=Paper=
{{Paper
|id=Vol-479/paper-12
|storemode=property
|title=Integrating System Dynamics with Conceptual and Process Modeling
|pdfUrl=https://ceur-ws.org/Vol-479/paper12.pdf
|volume=Vol-479
}}
==Integrating System Dynamics with Conceptual and Process Modeling==
https://ceur-ws.org/Vol-479/paper12.pdf
Integrating System Dynamics with Conceptual
and Process Modeling
P. Fiona Tulinayo
Institute of Computing and Information Sciences, Radboud University Nijmegen
Heyendaalseweg 135, 6525 AJ Nijmegen, The Netherlands, EU.
F.Tulinayo@science.ru.nl
Abstract. The purpose of this research is to provide a basis on which
stakeholders make their decisions, to improve System Dynamics (SD)
modeling by deploying methods and techniques from system develop-
ment, and to support in the creation of SD models where, static and
dynamic methods are applied together to achieve a common goal. To
achieve this we will take a stepwise approach: identify the key concepts
as used in different methods, map their constructs, derive transforma-
tions, create their syntax and semantics, and develop requirements spec-
ifications on which a tool can be based. For Conceptual modeling we
use Object-role modeling (ORM), a fact-oriented approach for model-
ing information at a conceptual level. For process modeling we use a
technique (workflow language) called Yet Another Workflow Language
(YAWL). YAWL works as an extension of Petri Nets with constructs
to address the multiple instances, advanced synchronization and cancel-
lation patterns. This research is under the supervision of Dr. S.J.B.A
(Stijn) Hoppenbrouwers, Dr.(Patrick) van Bommel and Prof. Dr. H.A.
(Erik) Proper.
Keywords: System Dynamics, Conceptual Modeling, Process Modeling
1 Introduction
Integration of methods involves mapping and defining different concepts with an
aim of using them under one umbrella. Paige [17] defines method integration as
an involvement in defining relationships between different methods so that they
may productively be used together to solve problems. He further gives more def-
initions inline with method integration in [18]. This study proposes to integrate
System Dynamics (SD) with conceptual and process modeling. For Conceptual
modeling we intend to use Object-Role Modeling (ORM) [10], a fact-oriented ap-
proach for modeling information at a conceptual level [11]. This is because of its
strong verbalization, conceptualization and a fully formal link to predicate logic.
ORM has graphical constraint notations that are far more expressive than, for
example, Unified Modeling Language (UML) class diagrams or industrial Entity
Relationships (ER) [11]. Halpin and Wagner further note that ‘‘although ORM
supports modeling of business terms facts, and many static integrity constraints
and derivation rules, it cannot model the reactive behavior of systems which can
be described using dynamic integrity constraints”. This explains our use of YAWL
2 Proceedings of CAISE-DC 2009
and SD to capture the dynamic part of the business process.
For process modeling we will use a technique (workflow language) called Yet
Another Workflow language (YAWL) [26] because its one of the general rep-
resentatives of process modeling. YAWL works as an extension of Petri Nets
with constructs to address the multiple instances, advanced synchronization,
and cancelation patterns. We will use Conceptual Modeling (ORM) and Pro-
cess Modeling (YAWL plus Petri Nets) in the service of creating an integrated
model.
We carry out this integration because it is hard to define complex dynamic
models in complex organizational settings therefore, we need support based on
ontology (conceptual structure). Secondly, for transferability purposes that’s in-
cases where information from one organization need to be reused by another.
lastly to be able to have a basis for the development of a tool that will aid in
understanding model behavior.
Figure.1 illustrates how we integrate the methods (SD, Conceptual (ORM)and
Fig. 1. Abstract View of the Integration
Process (YAWL plus Petri Nets) modeling). Two types of mappings are shown:
mappings between viewpoints are what we refer to as inter-viewpoint mappings,
and the Mappings between specific viewpoints and the integrated meta model
are refereed to as viewpoint meta model mappings. We use ORM, which is a
state and event reporting, fact oriented modeling technique, as a graphical rep-
resentation for the integrated meta-model. Petri nets Plus YAWL are used to
model a discrete flow, and SD to model a quantitative flow.
All in all, ORM will add high quality formal conceptualization to SD mod-
eling; YAWL and Petri Nets will serve to bridge the gap between static ORM
and Dynamic, flow-like aspects of SD.
2 Problem Definition and Research Motivation
Integrating system dynamics with conceptual and process modeling is the key
issue this research will focus on. In order to solve this problem a number of factors
need to be studied and analyzed to get a deeper understanding of the problem
Proceedings of CAISE-DC 2009 3
at hand. This issue has been identified in [21] where it is stated that; “....there
is a strong case for starting to apply systems dynamics methods more openly in
the BPM and MIS research fields, as I feel the tools and techniques available are
vastly under-rated in terms of their applicability and capability to provide novel
representations of real-world situations.....”. This statement is the main idea
behind this research. The proposed integration should enable the stakeholders
to attain a high level of understanding of the dynamics and statics within the
systems studied, enabling decision-makers to make more dependable decisions
at different levels.
With ORM focusing on conceptual modeling, YAWL on process modeling
and, SD on the dynamics within Business Process Modeling (BPM), this research
expects to achieve a well grounded method on which a tool can be based. This will
introduce a new breed within enterprize modeling where the static models are
merged with dynamic models to give a clear guide to process model development.
2.1 Research Questions
To address this issue, the following questions have been derived:
1. How can we integrate system dynamics with conceptual and process model-
ing methods?
2. How can the different interactions (concepts) used in these methods be
mapped and formally integrated to give a common foundation on which
a tool is to be developed?
3. How can the developed integration be used, conceptualized and validated ?
4. How can we give a complete description of the model to fit a tool to be
developed?
2.2 Research Objectives
To enable us answer the research questions, we have formulated Objectives to
guide us achieve the intended goal. These include;
1. To Integrate System Dynamics with Conceptual and process Modeling.
2. To develop requirements (functional and non functional) on the basis of
which a tool can be developed
3. To evaluate the use, conceptualization, and validation of the integrated mod-
els.
4. To operationalize the method by making the language usable and provide
procedures and guidelines where need be.
3 Literature Review
3.1 System Dynamics Modeling
SD as a method has been in existence since 1961, developed by Jay Forrester to
handle socio-economic problems with a focus on the structure and behavior of
4 Proceedings of CAISE-DC 2009
systems composed of interacting feedback loops. A review and history is given
in [8]. SD provides a high level view of the system emphasizing the interactions
between its constituent parts, as well as the impact of time on its dynamic
behavior [14]. As a method, it has its focus on the structure and behavior of
systems composed of interacting feedback loops. The art of SD modeling lies
in discovering and representing the feedback processes and other elements of
complexities that determine the dynamics of a system [22]. The dynamics arise
from the interaction of two types of feedback loops, positive and negative loops.
Positive loops tend to reinforce or amplify whatever is happening in the system.
Negative loops counteract and oppose change. These loops all describe processes
that tend to be self limiting, processes that create balance and equilibrium [19].
Simulation with SD models is used for learning about the dynamic complexity
of systems, identification of optimal policies in existing systems, improvement
of system behavior through parameter or structural changes. The method has
been applied to a wide range of domains, from the management of production-
distribution systems to the management of ecosystems. Comparisons between
SD and Discrete-event system [2], and between SD and Petri nets [6] have been
done. In these comparisons the main differences between SD and these methods
are highlighted. [20] further identifies issues for the future of system dynamics.
3.2 Conceptual modeling
The term conceptual modeling [27] is derived from a conceptual model which
is an invention to provide an appropriate representation of the target system,
appropriate in the sense of being accurate, consistent, and complete [16]. Con-
ceptual models are similar to intermediate causal models proposed by [28] which
were developed to capture the meaning of an application domain as perceived
by an individual; a precise study can be found in [12]. The method provides
artificial models which enable modelers bridge the gap between the experiential
world and the abstract mathematical world. Conceptual modeling involves the
representation of the entire information system content of the database being
designed in somewhat abstract terms relative to the way data is physically stored
[5]. A classic view of the conceptual modeling process is presented in [12] where
they give a clear description on the fundamental view on the process of con-
ceptual modeling. We use conceptual modeling because the methodologies are
well developed and have proven to be quite successful for building information
systems in a graphical way at the conceptual level [15].
3.3 Process modeling
Process modeling [4] becomes more and more an important task not only for the
purpose of software engineering, but also for many other purposes [1]. Before
defining what process modeling is, we start by defining what a process is; [13]
describes a process as “a set of partially ordered steps intended to reach a goal ”.
[4] notices that any component of a process is a process element and a process
step is “an atomic action of a process that has no externally visible substruc-
ture”. With that note a number of scholars have defined process modeling in
Proceedings of CAISE-DC 2009 5
different ways. [29] defines it as logically capturing and abstracting the systems
components, relationships and behavior, with respect to modeling objectives.
[4] defines a process model as an abstract description of an actual or proposed
process that represents selected process elements that are considered important
to the purpose of the model and can be enacted by human or machine. These
definitions all state that process models are abstract representations meaning
that the system depicts the behavior of an actual system in place. Abstraction
can help the modeler to study the behavior of any system with out tampering
with the operational system, hence enabling exploration of various options be-
fore decisions are made by Stakeholders. Under process modeling we opt to use
a Workflow language called YAWL as a workflow method [26]. Workflows are
used to define, validate, and automatically manage and monitor the execution of
operations (business processes) in organizations. They aim at formalizing activi-
ties involving the coordinated execution of multiple tasks performed by different
processing actors [3]. History and the various articles on YAWL can be found on
the YAWL website 1
4 Design Approach
4.1 Conceptual Linking and Transformations
We start by identifying the methods to be used in both conceptual (ORM) and
process modeling (Petri nets and YAWL). By so doing, we are able to have a clear
scope on which methods to use in this study and why. After that, we identify
the key concepts as used in these methods, then come up with conceptual links
between them. First we consider ORM and SD; then ORM, SD and Petri Nets.
We use Petri nets because it is the basis on which YAWL was developed. By
starting with Petri nets (foundations of YAWL) we give a strong foundation to
the integration. Having mapped the concepts and identified their transformation
statements, we then use the model elements and concepts developed to come up
with a generic meta-model plus semantics/syntax of the model. After that we
will develop the requirements specification on which a tool can be based.
4.2 Case Study and Experimental Modeling Sessions
Case study is an exploratory (single in-depth study) or explanatory (cross-case
analysis) research strategy, that involves an empirical investigation of a particu-
lar contemporary phenomenon within its real life context using multiple sources
of evidence [30]. We will use case study methodology to focus on understanding
the dynamics present within a single setting [7], and to understand them within
a particular context [31]. We chose to use this method because of its use of many
techniques when collecting empirical data.
We will also carry out experimental sessions in different settings to enable
us gauge the applicability of the model developed. In these sessions; SD, ORM,
Petri nets and YAWL will be used as they all have different but important roles
to play as explained earlier. They will also help in better understanding of the
integrated model.
1
http://www.yawl-system.com/
6 Proceedings of CAISE-DC 2009
4.3 Tool support
In the final development of the integrated model, we will consider the following:
Conceptualization links, Mapping, Transformation, Syntax/ Semantics and De-
scription of model integration to fit tool making. Modern SD packages will
be used to model the SD model because of their graphical interface making the
modeling of a complex system much easier. The SD model(s) will be built based
on the Case study results which provide a descriptive model on which the SD
conceptual feedback structure will be developed. The feedback structure model
will be developed with the help of a Causal Loop Diagram (CLD). CLDs will
be converted into Stock and Flow Diagram(SFD) which is a formal quantita-
tive model. Mathematical relationships between or among variables that enable
the simulation of the model will be defined thereafter, simulations of the key
variables will be run.
5 Preliminary Results
We have so far identified the conceptual links between SD and ORM [23]. To
achieve that we used a working example the procedures a paper might go through
en route from writing to publication. By using an example we came up with
different illustrations to clearly show the link and conceptualize the different
concepts as used in both methods.
In Table.1. we show a summary of the mapping, transition statements of
the key variables plus their elements. A detailed explanation is formulated and
is under review. Having achieved that, we are currently working on a meta-
model where we use the mappings plus model elements identified to derive the
syntax/semmantics of the methods.
Table 1. SD with ORM and Petri nets
System ORM Petri nets Transitional State- Elements
Dynamics ment
Stock Unary fact types Places They all contain “things” Containers
Quantity Objects Tokens These can be looked at as the Contents
things that flow with in the sys-
tem
Flows (Inflow Object types Transitions They all connect items: Stocks Homogeneous
and Outflow) (SD), Unary fact types, (ORM) connectors
and Places (Petri Nets)
Connectors Fact types Arcs They are all active and involve Heterogeneous
activities that cause a change to connectors
the recipient or destination
6 Conclusion
This study aims at integrating dynamic and static methods. The methods ap-
plied have been in existence for sometime, are well founded and widely applied
Proceedings of CAISE-DC 2009 7
to different settings in the modeling world. We chose to use these methods be-
cause of their complementary dynamic and static aspects in modeling systems.
During this study we will come up with a generic meta-model as illustrated in
figure.1, and requirements on which a tool can be based. We will apply the ap-
proach presented in context of various case domains. We will further develop and
refine the method (its models as well as the stepwise process): By devoting more
attention to integrate existing formalizations (syntax and semantics), but also
to operationalization of the modeling procedures. Finally, we intend to explore
further links between SD and process modeling (already initiated by the Petri
Net involvement), in particular with the YAWL method [25].
References
1. Becker, J., M., Rosemann, C. von Uthmann.: Guidelines of Business Process Mod-
elling, in: W.M.P van der Aalst et.al.,(Eds.), Business Process Management: LNCS
1806: 30–49, Springer-Verlag, Berlin (2000)
2. Brailsford, S.C. and Hilton, N.A.: ”A Comparison of Discrete Event Simulation
and System Dynamics for Modelling Healthcare Systems.” Proceedings of ORAHS,
Glasgow Caledonian University, pp. 18-39 (2000)
3. Chen, P. P., Thalheim, B. and Wong, L. Y.: Future Directions of Conceptual Model-
ing, in Conceptual Modeling, P. P. Chen, J. Akoka, H. Kangassalo, and B. Thalheim,
Eds. Springer LNCS 1565, Berlin, Germany (1999)
4. Curtis, B., Kellner, M. I., and Over, J.: ”Process Modeling,” Communications of
the ACM (35:9), 75–90 September (1992)
5. Date, C. J.: An Introduction to Database Systems, Volume 1, Fifth Ed, Reading:
Addison-Wesley (1990)
6. Duggan, J.: ”A Comparison of Petri Net and System Dynamics Approaches for Mod-
elling Dynamic Feedback Systems.” 24th International Conference of the Systems
Dynamics Society, Nijmegen, The Netherlands, July (2006)
7. Eisenhardt, K.M.: Building theories from case study research”, Academy of Man-
agement Journal, 14 (4): 532–50 (1989)
8. Forrester J.W.: Industrial Dynamics. The MIT Press, Cambridge (1961)
9. Guizzardi, G., Pires, L. F., Sinderen, M. J. v. (2002): On the role of Domain On-
tologies in the design of Domain-Specific Visual Modeling Languages. In: Proc. 17th
ACM Conference on Object-Oriented Programming, Systems, Languages and Ap-
plications (OOPSLA 2002)
10. Halpin, T.: Object-Role Modeling (ORM/NIAM), Handbook on Architectures of
Information Systems, Springer, Heidelberg, Ch. 4 (1998)
11. Halpin. T and wagner, G.: Modeling reactive Behavior in ORM. LNCS 2813, :567–
569. Springer-Verlag Berlin Heidelberg (2003)
12. Hoppenbrouwers, S.J.B.A, Proper, H.E., van der Weide, T.: A Fundamental View
on the Process of Conceptual Modeling. In: Conceptual Modeling - ER 2005 – 24
International Conference on Conceptual Modeling. Volume 3716 of Lecture Notes
in Computer Science. 128 - 143 (2005)
13. Humphrey, W.S. and Feiler, P.H: Software process development and enactment:
Concepts and definitions. Tech. Rep. SEI-92-TR-4. Pittsburgh: Software Engineer-
ing Institute, Carnegie Mellon University (1992)
8 Proceedings of CAISE-DC 2009
14. Hustache, J.–C., Gibellini, M., Matos, P.L.: A System Dynamics Tool for Economic
Performance Assessment in Air Traffic Management 4th USA/Europe Air Traffic
Management R and D Seminar Santa, December 3–7 (2001)
15. Jarrar, M., Demey, J., Meersman, R.: On Using Conceptual Data Modeling for
Ontology Engineering. In: Aberer K., March S., and Spaccapietra S., (eds.): Journal
on Data Semantics, Special issue on ”Best papers from the ER/ODBASE/COOPIS
2002 Conferences”, LNCS Vol. 2800, Springer. ISBN: 3-540-20407-5. October 185–
207 (2003)
16. Norman, D. A.: Some observations on mental models. In Dedre Gentner and A.
L. Stevens (Eds) Mental Models, Lawrence Erlbaum Associates, Hillsdale, NJ. 7-14
(1983)
17. Paige, R.: A meta-method for formal method integration. Proc. Formal Methods
Europe, LNCS 1313, Springer-Verlag.(1997)
18. Paige, R.: When are methods complementary? Information and Software Technol-
ogy 41(3), February, (1999)
19. Pfahl, D. and Lebsanft, K.: Integration of system dynamics modeling with de-
scriptive process modeling and goal-oriented measurement. Journal of Systems and
Software, 46: 135-150 (1999)
20. Richardson G.: Problems for the future of system dynamics. System Dynamics
Review 12: 141–157 (1996)
21. Sharif, A. M.: Industrial Viewpoint can systems dynamics be effective in modeling
dynamic business systems? Business Process Management Journal 11(5): 612–615
q Emerald Group Publishing Limited (2005)
22. Sharma, D., Sahay, B.S., Sachan, A.: Modeling Distributor Performance Index
Using System Dynamics Approach., 16(3) (2004)
23. Tulinayo, P.F, Hoppenbrouwers, S.J.B.A.(S.), and Proper, H.A.(E.).: Integrating
System Dynamics with Object-Role Modeling. IFIP International Federation for
Information Processing. J. Stirna and A. Persson (Eds.): PoEM 2008, LNBIP 15:
77-85 (2008)
24. van der Aalst. W.M.P, ter Hofstede, A.H.M., Kiepuszewski,B. and Barros A.P.:
Workflow Patterns. Distributed and Parallel Databases, 14(1):5-51 (2003)
25. van der Aalst, W.M.P, Hofstede, A.H.M.t.: YAWL: Yet Another Workflow Lan-
guage. Information Systems, 30(4), 245–275 (2005)
26. van der Aalst. W.M.P. and ter Hofstede, A.H.M.: YAWL: Yet Another Workflow
Language. QUT Technical report, FIT-TR-2002-06, Queensland University of Tech-
nology, Brisbane (2002)
27. Wand, Y., Storey, V., Weber, R.: An Ontological Analysis of the relationship Con-
struct in Conceptual Modeling. In: ACM Transactions on Database Systems, 24(4):
494–528 (1999)
28. White, B.: Intermediate causal models: A missing link for successful science educa-
tion? In R. Glaser (Ed), Advances in Instructional Psychology, volume 4, Lawrence
Erlbaum Associates, Hillsdale, NJ. 177-252 (1993)
29. Williams, D.: Challenges of System Dynamics to Deliver Requirements Engineering
Projects: faster, better and cheaper (2003)
30. Williams, D.: Dynamic Synthesis: A Theoretical Framework for Research in Re-
quirements Engineering Process Management. Operational Research Society, ISBN:
0 903440202 (2000)
31. Yin, R. K.: Case Study Research: Design and Methods Applied Social Research
Method, Sage, Beverly Hills, California (1988)