PREREQUISITES FOR DEVELOPING A NEW SIMULATION AND MANUFACTURING ARCHITECTURE

BULETINUL INSTITUTULUI POLITEHNIC DIN IA I Publicat de Universitatea Tehnic Gheorghe Asachi din Ia i, Tomul LVII (LXI), Fasc. 4, 2011 Sec ia CONSTRUC II DE MA INI PREREQUISITES FOR DEVELOPING A NEW SIMULATION AND MANUFACTURING ARCHITECTURE Received: June 12 th, 2011 Accepted for publication: July 25 th, 2011 BY BOGDAN PÎRVU and IOAN BONDREA Lucian Blaga University, Sibiu, Romania Department of Mechanics Abstract. Real process-oriented information is never exchanged when coupling the real world manufacturing line with its simulation. Also the perspectives over the same real manufacturing process are different regarding the production engineer and the automation engineer. By analyzing new emerging technologies, with future distributed controllers and OPC UA middleware, we will formulate the motivation, concepts and a first prototype implementation of a new simulation and manufacturing approach that could unify the two different engineering perspectives and offer a holistic view of the real processes and a better manufacturing systems simulation and automation connection. This approach contributes to manufacturing s systems development by indicating how the simulation and its real manufacturing system counterpart could be better harmonized, leading to an increased agility of the production facility. This is achieved with an enhanced transfer of information from manufacturing simulation to production that leads to a better view and an easier process construction of the real manufacturing process, both in simulation and at the controller level. Key words: SOA, OPC UA, IEC 61499, Digital Factory. 1. Introduction The path towards achieving increased company agility is formulated by different concepts and a recent one is the Digital Factory. The concept states, that in order to achieve attributes like adaptability, there should be a close connection between the real factory and its virtual counterpart; the vision is to Corresponding author: bogdan.pirvu@ulbsibiu.ro

204 Bogdan Pîrvu and Ioan Bondrea optimize all parameters in the virtual world (proactive or during operation) and then to migrate the data for the usage of the real production system. The concept is thoroughly presented in the VDI 4499 standard (VDI, 2008). One of the major issues, deriving from this ideal (the close relationship of the two worlds), is the relationship between production planning and the real system, concerning the communication between simulation and the real production line. Imagining an ideal case, a complete and error proof migration of an optimum planning scenario to a reconfigurable or flexible production facility (and vice-versa) should offer the agility needed perform at the highest level considering the turbulence of the current global market environment. Today, the communication between simulation and the real line requires the usage of OPC (Ole for Process Control), since only messages can be exchanged. Major reasons are on one side the automated line - with the heterogeneous and proprietary controller solutions (PLCs stem from different vendors) - and on the other side the simulation software tools (the material flow and DMU simulation is focused by this work). Both of which, from the beginning of development, were never intended to provide functionality or mechanisms to transfer results from one to the other; it lacks cohesion. Each perspective (regarding the automation and manufacturing simulation) over the same manufacturing processes (Fig.1) is very different and not complete (one deals with low level programming and the other with mechanical aspects kinematics, collisions, part design). Here is one of the major issues that do not allow an easy and smooth migration from one environment to the other. Fig.. 1 Simulation and controller perspectives over the same real process. This paper presents current and future controller, middleware and simulation technologies in order to formulate the prerequisites and the first prototype implementation of a new simulation architecture that has process orientation as its main focus point.

Bul. Inst. Polit. Ia i, t. LVII (LXI), f. 4, 2011 205 2. Current Simulation, Controller and Middleware Approaches Regarding simulation, one current possibility is to model the behavior of the system by internal logic within the simulation. By this, the various states of the system (relevant for the controller) are modelled by Sequential Function Charts (SFC) in Delmia (Schleipen et al., 2009) and global variables in Plant Simulation (Pîrvu et al., 2010). Then the SFCs or the global variables are then linked with I/Os (ports). When a certain event in simulation happens their state (SFC or the global variable) also changes (this will trigger an event via OPC - at the controller level). Another current option (in case of Delmia) would be to programme the logic (PLC programme) on the PLC and to connect I/Os from PLC s to the simulation ports (visualisation of the logic in an animated virtual world). Obviously a mix of those two is also possible. Dassault Systemes s tool Delmia Automation seems to offer better performance compared to the Siemens PLM Process Simulate platform by analyzing the practical virtual commissioning industry examples (e.g. Sanyo Machine Works, GM) and usage by researchers as a platform to test new ideas and approaches( e.g. (Leitao et al., 2009), (Cachapa et al., 2007)). Leitao et al. underlines in their paper that Delmia Automation (other tools offer even more limitations in terms of validating and migration process from virtual environment to the real world) enables validation of the control logic in a simulation environment and offers the possibility of programming various PLCs. But, the migration to the real line has the following obstacles (Leitao et al., 2009): validation is achieved by means of simulation model and a validation using formal language is missing ; deployment into controllers is dependent on developed interfaces, requiring more integration and interoperability compatibilities ; it is not prepared to allow multi-agent systems and service-oriented ecosystems. To conclude, even probably the best digital manufacturing tool today (Delmia Automation) doesn t offer a smooth transition from the virtual environment to the real world and vice versa. Regarding the controller aspects, the IEC 61131 standard has been largely adopted by the automation industry, but could reach the end of its technological lifecycle because it is not able to provide the agility needed today in manufacturing (Tata & Vyatkin, 2009). The IEC 61499 standard, which is at this time in quasi-experimental state, promises to offer manufacturing the level of agility needed to face current and future exigencies (Sunder et al., 2006). Worth mentioning is the ISaGRAF environment, which is the first commercial available tool to support the Function Block (FB) standard (IEC 61499) and also the IEC 61131 standard. Also a current trend is to use PC-based solutions for controlling manufacturing systems in the context of SOA (Service-Oriented Architectures) applied at the automation level. By using Web-based technologies (DPWS- Device Profile for Web Services, UPnP- Universal Plug

206 Bogdan Pîrvu and Ioan Bondrea and Play, etc.) it is possible to control an industrial manufacturing system using standard Web technologies; a remarkable example the doze maker was developed within the SIRENA (Service Infrastructure for Real-time Embedded Network Applications) project that demonstrated the capabilities of SOA applied to the automation systems. Regarding middleware, OPC UA (Unified Architecture) keeps all the functionality of Classic OPC but uses state of the art Web services technology and an optimized binary TCP protocol for high performance communication instead of Microsoft s COM/DCOM technology. Thus OPC UA becomes platform-independent overcoming one of the major issues of classical OPC. OPC UA specifies how data is exchanged, while standard information models specify what information is exchanged (Mahnke et al., 2009). OPC UA offers the possibility to use information models (some basic ones are defined). Using already defined information models one can define other information models which are custom. With the introduction of metadata it is possible not just the transmission of variables, but at the same time how the transmitted data should be interpreted (data over data). The major downside at this point is the data transmission speed with OPC UA. Because more information is exchanged the data transmission speed is low (compared at just a couple of variables, it is clearly slower than classical OPC) (Mahnke et al., 2009). To summarize, it can be concluded that: a smooth migration from the virtual environment to the real world is not currently available; modelling in most cases is imprecise, takes a lot of time and is difficult to build in the context of controller logic; there are always two perspectives (controller programmer and simulation engineer) that are rarely identical. This leads to false (not entirely realistic) simulation results (Pîrvu & Bondrea, 2011). 3. Problem Statement An approach that links manufacturing simulation with future controllers (IEC 61499 or PC- based) while having a holistic view of the real processes that take place in manufacturing is missing. The approach refers to a system architecture which is: process-oriented, component-based and relies on loose coupling of functionalities regarding manufacturing processes. Also, simulation (material flow and DMU simulation is regarded as a general process planning simulation tool that could be also used for virtual commissioning. Virtual commissioning and manufacturing planning are closely linked, because correct planning results are achieved only when the virtual model s behaviour mirrors the real manufacturing system (Pîrvu & Bondrea, 2011). The IEC 61499 standard defines a new controller architecture to solve the distributed controller technology problem using FBs, as the main logical unit, and, responsible for their activation is an event-based execution model. The IEC 61499 and SOA (regarding automation, using for example the DPWS technology) approach can be considered suitable in the above stated approach

Bul. Inst. Polit. Ia i, t. LVII (LXI), f. 4, 2011 207 allowing for a true process orientation approach. This way we can distinguish the real processes that are happening in the factory and to programme the controller accordingly (by having the overview of the process) and not focus on I/Os. This is achieved by splitting down the manufacturing process into atomsub-processes. They can be realized inside an IEC 61499 function block network or at the microcontroller/pc level enabling a way to realize subprocesses in a standardized way independent of proprietary PLC solutions. The way of putting together control programs resembles an orchestration on higher level abstraction. As a middleware solution OPC UA is suitable for enabling a better coupling between manufacturing simulation and real production systems because: it is standardized, capable of exchanging more data (custom information models can be build), with a better security and platform independent. 4. Target, Approach and First Prototype Implementation Considering the fore mentioned arguments, a process-oriented simulation architecture that links simulation with future controllers while having a holistic view of the real processes is the target of current and future work. One key element that brings together the real world of production and the simulation is the architecture of functionalities where the complete manufacturing process is broken down into small process steps that are context independent (Pîrvu & Bondrea, 2011). Such process decomposition is depicted in Fig. 2. Fig. 2 The architecture of functionalities example. In order to implement the architecture of functionalities, first, the process model should be derived for a given type of manufacturing equipment taking

208 Bogdan Pîrvu and Ioan Bondrea the constraints of state of the art soft-research results into consideration. The first prototype implementation focused a simple industrial pill filling process used in the pharmaceutical industry (Fig. 3). The automated module s functionality is to load pills in the boxes that come on the conveyor. The number of pills is read from an RFID tag which is glued on the pill box, with the read-transponder. After a pill box is loaded, on the RFID tag is written with the second transponder the write transponder - that the box is full. In the loading process the pills are unloaded with a disc that has special frontal places where only one pill fits (Pîrvu et al., 2010) The pills which are dropped from the wheel in the pre-loading pill buffer are counted with a light barrier attached to the source s structure (Fig. 4). The pre-loading pill buffer will open only if the pill box is still present on the container in the loading position. The presence of the box that has to be filled is sensed by an ultrasonic sensor and the containers with inductive sensors that activate or deactivate the stoppers. There can be only one container in the loading area; all others will wait until that container will leave the loading area. Fig. 3 Automated module functionality Fig. 4 Pill numbering process As a first prototype implementation of the new approach (Fig. 5) uses the current available technologies as a proof of concept. The orchestration of the simulation model and production system is done from the simulation (Plant Simulation 9) using an OPC interface (SimaticNET) for communication. Plant Simulation 9 is a discrete, event-oriented simulation program. In the simulation model there have been used standard objects or custom objects with particular functionality to fit the required needs of the modelling. The simulation model is constructed into a separate frame as the one of the orchestration. By doing so, an important degree of logical separation was achieved and the local aspects were encapsulated into objects within the simulation model and, the global aspects concerning the co-ordination of the sub-processes were encapsulated into the orchestration frame.

Bul. Inst. Polit. Ia i, t. LVII (LXI), f. 4, 2011 209 The OPC interface object from simulation is used to coordinate the events happening within the simulation model with the PLC (Siemens S7-300) that is in charge of the production line control. The next implementation focuses on the improvement of the implementation of the basic concept (Fig. 5) by orchestrating the simulation and manufacturing outside of the simulation using a Petri net engine (Matlab). This engine orchestrates based on a Petri net description of the real process. A major future step is the development of services; services will be developed in order to incorporate the basic ideas of distributed-controlling capable units (e.g. IEC 61499) and of simulation requirements as a technical implementation of the process modules in the real world and into simulation. This way, only the coordination of the service succession by an orchestrator will determine the entire functionality of the real manufacturing line also allowing for fast and precise way to build up the simulation. In the end the applicability of the new approach will be showed as a proof of concept implementation that will incorporate the two major steps. At the end the proof of concept will be evaluated in terms of the environment s set-up simulation efforts for the real manufacturing lines. Fig. 5 Basic architecture and first prototype implementation. 5. Conclusions The bibliographical research focused the identification of current and future controllers, middleware and manufacturing simulation solutions in order to underline the current limitations. Even more important, the justification behind a better approach that enables a closer link between manufacturing

210 Bogdan Pîrvu and Ioan Bondrea simulation with the real manufacturing systems - that leads to more agile systems is stated taking into consideration the new emerging technologies that will be available in the near future, mainly considering IEC 61499/ SOA enabling controllers and OPC UA middleware. An early prototype implementation of such architecture (links manufacturing simulation with real manufacturing systems) is shortly presented and it is using current available technologies (Fig. 5). Orchestration of the real manufacturing line and the simulation model is done in this early phase in the simulation environment (Plant Simulation). The real line is equipped with an IEC 61131 compliant controller (Siemens S7-300 controller) and it is being controlled by the simulation environment by using an OPC interface. The work which is stated within this paper contributes to manufacturing s science development because the novel approach should indicate how simulation and real manufacturing lines can be better harmonized, thus leading to more versatile companies. This is achieved with an enhanced transfer of information from simulation manufacturing to production and a better view of the real manufacturing process, both in simulation and at the controller level. Acknowledgements. I would like to thank Dr. Ing. Jochen Schlick - Deputy Director of IFS (Innovative Factory Systems) within the German Research Centre for Artificial Intelligence - for the continuous support and access to the latest technologies. Research done within the POSDRU/6/1.5/S/26 project co-financed from the European Social Fund through the Operational Programme Human Resources Development 2007-2013. REFERENCES VDI, (2008), Digitale Fabrik Grunflagen Blatt 1, Verein Deutscher Ingenieure, Düsseldorf, 1-52. Schleipen M., Sauer O., Friess N., Braun L., Shakerian K., (2009), Production Monitoring and Control Systems within the Digital Factory, Proceedings of the 6th CIRP-Sponsored International Conference on Digital Enterprise Technology, Hong Kong-China, 711-724. Pîrvu B., Bondrea I., Marin R., (2010), Modeling and Control of an Automated Module using Discrete Event Simulation and Object-Based Modeling, Academic Journal of Manufacturing Engineering, 8, 2, Timisoara, 63-69. Leitao, P., Mendes J. M., Colombo A. W., (2009), Smooth migration from the Virtual design to the real manufacturing control, 7th International Conference on Industrial Informatics, Cardiff-UK, 539-544. Cachapa D., Colombo A., Feike M., Bepperling A., (2007), An Approach for Integrating Real and Virtual Production Automation Devices Applying the Service-oriented Architecture Paradigm, 12th IEEE Conference on Emerging Technologies and Factory Automation, Patras-Greece, 309-314.

Bul. Inst. Polit. Ia i, t. LVII (LXI), f. 4, 2011 211 Tata P. & Vyatkin V., (2009), Proposing a novel IEC61499 Runtime Framework implementing the Cyclic Execution Semantics, 12th IEEE Conference on Emerging Technologies and Factory Automation, Patras-Greece, 416-421. Sunder C., Zoitl A., Christensen J.H., Vyatkin V., Brennan R.W., Valentini A., Ferrarini L., Strasser T., Martinez-Lastra J.L., Auinger F., (2006), Usability and Interoperability of IEC 61499 based distributed automation systems, IEEE International Conference on Industrial Informatics, Singapore, 31-37. Mahnke W., Leitner S.-H., Damm M., (2009), OPC Unified Architecture, Springer Verlag, Berlin-Heildelberg. Pîrvu B. & Bondrea I., (2011), Motivation behind a new Approach that links Manufacturing Simulation with future Controllers, International Conference on Mechanical Engineering and Technology, London-UK, to be published. PREMISELE PENTRU DEZVOLTAREA UNEI NOI ARHITECTURI DE SIMULARE SI FABRICATIE (Rezumat) Inter-conectarea sistemelor reale de fabrica ie cu modelele digitale (simularea) este dificil datorit perspectivelor diferite asupra aceluia i proces real de fabrica ie în cazul responsabilului cu programarea logicii de automatizare i a inginerului de proces. Astfel, migra ia datelor dintr-un mediu în cel lalt (simulare unitate de comand ) este dificil, incomplet i pu in transparent. În contextual dezvolt rii arhitecturilor de comand distribuite i noii specifica ii de middleware OPC UA, formul m motiva ia i premisele pentru elaborarea unei abord ri arhitecturale noi în privin a simul rii ce poate unifica cele dou perspective inginere ti dar i s ofere o perspectiv holistic i o rela ie de inter-conectare mai bun dintre simulare i sistemele reale de fabrica ie. Sunt prezentate elementele noi care sunt aduse de standardul IEC 61499 i IEC 62541, stadiul actual cu privire la rela ia simulare cu sistemele reale de fabrica ie, contextul în care noile concepte din standardele IEC ar putea fi utilizate în sprijinul ipotezei noii arhitecturi dar i o privire asupra modalit ii de a transpune noua abordare in realitate. Sunt abordate secven ial, stadiul general actual, solu iile de simulare actuale, abord rile actuale în cazul PLC-urilor i al solu iilor de middleware i concluziile despre modul actual de interconectare al simul rii cu automatele programabile. În urma analizei stadiului actual se formuleaz problemele actuale i viitoare. Dintre acestea se focalizeaz problema arhitecturii orientat pe proces i se formuleaz o posibil abordare cu privire la solu ionarea sa. Implementarea unui prim prototip ce prive te noua abordare arhitectural este i ea prezentat. Datorit disponibilit ii/implement rii reduse a OPC-UA i IEC 61499 în cadrul primului prototip de implementare se utilizeaz tehnologiile curente; astfel ca solu ie midlleware este OPC, iar ca unitate de comand un automat programabil Siemens S7-300 (conform IEC 61131). Orchestrarea sub-proceselor i modelul simulat se realizeaz amândou în simulare (pachetul software Plant Simulation) iar comanda liniei reale se face utilizând o interfa OPC. Astfel din simulare comand m atât modelul de simulare cât i automatul programabil cu care sistemul de fabrica ie este prev zut.