Erasmus 2020

Computers in Industry 123 (2020) 103297
Contents lists available at ScienceDirect
Computers in Industry
journal homepage: www.elsevier.com/locate/compind
Using business process models for the specification of manufacturing

operations
Jonnro Erasmus a,∗ , Irene Vanderfeesten b , Konstantinos Traganos a , Paul Grefen a
a
Eindhoven University of Technology, De Zaale, 5600MB, Eindhoven, the Netherlands
b
Open Universiteit, Valkenburgerweg 177, 6419AT, Heerlen, the Netherlands
a r t i c l e i n f o a b s t r a c t
Article history: Business process management (BPM) is often employed as a driver of integration, by articulating busi-
Received 11 November 2019 ness processes that cross the boundaries of individual business functions. Business process modelling
Received in revised form 19 July 2020 as part of BPM has shown its potential in administrative environments, such as banking and insurance
Accepted 22 July 2020
organizations. However, business process modelling remains unproven for all the processes encountered
in manufacturing enterprises, including the physical operations processes that transform input materi-
Keywords:
als into the required product. These processes have physical characteristics that make them essentially
Manufacturing operations
different from administrative processes with a highly digital nature, like limited physical buffers for inter-
Business process modelling
Cross-functional process management
mediate products and transportation times between activities. In this paper, we present an approach to
use business process models for the specification of these physical operations processes. Our approach is
based on a catalogue of common process fragments that are adapted to the physical nature of manufac-
turing. These process fragments serve as reusable building blocks for the specification of manufacturing
processes. We use the industry standard Business Process Model and Notation (BPMN) to represent both
fragments and processes. We demonstrate the use of our catalogue of fragments by modelling and enact-
ing the real-world processes of ten manufacturing organizations. The results show strong support for
the use of business process models for both representation and enactment of manufacturing processes.
This work closes part of the gap between the support for administrative processes and physical industrial
processes, thus contributing to the advent of the smart manufacturing concept in the context of Industry
4.0.
© 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
1. Introduction management is extensively studied, but these studies focus on the

alignment of planning to achieve maximum production throughput
The challenge and appeal of improved integration between (O’Leary-Kelly and Flores, 2002; Sale et al., 2017).
manufacturing operations and business management functions Business process management (BPM) is often employed to
are well understood (Hausman et al., 2002; Tang, 2010). The cross the boundaries of business functions and improve integra-
techniques, skills and information systems employed to man- tion (Berente et al., 2009; Hanson et al., 2002; Kobayashi et al.,
age business activities can differ significantly from those used to 2003). Improved integration can lead to higher levels of effective-
manage operations, leading to deficient throughput and flexibility ness and efficiency, but also to a better structure of processes that
(Sawhney, 2006; Tang, 2010). For example, the lean manufacturing is the basis for flexibility in operations. The same need exists in
principles often applied in operations management are not easily the manufacturing industry, torn between the drive for efficiency
transferrable to resource or financial management. Likewise, busi- (to be competitive) and the demand of more flexibility in pro-
ness process reengineering approaches for increasing the efficiency cesses (to cater for customer-orientation and mass-customization).
of highly digitized administrative processes cannot be directly In addition, the manufacturing industry suffers from disparate and
mapped to physical processes on the factory floor. The performance fragmented process management across multiple information sys-
benefit of improved integration between business and operations tems (Erasmus et al., 2018). Indeed, Prades et al. (2013) make the
case for integration between enterprise resource planning systems
and manufacturing execution systems, by using Business Process
Model and Notation (BPMN) for process modelling in both infor-
∗ Corresponding author. mation systems. Conversely, Gerber et al. (2014) investigate how
E-mail address: [email protected] (J. Erasmus).
https://doi.org/10.1016/j.compind.2020.103297
0166-3615/© 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
2 J. Erasmus, I. Vanderfeesten, K. Traganos et al. / Computers in Industry 123 (2020) 103297
process models can be converted between the information systems settings, a proper evaluation of this basis in practice is essential. We
that support business management and operations management. performed this evaluation by applying the developed catalogue of
More recently, a single business process management system for model fragments in 10 factories across Europe. These factories vary
business and operations management is proposed and demon- extensively with respect to their size and the products they pro-
strated (Pauker et al., 2018). duce. This shows the broad applicability of the developed artefact.
While these studies embrace the ambition of cross-functional This paper is structured as follows. In Section 2, we present the
process management in manufacturing, they offer little practical methodology used for the execution of our research and link the
guidance on the modelling of executable manufacturing processes: steps in this methodology to the structure of this paper. In Section
they provide concepts and tools, but not the process modelling 3, we discuss related work, paying attention to business process
approach. This article presents such practical guidance in the form management in manufacturing, the use of BPMN in this context,
of a structured catalogue of process model fragments. These frag- and general approaches in BPM that have some similarities with our
ments function as process building blocks that can be composed work. In Section 4, we analyze manufacturing operations to develop
to design executable manufacturing processes. In the context of a catalogue of model fragments that represent reusable building
design science research, Gregor and Hevner (2013) refer to this blocks for process modelling. We continue in Section 5 by showing
type of scientific contribution as nascent design theory: knowl- how the model fragments can be combined into executable process
edge as operational principles or architecture. As with all scientific models and then briefly discuss how process automation can be
research, the purpose of design science research is to create achieved with a business process management system (BPMS). In
knowledge. The knowledge presented in this article is prescrip- Section 6 we evaluate our work by demonstration in real factories
tive knowledge to help practitioners solve problems, rather than and in Section 7 we discuss our findings. Lastly, conclusions are
descriptive knowledge used to explain observed phenomena. drawn in Section 8.
For information systems research, Hevner et al. (2004) argues
that design science is inseparable from behavioral science. Behav-
2. Research methodology
ioral science attempts to explain phenomena related to business
needs, through the development and justification of theories. In
As already indicated in the introduction, the research presented
contrast, design science aims to meet the identified business need,
in this paper is classified as design science research because the
through the building and evaluation of artefacts. These artefacts are
goal is to create prescriptive knowledge for problems encountered
packages of prescriptive knowledge that help practitioners con-
in practice. We use the well-established framework for design sci-
front problems encountered in practice. March and Smith (1995)
ence in information systems of Hevner et al. (2004) to illustrate
define the following four types of artefacts:
the structure of our research methodology, as shown in Fig. 1.
The three columns in this figure are used to depict the environ-
Constructs: A language or framework within which problems
ment, research and knowledge base, according to the design science
and solutions can be studied, analyzed and communicated.
research framework. The environment column represents the prac-
Models: Applied constructs that represent a real-world scenario
tical context of the research, i.e. the manufacturing industry. Paying
or phenomenon. Models are used by practitioners to aid under-
adequate attention to this environment safeguards the relevance of
standing of problems, solutions or the relationship between
the research. The research column includes the activities performed
elements of a system.
in our work, i.e., the development of the catalogue of process frag-
Methods: Guidance on actions to be executed in practice. The
ments. Lastly, the knowledge base column represents the collective
guidance can range from textual descriptions of activities to be
scientific literature related to the research topic. Paying adequate
performed by humans to formal, mathematical algorithms to be
attention to the knowledge base safeguards the academic rigor of
executed by computers.
our work. The section indicators in Fig. 1 point to the sections in
Instantiations: Implementations of constructs, models or meth-
this paper where the research activities are reported.
ods. Implementations enable more thorough assessment of
The research in the center column of Fig. 1 is grouped into four
artefacts.
activities:
The design science research artefact presented in this article

is a catalogue of model fragments that can be used as building 1 The first activity creates a compilation of a catalogue of manu-
blocks to design executable manufacturing operations processes. facturing operations from scientific literature and international
The model fragments are created with BPMN2.0, as the de-facto standards.
standard for business process modelling (Decker and Barros, 2008; 2 The second activity entails the decomposition of those manufac-
Takemura, 2008), to maximize their potential utility. Each frag- turing operations into modelling concepts. This activity serves to
ment contains the structure of a specific manufacturing process increase transparency and reproducibility of the research, as it
construct. By connecting fragments with the appropriate ‘process shows how the catalogue of operations is translated into model
glue’, i.e. connecting control flow specification, we obtain a com- elements.
plete specification of a manufacturing process. We organize the 3 The third activity leverages the modelling concepts to yield a
fragments in a catalogue that follows a classification of operations catalogue of process model fragments using BPMN2.0.
in the manufacturing domain. The catalogue of reusable elements 4 The fourth and last research activity shows how the model
is a tool to deal with the complexity of process modelling in the fragments can be combined into executable manufacturing oper-
manufacturing domain and to advance standardization of process ations processes.
models. As such, the catalogue is also the basis for an instantiation
artefact. Our use of fragments in process specification is similar The catalogue of fragments is the primary contribution of this
to the use of business process patterns (Gschwind et al., 2008; van research and is evaluated in industrial practice, as shown in the
der Aalst et al., 2003, 2000) or business process snippets (Baumgraß environment column of the Fig. 1. Thus, the set of process model
et al., 2015) in the BPM domain, or the use of software patterns in fragments is the artefact, in design science research terminology,
the software engineering domain (Fowler, 1997). that is contributed to the knowledge base. The development is per-
As the purpose of this work to provide a scientifically sound basis formed according to the regulative cycle, as advocated by Wieringa
for structured business process design in practical manufacturing (2009). The integration problem experienced in practice (as dis-
J. Erasmus, I. Vanderfeesten, K. Traganos et al. / Computers in Industry 123 (2020) 103297 3
Fig. 1. Overview of research methodology.
cussed in Section 1) is translated into a knowledge problem (the advices, specified as process model fragments. At run-time, aspects
application of business process modelling for manufacturing opera- are interwoven with the main process, forming the to-be-followed
tions). The solution design is presented in the form of process model execution flow.
fragments for manufacturing operations. Additionally, it is shown Hewelt and Weske (2016) propose a hybrid approach for flex-
how those fragments can be combined into executable processes ible case management, addressing both modelling and execution.
and this is also demonstrated in practice, at ten factories across Their approach is based on modelling business scenarios as small
Europe. fragments of case models and at runtime using data object states
to combine fragments. Their use of fragments for modelling is
comparable to our approach, but their approach has three impor-
3. Related work
tant differences with respect to ours. Firstly, our approach relies
on design-time assembly of fragments into business processes,
This research presents a catalogue of model fragments that can
whereas their approach is driven by run-time assembly. Even
be used to model executable manufacturing processes. There are
though their approach may allow for more flexibility in process
several related studies with a similar objective. Production Case
composition than ours, it leaves important decisions to execu-
Management (PCM) sets a process goal that is refined in multiple
tion time, thereby making ex-ante resource allocation hard or
steps and there is no a priori end-to-end execution path (Meyer
even impossible, whereas this is important in the manufacturing
et al., 2014). However, PCM aims to reduce the need for a skilled
domain we address. Secondly, we propose a catalogue of manufac-
knowledge worker to push a case forward by using rigid process
turing primitives taking the physical nature of manufacturing very
fragments and object-centric process definitions. These process
explicitly into account. Hewelt and Weske (2016) address business
fragments are syntactically very similar to BPMN, though with more
processes with knowledge workers, so remain in the administrative
attention paid to data nodes. The data nodes are used to model
domain that is typical for most work in business process man-
pre- and post-conditions of activities, providing a basis for linkage
agement. Finally, the work in Hewelt and Weske (2016) focuses
between the process fragments. An implementation of PCM exists
on formal semantics of their assembly approach, with only a toy
and is presented in the work of Haarmann et al. (2015).
example in the academic domain – hence there is no evaluation in
Chung et al. (2003) developed a Task Based Process Manage-
actual practice. From the onset, our work has been contextualized
ment (TBPM) system that uses a library of plans that are linked
by industrial practice, including evaluation in this practice. That
together using process models. A plan represents one of poten-
said, the work of Hewelt and Weske (2016) is invaluable to use,
tially multiple ways of achieving a task, by breaking it down into
as we apply their rules of process progress to develop the model
a structure of sub tasks. The Aspect-Oriented (AO) approach (Jalali
fragments.
et al., 2013) attempts to reduce complexity of the main process by
separating concerns into aspects. An aspect contains one or more
Marengo et al. (2018) discuss an approach to model business are appealing to audiences confined to specific industry sectors.
processes that pays explicit attention to the representation of the Such extension proposals are prevalent enough to warrant a sur-
items on which process activities are to be executed. Their approach vey by Braun and Esswein (2014). The survey found 30 extensions,
is aimed at the construction domain. There is a link to our work in classified according to conformance to the BPMN2.0 standard and
the observation that we pay explicit attention to the physical nature organized per application domain. The survey found that four out
of items as well, as our fragments have been defined taking the of five extensions do not conform to the BPMN2.0 standard. Only
physical characteristics of manufacturing into account. Marengo one extension proposal was related to manufacturing (Zor et al.,
et al. (2018) base their work, however, on an extension and for- 2011) and it was found to lack an abstract syntax and contained
malization of a declarative way of specification, whereas we use semantic conflicts with the standard. Braun and Esswein (2014)
an approach with explicit control flow specification that is close to do not offer an explanation, but we can conclude that either there
the physical processes in manufacturing. Like in the work of Hewelt is little need for an extension for manufacturing, or there is little
and Weske (2016), Marengo et al. (2018) focus on formal semantics. need for BPMN in manufacturing. It is therefore necessary to inves-
Even though they use a case from construction practice (building a tigate existing process modelling and enactment techniques and to
hotel), the evaluation of their work is not based on usability in prac- scrutinize previous applications of BPMN in manufacturing.
tice, but on formal satisfiability checking, i.e., on checking that at
least one process execution exists that satisfies the process specifi- 3.1. Process modelling and enactment in manufacturing
cation. For the manufacturing domain, we do not consider this very
relevant, as the issue here is not whether one process execution can The Integration DEFinition (IDEF) series of notations is a main-
satisfy a specification, but whether most process executions will stay in many industries, including defense (US Air Force, 1969),
satisfy a specification construction (Kamara et al., 2000) and even software development
The activities specified in the model fragments must ultimately (Kim et al., 2003). The use of IDEF is typically confined to the cre-
be performed by resources. Some techniques aim to reduce the ation of systems though, rather than the utilization of systems.
specificity of the process models to empower the resources. Con- That is equally true for manufacturing, where IDEF is often used
Dec (Pesic and van der Aalst, 2006) refrains from specifying control to design and improve manufacturing systems, but not usually to
flow between activities and rather opts for constraints that must drive execution (Cheng-Leong et al., 1999).
be met as relationships between the activities. As most declarative Value stream mapping (VSM) is quite ubiquitous in the manu-
languages, ConDec takes a so-called outside-in approach, mean- facturing sector thanks to the sheer success of lean manufacturing
ing that all behavior is allowed unless explicitly forbidden by a principles (Grewal, 2008). It’s entrenchment in production enter-
constraint. Similarly, Dynamic Condition Response Graphs (DCR prises has even led to proposals for adoption in business functions
Graphs) (Hildebrandt et al., 2012) contain events and five types of of a more administrative nature (Keyte and Locher, 2004). Further-
relations between them. Nested sub graphs can alternate between more, as with most business and process modelling notations, it is
completed and uncompleted state, particularly suitable for the eventually found wanting and the inevitable proposals for exten-
manufacturing domain where rework can be a common occurrence sions arise (Braglia et al., 2006). More importantly, neither IDEF nor
in processes. A third technique focuses on offering microservices, VSM include formal execution semantic to enable process enact-
which provide a certain functionality. These microservices are ment.
called by autonomous agents that intend to achieve a (process) S-BPM is a relatively new development aimed at handling the
goal (Oberhauser and Stigler, 2017). Agents navigate the land- complexity of multi-agent systems, with emphasis on manufac-
scape of microservices, which can be represented as a dependency turing (Fleischmann et al., 2013). S-BPM is a process modelling
graph. The structure of microservices maps well to the physical technique which emphasizes the different perspectives of inde-
domain, where a machine or resource typically provides certain pendent agents and communication between them. The activities
functionality that is required as input requirement by a down- and states of agents are modelled as a unit (a subject in this
stream production step. terminology), with the possibility of extensive communication
Although BPM was born out of the principles of production engi- between agents. The concept of communication between inde-
neering, it has been most successful in industry sectors that process pendent agents is equally relevant for communication between
information, rather than physical material. Most prominently, BPM agents in a single organization or agents across organizations
has been implemented extensively and successfully in financial (Fleischmann et al., 2015).
service organizations (Brahe, 2007; Weerdt et al., 2013). Never- S-BPM shows promise in the manufacturing domain, espe-
theless, several other industry sectors have also seen benefit from cially in the new subject of smart factories (Cadavid et al., 2015).
the application of BPM, including sectors with a ‘physical nature’ Vertical integration between process management and individ-
like automotive (Grefen et al., 2009) and transportation (Baumgraß ual agents has been demonstrated (Neubauer et al., 2014) and
et al., 2015). However, these applications of BPM almost exclusively practical evaluations have been successful (Neubauer and Stary,
focus on the business management functions in those industry sec- 2017). Additionally, S-BPM benefits from its mathematical under-
tors, rather than the activities that ‘touch’ the product. A surprising pinnings and natural language (Subject, Predicate, Object) structure
outlier is the healthcare sector, where patient handling processes (Fleischmann et al., 2015).
are modelled and sometimes enacted using a BPMS (Braun et al., Given the apparent advantages of S-BPM, the limited uptake of
2014; Reichert, 2011; Reijers et al., 2010; Shitkova et al., 2015; Van the notation and approach is surprising. Practical demonstration
Gorp et al., 2013). The advent of the Internet-of-Things has sparked has been successful, but it remains comparatively isolated. Scien-
the combination of physical objects and business process manage- tific research on the topic has also been slow to spread outside its
ment (Grefen et al., 2019), but most of this work is in an early stage, place of origin, as evidenced by the overlap of authors referenced
without adequate attention to structured approaches for modelling in this section. It is perhaps too early to say whether critical mass
complex applications. will be reached, but one significant barrier stands in its way: the
Regarding the notation used in this research, the remarkable growing popularity and enthusiasm for BPMN. BPMN also offers the
penetration of BPMN, both as a description of processes and as possibility of modelling processes as independent units (pools in
notation for process automation, has inevitably led to extension this terminology) with communication between agents and is sup-
proposals. Although the notation is quite capable of represent- ported for enactment in many information systems. Simply stated,
ing complex processes, domain-specific symbols and constructs BPMN has reached a de-facto standard level of penetration and
enjoys global support from organizations (Decker and Barros, 2008; Lastly, the concepts are converted into BPMN2.0 model elements
Takemura, 2008; Chinosi and Trombetta, 2012). to yield the catalogue of process model fragments.
4.1. Catalogue of manufacturing operations

3.2. BPMN in manufacturing
The first step of this research, as shown in Fig. 1, is to create a cat-
BPMN has certainly been considered as an option for manu- alogue of manufacturing operations. Although a provably complete
facturing operations (Zor et al., 2010). Indeed, García-Domínguez list of operations is unattainable, we can at least endeavor for good
et al. (2012) compared BPMN2.0 with VSM and IDEF3 in terms of representation of all operations. Thus, we discuss how a catalogue
the modelling of activity sequences, timing constraints, resource of manufacturing operations is constructed from literature.
assignment, material flow and information flow. The study found It would be highly preferential for this research to reference
that BPMN2.0 is comparable to IDEF3, with the addition of process and build on a well-established and comprehensive catalogue of
participants, event handlers and message flow; however, BPMN2.0 manufacturing operations, but unfortunately such a catalogue is
lacks the ability to model the physical aspects of a manufactur- not available. The rise of computer integrated manufacturing (CIM)
ing system. As for VSM, BPMN2.0 is found to be complementary, in the early 1990s prompted several studies into the definition of
because VSM is more concerned with the flow of material and manufacturing systems (Doumeingts et al., 1995), but these studies
information, rather than the exact sequence of activities. The mostly deal with the design approach and notation, rather than the
more important advantage of BPMN is that it includes execution type of manufacturing operations (Brandimarte and Cantamessa,
semantic. IDEF and VSM are adequate analysis and improvement 1995). A few exceptions are noted though. The Purdue Enterprise
notations, but BPMN can be used to enact processes with a BPMS. Reference Architecture (Williams, 1994) is the result of one of
Witsch and Vogel-Heuser (2012) also compared BPMN to other these studies and eventually led to the IEC 62264:2013 standard
notations, but rather as the foundation for the formal specifica- (IEC, 2013) that is extensively referenced in this article. Böhms and
tion framework of manufacturing execution systems (MES). BPMN Tolman (1993) also define four types of manufacturing operations
compared favorably to flowcharts, petri-nets, Unified Modeling (transformation, transportation, storage and verification) as part
Language (UML) and Systems Modelling Language (SysML) This of their methodology to express a CIM reference architecture, but
effort resulted in extensive modification of BPMN though, to cater unfortunately no further detail is provided. Lastly, the Y-CIM ref-
for the specific requirements of the considered cases. Similarly, Zor erence architecture of Scheer (1994) provides extensive guidance
et al. (2011) present BPMN extensions for manufacturing processes, on the development and modelling of enterprise architecture for
but these are again specific to the single case study. a factory. These studies provide valuable insight and examples on
It is clear that BPMN is a candidate for manufacturing process the definition of manufacturing enterprise architecture, but none of
modelling and enactment. It has been considered from various per- them attempt to construct a comprehensive list of manufacturing
spectives, including as the formal execution semantic for an MES operations.
and as a modern replacement for IDEF3. However, these consider- The widely adopted international standard series IEC
ations are somewhat ad-hoc and disparate. This research aims to 62264:2013 (IEC, 2013) provides standardized terminology
give structure to the adoption of BPMN for manufacturing, by pro- and ontology for the manufacturing domain (Chen, 2005). It advo-
viding a set of process model fragments that practitioners can use cates a functional hierarchy to classify the activities performed in
to model executable manufacturing process and, perhaps, lay the a manufacturing enterprise. Level four of the functional hierarchy
foundation for further research and development in the use of BPM is concerned with business management, including the resource,
in manufacturing. financial and supply chain management functions. Level three
Two conclusions can be drawn from these research efforts: 1) is named manufacturing operations management and refers to
There is precedent for the development of process model fragments the work flow control to produce the desired products (Chen,
to assist with the modelling of processes, and 2) there is interest in 2005). All material processes are situated in level three of the
the more goal-oriented process modelling, giving process partici- hierarchy, as those are the processes that directly contribute
pants more autonomy to pursue those goals. This fits well with the value to the product. Level three is therefore the focus and scope
current rise of autonomous machines and robots seen in the man- of this research. The following four categories of manufacturing
ufacturing industry. The research presented in this article does not operations are included in level three of the functional hierarchy
contradict the cited studies, but rather tries to promote the use of (Chen, 2005):
a single notation to model manufacturing operations and business
processes. Production operations: the functions that convert raw materi-
als, energy, and in-formation into products, with the required
quality, safety, and timeliness.
4. Development of manufacturing process model Inventory operations: coordinates, directs, controls, and tracks
fragments inventory and material movement within manufacturing oper-
ations.
This research aims to provide prescriptive knowledge on the use Maintenance operations: the functions that maintain the equip-
of business process modelling to automate manufacturing oper- ment and tools to ensure their availability for manufacturing and
ations processes. The prescriptive knowledge is presented in the ensure scheduling for periodic or preventive maintenance.
form of model fragments that are intended to be used as building Quality operations: coordinates, directs, and tracks the functions
blocks that can be combined into executable process specifications. that test materials and equipment to measure and verify quality
This section presents the development of the model fragments, fol- measures.
lowing a three-step approach. First, a catalogue of manufacturing
operations is constructed from scientific literature and interna- While these four categories provide a good overview of man-
tional standards. This catalogue represents a relatively complete list ufacturing operations, additional detail is needed to identify the
of operations typically found in manufacturing processes. Second, concepts necessary to create process models. Unfortunately, IEC
each entry in the catalogue of manufacturing operations is decom- 62264:2013 (IEC, 2013) does not provide additional detail and a sin-
posed into concepts to guide development of model fragments. gle catalogue of manufacturing operations does not exist, because
Starting with inventory operations, Langford (2007) lists the

following four logistics functions: packaging, materials handling,
warehousing and storage, and transportation. In comparison,
Ghiani et al. (2013) use the term internal logistics and define it
as the activities carried out in the production plants, consisting of
the following: receiving and storing materials, collecting from the
warehouse to feed the production lines, moving the semi-finished
goods up to packaging, and finally storing the finished product.
Packaging, handling, storage and transportation (PHS&T) is a group-
ing that is widely adopted and used by logistics support engineers
(Defense Acquisition University, 2011; INCOSE, 2015) and com-
pares favorably to the four functions of Langford (2007). Thus,
these four activities will form the first level of decomposition under
inventory operations.
Maintenance and quality operations are often grouped together
in literature. It may even be difficult to determine whether an
operation is considered a maintenance or quality operation. For
example, testing can be performed on equipment or the product.
We apply this differentiation to separate the two operation groups,
as advocated by the category definitions of IEC 62264:2013 (IEC,
2013): operations related to equipment are grouped under mainte-
nance operations, while process and product related operations are
grouped under quality. Then, the Integrated Product Support Ele-
ment Guidebook (Defense Acquisition University, 2011) is queried
to define the detail of those operations. Maintenance is subdivided
into modifications, corrective, and preventive maintenance. Qual-
ity operations differentiates between process and product related
operations. The resulting hierarchy of manufacturing operations,
showing the first three levels of decomposition, is shown in Fig. 2.
The next level of decomposition is again pieced together from
several sources. For production operations, two sources (Groover,
2011; NA 152-06-10 AA National Committee, 2003) provide exten-
sive detail, including descriptions and example processes. Table 1
lists the ten operation types of production operations, each with a
description and its source.
Predictably, inventory operations prove difficult to delineate as
a catalogue. Ray (2008) distinguishes between three types of pack-
aging: shop, bulk and shipping containers. Conversely, Groover
Fig. 2. First three levels of the catalogue of manufacturing operations.
(2011) distinguishes between containers used to hold individual
items and equipment used to make up unit loads. For the purposes
scholarly work tends to focus on a subset of operations. Therefore, of manufacturing processes, we will apply those two categories:
a catalogue of manufacturing operations with definitions must be placing a single item in a container and placing multiple items in a
constructed from several sources, with good representation as a container (unitizing). These two types of packaging can be inversely
goal, rather than absolute completeness. applied for unpacking of containers.
Starting with ‘production operations’, the hierarchy of Groover Ray (2008) also extensively discusses manual and robotic
(2011) is highly cited and detailed. This catalogue is structured handling. The following five handling activities are identified:
according to the nature of the operations, with a first differentiation preparatory, feeding, positioning, manipulating and removing.
between shaping and non-shaping operations. Shaping operations Regarding storage, we distinguish between two types based on the
are then further subdivided between operations that conserve the purpose: buffering is intended to synchronize the flow of mate-
mass of materials, reduce the mass of materials, or join multiple rial between work centers that may have unequal throughput,
parts to form a new shape. Non-shaping production represents while preservation is intended to hold materials and products until
the type of operations that improve the surface of the material needed (Defense Acquisition University, 2011).
or enhances the material properties, without altering its geom- For transport, Stock and Lambert (2001) identifies the following
etry. In pursuit of good representation, the hierarchy of Groover five modes: road, rail, air, water, and pipeline. Their perspective was
(2011) is complemented with DIN 8580:2003−09 (NA 152-06-10 one of intra-company transport though, instead of inter-factory
AA National Committee, 2003) and Todd (1994). The resulting pro- transport. This breakdown is quite common for transport analysis
duction operations is shown at the top of Fig. 2. or optimization projects, such as Davidsson et al. (2005). Similarly,
Unfortunately, well documented and structured lists of inven- the European Union distinguishes between six modes by differ-
tory, maintenance and quality operations prove to be rather elusive. entiating between sea and inland waterways. The United Nations
Thus, several authoritative sources must be consulted to complete has a similar approach and differentiates between seven modes
the catalogue of operations. The referenced sources were selected of transport: Maritime, rail, road, air, multi-model, fixed installa-
for clarity and comprehensiveness. Every effort is made to con- tion and inland water transport (Centre for Trade Facilitation and
sult as many sources as possible, but manufacturing operations can Electronic Business, 2001). Fixed installation transport in this case
be described in any number of ways, so the goal is to establish a includes pipe and cable transport, such as for petroleum and elec-
single sensible and practical list, rather than attempt to address trical power. For the second level of aggregation of transport, we
contradictions and ambiguity in the literature. combine the equipment classification of Ray (2008) with the modes
Table 1
Operation types and descriptions of production operations, with corresponding references.
Operation group Operation type Description Reference
Casting and Creation of an initial shape from the molten, gaseous or formless solid DIN 8580
Mass-conserving shaping
molding state.
Forming The three-dimensional or plastic modification of a shape while DIN 8580
retaining its mass and material cohesion.
Material removal Remove excess material from the starting workpiece so that the Groover, 2011
Mass-reducing shaping
resulting shape is the desired geometry.
Separating Detaching or dividing work pieces from each other. DIN 8580
Permanent joining Form a joint between components that cannot be easily disconnected. Groover, 2011
Assembly
Mechanical Fasten two (or more) parts together in a joint that can be disassembled Groover, 2011
fastening if needed.
Heat treatment The application of thermal energy to enhance the properties of the DIN 8580
work material, without altering its shape or mass.
Non-shaping
Cleaning Processes that remove soils and contaminants that result from Groover, 2011
previous processing or the factory environment.
Surface coatings Application of a thin layer of material to the exterior surface of the Groover, 2011
work part.
Surface treatments Mechanical and physical operations that alter the part surface in some Groover, 2011
way, such as improving its finish or impregnating it with atoms of a
foreign material to change its chemistry and physical properties.
Table 2
Operation types and descriptions of inventory operations, with corresponding references.
Process group Operation type Description Reference
Individual packaging Single product is inserted into a container for transport or storage. Groover, 2011
Packaging
Unitizing (a.k.a. Multiple parts or products are inserted into a single container for transport Groover, 2011
Containerization) or storage.
Preparatory handling Bringing materials closer to the workplace and preparing the machine. Ray, 2008
Feeding Placing or directing materials closer to workplace or point of use. Ray, 2008
Handling Positioning Orienting materials in exact location, placing into fixture, jig or machine. Ray, 2008
Manipulating Handling of materials during actual manufacturing operation. Ray, 2008
Removing Taking material out of workplace, such as taking out of jig, fixture etc. Ray, 2008
Buffering Static or slow holding of materials, parts or products (buffering, queueing, IPS element
Storing
etc.) in preparation for further processing. handbook
Preserving Static holding of materials, consumables, parts or products until retrieved IPS element
for production or distribution. handbook
Vehicular transporting Directed physical movement of materials, consumables, parts or products Stock and Lambert,
Transporting
aboard vehicles (automated guided vehicle, forklift, etc.) 2001
Fixed installation Fixed, point-to-point physical movement (conveyor, cable, piping, etc.) of Stock and Lambert,
transporting materials, consumables, parts, products or energy. 2001
of transport. Road, rail, air and water are treated as vehicular trans- standard (Object Management Group, 2014) provides execution
port, while conveyors, pipelines, cables and cranes are part of the semantic for activities, gateways, events and sequence flow. This
fixed installation category. Table 2 shows the resulting operation subset of executable BPMN concepts is elaborated in detail by
types and descriptions for inventory operations. Dijkman and Van Gorp (2010). Therefore, the following object-
Maintenance and quality operations are again treated together. types of the BPMN2.0 meta-model are included: activity, gateway,
A single authoritative source, the INCOSE Systems Engineering control flow, and message flow. In the interest of clarity, the fol-
Handbook (INCOSE, 2015) is used to define the lowest level of lowing object-types are not included: data objects, data stores, and
detail of maintenance operations. Table 3 shows the operation annotations. The focus on the flow of work, as opposed to the flow of
types and descriptions for maintenance operations. The source col- information, matches well with the physical nature manufacturing
umn is omitted because only one source is used for all maintenance processes.
operations. The process model fragments are created by translating the con-
Table 4 shows the operation types and descriptions of quality cepts of manufacturing operations (see Section 4.1) into model
operations. Again, the source column is omitted, because all oper- elements (activities, events, gateways and sequence flow). We
ation types and descriptions are based on the systems engineering apply the five rules of process progress defined by Hewelt and
handbook (INCOSE, 2015). Weske (2016), with slight adjustments to account for the physi-
cal nature of manufacturing processes. These rules are translated
4.2. Manufacturing operations as a set of concepts for our purposes and summarized as the following:
This subsection is included in the interest of transparency 1 Activity start: a fragment is instantiated as part of the process
and reproducibility. An intermediate step in the development model, but the activity is only started once all preconditions are
is explained, instead of deriving the process model fragments satisfied. Activity pre-conditions are modelled as one or more
directly from the catalogue of manufacturing operations presented events in our fragments. Importantly, this also allows activities
in Section 4.1. The intermediate step entails the decomposition of to be performed multiple times in the case of repetitive tasks or
manufacturing operations into concepts that must be represented rework.
with BPMN2.0. As this research is primarily concerned with the 2 Activity termination: an activity is ended once all post-conditions
design of executable manufacturing process models, only concepts are satisfied. Again, post-conditions are modelled as events in
related to execution semantic will be considered. The BPMN2.0 our fragments. Activity termination does not remove the frag-
Table 3
Operation types and descriptions of maintenance operations, extracted from the INCOSE Systems Engineering Handbook (INCOSE, 2015).
Process group Operation type Description
Repairing An activity which returns the capability of an asset that has failed to a level of performance equal to, or
Corrective maintenance
greater than, that specified by its functions, but not greater than its original maximum capability.
Replacing A maintenance task to replace a component when it has failed or deteriorated to the point where
system performance is outside specified parameters.
Servicing An activity which returns the capability of an asset that has not failed to a level of performance equal
Preventive maintenance to, or greater than, that specified by its functions, but not greater than its original maximum capability.
Scheduled replacing A maintenance task to replace a component at a specified, pre-determined frequency, regardless of
the condition of the component at the time of its replacement.
Condition monitoring The use of specialist equipment to measure the condition of equipment to assess whether it will fail
during some future period.
Sustaining An activity which extends the expected life of an asset beyond its original expected useful life.
System modifications
Upgrading An activity which enhances the capability of an asset beyond its original maximum capability.
Table 4
Operation types and descriptions of quality operations, extracted from the INCOSE Systems Engineering Handbook (INCOSE, 2015).
Process group Operation type Description
Process measuring An activity to determine the value of a conformance or performance quantity of a process.
Equipment calibration Set or adjust a machine or tool according to product and production requirements.
Process quality control
Equipment inspection Conformity evaluation by observation and judgement accompanied as appropriate by measurement,
testing or gauging.
Equipment testing An activity in which a system or component is executed under specified conditions, the results are
observed or recorded, and evaluation is made of some aspect of the system or component.
Product inspection A static analysis technique that relies on visual examination of development products to detect errors,
Product quality control violations of development standards, and other problems.
Product measuring An analysis technique that relies on the use of tools to determine the physical conformance of a
material or product.
Product testing Determination of one or more performance quantities, according to a procedure.
ment from the instantiated model, because the model must allow Material: The matter from which a product is made.
rework or repetitive tasks. Work piece: A single part or product currently undergoing oper-
3 Event occurrence: event occurrences are used to indicate the ation.
satisfaction of conditions and, therefore, how different model
fragments are linked to each other. Each inventory operations can also be decomposed into input,
4 Gateway behavior: this rule applies the XOR, AND and OR gate- transformation and output. Table 6 shows the results of this
ways of BPMN2.0 as specified in the execution semantic (Object decomposition, making use of the same terminology as with pro-
Management Group, 2014). duction operations. Notably, only packaging operations transform
5 Fragment termination: Instantiations of fragments are termi- the inputs into a different output. All other inventory operations
nated along with the termination of process instantiations. This involve activities that move, manipulate or hold items, without
represents the only departure from Hewelt and Weske (2016), changing it in any way. Furthermore, the five types of handling
caused by the difference between design-time and run-time operations are collapsed into a single entry in Table 6, because there
compilation of process models. was no discernible difference in terms of input, transformation and
output.
Maintenance operations are difficult to describe in terms of
input, transformation and output, because of uncertainty regard-
By applying these five rules, we infer that model fragments are ing the activities involved. Most notably, the two modification
bound by pre- and post-conditions, modelled as start events and operations, sustaining and upgrading, can involve multiple activi-
end events, respectively. Between the start and event events, each ties, perhaps even performed by multiple people. Comprehensive
fragment must contain the activities to be performed. The activi- maintenance jobs may even be planned and managed as projects.
ties can be supplemented by gateways to enforce process logic as Nevertheless, the inputs and outputs can be inferred from the
necessary. Additionally, the pre- and post-conditions are used to descriptions in Table 3 and the transformation descriptions in
combine fragments into executable process models. For example, Table 7 are not limited to single activities.
the application of mechanical force during a forming operation is an Lastly, quality operations are also decomposed and shown in
activity performed by an assigned actor, whether human, machine Table 8. Predictably, the input and output of each quality oper-
or a combination of several actors. The forming activity is condi- ation does not change, because these operations involve various
tional on the arrival or availability of input material, represented verifications, rather than transformation of materials or products.
by a start event. The forming activity is ended once the work piece
is shaped, represented by an end event. Therefore, it is likely that 4.3. Manufacturing operations represented as process model
an inventory operation (e.g. transportation or handling) will pre- fragments with BPMN2.0
cede the forming operation to make the work piece available at the
forming station. This section presents the interpretation of process concepts,
Each operation type is described as inputs, activities and out- presented in Section 4.2, as BPMN2.0 model fragments. The cat-
puts, based on the descriptions from the literature. Table 5 shows alogue of fragments is the contribution of this research as it can be
the results of the decomposition for production operations. In the used as building blocks to model executable manufacturing pro-
interest of consistency and clarity, the following two terms are used cesses. The combination of several fragments to model a process
to refer to inputs and outputs: is illustrated in Section 5.1 and evaluated in Section 6.1. Notably,
Table 5
Production operations decomposed into input, activity and output elements.
Operation type Input(s) Activity(ies) Output(s)
Casting and molding Unit of formless material. Apply energy to melt or evaporate material, pour Shaped unit of material in solid state.
material into mold and allow to solidify.
Forming Unit of material in solid state. Apply mechanical force to change shape of Shaped unit of material in solid state.
material.
Material removal Unit of material in solid state. Apply mechanical energy to remove some of the Unit of material with reduced mass
material. AND removed material.
Separating Single unit of material in solid state. Apply mechanical energy to divide a unit of Multiple work pieces AND excess
material into individual work pieces. material.
Permanent joining Multiple work pieces. Apply thermal or chemical energy to permanently Single work piece.
attach two or more work pieces.
Mechanical fastening Multiple work pieces AND fastening Use fastening material to fasten multiple pieces Single work piece.
material. together.
Heat treatment Single work piece. Apply thermal energy enhance work piece Single work piece (enhanced).
properties.
Cleaning Single work piece. Apply mechanical or chemical energy to remove Single work piece (cleaned).
unwanted particles.
Surface coating Single work piece. Deposit a chemical material on the surface of the Single work piece (coated).
work piece.
Surface treatment Single work piece. Apply mechanical energy to enhance surface Single work piece (enhanced).
integrity.
Table 6
Inventory operations decomposed into input, activity and output.
Individual Single work item or unit of material. Insert single item into a container. Package containing single item.
packaging
Multiple packaging Multiple work items or units of Insert multiple items into a single container. Package containing multiple items.
material.
Handling (all five Single work item OR unit of material. Manipulate or move item within a single work unit Single work item OR unit of material.
types) or storage area.
Buffering Single or multiple work piece(s) OR Hold work piece(s) or package(s) until some Single or multiple work piece(s) OR
package(s). condition is met. package(s).
Preserving Single or multiple work piece(s) OR Hold work piece(s) or package(s) until retrieved. Single or multiple work piece(s) OR
package(s). package(s).
Vehicular Single or multiple work piece(s) OR Directed movement of work piece(s) or package(s) Single or multiple work piece(s) OR
transporting package(s). aboard a vehicle. package(s).
Fixed installation Single or multiple work piece(s) OR Ongoing, point-to-point movement of material or Single or multiple work piece(s) OR
transporting package(s). work pieces. package(s).
Table 7
Maintenance operations decomposed into input, activity and output.
Repairing Non-conforming asset. Restore capability of a failed asset to within its Repaired asset
specification.
Replacing Non-conforming asset AND Exchange a component of an asset. Conforming asset AND replaced part.
replacement part.
Servicing Conforming asset. Restore the capability of an asset to within Serviced asset.
specification.
Scheduled replacing Asset AND replacement part. Exchange a component of an asset at a Asset AND replaced part.
pre-determined time.
Condition monitoring Asset AND measurement equipment. Measure the condition of equipment. Asset AND measurement equipment.
Sustaining Asset. Extend the expected life of an asset beyond its Asset with extended life.
original expected useful life.
Upgrading Asset. Enhance the capability of an asset beyond its Asset with enhanced capability.
original maximum capability.
Table 8
Quality operations decomposed into input, activity and output.
Process measuring Manufacturing process. Determine the value of a conformance or Measured manufacturing process.
performance quantity of a process.
Equipment calibration Equipment. Set or adjust a machine or tool according to Calibrated equipment.
product and production requirements.
Equipment inspection Equipment. Inspect equipment for conformity. Inspected equipment.
Equipment testing Equipment. Test some aspect of the system or component. Tested equipment.
Product inspection Product. Visual examination of a product. Inspected product.
Product measuring Product. Determine the physical conformance of a material Measured product.
or product.
Product testing Product. Determine one or more performance quantities, of Tested product.
a product.
Table 9
Model fragments of production operations.
these process fragments are intended to give the expected execu- will be abundantly available at the place of application. For exam-
tion behavior, in addition to accurate representation of the process ple, coating operations are often performed by submerging a work
for understandability and communication purposes. piece in the coating material. In such a case, it does not make sense
Similar to Hewelt and Weske (2016), we note that the model to model the coating material as an inflow. As a result, the four
fragments are usually straight-forward, but their interplay allows non-shaping operations can be represented with the same process
for complex behavior. The fragments should be considered building model and are therefore listed as a single entry in Table 9.
blocks that helps to accelerate the modelling of complex pro- Inventory operations continues to be an interesting challenge.
cesses. The model fragments are again presented according to the Several of the inventory operations fragments, most notably pack-
categories of manufacturing operations, starting with production aging and storing operations, require advanced process modelling
operations in Table 9. constructs. Multiple packaging is modelled as a multi-instance task,
It can be argued that the permanent joining operation requires providing the repetitive behavior of inserting multiple items into
joining material, analogous to the fragment for mechanical fas- a container. Buffering makes use of an intermediate event to place
tening. However, permanent joining is typically achieved with the process in a holding state until some predefined condition is
continuous or standardized material, such as adhesive or welding met. Such a condition may be based on queuing logic or simply
gas. Conversely, mechanical fastening often involves discrete parts, time duration. Preserving uses a similar hold function, but the trig-
such as bolts and widgets. Thus, fastening material is shown as a ger to continue processing is a signal or message instead. Table 10
distinct inflow, whereas joining material is assumed to be present shows the process fragments for inventory operations.
at the workstation. The model fragment for fixed installation transport is quite
A similar assumption is made with all four non-shaping oper- unique. It is the only fragment that necessitates time passage on
ations, i.e. heat treatment, cleaning, surface coatings, surface a connecter, instead of an activity. It is modelled this way, because
treatment. These operations may require production consumables, the transport operation is not assigned to, and therefore not per-
such as coating materials, but it is assumed that such consumables formed by, a process participant. For example, a conveyor belt that
Table 10
Model fragments of inventory operations.
transports parts from one point to another is not assigned a task Except for fixed installation transporting, manufacturing oper-
to perform the transportation. The belt simply rolls, carrying the ations can be accurately represented with BPMN2.0. Thus far, the
parts with it. A conditional event is used to indicate that “task n” analysis is confined to individual manufacturing operations that
can only commence once the material arrives, but this still doesn’t must be used in conjunction to model real manufacturing pro-
capture the elapse of time. A timer event can’t be used, because it is cesses.
not an accurate representation of the process behavior. The timer
event is used to indicate a fixed duration or a future timestamp. In 5. Implementation of process model fragments
the case of fixed installation transporting, time elapses as a result of
material movement. “Task n” will initialize upon the arrival of the The model fragments presented in Section 4.3 must be imple-
material, not after the elapse of time. Therefore, this is considered mented to deliver the expected process execution behavior.
a deficiency, because time passage on a connector is not supported Implementation, in this case, entails two parts. First, the model
by BPMN2.0, but the conditional event can be used to circumvent fragments represent relatively small portions of an actual manufac-
the problem if it is possible to track when the material arrives. turing process and must therefore be combined to model a practical
The third category of operations, namely maintenance opera- process. Secondly, the resulting process model must be subjected
tions, are modelled as business processes and presented in Table 11. to a process management system to enact the process. This section
All seven maintenance operations can be modelled without diffi- details the implementation and is divided into the same two parts.
culty, albeit not without uncertainty. The uncertainty is due to the
extensible nature of maintenance work, especially work involv-
ing equipment modification. For example, the upgrading operation 5.1. Design of manufacturing processes as an aggregation of
is modelled as a single task in Table 11, but upgrading a piece of model fragments
equipment may involve several tasks, perhaps even performed by
multiple people. Significant maintenance work is typically man- To guide the combination of fragments into useful manufactur-
aged as a project, subject to planning of the tasks to be performed. ing processes, we again draw on the work of Hewelt and Weske
The process fragments presented in Table 11 represent single (2016). Although Hewelt and Weske (2016) demonstrates run-time
maintenance tasks that may be duplicated or combined for more combination of fragments into process models, the rules that gov-
significant maintenance jobs. ern fragment combination can be equally applied for design-time
Lastly, quality operations are also modelled as process frag- combination of fragments. Therefore, the pre- and post-conditions
ments and presented in Table 12. The seven quality operations of the model fragments, represented by start and event events, are
can be conveniently grouped into process, equipment and prod- used to link fragments to each other. The difference is that in the
uct related operations. These groupings allow us to use only three work of Hewelt and Weske (2016) the initiation and termination
process fragments for quality operations. Equipment calibration, of activities are conditional on a specific data object entering a pre-
inspection and testing can be modelled as a single fragment and the defined lifecycle state. In our work, due to the material nature of
same can be done for product inspection, measuring and testing. manufacturing processes, we swap the initiation and termination
conditions to the state of a physical object, such as the arrival of a
Table 11
Model fragments of maintenance operations.
Table 12
Model fragments of quality operations.
Fig. 3. Sand casting process according to Groover (2016).
work piece at the work cell. Furthermore, we do not apply run-time in Section 4.2. Fig. 3 shows a portion of an illustrative sand-casting
compilation of model fragments, but rather opt for design-time process with the following three model fragments:
compilation. This is a conscious decision, that has two advantages:
A mechanical fastening fragment used to represent the joining

The formal semantic that dictates fragment combination is
of the core and mold.
less important, because the process modeler can apply expert
A casting and molding fragment that represents the melting,
knowledge during modelling.
pouring and solidification of material.
The process modeler has more freedom to include scenario-
A handling operation fragment to remove the sand mold and
specific elements, such as time-out events or compensation for
expose the work piece.
non-conformances.
To explain how fragments are combined into process models, a To extend the example further, Fig. 4 shows a sand-casting pro-
sand-casting process is used as an example. Any casting process is cess with some ancillary operations from Groover (2016). In this
centered around the pouring of liquid metal and allowing it to solid- process, it is shown that ‘mold making’ and ‘core making’ are inter-
ify, but sand-casting also requires preparation of the mold and core jected mid-way into the ‘casting’ operation (between the melting
(as opposed to permanent mold casting). Additionally, any cast- and pouring activities). This would represent a violation of the cast-
ing process is surrounded by several other operations to prepare ing model fragment, but it does not change any process logic. It is
and remove material. The model fragments can be used to model simply an efficiency enhancement in the process, as it is under-
the sand-casting process using BPMN2.0. The fragments are com- stood that the material melting can start before the mold and core
bined by stitching together the pre- and post-conditions described are positioned.
Fig. 4. Sand casting process according to Groover (2016).
Fig. 5. Sand casting process of Groover (2013) modelled as a combination of model fragments.
Fig. 5 shows a model of a sand-casting process, as a combina- cesses using business process modelling. Hence, the process models
tion of manufacturing process fragments with BPMN2.0. The BPMN derived from the fragments must be subjected to a process man-
model shown in Fig. 5 is certainly more complicated than the dia- agement system. Furthermore, the process management system
gram shown in Fig. 4, but also less ambiguous. In Fig. 4 it is not clear must be connected to the resources that will perform the activi-
whether all inflows to the pouring activity must be simultaneously ties specified in the process model, to relay commands and receive
active or only a single inflow can trigger the pouring activity. In data from the factory floor. Such an arrangement represents a com-
Fig. 5 this relationship is clarified by an AND-gateway, clearly indi- plex information system, as it provides functionality for process
cating that mold and core must be ready, and the material must enactment, control of individual resources, coordination between
be melted, before pouring can commence. A realistic manufactur- multiple resources and, of course, communication between the pro-
ing process can thus be accurately represented with BPMN2.0, as a cess engine and resources. In this section we provide guidance, still
combination of model fragments. in the form of prescriptive knowledge, on the realization of such
an information system. However, this section is kept very brief, as
5.2. Enactment of manufacturing operations processes it isn’t the focus of this article. Instead of a lengthy discussion, we
provide an overview with references to our other work focused
The purpose of the model fragments presented in Section on the development and realization of a manufacturing operations
4.3 is to support automation of manufacturing operations pro- management system.
Fig. 6. Software aspect of the HORSE System (Grefen et al., 2016).
The separation between process control and resource control nologies include human-robot collaboration, augmented reality,
is an important consideration in the design of a manufacturing autonomous guided vehicles and teaching robots by demonstra-
operations management system. Manufacturing operations pro- tion. The primary outcome of the HORSE Project is a modular
cesses are comprised of multiple unitary activities, as illustrated information system that integrates the emerging technologies with
by the set of model fragments presented in Section 4.3. Process a BPMS. The HORSE System architecture is shown in Fig. 6, at
control is concerned with the initiation of activities according to a fairly high level of aggregation. In the interest of brevity, only
the logic encoded in the process model. As those activities are per- process-related aspects are discussed here, as the system design
formed by resources, process control is also concerned with the is substantially covered in our other work (Erasmus et al., 2020;
assignment of resources to activities. Once a resource, or team Grefen et al., 2017).
of resources, are assigned to an activity, the actions performed The HORSE System has a layered architecture pattern, with
by those resources are governed by resource control. The separa- a global orchestration layer and a local control layer (Grefen
tion between process control and resource control is particularly et al., 2016). The global layer includes functionality to design and
useful in the manufacturing domain, because of resource diver- enact processes and the local layer is responsible for control of
sity. A factory floor may be populated by various of autonomous individual resources, in the form of machine control systems, aug-
and human-operated machines. Furthermore, versatile robotics are mented reality and graphical user interfaces. The global layer is
increasingly affordable, human-robot collaboration is now viable in based on Camunda BPM, an open-source BPMS, that provides pro-
specialized cases and emerging technologies (e.g. the Internet-of- cess modelling and enactment functionality. The following three
Things and augmented reality) are also appearing on the factory enhancements were made to Camunda BPM to account for the vast
floor (Kang et al., 2016; Lu, 2017; Qu et al., 2016; Tao and Qi, 2018). differences between resources on the factory floor:
This distinction between process control and resource control is
perfectly aligned to the functional hierarchy of IEC62264:2013 (IEC, 1 A new resource design module (named ‘Agent Design’ in Fig. 6)
2013), as introduced in Section 4.1. Level 3 of the hierarchy refers to specify the capabilities of resources.
to the coordination and monitoring of the workflow across the fac- 2 A new advanced resource assignment module (as part of ‘Global
tory floor, while level 2 represents the control systems of individual Execution in Fig. 6) to select the most appropriate resource (or
resources (or minds in the case of human resources). In fact, we team of resources) to perform a task.
use this distinction as a primary motivator for the use of business 3 A new ‘Global Awareness’ module that performs complex event
process management in manufacturing operations (Erasmus et al., processing to identify patterns and trends on the factory floor,
2018). such as a gradual decrease in productivity or process quality.
The process management system used to enact the designed
manufacturing processes was developed as part of the HORSE As a high-level description, individual activities are designed
Project (www.horse-project.eu), a multi-year research and inno- with the HORSE Config Local subsystem. The HORSE Design Global
vation project funded by the European Commission under the subsystem is then used to specify processes and resource capabil-
Horizon 2020 program. The project aims to make advanced man- ities. Those specifications are then used by the HORSE Exec Global
ufacturing technology more accessible to small and medium subsystem to enact the processes and coordinate the resource
manufacturing enterprises (Vanderfeesten et al., 2016). These tech- assignment. Once a resource is assigned, the HORSE Exec Local
Fig. 7. Process model for the wiper system inspection and packaging.
subsystem controls the actions of that resource, until the assigned as indicated in Fig. 7. Unfortunately, five of the seven fragments are
activity is completed (or terminated in case of failure). The global handling operations, part of the inventory operations category. As
and local layers of the HORSE System are connected via middle- such, these five fragments are single tasks connected with singu-
ware named OSGi, conceptually represented by the global and local lar inflows and outflows. The remaining two fragments are both
abstraction layers in Fig. 6. This middleware facilitates communi- product inspection fragments, from the quality operations cate-
cation between the layers, by allowing resources to subscribe and gory. Therefore, these two fragments consist of single tasks with
listen for messages with pre-defined identifiers. Importantly, the multiple conditional outflows. The first quality task, named “visual
communication middleware allows machines and robots from var- inspection”, has two potential outflows and the second task, named
ious vendors to collaborate in the same process and driven by the “evaluated detected defect”, has three possible results.
singular process management system. The middleware also allows The process model shown in Fig. 7 includes elements not com-
the resources to communicate with the global layer, to report on posed of any process fragments presented in section 4.3. These
the outcome of activities. elements cater for scenario specific requirements as elicited from
the process owner. Table 13 lists five scenario requirements that are
6. Evaluation translated into seven model elements that do not originate from the
process fragments.
The research is extensively evaluated, with an emphasis on prac- As a detailed example, the inspection task named “evaluate
tical relevance as part of the HORSE Project. The HORSE System detected defect” has two boundary timer events attached to it. This
is implemented and demonstrated at ten factories across Europe. is a requirement, specified by the process owner, that a supervisor
Only one of the demonstrations is included in this article, but addi- must be informed if a detected defect is not evaluated promptly.
tional demonstrations are described in our other work (Erasmus Therefore, a priority message is sent after one minute and the task is
et al., 2020) and all ten cases are elaborated on the HORSE Project interrupted after four minutes. The following elements are present
website (http://horse-project.eu/Experiments). The evaluation is in the process model of Fig. 7 that do not come from one of the
presented in two parts, based on the dual purpose of process mod- fragments in Section 4.3:
els: enactment and comprehension. The first part demonstrates
how a real process at the factory is modelled as a composition A boundary, non-interrupting, conditional event to initiate a task
of process fragments, and then enacted to demonstrate that the to replace the packaging box;
correct process behavior is observed. Secondly, comprehension is A boundary, non-interrupting, conditional event to initiate a task
gauged by assessing the ease with which typical users understand to replace the carton sheet in the box, to allow placing of a new
the process models. layer of products;
A boundary, non-interrupting, timer event that triggers if a
6.1. Assessment of the enactment of aggregated process models detected defect is not evaluated within one minute;
A message end-event to send a priority message if a defect is not
The process execution evaluation is done by modelling the pro- evaluated within one minute;
cess as a combination of the process fragments presented in section A boundary, interrupting, timer event that triggers if a detected
4.3. The models are then used to enact the processes using the defect is not evaluated within four minutes;
HORSE System (introduced in Section 5.2). Successful execution of A boundary, interrupting, conditional event that triggers if there
the manufacturing processes, in commercial factories, is consid- is more than one unevaluated defect in the work cell;
ered adequate evidence for practical use of the model fragments to A termination event that stops process execution if there is more
design executable manufacturing processes. than one unevaluated defect in the work cell;
Fig. 7 shows the process model for the final inspection and pack-
aging of wiper system assemblies. Confidential names are replaced The seven model elements listed in Table 13 do not invalidate
by generic labels, but all other process details are shown as origi- the model fragments of Section 4.3. All seven elements are events
nally captured. This model is composed of seven process fragments, that correspond to specific business rules. All manufacturing oper-
Table 13
Five scenario specific requirements as specified by the process owner.
Requirement Process interpretation Additional model elements
The box that the product is packaged in must Keep track of the number of products in the A boundary, non-interrupting, conditional event that
be replaced when it is full to ensure box and an event to initiate a task to replace triggers when the box is full.
continued production. the packaging box.
A carton sheet must be placed between layers Keep track of the number of products in the A boundary, non-interrupting, conditional event to that
of products to protect the product from current layer and an event to initiate a task to triggers when the layer is full.
transport damage. replace the carton sheet in the box.
Detected defects must be resolved within one Measure the time since the detection of a A boundary, non-interrupting, timer event that triggers if a
minute after detection. defect and send a message if the defect is not detected defect is not evaluated within one minute.
resolved within one minute. A message end-event to send a priority message if a defect
is not evaluated within one minute
The product must be discarded if a detected Measure the time since the detection of a A boundary, interrupting, timer event that triggers if a
defect is not resolved within four minutes. defect and discard the product if the defect is detected defect is not evaluated within four minutes
not resolved within four minutes.
Production must be stopped if there is more Keep track of the number of unevaluated A boundary, interrupting, conditional event that triggers if
than one unevaluated defect in the process. defects in the process and stop process there is more than one unevaluated defect in the work cell.
execution if it exceeds one. A termination event that stops process execution if there is
more than one unevaluated defect in the work cell.
Table 14
Mapping of Prat et al. (2014) and Moody (2009) criteria.
Prat et al., 2014 Moody, 2009
Self-reported competence Expert-novice differences The competence level of the person using the notation.
Simplicity Complexity Management The ability of a visual notation to represent information without overloading the human mind.
Perceptual Discriminability The ease and accuracy with which graphical symbols can be differentiated from each other.
Clarity
Semantic Transparency The extent to which the meaning of a symbol can be inferred from its appearance.
Homomorphism Semiotic Clarity One-to-one correspondence between symbols and their referent concepts.
Level of detail Visual Expressiveness The number of visual variables used in a notation.
Consistency Cognitive Integration Cognitive and perceptual integration of information from different diagrams.
ations of the process correspond to model fragments, as it should be ipants are not necessarily considered as part of the intended
for a catalogue of manufacturing operations. The process fragments audience, because factory workers will often undergo on-the-job
are, by their nature, approximations of common manufacturing training or at least perform tasks without detailed knowledge of
operations. Real processes will invariably have specific require- the process model.
ments that are not common enough to warrant a standardized Process models are not necessarily used as instruction or
fragment. training material. Therefore, it is pertinent to evaluate the compre-
The success of this evaluation is that a very specific process hension of people involved in process improvement and equipment
model was obtained as a composition of the process fragments, installation, to determine whether a business process model is an
in conjunction with scenario specific requirements. Therefore, the accurate and understandable representation of a manufacturing
process fragments are useful for creating a specific, executable pro- process.
cess model. A video of the operational process is available online*, Prat et al. (2014) recommends eight criteria for the evaluation
showing direct evidence of manufacturing operations orchestrated of models: self-reported competence, completeness, simplicity,
by a BPMS, based on a BPMN2.0 process model. Interestingly, clarity, style, homomorphism (fidelity of a model to modelled phe-
this process requires several smart manufacturing innovations, nomena), level of detail, and consistency. The physics of notation
including advanced robotics, augmented reality and human-robot advocated by Moody (2009) is intended to help creation of new
collaboration during the correction of defective products. Such low- notations, but it can also be used to evaluate an existing notation.
level synchronization can’t easily be controlled by a BPMS, because More importantly, Moody (2009) also provides clear descriptions
of slow response times. Instead, we deployed a dedicated con- of the criteria, helping us to create a questionnaire. Table 14 shows
trol system that uses state machine modelling for this task. This the mapping of criteria of Prat et al. (2014) and Moody (2009), with
points to a general lower limit for the process management engine. descriptions.
Actions that require sub-second synchronization can’t dependably We created a form with ten questions and space for comments
be controlled by a BPMS. from the descriptions in Table 14. The form was completed by
18 respondents from 11 different organizations. All 18 respon-
6.2. Assessment of the comprehensibility of aggregated process dents were involved in the modelling of executable manufacturing
models processes. The 11 organizations included four manufacturing com-
panies, four universities and three research organizations. In the
Apart from obtaining correct process behavior, process models interest of transparency, Table 15 presents the full results of the
are also used as a facilitator of comprehension and communication. survey, followed by summarizing graphs in Fig. 8 and Fig. 9.
For this purpose, it is important that the model accurately repre- Fig. 8 shows overwhelmingly positive results for homomor-
sents the process and that it is understandable for the intended phism. The four questions assess the fidelity of a model to a
audience. In the case of manufacturing operations processes, the modelled phenomenon and are directly based on the criteria advo-
intended audience is any party that is involved with process cated by Moody (2009). Similarly, Fig. 9 shows highly positive
improvement or equipment installation. Notably, process partic- results regarding model understandability.
Fifteen respondents also added comments when completing
the questionnaire. The comments reflected the general enthusi-
∗
https://youtu.be/P6XhSHQJj s asm of the respondents, but a few cautionary entries were also
Table 15
Results of the survey regarding comprehension of process models.
Question / Respondent 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
J. Erasmus, I. Vanderfeesten, K. Traganos et al. / Computers in Industry 123 (2020) 103297

What is your competence level (skill, I N N N I N N I E I E I N E E N N N
knowledge and experience) with BPMN2.0?
(Novice; Intermediate; Expert)
One element can represent more than one S S N N S N D N S N A S A N S N N S
manufacturing concept. (Never; Sometimes;
Always; Don’t know)
One manufacturing concept can be A N S N S S D S N A A S S S S S A A
represented with more than element.
(Never; Sometimes; Always; Don’t know)
An element doesn’t represent any N S D N N S D D S N S S D S N S A D
manufacturing concept. (Never; Sometimes;
Always; Don’t know)
A manufacturing concept can’t be represented S S D N N S N A S S S N N D S N N D
with any available element or combination
of elements. (Never; Sometimes; Always;
Don’t know)
I find it easy to tell one model element from 1 2 1 2 −1 1 2 2 2 1 2 1 1 −1 2 1 2 1
another. (Range: completely disagree = -2 to
completely agree = +2)
The notation gives enough information about −1 1 1 2 2 2 1 2 1 1 −1 1 1 −1 1 1 1 −1
the process. (Range: completely disagree =
-2 to completely agree = +2)
I find it easy to understand the meaning of a 1 1 1 1 1 1 0 1 −2 1 2 0 2 1 1 0 0 −1
model element, from its appearance (how it
looks). (Range: completely disagree = -2 to
completely agree = +2)
I find the process models simple enough (not 1 1 1 2 1 −1 −1 2 −1 0 2 1 2 2 0 2 1 0
too complicated). (Range: completely
disagree = -2 to completely agree = +2)
I find it easy to understand the relationship 1 2 2 2 1 1 0 2 1 0 2 2 2 2 1 2 1 0
between different models. (Range:
completely disagree = -2 to completely agree
= +2)
17
Fig. 8. Results of homomorphism related questions.
Fig. 9. Questions and results related to model understandability.
recorded. In the interest of brevity, only a summary of the com- of the respondents supports the argument that the notation is not
ments is discussed here. One respondent remarked that it’s easy to intuitive for manufacturing processes. While all concepts can be
create business models but difficult to create executable models, represented with BPMN, the process fragments are rather obscure
while another respondent appreciates the power of subprocesses and difficult to understand.
to limit the number of elements on a diagram. Three respondents The comment regarding the development of executable mod-
mentioned that BPMN is easy to learn, but also complained about els is also worth discussing. It is clearly more difficult to create
a lack of good learning material. Two respondents found it diffi- executable models, considering the need to control the activities
cult to distinguish manufacturing tasks from each other, because of humans, robots and machines. However, the expressiveness of
the same symbol is used for any task. It is especially difficult to see BPMN helps with the complexity of manufacturing processes. For
which tasks are performed by humans or machines. example, the repetitive nature of some manufacturing activities
Most notably, three respondents found it difficult to relate to can be modelled as multi-instances, as is done for the multiple
the notation, due to a lack of manufacturing specific symbols. They packaging operation, shown in Table 10.
commented that the following concepts can’t easily be represented This approach to the modelling of manufacturing operations
with BPMN: buffers, queues and flow of material. Although those allows extensive flexibility and design freedom. All activities shown
concepts can be captured in a roundabout way, as we found in on a single view of the process do not have to represent similar lev-
Section 4.3, such techniques are not intuitive. els of detail. For example, the “removal of sand mold” and “transport
to storage area” activities shown in Fig. 5 may represent different
7. Discussion amounts of work. “Removal of sand mold” may involve a compli-
cated work instruction, while the transport activity is a singular
The comments from the respondents in the comprehension action. This freedom makes it possible to use multiple instances
evaluation (Section 6.2) raise a few interesting topics. The men- at different levels of process aggregation to accurately capture the
tioned notation deficiencies contradict the set of model fragments repetitive nature of certain manufacturing operations.
(Section 4.3). The flow of physical material was found lacking The use of business process management in manufacturing
in BPMN2.0 (see the fixed installation transporting operation in operations will not replace current practices. Detailed scheduling
Table 10), but the other two deficiencies (buffers and queues) were and resource management will remain as important and expedi-
not identified. In fact, the buffering operation fragment in Table 10 ent for the foreseeable future. The prospect is rather to add an
explicitly makes use of the concept of queuing to release mate- additional perspective that can be used to view and manage the
rial according to a predetermined condition. Thus, the difference activities of the manufacturing system, independent of the location
between the manufacturing process fragments and the experience where those activities happen and who/what is involved in those
activities. The additional perspective is contemplated in anticipa- types, such as continuous production. The same research method-
tion of the continued increase of complexity in factories (Ugarte ology can be applied in other domains related to manufacturing.
et al., 2009). More interestingly, the set of model fragments can be implemented
as a catalogue of operations in a process modelling tool to help users
8. Conclusion design manufacturing process models.
This work is part of a larger effort to develop new manu-
This research demonstrates the use of business process mod- facturing process management techniques and tools. The rise of
elling for manufacturing operations, using a catalogue of model new technologies such as the Internet-of-Things, versatile robotics
fragments that capture the essential elements of operations. The and augmented reality opens new possibilities for operations
research is motivated by the promise of seamless cross-functional management. Instead of locally controlled work cells, a single pro-
process management that can be achieved by using the same cess management system can be used to centrally orchestrate
notation for business management and operations functions. To all activities, including business management and manufacturing
align with the most common industry standards, we have cho- operations activities. Thus, the set of process model fragments pre-
sen IEC62264:2013 for the description of manufacturing operations sented in Section 4.3 represents a significant step towards the
and BPMN2.0 as the process specification language in our work. unification of process modelling and management across all busi-
The catalogue of model fragments is developed and presented in ness functions, driving integration and cooperation.
a structured and reproducible manner. Firstly, a single list of manu-
facturing operations, with accompanying descriptions, is compiled CRediT authorship contribution statement
to represent common activities in manufacturing processes. This
list of manufacturing operations can act as a frame of reference Jonnro Erasmus: Conceptualization, Methodology, Valida-
for further research endeavors. Secondly, the manufacturing oper- tion, Investigation, Writing - original draft, Visualization. Irene
ations are decomposed into concepts that must be represented Vanderfeesten: Methodology, Writing - review & editing, Super-
with BPMN2.0 primitives. Thirdly, the concepts are translated to vision, Project administration, Funding acquisition. Konstantinos
BPMN2.0 and illustrated as process model fragments. Fourth, and Traganos: Software, Validation, Investigation, Writing - review &
lastly, it is shown how the process model fragments can be com- editing, Visualization. Paul Grefen: Conceptualization, Methodol-
bined into practical, executable manufacturing processes. ogy, Writing - review & editing, Supervision, Funding acquisition.
The contribution of this work is the catalogue of manufacturing
process model fragments, presented in Section 4.3. These fragments Declaration of Competing Interest
can be used as building blocks to create executable process models.
This way of working is demonstrated by modelling and enacting The authors report no declarations of interest.
a inspection and packaging process at a factory. While only one
demonstration is included in this article, the same evaluation was Acknowledgements
performed at nine other factories in Europe.
In conclusion, the use of BMPN2.0 to specify executable man- The work described in this paper was part of the HORSE Project
ufacturing processes was successfully demonstrated. The only and has received funding from the European Union’s Horizon
difficulty identified with respect to the modelling of manufacturing 2020 research and innovation program under grant agreement no.
operations is related to the flow of materials. More specifically, the 680734.
time it takes for material to flow. If some physical items must flow
from one location to another between two manufacturing opera- References
tions, BPMN2.0 does not include the primitives to represent the
time that elapses for the material to move – something that is Baumgraß, A., Dijkman, R., Grefen, P., Pourmirza, S., Völzer, H., Weske, M., 2015]. A
software architecture for transportation planning and monitoring in a collabo-
explainable by the fact that the notation was designed for processes rative network. In: Camarinha-Matos, L.M., Bénaben, F., Picard, W. (Eds.), Risks
with an administrative character. We opted for an approximate rep- and Resilience of Collaborative Networks: 16th IFIP WG 5.5 Working Confer-
resentation of this construct, by using a conditional intermediate ence on Virtual Enterprises, IFIP Advances in Information and Communication
Technology. Springer International Publishing, Cham, pp. 277–284.
event to determine when the material arrives at the next work cell.
Berente, N., Vandenbosch, B., Aubert, B., 2009]. Information flows and business pro-
This may not be a perfect representation, but it is considered good cess integration. Bus. Process. Manag. J. 15, 119–141, http://dx.doi.org/10.1108/
enough to avoid the introduction of yet another BPMN extension. 14637150910931505.
Apart from the difficulty with the flow of material, some Böhms, H.M., Tolman, F.P., 1993]. A methodology for expressing CIM reference archi-
tectures. Robot. Comput. Manuf. 10, 131–140, http://dx.doi.org/10.1016/0736-
respondents of the evaluation questionnaire found the notation 5845(93)90035-I.
unintuitive for manufacturing processes, due to a lack of specific Braglia, M., Carmignani, G., Zammori, F., 2006]. A new value stream mapping
manufacturing symbols. This is not surprising, given the novelty approach for complex production systems. Int. J. Prod. Res. 44, 3929–3952,
http://dx.doi.org/10.1080/00207540600690545.
of using a business process modelling notation for manufacturing Brahe, S., 2007]. BPM on Top of SOA: experiences from the financial industry. In:
operations, as well as the focus on physical aspects in manufactur- Alonso, G., Dadam, P., Rosemann, M. (Eds.), Business Process Management: 5th
ing operations design. At this time, we prefer not to speculate on a International Conference, Lecture Notes in Computer Science. Springer Berlin
Heidelberg, Berlin, Heidelberg, pp. 96–111.
potential solution to this problem, but rather relegate it to future Brandimarte, P., Cantamessa, M., 1995]. Methodologies for designing CIM sys-
work, as this will require substantial effort. tems: a critique. Comput. Ind. 25, 281–293, http://dx.doi.org/10.1016/0166-
This research presented in this article is positioned as a 3615(94)00037-Q.
Braun, R., Esswein, W., 2014]. Classification of domain-specific BPMN extensions.
stepping-stone towards more research and practical implementa- In: The Practice of Enterprise Modeling, Lecture Notes in Business Information
tion. The current set of process model fragments utilizes very few Processing, Springer, Berlin, Heidelberg, pp. 42–57, http://dx.doi.org/10.1007/
of the primitives of BPMN2.0. For example, message flow and data 978-3-662-45501-2 4.
Braun, R., Schlieter, H., Burwitz, M., Esswein, W., 2014]. BPMN4CP: design and imple-
items were not explicitly excluded from our assessment, but these
mentation of a BPMN extension for clinical pathways. In: 2014 IEEE International
concepts were not needed for the catalogue of manufacturing oper- Conference on Bioinformatics and Biomedicine (BIBM), IEEE, Belfast, UK, pp.
ations. Such concepts can perhaps be used to provide more precise 9–16, http://dx.doi.org/10.1109/BIBM.2014.6999261.
guidance on the combination of process fragments into practical Cadavid, J., Alférez, M., Gérard, S., Tessier, P., 2015]. Conceiving the model-driven
smart factory. In: Proceedings of the 2015 International Conference on Software
process models. Furthermore, the current research is limited to dis- and System Process, ICSSP 2015, ACM, New York, NY, USA, pp. 72–76, http://dx.
crete manufacturing operations, thus excluding other production doi.org/10.1145/2785592.2785602.
Centre for Trade Facilitation and Electronic Business, 2001]. Codes for Modes of Haarmann, S., Podlesny, N.J., Hewelt, M., Meyer, A., Weske, M., 2015]. Production
Transport, second. ed. United Nations, Geneva, Switzerland. case management: a prototypical process engine to execute flexible business
Chen, D., 2005]. Enterprise-control system integration - an international standard. processes. In: CEUR Workshop Proceedings, Innsbruck, Austria, pp. 110–114.
Int. J. Prod. Res. 43, 4335–4357, http://dx.doi.org/10.1080/00207540500142399. Hanson, J.E., Nandi, P., Kumaran, S., 2002. Conversation support for business process
Cheng-Leong, A., Pheng, K.L., Leng, G.R.K., 1999]. IDEF*: a comprehensive modelling integration. In: Proceedings of the Sixth International Enterprise Distributed
methodology for the development of manufacturing enterprise systems. Int. J. Object Computing Conference, IEEE, Lausanne, Switzerland, pp. 65–74, http://
Prod. Res. 37, 3839–3858, http://dx.doi.org/10.1080/002075499189790. dx.doi.org/10.1109/EDOC.2002.1137697.
Chinosi, M., Trombetta, A., 2012]. BPMN: an introduction to the standard. Comput. Hausman, W.H., Montgomery, D.B., Roth, A.V., 2002]. Why should marketing and
Stand. Interfaces 34, 124–134, http://dx.doi.org/10.1016/j.csi.2011.06.002. manufacturing work together?: some exploratory empirical results. J. Oper.
Chung, P.W.H., Cheung, L., Stader, J., Jarvis, P., Moore, J., Macintosh, A., 2003]. Manag. 20, 241–257, http://dx.doi.org/10.1016/S0272-6963(02)00010-4.
Knowledge-based process management—an approach to handling adaptive Hevner, A.R., March, S.T., Park, J., Ram, S., 2004]. Design science in information sys-
workflow. Knowledge Based Syst. 16, 149–160, http://dx.doi.org/10.1016/ tems research. MIS Q. 28, 75–105.
S0950-7051(02)00080-1. Hewelt, M., Weske, M., 2016]. A hybrid approach for flexible case modeling and exe-
Davidsson, P., Henesey, L., Ramstedt, L., Törnquist, J., Wernstedt, F., 2005]. An anal- cution. In: La Rosa, M., Loos, P., Pastor, O. (Eds.), Business Process Management
ysis of agent-based approaches to transport logistics. Transp. Res. Part C Emerg. Forum. Springer International Publishing, Cham, pp. 38–54, http://dx.doi.org/
Technol. 13, 255–271, http://dx.doi.org/10.1016/j.trc.2005.07.002. 10.1007/978-3-319-45468-9 3.
Decker, G., Barros, A., 2008]. Interaction modeling using BPMN. In: Proceedings of Hildebrandt, T., Mukkamala, R.R., Slaats, T., 2012]. Nested dynamic condition
the 2007 International Conference on Business Process Management, BPM’07, response graphs. In: Arbab, F., Sirjani, M. (Eds.), Fundamentals of Software Engi-
Springer-Verlag, Berlin, Heidelberg, pp. 208–219. neering, Lecture Notes in Computer Science. Springer, Berlin Heidelberg, Tehran,
Defense Acquisition University, 2011. Integrated Product Support Element Guide- Iran, pp. 343–350, http://dx.doi.org/10.1007/978-3-642-29320-7 23.
book. Defense Acquisition University, USA. IEC, 2013]. Enterprise-control System Integration - Part 1: Models and Termi-
Dijkman, R., Van Gorp, P., 2010]. BPMN 2.0 execution semantics formalized as graph nology, 2nd ed. The International Electrotechnical Commission (IEC), Geneva,
rewrite rules. In: Mendling, J., Weidlich, M., Weske, M. (Eds.), Business Process Switzerland.
Modeling Notation. Springer, Berlin Heidelberg, Berlin, Heidelberg, pp. 16–30. INCOSE, 2015]. Systems Engineering Handbook: a Guide for System Life Cycle Pro-
Doumeingts, G., Vallespir, B., Chen, D., 1995]. Methodologies for designing CIM cesses and Activities, fourth. ed. John Wiley & Sons, Inc., Hoboken, NJ.
systems: a survey. Comput. Ind. 25, 263–280, http://dx.doi.org/10.1016/0166- Jalali, A., Wohed, P., Ouyang, C., Johannesson, P., 2013]. Dynamic weaving in aspect
3615(94)00036-P. oriented business process management. In: On the Move to Meaningful Internet
Erasmus, J., Vanderfeesten, I., Traganos, K., Grefen, P., 2018]. The case for unified Systems: OTM 2013 Conferences, Lecture Notes in Computer Science. Springer
process management in smart manufacturing. In: EDOC 2018. Presented at Berlin Heidelberg, Berlin, Heidelberg, pp. 2–20, http://dx.doi.org/10.1007/978-
the 22nd International Enterprise Distributed Object Computing Conference, 3-642-41030-7 2.
IEEE, Stockholm, Sweden, pp. 218–227, http://dx.doi.org/10.1109/EDOC.2018. Kamara, J.M., Anumba, C.J., Evbuomwan, N.F.O., 2000]. Process model for client
00035. requirements processing in construction. Bus. Process. Manag. J. 6, 251–279,
Erasmus, J., Vanderfeesten, I., Traganos, K., Keulen, R., Grefen, P., 2020]. The HORSE http://dx.doi.org/10.1108/14637150010325462.
project: the application of business process management for flexibility in smart Kang, H.S., Lee, J.Y., Choi, S., Kim, H., Park, J.H., Son, J.Y., Kim, B.H., Noh, S.D., 2016].
manufacturing. Appl. Sci. 10, 4145, http://dx.doi.org/10.3390/app10124145. Smart manufacturing: past research, present findings, and future directions. Int.
Fleischmann, A., Kannengiesser, U., Schmidt, W., Stary, C., 2013]. Subject- J. Precis. Eng. Manuf. Technol. 3, 111–128, http://dx.doi.org/10.1007/s40684-
oriented modeling and execution of multi-agent business processes. In: 2013 016-0015-5.
IEEE/WIC/ACM International Joint Conferences on Web Intelligence (WI) and Keyte, B., Locher, D., 2004. The Complete Lean Enterprise: Value Stream Mapping
Intelligent Agent Technologies (IAT), IEEE, Atlanta, GA, USA, pp. 138–145, http:// for Administrative and Office Processes, 1st ed. Productivity Press, New York,
dx.doi.org/10.1109/WI-IAT.2013.102. http://dx.doi.org/10.1201/b16650.
Fleischmann, A., Schmidt, W., Stary, C., 2015]. Subject-oriented business process Kim, C.-H., Weston, R.H., Hodgson, A., Lee, K.-H., 2003]. The complementary use of
management. In: J, vomBrocke, M, Rosemann (Eds.), Handbook on Business IDEF and UML modelling approaches. Comput. Ind. 50, 35–56, http://dx.doi.org/
Process Management 2: Strategic Alignment, Governance, People and Culture, 10.1016/S0166-3615(02)00145-8.
International Handbooks on Information Systems. Springer, Berlin Heidelberg, Kobayashi, T., Tamaki, M., Komoda, N., 2003]. Business process integration as a solu-
Berlin, Heidelberg, pp. 601–621, http://dx.doi.org/10.1007/978-3-642-45103-4 tion to the implementation of supply chain management systems. Inf. Manag.
25. 40, 769–780, http://dx.doi.org/10.1016/S0378-7206(02)00102-7.
Fowler, M., 1997]. Analysis Patterns: Reusable Object Models, the Addison-wesley Langford, J.W., 2007]. Logistics: Principles and Applications, 2nd ed. McGraw-Hill
Series in Object-oriented Software Engineering. Addison Wesley, Menlo Park, SOLE Press series. SOLE Press/McGraw-Hill, New York.
Calif. Lu, Y., 2017]. Industry 4.0: a survey on technologies, applications and open research
García-Domínguez, A., Marcos-Bárcena, M., Medina, I.V., 2012. A comparison of issues. J. Ind. Inf. Integr. 6, 1–10, http://dx.doi.org/10.1016/j.jii.2017.04.005.
BPMN 2.0 with other notations for manufacturing processes. In: The 4th Man- Marengo, E., Nutt, W., Perktold, M., 2018]. Construction process modeling: repre-
ufacturing Engineering Society International Conference, AIP Publishing, Cadiz, senting activities, items and their interplay. In: Weske, M., Montali, M., Weber,
Spain, pp. 593–600, http://dx.doi.org/10.1063/1.4707613. I., vom Brocke, J. (Eds.), Business Process Management. Springer International
Gerber, T., Theorin, A., Johnsson, C., 2014]. Towards a seamless integration between Publishing, Cham, pp. 48–65, http://dx.doi.org/10.1007/978-3-319-98648-7 4.
process modeling descriptions at business and production levels: work in Meyer, A., Herzberg, N., Puhlmann, F., Weske, M., 2014]. Implementation frame-
progress. J. Intell. Manuf. 25, 1089–1099, http://dx.doi.org/10.1007/s10845- work for production case management: modeling and execution. In: 18th
013-0754-x. International Enterprise Distributed Object Computing Conference, IEEE, Ulm,
Ghiani, G., Laporte, G., Musmanno, R., 2013]. Introducing logistics. In: Introduction Germany, pp. 190–199, http://dx.doi.org/10.1109/EDOC.2014.34.
to Logistics Systems Management. John Wiley & Sons, Ltd, Chichester, UK, pp. Moody, D., 2009]. The “Physics” of notations: toward a scientific basis for construct-
1–43, http://dx.doi.org/10.1002/9781118492185.ch1. ing visual notations in software engineering. Ieee Trans. Softw. Eng. 35, 756–779,
Grefen, P., Brouns, N., Ludwig, H., Serral, E., 2019]. Co-location specification for http://dx.doi.org/10.1109/TSE.2009.67.
IoT-aware collaborative business processes. In: Cappiello, C., Ruiz, M. (Eds.), NA 152-06-10 AA National Committee, 2003]. DIN 8580:2003-09 Manufacturing
Information Systems Engineering in Responsible Information Systems. Springer Processes - Terms and Definitions, 2003rd-09 Ed. Beuth Verlag, Berlin, Germany.
International Publishing, Cham, pp. 120–132. Neubauer, M., Stary, C. (Eds.), 2017]. S-BPM in the Production Industry. , first. ed.
Grefen, P., Mehandjiev, N., Kouvas, G., Weichhart, G., Eshuis, R., 2009]. Dynamic Springer International Publishing, Cham, Switzerland.
business network process management in instant virtual enterprises. Comput. Neubauer, M., Stary, C., Krenn, F., 2014. Subject-oriented process design across
Ind. 60, 86–103, http://dx.doi.org/10.1016/j.compind.2008.06.006. organizational control layers. In: 2014 12th IEEE International Conference on
Grefen, P., Vanderfeesten, I., Boultadakis, G., 2017]. D2.2A - HORSE Complete System Industrial Informatics (INDIN), IEEE, Porto Alegre, Brazil, pp. 418–423, http://
Design - Public Version (Project Deliverable No. HORSE-D2.2a). HORSE Consor- dx.doi.org/10.1109/INDIN.2014.6945549.
tium. Brussels, Belgium. Oberhauser, R., Stigler, S., 2017]. Microflows: enabling agile business process mod-
Grefen, P., Vanderfeesten, I., Boultadakis, G., 2016]. Architecture Design of the HORSE eling to orchestrate semantically-annotated microservices. In: Proceedings of
Hybrid Manufacturing Process Control System (Design Report). Technische Uni- the 7th International Symposium on Business Modeling and Software Design,
versiteit Eindhoven, Eindhoven, The Netherlands. Barcelona; Spain, pp. 19–28.
Gregor, S., Hevner, A.R., 2013]. Positioning and presenting design science research Object Management Group, 2014]. Business Process Model and Notation (BPMN),
for maximum impact. MIS Q. 37, 337–356. 2.0.2. Ed. Object Management Group, Inc., USA.
Grewal, C., 2008]. An initiative to implement lean manufacturing using value stream O’Leary-Kelly, S.W., Flores, B.E., 2002]. The integration of manufacturing and mar-
mapping in a small company. Int. J. Manuf. Technol. Manag. 15, 404–417, http:// keting/sales decisions: impact on organizational performance. J. Oper. Manag.
dx.doi.org/10.1504/IJMTM.2008.020176. 20, 221–240, http://dx.doi.org/10.1016/S0272-6963(02)00005-0.
Groover, M.P., 2016]. Principles of Modern Manufacturing: SI Version, 6th ed. John Pauker, F., Mangler, J., Rinderle-Ma, S., Pollak, C., 2018]. Centurio.Work - modular
Wiley & Sons, Singapore. secure manufacturing orchestration, in: proceedings of the dissertation award,
Groover, M.P., 2011]. Fundamentals of Modern Manufacturing: Materials, Processes, demonstration, and industrial track at BPM 2018. In: Presented at the 16th Inter-
and Systems, 4th ed. J. Wiley & Sons, Hoboken, NJ. national Conference on Business Process Management, CEUR-WS.Org, Sydney,
Gschwind, T., Koehler, J., Wong, J., 2008]. Applying patterns during business process Australia.
modeling. In: Dumas, M., Reichert, M., Shan, M.-C. (Eds.), Business Process Man- Pesic, M., van der Aalst, W.M.P., 2006]. A declarative approach for flexible busi-
agement. Springer, Berlin, Heidelberg, pp. 4–19, http://dx.doi.org/10.1007/978- ness processes management. In: Eder, J., Dustdar, S. (Eds.), Business Process
3-540-85758-7 4. Management Workshops, Lecture Notes in Computer Science. Springer, Berlin
Heidelberg, Vienna, Austria, pp. 169–180, http://dx.doi.org/10.1007/11837862 Todd, R., 1994. Manufacturing Processes Reference Guide, 4th ed. Industrial Press,
18. Inc., USA.
Prades, L., Romero, F., Estruch, A., García-Dominguez, A., Serrano, J., 2013]. Defin- Ugarte, B.Sde, Artiba, A., Pellerin, R., 2009]. Manufacturing execution system – a
ing a methodology to design and implement business process models in BPMN literature review. Prod. Plan. Control. 20, 525–539, http://dx.doi.org/10.1080/
according to the standard ANSI/ISA-95 in a manufacturing enterprise. Procedia 09537280902938613.
Eng. 63, 115–122, http://dx.doi.org/10.1016/j.proeng.2013.08.283. US Air Force, 1969]. MIL-STD-499 System Engineering Management, first. ed. US Air
Prat, N., Comyn-Wattiau, I., Akoka, J., 2014]. Artifact evaluation in information sys- Force, Washington, DC.
tems design-science research – a holistic View. In: Proceeding of the 19th Pacific van der Aalst, W.M.P., Barros, A.P., ter Hofstede, A.H.M., Kiepuszewski, B., 2000].
Asia Conference on Information Systems, Association for Information Systems, Advanced workflow patterns. In: Scheuermann, P., Etzion, O. (Eds.), Cooperative
Chengdu, China. Information Systems. Springer Berlin Heidelberg, Berlin, Heidelberg, pp. 18–29,
Qu, T., Lei, S.P., Wang, Z.Z., Nie, D.X., Chen, X., Huang, G.Q., 2016]. IoT-based real-time http://dx.doi.org/10.1007/10722620 2.
production logistics synchronization system under smart cloud manufacturing. van der Aalst, W.M.P., ter Hofstede, A.H.M., Kiepuszewski, B., Barros, A.P., 2003].
Int. J. Adv. Manuf. Technol. 84, 147–164, http://dx.doi.org/10.1007/s00170-015- Workflow patterns. Distrib. Parallel Databases 14, 5–51, http://dx.doi.org/10.
7220-1. 1023/A:1022883727209.
Ray, S., 2008]. Introduction to Materials Handling. New Age International (P) Ltd., Van Gorp, P., Vanderfeesten, I., Dalinghaus, W., Mengerink, J., van der Sanden, B.,
Publishers, New Delhi. Kubben, P., 2013]. Towards generic MDE support for extracting purpose-specific
Reichert, M., 2011]. What BPM technology can Do for healthcare process support. healthcare models from annotated, unstructured texts. In: Weber, J., Perseil,
In: Peleg, M., Lavrač, N., Combi, C. (Eds.), Artificial Intelligence in Medicine: 13th I. (Eds.), Foundations of Health Information Engineering and Systems: Second
Conference on Artificial Intelligence in Medicine, Lecture Notes in Computer International Symposium, Lecture Notes in Computer Science. Springer Berlin
Science. Springer Berlin Heidelberg, Berlin, Heidelberg, pp. 2–13. Heidelberg, Berlin, Heidelberg, pp. 213–221.
Reijers, H.A., Russell, N., van der Geer, S., Krekels, G.A.M., 2010]. Workflow for Vanderfeesten, I., Erasmus, J., Grefen, P., 2016]. The HORSE project: IoT and Cloud
healthcare: a methodology for realizing flexible medical treatment processes. solutions for dynamic manufacturing processes. In: Lazovik, A., Schulte, S. (Eds.),
In: Rinderle-Ma, S., Sadiq, S., Leymann, F. (Eds.), Business Process Management Proceedings of the 5th European Conference on Service-Oriented and Cloud
Workshops. Springer Berlin Heidelberg, Berlin, Heidelberg, pp. 593–604, http:// Computing. Springer, Berlin, Heidelberg, Germany, Vienna, Austria, pp. 303–304,
dx.doi.org/10.1007/978-3-642-12186-9 57. http://dx.doi.org/10.1007/978-3-319-72125-5.
Sale, R.S., Mesak, H.I., Inman, R.A., 2017]. A dynamic marketing-operations interface Weerdt, J.D., Schupp, A., Vanderloock, A., Baesens, B., 2013]. Process Mining for
model of new product updates. Eur. J. Oper. Res. 257, 233–242, http://dx.doi. the multi-faceted analysis of business processes - A case study in a finan-
org/10.1016/j.ejor.2016.07.051. cial services organization. Comput. Ind. 64, 57–67, http://dx.doi.org/10.1016/
Sawhney, R., 2006]. Interplay between uncertainty and flexibility across the value- j.compind.2012.09.010.
chain: towards a transformation model of manufacturing flexibility. J. Oper. Wieringa, R., 2009]. Design science As nested problem solving. In: Proceedings of
Manag. 24, 476–493, http://dx.doi.org/10.1016/j.jom.2005.11.008. the 4th International Conference on Design Science Research in Information
Scheer, A.-W., 1994]. CIM: Towards the Factory of the Future, third. ed. Springer, Systems and Technology, DESRIST’ 09, ACM, New York, NY, USA, p. 8, http://dx.
Berlin. doi.org/10.1145/1555619.1555630, 1–8:12.
Shitkova, M., Taratukhin, V., Becker, J., 2015]. Towards a methodology and a tool for Williams, T.J., 1994]. The Purdue enterprise reference architecture. Comput. Ind. 24,
modeling clinical pathways. Procedia Comput. Sci. 63, 205–212, http://dx.doi. 141–158, http://dx.doi.org/10.1016/0166-3615(94)90017-5.
org/10.1016/j.procs.2015.08.335. Witsch, M., Vogel-Heuser, B., 2012]. Towards a formal specification framework
Stock, J.R., Lambert, D.M., 2001]. Strategic Logistics Management, 4th ed. The for manufacturing execution systems. IEEE Trans. Industr. Inform. 8, 311–320,
McGraw-Hill/Irwin series in marketing. McGraw-Hill/Irwin, Boston. http://dx.doi.org/10.1109/TII.2012.2186585.
Takemura, T., 2008]. Formal semantics and verification of BPMN transaction and Zor, S., Görlach, K., Leymann, F., 2010]. Using BPMN for modeling manufacturing
compensation. In: IEEE Asia-Pacific Services Computing Conference, IEEE, Yilan, processes. Proceedings of 43rd CIRP International Conference on Manufacturing
Taiwan, pp. 284–290, http://dx.doi.org/10.1109/APSCC.2008.208. Systems, 515–522.
Tang, C.S., 2010]. A review of marketing–operations interface models: from co- Zor, S., Schumm, D., Leymann, F., 2011]. A proposal of BPMN extensions for the man-
existence to coordination and collaboration. Int. J. Prod. Econ. 125, 22–40, http:// ufacturing domain. In: Proceedings of the 44th CIRP International Conference on
dx.doi.org/10.1016/j.ijpe.2010.01.014. Manufacturing Systems, CIRP: Paris, France, Madison, MI, USA.
Tao, F., Qi, Q., 2018]. New IT driven service-oriented smart manufacturing: frame-
work and characteristics. IEEE Transactions on Systems, Man, and Cybernetics:
Systems, 1–11, http://dx.doi.org/10.1109/TSMC.2017.2723764.

Erasmus 2020

Uploaded by

Copyright:

Available Formats

Erasmus 2020

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Erasmus 2020

Uploaded by

Copyright:

Available Formats

Computers in Industry 123 (2020) 103297

Contents lists available at ScienceDirect

Using business process models for the speciﬁcation of manufacturing

1. Introduction management is extensively studied, but these studies focus on the

The design science research artefact presented in this article

Fig. 1. Overview of research methodology.

4.1. Catalogue of manufacturing operations

Starting with inventory operations, Langford (2007) lists the

Operation group Operation type Description Reference

Process group Operation type Description Reference

Process group Operation type Description

Process group Operation type Description

Operation type Input(s) Activity(ies) Output(s)

Operation type Input(s) Activity(ies) Output(s)

Operation type Input(s) Activity(ies) Output(s)

Operation type Input(s) Activity(ies) Output(s)

Fig. 3. Sand casting process according to Groover (2016).

A mechanical fastening fragment used to represent the joining

Fig. 4. Sand casting process according to Groover (2016).

Fig. 6. Software aspect of the HORSE System (Grefen et al., 2016).

Requirement Process interpretation Additional model elements

Prat et al., 2014 Moody, 2009

J. Erasmus, I. Vanderfeesten, K. Traganos et al. / Computers in Industry 123 (2020) 103297

Fig. 8. Results of homomorphism related questions.

Fig. 9. Questions and results related to model understandability.

You might also like