (an examplle is given affter Figure 3)).
Figure 3.
3 On board electronic
e
card
d case study FF-DMM
F
As an exxample for Power
P
Electrronic Controlller Case Stu
udy, we pressent an exam
mple of four functions
(Figure 3):
3
“Generaate power” iss a function going from electronics (indicted
(
by a numeral “1” in the first row of
Figure 3),
3 where pow
wer is regullated, to subsstrate (“2”), then to connnectors (“3””), and then to
t wiring
(“4”), where
w
motor control
c
is coonnected; theere are no oth
her possibilities to achieeve this functtion even
in the caase of the othher concepts.
The consstraint functiion “withstannd pressure” can have tw
wo alternativee paths depennding on the concepts
selected:: pressure caan be appliedd on the boxx and on the connectors or on the coollar. Functio
onal flow
alternativves are reprresented in parallel
p
row
ws: pressure applied to the
t collar is noted with a single
numeral “1” in the correspondingg module, annd pressure applied
a
to thee box and coonnectors is indicated,
i
in anothher row, withh the numeraal “1” in botth the box and
a connectoors modules (the “2 non--hermetic
integrateed” connectoors field is unnmarked becaause they can
nnot withstannd/endure prressure).
Experim
ments carried out by engineers showed that shoccks have twoo different ppropagation pathways
p
within thhe PEC case study. Thereefore, the funnction “resistt shock” has two differennt functional flows. In
both cases, shocks prropagate from
m the collar (indicated in
n a merged field
f
by a num
meral “1” in
n the final
two row
ws of Figure 3) to electrronics (“5”).. Based on internal funcction analysis, shocks can
c either
propagatte through thhe chassis (““2”), box (“33”), and subsstrate (“4”) or,
o as indicatted in another matrix
row, throough wiring (“2”) and coonnectors (“33”).
“Dissipaate heat” is the
t most com
mplex functiion to repressent with a FF-DMM m
matrix, becau
use it has
aspects of
o both a parrallel and ann alternative function, dep
pending on the
t concept sselected. In any case,
this funcction starts frrom electronnics (indicated in a mergeed field by a numeral “1”” in Figure 3)). Then it
can be propagated
p
i parallel thhrough the box
in
b (“2”) an
nd the chasssis (“3”), whhere it can either be
evacuateed or it can go
g into the collar
c
(“4”) before
b
it is dissipated.
d
O the functioon can be prropagated
Or
through the substratee (“2”), box (“3”), and chhassis (“4”), where it cann either be eevacuated or it can go
into collar (“5”) befoore it is dissippated.
m experts reggarding the different
d
conncepts, the F
FF-DMM maatrix also
Aside frrom capturinng data from
aims to validate the capacity of different braainstormed concepts
c
to satisfy
s
the fuunctions requ
uested by
the cliennt. For exam
mple, a conceept composedd of a “Pivo
ot” type box and “2 non-hermetic in
ntegrated”
118
connectoors cannot achieve
a
the function
f
“enndure/withstaand pressuree” when muud enters thrrough the
collar (thhe second row
w of the funcction “endurre/withstand pressure” in Figure 4).
Figure 4. Functional ability
a
of initiaal set of concep
pts presented in the FF-DM
MM matrix
Rows annd columns can
c easily bee added to the FF-DMM in order to trrack project progress and
d change.
In order for the FF-D
DMM to rem
main clear andd as simple as
a possible too understandd, columns arre hidden
accordinng to the connvergence off the initial seet of concepts. A part off the matrix ccan be extracted by a
given deesign departm
ment, for more precise annalysis of its own objectiives, and thenn brought baack to the
original FF-DMM with
w more dettails.
6
C
CONCLUSI
IONS
This papper presents our
o defined Functional
F
F
Flow
– Domaain Mappingg Matrix (FF--DMM) mattrix based
on ontollogy, processs for their filling
f
by deesign team member
m
andd taxonomy for their com
mpletion.
Ontologyy and taxonnomy represeent our princcipal contrib
bution of thee literature reeview aboutt Domain
Mappingg Matrix. Ouur aim by intrroducing Funnctional Flow
w into DMM
M so called FF
F-DMM con
ncerns the
ability too represent fuunction path through product architeccture.
REFER
RENCES
Daniloviic, M. and Browning,
B
T
T.R.
(2007). Managing complex prooduct develoopment projeects with
deesign structuure matrices and domaiin mapping matrices, International
I
l Journal off Project
Ma
Management.
Eigner, M.
M and Maleetz, M. (20008). Potentialls of DSM, DMM
D
and MDM
M
for reqquirements modeling,
m
DS
SMC.
Gorbea, C., Spielmaannleitner, T.,
T Lindemannn, U., and Fricke
F
E. (2008). Analyysis of hybrid
d vehicle
u
multiple domain maatrices, DSM
MC.
arcchitectures using
Holley, V.,
V Yannou, B., and Jankkovic, M. (20008). Robusttesse d’un QF
FD en phasee de choix dee concept,
CO
ONFERE.
Holley, V., Yannou,, B., and Jannkovic, M. (2010).
(
Towaards an enricched represeentation of functional
fu
M
(DMM
M), DSM Co
onference.
paath in Domainn Mapping Matrix
Lindemaann, U., Mauurer, M., and Braun, T. (22009). Structtural Compleexity Manageement, Springer.
Wyatt, D.,
D Wynn, D., and Clarksson, J. (2008). Synthesis of product architectures
a
M/DMMusing a DSM
baased approachh, DSMC.
V
Holleey
Contact: Vincent
Ecole Cenntrale Paris
Laboratoiire de Génie Industriel
I
Grandes Voies
V
des Viggnes
92295 Chhatenay-Malabbry
France
+33 1 455 37 72 34, [email protected]
119
INVEST ON VISUALIZATION
Using the FF-DMM Matrix to
Represent Functional Flow in
Product Architecture
Vincent Holley1,2, Bernard Yannou1 and Marija Jankovic1
1Ecole
Centrale Paris, Laboratoire de Génie Industriel, Paris, France
Riboud Product Centre – SRPC, Clamart, France
2Schlumberger
INVEST ON VISUALIZATION
Oil Market
6
•
Extreme conditions
– High pressures
– High temperatures
– High shock and vibration
6 5 designers
years p
projects
j
6 7 to 15 y
6 5 to 10 million $/year
6
•
Naturally constrained
– Mud, oil, gas and acid
– Within
Withi a smallll diameter
di
t
(typically 5 to 15 cm)
Schlumberger projects
Issues
6 Duration lengthened
about 40% to 150%
6 Cost may be x2
6 Reliability need
2 to 3 years of
i
i
re-engineering
13th International DSM Conference 2011- 2
120
INVEST ON VISUALIZATION
Audit and Diagnosis of Design Project Management
Action Research approach
•
Audit concerns about 14 projects and 25 jobs
– Model design tasks including job interactions
•
Our diagnosis
– Design process very loosely: Extreme variability
– No prescribed design tools: No FMECA
– No collaborative platform: No multi-disciplinary management
– Engineers are experts in their area of expertise
•
This article takes part of a Ph.D. work look for the improvement
of design process by a simple user-friendly method to manage
design collaboration: highlight highly constrained architectural
zones
13th International DSM Conference 2011- 3
INVEST ON VISUALIZATION
Our Global Approach: Multi
Multi-Physics
Physics Design Scorecards Method
121
INVEST ON VISUALIZATION
Literature Review: Mapping Functions-Components
Gorbea, Spielmannleitner, Lindemann, Fricke,
DSM, 2008
13th International DSM Conference 2011- 5
INVEST ON VISUALIZATION
Literature Review: Mapping Functions-Components
Gorbea, Spielmannleitner, Lindemann, Fricke,
DSM, 2008
13th International DSM Conference 2011- 6
122
INVEST ON VISUALIZATION
Analysis
of Multi-Physics
Concepts:
Data Gathering
y
y
p
g
INVEST ON VISUALIZATION
Data Gathering Ontology
13th International DSM Conference 2011- 8
123
INVEST ON VISUALIZATION
FF-DMM Protocol
• Data collection process for the FF-DMM:
1 Add ffunction
1.
i names,
2. List technical solutions below their
modules and assign a color to design
departments,
3. Fill in the body of the matrix with the
Functional Analysis.
13th International DSM Conference 2011- 9
INVEST ON VISUALIZATION
FF-DMM Illustration
Mechanics
Scope of
study
Packaging
Scope of
study
Electronics
Scope of
study
Modules
Technical
Solutions
Functions
13th International DSM Conference 2011- 10
124
INVEST ON VISUALIZATION
FF-DMM Illustration
INVEST ON VISUALIZATION
FF-DMM Illustration
125
INVEST ON VISUALIZATION
Conclusion
•
Defined Functional Flow – Domain Mapping Matrix (FF-DMM)
– Ontology,
– Filling process,
– Taxonomy.
•
Contribution of the literature review
– Ontology,
– Taxonomy.
•
FF-DMM ability
– To represent function path through product architecture.
13th International DSM Conference 2011- 13
126