The Concept of a Real-Time Enterprise in Manufacturing
form The Concept of a Real-Time Enterprise in Manufacturing
The subject of the dissertation of Mr. Daniel Metz at the Business & Information Systems Engineering (BISE) Institute at the University of Siegen is the analysis sur- rounding the concept of real-time enterprise (RTE), and supporting technologies in the last decade, with the main intention to identify shortcomings. Subsequently, Mr. Metz has developed a reference architecture that overcomes temporal and semantic vertical integration gaps, which is crucial to realize the concept of an RTE, across different enterprise levels by exploiting the paradigm of event-driven architecture (EDA) and complex event processing (CEP). The developed reference architecture has been implemented and validated in a foundry, which has typical characteristics of small- and medium-sized enterprises (see the following foreword from industry).
The reference architecture integrates the real-time process data from different re- sources located on the shop floor with the corresponding (offline) transactional data from the enterprise applications, like an enterprise resource planning (ERP) system. The integrated data is utilized in manifold ways–computation of key performance indicators (KPIs), visualization to enterprise members, creation of logistic informa- tion (i.e., tracking and tracing of different enterprise entities), and analysis of the integrated data/event streams with the support of a CEP engine. Especially, CEP is employed to identify any deviations between the planned data and actual values. In addition to these classical reactive/feedback approaches, the reference architecture also supports proactive approaches to enhance the performance of an enterprise.
Further, the reference architecture has been implemented and validated in a state- of-the-art aluminum sand casting enterprise. The implementation utilizes modern methods and techniques of software engineering (e.g., .NET framework, EsperTech CEP engine). Additionally, the implementation aligns with numerous national and international standards and models (e.g., IEC 62264, MESA Model).
The book may contribute to further understanding and development of the con- cept and realization of RTE and CEP.
1 Introduction
In this chapter, the problem area of the presented research work is motivated, the deduced research goals are defined, the basic positions concerning philosophy of sciences are provided, and the structure of the research work is described.
1.1 Motivation
1.2 ResearchGoals
1.3 PositioningconcerningthePhilosophyofSciences
1.4 StructureoftheWork
2 Problem Description and Fundamentals
In the following chapter, the RTE is introduced as an organizational concept for manufacturing enterprises. The presentation of the RTE starts with a broader dis- cussion of RTE’s motivations, fundamentals and principles. Further, the emphasis of presented research and status of development in the realm of RTE are presented. Here, the realization of the RTE in manufacturing can be identified as requiring further attention in ISR.
The event-based processing of information, vertical integration of a manufactur- ing enterprise and real-time alignment of planned and actual manufacturing process execution are recognized as primary requirements for the realization of an RTE. MES is introduced as a contribution of the engineering community towards the im- plementation of RTE in manufacturing. In addition, CEP is unveiled as an IT build- ing block for the RTE. However, the liaison between MES and CEP requires further attention in research, which is elaborated in this research work.
2.1 Real-Time Enterprise – Organizational Concept for Manufacturing Enterprises
2.1.1 Vision,Principles,andDefinition
2.1.2 ResearchScopes,Gaps,andChallenges
2.1.3 EmergentEnablers
2.1.4 SummaryandResearchGoals
2.2 PrinciplesandFundamentalsofEventProcessing
2.2.1 Events and their Processing as a Fundamental Requirement forBusiness
2.2.2 DefinitionsandCircumscriptionsofEvents
2.2.3 Event Clouds, Event Streams, and Complex Events
2.2.4 Historical Background pertaining to Processing of Events
2.2.5 Event-DrivenArchitecture
2.2.6 ComplexEventProcessing
3 Requirements Analysis
Fundamentals and principles of the RTE and its application to manufacturing enter- prises were introduced in Chap. 2. The main building blocks of an RTE (i.e., (verti- cal) integration, (process) automation, and individualization) have been discussed in general. The RTE has been motivated as an appropriate strategy to address business pressures (cf. Fig. 2.2). As a result, (manufacturing) enterprises strive for achieving celerity, agility, and information availability. As outlined in the introduction of this research work (cf. Fig. 1.1), the transmission of the RTE vision to manufacturing enterprises is affected by managerial, engineering, and information technological issues. Henceforth, problem perspectives, methodologies, and solution approaches of the management, engineering, and computer science community are elaborated in more depth. This presentation reveals the interdisciplinary character of the pre- sented research work that is typical for many problems and methodologies presented by the ISR community. In addition, the application area for implementing/testing the presented research work is described. Thereby, from the perspective of philoso- phy of sciences, the slice of reality/universe of discourse is disclosed.
3.1 Description of Application Area
3.1.1 Manufacturing – Economic Relevance and Classification
3.1.2 Small and Medium Sized Enterprises – Structure, Traits, andChallenges
3.2 Management Perspective
3.3 Engineering Perspective
3.3.1 PrinciplesofFeedbackControl
3.3.2 Multiple Control Loops across a Manufacturing Enterprise
3.4 Computer Science Perspective
3.5 Discussion of Perspectives
4 Event-Driven Framework for Real-Time Enterprise
The previous discussion on the realization of RTE for manufacturing enterprises– also considering managerial, engineering and computer science perspectives–reveals the urgent need for (i) vertical integration of various enterprise levels (i.e., from shop floor to top floor); and (ii) realization of multiple control loops between these enterprise levels. Consequently, the aim of the presented research work is the devel- opment of an IT-framework for real-time monitoring and control of manufacturing processes. Thereby, previous achievements that have been made by management, engineering and computer science communities are incorporated.
4.1 Literature Review
4.1.1 Architectural Styles for Manufacturing Control
4.1.2 Research on Intelligent and Decentralized Control Approaches
4.1.3 Service-Orientation for Manufacturing
4.1.4 Convergence of Control and Physical Systems
4.1.5 Event Processing in Manufacturing
4.1.6 Summary
4.2 Process Model toward the Realization of RTE
4.2.1 Analysis and (Re-) Design of Value Creation Processes
4.2.2 EnterpriseDataModelandDataFlowDiagrams
4.2.3 Knowledge Identification
4.2.4 Knowledge-BasedMonitoring and Control of Manufacturing Processes
4.3 Event-Driven Framework for Real-Time Monitoring and Control
4.3.1 Outline of Architectural Components
4.3.2 Real-Time Acquisition of Process Data from Manufacturing Resources
4.3.3 Aggregation of Process Data for Forward and Backward Traceability
4.3.4 Online Tracking of Enterprise Process Entities
4.3.5 Detection of (Critical) Process Situations using CEP
5 Implementation of Solution Approach and Evaluation in a Foundry
The aforementioned IT framework has been developed in cooperation with Ohm & Ha ̈ner Metallwerk GmbH & Co. KG, Olpe, Germany. The enterprise is family- owned, employs more than 400 workers and realizes a turnover of more than 50 mil- lion Euros. The working foundry manufactures castings with a total weight starting at 20 grams and goes up to 2000 kilograms (cf. [196]).
State-of-the-art machinery is employed in the new plant1 in Drolshagen, Ger- many, to produce components in small and medium lot sizes. The company serves more than 400 customers per year, thus has to cope with a high product variety. The layout of the new plant primarily follows product layout, but partially, also cellular and process layouts are employed. The production encompasses both continuous and discontinuous flow of material. The outcome of the discrete manufacturing pro- cesses are either bulk goods (e.g., molten metal) or piece goods (e.g., castings, sand cores).
In summary, the manufacturing processes of the considered foundry can be char- acterized as depicted in Fig. 3.2. Further, according to the characteristics mentioned by Mo ̈nch, the established sand casting processes in Drolshagen are complex man- ufacturing processes (cf. [189] and Sect. 4.2.2). Following, fundamentals of metal casting in general and sand casting in particular are elaborated
5.1 Sand Casting
5.1.1 Principle of Sand Casting
5.1.2 Sand Casting Process
5.2 Analysis of Business and Manufacturing Processes
5.3 Design of an Enterprise Data Model
5.4 Implementation of the Framework
5.4.1 Process Coverage
5.4.2 Tracking of Process Entities
5.4.3 Process Analysis using Historical Process Data
5.4.4 Real-Time Controlof Sand Casting Processes
5.5 Evaluation of Manufacturing Process Improvements
5.5.1 Non-Measurable Improvements
5.5.2 Measurable Improvements
5.5.3 Comparison with Requirements
6 Summary, Conclusions and FutureWork
The necessity to (re-) act on internal and external events in (near) real-time has been formulated as part of the RTE vision. This vision requires an enterprise with fully integrated, automated and individualized value creation processes. With respect to manufacturing enterprises, the following problem areas could be identified in this research work: (i) standardization of shop floor interfaces to establish a vertically integrated enterprise; (ii) conceptualization of real-time monitoring and control con- cepts based on EDA and CEP; and (iii) their implementation in the realm of MES.
The seamless horizontal and vertical integration is a prerequisite for an immedi- ate transfer of information from its POC to an appropriate POA. Further, the RTE requires measures to select and analyze relevant information, thus avoids an indis- criminant flood of data. Finally, a continual alignment of planned and actual process execution is envisioned by an RTE.
Overall, the realization of the aforementioned RTE in manufacturing requires the consideration of management, computer science, and engineering perspectives. The management community’s view of a manufacturing enterprise and its production management is focused on topmost enterprise levels (i.e., strategic and tactical en- terprise levels). Further, from a management perspective, ERP is considered as a suitable approach for support of value creation processes.
However, a deeper and broader integration of the manufacturing level (i.e., shop floor) has been researched by the engineering community. Their research and de- velopment activities have led to various standards and implementations of MES. Recently, MES has been mentioned as a means to establish an RTE in manufactur- ing. In addition, control engineering has provided principles for the realization of multiple closed-loop controls (i.e., feedback controls). At this point, it is notewor- thy that, for instance, cybernetics has been considered as a theoretical foundation and principle of the RTE. Nevertheless, major issues remain open with respect to the interface to shop floor resources.
The RTE was formulated by Gartner as a vision, thus does not explicitly define with which IT it has to be implemented. Nevertheless, the RTE’s behavior has been described as being event-driven. Therefore, EDA and CEP have been identified as enablers of an RTE. These paradigms enable IT systems, which adhere to the following system requirements of an RTE to be built: (i) agility; (ii) timeliness; and (iii) availability of information. Unfortunately, so far, CEP has been employed only for financial and administrative processes within and across enterprises. Concepts and approaches for real-time monitoring and control of manufacturing processes that capitalize on CEP are rare. Also, the liaison of MES and CEP requires further attention in research.
A framework for the realization of an RTE in manufacturing has been concep- tualized, implemented, and evaluated in this research work. This framework has been designed for manufacturing enterprises, which (i) produce tangible products; (ii) are SMEs; and (iii) are part of the German industry. Thus, the manufacturing en- terprises and their manufacturing processes, which are supported by the developed framework, have been characterized. Thereby, the relevance of the manufacturing sector for the German economy has been discussed.
Before delving into the details of the developed framework, a related work on intelligent monitoring and control approaches, like MAS, HMS, and so forth, has been provided. Some ideas and concepts of these approaches could be transferred to the presented event-driven framework. The framework consists of two parts: (i) a process model that describes the methodology for the introduction of the IT architecture; and (ii) the design and implementation of the IT architecture for the realization of an RTE in manufacturing.
The process model entails (i) analysis and (re-) design of business and manufac- turing processes; (ii) design of an enterprise data model and modeling of data flows between (IT) systems; (iii) identification of control-related knowledge employing, for instance, a KDD process and interviews with domain experts; and (iv) the use of this knowledge for monitoring and control of manufacturing processes.
The last two steps of the process model assume the existence of integrated pro- cess data, event-driven monitoring and control mechanisms. Therefore, an IT ar- chitecture has been developed that capitalizes on EDA and CEP. Process data is acquired from manufacturing resources and integrated with transactional data from enterprise applications. This functionality can be used to realize backward and for- ward traceability. Further, a real-time tracking of enterprise entities, like production orders, products, batches, and so forth, has been developed.
The stream of tracking objects can be interpreted as an event stream. This event stream is analyzed by a CEP engine to detect critical process situations and de- duce appropriate (re-) actions. Noteworthy, the tracking objects contain actual process data as well as planned data from enterprise systems (e.g., ERP system). Therefore, the established control approach is based on an integrated enterprise, which is demanded by an RTE. Also, the remaining RTE principles, i.e., automa- tion of decision-making processes and the immediate availability of information at the POA, are fulfilled by the developed IT architecture.
The process model and the event-driven framework have been implemented for a foundry. The framework is employed to monitor and control highly automated aluminum sand casting processes of the Ohm & Ha ̈ner Metallwerk GmbH, Olpe, Germany, in their new plant in Drolshagen, Germany. The current release of the implemented framework has led to a significant reduction of the average measured takt time by around 18 % and a return on investment in less than one year.
So far, the presented framework is limited to manufacturing enterprises and man- ufacturing processes that have been characterized in Sect. 3.1. Further, the vision of an RTE entails other aspects of an enterprise like personnel, sales, and so forth. Up to now, the framework has not dealt with these aspects. Consequently, the extension of the framework to cover these aspects can be part of future work. In addition, the framework can be extended to implement more tasks/functions of MES.
The extension of the framework to other manufacturing processes of the consid- ered foundry has been envisaged. The mechanical processing (e.g., turning, milling) especially has to be implemented to complete the coverage of the overall value cre- ation process. In addition, control actions have to be implemented that directly influence manufacturing resources. An example is the blocking of a pouring ma- chine, if the wrong alloy is used or the (measured) pouring temperature is expected to not adhere to defined product specifications. Also, the transfer of the control ap- proach to manufacturing resources has been planned. This development fits into the recently promoted research pertaining to Industry 4.0 and CPS.
コメント