Open Journal of Safety Science and Technology, 2011, 1, 10 8-114
doi:10.4236/ojsst.2011.13012 Published Online December 2011 (http://www.SciRP.org/journal/ojsst)
Risk Analysis Model Using UML and MADS Model
Hafida Bouloiz1, Emmanuel Garbolino2, Mohamed Tkiouat3
1Industrial Engineering Department, National School of Applied Sciences, E.N.S.A., Agadir, Morocco
2Crisis and Risk Research Centre, Mines ParisTech,
Sophia Antipolis, France
3Industrial Engineering Department, En gi nee r s’ Mohammadia School, Rabat, Morocco
Received October 24, 2011; revised November 17, 2011; accepted November 25, 2011
Abstract
The purpose of this paper is to propose a model of risk analysis which combines two tools belonging to a
different context. These both tools are MADS (Model of Analysis of Dysfunctional Systems) and UML
(Unified Model Language). The proposed method aims to integrate UML language, especially the collabora-
tion diagram, in the MADS model. We represent the danger source system of MADS model with the colla-
boration diagram in order to define and model the scenarios of risk. The application of this method is illus-
trated with an example of a storage unit of chemicals. On the one hand, the proposed model provides a com-
prehensive view that facilitates the understanding of the organization of an industrial system, and on the an-
other hand, it leads to more effective analysis of risks taking into account the interactions between the system
components.
Keywords: Risk Analysis, Safety, MADS Model, UML, Interactions, Collaboration Diagram
1. Introduction
Industrial processes, particularly chemical industries, are
almost daily news headlines with the existence of poten-
tial risks that could cause accidents, considering the ha-
zardous nature of the chemicals. Indeed, zero risk does
unfortunately not exist in these industrial activities. This
explains the large number of methods of risk analysis
that have been developed in order to control the risks.
Many studies describe the content of these methods of
risk analysis such as [1-9]. Most of these methods are
from field of dependability. Among of these methods:
Hazard and Operability Study-HAZOP [6], Failure Mode
Effect analysis-FMEA (24), Fault Tree Analysis-FTA
[10], Preliminary Risks Analysis-PRA [6,11].
The occurrence of industrial accidents such as Bhopal
(1984), Piper Alpha (1988), Chernobyl (1986), Seveso
(1976), Mexico gas explosion (1984), Three Mile Island
(1979), has shown the limits of these methods which
describe an accident by the a series of events linked by
cause and effect and does not take into account the in-
teractions within the system so the complexity of a sys-
tem. They do not give a good understanding of risks in
the complex systems. According to [12], system is “a set
of interacted elements”. A similar view comes from or-
ganizational theory, where an industrial system is re-
garded as complex as its parts are in interactions [13].
Complex systems are characterized by emergent beha-
vior due to interactions between the various components
of the system seen at different levels of organization
[14].
The goal of risk analysis is to define and identify the
measures of risk control. It is the important step in the
process of risk control and industrial safety. Therefore,
risk analysis must take into account the complexity of
system; especially that complex system is subject of ac-
cidents caused by dysfunctional interactions between sy-
stem components [15]. In this context, the systemic ap-
proach is the demarche which allows analyzing the sys-
tem and to formalize the interactions between its com-
ponents. This approach, appeared at the end of 1960 [16],
is an interdisciplinary joint makes it possible to under-
stand and describe the complexity and it became an ap-
proach, a language or technical ensuring the modeling of
complex systems.
This paper proposes a new model of risk analysis
which contains two systemic tools which are UML mo-
del (Unified Modeling Language) and MADS model

H. BOULOIZ ET AL.109
(Analysis Method of Dysfunctional Systems). The pur-
pose of this approach is to integrate UML formalism,
especially collaboration diagram in the MADS model.
We represent the danger source system of MADS model
with collaboration diagram in order to define risk sce-
narios. A comparison between UML and MADS model
in the context of industrial risk analysis has been de-
scribed in [17].
This paper is organized as follow: Section 2 presents
and defines the two tools corresponding to MADS model
and UML language. Section 3 present case study which
is a storage unit of chemicals. In Section 4, we present
proposed approach and it illustration through the case
study, and Section 5 concludes the paper.
2. Proposed Method
The proposed method is founded on the UML and
MADS model. In the next section, we present these both
models.
2.1. MADS Model
MADS (Analysis Method of Dysfunctional Systems)
model is a model of industrial risk analysis founded on
the systemic approach [18]. This model is built on the
basis of principles of the systemic modeling developed
by [12]. It is composed of two systems called danger
source system and target system. MADS model shows
that the occurrence of any undesirable event passes by an
events process. As shown in Figure 1, this process starts
from a source of danger in the form of hazard flow and
reaches a hazard target (target system).
MADS model presents a vocabulary which highlights
a sequence of events: initiating event, initial event, flow
of danger and final event. Therefore, with MADS model,
the scenarios of risk are represented as a process of events,
beginning with an initiating event in the system.
2.2. UML Model
UML model is a graphic modeling language in the field
of software engineering, standardized by the OMG (Ob-
ject Management Group). It became a standard of object
modeling [19], which aims to build, to visualize, and to
specify the information systems [20]. UML model in-
cludes a set of graphical notation techniques in order to
create multiple views allowing expressing static, dyna-
mic and functional aspect of the system (the different
modeling diagrams are explained in detail in UML Nota-
ntion guide) [21].
UML model has been used in a wide variety of appli-
cations. In [18], the author uses UML language in order
Figure 1. MADS model (Périlhon, 1999, adapted).
to model an information system of natural hazards. The
authors in [22] propose a design for the plant safety
model that is fully integrated within the plant lifecycle
model using UML language. Reference [23] presents a
model of railway system using UML in order to study its
reliability. In [24], the authors use UML model in order
to show the link between risk analysis and maintenance.
In [25] a sequence diagram of UML is used to model the
behavior of actors in a situation of decision-making. In
[19], UML is used as an operational tool which formal-
izes the interactions within an ind ustrial system and con-
tributes to analyze its risks. In this paper, UML model
comprises collaboration diagram is used to model sce-
narios of risk. The objective of collaboration diagram is
to define the interactions with a dynamic point of view
between the system objects. It represents these interac-
tions through a chronological representation by sending
messages between the objects in order to realize a func-
tion also called use case.
3. Case Study
The case study in this paper correspond to a storage unit
of chemicals which belongs to an industry specialized in
the manufacture of chemical substances for industrial use
located in the industrial area in Casablanca (Morocco).
This industry is part of an industrial group which is a
global leader in the field of chemical specialty. For rea-
sons of confidentiality, we do not quote its name. The
storage unit studied in this paper corresponds to the
warehouse. This unit contains three storage depots, cor-
responding to the three types of the stored materials,
which are: monomers in liquid state, peroxides in solid
state and flammable products in liquid state. The mono-
mers are products presen ting the risk of a polymerization
which is strongly exothermic and can cause an explosion
or a fire. The peroxides are characterized by their oxi-
dizing and combustive properties. They can activate the
combustion of a combustible substance. The flammable
products have the characteristic to ignite in air and con-

H. BOULOIZ ET AL. 110
 to store the products; tinue to burn. Drums and containers are stored at tem-
peratures specific to type of chemical product. Mono-
mers must be stored in a temperature range between 16˚
and 25˚ to avoid polymerization or solidification of these
materials. The temperature of peroxides should not ex-
ceed 30˚, and the temperature of storage of the flamma-
ble products should not exceed 35˚. Therefore, tempera-
ture in the warehouse is the important parameter to con-
trol. We consider the storage unit as a complex system
composed of a set of components in interaction. Figure 2
presents a configuration of this system. Human factor has
a level of responsibility on an adequate achievement of
the storage. Procedures constitu te an information support
corresponding to th e maintenance, th e instructions of the
products storage, the safety data sheets, the protection
measures in case of an accidental spill of products, the
safety check list, the location sheets, etc. Safety devices
correspond to the prevention equipments (alarm, smoke
detector, detector of temperature) and to the protection
equipments (sprinkler, individual protection equipments).
Regarding stored products, three types of chemical pro-
ducts are stored: monomers, peroxides, and flammable
produ cts and each typ e of chemical is stored in a spec ific
temperature.
 to maintain the safety devices;
 to control the stored products;
 to control the temperature of storage.
These functions must be performed and all system
components must be organized according to a goal which
is to ensure an adequate storage of the products, by
avoiding any situation being able to present a risk.
4. Proposed Method and Application
As previously mentioned the proposed model combines
UML a We represent the danger source system of MADS
model with the collaboration diagram (Figure 3) in order
to define and model the scenarios of risk. We present
these scenarios in the form of events process as defined
in the MADS model. A scenario starts from an initiating
event in the collabor ation diagram which leads to an ini-
tial event.
The danger which flows fro m this initial event reach es
target system causing a final event corresponding to da-
mage and consequences.
In this paper, we present three examples of risk scena-
rios. Two scenarios in the function “to maintain the sa-
fety device” and a scenario in the function “to control the
In this storage unit, four functionalities are defined:
Figure 2. Schema showing a set of interactions in the system corresponding to storage unit of chemicals.

H. BOULOIZ ET AL. 111
Figure 3. Propose d model founded on the MADS and UML
model.
stored pro duct s”.
4.1. Risk Scenario in the Function “to Control”
the Stored Product
Figure 4 shows this scenario which presents the case
where operator of storage control did not pay attention to
default corresponding to a degradation of storage shelves.
The flow of danger associated to this no conforming ac-
tion of operator is fall of storage containers which may
lead to physical wounds or also an explosive atmosphere
when it about a flammable product.
4.2. Risk Scenarios in the Function “to Maintain
Safety Device”
This scenario (Figure 5) presents the case where the pro-
cedure of maintenance is not adequate. This irrelevance
of the maintenan ce procedure leads to dysfunction of the
safety equipments (i.e. temperature detector of storage).
When the temperature of storage (i.e. monomers) ex-
ceeds the prescribed temperature, whereas it is not de-
tected by the temperature detector, there is a risk of an
exothermic reaction of the stored monomers, which can
lead to an explosion.
Another scenario may be identified (Figure 6). For
example the operator did not respect the procedure of
maintenance, which is used to describe the instructions
of reliable functioning of safety equipments (i.e. tempe-
rature detector of storage). This not conforming action of
the operator can generate an inadequate functioning of
the safety devices.
The risk scenarios represented with the method which
combines MADS model and collaboration diagram allow
defining the possible scenarios which generate risk, by
Figure 4. A scenario of risk in a function of storage control due to human error.

H. BOULOIZ ET AL. 112
Figure 5. A scenario of risk in a function of maintenance due to use an inadequate procedures of maintenance .
Figure 6. A scenario of risk in a function of maintenance due to inadequate behavior of operator.

H. BOULOIZ ET AL. 113
taking account of initiating events in the interacted com-
ponents and in the each function of system.
5. Conclusions
The proposed method aims to integrate the UML model
in the MADS model. Danger source system of MADS
model is represented with collaboration diagram which
make it possible to identify all possible scenarios at each
function or use case of the system. It does specify the
function of the system on which this scenario is identi-
fied, allowing a comprehensive identification of risk sce-
narios. This method presents several interests. It repre-
sents a mean to support the risk analysis with a systemic
method taking into account the interactions between sy-
tem components. In addition to risk analysis, this model
is a particularly powerful tool that facilitates the under-
standing of the organizatio n of an industrial system. This
understanding is due to use of collaboration diagram
which define the interactions between system compo-
nents and these interactions are represented through a
chronological representation by sending messages be-
tween the components in order to realize a function of
the system.
The future work is to develop a computing platform
which allows implementing this model.
6. References
[1] C. Cesics, “Introduction to Risk Analysis: Systematic
Methods,” Series of Publications of the Committee of
Experts for Safety in the Chemical Industry, Notebook
No. 4, 1981, pp. 1-8.
[2] C. Ccps, “Guidelines for Hazard Evaluation Procedure
with Worked Examples,” 2nd Edition, CCPS AIChE,
New York, 1992.
[3] A. Villemeur, “Reliability, Availability, Maintainability
and Safety Assessment”, Vol. 1: Methods and Techniques,
Vol. 2: Assessment, Hardware, Software and Human
Factors, John Wiley, Hoboken, 1992.
[4] M. Monteau and M. Favaro, “Bilan des Méthodes
D’Analyse a Priori Des Risques, Partie 2: Principales
Méthodes de la Sécurité des Systèmes,” Cahiers des
Notes Documentaires INRS, Vol. 139, 1990, pp. 363-389.
[5] F. Craweley, M. Preston and B. Tyler, “Hazop: Guide to
Best Practice,” Institution of Chemical Engineers and
European Process Safety Centre, Rugby, 1999.
[6] R. L. Rogers, “Methodology for the Risk Assessment of
Unit Operations and Equipment for Use in Potentially
Explosive Atmospheres,” EU RASE Project n SMT4-
CT97, Library Area SAFETYNET, Inburex GmbH Hamm,
2000.
[7] Y. Mortureux, “Preliminary Risk Analysis,” Technical
Engineering, Industrial Enterprise, Safety and Risk Man-
agement (CD Rom), 2002.
[8] A. Desroches, A. Leroy and F. Vallé, “Risk Manage-
ment,” Hermès and Lavoisier, Paris, 2003.
[9] J. Tixier, G. Dusserre, O. Salvi and D. Gaston, “Review
of 62 Risk Analysis Methodologies of Industrial Plants,”
Journal of Loss Prevention in the Process Industries, Vol.
15, No. 4, 2002, pp. 2291-303.
[10] F. I. Khan and S. A. Abbasi, “Techniques and Method-
ologies for Risk Analysis in Chemical Process Indus-
tries,” Journal of Loss Prevention in the Process Indus-
tries, Vol. 11, No. 4, 1998, pp. 261-277.
doi:10.1016/S0950-4230(97)00051-X
[11] M. Nicolet-Monnier, “Integrated Regional Risk Assess-
ment: The Situation in Switzerland,” International Jour-
nal of Environment and Pollution, Vol. 6, No. 4-5, 1996,
pp. 441-461.
[12] V. L. Bertalanffy, “General System Theory,” Dunod,
Paris, 1993, p. 53.
[13] J. G. March and A. Herbert, “Organizations,” 2nd Edition,
Blackwell Business, Cambridge, 1993.
[14] S. Sheard, “Definition of the Sciences of Complex Sys-
tems,” Insight INCOSE, Vol. 9, 2006, pp. 25-26.
[15] N. A. Leveson, “New Accident Model for Engineering
Safer Systems,” Safety Science, Vol. 42, No. 4, 2004, pp.
237-270. doi:10.1016/S0925-7535(03)00047-X
[16] V. L. Bertalanffy, “General Systems Theory, Foundation,
Development, Applications,” George Braziller, Inc., New
York, USA, 1969.
[17] H. Bouloiz, E. Garbolino and M. Tkiouat, “Contribution
of a Systemic Modeling Approach Applied to Support
Risk Analysis of a Storage Unit of Chemical Products in
Morocco,” Journal of Loss Prevention in the Process In-
dustries, Vol. 23, No. 2, 2009, pp. 312-322.
doi:10.1016/j.jlp.2009.12.001
[18] A. Napoli, “Forest Fire Metrology: Methodological and
Technological Approach to Support the Experimental Pro-
cess in Forest Fire Behaviour Modelling,” In Actes du
Symposium International DELFI Feux de Forêts: Besoins
et Innovations, Athènes, 1999, pp. 216-223.
[19] G. Booch, J. Rumbaugh and I. Jacobson, “The Unified
Software Development Process,” Adisson-Wesley, New
York, 2000.
[20] Object Management Group, UML 2.1 Superstructure Spe-
cification, Document: ptc/06-04-02, 2006.
[21] UML Notation guide, OMG, 2006.
http://cgi.omg.org/uml/
[22] H. A. Gabbar, K. Suzuki and Y. Shimada, “Design of
plant Safety Model in Plant Enterprise Engineering En-
vironment,” Journal of Reliability Engineering and Sys-
tem Safety, Vol. 73, No. 1, 2001, pp. 35-47.
doi:10.1016/S0951-8320(01)00029-1
[23] J. L. Boulanger, P. Bon and G. Mariano, “Semi Formal
Modeling and Formal Specification: UML & B in Simple
Railway Application,” International Conference on Soft-
ware and Systems Engineering and their Applications,
Toulouse, 12-14 October 2004.
[24] A. Mili, S. Basseto, A. Siadat and M. Tollenaere, “Dy-

H. BOULOIZ ET AL. 114
namic Risk Management Unveil Productivity Improve-
ments,” Journal of Loss Prevention in the Process Indus-
tries, Vol. 22, No. 1, 2008, pp. 25-34.
doi:10.1016/j.jlp.2008.07.011
[25] A. Refsdal and K. Stolen, “Extending UML Sequence
Diagrams to Model Trust Dependent Behavior with the
Aim to Support Risk Analysis,” Journal of Electronic
Notes in Theoretical Computer Science, Vol. 197, No. 2,
2008, pp. 15-29. doi:10.1016/j.entcs.2007.12.014