Digital factory system for dynamic manufacturing network supporting networked collaborative product development

Abstract

During the last years, important research investments have been made by Airbus Group Innovations for the establishment of sustainable Product and Process data interoperability based on open standards. Driven successively by concurrent engineering, collaborative product design in the virtual enterprise or digital behavorial aircraft, it was capitalize through the establishment of a federative interoperability framework. Driven by factory of the future-related research, the dynamic manufacturing network (DMN) concept enriched the framework, which aims at providing agile infrastructure for networked collaborative product development. For such networks, protocols based on open eBusiness product lifecycle management (PLM) standards for exchange and sharing of product and process data between the implied organizations, their processes and the technical enteprise applications supporting these processes are needed This paper presents a new way of combining model-based enterprise platform engineering, model-driven architecture, and system engineering in order to adress the establishment of a sustainable interoperability within DMN. Based on relevant litterature interoperability issues, this paper describes the new approach, which relies on the association of effective existing technologies coupled with research results on the fields of model-driven engineering (MDE), enterprise interoperability, system engineering and PLM. The new approach relies on the concept of digital factory system (DFS) coupled with DMN in order to produce sustainable and agile collaborative infrastructure for manufacturing digital ebusiness ecosystem. The approach is then illustrated through use case coming from the IMAGINE project and an outline is provided on how it will be used and developed further for the assessment of PLM standards and their implementation in the Standard Interoperability PLM project at IRT-Systemx.

Introduction

During the last years, important research investments have been made by Airbus Group Innovations for the establishment of sustainable product and process data interoperability based on open standards. Driven successively by concurrent engineering, collaborative product design in the virtual enterprise or digital behavorial aircraft, it was capitalized through the establishment of a federative interoperability framework [1]. This approach [1] consists first in comparing different interoperability frameworks and characterize the ideal collaborative system according these different frameworks [2]. It also defines an approach for achieving efficient eBusiness collaboration between enterprises, relying on ATHENA [3] interoperability framework [4], [5] but extending [6] it on the basis of analysis of interoperability brakes leading to non-interoperability – which are to be addressed – and providing a set of interoperability enablers, which are to be considered [7]. In this way, the importance of open source coupled with open standards and also their strategic nature are highligthed [1]. Building collaboration based on open standards is conditionned by the availability of commodities on the web, i.e., industrial quality software solutions that are open source and that are implementing open standards [8]. Such availibility is a key criteria when assessing maturity and usability of a standard for establishment of sustainable interoperability required for PLM. Governance is required within the enteprise, through controlled urbanism of the information system based on visual enteprise modeling. Governance is required within digital ebusiness ecosystems (e.g., eHealth in Australia governed through NEHTA interoperability framework [9]). Existence of such governance, with associated interoperability framework of reference, is a key criteria when assessing the interoperability maturity level [10] of a community. When sustainable interoperability is to be established on the long term (e.g., 30 to 50 years of an Aircraft program), Figay [1] also proposes a strategic approach, which consists of alterning research and operational projects in order to allow mastered evolution of the interoperability framework of a digital ebusiness community (e.g., Aeronautic, Space and Defense in Europe, through ASD Strategic Standardization Group [11]), establishing links with the other communities contributing to this interoperability (e.g., ISO TC184 SC4, European research working on Enteprise Interoperability [12], Object Management Group Mantis [13], OASIS PLCS TC TC [14], OASIS WSRP TC [15], Boost Aerospace private cloud of the European Aerospace and Defense industry [16], etc). Silos between operations and research were identified as a source for non-interoperability, as well as disruptive research without consideration on the innovation path allowing sustainable interoperability. The different successive research and operational projects which feed the federative interoperability framework are described in the introduction of Figay [1], the last update being provided by Figay et al. [17], [18].

This paper present the concept of digital factory system (DSF) for DMN supporting networked collaborative product development (NCPD). It results from research performed in the research context previously described. It aims at extending the federative interoperability framework proposed by Figay [1]. It relies on the concept of extended hypermodel for interoperability defined by Figay [1], [7], which aims at mastering data lost coming from translation of information models between the different languages used in a model-driven approach relying on model-driven architecture in order to preserve semantic. It is based on assessment of the results of the IMAGINE project [19], for which the industrial context is provided in the next section. It will be developed further and assessed through the Standard Interoperabilty PLM (SIP) project [20], as described within persperctives of the conclusion of this paper. The brake identified for IMAGINE is missing methodology for addressing interoperability within a DMN between thousands of applications distributed in different organizations and information systems. The brake addressed by SIP is missing test based approach for developing and accessing standards and their implementations within a DMN context. While IMAGINE and SIP aim at adressing these two interoperability brakes, which are to be added to those defined by Figay [1], problem adressed by this paper, related to the combination of model-driven engineering (MDE), enterprise interoperability, system engineering and PLM, will be more precisely defined in the conclusion of the state of the art. Our proposal will rely on the methodology defined by Figay [1], completed with technical solutions assessment within the field of domain-specific languages and model-driven architecture. Then digital factory system for DMN concepts will be described, and its applicationt within the IMAGINE context will be analyzed.

It is to be noticed that addressing PLM interoperability within a DMN implies combining numerous research and expertise fields. It cannot rely on a single simplistic model. It implies combining numerous models used by the numerous actors, which contribute to interoperability and cause non-interoperability. We consider a DMN as a complex system of systems, considering alternatively the whole system (holistic approach) and some sub-systems, according different viewpoints, aspects or fields. For each contribution we provide, current industrial contexts that reflect the state of the practice and research contexts that define the state of the art and vision of the future are to be considered. Each is evolving quicker and quicker, challenging us for producing evolutionary approach that is adapted to the continuously evolving environment. This approach is formalized by Figay [1]. Because our proposal cannot be understood without a minimum knowledge of the approach and of the complexity of the addressed fields, we structured this research paper a way which can be considered unusual and difficult to understand without a minimum knowledge of the business context and of the research context. The state of the art to be realized is not related to a single domain, but to multiple domains our approach can benefit not individually, but through their combination for achieving needs of mastering the complexity of PLM interoperability in a DMN.

The main aim of this study was to present the latest in this field not as a collection of independent articles, but as a whole, in an integrated manner. We provide the relevant literature review of methods and approaches developed to understand various issues raised by the problem of interoperability. Our intention is to cover the full spectrum of the interoperability issues and providing a set of interoperability enablers which are to be considered.

Within the current economic manufacturing context, important research investments are made on the factory of the future. The DMN concept is emerging. It is defined, according to Viswanadham [21] and Papakostas et al. [22], as a coalition, either permanent or temporal, comprising production systems of geographically dispersed small and medium enterprises and/or original equipment manufacturers that collaborate in a shared value-chain to conduct joint manufacturing. The IMAGINE research project responds to great interest for creating such networks and take advantage of them both for NCPD and for supply chain optimization, as illustrated by [23].

Airbus Group Innovation (AGI) contributing to this project was motivated by importance of integrating processes and solutions related to DMN with PLM and system engineering. Indeed, within the aeronautic, space and defense domain, such collaborative development cannot be envisaged without integration of these processes and of the underlying associated standards. Some of the relevant open standards to be considered and combined were identified for building the federated interoperability framework of the aeronautic, space and defense community. Shared best practices in system engineering are reflected by the System Engineering Process Framework ISO standard (ISO 15288) [24], produced by the system engineering community. It provides the definition of system engineering we are considering, associated to a common framework for describing the life cycle of systems created by humans, which consists of a set of processes and associated terminology. These processes are applicable at any level in the hierarchy of a system's structure and throughout the life cycle for managing and performing the stages of a system's life cycle. PLM considered definition is the one provided by CIMDATA [25], “a strategic business approach that applies a consistent set of business solutions that support the collaborative creation, management, dissemination, and use of product definition information within the extended enterprise (customers, design and supply partners, etc.), spanning from concept to end of life of a product or plant and integrating people, processes, business systems, and information.”

In order to establish NCPD, the strategic importance of manufacturing data standards was identified by the ASD Strategic Standardization Group [26]. A particular focus is put on exchange, sharing and long-term retention of product and process data for collaborative design and for integrated logistic chains. Before IMAGINE, production process was not in the scope of SSG, and one AGI motivation for participation in IMAGINE was to complete SSG scope with standards related to production processes (e.g., ISA95 [27]) and supply chain processes (e.g., SCOR [28]), and to study how they can be combined with PLM and System Engineering standards for obtaining through life cycle interoperability.

AGI's motivation also came from the fact that it was identified that manufacturing standards for product and process data exchange, sharing and long-term archiving are not sufficient when willing ensuring PLM interoperability within a DMN. They do not standardized processes and software. However, the expected interoperability concerns PLM solutions (processes, application and software products) and requires clearly established links with business motivations and industrial program managements. Such interoperability is also to be ensured between the different implied specific domain, in particular those concerning disciplines involved in manufactured product engineering, such as mechanical design, simulation, product planning, production control, etc.

Finally, it was identified that establishing the needed DMN interoperability is complex and challenging due to the large number of involved domains, and to related domain-specific languages (DSL) used by the manufacturing standards. In addition, specifying and implementing the usage of these standards in order to achieve sustainable interoperability within a DMN remain a challenge, even if relying on model-driven engineering for the PLM hub.

IMAGINE project provided the opportunity for extending the federated interoperability framework and associated NCPD platform, also called the cPlatform. In addition, DSL and MDA [32] technical solutions were assessed during the project, in order to identify appropriate set of commodities, i.e., open source solutions of industrial quality based on open standards. Such commodities should extend the cPlatform and associated processes for better dynamicity and adaptation within a DMN context.

A DSL is a language tailored to a specific application domain. Different kinds (domain-specific markup, modeling or programming languages) of DSLs exist. DSLs have become popular with the rise of domain-specific modeling (DSM). DSM is an engineering methodology for systems development, being computer softwares systems, with model-driven architecture (MDA), or manufactured systems with computer-aided design (CAD), computer-aided manufacturing (CAM) or model-based system engineering (MBSE). DSM languages provide higher level of abtraction than general purpose modeling languages. They consequently require less effort and less details when specifying a given system. They also support code generation from models created with the DSM languages. The generated code is for execution by automates, being computers or numerical control machine tools. Such generation reduced the number of defects when comparing with manual coding, and is also quicker.

Today, DSM languages are not anymore built-in specific code generators products, as it was the case in the 1980s. They can be defined using numerous technologies providing meta-meta languages, such as entity relationship diagrams, formal languages, Extended Backus–Naur Form (EBNF [29]), ontology languages, XML schema and meta-object facility (MOF [30]). The strengths of these languages tend to be in the familiarity and standardization of the original language. MOF and associated model-driven architecture (MDA [31]) technologies are standards of object management group (OMG). MDA is based in particular on the four-layer MOF-based metadata architecture, which is a key enabler for model transformation, model interchane and model-driven engineering. As illustrated in Fig. 1, data of the real world (M0-reality) are modeled (M1-layer model) using modeling language (M2-layer metamodel or modeling language). The M3 layer, meta-metamodel, is about languages for describing languages, such as MOF. MOF is a modeling language that allows to model constructs of M2 and M1 languages and their relationships.

Several commercial and open source products are today supporting the standards related to model-driven engineering. It demonstrates maturity of these standards, which makes them suitable for preparing and building interoperability (cf. [1]). Our assessment focuses mainly on open source solutions implementing open standards, as Figay [1] identified open source and open standards as interoperability enablers. Open source and free MDA environments are available today, supported by the eclipse modeling tools (EMT at [33]) community. The EMT community delivers sustainable components allowing the creation of visual modeling platform. It includes eclipse modeling framework (EMF), model-to-model transformation components based as ATLAS transformation language (cf. ATL tranformation-based model management framework defined by Bezivin et al. [62]) or OMG's query view transformation (QVT [34]), model to text components such as Acceleo (at [33]), components supporting specficiation and validation of contraints based on OMG's object constraint language (OCL) or unified modeling language (UML [35])-based modeler (e.g., Papyrus at [33]). Visual modeling is supported, or by defining a DSL by means of a UML profiles (e.g., SySML), or by defining a DSL by means of EMF (at [33]) and then taking advantage of a new EMT component, Sirius, which support definition of a visual modeling component coupled with EMF-based DSL model editor. Very recently, OBEO company (http://www.obeo.fr) provided as open source solution a platform, OBEO Designer, integrating all the [33] components in order to obtain a solution for creation of visual modeling platform based on DSLs. The EMT community is supported by industrial companies aiming at using model-based system engineering supporting the EMT community. It includes CEA (Commissariat à l'Energie Atomique et aux énergies with Papyrus), Airbus (with Polarsys) or THALES (with Capella).

Other open source projects, out of the EMT community, provide platforms that allow, from UML profiled models, generation of code, compilation, packaging and artifacts deployment on enterprise application platform. One of the most interesting is AndroMDA [36], which target several technological platform, in particular Java for Enterprise (J2EE) servers with encompass Enterprise Java Beans and Portlets components based on Java Community open specifications.

Considering PLM standards and associated modeling technologies, respectively, ISO 10303 EXPRESS and W3C's Ontology Web Language (OWL), OMG communities provided DSLs based on MOF, respectively, the Reference Metamodel for the EXPRESS Information Modeling Language ([37]) and the Ontology Definition Metamodel (ODM [38]).

Concerning enterprise modeling and controlled urbanism of enterprise modeling systems, numerous frameworks and reference models have been produced the last years. One of the most accurate new languages, aligned with controlled urbanism, is Open Group's ArchiMate language [39]. ArchiMate is an open and independent enterprise architecture modeling language, which supports the description, analysis and visualization of architecture within and across business domains in an unambiguous way. With the last version came the Archimate model exchange file format. ArchiMate is supported by numerous tools, and in particular the Archi free ArchiMate Modeling tool [40], generated on top of EMT.

Concerning business modeling, a number of languages have been produced by OMG, such as Business Process Modeling Notation (BPMN [41]), Semantic of Business Vocabulary and Rules (SVBR [42]) or Case Management Model and Notation (CMMN [43]). They are made available as UML profiles or as DSL based on MOF.

Concerning applications on the shelves, some families of components are emerging, relying on open specifications. It demonstrates the maturity of the market and of the solutions for these families. It is the case for enterprise portals, relying on specifications such as OASIS' Web Service Remote Portlets (WSRP [44]) or Java Community's JSR 168 [45] and JSR 286 [46]. It is the case of enterprise service bus, relying on W3C standards and Java Business Integration (JBI [47]) specification. It is the case for enterprise workflow systems, relying on Workflow Management Coalition [48] standards. It is the case for enterprise application servers, relying on the OMG's Corba Component Model (CCM [49]) and derived Enterprise Java Beans specification (EJB [50].)

Concerning information and communication technologies, solutions are emerging through the effort of the Cloud community, allowing dealing with virtualization of enterprise networks (e.g., ProxMox [51]) and virtualization of machines (e.g., VirtualBox [52]). Important standardization efforts are undertaken, as illustrated by the NIST Cloud Computing Standard roadmap [53] and by the OpenStack initiative [54].

The assessment described in this section was performed according the approach defined by Figay [1]. This approach allowed defining identification of technical solutions for building a platform for demonstration and experimentation of open standard-based PLM interoperability in a DMN. Today, the availability of meta-models for languages used for PLM data/enterprise modeling standards and of MDA platforms creates the condition for assessment of these standards for the model-based generation of interoperable DMN collaborative platforms.

The IMAGINE project offered the opportunity for assessing association of these technologies and standards in order to deliver platforms enabling manufacturing enterprise applications interoperability: collaborative manufacturing PLM hubs, relying on open standards. ArchiMate and Archi were used as a means of aggregating motivation, business, applicative and ICT viewpoints a simple way for supporting discussion between product architects, business architects, information system architects and ICT architects within the project. Such aggregation has been studied at different scales, being for digital business ecosystems, DMN related to a single manufactured product, enterprise information system governance or application development project. A generic collaborative platform (cPlatfom) was defined for DMN support as an open standard-based aggregation of enterprise portal, enterprise service bus and enterprise application server according the model illustrated by Fig. 2.

A system engineering approach was adopted. A modeling and a development environment complete the cPlatform as the operating platform. Fig. 3 shows an Archi applicative view of the designing, development and usage environments, which are the “supporting” systems (as defined in [24]) of the system of interest: the DMN.

These models were structured according the interoperability framework defined by [1]. The realization of the DMN platform followed principles defined by [1] for preparing and building an effective and sustainable operational interoperability at an acceptable price. It is first relying on enterprise modeling and MDA as interoperability enablers. It is second relying on assessed commodities on the Web. Archi was elected as open source visual modeler of an open enterprise modeling standard. Archi is developed on top of the Eclipse platform. Its development is based on the usage of EMT open source technologies, which implement MDA. Similarly, we identified corresponding commodities on the web for each component of the DMN applicative model, a common commodity on the web is defined by Figay [1] as an open source realization of the applicative components of industrial quality and implementing the elected open standards. The existence of commodities on the web for a standard reflects the standard maturity level. For the enterprise ESB, used commodity on the web was OpenESB. For the enterprise portal, it was Liferay. For the workflow system, it was Shark and JaWe. Some components have been specified and developed in order to ensure the aggregation of all these components. Fig. 4 Illustrates how the DMN environment was realized in IMAGINE.

In terms of methodology, we extended and adapted the DMN methodology defined by IMAGINE partners, addressing the production process only, to the design process within a PLM and system engineering context. The “DMN blueprint” concept, introduced by IMAGINE partners, was extended by AGI for supporting the aerospace and defense living labs business cases. DMN blueprints are models of the DMN cross-organizational collaborative enterprise processes, of their participants and of their applicative and ICT capabilities. These models describe as well the AS IS situation, with current baseline of collaborative capabilities, than the TO BE situation, with platforms targeted according a strategic roadmap. All these blueprints are represented as ArchiMate views of an enterprise model, structured according ArchiMate viewpoints. ArchiMate viewpoints are built according practices defined in [55] for a set of stakeholders having well defined concerns (cf. standardized viewpoints in [39]). These views can be aggregated by means of enterprise architecture modeling and managed relying on the DMN methodology, which extends the ArchiMate viewpoints with some related to DMN methodology adapted and extended by AGI. The DMN methodology combines controlled urbanization of the information system, enterprise application integration, PLM and system engineering. System engineering and PLM are completing traditional urbanization and enterprise modeling approaches the following way. From a model of the manufactured product breakdown, based on configuration item, cross-organizational collaborative workflow process models and associated participants are derived from workflow templates, and deployed on the cPlatform, which combines enterprise portal, enterprise service bus and enterprise application server. When enacted and each time an instance of collaboration is launched, it is possible to check if expected properties of workflow participants related to interoperability are aligned with characteristics of actual participants.

From the operational enterprise collaboration platform, it is also possible to derive manually the model of the “AS IS” DMN, as the platform allows capturing and managing organizations, business processes and applications involved in the collaboration.

The problem addressed in this paper come from the fact that currently, deriving operational platform from DMN models and deriving DMN models from operation platform is not automated, making it too expensive and leading to insufficient quality. An efficient MDE tool chain is to be put in place. It should support generation of consistent models, code and then execution artifacts, which will be deployed at the appropriate place. The different generated information sets being static data or piece of programs will have to be formalized using precise ICT standards. Resulting artifacts will have to be produced and consumed by the different components of the whole DMN platform, i.e., specification, design, development and execution components. In the reverse, it should be possible to generate the models describing the AS IS platform from the operational platform itself. Consequently, MDE transformation will have to be defined appropriately, relying on MDE commodities such as EMT or AndroMDA.

Without such MDE tool chain and associated methodology, it is not possible to combine an efficient way model-driven engineering, enterprise modeling, PLM and system engineering in order to prepare and construct sustainable operational interoperability within a DMN at an acceptable price. The following state of the art investigates section points out from the literature the most advanced research activities addressing a similar need and shows why a new approach is required.