Smooth Migration From The Virtual Design
Smooth Migration From The Virtual Design
Abstract- The Virtual manufacturing concept is based on the use agent-based applications, taking advantage of its powerful
of computational resources during the design of products and mathematical foundation. In fact, the use of High-level Petri
manufacturing systems, reducing the ramp-up time to market and the
costs for the system’s design, operation and reconfiguration. nets allows achieving a formal approach that covers the
Combined with emergent paradigms for manufacturing control, such complete life-cycle of the development process, from the
as multi-agent systems and service-orientation principles, and using a requirements-analysis through the design-validation and
formal language to describe the system behavior, e.g. Petri nets, a implementation of collaborative automation systems and their
truly and complete virtual production environment is achieved. This control systems in an integrated manner [8].
paper discusses this powerful approach to the virtual design and
production environment driven by formal models, allowing users to During the development life-cycle, the simulation of the
design and deploy collaborative and re-configurable automation control applications assumes a crucial role, allowing the
applications in a faster and cost-effective fashion. A special attention verification of the correctness of the achieved solution and
is devoted to the process of migration from the virtual environment to even testing various “what-if” scenarios to continuously
the real world, aiming to reduce the transition time and increasing the improve/optimize it during the design phase. High-level Petri
efficiency of the process.
nets formalism is suitable to execute the simulation, taking into
I. INTRODUCTION account the system specifications and considering the time
parameter associated to transitions. It supports the debug of the
Today’s marketplace is characterized by an increasing global system’s behavior by easily reproducing abnormal conditions
competition, demanding highly customized products, shrinking (or at least, conditions that cannot be easily created in real
product lifetimes and increasing the products complexity. In world). The information extracted from the temporal evolution
this scenario, industries need to address quickly the customers’ of the High-level Petri nets control model reflects the temporal
needs by manufacturing the desired products at the right time sequence of the system operation, being easy to check the
and at competitive costs, under the penalty of no longer being system compliance with the specified performance indexes,
competitive. A new class of reconfigurable systems is required, discover cyclic evolution and existence of bottlenecks.
being the traditional centralized, top-down, rigid and static Integrating this formal approach to describe and validate the
control replaced by the decentralized control over distributed system’s behavior with 2D/3D software tools, a truly and
autonomous, cooperative and intelligent entities. Multi-agent interactive virtual production environment is achieved. This
systems [1], Holonic manufacturing systems [2-3] and more virtual engineering environment uses intensively
recently Service-oriented Architectures [4] are emergent computational resources to perform realistic design analysis
paradigms that address the referred requirements. and verification under different scenarios and conditions, and
Several implementations of applying the referred emergent generates feedback in real time.
paradigms were reported in the literature (see [5-6] and the Using virtual environments to emulate the real world and to
references therein) but in spite of their promising potential, support the simulation, it is important to guarantee an easy and
their industrial adoption is being very slow. Among other smooth migration of the control applications from the virtual
reasons, an issue that contributes for this weak adoption is the environment to the real world. Among others, it is important to
missing of formal, mature, structured and integrated guarantee that the control application models behave in a
engineering frameworks to support the specification, design, similar manner when deployed into the industrial controllers to
verification and implementation of collaborative (agent-based run in the real world. If, during the migration process, the
or service-oriented) production control applications, in an easy, designed behavioral models need to be significantly adjusted,
modular and rapid manner. Aiming to address this weakness, the results achieved during the simulation process do not
High-level Petri nets formalism [7] is used in this work as the reflect the real system behavior. Additionally, a smooth
formal language to specify and validate service-oriented and migration can also imply benefits in terms of re-
978-1-4244-3760-3/09/$25.00
c 2009 IEEE 539
Authorized licensed use limited to: Instituto Politecnico de Braganca. Downloaded on January 15, 2010 at 18:01 from IEEE Xplore. Restrictions apply.
configurability, since a higher performance in the migration 373 airplanes to improve the skin quality after the final
process may decrease the costs and time associated to perform installation, allowed to minimize the waviness and to increase
the re-configuration/adaptation process. the fuel economy of the plane, leading to saves of more than 2
This paper discusses the advantages of using virtual million dollars per year. Another well-known example is the
production environments to support the simulation and use of robotic simulation and off-line programming
verification during the design phase. The behaviors are capabilities. However, in the context of this work, the focus is
described by the introduction of a formal and structured the production-centered and especially the control-centered
approach to support the specification and validation of Virtual Manufacturing, which uses additional simulations to
automation systems using the High-level Petri nets as formal control models and actual processes allowing seamless
language. Integrating the powerful capabilities of High-level simulation for optimization during the actual production cycle
Petri nets to support formal verification and simulation with [11].
2D/3D software tools, a computational framework is achieved As stated by Cachapa et al. [13], a computational simulation
supporting users to create a virtual design and production is an attempt to model a real-life situation on a computer so
environment for the convenient development of automation that the behavior of the system can be verified. Simulation has
systems and their supervisory control. This virtual and formal been commonly used to study the behavior of real world
environment permits to achieve an easy and smooth migration manufacturing systems to gain better understanding of
from the virtual to the real production environments. underlying problems and to provide recommendations to
The rest of the paper is organized as follows. Section II improve those systems [14]. Being the observation of real
overviews the virtual manufacturing concept and analyses the systems very expensive and sometimes cumbersome, a
requirements for a smooth migration. Section III discusses a simulation model is an easier way to identify bottlenecks and
formal and integrated approach, using High-level Petri nets, to to enhance system performance in terms of productivity,
support the virtual design of automation systems and the queues, resources utilization, cycle times and lead times [14].
smooth migration to the real world. Section IV presents an The advantages of using the simulation of the system
experimental case study scenario to illustrate the proposed through the use of virtual production environments (i.e.
concepts. At last, Section V rounds up the paper with the connecting virtual equipment to the control application) are
conclusions. mainly the following [15]:
− A complete virtual factory automation system model can
II. VIRTUAL DESIGN OF PRODUCTION SYSTEMS be built, including the logic control that can be simulated
before to be implemented in the physical factory plant.
The virtual design of production systems aims to provide a − The complete system or partial pieces of the system (e.g.
computational environment allowing users to develop lines, cells or equipments) can be debugged and verified
applications in an emulated real world. As the term “virtual” without the need to use the (real) physical devices.
suggests, these environments can be created, simulated or − It is easier to reproduce abnormal conditions and to debug
carried on by means of a computer. In fact, informatics and with conditions that cannot be easily created in the real
computational resources play crucial role to provide world (for example introducing “what-if” scenarios).
mechanisms for the visualization and simulation of models Especially, dangerous tests in the real world can be done
describing the system behavior. safely in this virtual world.
A. Virtual Design Environments − The data can be reused for operator training and
Virtual Manufacturing is a concept that has been used by maintenance, and the simulations can be repeated as
some companies to improve their processes, introducing more many times as necessary to the correct understanding and
quickly new products in the market. In spite of the several tuning of the system control or as a feasible study to reuse
definitions founded in the literature, see [9-12], it can be logic control to other systems.
briefly defined as the use of computer models and simulations − The control strategies can be improved/optimized using
to model and design manufacturing processes for the purpose iterative simulation scenarios and before its practical
of analyzing and understanding them. The strategy is to create realization.
an integrated environment, composed by several software tools Fig.1 illustrates the virtual and real manufacturing
and systems, aiming to generate a new method to develop environments, with special focus to the iterative process till
products and systems. reaching the correctness and maturity of the solution, using
Three paradigms of Virtual Manufacturing can be identified information technologies as the basis platform. After the
[11]: design-centered, production-centered and control- complete verification of the correctness of the system behavior
centered. Probably, the most well-known application area of in the virtual environment, the designed system is migrated to
Virtual Manufacturing is the design-centered, where Computer run in the real world, e.g. connecting to the real physical
Aided Design (CAD) and Computer Aided Engineering (CAE) devices.
software tools are widely used. As an example, the use of
Finite Element Analysis (FEA) during the design of the Boeing
540 2009 7th IEEE International Conference on Industrial Informatics (INDIN 2009)
Authorized licensed use limited to: Instituto Politecnico de Braganca. Downloaded on January 15, 2010 at 18:01 from IEEE Xplore. Restrictions apply.
9LUWXDO 5HDO Delmia Automation tool from Dassault Systèmes provides a
(QYLURQPHQW :RUOG
3D environment where virtual systems can be modeled and
5DZPDWHULDO
2SSRUWXQLWLHV
GHPDQGV LQIRUPDWLRQ controlled, being part of a 3D Product Lifecycle Management
HQHUJ\
(PLM) which supports the product lifecycle from product
&RQWURODQG design to planning and realization [17]. Delmia Automation
LPSURYH
,WHUDWHWR
tool already enables the validation of the control logic in a
DFKLHYH
FRUUHFWQHVV
simulation environment and the programming of various
DQG Programmable Logic Controllers (PLCs), but the migration to
PDWXULW\ 0RQLWRULQJDQG
IHHGEDFN the real world still presents some obstacles:
− The validation is achieved by the simulation of the model
DQDOV\LVVLPXODWLRQ SURGXFWV but a validation using formal languages is missing.
FRQWURORSWLPL]DWLRQ
− The deployment into industrial controllers is strongly
³0DNLQJLW´
³0DNLQJLW´RQ
SK\VLFDOO\RQ dependent of the developed interfaces, requiring more
&RPSXWHUV IDFWRU\SODQWV integration and interoperability compatibilities.
Fig. 1. Virtual and Real Manufacturing Environments (adapted from [16]). − It is not prepared to accommodate multi-agent systems
and service-oriented automation ecosystems
B. Migration from the Virtual to the Real World Other software tools present even more limitations in terms
The migration process is well known within people and of validation and migration processes.
animals, representing respectively, the movement from one In terms of existing approaches to develop automation
country or region to another and the movement, usually control solutions, a comparison is illustrated in Fig. 2.
periodically, from one region or climate to another for feeding 6WDWH )XQFWLRQ ,(& ,(& +LJKOHYHO
80/ 3HWULQHWV
and breeding. The migration of birds is one typical example of 'LDJUDP &KDUWV
seasonal migration. In the context of this work, the concept of VHTXHQWLDO
EHKDYLRUV
migration concerns to the movement of a production control FRQFXUUHQW
EHKDYLRUV
application from the virtual environment to the real world,
FODULW\
considering hardware and software technologies.
FRPSOH[LW\
The achievement of an easy and smooth migration from the
WHVWDELOLW\
virtual environment to the real world requires the
GLVWULEXWLRQ
accomplishment of the following main requirements:
− Accuracy: the virtual model should be “calibrated” to YDOLGDWLRQ
represent exactly the real world, namely in terms of Fig. 2. Comparison of Modeling Methods for Automation Applications.
layout and device disposition, but also in terms of
structural and behavioral control aspects, considering the It is clear that State Diagrams, Function Charts and IEC
environment context founded in the real world. 61131-3 modeling approaches present several drawbacks to
− Hardware representation: the hardware constraints model and validate efficiently collaborative automation
should be taken into consideration during the virtual systems. IEC 61499 allows analyzing models of control
simulation, i.e. virtual devices should emulate correctly systems and proves their properties, but it requires the use of
the functionalities of the physical relatives, e.g. showing a external languages, such as Prolog and Net Condition/Event
correct mapping of the sensors and actuators connected to Systems [18]. In fact, only Petri nets allow simulating the
the industrial controllers I/Os. behavior model and consequently formally verifying the
− Integration: the deployment mechanism should simplify system specifications, using its own formalism. This capability
the migration of the control solution to the real world, to support the simulation and verification of the system
considering interfaces allowing integrating various correctness permits to achieve a sooth migration from the
industrial controllers available in the market. virtual design to the real control environment.
− Re-configuration: the use of model-driven approaches in
the development of control solutions allows simplifying III. FORMAL AND INTEGRATED APPROACH FOR SMOOTH
the re-configuration capability by reducing the MIGRATION
complexity and costs in the deployment phase. As previously referred, the development of production
A pertinent question is how existing software tools and control systems is usually carried out manually and not derived
approaches to develop collaborative automation solutions from a model-like description of the production system. The
address the smooth migration problem, fulfilling the described correctness of such design can only be validated in the
requirements. implementation phase, presenting high rates of
The use of 2D/3D software tools allows creating virtual misunderstanding and mistakes, and, as a consequence, it is
environments for production systems, presenting generally very expensive [8].
some limitations in terms of migration. As an example, the
2009 7th IEEE International Conference on Industrial Informatics (INDIN 2009) 541
Authorized licensed use limited to: Instituto Politecnico de Braganca. Downloaded on January 15, 2010 at 18:01 from IEEE Xplore. Restrictions apply.
A. High-level Petri nets Approach Once the control system is designed and simulated, being
The approach proposed in this work considers the High-level its specification verified in the virtual environment, it is
Petri nets as the formal language for the description of the required to set the system into operation by migrating the
behavioral models of the production system. Being a formal control solution to the real world. The methodology introduces
method to describe and validate certain typical relationships, as innovation the integration of modeling, analysis and
such as concurrency and parallelism, synchronization, resource validation processes, leading to a reduction of the development
sharing, mutual exclusion, monitoring and supervision [19], costs and time effort, supporting also the adaptation and re-
High-level Petri nets are suitable for the formal specification configuration needs.
and validation of production control systems. Using the
B. Continuum Development Tools
mathematical foundation associated to High-level Petri nets,
The application of Petri nets can range typical from systems
the analysis, validation and simulation can be performed: the
with predictable behavior to more complex ones with
first ones using the functional algebra theory, e.g. analyzing the
distributed participants. In any case, system engineering and
T- and P- invariants, and the other one by means of a token
associated tools are required to facilitate the developer's
game (simulation of the Petri net according to its rules)
intervention. From the Petri nets side, the practical usage is
considering the time parameter associated to the transitions.
limited by the lack of computer tools which would allow
handling large and complex nets in a comfortable way [20].
Therefore, the Continuum Development Tools has being
developed according to the described engineering approach to
provide a user-friendly environment for several engineering
tasks of service-oriented/multi-agent automation systems. It
covers the specification and configuration of automation
systems, the analysis and simulation, and also the operation of
the system.
As an example, Fig. 4 represents a screenshot illustrating the
simulation of the control logic for an automation system,
represented using a Petri net model.
The proposed High-level Petri nets engineering methodology Fig. 4. Main GUI of the Continuum Development Tools.
provides a catalogue of services that simplifies the
development life-cycle of collaborative automation systems, The integration of 2D/3D visualization tools, such as Delmia
from the design to the operation and re-configuration [8], as Automation and Arena, with the High-level Petri nets based
illustrated in the Fig. 3. It simplifies the definition and formal service-oriented engineering framework establishes a complete
specification of an “encapsulation process” in industrial support for the development life-cycle, offering the possibility
collaborative production systems. The dynamic behavior of to experiment in a virtual 2D/3D environment the production
collaborative (agent-based or service-oriented) control units is system, helping to achieve maximum production efficiency,
formally modeled and validated using High-level Petri nets, lower cost, better quality and short time to market.
which ensure a rigorous specification due to its powerful The resulted virtual simulation is driven by the High-level
mathematical foundation. As an example, when dealing with Petri nets based logic control engine provided by the
service-oriented systems, the services are associated to Continuum framework. In the proposed engineering
transitions that can be stepwise refined to get more detailed framework, the simulation process is executed iteratively,
models. modifying and adjusting the production control system
parameters, till reaching the correctness of the control system
542 2009 7th IEEE International Conference on Industrial Informatics (INDIN 2009)
Authorized licensed use limited to: Instituto Politecnico de Braganca. Downloaded on January 15, 2010 at 18:01 from IEEE Xplore. Restrictions apply.
behavior. The validation and simulation of the control system continue straight on to the end lifter, or turn in the direction of
during the design phase, allows reducing significantly the costs one of the two workstations.
and the number of mistakes detected in the operation phase.
After the complete verification of the correctness of the
virtual production system behavior, the production control
system is ready to be migrated from virtual environment to the
real world, as illustrated in Fig. 5.
power on the
system
virtual
simulation
Fig. 6. Physical and Virtual Representation of the Case Study Scenario.
2009 7th IEEE International Conference on Industrial Informatics (INDIN 2009) 543
Authorized licensed use limited to: Instituto Politecnico de Braganca. Downloaded on January 15, 2010 at 18:01 from IEEE Xplore. Restrictions apply.
nets model inside the Continuum platform, the status of the ACKNOWLEDGMENT
High-level Petri nets model is made visible in the graphical
The authors would like to thank the European Commission
Petri nets editor, giving information about the current marking
and the partners of the EU IST FP6 “Service-Oriented Cross-
and the enabling- and the firing-modes of the executed Petri
layer infrastructure for Distributed smart Embedded devices”
nets model. At this stage, alternative simulations can be
(SOCRADES) project, the EU FP6 “Innovative Production
performed in the virtual environment by changing the system
Machines and Systems” (I*PROMS) Network of Excellence,
parameters, introducing the degradation of the conveyors and
and the ICT FP7 “Cooperating Objects” (CONET) Network of
including more virtual pallets that leads to the increase of the
Excellence, for their support.
traffic congestion.
In spite of being an early prototype experimentation, some REFERENCES
important concepts were observed, namely, that complex [1] Wooldridge, M., “An Introduction to Multi-Agent Systems”, John Wiley
systems can be easily modeled, analyzed and simulated in a & Sons, 2002.
truly virtual environment, that combines 2D/3D simulation [2] Koestler, A., “The Ghost in the Machine”, Arkana Books, London, 1969.
[3] Leitão, P., Restivo, F., “ADACOR: A Holonic Architecture for Agile and
software tools and the High-level Petri nets computational Adaptive Manufacturing Control”, Computers in Industry, 57 (2), 2006,
environment. A demonstration video including this approach pp. 121-130, Elsevier.
can be seen at [4] Jammes, F. and Smit, H., “Service-oriented Paradigms in Industrial
Automation”, IEEE Transactions on Industrial Informatics, 1(1), 2005,
http://www.youtube.com/watch?v=0aRRvqEln2I. pp. 62–70.
Important to this work is the reference that the transition to [5] Leitão, P., “Agent-based Distributed Manufacturing Control: A State-of-
the real system is done in such a way that the control models, the-art Survey”, to appear in the International Journal Engineering
Applications of Artificial Intelligence, 2009, Elsevier.
previously analyzed in the virtual environment, remain the
[6] Babiceanu, R. and Chen, F., “Development and Applications of Holonic
same, being not necessary the re-engineering of the control Manufacturing Systems: a Survey”, Journal of Intelligent Manufacturing,
system. In the experimental case, a PC was used to host the nº 17, 2006, pp. 111-131.
control application running with the real devices. Moreover, [7] ISO/IEC 15909, 2000. High-level Petri Nets - Concepts, Definitions and
Graphical Notation. Final Draft International Standard ISO/IEC 15909.
the use of Web services between the control applications and [8] Leitão, P. and Colombo, A.W., “Petri net based Methodology for the
the virtual/real equipment facilitates not only the Development of Collaborative Production Systems”, Proceedings of the
communication, but also the integration of these virtual/real 11th IEEE International Conference on Emerging Technologies and
Factory Automation, 2006, pp. 819-826.
scenarios. These observations allow to conclude, in a [9] Iwata, K., Onosato, M., Teramoto, K., and Osaki, S., “A Modelling and
preliminary stage, that the control solution was smoothly Simulation Architecture for Virtual Manufacturing Systems”, Annals of
migrated from the virtual environment to the real environment. the CIRP, 44 (1), 1995.
[10] Bowyer A., Bayliss, G., Taylor, R., and Willis, P., “A Virtual Factory”,
International Journal Shape Model, 2(4), 1996, pp. 215-226.
V. CONCLUSIONS [11] Lin, E., Minis, I., Nau, D. and Regli, W., “Virtual Manufacturing User
Workshop”, Lawrence Associates Inc., July, 1994.
This paper discusses the use of virtual engineering [12] Marinov, V., “What Virtual Manufacturing is?”;
http://bosphorus.eng.emu.edu.tr/vmarinov/VM/VMdef.htm, 2000.
environments to support the fast, easier and structured design [13] Cachapa, D. Colombo, A. Feike, M. and Bepperling, A., “An Approach
of production control systems, taking special attention to the for Integrating Real and Virtual Production Automation Devices
requirements for an easy and smooth migration from the virtual Applying the Service-oriented Architecture Paradigm”, Proceedings of
the 12th IEEE International Conference on ETFA, 2007, pp. 309-314.
to the real production system.
[14] Ali, S. A., Seifoddini, H. and Sun, H., "Intelligent Modeling and
Aiming to achieve a formal and structured framework for the Simulation of Flexible Assembly Systems", Proceedings of the 37th
development of collaborative automation systems, the Conference on Winter Simulation Conference, 2005, pp. 1350-1358.
proposed approach considers the use of High-Level Petri nets [15] Leitão, P., Mendes, J.M., Bepperling, A., Cachapa, D., Colombo, A.W.
and Restivo, F., “Engineering Tools for the Integration of Service-
to drive the system's behavior (both in virtual and real worlds), oriented Production Systems”, to appear in the Proceedings of the 13th
making easier and reducing the effort in the transition from the IFAC Symposium on Information Control Problems in Manufacturing,
design and analysis phases done in the virtual environments to Moscow, Russia, 2009.
[16] Saadour, M. and Sandoval, V., “Virtual Manufacturing and its
the real-time operation phase. In fact, taking advantage of the Implications”, Virtual Reality and Prototyping, June, 1999.
powerful mathematical foundation associated to the Petri nets [17] ARC Advisory Group, “DELMIA V5 Automation Platform: Merging
formalism, a rigorous specification, analysis and verification of Digital Manufacturing with Automation”, ARC White Paper, February
2006.
the system correctness can be performed during the design [18] Vyatkin, V. and Hanisch, H.-M., “A Modeling Approach for Verification
phase and before to deploy the control solution into practice. of IEC 61499 Function Blocks using Net Condition/Event Systems”,
The simulation procedure can be combined with the integration Proceedings of the 7th IEEE International Conference on Emerging
with 2D/3D software tools, allowing an easy and user-friendly Technologies and Factory Automation (ETFA´99), 1999, pp. 261-270.
[19] Murata, T., “Petri nets: Properties, Analysis and Applications”, IEEE,
interaction during the design and simulation processes. vol. 77, 1989, pp. 541-580.
The proposed concepts were illustrated using the Continuum [20] Suraj, Z., Fryc, B., Matusiewicz, Z. and Pancerz, K., “A Petri Net System
platform to develop the service-oriented control application for -an Overview”, Fundamenta Informaticae, 71 (1), IOS Press Amsterdam,
The Netherlands, January 2006, pp. 10 -119.
the experimental case study, emulated in the Delmia [21] Microsoft, 2006. Devices Profile for Web Services. Avail. at
Automation tool and hosting virtual services. http://schemas.xmlsoap.org/ws/2006/02/devprof/
544 2009 7th IEEE International Conference on Industrial Informatics (INDIN 2009)
Authorized licensed use limited to: Instituto Politecnico de Braganca. Downloaded on January 15, 2010 at 18:01 from IEEE Xplore. Restrictions apply.