EU H2020 MSCA RISE Project FIRST

“virtual Factories:

Interoperation suppoRting buSiness innovation”

Stephan Boese

, Giacomo Cabri

, Norbert Eder

Federica Mandreoli

, Alexander Lazovik

, Massimo Mecella

Keith Phalp

, Paul de Vrieze

, Lai Xu

, and Hongnian Yu

GK software, Germany

{SBoese,neder}@gk-software.com

Università di Modena e Reggio Emilia, Italy

{gcabri,fmandreoli}@unimore.it

University of Groningen, The Netherlands

a.lazovik@rug.nl

SAPIENZA Università di Roma, Italy

mecella@dis.uniroma1.it

Bournemouth University, U.K.

{kphalp,pdevrieze,lxu,hyu}@bournemouth.ac.uk

Abstract. FIRST – “virtual Factories: Interoperation suppoRting buSiness inno-

vation”, is a European H2020 project, founded by the RESEARCH AND

INNOVATION STAFF EXCHANGE (RISE) Work Programme as part of

the Marie Skłodowska-Curie actions. The project concerns with Manufacturing

2.0 and aims at providing the new technology and methodology to describe man-

ufacturing assets; to compose and integrate the existing services into collabora-

tive virtual manufacturing processes; and to deal with evolution of changes. This

Chapter provides an overview of the state of the art for the research topics related

to the project research objectives, and then it presents the progresses the project

achieved up to now towards the implementation of the proposed innovations.

1 Introduction

The manufacturing industry is entering a new era in which new ICT technologies and

collaboration applications are integrated with traditional manufacturing practices and

processes to increase flexibility in manufacturing, mass customization, increase speed,

better quality and to improve productivity.

Virtual factories are key building blocks for Manufacturing 2.0, enabling the creation

of new business ecosystems. In itself, the concept of virtual factories is a major expan-

sion upon virtual enterprises in the context of manufacturing, which only integrates

collaborative business processes from different enterprises to simulate, model and test

different design options, to evaluate performance, thus to save time-to-production [1].

Boese, S., Cabri, G., Eder, N., Mandreoli, F., Lazovik, A., Mecella, M., Phalp, K., de Vrieze, P., Xu, L. and Yu, H.

EU H2020 MSCA RISE Project FIRST - â

AIJvirtual Factories: Interoperation suppoRting buSiness innovationâ

DOI: 10.5220/0008861700030019

In OPPORTUNITIES AND CHALLENGES for European Projects (EPS Portugal 2017/2018 2017), pages 3-19

ISBN: 978-989-758-361-2

Creating virtual factories requires the integration of product design processes, manu-

facturing processes, and general collaborative business processes across factories and

enterprises. An important aspect of this integration is to ensure straightforward com-

patibility between the machines, products, processes, related products and services, as

well as any descriptions of those.

Virtual factory models need to be created before the real factory is implemented to

better explore different design options, evaluate their performance and virtual commis-

sion the automation systems thus saving time-to-production [2].

Fig. 1. FIRST research objectives.

Providing new technology and methodologies to describe manufacturing assets, to

compose and integrate existing services into collaborative virtual manufacturing pro-

cesses, to deal with evolution of change is the main goal of the FIRST project.

FIRST – “virtual Factories: Interoperation suppoRting buSiness innovation”, is a

European H2020 project, founded by the RESEARCH AND INNOVATION STAFF

EXCHANGE (RISE) Work Programme as part of the Marie Skłodowska-Curie actions.

The RISE scheme promotes international and cross-sector collaboration through ex-

changing research and innovation staff, and sharing knowledge and ideas from research

to market (and vice-versa). The project consortium includes five University partners

and two industrial partners from Europe and China.

The main innovations the project aim to introduce are depicted in Fig. 1. The related

Research Objectives (RO) are the following:

RO1. The design of a new semantic manufacturing asset/service/process description

languages and manufacturing asset and (sub-) process discovery methods to ena-

ble on-the-fly service-oriented process verification.

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

RO2. An on-the-fly manufacturing service-oriented process verification method.

Human modellers are not perfect in their modelling, or awareness of physical con-

straints. On-the-fly service-oriented process verification is thus essential to create

valid models and to prevent expensive misconfiguration of machinery.

RO3. The design of an interoperability framework providing support for compati-

bility and evolution to enable global manufacturing collaboration; improving flex-

ibility of existing product design and manufacturing processes; supporting (new)

product-service linkages; and improved management of distributed manufactur-

ing assets.

RO4. Integration for seamless matchmaking of business opportunities, co-creation

of product innovation, and creation of novel business models. Innovation requires

special expertise, domain knowledge of the industry sector, technical knowledge,

business models, finances, and markets.

In the rest of this Chapter, we first provide an overview of the state of the art for the

research topics related to the project research objectives then we present the progresses

the project achieved up to now towards the implementation of the proposed innova-

tions.

2 Overview of Manufacturing Assets and Services

Classification and Ontology

This section describes the state of the art as well as the progress that can be envisaged

beyond the state of the art, in relation to two major research topics related to the FIRST

project.

2.1 Relevant Technologies, Standards and Frameworks of Product Lifecycle

Management

Product lifecycle management is the process of dealing with the creation, modification,

and exchange of product information through engineering design and manufacture, to

service and disposal of manufactured products. In this section, we review the economic

and technical aspects of an interoperation framework for product lifecycle manage-

ment, related standards, technologies, and projects.

STEP (Standard for the Exchange of Product Model Data) ISO 10303. ISO 10303,

also known as STEP (Standard for the Exchange of Product Model Data), is an inter-

national standard for industrial automation systems and integration of product data rep-

resentation and exchange. It is made up of various parts that offer standards for specific

topics. Part 242:2014 refers to the application protocol for managing model-based 3D

engineering (ISO 2014). The standard will be essential to implementing a digital factory

based model.

EU H2020 MSCA RISE Project FIRST - â

AIJvirtual Factories: Interoperation suppoRting buSiness innovationâ

Open Services for Lifecycle Collaboration (OSLC). Open Services for Lifecycle

Collaboration (OSLC) is an open community that creates specifications for the

integration of tools, such as lifecycle management tools, to ensure their data and

workflows are supported in the end-to-end processes. OSLC is based on the W3C

linked data (W3C 2015).

Reference Architecture Model for Industry 4.0. Reference Architecture Model for

Industry 4.0 [3] defines three dimensions of enterprise system design and introduces

the concept of Industry 4.0 components [3]. The RAMI4.0 is essentially focused on the

manufacturing process and production facilities; it tries to focus all essential aspects of

Industry 4.0. The participants (a field device, a machine, a system, or a whole factory)

can be logically classified in the model and relevant Industry 4.0 concepts described

and implemented.

The RAMI4.0 3D model includes hierarchy levels, cycle and value stream, and lay-

ers. The layers represent the various perspectives from the assets up to the business

process, which is most relevant with our existing manufacturing asset/service classifi-

cation.

Currently RAMI4.0 does not provide detailed, strict indication for standards related

to communication or information models. The devices/assets are provided using Elec-

tronic Device Description (EDD) (also see section 3.1) [4], which includes the device

characteristics specification, the business logic and information defining the user inter-

face elements (UID – User Interface Description).

The optional User Interface Plugin (UIP) that defines programmable components

based on the Windows Presentation Foundation specifications, to be used for develop-

ing UI able to effectively interact with the device.

The Functional and Information Layer the Field Device Integration (FDI) [5] spec-

ification as integration technology. The FDI is a new specification that aims at over-

coming incompatibilities among some manufacturing devices specifications. Essen-

tially the FDI specification defines the format and content of the so-called FDI package

as a collection of files providing: the device Electronic Device Description (EDD), the

optional User Interface Plugin (UIP), and possible optional elements useful to config-

ure, deploy and use the device, e.g. manual, protocol specific files, etc.

An FDI package is therefore an effective mean through which a device manufacturer

defines which data, functions and user interface elements are available in/for the device.

Semantics for Product Life-cycle Management (PLM) Repositories. OWL-DL is

one of the sublanguages of OWL

. OWL-DL is the part of OWL Full that fits in the

Description Logic framework and is known to have decidable reasoning. In building

product lifecycle management repositories, OWL-DL is used to extract knowledge

from PLM-CAD (i.e CATIA) into the background ontology automatically, other non-

standard parts, i.e. not from CATIA V5 catalogue, manually into the background on-

tology. OntoDMU is used to import standard parts into concepts of the ontology.

An ontological knowledge base consists of two parts offering different perspectives

on the domain. In Fig. 2, the structural information of a domain is characterized through

https://www.w3.org/TR/owl-guide/

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

its TBox (the terminology). The TBox consists of a set of inclusions between concepts.

The ABox (the assertions) contains knowledge about individuals, e.g. a particular car

of a given occurrence of a standard part in a CAD model. It can state either that a given

named individual (i.e. ‘myCar’) belongs to a given concept (e.g., that myCar is, in fact,

a car) or that two individuals are related by a given property (e.g. that myCar is owned

by me).

Fig. 2. Ontology based on PLM Repositories [7].

Ontology Mediation for Collaboration of PLM with Product Service Systems

(PSS). The PSYMBIOSYS

EU Project addresses collisions of design and manufactur-

ing, product and service, knowledge and sentiments, service-oriented and event-driven

architectures, as well as business and innovations. Each lifecycle phase covers specific

tasks and generates/requires specific information. Ontology mediation is proposed is

proposed as a variant of ontology matching since the level of matching can be rather

complex.

Fig. 3. Ontology Mediation [6].

When matching two different modelling languages, such as Modelica and SysML

in, the issue of completeness makes the mapping task impossible. The two languages

are significant differences and overlaps. Fig. 3 above presents an ontology mediation

approach, which Basic Structure Ontology (BSO) is at the centre, and the mediation

http://www.psymbiosys.eu/

EU H2020 MSCA RISE Project FIRST - â

AIJvirtual Factories: Interoperation suppoRting buSiness innovationâ

among three different tools was working through three matching sets that connected the

common structure ontology which each of the tools: Medelica tool, SysML tool and a

party proprietary tool [6].

Interoperability of Product Lifecycle Management. Integrating among heterogene-

ous software applications distributed over stakeholders in closed-loop PLM. The capa-

bilities of the Internet of Things are being extended to Cyber-Physical-Systems (CPS),

which divide systems into modular and autonomous entities. The systems are able to

communicate, to recognize the environment and to make decisions. Different compa-

nies with different IT-infrastructures adopt different roles in the product lifecycle.

In order to manage the interoperability of heterogeneous systems throughout the

product lifecycle, different approaches could be used [7]:

 Tightly coupled approaches implement federated schema over the systems to be

integrated. A single schema is used to define a combined (federated) data model

for all involved data sources [7. Any change of the individual system’s data mod-

els need to be reflected by a corresponding modification of the entire federated

schema.

 Object-oriented interoperability approaches are closely related to tightly couple

ones. Different types of these approaches are described in (Pitoura, Bukhres, &

Elmagarmid, 1995). Object-oriented interoperability approaches use common

data models which are a similar problem of dealing with modification of the in-

dividual system.

 Loosely coupled interoperability approaches are more suitable to achieving scal-

able architecture, modular complexity, robust design, supporting outsourcing ac-

tivities, and integrating third party components. Using Web services for a com-

munication method among different devises, objectives, or databases is one of

such loosely coupled interoperability approaches. The semantic meaning of a

Web service can be described using OWL (Web Ontology Language). Web ser-

vices described over third party ontologies [9] are called Semantic Web Services.

 Service Oriented Architecture (SOA) has emerged as the main approach for deal-

ing with the challenge of interoperability of systems in heterogeneous environ-

ment [10,11]. SOA offers mechanisms of flexibility and interoperability that al-

low different technologies to be dynamically integrated, independently of the

system's PLM platform in use [12]. Some of standards for PLM using SOA are:

OMG PLM Services [13] and OASIS PLCS PLM Web Services [14].

OMG PLM Services. The current version, PLM Services 2.0 [14], covers a superset

of the STEP PDM Schema entities and exposes them as web services. This specification

resulted from a project undertaken by an industrial consortium under the umbrella of

the ProSTEP iViP Association. Its information model is derived from the latest ISO

10303-214 STEP model (which now includes engineering change management pro-

cess) by an EXPRESS-X mapping specification and an EXPRESS-to-XMI mapping

process. The functional model is derived from the OMG PDM Enablers V1.3. The spec-

ification defines a Platform Specific Model (PSM) applicable to the web services im-

plementation defined by a WSDL specification, with a SOAP binding, and an XML

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

Schema specification. More details on architecting and implementing product infor-

mation sharing service using the OMG PLM Services can be found in [15].

OASIS PLCS PLM Web Services. Product Life Cycle Support (PLCS) is the phrase

used for the STEP standard ISO 10303-239 (Product Life Cycle support (PLCS) Web

services V2) [16]. After the initial STEP standard was issued by ISO, a technical com-

mittee was formed in the OASIS organization to develop this further. A set of PLCS

web services has been developed by a private company (Eurostep) as part of the Euro-

pean Union funded VIVACE project

. Eurostep has put this forward on behalf of

VIVACE to the OASIS PLCS committee for consideration as the basis for an OASIS

PLCS PLM web services standard.

ISA-95/OAGIS SOA in Manufacturing. ISA-95

and OAGi are jointly working on

standards for manufacturing systems integration. They are actively looking into the

suitability of SOA for such integration in manufacturing.

2.2 Manufacturing Assets/Services Classification

Digital Manufacturing Platforms will be fundamental for the development of Industry

4.0 and Connected Smart Factories. They are enabling the provision of services that

support manufacturing in a broad sense by aiming at optimising manufacturing from

different angles: production efficiency and uptime, quality, speed, flexibility, resource-

efficiency, etc. [17]. The available services collect, store, process and deliver data that

either describe the manufactured products or are related to the manufacturing processes

and assets that make manufacturing happen.

As pointed out in [17], pre-requisites for digital platforms to thrive in a manufactur-

ing environment include the need for agreements on industrial communication inter-

faces and protocols, common data models and the semantic interoperability of data, and

thus on a larger scale, platform inter-communication and inter-operability. The achieve-

ment of these objective will allow a boundaryless information flow among the single

product lifecycle phases [18] thus enabling an effective, whole-of-life product lifecycle

management (PLM). Indeed, the most significant obstacle is that valuable information

is not readily shared with other interested parties across the Beginning-of-Life (BoL),

Middle-of-Life (MoL), and End-of-Life (EoL) lifecycle phases but it is all too often

locked into vertical applications, sometimes called silos. Moreover, these objectives are

strictly related to the need of achieving the full potential of the Internet of Things in the

manufacturing industry. Indeed, without a trusted and secure, open, and unified infra-

structure for true interoperability, the parallel development of disparate solutions, tech-

nologies, and standards will lead the Internet of Things to become an ever-increasing

web of organization and domain-specific intranets.

The EU PROMISE project

developed the foundation of the Quantum Lifecycle

Management (QML) Technical Architecture to support and encourage the flow of

https://cordis.europa.eu/project/rcn/72825_en.html

https://isa-95.com/

The PROMISE Project (2004-2008): A European Union research project funded under the 6th

Framework Program (FP6) which focused on information systems for whole-of-life product

lifecycle management.

EU H2020 MSCA RISE Project FIRST - â

AIJvirtual Factories: Interoperation suppoRting buSiness innovationâ

lifecycle data between multiple enterprises throughout the life of an entity and its com-

ponents. QML was further developed by the Quantum Lifecycle Management (QLM)

a Work Group of The Open Group whose members work to establish open, vendor-

neutral IT standards and certifications in a variety of subject areas critical to the enter-

prise. The three main components of QML are the Messaging Interface (MI), the Data

Model (DM), and the Data Format (DF) [19]. The Message Interface provides a flexible

interface for making and responding to requests for instance-specific information. The

Data Model, instead, enables detailed information about each instance of a product to

be enriched with “field data”; i.e., detailed information about the usage and changes to

each instance during its life. Finally, the Data Format represents, through an XML

schema, the structure of the message exchanged between many products and/or sys-

tems. The structure of the message is similar to the Data Model schema so that it could

be easily recognize by a system QLM DM compatible, thereby automating the data

collection.

Various works adopt QLM for manufacturing assets representation and classifica-

tion. For instance, the paper [20] proposes data synchronization models based upon

QLM standards to enable the synchronization of product-related information among

various systems, networks, and organizations involved throughout the product lifecy-

cle. These models are implemented and assessed based on two distinct platforms de-

fined in the healthcare and home automation sectors. Främling, Kubler, & Buda [21]

describe two implemented applications using QLM messaging, respectively, defined in

BoL and between MoL-BoL.

The former is a real case study from the LinkedDesign EU FP7 project, in which

different actors work on a production line of car chassis. This process segment involved

two robots to transfer the chassis part from machine to machine. The actors involved in

the manufacturing plan expressed, on the one hand, the need to check each chassis part

throughout the hot stamping process and, on the other hand, the need to define commu-

nication strategies adapted to their own needs. Accordingly, scanners are added be-

tween each operation for the verification procedure, and QLM messaging is adopted to

provide the types of interfaces required by each actor. The latter, instead, involves ac-

tors from two distinct PLC phases: 1) In MoL: A user bought a smart fridge and a TV

supporting QLM messaging; 2) In BoL: The fridge designer agreed with the user to

collect specific fridge information over a certain period of the year (June, July, August)

using QLM messaging. Also in this case, the appropriate QLM interfaces regarding

each actor have been set up in such a way that the involved actors can get the required

information about the smart objects.

In most applications scenarios, taxonomies are usually adopted as common ground

for semantic interoperability. Classifying products and services with a common coding

scheme facilitates commerce between buyers and sellers and is becoming mandatory in

the new era of electronic commerce. Large companies are beginning to code purchases

in order to analyse their spending. Nonetheless, most company coding systems today

have been very expensive to develop. The effort to implement and maintain these sys-

tems usually requires extensive utilization of resources, over an extended period of

http://www.opengroup.org/subjectareas/qlm-work-group

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

time. Additionally, maintenance is an on-going, and expensive, process. Another prob-

lem is that company’s suppliers usually don’t adhere to the coding schemes of their

customers, if any.

Samples of taxonomy including the description and classification of manufacturing

assets and services are: eCl@ss, UNSPSC, and MSDL. eCl@ss

is an international

product classification and description standard for information exchange between cus-

tomers and their suppliers. It provides classes and properties that can be exploited to

standardise procurement, storage, production, and distribution activities, both intra-

companies and inter-companies. It is not bound to a specific application field and can

be used in different languages. It is compliant to ISO/IEC. It adopts an open architecture

that allows the classification system to be adapted to an enterprise’s own internal clas-

sification scheme, so granting flexibility and standardization at the same time. Thanks

to its nature, it can be exploited in the Internet of Things field in order to enable in-

teroperability among devices of different vendors. As of October 2017, there are about

41,000 product classes and 17,000 uniquely described properties which are categorized

with only four levels of classification; this enables every product and service to be de-

scribed with an eight-digit code. One of the aims of eCl@ass is to decrease inefficien-

cies, so that packaging and distribution take place automatically, relying on the classes

and identifier available by the standard. The nature of eCl@ss enables the definition of

several aspects in virtual factories.

The United Nations Standard Products and Services Code (UNSPSC)

provides an

open, global multi-sector standard for efficient, accurate classification of products and

services. The UNSPSC was jointly developed by the United Nations Development Pro-

gramme (UNDP) and Dun & Bradstreet Corporation (D & B) in 1998. It has been man-

aged by GS1 US since 2003. UNSPSC is an efficient, accurate and flexible classifica-

tion system for achieving company-wide visibility of spend analysis, as well as, ena-

bling procurement to deliver on cost-effectiveness demands and allowing full exploita-

tion of electronic commerce capabilities. Encompassing a five-level hierarchical clas-

sification codeset, UNSPSC enables expenditure analysis at grouping levels relevant to

the company needs. The codeset can be drilled down or up to see more or less detail as

is necessary for business analysis. The UNSPCS classification can be exploited to per-

form analysis about company spending aspects, to optimize cost-effective procurement,

and to exploit electronic commerce capabilities.

The Manufacturing Service Description Language (MSDL) [22] is a formal ontology

for describing manufacturing capabilities at various levels of abstraction including the

supplier-level, process-level, and machine-level. It covers different concepts like ac-

tors, materials, like ceramic and metal, physical resources, tools, and services. Descrip-

tion Logic is used as the knowledge representation formalism of MSDL in order to

make it amenable to automatic reasoning. MSDL can be considered an “upper” ontol-

ogy, in the sense that it provides the basic building blocks required for modeling domain

objects and allows ontology users to customize ontology concepts based on their

speciﬁc needs; this grants flexibility and standardization at the same time. MSDL is

composed of two main parts: 1) MSDL core and 2) MSDL extension. MSDL core is

http://www.eclasscontent.com/index.php?language=en&version=7.1

http://www.unspsc.org/

EU H2020 MSCA RISE Project FIRST - â

AIJvirtual Factories: Interoperation suppoRting buSiness innovationâ

the static and universal part of MSDL that is composed of basic classes for manufac-

turing service description; MSDL extension is dynamic in nature and includes a collec-

tion of taxonomies, sub-classes and instances built by users from different communities

based on their speciﬁc needs; MSDL extensions drive evolution of MSDL over time.

3 On-the-fly Manufacturing Service-oriented Process

Verification

The increasing digital interconnection of people and things, anytime and anywhere, is

referred to as hyperconnectivity. Advances in connectivity are already leading to strong

development and enhancement of networked service-oriented collaborative organisa-

tional structures. Hyperconnectivity, including developments in the areas of Internet of

Things and Cyber-Physical Systems, provides new opportunities to manufacturing,

value-added services, e-Health and care, crisis/disaster management, logistics, etc. Col-

laboration and networking are critical to hyperconnected world and intelligent autono-

mous systems. BU research is thus focused on design effective collaborative processes

for Internet scale and verifying correctness of distributed collaborative processes at

runtime, as well as deals with interoperability of multi-party collaborative processes at

a change environment, which is key enablement of intelligent autonomous systems,

such as virtual factory.

To be able to verify distributed collaborative business processes [29], the process

description modelling language goes beyond traditional activity level, which into in-

voked services. Therefore, avoiding conflicts of conditions of invoked service could be

checked at control flow level, which is beyond state-of-the-art. Taking recent ad-

vantages in Big Data, our research demonstrates the ability to collect large amounts of

data and metrics on a large variety of processes, including distributed collaborative pro-

cesses [30]. To make maximum use of this data, the new runtime verification tools are

created and designed to take this data into account, which is the key issues of WP4 of

the FIRST project.

Compliance constrains business processes to adhere to rules, standards, laws and

regulations. Non-compliance subjects enterprises to litigation and financial fines. Col-

laborative business processes cross organizational and regional borders implying that

internal and cross regional regulations must be complied with. To protect customs’ data,

European enterprises must comply with the EU data privacy regulation (general data

protection regulation - GDPR) and each member state’s data protection laws. Compli-

ance verification is thus essential to deploy and implement collaborative business pro-

cess systems. It ensures that processes are checked for conformance to compliance re-

quirements throughout their life cycle. BU research also looks at checking authorisation

compliance among collaborative business processes in the context of virtual facto-

ries/enterprises. Security is an important issue in collaborative business processes, in

particular for applications that handle sensitive personal information and checking com-

pliance of collaborative business processes in the virtual factory/environment context

as well as providing traceable commercial sensitive data distribution for using new

technologies, such as blockchain, etc.

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

The design of an interoperability framework provides support for compatibility and

evolution, which is essential for designing intelligent autonomous systems. Building

upon the results of semantic asset/service/process description languages and service-

oriented distributed collaborative process verification methods, a common schema and

schema evolution framework is used to facilitate interoperability on data/information,

services and processes respectively. It enables: global (manufacturing) process collab-

oration; improving flexibility of existing product design and manufacturing processes;

supporting (new) product-service linkage; and improved management of distributed

manufacturing assets. BU research particularly focus on identifying related concepts of

factories of the future and research challenges of interoperability of virtual factory are

addressed. Following BU previous research on resilience of SOA collaborative process

systems, paper [23] and architecture design of collaborative processes for managing

short term, low frequency used collaborative processes; the results of our research pro-

vide the good foundation for identifying research requirements for D1.1 of the FIRST

project. More specific explanations of each scientific papers are followed.

The internet and pervasive technology like the Internet of Things (i.e. sensors and

smart devices) have exponentially increased the scale of data collection and availability.

This big data not only challenges the structure of existing enterprise analytics systems

but also offer new opportunities to create new knowledge and competitive advantage.

Businesses have been exploiting these opportunities by implementing and operating big

data analytics capabilities. Social network companies such as Facebook, LinkedIn,

Twitter and Video streaming company like Netflix have implemented big data analytics

and subsequently published related literatures. However, these use cases did not pro-

vide a simplified and coherent big data analytics reference architecture as well as cur-

rently, there still remains limited reference architecture of big data analytics. [30] aims

to simplify big data analytics by providing reference architecture based on existing four

use cases and subsequently verified the reference architecture with Amazon and Google

analytics services.

The users of virtual factory are not experts in business process modelling to guaran-

tee the correct collaborative business processes for realising business process execu-

tion. To enable automatic execution of business processes, verification is an important

step at the business process design stage to avoid crashes or other errors at runtime.

Research in business process model verification has yielded a plethora of approaches

in form of methods and tools that are based on different technologies like Petri nets

family and temporal logic among others. From the literature no report specifically tar-

gets and presents a comparative assessment of these approaches based on criteria as one

we propose. [29] therefore presents an assessment of the most common verification

approaches based on their expressibility, flexibility, suitability and limitations. Further-

more, we look at how big data impacts the business process verification approach in a

data-rich world.

[31] proposes architecture of collaborative processes for managing short term, low

frequency collaborative processes. A real world case of collaborative processes is used

to explain the design and implementation of the cloud-based solution for supporting

collaborative business processes. Service improvement of the new solution and com-

puting power costs are also analysed accordingly.

EU H2020 MSCA RISE Project FIRST - â

AIJvirtual Factories: Interoperation suppoRting buSiness innovationâ

In paper [23], we have proposed resilience analysis perspectives of SOA collabora-

tive process systems, i.e., overall system perspective, individual process model perspec-

tive, individual process instance perspective, service perspective, and resource perspec-

tive. A real world collaborative process is reviewed for illustrating our resilience anal-

ysis. This research contributes to extend SOA collaborative business process manage-

ment systems with resilience support, not only looking at quantification and identifica-

tion of resilience factors, but also considering ways of improving the resilience of SOA

collaborative process systems through measures at design and runtime.

Concepts and research challenges of interoperability of virtual factory are addressed

[32]. We present a comprehensive review on basic concepts of factories of the future,

i.e. smart factory, digital factory and virtual factory. The relationships among smart

factory, digital and virtual factory are studied. Interoperability of virtual factories is

defined. Challenges of interoperability of virtual factories are identified: lack of stand-

ards of virtual factories; managing traceability of sensitive data, protected resources

and applications or services are critical for forming and using virtual factories; handling

multilateral solutions and managing variability of different solutions/virtual factory

models are also impact to the usability of the virtual factory. In short, the interoperabil-

ity of virtual factory related to many newly developed ICT of the hardware and software

innovation. An interoperation framework allows evolutional and handling changes,

which is crucial for generating and maintaining virtual factories among different indus-

trial sectors.

4 A Reliable Interoperability Architecture for Virtual Factories

One of the key issues in digital factories is to provide, manage and use the different

services and data that are connected to the production processes. Manufacturing ma-

chines typically provide data about their status and services. These services are usually

exploited at the digital factory level together with data and services coming from other

departments, such as purchasing and marketing. We face heterogeneous situations:

from the one hand, machines are from different vendors and, even if not proprietary,

they are likely to adopt different standards and vocabulary, and data are managed by

different systems as well; from the other hand, services can be provided at different

levels of granularity, from very fine grained one (in terms of functionalities) to very

coarse. The role of the digital factory is to integrate the different services and data and

to combine them in order to make the whole process as efficient and competitive as

possible in the achievement of the specific goals.

Another importation issue to be faced is the fact that the process can cover a space

wider than the single factory (it supports a supply chain): usually a factory gets the raw

material from suppliers and provide products or semi-finished products to customers,

through delivery agents, requiring the corresponding services and data to integrate to

each other or at least to be able to interact in a scalable and flexible way.

We propose to achieve this through a general three-layer interoperability framework,

i.e., based on processes, services, and data, and assists users in the achievement of their

objectives through the discovery of service and data flows that best fit the expressed

requirements. In the following the three layers are detailed.

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

Process Space Layer - Goal-oriented Process Specification

The top layer of the proposed architecture deals with the goals and the processes able

to achieve such goals. Notably, companies would like to define, on the basis of such

goals, specific KPIs – Key Performance Indicators, which qualify the QoS of the pro-

duction process. Clearly goals and KPIs are defined over many aspects, including the

interactions with external companies being part of the process.

Service Space Layer - Dynamic Service Discovery and Composition

Starting from the goals and processes defined in the process layer, services must be

dynamically composed to achieve the goal(s). OpenAPIs are exposed by such services

in order to control, discover, and compose them in a dynamic way. Rich semantic de-

scriptions of the services should be available in the interoperability platform, in order

to support both the discovery of the services and their execution/invocation. The de-

scriptions should include some keywords that identify the context of the service (e.g.,

“food”, “cooking”), the equipment (e.g., “oven”, “mixer”), the performed operation

(e.g., “turn-on”, “speedup”), and the parameters (e.g., “temperature”, “speed”).

With regard to the discovery phase, the semantic description is exploited to search

for specific services without knowing their exact name and their syntax a priori. Se-

mantic techniques can be exploited to find synonyms and keywords related to the words

searched for in this phase. Searches can be performed either automatically by the pro-

cess layer, in particular by the orchestration engine enacting processes, or by a human

operator acting in the factory, which may be involved when needed (e.g., the adaptation

techniques realized in the process layer fail, and a human intervention is needed in order

to make the process progress) [24].

But the semantic descriptions can be exploited also in the composition phase. Being

the composition dynamic, the platform must not only find but also exploit the needed

service in an automatic way or providing an effective support to the human operator.

To this purpose, the semantic description of the service parameters is needed in order

exploit the meta-services of the data layer to adapt the client service invocation to the

server syntax (see next subsection). Some proposals and examples of semantic service

descriptions exist, such as in the SAPERE project [25] mentioned later.

The dynamism is useful to handle unexpected situations, often notified by a human

operator. Clearly, the platform must also consider failure situations, such as an equip-

ment out of work, and so on. These issues require the frequent involvement of humans

in the loop in order to deal with them in an effective way.

Data Space Layer - Service-oriented Mapping Discovery and Dynamic

Dataspace Alignment

Data are managed and accessed in a data space. The data space must be able to deal

with a huge volume of heterogeneous data by autonomous sources and support the dif-

ferent information access needs of the service level. In particular, a large variety of data

types should be managed at the dataspace level. According to the level of dynamicity,

EU H2020 MSCA RISE Project FIRST - â

AIJvirtual Factories: Interoperation suppoRting buSiness innovationâ

data can be static such as data available in traditional DBMSs but also highly dynamic

like sensor data that are continuously generated. Moreover, it should accommodate data

that exhibit various degrees of structures, from tabular data like relational data and CSV

data to fully unstructured data like textual data. Finally, it should cope with the very

diversified data access modalities sources offer, from low level streaming access to high

level data analytics.

To this extent, the data modelling abstraction we adopt to represent the data space is

fully decentralized, thereby bridging, on the one hand, existing dataspace models that

usually rely on a single mediated view and, on the other hand, P2P approaches for data

sharing [26]. The dataspace is therefore a collection of heterogeneous data sources that

can be involved in the processes, both in-factory and out-factory. Those data are either

describing the manufactured products or the manufacturing processes and assets (ma-

terial, machine, enterprises, value networks and factory workers) [27]. Each data source

has its data access model that describes the kind of managed data, e.g., streaming data

vs. static data, and the supported operators. As an example, sensed parameters such as

temperature in an oven, temperature in a packing station, etc. are all streaming data

needed in the dataspace that can be accessed only through simple windowing operators

on the latest values. On the other hand, supplier data can be recorded in a DBMS that

offers a rich access model both for On Line Transaction Process (OLTP) operations and

On Line Analytical Process (OLAP) operations.

Data representation relies on the graph modelling abstraction. This model is usually

adopted to represent information in rich contexts. It employs nodes and labelled edges

to represent real world entities, attribute values and relationships among entities.

The main problem the interoperability platform must cope with when dealing with

data is data heterogeneity. Indeed, the various services gather data, information and

knowledge from sources distributed over different stakeholders and external sources,

e.g., the delivery agents and the Web. All these sources are independent, and we argue

that a-priori agreements among the distributed sources on data representation and ter-

minology is unlikely in large digital supply chains over several digital factories.

Data heterogeneity can concern different aspects: (1) different data sources can rep-

resent the same domain using different data structures; (2) different data sources can

represent the same real-world entity through different data values; (3) different sources

can provide conflicting data. The first issue is known as schema heterogeneity and is

usually dealt with through the introduction of mappings. Mappings are declarative spec-

ifications describing the relationship between a target data instance and possibly more

than one source data instance. The second problem is called entity resolution (a.k.a.

record linkage or duplicate detection) and consists in identifying (or linking or group-

ing) different records referring to the same real-world entity. Finally, conflicts can arise

because of incomplete data, erroneous data, and out-of-date data. Returning incorrect

data in a query result can be misleading and even harmful. This challenge is usually

addressed by means of data fusion techniques that are able to fuse records on the same

real-world entity into a single record and resolve possible conflicts from different data

sources.

Traditional approaches that address data heterogeneity propose to first solve schema

heterogeneity by setting up a data integration application that offers a uniform interface

to the set of data sources. This requires the specification of schema mappings that is a

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

really time- and resource-consuming task entrusted to data curation specialists. This

solution has been recognized as a critical bottleneck in large scale deeply heterogeneous

and dynamic integration scenarios, as digital factories are. A novel approach is the one

where mapping creation and refinement are interactively driven by the information ac-

cess needs of service flows and the exclusive role of mappings is to contribute to exe-

cute service compositions [28]. Hence, we start from a chain of services together with

their information needs expressed as inputs and outputs which we attempt to satisfy in

the dataspace. We may need to discover new mappings and refine existing mappings

induced by composition requirements, to expose the user to the inputs and outputs

thereby discovered for their feedback and possibly continued adjustments. Therefore,

the service composition induces a data space orchestration that aims at aligning the data

space to the specific service goals through the interactive execution of three steps: map-

ping discovery and selection, service composition simulation, feedback analysis. Map-

pings that are the outcome of this process can be stored and reused when solving similar

service composition tasks.

5 Concluding Remarks

The FIRST project will continue till the end of 2020. In this paper we have presented

the initial outcomes of the project, which will be improved and refined over the next

years.

References

1. EFFRA. ‘’Factories of the Future 2020 Roadmap Multiannual roadmap for the contractual

PPP under Horizon 2020. (2013)

2. Debevec, M., M. Simic, and N. Herakovic. "Virtual factory as an advanced approach for

production process optimization" International journal of simulation modelling 13, no. 1

(2014): 66-78.

3. Reference Architecture Model Industrie 4.0 (RAMI4.0). https://www.zvei.org/fileadmin/

user_upload/Presse_und_Medien/Publikationen/2016/januar/GMA_Status_Report__Refer

ence_Archtitecture_Model_Industrie_4.0__RAMI_4.0_/GMA-Status-Report-RAMI-40-

July-2015.pdf .

4. Naumann, F., & Riedl, M. (2011). EDDL - Electronic Device Description Language.

Oldenbourg Industrieverl.

5. Field Device Integration Technology. http://www.fdi-cooperation.com/tl_files/images/

content/Publications/FDI-White_Paper.pdf.

6. Shani, U., Franke, M., Hribernik, K. A., & K. D. Thoben. (2017). Ontology mediation to

rule them all: Managing the plurality in product service systems. 2017 Annual IEEE

International Systems Conference (SysCon), (pp. 1-7). Montreal, QC: IEEE.

7. Franke, M., Klein, K., Hribernik, K., Lappe, D., Veigt, M., & Thoben, K. D. (2014).

Semantic web service wrappers as a foundation for interoperability in closed-loop product

lifecycle management. . Procedia CIRP, 22, 225-230.

8. Pitoura, E., Bukhres, O., & Elmagarmid, A. (1995). Object orientation in multidatabase

systems. ACM Computing Surveys (CSUR), 27(2), 141-195.

EU H2020 MSCA RISE Project FIRST - â

AIJvirtual Factories: Interoperation suppoRting buSiness innovationâ

9. Martin, D., Domingue, J., Sheth, A., Battle, S., Sycara, K., & Fensel, D. (2007). Semantic

web services, part 2. IEEE Intelligent Systems, 22(6), 8-15.

10. Srinivasan, V., Lämmer, L., & Vettermann, S. (2008). On architecting and implementing a

product information sharing service. Journal of Computing and Information Science in

Engineering, 1-11.

11. Wang, X., & Xu, X. ( 2013). An interoperable solution for Cloud manufacturing. Robotics

and Computer-Integrated Manufacturing, 29(4), pp.232-247.

12. Jardim-Goncalves, R., Grilo, A., & Steiger-Garcao, A. (2006). Challenging the

interoperability between computers in industry with MDA and SOA. Computers in Industry,

57(8), 679-689.

13. PLM Services 2.1 . (2011, May). Object Management Group.

14. Product Life Cycle support (PLCS) Web services V2. (n.d.). OASIS . http://www.plcs-

resources.org/plcs_ws/v2/ .

15. Srinivasan, V., Lämmer, L., & Vettermann, S. (2008). On architecting and implementing a

product information sharing service. Journal of Computing and Information Science in

Engineering, 1-11.

16. ISO 10303-239 2005. Industrial automation systems and integration–Product data

representation and xchange–Part 239: Application protocol: Product life cycle support.

International Organization for Standardization, Geneva (Switzerland).

17. EFFRA, H. 2. (2016). Factories 4.0 and Beyond Recommendations for the work programme

18-19-20 of the FoF PPP. Retrieved from Horizon 2020 EFFRA:

http://effra.eu/sites/default/files/factories40_beyond_v31_public.pdf

18. QLM. (2012). An introduction to Quantum Lifecycle Management (QLM) by The Open

Group QLM work Group. Retrieved from http://docs.media.bitpipe.com/io_10x/

io_102267/item_632585/Quantum%20Lifecycle%20Management.pdf

19. Parrotta, S., Cassina, J., Terzi, S., Taisch, M., Potter, D., & Främling, K. (2013). Proposal

of an interoperability standard supporting PLM and knowledge sharing. In IFIP International

Conference on Advances in Production Management Systems (pp. 286-293). Springer.

20. Kubler, S., Främling, K., & Derigent, W. (2015). P2P Data synchronization for product

lifecycle management. Computers in Industry, 66, 82-98.

21. Främling, K., Kubler, S., & Buda, A. (2014). Universal messaging standards for the IoT

from a lifecycle management perspective. IEEE Internet of things journal, 1(4), , 319-327.

22. Ameri, F., & Dutta, D. (2006). An upper ontology for manufacturing service description. In

ASME Conference, (pp. 651-661).

23. de Vrieze P., Xu L. (2018) Resilience Analysis of Service Oriented Collaboration Process

Management systems. Service Oriented Computing and Applications. Springer. 2018.

24. Marrella, A., Mecella, M., & Sardiña, S. (2017). Intelligent Process Adaptation in the

SmartPM System. ACM Transaction on intelligent systems, 8(2).

25. Castelli, G., Mamei, M. Rosi, A., & Zambonelli F. (2015). Engineering pervasive service

ecosystems: The SAPERE approach. ACM Transactions on autonomous and adaptive

systems, 10(1).

26. Penzo, W., Lodi, S., Mandreoli, F., Martoglia, R., Sassatelli, S. (2008). Semantic peer, here

are the neighbors you want!. Proceedings of ACM EDBT.

27. EFFRA Factories 4.0 and Beyond Recommendations for the work programme 18 - 19 - 20

of the FoF PPP under Horizon 2020, Online: http://www.effra.eu/sites/default/files/

factories40_beyond_v31_public.pdf

28. Mandreoli, F. (2017). A Framework for User-Driven Mapping Discovery in Rich Spaces of

Heterogeneous Data. Proc. 16th OTM Conferences (2), pp. 399-417.

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

29. Kasse, J.P., Xu, L,, de Vrieze, P, (2017) A Comparative Assessment of Collaborative

Business Process Verification Approaches. In:18th IFIP Working Conference on Virtual

Enterprises (PRO-VE 2017), Vicenza, Italy, 18 Sep 2017 - 20 Sep 2017. Springer

30. Sang, G., Xu, L., de Vrieze, P. (2017) Simplifying Big Data Analytics System with

Reference Architecture. In: 18th IFIP Working Conference on Virtual Enterprises (PRO-VE

2017), Vicenza, Italy, 18 Sep 2017 - 20 Sep 2017. Springer.

31. Xu, L, de Vrieze, P. (2018) Supporting Collaborative Business Processes: a BPaaS

Approach. International Journal of Simulation and Process Modelling. Inderscience.

32. Xu, L., de Vrieze, P., Yu, H., Phalp, K., and Bai, Y. (2018) Interoperability of Virtual

Factory: an Overview of Concepts and Research Challenges. International Journal of

Mechatronics and Manufacturing Systems.

EU H2020 MSCA RISE Project FIRST - â

AIJvirtual Factories: Interoperation suppoRting buSiness innovationâ