Using organizational analysis and IDEF0 for
enterprise mode lling in SMEs
MARCO CANTAMESSA and EMILIO PAOLUCCI
Abstract. This paper deals with enterprise mode lling, and paysspecial attention towards the requirements of Small and
Medium Enterprises ( SMEs) . Among existing methodologies,SADT/ IDEF0 is examined and chosen as a starting point. After
having discussed some of the problems connected to the use of
such methodology, a new approach is propose d. Th is methodmay be seen as a bottom-up variant of SADT, which allows us toreduce the mode l development time, decrease the influence
of subjective viewpoints during the mode lling proce ss, andincrease the ease with which information in the model may be
subse quently processe d. Finally, the application of the
proposed methodology in a medium-sized enterprise is brieflyreported.
1. Introduction
The lite rature on enterprise mode lling and en-
gin ee ring is rich in terms of proposed methods and
approache s. This te stifies to the existence of a very live ly
re search activity in th is field. Howe ve r, as conce rns
in dustrial acce ptan ce of th e se m e th ods, th e fe w
em pirical studie s availab le in the lite rature show that
the se proposals are not wide ly applied in practice . For
instance , Deve reux et al. ( 1995) , in a survey cove ring
both `hard ’ and `soft’ innovations to m anufacturing
system s, reported that wh ile 70% of the compan ie s in
the sam ple used som e form of proce ss flowcharting in
support of the de sign proce ss, on ly 20% used a system
design methodology coming from lite rature or consult-
ing firms, and the re st used a m ethod deve loped `in
house ’ . Wh ile the relative youth of enterprise m ode lling
as a discipline partially explain s the lim ited diffusion of
th e se m e th ods, it m ay be valu able to in ve stigate
whethe r they may be improve d in order make them
more acce ssible to end users. Th is is particularly true in
the case of Small and Medium Enterprise s ( SMEs) ,
whose problem s and requirements are obviously some-
what differen t than those of large com pan ies.
Eve n though SMEs are not as com plex as large
com panies, the need for analysin g and redesign ing
the ir operations is no le sser than for the latte r, in orde r
to have them operate in an e fficien t and flexible way,
and to avoid, as often happens, that the ir proce sses
sim ply em erge from the unp lanned evolution of past
custom s and habits. Investing in enterprise mode lling
m ethods for SMEs is worthwh ile because of the we ll-
known economic re levan ce of such com pan ie s. More -
over, with today’ s blurring of the organ izational borders
along m anufacturing value chain s, or so-calle d `ex-
tended enterprises’ ( Porter 1985 , Browne et al. 1995) ,
re lative ly small and strongly focused com panie s tend to
becom e leading actors and building blocks of a new
industrial organ ization ( Howard 1990) . For such
com panies, expe rience with enterprise mode lling and
engin ee ring m ay becom e a strategic asse t enabling
them to in teract closely, e fficien tly and flexibly with
othe r firms. In th is pe rspective , Section 2 of this pape r
deals with a brie f revie w of e nterprise m ode lling
techn iques focussing on SADT/ IDEF0 as a particularly
suitable approach for SMEs. In Section 3, a nove l
approach for IDEF0 m ode l authoring is proposed, to
overcome som e disadvan tages of the traditional SADT
m ethod. Section 4 discusses how this approach may be
used in an ente rprise engin ee ring proje ct and, finally,
an application to a medium -sized company is described
in Section 5.
2. En terp rise m ode lling ap proache s
A num ber of re levant approache s for modelling an
enterprise in te rm s of its proce sses and com ponents
m ay be found in literature ( revie ws may be found in
William s et al. 1994 , CEN TC310 WG1 1994 , Brandi-
m arte and Can tam essa 1994) . The com ple xity of
e n te rprise m ode lling is such that the se d iffe ren t
INT. J. CO MPUTER INTEGRATED MANUFACTURING, 1998, VO L. 11, NO . 5, 416 ± 429
0951-192X/ 98 $12.00 Ó 1998 Taylor & Francis Ltd
Authors: Marco Can tam essa and Emilio Paolucci, Dipartimento di Sistem i di
Produzione ed Economia dell’Azienda, Polite cnico di Torino, Corso Duca degli
Abruzzi 24, 10129 Torino, Italy.
approache s may be assigned to different conceptual
leve ls and classified according to the following schem e
( Doum e ingts et al. 1995 ) .
( 1) Architectures and reference models ( e .g. CIMO SA
and GERAM) provide a standardized and
ge ne ric de scription of the enterprise e lem ents
and their re lationships. They are m ainly used
to clarify the hypothe se s upon what a gene ric
com pany is assum ed to be , so that furthe r
d e ve lopm e n ts m ay be base d upon th e m
( com m ercial system s in the case of software
vendors, and mode lling fram eworks and struc-
tured approache s in the case of re se arch
projects) ;
( 2) Modellin g formalisms are languages wh ich support
the representation, ge nerally in a graphical and
in tuitive manner, of various e lem ents of the
enterprise and under differen t pe rspective s.
Exam ples of such form alism s are IDEF0 ( Marca
and McGowan 1988 ) and Data Flow Diagram s
for repre sen ting functions, En tity ± Re lationship
diagrams and IDEF1X for represen ting in form a-
tion , IDEF3 ( Maye r et al. 1992) and GRAI Nets
( Doum eingts 1989) for repre sen ting behaviour.
More formal language s have also been proposed
for en terprise m ode lling, e .g. the CIMO SA
proce ss mode l ( Ve rnadat 1993) ;
( 3) Modellin g frameworks. An enterprise model ge n-
e rally require s the deve lopm ent of a number of
partial mode ls, deve loped with different formal-
ism s, each dealing with a specific perspective
and in a specific phase of the m ode lling proce ss.
Mode lling frameworks, e .g. CIMO SA ( Jorysz and
Vernadat 1990, AMICE 1993) , are conceptual
structure s used to coordinate the deve lopment
of these partial m odels;
( 4) Structured approaches or methodologies sugge st the
steps to be followe d when using a m ode lling
framework in an enterprise enginee ring proje ct.
Exam ples of such m ethodologie s are SSADM
( Cutts 1991 ) , GIM ( GRAI Integrated Me thodol-
ogy) ( Doum e ingts 1989 ) and PERA ( William s
1992) .
SMEs see king to im prove the ir m anufacturin g
operations m ay obviously find them se lve s a bit lost in
front of the vastne ss and com plexity of such an offer.
Howeve r, the se le ction of an appropriate approach can
be m ade easie r by clarifying the obje ctives of the
enterprise enginee ring project. In particular, it is
importan t to decide whe the r the focus is specifically
upon the deve lopment of a CIM system , or whethe r an
ove rall approach to com pan y re de sign should be
pe rformed, eve ntually having computer-based in tegra-
tion as one e lement.
The strong IT orientation of the form er choice
naturally leads towards the adoption of a fully-fledged
structured approach , e .g. CIMO SA or GRAI-GIM. In
such case s, en te rprise m ode lling require s to conside r
d iffe re n t vie ws upon the com pany ( e .g. function ,
in formation or behaviour vie ws) , and to reach a leve l
of de tail adequate to support the specification and
im ple m e n tation of a software syste m . Th e costs
attached to the enterprise mode lling e ffort may in th is
case becom e sign ificant in absolute te rms, but not when
com pared to the ove rall inve stment in CIM. Additional
e ffort spent upon in itial m ode lling m ay re sult in
conside rable savings of rework avoided in the subse -
quent phases of the project.
In practice , ve ry often , SMEs come to enterprise
m ode lling without having the clear in itial aim of
deve loping a CIM system . The ir e ffort is rathe r m ore
dire cted toward something m ore sim ilar to a Busine ss
Proce ss Re -e ngine e ring in itiative ( Ham m e r 1990,
McSwiney 1995) : the objective in th is case is to
obtain a pre lim inary understanding of `what is going
on ’ in the com pany, and to identify the areas in
which organ izational and proce ss rede sign, toge the r
with IT system s rede sign, m ay contribute to im proving
operational perform ance . Afte r this has been accom -
plished, it may be possible to go some steps forward
by using m ore structure d and de tailed approaches. In
oth e r words, SMEs particu lar ly ne e d e n te rprise
m ode lling approaches for perform ing analysis at a
conceptual de sign leve l. Such pre liminary studie s
m ostly deal with functional aspects, because they are
the m ost apparent and bring busine ss processes in to
focus. They should be pe rformed without too much
e ffort, wh ile at the same tim e providing a sound basis
for further deve lopments ( e .g. adding othe r m odel-
ling vie ws and going deepe r in te rm s of leve l of
de tail) . More specifically, the following requirements
m ay be proposed ( see Colquhoun et al. 1996 for a
sim ilar list) :
( 1) the m ode lling approach must be simple to use
by non-expe rts ( SMEs ge ne rally do not have
enough resource s to dedicate specially trained
staff for enterprise m ode lling tasks) ;
( 2) mode ls must be easily inte rpre ted and discussed
by all the actors involve d in the innovation
process in orde r to increase their comm itm ent;
( 3) mode ls should be used as a basis for furthe r
deve lopments with min im al adaptation;
( 4) mode ls should he lp to identify opportunitie s for
syste m im prove m en t ( Dave nport and Short
1990) ;
Enterprise modellin g in SMEs 417
( 5) the m ode lling process must be as quick and
economical as possible , thus reducing expenses
in exte rnal consultancy or in distracting in te rnal
pe rsonne l from their usual work;
( 6) mode ls must be agile and concise, so that the ir
continuous m ain te nance may be viable and
economic. Th is is e ssen tial for initiating a true
`continuous im provem ent’ proce ss.
Despite some unden iable lim itations ( Billo et al.
1994) , the approach that probably ge ts close r to
fulfilling such requirements, and especially the former
four, is SADT/ IDEF0. Th is structured analysis approach
has become , since its first proposal in the early 1970 s, a
de facto standard for proce ss mode lling in industrial and
busine ss environments ( Colquhoun et al. 1993) . O f
particular in terest is IDEF0’ s un ive rsality, i.e . a language
wh ich has been used in academia and industry alike ,
across a wide range of busine ss sectors and with a
worldwide diffusion . Among the advan tage s of IDEF0,
one can m ention its expre ssive power, the intuitivene ss
of its use and the possibility of using it for functional
mode lling within a num ber of en terprise m ode lling
fram eworks. Am ong these , one can rem ember the IDEF
fam ily of formalisms ( where it is com monly com ple -
men ted with IDEF1x for in form ation m ode ls and
IDEF3 for behavioural m ode ls) and GIM ( in connec-
tion to GRAI grids and nets) . Moreove r, the building
blocks of IDEF0 are close to the enterprise activity
construct used in the CIMOSA function view.
Proper use of IDEF0, howe ve r, is not exem pt from
shortcomings, e specially with re spect to the two latte r
requirem ents. The original proposers of SADT have
always stressed the importance of taking great care of
the m ode lling proce ss and of inve sting suitable effort
into it. The user must be conscious that the final results
will essen tially depend upon the quality of such a
proce ss ( Marca and McGowan 1988) . Unfortunate ly,
th is recommendation often goes unheeded by m any
analysts who, pressed by tigh t time constraints, use
IDEF0 in an approxim ate and in form al way; in th is case ,
the in tuitivene ss and simplicity with wh ich m ode ls are
laid down may become misleading, and induce the
deve lopm ent of incorre ct or poorly focused mode ls.
More specifically, a critical aspect of IDEF0 m ode l
authoring is due to the subje ctive bias that the analyst
naturally tends to cast upon the mode l, based upon
prior experience and to the way with wh ich he or she is
accustom ed to in te rpre ting and looking at re ality
( Busby and William s 1993) . Th is phenom enon will be
enhanced in the case of low involve m ent by the
organ ization ’ s m embers. The risk of subjective bias
has a connection to the fact that IDEF0 m ode ls are
gene rally deve loped using a top-down decomposition.
Working top down require s a sign ifican t conceptual
e ffort in identifying a proper decom position of the
system into a subsystem hie rarchy, and it is ve ry easy for
the analyst to pe rform such a task rathe r roughly, by
drawin g new models from previously deve loped ones
and without payin g enough attention to the ir appro-
priate ne ss to the pre se n t case . Anoth e r fre quen t
problem in IDEF0 mode lling is connected to incon-
sistencie s in in formation flow wh ich are com monly
found in manufacturing system s ( i.e . activities that
re ce ive m ore or le ss inform ation than required, or that
produce in form ation which goes unused) . Actors in the
system , pressed by an analyst attem pting to deve lop an
orde rly and consistent model, often tend to allow him
to `fix’ such problems. Th is leads to a represen tation
which , although desirable , is not a faith ful picture of
the system and will h ide these problem s when the
m ode l will late r be analyse d.
To avoid such risks, the `golden rule ’ of IDEF0
m ode lling prescribes that the authoring proce ss be
m ade up of a lengthy sequence of m ode l writing and
revising phase s, during wh ich the analyst and mem bers
of the organ ization should be ite rative ly involve d and
com m unicate rathe r in tensive ly. As a re sult, deve loping
a good mode l often require s a few m onths time eve n for
a m ode l of limited size ; for a sm all company th is
gene rally becomes a far too demanding task from both
the organ izational and econom ic points of view. More -
over, extending the mode lling proce ss duration has a
number of othe r undesirable consequence s, e .g. a
decrease in comm itm ent on behalf of the organ iza-
tion ’ s m embers, the deve lopment of a m odel wh ich will
be obsole te from the start, and, finally, the inclination
to pe rform an unnece ssarily de tailed analysis, thus
m aking the mode l extrem e ly cumbersome and difficult
to understand.
To face these problems, proposals have been made
to re duce th e e ffort con ne cte d to IDEF0 m ode l
authoring, principally by followin g two differen t paths.
O n one side , som e authors have worked on IDEF0
m etamode ls ( Fung 1993 , Ang et al. 1994 ) : the se are
gene ric m ode ls which should be able to de scribe the
m ain feature s which are peculiar to com pan ie s in a
ce rtain industry, and essen tially propose a standard
h ierarch ical decomposition , toge ther with the principal
in formation and m aterial flows, controls and resource s.
In orde r to deve lop a specific mode l, the analyst should
adapt the m etam ode l’ s structure ( ge ne rally by pruning
or merging diagrams) and then enrich the re sulting
m ode l with company-specific activitie s and entities. The
proce ss followe d, the re fore, seems to be ve ry sim ilar to
CIMO SA’ s conce pt of `stepwise instantiation ’ of a
specific system by starting from a re fe rence architec-
ture . Toge the r with metam ode ls, authors often propose
M. Can tamessa and E. Paolu cci418
more or le ss com plex guide line s for mode l instantiation
and have deve loped software tools to support them,
som e tim es usin g Artificial Inte lligence techn ique s.
Me tamode ls speed up the mode lling proce ss and, in
principle , allow unskilled analysts to carry out com plex
mode lling tasks, because m uch of the e xpe rie nce
re quired is incorporated in the m etam ode l itse lf.
Howeve r, the risks associated with forcing the com plex-
ity of a specific en terprise to fit into a prefabricated
scheme re sulting from past expe rience now becom e
even h ighe r: expe rience , by its own nature , at least
modifie s itse lf ove r tim e , while a rigid scheme such as a
metamode l is more difficult to change . In conclusion,
me tamode ls may be a precious support indeed, but only
if the analyst has sufficient critical ability to find out the
ine vitable inconsistencie s wh ich e xist be tween the
metamode l and the real system, and to change the
former where nece ssary.
O the r authors, in stead, have proposed modifica-
tions to the IDEF0 language in orde r to take out some
of its difficultie s, mostly by sim plifying model structur-
ing. Among these , Billo et al. ( 1994) propose diagram s
based upon the so-calle d `triple diagonal’ , where the se
diagonals, re spective ly, correspond to plann ing, control
and execution activitie s. Triple diagonal m ode ls are not
hie rarch ically structured and the re fore show all re le -
vant data upon a single diagram; furthe r de tails m ay be
mode lled by using anothe r tool, e .g. the `use r concept
diagram ’ ( Hostnik et al. 1985) . Th is is a sort of iconic
enhancem ent of standard proce ss flow-charts, in wh ich
cycle tim es and areas of re sponsibility are m ade explicit.
This and other proposals seem to be prom ising, indeed
under m ultiple poin ts of vie w, e.g. the ease of writing,
validating and also in terpre ting mode ls; howeve r, they
have not yet evolve d in to a wide ly accepted standard,
and the ir industrial applicability may at presen t be
hinde red by this.
3. A bottom -up ap p roach to IDEF0 m od e l au thoring
In order to make the potential of IDEF0 more easily
availab le to SMEs, we have m ove d in to a third dire ction
by revisiting the usual m ode lling approach. Inspiration
for th is approach has com e from organ izational analysis
technique s ( Leife r and Burke 1994 ) , in which in form a-
tion on the organ ization and its busine ss proce sses is
systematically gathe red, loaded into a database and,
finally, analyse d with statistical techniques. The hin t is
to do the sam e for enterprise m ode lling, by colle cting
de tailed in form ation about the activities pe rform ed in
the system through personal inte rviews and/ or ques-
tionnaire s, and then to use such a database as an
in termediate model from wh ich to build other m odels of
the system , IDEF0 in primis. In othe r words, the idea is
the following ( figure 1) : the IDEF0 mode l ( wh ich we
will nam e `X’ ) is a view of reality `R’ , re stricted to
functional aspects and hierarch ically structured. Now, if
one deve lops a wider in te rmediate m ode l of reality, `Y’ ,
then the model X can be derived from Y instead of R
( by `wide r’ we m ean th at Y e ncom passe s all the
in form ation contained in X) . Furthe rm ore , if the
proce sses of obtain ing Y from R, and X from Y are
colle ctive ly le ss burdensome than the proce ss which
dire ctly extracts X from R, then we have found a
conve n ient alternative approach to IDEF0 m odelling.
Working with an in termediate m ode l has othe r
advan tage s. First, when othe r mode ls will have to be
deve loped to cove r different vie ws of the system , the se
m ay be drawn directly from the inte rmediate mode l,
without havin g to start a complete authoring proce ss for
e ach of them ( thus savin g m any tim e -consum ing
sessions for inte rviews and revisions with individuals
and groups) . More ove r, the com mon interm ediate
m ode l m ay he lp to main tain consistency be tween the
various partial mode ls of the sam e system . A som ewhat
sim ilar concept, but re lated to a much wide r scale
framework, m ay also be found in the CIMO SA process-
based mode lling approach ( Vernadat 1993) , as well as
in some othe r re cent proposals for en terprise enginee r-
ing ( Agu iar and Weston 1995 , Koonce et al. 1996) .
Finally, through the inte rmediate mode l, it is re lative ly
easy to obtain mathem atical repre sen tations of the
m anufacturing system ’ s entities and their re lationsh ips,
thus allowing us to analyse the system with quan titative
m ethods. Th is option offers a powe rful tool for high -
ligh ting problem s and solution opportun itie s in the
system , and it makes analysis tasks easy and straightfor-
ward, com pared to going through the large number of
diagram s wh ich m ake up an IDEF0 mode l.
Enterprise modellin g in SMEs 419
Figure 1. Mode l authoring through an intermediate mode l.
Th e in te rm e d iate m ode l has to be de sign e d
according to the m ode ls which will be based upon it,
and the formalisms with wh ich they will be deve loped.
Be cause this pape r focuses upon IDEF0, the in te rm edi-
ate mode l is somehow sim ple and is depicted as an
Entity-Relationsh ip Diagram in figure 2 ( for simplicity,
en tity attributes have not been listed in the diagram) .
Of course , the mode l may have to be augmented in the
case when othe r partial mode ls have to be deve loped in
coordination with the IDEF0 one, e .g. an IDEF1x m ode l
for de scribing inform ation , and/ or an IDEF3 model for
behaviour. Finally, it m ay be noted that the mode l in
figure 2 is sim ilar to the CIMOSA defin ition of function
vie w concepts; its m ain elem ents are the following:
· activ ities ( IDEF0 `boxe s’ ) are the elem entary un its
wh ich m ake up busine ss proce sses. Like in IDEF0,
each activity is characte rized by its inputs and
outputs ( wh ich may be both physical en tities or
in form ation) , as we ll as by the re source s and
controls which, respective ly, support and in flu-
ence its exe cution ;
· entities and in formation ( IDEF0 inputs, outputs and
controls) ; processes are made up of a flow of
physical en tities ( e .g. parts, raw m aterials, fixtures,
e tc.) and information . The in termediate m ode l
describe s each of the se flows and records wh ich
resource Ð if any Ð is in charge of it ( th is is an
additional in form ation re spect to IDEF0) ;
· resources ( IDEF0 m echan isms) ; activitie s are exe -
cuted by the inte rvention of re source s, be they
hum an or m ach ines, wh ich are de scribed in the
in te rm ediate mode l. As mentioned above, each
resource is also directly re lated to entity and
in form ation flow am ong the activitie s they are
involve d in;
· departmen ts; in orde r to make it easier to apply the
mode l to industrial ente rprises, each resource is
associated with a com pany departm ent. Th is
allows us to consider the authority and responsi-
bility flows to wh ich each resource is subordinated.
When such an in termediate m ode l is availab le , the
IDEF0 m ode l m ay be deve loped by extracting the
in formation it contain s. Apparently, the re is a problem
here in that such information is re lated to low-leve l
activitie s, and in IDEF0 m ode lling it is custom ary to
vie w these activitie s as the product of a top-down
decom position process, rathe r than as a starting point.
Now, the proposal wh ich comes from this in te rm ediate
m ode l-based approach is to re th ink th is procedure
and deve lop IDEF0 m odels from the bottom up by
aggre gatin g, rath e r th an de com posin g, ac tivitie s.
There are advan tage s to th is, which allow us to
overcome som e of the previously discussed problem s
connected to IDEF0 mode lling. The main bene fit is
that low-leve l activitie s and the ir in te ractions are
identified directly by the actors involve d in the system
and with m in im al in te rve ntion on behalf of the
analyst. There fore, inconsistencies in the in form ation
flow will appear right away, without tem pting the
analyst in to `fixing’ them to ensure mode l consistency.
Moreove r, the m ode l structure built by bottom -up
aggregation will fit a give n and stable se t of activitie s,
while in the usual top-down approach the analyst will
tend to influence activity identification in orde r to
m ake them fit the mode l structure wh ich he or she
has in mind and has deve loped up to that mom ent.
The concept m ay be clarified, eve n if not form ally
proved, through a toy exam ple , wh ich shows the kind
of problems encountered ( even though the scale of
the se problems is so reduced that they m ay appear
trivial) .
By in tervie wing its CEO , the `top-down analyst’
finds out that com pany XYZ has two departm ents,
M. Can tamessa and E. Paolu cci420
Figure 2. Simplified Entity ± Re lationship representation of the intermediate model.
`product deve lopm ent’ ( PD) and `production’
( P) . The analyst then create s the A0 diagram of
figure 3( a) and in te rviews the m anagers in charge
of the se two departm ents. The PD m anager
identifies h is department’ s activitie s as `product
de sign ’ and `process de sign ’ , wh ile the P m anager
declare s his workforce is engage d in `proce ss
de sign ’ and `m anufacturing’ ( the reason is that
proce ss de sign is pe rformed by product designe rs,
with the aid of shop floor pe rsonnel) . Now, the
m ode l de rivin g from th e A0 decom position
should be revised because it is doubtful that the
two-way partition of operations in PD and P may
still be accepted in the ligh t of the se new findings.
The analyst may anyway be tempted to avoid th is
revision by forcing `proce ss de sign ’ e ither in the
PD activity box, or in the P activity box or, more
creative ly, by placing two `proce ss de sign ’ activitie s
in both PD and P.
A `bottom -up analyst’ would instead go directly
for low-leve l activitie s and understand that there
are three of them , i.e . `product de sign ’ , `proce ss
de sign ’ and `m anufacturing’ , the second of wh ich
be ing join tly pe rform ed by people be longing to
two differen t departments. The proposed decom -
position ( or, more accurate ly, com position) of the
system would immediate ly appear as that of figure
3( b) . The top-down analyst would obtain such a
proposal on ly through a revision of h is in itial
proposal, not accounting for the risk of not doing
so, and the re fore deve loping a distorted m ode l.
The first result of bottom -up aggregation , the refore,
is an IDEF0 m odel that m ore faith fully represen ts the
real system. Moreove r, in the ligh t of experience gain ed
with th is me thod, and as will be de scribed in Section 5,
the mode lling process becom es quicker, because having
de fined low-leve l activitie s since the beginn ing allows us
to reduce the num ber of authoring ite rations with
actors of the system. These ite rations also change in
nature , and gene rally deal with sim ple issues such as
clarifying te rminology, rathe r than with m ore complex
problem s, such as in te rpre tin g syste m beh aviour.
Conce rn ing the h ie rarch ical aggre gation of activitie s,
the approach followed up to now is pure ly m anual and
is straightforward to accom plish . If de sired, it is also
possible to use autom atic approach e s for m ode l
structuring, such as the one proposed by Kusiak et al.
( 1994 ) .
Without any change to the in te rm ediate m odel
shown in figure 2, it is also possible to obtain
another inte resting vie w of the m anufacturing system ,
based upon incidence and adjacency matrice s ( see
figure 4) , wh ich allows us to m athematically repre -
sent inte ractions be tween the four obje ct classe s
conside red:
· the Resource / Activity matrix shows wh ich activ-
itie s are performed by each resource . Analysis of
the matrix m akes it possible to identify exce ssive
or insufficient workloads cast upon people , as we ll
as activitie s in wh ich too few ( or too many)
re source s are involved. Proce ssing th is incidence
matrix also allows us to find clusters of re source s
and activitie s, thus supporting the formation of
workgroups; th is is analogous to the way in which
mach ine s and parts are grouped toge the r in the
Production Flow Analysis approach in Group
Enterprise modellin g in SMEs 421
Figure 3. Top down versus bottom up development of an IDEF0 mode l.
Technology, and m any algorithm s deve loped in
th is area can be readily adapted for this type of
problem ( a com prehensive survey of such meth-
ods m ay be found in Offodile et al. 1994 ) ;
· the Department/ Activity matrix is similar to the
previous one , but aggregate s inform ation at a
department leve l, by showing which departments
are in volve d in each activity and h ow m any
resource s are deployed;
· the Resource / Re source matrix allows to us analyse
in form ation and mate rials exchange d am ong each
couple of re source s in the system . Analysing th is
matrix allows us to be tter understand the system ’ s
structure , and to rede sign both the in form ation
flow and the media with wh ich inform ation is
exchanged;
· the Departm ent/ Departm ent m atrix is sim ilar to
the previous one, but ope rate s on department-
leve l inform ation .
4. Using the prop osed m e thod in en terp rise
engin ee ring
It is now possible to brie fly de scribe how the
proposed approach to IDEF0 m ode lling may be used
with in an enterprise enginee ring proje ct. The metho-
dology used ( figure 5) is a fairly standard one , and is
based upon the m ode lling of the existing ( `As-is’ )
system , the design of a `Should-be ’ targe t and the
de fin ition of a migration plan leading to the desired
goal. There fore , the de scription will be concise and
focus upon the specific characte ristics connected to the
m ode lling approach being proposed.
4.1. Phase 1: bu ilding the `As-is’ model
The main objective of th is phase is to mode l the
curren t situation of the system . The in te rmediate m odel
is to be deve loped at first; afte rwards, two othe r `As-is’
m ode ls will be de rived from the former, re spective ly,
using the IDEF0 language, and the incidence and
adjacency matrice s. The `As-is’ mode l may be built
according to three m ain steps.
· Preliminary analysis. The analyst ach ieve s a pre -
lim inary understanding of the com pany and its
structure ; he or she circum scribes the part of the
com pany upon wh ich to act and, with in th is
portion , identifies the main actors, ope rations
and in teractions with the re st of the com pany.
· Activity analysis. The analyst system atically inte r-
views, e ithe r ve rbally or using questionnaire forms,
M. Can tamessa and E. Paolu cci422
Figure 4. Examples of Activity/ Resource, Activity/ Department, Resource/ Resource and Department/ Department matrices
( clockwise , from uppe r left) .
the actors involve d in the system be ing studied.
The inform ation gathe red is the sam e as indicated
by figure 2, and concerns the activitie s pe rform ed
by actors and the in teractions am ong them.
· `As-is’ model defin ition . The comple ted database
now allows us to produce two differen t vie ws of the
system : the IDEF0 mode l, deve loped by aggregat-
ing activitie s h ierarch ically, and the tabular mode l,
deve loped by extracting the incidence and ad-
jacency m atrice s.
4.2. Phase 2: `As-is’ model analysis an d `Shou ld-be’ model
bu ilding
The aim is to analyse the `As-is’ mode l, so as to
pinpoin t the curren t de ficiencie s and limitations of the
manufacturing system. Th is will lead to a `Should-be ’
mode l, that is a represen tation of the de sired goal towards
wh ich the company should tend. By com paring the `As-is’
and `Should-be ’ m ode ls, a sequence of innovative actions
may then be de signed and implem ented. Th is phase may
be performed through three m ain steps.
( 1) Defin ition of objectives and performan ce measu res.
Th is step is needed so that the curren t and
desired system s may be evaluated with re spect to
a stable fram e of re fe rence . O f course , the
se tting of obje ctives is a strategic matte r which
stands far ahead of the re -enginee ring task
wh ich is being addre ssed in th is pape r, and the
reade r should the re fore refe r to related lite ra-
ture ( Govindarajan and Shank 1992) . The sam e
holds for deciding, give n such objectives, which
pe rformance m easures should be used.
( 2) `As-is’ model analysis. The `As-is’ mode l is to be
analyse d in orde r to find its main drawbacks,
e specially in re lation to the previously e stab-
lished perform ance measure s. Now, such pro-
b le m s m ay h ave d i ffe r e n t o r ig in s a n d
m agnitude s; in orde r to point them out, the
two mode ls are complem entary and have to be
used alte rnative ly. Problem s at a minor scale
m ay be found using low-leve l IDEF0 diagrams or
ve ry specific querie s upon the tabular vie w;
conve rse ly, de ficiencie s connected to system
structure are spotted by working on e ither the
Enterprise modellin g in SMEs 423
Figure 5. Outline of the proposed methodology.
IDEF0 hie rarchy or the ove rall tabular mode l.
The kind of e ffort to be spent in analysing the
m ode ls the refore depends on the desired kind
of innovation , whe ther it is a radical in tervention
( e .g. a comple te busine ss process re -enginee ring
project) , or an incremental innovation ( e .g. an
activity to be perform ed at fixed inte rvals with in
a continuous improve ment program) .
As a m atte r of fact, the re is no fixed rule for
analysin g th e `As-is’ m ode l. Howe ve r, it is
conve n ient to define a se t of comm on criteria
an d d iagn ostic que stion s for spottin g th e
system ’ s ine fficiencie s, which may be gradually
incre m en te d wh ile the an alyst’ s e xpe rience
grows. Such analysis m ay be facilitated if the
database contain ing the in te rm ediate mode l
allows us to form ulate detailed querie s with
e ase , sin ce a com ple te an alysis com m on ly
requires us to try out a long series of such
in te rrogations ( e .g. `extract all com munications
of type chan ge to prev iou s order wh ich are
pe rform ed verbally between departmen t X and
department Y’ ) . As a sam ple of such questions,
four com mon and useful one s may be listed :
· the Activity/ Re source incidence m atrix al-
lows us to see wh ich people have to support
too m any activitie s or, conve rsely, to find out
activitie s wh ich are e ithe r understaffed or
perform ed by too many people at the sam e
time ( probably without any coordination) .
Moreove r, workgroups may be formed by
proce ssing the incidence m atrix, e ithe r by
sigh t, or with clustering algorithm s able to
sort out the `natural’ groups of resource s in
the system;
· re source / Re source and Departm ent/ Depart-
ment adjacency matrices allow us to spot
opportunitie s for e stablishing in terfunctional
teams, whenever it is found that a ce rtain
group of people have in tense and frequent
comm unications among themse lve s;
· adjacency matrice s comm only help in identify-
ing people who spend a sign ifican t part of the ir
time givin g advice and in form ation to collea-
gue s. In th is case , it m ay becom e necessary to
support th is in form ation tran sfer ( e .g. by
inse rting such in formation in a database or
sim ply posting it on a billboard) , and to value
the role of such provide rs of expert knowledge;
· by analysin g the IDEF0 mode l it m ay be
possible to find activitie s wh ich are unnece s-
sary ( often called non value -adding ones) ,
duplicate ( be ing exe cuted separate ly by dif-
feren t people at the sam e time ) , bottlenecks
( when too many activitie s depend upon the ir
com ple tion) , candidate s for be ing put in
paralle l ( it often happens that the sequencing
of activitie s is ju st a h abit, but n ot an
operational nece ssity) .
( 3) `Shou ld-be’ model defin ition ; afte r having analyse d
the `As-is’ m ode ls, and keeping in mind ge ne ral
obje ctive s and constraints, the aim is to de fine a
goal state for the system ( represen ted by the
`Should-be ’ mode l) . Th is may be done by going
ove r the list of deviations found in the previous
phase and, starting from the large r-scale ones,
corre cting them one at a tim e. The re sulting
`Should-be ’ m ode l is, as usual, give n by the
in te rmediate m ode l, and the two m ode ls, IDEF0-
and m atrix-based.
4.3. Phase 3: defin ition and implementation of the `Migration
Plan ’
The objective of this phase is to compare the actual
and the de sired system state ( i.e . the `As-is’ and the
`Should-be ’ m ode ls) . Such comparison allows us to
de fine a sequence of steps that may gradually lead to
the `Shou ld-be ’ state . Proposed m odifications m ay
essen tially be of thre e kinds: organ izational, procedural
and related to in formation system s. In the presen t
phase , it is wise to set priorities for such innovations by
using the previously de fined performance metric, and
conside ring both costs and technological constraints.
Im plementation of the migration plan require s that,
from a simple list of actions, a comple te project be
planned and exe cuted; it is clear that, during th is
phase , Proje ct Man age ment techn ique s m ay be of great
he lp. Moreover, im plem entation of the Migration plan
require s attentive m anage m ent conce rning the alloca-
tion of authority, e specially if the plan regards more
than one of the company’ s departments.
4.4. Phase 4: auditing of the process
The last phase has the obje ctive of ve rifying all the
previous re sults and, specifically, the proposed solutions
and the ir gradual implem entation . Th is activity has the
imm ediate aim of finding corrections wh ich may be
in troduced while the innovation process is still ongoing.
However, the true aim is to initiate a continuous
improvem ent proce ss wh ich may continue long afte r
the present in te rve ntion has been comple ted. In fact, a
com m only raised criticism on enterprise enginee ring
M. Can tamessa and E. Paolu cci424
methods is that they lead to an overwhe lm ing amount
of mode ls and recomm endations, which very soon
becom e forgotten and unheeded. The current proposal
trie s to overcom e this problem , on one side by havin g
facilitated mode l deve lopment and analysis, and on the
othe r side by inse rting this auditing phase .
5. Case stud y
Th is section will brie fly describe the application of
the proposed m ethodology in Com pany E, an Italian
medium-size firm located close to Turin, employin g
about 1100 workers. E produce s e le ctromechanical
components for the automotive and e le ctron ic appli-
ance s industries, and is a European market leader in
th is sector. The me thodology was applied in one
departm ent of the company having a workforce of
about 100 persons, thus justifying the connection with
SMEs.
E is an in n ovative com pan y wh ich draws its
competitivene ss from continuous innovation and qual-
ity of products. In orde r to reinforce th is capability, an
in te rnal program for `continuous im prove m ent’ was
started in late 1992 , and a dedicated staff directly
depending from the CEO was appoin ted. During 1993
and 1994, the staff re -de signed the work flow and
busine ss proce sses of about half of E’ s m anufacturing
departm ents. Howe ve r, the lack of an accurate and
e fficient m e thodology m ade th is proce ss tim e consum -
ing and expensive . The in itial analysis of each depart-
m ent was laborious, and the re was no tool availab le to
re -use part of the expe rience gain ed in previous
analysis.
In orde r to ove rcom e the se problems, E decided to
try out the proposed methodology on the plastic
in je ction moulding departme nt in 1995. Th is depart-
m ent is particularly critical for the com pany, both
because of the technological complexity and varie ty of
its activitie s, and for the impact it has upon down-
stream departments, conce rn ing ove rall lead-time and
quality. Currently, E’ s customers require a few part-
pe r-m illion de fect rate s and a JIT de live ry lead-time of
three days, de spite considerable geograph ical distance s
Enterprise modellin g in SMEs 425
Figure 6. `As-is’ A0 diagram of the case study.
and a rather uneve n dem and. A late 1994 e stimate
reve ale d th at une xpecte d eve nts, and in e fficien t
organ ization and manage ment of the production flow
we re causing an ave rage equipm ent idle time of about
20% , with a per-ann um cost of about 600 000 USD. Due
to increasing dem and, idle tim e reduction was there -
fore place d at the top of the list of obje ctive s to be
reached.
In orde r to ve rify the potential of the m ethodology
for non-expe rt use rs, two postgraduate students with
little practical expe rience in enterprise engin ee ring
we re chosen to apply it. In orde r to facilitate the
application of the m ethodology, a prototype software
tool for managin g the inte rmediate and m atrix-based
mode ls was deve loped by using standard PC-based
environ ments ( Microsoft Visual Basic 3.0 Pro and
Acce ss 2.0) . The IDEF0 mode l has been worked upon
by using a standard editing tool ( Design / IDEF by
Metasoftware Co.) . The `As-is’ m odel ( figure 6 shows
the A0 diagram of the IDEF0 mode l) was com ple ted in
three weeks, wh ich is much shorte r than the tim e
required by the traditional SADT methodology; above
all, shorte r than the standard period usually required by
E’ s innovation staff. Afte r its deve lopment, the `As-is’
m ode l was validated by department m anagers and
workers.
The prototype software facilitated the proce ssing of
the large am ount of in form ation availab le in the
database by allowin g flex ible que rie s and m atrix
clustering ( the we ll-known RO C algorithm was used
for this purpose , King 1980) . Th is analysis high ligh ted
the m ain problems of the departm ent, pointing out
areas of ine fficiency in in form ation managem ent and
proce ssing.
In the following two we eks, and on the basis of the
previous analysis, the `Should-be ’ mode l and m igration
plan we re prepared. Figure 7 shows the A0 diagram of
th is model. It is easy to notice that the most sign ificant
innovation in troduced was to separate line activitie s
( i.e . moulding and control) from staff activitie s ( i.e .
problem-solving and production plann ing activities) .
This separation was obtained by assigning staff activitie s
to a new team made up of expe rienced workers; in the
`As-is’ mode l, such workers we re involve d full tim e in
M. Can tamessa and E. Paolu cci426
Figure 7. `Should-be ’ A0 diagram of the case study.
line activities, so that the ir contribution to department
problem solving was e rratic and ill-structured. Expected
improve ments of the proposed work organ ization were
the following:
( 1) simplification of line activitie s and the acknowl-
edgem ent of problem-solving activitie s’ re le -
van ce ;
( 2) more e ffective problem -solving proce sses;
( 3) reduced equipm ent idle times, as a consequence
of m ore efficient line activitie s and decision
proce sses;
( 4 ) in c re ase d e xch an ge rate of com pe te n cie s
am ong workers with differen t specialtie s.
The Migration Plan was defined by comparing
the `As-is’ and `Should-be ’ m ode ls; the plan ( figure
8) m ain ly contain s organ izational and workflow-
re lated change s, and only one IT-re lated innovation
requiring minor inve stment. E’ s manage m ent has
approved it, and the plan has bee n im plemented. A
first asse ssment of re sults was made four m onths
afte r the Migration Plan im plementation start-up. In
that short period, equipm ent idle tim e has already
d e c re ase d by 30% , with e stim ate d savin gs of
200 000 USD per year.
6. Conclud ing rem arks
Th is paper brie fly discusses some issues in the fie ld
of en terprise mode lling and engin ee ring, outlining
the ir re levan ce within the managem ent of organ iza-
tion al an d tech n ological in novation in SMEs. In
Enterprise modellin g in SMEs 427
Migration Plan
Changes in organizational structure
· Dies are the scarce resource of the department, and planning their utilization requires a large amount of information andknowledge which is difficult to structure, since it changes very quickly and is owned by different persons.
Þ A new `Dies and Problems’ (DP) team is to be introduced, consisting of the responsible for die management, the diewarehouseman, process and tooling engineers. The task of this group is the retrieval, processing, storage and maintenance ofinformation concerning technological and organisational problems connected to dies.
· The responsibility of procedures concerning shift changeover is split between shift foreman and other departmental managers;this authority overlap causes conflicts and problems in product quality.
Þ The role of shift foremen will be removed, and shift changeover will be directly managed by operator team leaders of the endingand beginning shift.
· The enlargement of responsibilities given to operator team leaders induces new potential organizational conflicts.Þ The increased autonomy of operator team leaders will be balanced by increasing the plant manager’s line authority.
Changes in work-flow procedures
· The die warehouseman is involved in too many activities; he is often late in delivering required dies, increasing equipment idletime.
Þ Clerical activities connected to warehouse management will be accomplished directly by operators asking for dies.
· As consequence of the absence of required information, and of fragmentation in authority allocation within the department,production plans are often incompatible with molding technological constraints.
Þ Production schedules will be discussed among the plant foreman and the DP team.
· New-product pilot runs cause many conflicts in production activity control.Þ Production planning will give the required resources to new-product pilot runs and will transmit results to the DP team
· Set-up time must be reduced to improve shop-floor flexibility and reduce lot size.Þ Activities related to die change (including machine cleaning, loading, etc.) will be jointly executed by all workers involved.
Changes in Information System
· Production activity control isn’t currently supported by any software, this is preventing integration with the company’s MRPsystem.
Þ A new software application for production planning in the department is to be developed. The main expected features are:¨ to integrate department production activity control with the company’s MRP system;¨ to support the processing of daily department production plans;¨ to improve the availability of sharp information ion production activity control.
· Production flow must be controlled more closely.Þ A person will be dedicated full time to monitoring production flow, die availability, and coordination with internal and external
customers.
Figure 8. Migration plan proposed for the case study.
particular, special attention is cast upon methodologie s
that may be easily and e ffective ly adopted by SMEs. The
proposal m ade in th is pape r m ay be seen as a revision of
the wide ly-known SADT approach , whose objective is to
dramatically reduce the deve lopment lead-time of the
mode l, to make it exem pt from subjective vie wpoints, as
much as possible , and increase the ease of proce ssing
in formation con tained in it. Lead-time shorten ing
reduces the amount of re sources reque sted by m ode l
deve lopm ent and analysis, and makes its m ain tenance
easie r. The proposed methodology has been succe ss-
fully applied in a medium-sized enterprise , supporting
its application with a software tool designed for the
required data collection and analysis tasks. A furthe r
step will be the integration of these two package s in
order to allow transparent exchange of in form ation
between them .
Acknowled gm ents
The authors wish to thank G. Moretti and S. Ruffini
for the ir precious work in te sting the methodology.
Thanks are also due to Mr P. Aragn o for his support in
carrying out the case study in com pany E.
Re fe rence s
AGUIAR, M. W. C. and WESTO N, R. H., 1995, A mode l-drivenapproach to enterprise integration. International Journal of
Computer-Integrated Manufacturing, 8, 210 ± 224.
AMICE, 1993, CIMOSA: Open System Architecture for CIM
(Be rlin: Springer) .ANG, C. L., LUO , M. and GAY, R. K. L., 1994, Deve lopment of a
knowledge-based manufacturing mode lling system basedon IDEF0 for the metal-cutting industry. International Journal
of Production Economics, 34, 267 ± 281.
BILLO , R. E., RUCKER, R. and PAUL, B. K., 1994, Three rapid and
effective requirements de fin ition mode lling tools: evolvingtechnology for manufacturing system investigations. Inter-
national Journal of Computer Integrated Manufacturing, 7, 186 ±199.
BRANDIMARTE, P. and CANTAMESSA, M., 1994, Methodologie s forde signing CIM systems: a critique. Computers in Industry, 25,
281 ± 293.BROWNE, J., SACKETT , P. J. and WORTMANN, J. C., 1995, Future
manufacturing systemsÐ Towards the extended enterprise.Computers in Industry, 25, 235 ± 254.
BUSBY, J. S. and WILLIAMS, G. M., 1993, The value andlim itations of using process m ode ls to describe the
manufacturing organization. International Journal of Produc-
tion Research, 31, 2179 ± 2194.
CEN TC310 WG1, 1994, An evaluation of CIM mode llingconstructs. Evaluation report of constructs for views accord-
ing to ENV 40 003. Computers in Industry, 24, 159 ± 236.
COLQ UHOUN, G. J., BAINES, R. W. and CROSSELY, R., 1993, A stateof the art review of IDEF0. International Journal of Computer-
Integrated Manufacturing, 6, 252 ± 264.COLQ UHOURN, G. J., BAINES, R. W. and CROSSELY, R., 1996, A
composite behavioral mode lling approach for manufactur-ing enterprises. International Journal of Computer-Integrated
Manufacturing, 9, 463 ± 475.CUTTS, G., 1991, SSADM Structured Systems Analysis and Design
Methodology( Oxford: Blackwell Scientific) .DAVENPO RT, T. H. and SHORT, J. E., 1990, The new industrial
engineering: information technology and business processrede sign. Sloan Management Review, 11, 11 ± 27.
DEVEREUX , S., SMITH, P. and WOO D, D., 1995, A survey of the useof de sign methodologie s for implem enting change in
manufacturing companies in the United Kingdom. Interna-
tional Journal of Manufacturing System Design , 1, 51 ± 58.
DOUMEINGTS, G., 1989, GRAI approach to designing andcon trolling advanced manufacturing system in CIM envir-
onment. In S. Y. Nof and C. L. Moodie ( eds) Advanced
Information Technologies for Industrial Material Flow Systems
(Berlin Heidelberg: Springer) , pp. 196 ± 214.
DOUMEINGTS, G., VALLESPIR, B. and CHEN, D., 1995, Methodolo-gies for de signing CIM systems: a survey. Computers in
Industry, 25, 263 ± 280.
FUNG, R. Y. K., 1993, A structured analysis of the management
of high complexity jobbing production. Proceedings of the IX
CAPE Conference, Nanjing, China, July 1993 (China: AviationIndustry Press) , pp. 23 ± 28.
GOVINDARAJAN, V. and SHANK, J. K., 1992, Strategic Control
Management: tailoring controls to strategies. Journal of Cost
Management, Fall, 14 ± 24.
H AMMER, M., 1990 , Re -e n gin e e ring work: Don ’ t au to-mate , obliterate . Harvard Business Review, July ± August,
104 ± 112.
HOSTNIK, C. J., BILLO , R. E. and RUCKER, R., 1985, Making the
most of structured analysis in manufacturing informationsystem design : application of icons and cycle time . Computers
in Industry, 16, 267 ± 278.HOWARD, R., 1990, Can small business he lp countries com-
pe te?. Harvard Business Review, January ± February.JORYSZ, H. R. and VERNADAT, F. B., 1990, CIM-OSA part 1: total
en terprise mode lling and function view. International Journal
of Computer-Integrated Manufacturing, 3, 144 ± 156.
KING, J. R., 1980, Machin ing-component grouping in produc-tion flow analysis: an approach using a Rank O rder
Clustering algorithm . International Journal of Production
Research, 18, 213 ± 232.
KOONCE, D. A., JUDD, R. P. and PARKS C. M., 1996, Manufactur-ing systems engineering and design: an intelligen t, multi-
mode l, in tegration arch itecture . International Journal of
Computer-Integrated Manufacturing, 9, 443 ± 453.
KUSIAK, A., LARSO N, T. N. and WANG, J. R., 1994, Reengineeringof design and m anufacturing processes. Computers in
Industrial Engineering, 26, 521 ± 536.LEIFER, R. and BURKE, W. J., 1994, Organizational activity
analysis: a me thodology for analyzing and improvingtechnical organization. IEEE Transactions on Engineering
Management, 41, 234 ± 244.MARCA, D. A. and MCGOWAN, C. L., 1988, Structured Analysis and
Design Techn ique ( New York: McGraw Hill) .
M. Can tamessa and E. Paolu cci428
MAYER, R. J., CULLINANE, T. P., DEWITTE, P. S., KNAPPENBERGER , W.B., PERAKATH, B. and WELLS, M. S., 1992, Information
Integration for Concurrent Engineering ( IICE) IDEF3Process Description Compute r Method Report. Armstrong
Laboratory, Wright-Patterson Air Force Base , O hio 45433,AL-TR-1992-0057.
MCSWINEY, J., 1995, BPR for SMEs. In J. Browne , and D.O’ Sullivan ( eds) Re-engineering the Enterprise ( London: Chap-
man and Hall) .OFFO DILE, O. F., MEHEREZ, A. and GRZNAR, J., 1994, Cellular
manufacturing: a Taxonomic Review Framework. Journal of
Manufacturing Systems, 13, 196 ± 220.
PORTER, M. E., 1985, Competitive Advantage (New York: The FreePre ss) .
VERNADAT, F., 1993, CIMOSA: Ente rprise mode lling andenterprise integration using a proce ss-based approach.
Proceedings of the JSPE-IFIP WG 5.3 Workshop on The Design of
In formation In fra stru ctu re Systems for Man u factu rin g
(DIISM’93), Tokyo, Japan , November 1993, pp. 65 ± 79.WILLIAMS, T. J., 1992, The Purdue Enterprise Reference Architecture
( Re search Triangle Park, NC: Instrume nt Socie ty ofAmerica) .
WILLIAMS, T. J., BERNUS, P., BROSVIC, J., CHEN, D., DOUMENINGTS, G.,NEMES, L., NEVINS, J. L., VALLESPIR, B., VLIETSTRA, J. and
ZOETEKO UW, D., 1994, Arch itectures for integrating manu-facturing activities and enterprises. Computers in Industry, 24,
111 ± 139.
Enterprise modellin g in SMEs 429