Scholarly article on topic 'Continuous Integration of Field Level Production Data into Top-level Information Systems Using the OPC Interface Standard'

Continuous Integration of Field Level Production Data into Top-level Information Systems Using the OPC Interface Standard Academic research paper on "Computer and information sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Procedia CIRP
OECD Field of science
Keywords
{Interoperability / "Information integration" / Ontology / "Semantic data" / OPC / "OPC UA"}

Abstract of research paper on Computer and information sciences, author of scientific article — Max Hoffmann, Christian Büscher, Tobias Meisen, Sabina Jeschke

Abstract On the way to the fourth industrial revolution, one major requirement lies in reaching interoperability between hardware and software systems. Especially real-time propagation of shop floor information in top-level production planning and control systems as well as the consolidation of distributed information into a consistent data basis for comprehensive data analysis are still missing in most production environments. Existing approaches to serve interoperability through standardized interfaces are limited by proprietary data exchange protocols and information models. Within industrial manufacturing and automation, standardization attempts between these systems are primarily focused on industrial interfaces like OPC/OPC-UA. However, the aggregation of data created by devices like sensors or machinery control units into useful information has not been satisfactorily solved yet as their underlying models are carried out using different modeling paradigms and programming languages, thus intercommunication is difficult to implement and to maintain. In this work, an integration chain for data from field level to top-level information systems is presented. As Manufacturing Execution Systems or Enterprise Resource Planning tools are implemented in higher programming languages, the modeling of field level information has to be adapted in terms of a semantic interpretation. The approach provides integration capabilities for OPC-conform data generated on the field level. The information is extracted from low level information systems, transformed according to object-oriented programming paradigms and object-relational standards and finally integrated into databases that allow full semantic annotation and interpretation compatible to a common information model. Hence, users on management levels of the enterprise are able to perform holistic data treatment and data exploration along with personalized information views based on this central data storage by means of a reliable and comfortable data acquisition. This increases the quality of data and of the decision support itself, as more time remains for the actual task of data evaluation.

Academic research paper on topic "Continuous Integration of Field Level Production Data into Top-level Information Systems Using the OPC Interface Standard"

ELSEVIER

48th CIRP Conference on MANUFACTURING SYSTEMS - CIRP CMS 2015

Continuous integration of field level production data into top-level information systems using the OPC interface standard

Max Hoffmann^*, Christian Büschera, Tobias Meisena, Sabina Jeschkea

a Insitute of Information Management in Mechanical Engineering (IMA) ofRWTH Aachen University, Dennewartstrasse 27, 52068Aachen, Germany * Corresponding author. Tel. : +49-241-80911-34; fax: +49-241-80911-22. E-mail address: max.hoffmann@ima.rwth-aachen.de

Available online at www.sciencedirect.com

ScienceDirect

Procedía CIRP 41 (2016) 496 - 501

www.elsevier.com/looate/procedia

Abstract

On the way to the fourth industrial revolution, one major requirement lies in reaching interoperability between hardware and software systems. Especially real-time propagation of shop floor information in top-level production planning and control systems as well as the consolidation of distributed information into a consistent data basis for comprehensive data analysis are still missing in most production environments. Existing approaches to serve interoperability through standardized interfaces are limited by proprietary data exchange protocols and information models. Within industrial manufacturing and automation, standardization attempts between these systems are primarily focused on industrial interfaces like OPC/OPC-UA. However, the aggregation of data created by devices like sensors or machinery control units into useful information has not been satisfactorily solved yet as their underlying models are carried out using different modeling paradigms and programming languages, thus intercommunication is difficult to implement and to maintain.

In this work, an integration chain for data from field level to top-level information systems is presented. As Manufacturing Execution Systems or Enterprise Resource Planning tools are implemented in higher programming languages, the modeling of field level information has to be adapted in terms of a semantic interpretation. The approach provides integration capabilities for OPC-conform data generated on the field level. The information is extracted from low level information systems, transformed according to object-oriented programming paradigms and object-relational standards and finally integrated into databases that allow full semantic annotation and interpretation compatible to a common information model. Hence, users on management levels of the enterprise are able to perform holistic data treatment and data exploration along with personalized information views based on this central data storage by means of a reliable and comfortable data acquisition. This increases the quality of data and of the decision support itself, as more time remains for the actual task of data evaluation.

© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.Org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of the scientific committee of 48th CIRP Conference on MANUFACTURING SYSTEMS - CIRP CMS 2015

Keywords: Interoperability; Information integration; Ontology; Semantic data; OPC; OPC UA

1. Introduction

In terms of the fourth industrial revolution - also referred to as Industry 4.0 - it is the aim of the manufacturing industry to bring together all information from distributed production facilities and levels and in a holistic and continuous way to provide decision support on various levels and throughout the entire process chain. This includes not only the gathering of information from different levels and areas of the production. One major focus also consists in creating flexible automation systems and vertical control flows to enable reconfiguration capabilities of the production in real-time.

However, most automation systems in existing factories are characterized by static hierarchies and rigid control/automation flows. In most cases, control flows are organized top-down, whereas information flows and the gathering of data are bottom-up. Traditionally, the planning of processes and information exchange in industrial environments is inspired by the automation pyramid [1], which serves as a general design pattern for creating information and communication infrastructures (ICT) for the industry. A simple representation that focuses on the main layers of the automation pyramid is depicted in Figure 1. On top of the automation pyramid the Enterprise Resource Planning (ERP) systems are located.

2212-8271 © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.Org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of the scientific committee of 48th CIRP Conference on MANUFACTURING SYSTEMS - CIRP CMS 2015 doi: 10.1016/j.procir.2015.12.059

DCS ISCADA

Field Level

Fig. 1. Automation pyramid for static control and information flows

ERP systems perform the resource planning of a company in terms of human and material resources. They are primarily used for long-term planning, e.g. in terms of process chains [2]. Manufacturing Execution Systems (MES) are responsible for mid-term production planning and execution. During the manufacturing, Process Distributed Control Systems (DCS) as well as Supervisory Control and Data Acquisition (SCADA) systems control conditions and system states during operation to avoid critical problems or serious failures in the production flow. Programmable Logic Controllers (PLC) are digital computers that are in charge of controlling signals of field devices like the machinery or other components of industrial environments. The field level is characterized by low level devices that are represented by the machines or sensors. These devices generate data that is needed to perform process optimization or to detect problems in the production flow.

Although there is a connection especially between the field level devices and PLC as well as SCADA systems, the information flow is mostly limited to rigid interfaces that cannot dynamically interpret and aggregate information in an appropriate way. In most cases, the data flow is limited to direct control of the production, i.e. MES are mostly not provided with real-time information from the field level, e.g. to perform rescheduling, real-time or even self-optimization [3]. This issue is due to the lack of powerful interfaces between low level and high level production systems. Another reason for preferring rigid interfaces over flexible automation systems is the aim for a complexity reduction on the field level, which is highly connected to robust and well-determined operational processes in the production to make processes more secure. However, whereas low level networks are primarily based on the communication over proprietary bus systems, high level information systems require networking connections that are based on (Industrial or office) Ethernet technologies.

In industrial practice, most of low level production systems are based on traditional data acquisition technologies as the Object Linking and Embedding (OLE) for Process Control -also known as OPC - that is performant and reliable on the one hand, however, on the other hand, very limited in terms of intelligent classification or annotation of data. Another drawback is lacking connectivity to networks of higher information systems. As a result, data from field level devices can only be manually integrated. This limits the usage of production information to batch-wise processing. This issue constitutes a

major issue in providing real-time functionalities and optimization capabilities for production environments, which are required for addressing the ideas of the Industry 4.0.

In order to address this problem, production environments could be equipped with novel technologies like OPC Unified Architecture (OPC-UA), which involves an adaptation of the entire information and communication infrastructure (ICT) of a manufacturing enterprise. As these major changes of ICT environments within companies are usually connected to a high amount of time and money, most enterprises refuse a radical modernization of their running systems.

As a result, the true challenge of enabling scenarios of the Industry 4.0 in existing production sites is to extend capabilities of the hardware infrastructure that is in use to implement modern ways of information management in currently running environments without designing an entirely new process.

It is the aim of the present paper to introduce an approach to interconnect low level production networks that are based on traditional interface technologies like OPC with higher level information systems like MES or ERP by bridging OPC-based production information over an OPC-UA interface. By enabling full interoperability between these technologies, the data from field devices can be appropriately enriched with semantic information of the production environment or with additional meta-data that is important for data treatment steps at a later point or for real-time optimization purposes.

A real-time simulation of a combined OPC/OPC-UA production network is provided in order to proof the added value of creating generic interoperability between the technologies. While addressing these demands for interoperability, the paper attempts to answer the following research questions:

• Is it possible to couple OPC and OPC-UA based system within a production environment in a generic way?

• How can OPC components or sensors be appropriately simulated in order to investigate a holistic usage of data from OPC-based devices in higher information systems?

• Is it possible to simulate the behavior of such production networks under real conditions, i.e. in real-time?

To find resilient answers to these questions, the state of the art of automation systems with a focus on OPC technologies is summarized in section 2. In section 3, an approach for simulating a combined OPC/OPC-UA production network with dynamic data exchange and enrichment capabilities is described in detail. Section 4 validates the modelling approach by simulating a manufacturing environment based on a use case that has been derived from information of a real production environment. Section 5 concludes the results of the present paper and investigates further research opportunities in order to expand our research results into industrial practice.

2. State of the Art

In traditional automation, low level production facilities are integrated into the ICT in a manual and rarely standardized way. To connect field devices with higher systems of the automation pyramid, interfaces/drivers have to be carried out specifically for each different device in the machine layer.

To facilitate the workflow of embedding devices into the ICT, standard interfaces or data protocol conventions are used

to enable plug-and-play or - in terms of the manufacturing area - plug-and-produce capabilities, which work similar to the connection of devices in computational environments.

Thus, in the beginning of the 1990s, manufacturing enterprises and members of the automation industry attempted to carry out a standard interface that was based on the Windows NT standard as most widespread operating system throughout all companies. This Operating System (OS) provides the OLE technology to interconnect multiple applications on the OS. The aim of the manufacturing companies was to establish similar approaches for connecting field devices with control systems of the automation layer. In 1995, a task force of big companies and automation providers like Siemens, Rockwell Automation, General Electric and ABB came up with the OLE for Process Control (OPC) standard, which uses the Distributed Component Object Model (DCOM) [4] for the linking of production facilities [5]. The central specification of this approach is the OPC Data Access (OPC DA) specification and has been published in 1996 [6]. By standardizing the access to data in automation systems, it is possible to embed driver or interface specific information directly into the field devices. This approach enables plug-and-produce capabilities without manually configuring each device.

The basic communication functionalities of OPC DA are based on the server-client principle. The OPC Server reads and propagates data from field level or from other data generating devices. In terms of building an OPC network, the OPC Client creates an instance of the server. This server represents a single device or a group of data sources (Figure 2).

The OPCItems represent concrete objects, e.g. sensors in an automation system. These can be grouped to OPCGroups for similar items. The OPCServer represents these items and propagates single or aggregated data into automation systems. OPC Servers can store information, whereas OPC Clients read and redistribute these information. Besides, every OPC Client can also function as an OPC Server.

Due to the success of this approach, the industrial users demanded for additional functionalities for the OPC standard, necessary for a full migration of the automation environment. These demands resulted in two additional OPC specifications, the OPC Alarms & Events (OPC A&E) and the OPC Historical Data Access (OPC HDA) specification in 1999 and 2001. The OPC A&E specification provides services to trigger realtime actions based on events or critical system states that could be harmful to the process or to the automation systems. The OPC HDA specification provides functionalities to access data from previous processes for data acquisition purposes.

The success of OPC in the 1990s was mainly due to the

Object / Instance

OPC Client creates OPCServer

OPCGroup OPCGroup

OPC Item II ■ OPC Item

Fig. 2. OPC server-client principle and item organization

high performance of the DCOM technology and due to the robustness of the resulting automation system [6]. However, the composition of OPC-based automation systems is rather rigid and hierarchical. Another drawback of the OPC standard is the lack of communication capabilities in networks as the DCOM standard cannot be properly configured to work with firewalls. As a result, the OPC consortium figured out ways to communicate over internet based systems, accordingly publishing the OPC XML-DA standard, which enables the propagation of OPC-based information via web services [5].

However, OPC XML-DA turned out to be comparatively slow, which is mainly due to the high overhead of information that has to be integrated in each XML message for each piece of information sent between OPC servers and clients.

Due to the resulting interoperability problems of classical OPC solutions, and due to the dependency on Windows-based systems and bad configurability characteristics in terms of industrial networks, the demand for a modernized version of OPC rose up from the industry. As a consequence, the OPC Unified Architecture (OPC-UA) standard had been carried out, which attempts to combine all specifications of the OPC Classic standard, and to extend their functionalities while enabling platform independence and interoperability [7].

One major functionality of OPC-UA is the possibility to configure lightweight and quick connections over the internet. As OPC-UA is not solely based on DCOM, the communication is performed over network borders and firewalls. This new approach solves major security problems and allows data exchange between distributed systems, as an encapsulation of system devices into a dedicated network, which is - unlike to OPC Classic - not necessary when using OPC-UA [8].

By providing abstract services, OPC-UA delivers basic functionalities to create a service-oriented architecture (SOA) of the automation system [8,9]. OPC-UA integrates different specifications of OPC into a single set of services. Thus, components and automation networks based on OPC Classic can theoretically be integrated into OPC-UA networks [10].

Another advantage of OPC-UA is the configurability of the new standard. The OPC-UA specification only specifies the message format of the information that is send. Unlike to OPC Classic, OPC-UA does not specify an API [8]. Hence, the user of the OPC-UA server is able to use or implement an API of his personal preference. All communication in the OPC-UA standard is performed using the Communication Stack. There is a client-side and a server-side communication stack. Both communication stack APIs can be developed in individual programming languages, as long as their concepts support the technology mapping given by the OPC-UA specification [8].

The described configurability of the OPC-UA application programming interface provides the user with a freedom of choice in terms of the technology or the programming language that is used to access the information in OPC-UA networks. This way, OPC-UA delivers a flexible usability in different environments and for different purposes. This availability is linked to several advantages:

• The application or the purpose of an information system determines the properties of the OPC-UA communication stack API, e.g. whether the application is complex or lightweight for the usage in small embedded devices.

• The communication with OPC-UA network can be integrated into existing enterprise communication systems.

• Tool interoperability with other systems or components of the factory is guaranteed as the programming language is of free choice. This enables the embedding of OPC-UA systems in business process, e.g. data integration chains.

The advantages of OPC-UA are able to address challenges that are introduced by the Industry 4.0. Thus, many scenarios that are in the scope of the fourth industrial revolution can be realized, if automation systems are fully equipped with UA capable devices and according infrastructures. The embedding of OPC-UA into industrial environments is the first step towards the goals of the Industry 4.0, however there are many more challenges to meet besides technical and syntactical heterogeneity in industrial production. The next steps would consist in creating structural and semantic standards.

However, existing production and automation systems are still in use of traditional automation systems, e.g. based on OPC or other rigid interface standards. Most manufacturers do not want to change their entire automation system or simply cannot take the risk of quitting a running system.

This results in a demand of embedding devices like sensors or automation components that are still based on traditional interfaces into an infrastructure that is designed according to the OPC-UA specification. The migration of DCOM-based OPC servers and clients into an OPC-UA infrastructure can be performed in several ways. The first way is to wrap the OPC components using the OPC-UA web service standards, the second approach is to directly implement the OPC Classic protocol according to the OPC-UA specification. [8]

The wrapping approach is performed by storing the COM/ DCOM-based information into a web service envelope. This solution is not only slow due to the recombination of two different data formats, there are also security-related issues as the number of exploits to consider increases significantly.

In contrast to the wrapping of OPC DA components, the direct implementation provides several advantages in terms of homogeneity of the resulting application environment. The arising systems are easier to maintain and easier to update as there is only one technology to consider, which is OPC-UA.

However, there are no sufficiently generic approaches to migrate and examine the implementation of OPC/OPC-UA infrastructure as existing solutions require a manual configuration. The different paradigms of OPC and OPC-UA require a consistent mapping into one programming language and application environment to examine the cooperation of OPC and OPC-UA based networks under real conditions.

The scope of this work is to combine the wrapping approach with direct implementation of OPC devices in terms of an object-oriented programming concept. These devices are consequently embedded into an OPC-UA network.

3. Continuous integration of OPC Classic components into an OPC-UA production network

The generic integration of OPC Classic components into an ICT based on the OPC-UA interface standard is performed in several steps that are based on a virtual representation of the components and the network for the embedding of a combined OPC/OPC-UA infrastructure:

1. Virtual representation of general OPC components through object-oriented programming concepts (OPC DA Clients),

2. detailed mapping of OPC Classic networks by embedding of their components into a suitable server environment,

3. configuration of an OPC-UA production network by using the same object-oriented programming approach as well as by creating an interoperable API,

4. mapping of the OPC Data Access data model on OPC-UA by considering requirements of the OPC specification, and

5. distribution of simulated data over OPC-UA networks and OPC Classic based consolidation of consistently annotated information into a database system for further processing.

In Figure 3 the abstract idea of integrating OPC Classic components into an OPC-UA network is visualized. As pointed out, the OPC Classic production environment, which may be present in many industrial applications, is located within the system limits of the OPC-UA production network.

For our scenario, the Java programming language has been identified as suitable object-oriented approach to realize the combination of objects from various architectural approaches. In the first step, the OPC Classic components (OPC DA) are represented by Java classes based on the openSCADA [11] implementation of OPC. This openSCADA library provides basic functionalities of SCADA and delivers a full implementation of the OPC Classic standard. In terms of our scenario, the library is used to simulate an OPC DA Client that receives simulated production information from an OPC server. Data values are generated by simulated sensors and annotated with meta-data to meet format specifications of an OPC DA client.

The mapping of the OPC Classic network that comprises of multiple OPC DA clients and sensors is realized by the creation of a communication context, in which the different components of OPC Classic can interact with each other. This environment is encapsulated and is part of a higher OPC-UA information system context. The connection between the OPC Classic context and the OPC-UA environment is performed by a communication of the OPC DA server with an OPC-UA client that delivers functionalities of OPC Classic.

The configuration of the OPC-UA production network is also performed by the Java language. In terms of OPC-UA, a library of the Prosys OPC-UA implementation [12] is used to map OPC-UA specifications on Java objects. As mentioned, the API of OPC-UA is not standardized for an access by object-oriented programming languages. Thus, in terms of the OPC-UA server representation and control, we developed an

Fig. 3. Concept of OPC integration within an OPC-UA production network

API, which maps the Address Space information model of OPC-UA on Java classes that represent machines and devices of the production environment. This implementation contains all nodes, variables and references that are required to map the production information on suitable Java objects that are designed according to the OPC-UA standard.

The technologies and implementations described have the capability of properly simulating a combined OPC/OPC-UA production network. Details of the simulation of such network are explained in the following section.

4. Simulation of an OPC/OPC-UA production network

In order to create a combined simulation of a production network that contains both, an emulation of traditional OPC components as well as novel OPC-UA servers and clients, an appropriate modeling environment and language syntax has to be selected. In the first place, the modeling environment needs to integrate object oriented approaches as concrete instances or objects of real devices need to be created for the simulation. Secondly, the modeling environment needs to provide an integration of database systems in order to enable a real-time propagation of production data in the course of the simulation. The entire ICT of the simulated OPC/OPC-UA production network is visualized in Figure 4.

The simulation of the production environment is initiated by a local repository that contains historical data of an enterprise. In our case, the data storage contains discrete sensor data of a machine that is used in the manufacturing industry for the production of structural components. The repository contains signals of eight sensors that are directly associated to the machine or to its close environmental conditions.

The historical data consists of thousands of data sets that have been collected over several weeks. Each of these data sets contains sensor data that were recorded for a determined duration. Between each pair of data sets, a fixed waiting period is included. Each data set has a unique identifier (Id) and can be selected separately, whereby each set consists of 30000 values which represent 30 seconds of the process - one value per millisecond. The goal of the simulation is to pick data sets randomly from the local repository and to propagate the values in real-time. This approach guarantees a realistic system behavior as the data sets are executed without a fixed sequence, however the data sets itself contain data in chronological order. The implementation and execution of the Java-based OPC-UA server application is described as follows.

Simulator

Data base System openSCA DA OPC Classic DA Client Fig. 4. Realization of an OPC/OPC-UA production environment simulation

+ SimulatorO + getCurrentData(): Data + startSimulation(): void + stopSimulation(): void

-data 1..* -dataStructure 1."

-dbHandler DbHandler

+ DbHandlerG + loadDataQ: DataStructure

Data -dataSet ^ DataStructure

1..* #

Fig. 5. Simulator for randomized selection and propagation of data sets

The application of the OPC-UA server and the according API has been implemented using the Prosys Java OPC-UA Client SDK. Based on our API, a server simulator class was developed that analyses the structure of the historical data and extracts data from the repository in discrete time steps. The data and the structure of the data sets are configured by Java classes that map production information on data streams that are conform to OPC-UA specifications. The implementation of the server application is depicted in Figure 5.

The simulator concept contains the DbHandler class that utilizes a method to randomly pick data sets from the local repository. The data mapping is performed by a Data class whereas the DataStructure class determines the according data format of the data sets that is transferred to the simulator. The simulator creates the Address Space Model of the OPC-UA server and loads the values of the current data set into the cache of the server. The single data points of the actual data set are then propagated into the simulated production system.

The data that is propagated into the production network can be accessed by any OPC-UA client located within this network. In our case, the Toolbox TOP Server [13] is used to emulate a fully functional OPC-UA client that is able to communicate via web services in a Microsoft Windows architecture or between two computers in the same network. The OPC-UA client is able to browse for OPC-UA servers that are located within the same network domain and is accordingly able to find the running emulated OPC-UA server.

Besides OPC-UA clients, TOP Server also support functions of OPC Classic DA servers. Thus, through the server implementation, it is possible to utilize the native downward compatibilities of OPC-UA. The OPC-UA data stream from web services and initiated by an OPC-UA server environment is accordingly transferred into an OPC Classic based signal.

In order to browse for the OPC DA server that is emulated by the TOP Server, the openSCADA based OPC Classic DA client application is used. By using the Utgard module [14] of the openSCADA framework, a fully functional OPC DA client was implemented that can browse for OPC DA servers and nodes within an OPC Classic production network. The data stream between the emulated OPC DA server and the OPC DA client is exchanged via COM/DCOM technologies as the communication is based on pure OPC Classic.

As highlighted before, the communication in OPC Classic networks is difficult to establish, as the firewall setup in terms of DCOM is a very complicated task. Although we accomplished to browse for OPC DA servers, which are on a different computer in the same network, via DCOM, we conclude that a configuration of the OPC DA client on the same machine as the TOP Server OPC DA server is easier to perform.

After setting up the simulated production environment in the described way, the combined OPC/OPC-UA production simulator is started based on data in the local repository. The data stream is propagated as OPC-UA signal and transformed through the TOP Server. The data is then accessed by the OPC DA client. The client implementation that is also carried out in pure Java uses an object-relational database mapping library that enables a consolidation of incoming data stream into multi-dimensional database systems.

5. Validation and evaluation of the integration process

The data consolidation enables an annotation of production data with meta-information. This information can e.g. contain sensor or component types, the installation place of sensors or additional information about the data, which is useful for later analysis steps. In order to validate the data integration process based on the implemented OPC DA client, we conducted data performance measurements concerning the data integration process as well as analysis and data mining operations on the consolidated production data. The data analysis itself is not in the scope of the current paper, however we have proven that a target-oriented analysis of the simulation production data is possible by using the OPC-UA based data generation.

The performance measurements aim at a validation of the real-time capabilities of the solution that have been carried out. In terms of the performance penetration tests, the consolidation step of the annotated information has been determined as the major bottleneck of the integration chain. As the streaming of the production data can be executed on a high-frequent basis, the critical point of the integration consists in the annotation of production data together with the subsequent consolidation into the multi-dimensional database system.

The according tests have been carried out using different data stream frequencies to determine the limits of the integration process. As the data sets contain 1000 values for each second, frequencies of the simulation can be adjusted to an extent that is able to challenge the real-time capabilities of the data integration chain. Hence, the frequency rate of the data propagation was adapted gradually until the limits of the data base integration of the OPC DA client were determined. The results have shown that data stream processing is smoothly possible to an integration frequency of 0.5 seconds. This result implies that about 500 values (for 500 ms) have to be buffered prior to each data integration operation. Further detailed performance measurement tests have been performed in order to determine the real-time capabilities in terms of an integration of multiple values and data streams. The results of these tests are beyond the scope of this architectural paper. In terms of an overall evaluation of these test scenarios, it can be concluded that the system architecture carried out in terms of this work is principally capable of real-time data integration.

These satisfactory results are due to the fact that we carried out an integration chain that is based on a consistent usage of tools and technologies. The implementation of the simulation environments and the APIs for the utilized application ensures the integrity and interaction of the used solutions. Thus, the entire production environment turns out to be interoperable, although entirely different underlying technologies are in use.

6. Conclusion and Outlook

The aim of this work is to demonstrate generic interoperability of OPC-UA production networks by showing a simulated combined OPC/OPC-UA environment for the usage in real-time fashion. An integrated tool chain was developed based on pure Java applications that is capable of integrating data from OPC-UA sensors or components. The data that is integrated using an OPC Classic DA client contains all necessary production and meta-information to perform extensive data analysis in terms of the production optimization.

In the next steps, the focus will be on further integration of OPC components in OPC-UA environments. Extensible web services will be enabled that are enriched by domain or vendor specific meta-information. The domain specific knowledge that is integrated by expanding the communication stack of OPC-UA is realized by integrating ontologies into intelligent obj ects or smart devices in the production environment.

Together with full interoperability of OPC components with OPC-UA described in this work the capabilities of future smart OPC-UA devices can be fully exploited and implemented into older manufacturing environments. The described interoperability is a major step towards the realization of Industry 4.0 in existing production sites by creating the basis for the realization of Cyber Physical Production Systems.

Acknowledgements

This work is support by the German Research Foundation in the Excellence Cluster "Integrative Production Technology for High-Wage Countries" at the RWTH Aachen University.

References

[1] International Electrotechnical Commission. IEC 62264 - Enterprise-control system integration.

[2] Hoffmann M, Meisen T, Schilberg D, Jeschke S. Multi-dimensional production planning using a vertical data integration approach: A contribution to modular factory design. 10th Int. Conf. & Expo on Emerging Technologies for a Smarter World (CEWIT). Melville, NY, USA; 2013.

[3] Brecher C, Müller S, Breitbach T, Lohse W. Viable System Model for Manufacturing Execution Systems. 46th CIRP CMS 2013 2013:461-6.

[4] Brown N, Kindel C. Distributed Component Object Model Protocol -DCOM/1.0: Microsoft Corp; 1998.

[5] Lange J, Iwanitz F, Burke TJ. OPC: From Data Access to Unified Architecture. 4th ed. Berlin, Offenbach: VDE-Verl; 2010.

[6] Mahnke W, Damm M, Leitner S. OPC unified architecture. Berlin [u.a.]: Springer; 2009.

[7] Enste U, Mahnke W. OPC Unified Architecture: Die nächste Stufe der Interoperabilität. at - Automatisierungstechnik 2011 ;59:397-404.

[8] Leitner S, Mahnke W. OPC UA - Service-oriented Architecture for Industrial Applications. ABB Corporate Research Center 2006.

[9] Hensel R. Industrie-4.0-Konzepte rütteln an der Automatisierungspyramide. [10.04.2013]; Available from: http://www.ingenieur.de /Themen/Produktion/Industrie-4.0-revolutioniert-Produktion.

[10] Hannelius T, Salmenpera M, Kuikka S. Roadmap to adopting OPC UA. 6th IEEE Int. Conference on Industrial Informatics 2008:756-61.

[11] openSCADA. open source Supervisory Control And Data Acquisition System; Available from: http://openscada.org/.

[12] Prosys. Prosys OCP UA Java SDK; Available from: http://www.prosysopc.com/products/opc-ua-java-sdk/.

[13] Toolbox. TOP Server OPC & I/O Server; Available from: http://www.toolboxopc.com/index.html.

[14] openSCADA. Utgard - vendor-independent, 100% pure JAVA OPC Client API; Available from: http://openscada.org/projects/utgard/.