Semantic Capability Model for the Simulation of Manufacturing
Processes
Jonathan Reif
1 a
, Tom Jeleniewski
1 b
, Aljosha K
¨
ocher
1 c
, Tim Frerich
2 d
, Felix Gehlhoff
1 e
and Alexander Fay
3 f
1
Institute of Automation Technology, Helmut Schmidt University, Holstenhofweg 85, Hamburg, Germany
2
CTC GmbH (An Airbus Company), Airbusstraße 1, Stade, Germany
3
Chair of Automation, Ruhr University Bochum, Universit
¨
atsstraße 150, Bochum, Germany
Keywords:
Simulation, Simulation Sequence, Capabilities, Ontologies, Semantic Information Model, Industry 4.0.
Abstract:
Simulations offer opportunities in the examination of manufacturing processes. They represent various aspects
of the production process and the associated production systems. However, often a single simulation does
not suffice to provide a comprehensive understanding of specific process settings. Instead, a combination
of different simulations is necessary when the outputs of one simulation serve as the input parameters for
another, resulting in a sequence of simulations. Manual planning of simulation sequences is a demanding
task that requires careful evaluation of factors like time, cost, and result quality to choose the best simulation
scenario for a given inquiry.
In this paper, an information model is introduced, which represents simulations, their capabilities to generate
certain knowledge, and their respective quality criteria. The information model is designed to provide the
foundation for automatically generating simulation sequences. The model is implemented as an extendable and
adaptable ontology. It utilizes Ontology Design Patterns based on established industrial standards to enhance
interoperability and reusability. To demonstrate the practicality of this information model, an application
example is provided. This example serves to illustrate the model’s capacity in a real-world context, thereby
validating its utility and potential for future applications.
1 INTRODUCTION
In the design phase of production processes, espe-
cially those marked by high complexity or substan-
tial unit costs, testing various process parameter con-
figurations with physical prototypes is both time-
consuming and costly (Naresh et al., 2020). This is-
sue is not limited to manufacturing companies creat-
ing entirely new production systems but also affects
those aiming to improve aspects like material proper-
ties, energy efficiency, or emission levels. As a result,
manufacturing companies are constantly redesigning
processes.
Simulations emerge as a powerful tool in this con-
a
https://orcid.org/0009-0001-2079-8967
b
https://orcid.org/0009-0007-0360-4108
c
https://orcid.org/0000-0002-7228-8387
d
https://orcid.org/0000-0001-9097-712X
e
https://orcid.org/0000-0002-8383-5323
f
https://orcid.org/0000-0002-1922-654X
text as they offer the possibility to generate process
information virtually and enable cost-effective testing
of different process parameter configurations within
reasonable time frames (Mourtzis, 2020).
However, the utility of a single simulation is often
limited. It may not suffice to generate necessary pro-
cess information. Instead, one simulation often relies
on specific input parameters derived from preceding
simulations. This interdependent connection between
simulations necessitates an execution of simulation
sequences. In this context, the term parameter specif-
ically refers to input and output parameters used in
a simulation corresponding to process parameters of
a corresponding manufacturing process. It does not
refer to parameters that are specific to the simulation
itself, such as step size.
In practice, these sequences are manually created
by experts. One of the primary considerations here is
to ensure the general suitability of simulations to an-
swer specific questions. Another aspect relates to the
quality of results and how suitable simulations can be
Reif, J., Jeleniewski, T., Köcher, A., Frerich, T., Gehlhoff, F. and Fay, A.
Semantic Capability Model for the Simulation of Manufacturing Processes.
DOI: 10.5220/0012945300003838
Paper published under CC license (CC BY-NC-ND 4.0)
In Proceedings of the 16th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2024) - Volume 2: KEOD, pages 39-50
ISBN: 978-989-758-716-0; ISSN: 2184-3228
Proceedings Copyright © 2024 by SCITEPRESS – Science and Technology Publications, Lda.
39

Simulation Complexity
Object of
Consideration
Simulation Simulation sequence
Figure 1: Schematic illustration of available simulations.
effectively combined for this purpose. Often simula-
tions representing the same object of consideration,
e.g., the same process step, are available in differ-
ent complexities, i.e., different levels of abstraction,
necessitating a selection with regard to time and re-
source constraints. Figure 1 illustrates this relation-
ship. In this figure, various simulations (depicted as
gray boxes) with increasing complexity along the X-
direction are assigned to different objects of consid-
eration along the Y-direction. Note, that the simula-
tions objects of consideration can be process steps,
but there can also be different simulations for certain
objects of consideration within a process step, e.g.,
different process parameters, like pressure or tem-
perature. These individual simulations are intercon-
nected through the previously described input-output
relationships and can be combined to generate pro-
cess information. This is exemplified by the colored
lines. The figure is merely an excerpt. In practice, a
greater number of simulations with varying levels of
complexity, as well as more process steps, are avail-
able. This increases the complexity of selecting and
generating simulation sequences.
A general assumption is that the largest part of a
system should be simulated at the lowest complexity
and only specific parts of interest should be simulated
at a higher level of detail if necessary (Puntel-Schmidt
and Fay, 2015). Due to these considerations, a manual
creation of simulation sequences becomes a complex,
time-consuming, and error-prone activity. It requires
a deep understanding of simulations, their parameters,
and the overall production process.
In order to automatically derive a sequence of sim-
ulations, a machine-interpretable model is required.
This model must describe simulations, their function-
alities and relevant quality criteria, like simulation
time, result quality, etc. Describing the functionali-
ties of simulations corresponds to modeling capabil-
ities of flexible manufacturing resources. A capa-
bility is defined as ”an implementation-independent
specification of a function in industrial production to
achieve an effect in the physical or virtual world”
(K
¨
ocher et al., 2023). Simulations can be considered
to achieve an effect in the virtual world, therefore the
following question arises:
How should a machine-interpretable model for plan-
ning simulation sequences be structured, building on
preliminary work from the capability context and rel-
evant standards for simulations?
For this purpose, first specific requirements for
such a model are deducted building on (Reif et al.,
2023) in Section 2, before drawing a comparison to
related research in this field in Section 3. Based on
this, the created information model is described and
explained in detail in Section 4, pointing out its dif-
ferent components. Afterwards Section 5 describes an
application example from industrial practice to show
the practical potential of the created semantic model.
The paper concludes with a summary and an outlook
for potential future research.
2 REQUIREMENTS
As pointed out in (Reif et al., 2023) there are several
requirements for an approach to automating the gen-
eration of simulation sequences:
• A formal description of simulations and their pa-
rameters in a machine readable format to enable
an automated selection of suitable simulations
based on required process information.
• The consideration of quality criteria to ensure that
objectives align with the capabilities of selected
simulations.
• The consideration of parameter influences on
quality criteria to ensure a useful selection of sim-
ulations.
From this emerges the necessity for a semantic model
that describes simulations and their relevant proper-
ties, to facilitate the automated creation of simulation
sequences (Reif et al., 2023). The requirements for
this semantic model are explained in the following.
They can all be derived from the goal to select a sim-
ulation, respectively a sequence of simulations, that
can supply a user with the right process information
in a demanded quality with minimal effort.
R1: Description of Simulations and Their Capa-
bilities in a Machine Readable Format
The first requirement that the model needs to meet
is the formal representation of simulations and their
properties in a machine readable format. This format
is crucial as it facilitates an automatic identification
and arrangement of appropriate simulations and
KEOD 2024 - 16th International Conference on Knowledge Engineering and Ontology Development
40

simulation sequences based on required process
information. For this automation, simulations must
be characterized by their input and output parameters
as well as their capabilities, i.e., their ability to
produce specific process information. The potential
to produce a designated output from a given input is
referred to as a capability. Consequently, this allows
for the pinpointing and selection of a fitting simula-
tion, or a sequence of simulations, to produce desired
process information through capability matchmaking,
which describes the process of matching product
requirements against resource capabilities (J
¨
arvenp
¨
a
¨
a
et al., 2017).
R2: Description of Selection Criteria for Simula-
tions
The second requirement involves detailing the criteria
for selecting simulations. This is essential for choos-
ing suitable simulations and planning sequences. The
aim is to not only choose simulations that provide re-
quired process information for a specific inquiry but
also to ensure that this information meets quality cri-
teria and is obtained with minimal resource effort.
Thus, the criteria to be described include quality
criteria to guarantee that simulations deliver informa-
tion at the desired level of quality. This also includes
other selection criteria like, e.g., simulation time, in-
cluding run time and lead time. Additionally, it is
necessary to consider the influence of specific input
parameters on selection criteria, such as the quality
of the simulation output. This involves documenting
how input parameters affect the output to determine
their influence on quality criteria.
If an input parameter has a minor impact on the
overall result, less stringent quality requirements may
be applied to it, and lower fidelity models may be ap-
plied to represent it. Conversely, parameters that sig-
nificantly affect the outcome should meet higher qual-
ity standards as the simulation results can only be as
good as its input parameters. (Reiter et al., 2012)
By adhering to this principle, the process of
generating process information can be optimized,
essentially, the resource effort can be minimized,
since inputs that have less impact on quality criteria
can be generated using simulations that are less
resource-intensive in a sequence of simulations.
R3: Interoperability and Reusability of the Infor-
mation Model
The third requirement is the interoperability and
reusability of the information model. Models of-
ten lack the adaptability and extensibility needed for
reusing them in different use cases. For this reason, it
is advisable to implement the information model in a
format that offers the possibility to extend and adapt it
(Vogel-Heuser et al., 2015). Furthermore, models cre-
ated for specific scenarios often face challenges when
applied to different situations due to varying defini-
tions and naming conventions, making it recommend-
able to adhere to common semantics used in a domain
(Hildebrandt et al., 2020).
Additionally, to enhance acceptance in the indus-
try, it is advantageous to utilize established industrial
standards. Consequently, the model should be uni-
versally applicable to the considered domain and use
terms that are widely accepted. This is achievable by
conforming to industry guidelines and standards that
are based on expert knowledge.
Another advantage of adhering to established
standards is that in this way it is ensured that sim-
ulations of manufacturing processes are described in
a consistent manner. This consistency allows simu-
lations to be linked with real-world processes using
the same input and output parameters, described with
common semantics. This linkage not only enables a
use of simulated process information as input for sim-
ulations but also allows for the integration of informa-
tion from real processes, e.g. historical process data,
or expert knowledge, like commonly used parameter
values, particularly when this information meets the
set quality criteria.
3 RELATED WORK
In this section, which is divided into the three sub-
sections Description of Simulations and Their Capa-
bilities, Quality Criteria for Simulation Assessment
and Description of Simulation Parameter Influences,
different approaches relevant to the work presented
are described and compared to the requirements men-
tioned in the previous section.
3.1 Description of Simulations and
Their Capabilities
(Grolinger et al., 2012) focus on utilizing ontologies,
defined by (Studer et al., 1998) as ”formal, explicit
specification of a shared conceptualization”, for rep-
resenting simulation models across various domains.
Ontologies are used to provide a common understand-
ing and facilitate operations such as querying, com-
parison, and inference on simulation models. The
main challenges addressed include the difficulty of
extracting specific information from simulation mod-
els due to their proprietary formats and terminologies,
and the complexity involved in comparing and shar-
ing models across different simulation engines. The
Semantic Capability Model for the Simulation of Manufacturing Processes
41

proposed solution is an ontology-based representation
system that maps domain-specific simulation models
to instances within a structured ontology. However,
this method does not take into account a selection or
planning of simulations or sequences of simulations,
nor does it contemplate any form of simulation eval-
uation. Furthermore, it does not comply with estab-
lished standards or common semantics.
(
ˇ
Sindel
´
a
ˇ
r and Nov
´
ak, 2016) present the develop-
ment of an ontology specifically tailored for the field
of modeling and simulation. The goal is to provide a
centralized, structured knowledge base that can sup-
port the interconnection, interoperability, integration,
and reuse of simulation artifacts through Semantic
Web Technologies. Probable applications described
include improved model discovery and sharing, en-
hanced capability for semantic searches, and the fa-
cilitation of knowledge transfer and reuse across dif-
ferent simulation studies and platforms. However, the
authors do not deal with the topics of simulation se-
lection or planning. Furthermore the presented ontol-
ogy does not rely on industrial standards applicable
for the domain.
(Listl et al., 2023) present a framework for man-
aging discrete-event simulation models through the
use of knowledge graphs. This approach addresses
the challenges associated with managing large and
heterogeneous data sets in the context of production
system simulations. The framework facilitates the
integration of multiple data sources, supports model
reuse, and enables various applications to leverage the
structured representation of simulation models and
their data. By adhering to the industrial standard ISA-
95, the knowledge graph ensures that the data repre-
sented within the simulation models are in alignment
with widely recognized manufacturing standards. By
incorporating the industrial standard VDI 3633 (VDI
3633:1, 2014) for simulation of logistics, material
flow and production systems, the framework ensures
that its approach to simulation management aligns
with established best practices in the field. In spite of
that, the approach mainly focuses on managing and
integrating simulation models using structured data
representations, with a significant emphasis on data
management, reuse, and interoperability, without pur-
suing planning approaches.
Overall, various methods exist for integrating sim-
ulations into information models. However, most of
these methods tend to not utilize industrial standards.
Some of these approaches address the selection of
simulations, though, the detailed planning of simula-
tion sequences, particularly considering selection cri-
teria and the influence of parameters, is not consid-
ered.
In (K
¨
ocher et al., 2023) a technology-independent
model aimed at enhancing adaptability and flexibil-
ity in manufacturing is presented. This so called
Capability-Skill-Service (CSS) model aims at describ-
ing, organizing, and executing manufacturing func-
tions in a way that supports adaptability, reusability,
and efficient management of production resources.
The capability aspect of this CSS model is a central
element that bridges the gap between abstract require-
ments of a production process and concrete function-
alities provided by physical or virtual resources. Ca-
pabilities are abstract descriptions of functions that
are required by processes and provided by resources.
According to the CSS model, purely virtual function-
alities may also be described as capabilities. How-
ever, the focus has so far been on manufacturing pro-
cesses and the description of simulations using the
CSS model has not been considered yet.
3.2 Quality Criteria for Simulation
Assessment
The industrial standard ISO/IEC 25010 (ISO/IEC
25010:2011, 2011) introduces a Product Quality
Model specifically designed for software, organiz-
ing software quality attributes into eight major cate-
gories, such as Functional Suitability, Performance
Efficiency, and Compatibility, among others. Each
category is further detailed through various sub-
characteristics. Additionally, other researchers have
proposed comparable metrics for evaluating mod-
els and choosing simulation software, including at-
tributes like Correctness, Efficiency, and Functional-
ity (Mohagheghi and Dehlen, 2009; Fumagalli et al.,
2019). (Barth et al., 2023) furthermore address the
critical need for a systematic evaluation of the quality
of simulation models, particularly in the context of
digital twins and model-based systems engineering.
As simulation models become integral to system de-
velopment and delivery, the necessity for clear quality
requirements, akin to those for physical components,
becomes paramount. Yet, such standards for simula-
tion models are often undefined due to their novelty
and the absence of universally accepted quality crite-
ria.
(Barth et al., 2023) introduce a metric for objec-
tively assessing simulation model quality, recognizing
the challenge posed by the intrinsic novelty of simu-
lations as deliverables and the existing gap in defined
quality criteria. It emphasizes that while simulation
models are assumed to possess high quality, explicit
requirements or specifications are rarely articulated.
Building on that (Barth et al., 2023) continue to ex-
plore the systematic evaluation of simulation model
KEOD 2024 - 16th International Conference on Knowledge Engineering and Ontology Development
42

quality, focusing on expanding and clustering qual-
ity criteria. The presented approach utilizes ISO/IEC
25010 and thus adapts and extends quality criteria
from software quality domains to suit the unique as-
pects of simulation models, proposing a hierarchy of
attributes that reflect various technical requirements
and stakeholder perspectives.
3.3 Description of Simulation
Parameter Influences
(Saltelli et al., 2008) offer a comprehensive introduc-
tion to sensitivity analysis (SA). The authors provide
a systematic approach to SA, which is a method cru-
cial for understanding how various inputs influence
the outputs of models. The authors point out how SA
can be used to gain understanding about influences of
specific parameters on a simulation. It allows mod-
elers to understand which parameters have the most
influence on the outcome of a simulation. This knowl-
edge is vital for focusing attention on the most criti-
cal parameters, thus optimizing the design and execu-
tion of further experiments or simulations. SA thus
is a pivotal method that bridges theoretical modeling
and practical application in simulations, enabling re-
searchers and practitioners to discern how variations
in input parameters influence simulation outcomes.
(Saltelli et al., 2008)
(Jeleniewski et al., 2023) present a semantic
model developed to articulate the interdependencies
between process parameters in manufacturing set-
tings. This so called Parameter Interdependency
(ParX) ontology incorporates Ontology Design Pat-
terns (ODPs) derived from industrial standards, al-
lowing for a structured and machine-readable rep-
resentation of process knowledge that can be ex-
tended and reused. The semantic model facilitates
process redesign by enabling calculations and predic-
tions of process outputs from specific inputs. The
model incorporates the OpenMath (OM) standard to
express and manage interdependencies between pro-
cess parameters within the semantic model. OM is
a standard for representing mathematical content, en-
abling an exchange of mathematical expressions in a
machine-readable format. An OM Resource Descrip-
tion Framework (RDF) representation is presented
in (Wenzel, 2021). This standard is pivotal for the
model’s ability to detail the quantitative interdepen-
dencies and calculations associated with manufactur-
ing processes as it is utilized to connect mathematical
functions to specific process steps. (Jeleniewski et al.,
2023)
4 SIMULATION ONTOLOGY
This section outlines the Simulation Support (SiS)
information model, an ontology designed to de-
scribe simulations including their capabilities and
their properties relevant for the automated planning of
simulation sequences as described in Section 1. The
model was designed according to the requirements
outlined in Section 2.
The information model is implemented as an on-
tology. This takes R1 into account, as an ontology is
suited to describe simulations and their parameters in
a machine readable, extensible and reusable format.
To create this model, the methodical approach from
(Hildebrandt et al., 2020) was followed. This ap-
proach provides a structured method for building on-
tologies specifically tailored for manufacturing com-
panies. In this method, domain experts play a cru-
cial role by formulating competency questions and
corresponding answers. Competency questions hence
serve as requirements that an ontology must address.
This ensures that the developed ontology considers
the requirements for the automated generation of sim-
ulation sequences. However, it still needs to be val-
idated afterward. In addition, ODPs are leveraged
to enhance generalizability, standardization, and re-
duce modeling effort. Unlike starting from scratch,
the approach of (Hildebrandt et al., 2020) allows for
high reusability of existing models, which takes into
account R3. According to (Gangemi and Presutti,
2009), an ODP is a solution for recurring ontology
design challenges. ODPs are maintained separately
and integrated into an alignment ontology. This inte-
gration enables the incorporation of additional stan-
dards and ODPs in future developments, ensuring ex-
pandability. ODPs relevant for the concept presented
were identified and aligned. The ODPs used and their
alignment are shown in Figure 2.
As can be seen from the prefixes used in Fig-
ure 2 the ODPs used for the SiS ontology include
the CSS model (K
¨
ocher et al., 2023), the DIN EN
61360 (Property Descriptions of Technical Systems)
(DIN EN 61360-1, 2018) and the VDI 3633 (Simu-
lation of systems in materials handling, logistics and
production) (VDI 3633:1, 2014) as well as the ParX
ontology (Jeleniewski et al., 2023), which uses the
OM standard.
The information model, which is available on
GitHub
1
can be divided into three segments:
• The description of simulations and their capabili-
ties (R1, R3)
• The description of quality criteria (R2)
1
https://github.com/JonathanReif/SiS
Semantic Capability Model for the Simulation of Manufacturing Processes
43

VDI3633:Simulation
CSS:Capability
VDI3633:hasProcessQuantity
VDI3633:hasResultsData
CSS:executes
ParX:hasApplication
CSS:Process
SiS:QualityCriteria
DINEN61360:hasDataElement VDI3633:Data
DINEN61360:hasDataElement
CSS:Resource
OM:arguments
OM:operator
ParX:Interdependency
OM:ObjectList
OM:Operator
OM:Object
ParX:isDataFor
DINEN61360:DataElement
SiS:hasInlfuenceOn SiS:isInfluenceFor
SiS:SensitivityIndex SiS:InfluenceScore SiS:Interdependency owl:equivalentClass ParX:Interdependency
SiS:hasQualityCriteria
CSS:requiresCapability
SiS:Influence
SiS:hasInfluence
CSS:providesCapability
Figure 2: Alignment of the different components of the proposed semantic model for the description of simulations.
• The description of parameter influences (R2)
In the following, these components of the overall
model, as well as the industrial standards used in the
information model, are described.
The first segment focuses on the description of
simulations and their capabilities. According to
(K
¨
ocher et al., 2023), a CSS:Process is defined by
its ability to convert inputs into outputs, correspond-
ing to the fundamental process of transforming input
parameters into simulation results. A CSS:Resource
is used for the execution of a CSS:Process it is con-
nected to via the property CSS:executes.
In the presented approach, a simulation is catego-
rized as a subclass of a CSS:Resource. It is repre-
sented by the class VDI3633:Simulation (1).
VDI3633:Simulation ⊆ CSS:Resource (1)
In the SiS ontology, a CSS:Process is utilized
to describe the process of generating information by
simulation, which is the output derived from specific
input parameters. As defined in the CSS model, capa-
bilities are the abstract description of a CSS:Process,
i.e. the process of simulating information by a specific
VDI3633:Simulation
According to the CSS ontology, a resource
may provide capabilities using the object property
CSS:providesCapability and processes may re-
quire capabilities using CSS:requiresCapability.
It follows from (1) that a simulation may provide a
(simulation) capability, which is required by a pro-
cess. Each process, i.e. each description of a sim-
ulation process, is further characterized using prop-
erties and classes to model input and output pa-
rameters according to VDI 3633. Input parame-
ters are linked by VDI3633:hasProcessQuantity,
output parameters by VDI3633:hasResultsData
as VDI 3633 makes this differentiation. Hence,
VDI3633:hasProcessQuantity has a domain and
range defined as shown in (2) and (3), respectively.
∃ VDI3633:hasProcessQuantity.⊤ ⊆ CSS:Process (2)
⊤ ⊆ ∀ VDI3633:hasProcessQuantity.VDI3633:Data (3)
Domain and range of VDI3633:hasResultsData are
defined analogously.
Both input and output parameters are mod-
eled as VDI3633:Data, which is linked to a
DINEN61360:DataElement by the object property
DINEN61360:hasDataElement.
The DIN EN 61360 (DIN EN 61360-1, 2018)
standard is utilized to provide a detailed de-
scription of data elements along with their asso-
ciated types and instances. The DataElement
class is employed to characterize the properties
of any data element. Each data element pos-
sesses a DINEN61360:TypeDescription and a
DINEN61360:InstanceDescription.
The ontology’s second segment focuses on out-
lining quality criteria. R2 formulates the necessity
to consider selection criteria to ensure that objec-
tives align with capabilities of selected simulations.
Selection criteria in this context means on the one
hand quality criteria that describe certain quality as-
pects of a respective simulation and on the other hand
parameter influences. This means the influence of
KEOD 2024 - 16th International Conference on Knowledge Engineering and Ontology Development
44

input parameters on output parameters of a simula-
tion, respectively on different selection criteria. A
consideration of both aspects is necessary to ensure
that selected simulations can produce information un-
der consideration of quality criteria relevant for a use
case.
To set quality criteria in relation to a simulation,
the object property SiS:hasQualityCriteria with
the domain VDI3633:Simulation (4) and the range
SiS:QualityCriteria (5) is introduced, as can be
seen in Figure 2, to make it possible to assign qual-
ity criteria to the respective simulation. The quality
criteria can express different quality attributes.
∃SiS:hasQualityCriteria.⊤ ⊆ VDI3633:Simulation (4)
⊤ ⊆ ∀SiS:hasQualityCriteria.SiS:QualityCriteria (5)
As SiS:QualityCriteria is a subclass
of DINEN61360:DataElement, these differ-
ent attributes can be further described by
a DINEN61360:TypeDescription elaborat-
ing on the SiS:QualityCriteria and by a
DINEN61360:InstanceDescription specifying a
value for the respective SiS:QualityCriteria.
In this way the model allows to resemble differ-
ent quality criteria depending on what is deemed im-
portant for the use case considered. For example the
quality criteria for simulation models described by
(Barth et al., 2023) could be used here to describe the
quality of simulation models.
The third segment of the ontology resembles the
description of parameter influences. Influences rep-
resent the impact of an input parameter of a sim-
ulation process on the quality criteria associated
with this simulation process. For this description
the object property SiS:hasInfluence with the do-
main VDI3633:Data and the range SiS:Influence
is introduced. Additionally, the object properties
SiS:hasInfluenceOn and SiS:isInfluenceFor
are defined with the domain SiS:Influence and the
range DINEN61360:DataElement or CSS:Process,
respectively. This allows for a specification of quality
criteria, respectively the simulation process that is af-
fected by SiS:Influence. An influence is expressed
by the different subclasses of SiS:Influence shown
in Figure 2, see (6).
SiS:SensitivityIndex,
SiS:InfluenceScore,
SiS:Interdependency ⊆ SiS:Influence
(6)
Several possibilities for the description of influences
of input parameters on output parameters were iden-
tified.
The first possibility is the description of influ-
ences by mathematical expressions, mirroring the
impact of input parameters on output parameters. For
instance consider a composite production process,
in which the glass transition temperature of resin
is an important parameter. The influence of the
glas transition temperature may differ depending
on whether the process temperature is in a holding
phase or a heating phase. This variable influence can
be depicted using a mathematical function. Those
functions typically result from expert knowledge.
For a description of such influences the concept
described by (Jeleniewski et al., 2023) was im-
plemented in the information model. This was
realized by using the ParX ontology, which is an
approach for a semantic model representing process
parameters and their interdependencies (Jeleniewski
et al., 2023). It is aligned with the SiS ontology
through the property ParX:hasApplication
with the domain CSS:Process and the
range ParX:Interdependency. The class
ParX:Interdependency is equivalent to the
class SiS:Interdependency.
The second possibility provided to express param-
eter influences is the option to express the influence
not as a mathematical expression but as a single
index. These indices can on the one hand be assessed
by the analysis of the simulation model, e.g. by a
sensitivity analysis, which is a common method to
determine the influence of input parameters on output
parameters of a simulation (Saltelli et al., 2008). The
sensitivity index that is the result of this analysis then
can be expressed by the SiS:SensitiviyIndex.
As SiS:SensitivityIndex is a subclass
of DINEN61360:DataElement it has a
DINEN61360:TypeDescription as well as a
DINEN61360:InstanceDescription. The for-
mer is used to give further information on the
SiS:SensitivityIndex the latter to express the
influence as a value in the value range 0 ≤ I
n
≤ 1
with n ∈ N. Note, that the indices only describe
a fractional contribution of one parameter on a
specific quality criteria of an output o which has the
implication, that the sum of all parameter indices in
the set N for a CSS:Process is 1 as described in (7).
N
∑
n=1
I
n,o
= 1 ∀o ∈ O (7)
If an automated analysis of a simulation is not fea-
sible, e.g., due to the complexity of the simulation,
there is the option to have the influence value evalu-
ated by a domain expert, which can then be expressed
as a SiS:InfluenceScore, in the same way and in
the same value range as the sensitivity index.
Either way, through this description it is possible
to assess whether an input parameter has a great influ-
ence on the output of a simulation and thus whether it
Semantic Capability Model for the Simulation of Manufacturing Processes
45

has to be available in a certain quality or if it can be
neglected.
5 APPLICATION EXAMPLE
The model presented in this paper was developed in
collaboration with an industry partner specializing in
aerospace components and is evaluated within this do-
main. The primary focus in the considered use case is
on modeling process knowledge of the Resin Transfer
Molding (RTM) process for aircraft components dur-
ing the initial design phase. In the course of this, the
possibilities of process information generation, i.e.
the respective simulations, are also to be described,
mapped and made usable.
In this section, the application of the presented in-
formation model through an exemplary use case is
described. The RTM process, which is employed to
manufacture fiber-reinforced plastics, consists of in-
jecting resin under pressure into a mold that holds
a fiber preform, and then proceeding with a curing
phase. The process step exemplified in the presented
example is the infusion of resin into a mold.
The application example follows the method pre-
sented in (Reif et al., 2023) and demonstrates that the
information model introduced in Section 4 can serve
as the foundation for this method.
The method starts with identifying suitable sim-
ulations for the generation of a required process in-
formation. In this application example, a user wants
to simulate the local fill time in the process, i.e., the
time until resin arrives at every point of the part.
The information model provides the possibility for
a user to evaluate which simulations are capable of
generating the desired information from the provided
inputs. First, the process for which the parameter
ex:LocalFillTime is an output must be identified.
Next, the capability required by the process needs to
be determined. Finally, this required capability can be
matched to the simulations that provide it.
This can be achieved via the SPARQL Proto-
col and RDF Query Language (SPARQL) query dis-
played in Listing 1. In this way this selection, nor-
mally done manually, can be supported in an auto-
mated fashion.
Listing 1: A SPARQL query to identify simulations able to
generate a specific output.
SELECT ? pr ? cap ? sim WHERE {
? pr V DI 36 3 3 : ha s Re s ul t sD a t a ex :
Lo c al F il l Ti m e .
? pr CSS : re qu i r e s C ap a b il i t y ? cap .
? ca p a CSS : Ca pa b il i t y .
? si m CSS : p r o vi d e s C a pa b i li t y ? c ap .
? si m a VDI 3 6 3 3 : Si mu l at i o n .}
ex:SimulatingThermosetInfiltration1
CSS:Process
ex:InfiltrationSimulationDL
VDI3633:Simulation
ex:InfiltrationSimulationTPDL
VDI3633:Simulation
ex:SimulatingInfiltration1
CSS:Capability
ex:Geometry
VDI3633:Data
ex:TextilePermeability
VDI3633:Data
ex:ResinViscosity
VDI3633:Data
ex:LocalFillTime
VDI3633:Data
ex:DryFiberForming
CSS:Process
ex:SimulatingDraping1
CSS:Capability
ex:KinematicDraping
VDI3633:Simulation
ex:ShellFEModel
VDI3633:Data
CSS:requiresCapability
CSS:providesCapability
VDI3633:hasProcessQuantity
VDI3633:hasProcessQuantity
VDI3633:hasProcessQuantity
VDI3633:hasResultsData
VDI3633:hasResultsData
CSS:requiresCapability
CSS:providesCapability
CSS:providesCapability
VDI3633:hasProcessQuantity
Figure 3: Exemplary graph excerpt resembling the capabil-
ity of the simulation.
In the use case considered, two available simula-
tions to simulate the parameter ex:LocalFillTime
can be identified. These two simulations provided in
the information model can be seen in Figure 3, an in-
filtration simulation based on Darcy’s law, which de-
scribes the flow of a fluid through a porous medium
and an infiltration simulation based on two phase
Darcy’s law, which describes immiscible two-phase
flow in porous media.
As can be seen in Figure 3, the simulations
are expressed as VDI3633:Simulation. As both
simulations for the thermoset infiltration can
generate the same output from the same input
parameters they are both connected to the same
capability ex:SimulatingInfiltration1 via the
property CSS:providesCapability. The sim-
ulation process generating ex:LocalFillTime
out of the respective inputs is expressed as
ex:SimulatingThermosetInfiltration1.
As the respective object properties ex-
press whether VDI3633:Data is an input
or an output of the process the instances
ex:Geometry, ex:TextilePermeability
and ex:ResinViscosity are connected to
ex:SimulatingThermosetInfiltration1
via VDI3633:hasProcessQuantity and
KEOD 2024 - 16th International Conference on Knowledge Engineering and Ontology Development
46

ex:LocalFilltime is connected via
VDI3633:hasResultsData.
The next step is the selection of a suitable simula-
tion based on use case specific quality criteria. This
quality criteria can be queried via SPARQL as well
as can be seen in Listing 2. In Listing 2 only the
quality criteria for ex:InfiltrationSimulationDL
is queried, the quality criteria for other simulations
can be queried accordingly.
Listing 2: A SPARQL query to get quality criteria of a sim-
ulation.
SELECT ? qua l ? id ? val
WHERE { ex : I n f i l t r a t i o nS i m ul a t io n D L a
VD I3 63 3 : S i mu la t io n .
? si m SiS : h a s Qu a l i t y Cr i t er i a ? q ua l .
? q ual D IN EN 6 13 6 0 :
ha s _ I n s t a n c e _ D e s c ri p t io n ? id .
? id D I NE N 6 1 36 0 : v al ue ? v a l .}
For the selection of a simulation the quality
criteria are significant. The quality criteria relevant
to this use case include lead time for preparing the
simulation, simulation time, technological maturity
of the simulation, and the result’s accuracy. For
simplification purposes the exemplary graph in Fig-
ure 4 only shows the quality criteria result accuracy.
It is expressed as ex:ResultAccuracyInfSimDL
for the simulation InfiltrationSimulationDL
and as ex:ResultAccuracyInfSimTPDL for the
simulation InfiltrationSimulationTPDL as both
simulations have a different accuracy. Accuracy
in this case means the consideration of phenomena
such as pore formation, which can be simulated
with ex:InfiltrationSimulationTPDL. Further-
more, both instances of SiS:QualityCriteria
have a DINEN61360:InstanceDescription,
which makes it possible to express a value for
the accuracy. As can be seen in Figure 4, the
returned value for the quality criterion ”Result
Accuracy” for ex:InfiltrationSimulationDL
would be 0.7 in this example. The value for
ex:InfiltrationSimulationTPDL would be 0.8,
indicating a higher result accuracy for the latter.
This helps users to make an informed deci-
sion when choosing the appropriate simulation for
their needs. For this example the user chooses the
ex:InfiltrationSimulationDL because a higher
result accuracy is not needed for this use case. If
the consideration of pore formation is relevant a
higher result accuracy should be chosen and the sim-
ulation ex:InfiltrationSimulationTPDL would
be better suited, which is information, that can
be provided in the DINEN61360:TypeDescription
of the quality criteria es:ResultAccuracy. The
trade-off here is the additional simulation time
ex:InfiltrationSimulationDL
VDI3633:Simulation
ex:ResultAccuracyInfSimDL
SiS:QualityCriteria
ex:ResultAccuracyInfSimDL ID
DINEN61360:InstanceDescription
ex:ResultAccuracy
DINEN61360:TypeDescription
0.7
ex:ResultAccuracyInfSimTPDL
SiS:QualityCriteria
ex:ResultAccuracyInfSimTPDL ID
DINEN61360:InstanceDescription
0.8
ex:InfiltrationSimulationTPDL
VDI3633:Simulation
SiS:hasQualityCriteria
DINEN61360:hasInstanceDescription
DINEN61360:hasTypeDescription
DINEN61360:value
DINEN61360:hasTypeDescription
DINEN61360:hasInstanceDescription
SiS:hasQualityCriteria
DINEN61360:value
Figure 4: Exemplary graph excerpt resembling the quality
criteria of the simulation.
this simulation requires, which is also modeled as
SiS:QualityCriteria but is omitted, for clarity
reasons, in Figure 4.
The next step in the method is the identification of
necessary inputs for the simulation process. This can
be done via the SPARQL query displayed in Listing 3
Listing 3: A SPARQL query to get inputs of a simulation.
SELECT ? cap ? pr ?pq WHERE {
ex : In f i l t r a t i o n S i m u l at i o nD L CSS :
pr o v id e s C a p ab i l i t y ? cap .
? pr CSS : re qu i r e s C ap a b il i t y ? cap .
? pr V DI 36 3 3 : ha s P r o c es s Q ua n t i t y ? pq .}
In the presented example, the simulation uses the
geometry of the textile (i.e., the composite material)
discretized by a polygon mesh, the textile permeabil-
ity, and the resin viscosity as inputs as can be seen in
Figure 3.
The next step in the method is to determine if
the identified input parameters are known to the user
or if they must be generated via simulation. For
this example the user knows the necessary input pa-
rameters except for the textile permeability, which
the user has to simulate then. In this way the cy-
cle starts again and the user has to identify a suit-
able simulation again as shown in Listing 1. The
permeability of the fiber is dictated by the draping
of said fiber. Therefore a draping simulation can
be used to determine the permeability as opposed to
measuring it experimentally. The model takes this
into account by representing the required input pa-
rameter ex:TextilePermeability of the simulation
process ex:SimulatingThermosetInfiltration1
as the output parameter of the simulation process
ex:DryFiberForming which requires the capabil-
ity ex:SimulatingDraping1 offered by the simu-
Semantic Capability Model for the Simulation of Manufacturing Processes
47

lation ex:KinematicDraping. This simulation in
turn requires a Shell-FE-Model as an input. In this
way it is possible to express sequences of simula-
tions resulting from this relation between input and
output parameters of connected simulations. The
output ex:LocalFillTime of the simulation pro-
cess ex:SimulatingThermosetInfiltration1 can
in turn be used as further input for succeeding simu-
lations like, e.g., a distortion prediction with a ther-
mochemical simulation, which is not pictured in Fig-
ure 3 for clarity reasons. The user applies the pre-
sented method again taking into account the quality
criteria important for him. Another relevant aspect in
this context is the influence that an input parameter
has on the important quality criteria. If the influence
is high there should be more consideration in choos-
ing the right simulation to generate required input pa-
rameters than if the influence is low. The influence of
the input parameters for a specific simulation can be
queried as shown in Listing 4.
Listing 4: A query formulated in SPARQL to identify pa-
rameter influences.
SELECT ? cap ? pr ? inf ? in p ? id ? va l
WHERE {
ex : In f i l t r a t i o n S i m u l at i o nD L CSS :
pr o v id e s C a p ab i l i t y ? cap .
? pr CSS : re qu i r e s C ap a b il i t y ? cap .
? in f SiS : i s I n f l u e n ce F or ? pr .
? in p SiS : h a sI n fl u en c e ? in f .
? in f D I N E N6 1 36 0 :
ha s _ I n s t a n c e _ D e s c ri p t io n ? id .
? id D I NE N 6 1 36 0 : v al ue ? v a l .}
In this application example, the influence of the
input parameters on the output parameters was as-
sessed by a domain expert. It was determined that
textile permeability and quality of the geometry dis-
cretized by a polygon mesh have a high influence
on the simulation output and therefore get a score
of 0.4 each, while resin viscosity has a medium in-
fluence and therefore gets a score of 0.2. In Fig-
ure 5 only the influence of textile permeability for
SimulatingThermosetInfiltration1 is pictured
for simplification purposes. Other influences are
modeled accordingly.
The result of this method is a simulation se-
quence in which certain simulations have to be ex-
ecuted before others to deliver necessary input pa-
rameters for the former. In this example, the simula-
tion ex:KinematicDraping must be executed before
the simulation ex:InfiltrationSimulationDL or
ex:InfiltrationSimulationTPDL, as shown in
Figure 3.
The information model presented allows repre-
senting the connection between different simulations
by modeling both input and output parameters of
ex:TextilePermeability
VDI3633:Data
ex:InfluenceScoreTePe
SiS:InfluenceScore
ex:SimulatingThermosetInfiltration1
CSS:Process
ex:InfluenceScoreTePe ID
DINEN61360:InstanceDescription
ex:InfluenceScore
DINEN61360:TypeDescription
0.4
SiS:hasInfluence
SiS:isInfluenceFor
DINEN61360:hasInstanceDescription DINEN61360:hasTypeDescription
DINEN61360:value
Figure 5: Exemplary graph excerpt resembling the parame-
ter influences of the simulation.
those simulations as VDI3633:Data, facilitating the
description of simulation sequences. Describing qual-
ity criteria as shown in Figure 4 not only acknowl-
edges the overall ability of simulations to produce
specific knowledge, but also allows for the consider-
ation of these quality criteria when determining the
most suitable simulation for a particular scenario.
The structured description of quality criteria sup-
ports the consideration of such criteria in simulation
selection and planning, a complex process done by
experts. By describing it in a machine readable format
this consideration can be supported in an automated
fashion as well. Another aspect the model supports
is the evaluation and comparison of different planned
simulation sequences based on relevant quality crite-
ria.
Evaluating and describing the impact of specific
parameters allows for the planning of efficient simu-
lation sequences true to the motto as precise as nec-
essary but as imprecise as possible. In this example
result accuracy is of high importance. For this reason
it can be derived that TextilePermeability, having
a high influence, should be taken into particular con-
sideration, which in turn sets quality requirements for
the simulation ex:KinematicDraping that generates
this input parameter.
The application of the SiS ontology to the exam-
ple described in this section was carried out in collab-
oration with a simulation expert from the manufac-
turing domain. This collaboration confirmed the as-
sumption that the ontology is suitable for describing
simulations for manufacturing processes, considering
important aspects for selection and planning. Conse-
quently, the ontology supports the selection and plan-
ning of simulations by structuring necessary informa-
tion and making it accessible to users.
The modeled example including the mentioned in-
stances not pictured in the exemplary graph excerpts
can be found on GitHub
1
.
KEOD 2024 - 16th International Conference on Knowledge Engineering and Ontology Development
48

6 CONCLUSION AND FUTURE
WORK
In conclusion, this paper has presented a novel infor-
mation model designed to address the challenges in-
herent in the planning of simulation sequences, ac-
cording to the question raised at the beginning. The
model represents simulations, their ability to generate
specific knowledge, and their respective quality cri-
teria, thereby facilitating the generation of simulation
sequences. The development of the model was guided
by specific requirements and has proven practical in
an application example. This underscores the model’s
utility in providing necessary information to support
users in simulation selection and planning. However,
instantiating the information model requires expertise
in both the semantic web domain and the simulation
domain, as well as a solid understanding of the pro-
posed information model. Furthermore, the model on
its own only offers limited added value to a user, as
it is intended to serve as a foundation for further de-
velopments and applications that build upon it. In
this way the information model presented in this pa-
per paves the way for further research into automated
support for simulation sequence creation, with the ul-
timate goal of reducing the effort required to generate
necessary process information.
In terms of future work, there are several potential
extensions and developments building on the work
presented in this paper. The first aspect is the integra-
tion of the presented information model into a system
that provides an intuitive user interface. This allows
users to interact with the information model in a way
that is significantly useful to them, as the formula-
tion of SPARQL queries is not feasible for most users.
In this context, the question of how the information
model can be automatically instantiated for existing
simulations and during simulation creation could be
examined more closely. The second important area
of development is the creation of a planning logic for
automated selection of simulations and planning of
simulation sequences, as users would benefit signif-
icantly from a higher degree of automation in simu-
lation planning. In this scenario, it could be benefi-
cial to examine the evaluation of quality criteria for
specific use cases more closely, as well as the uncer-
tainties that are inherently associated with the assess-
ment of this quality criteria. Developing a prototype
for the automated generation of simulation sequences
is a practical step towards implementing this plan-
ning logic. As a possible extension to the informa-
tion model it could be beneficial to incorporate a skill
component, meaning the executable implementation
of the abstract functions in addition to the capabili-
ties describing these functions. (K
¨
ocher et al., 2023)
This would allow for a representation of how to exe-
cute simulations. In this context, it would be valuable
to assess the feasibility of automating the execution
of simulations, which would involve evaluating the
extent to which simulation parameterization and ex-
ecution can be automated, identifying limitations or
challenges that may arise.
ACKNOWLEDGEMENTS
This contribution originates from the LaiLa project,
funded by dtec.bw – Digitalization and Technology
Research Center of the Bundeswehr which we grate-
fully acknowledge. dtec.bw is funded by the Euro-
pean Union – NextGenerationEU
REFERENCES
Barth, M., Ristic, M., and J
¨
akel, J. (2023). A role-based
Metric to determine the Quality of Simulation Mod-
els. In 2023 IEEE 28th International Conference
on Emerging Technologies and Factory Automation
(ETFA), volume 2023. IEEE.
DIN EN 61360-1 (07.2018). Standard data element types
with associated classification scheme - Part 1: Defini-
tions - Principles and methods (IEC 61360-1:2017).
Fumagalli, L., Polenghi, A., Negri, E., and Roda, I. (2019).
Framework for simulation software selection. Journal
of Simulation, 13(4):286–303.
Gangemi, A. and Presutti, V. (2009). Ontology Design Pat-
terns. In Staab, S. and Studer, R., editors, Handbook
on Ontologies, pages 221–243. Springer Berlin Hei-
delberg, Berlin, Heidelberg.
Grolinger, K., Capretz, M. A. M., Marti, J. R., and Sri-
vastava, K. D. (2012). Ontology-based Representa-
tion of Simulation Models. In Proceedings of the 24th
International Conference on Software Engineering &
Knowledge Engineering (SEKE’2012), pages 432–
437. Knowledge Systems Institute Graduate School.
Hildebrandt, C., K
¨
ocher, A., Kustner, C., Lopez-Enriquez,
C.-M., Muller, A. W., Caesar, B., Gundlach, C. S.,
and Fay, A. (2020). Ontology Building for Cyber–
Physical Systems: Application in the Manufacturing
Domain. IEEE Transactions on Automation Science
and Engineering, 17(3):1266–1282.
ISO/IEC 25010:2011 (2011). Systems and software egni-
neering - Systems and software Quality Requirements
and Evaluation (SquaRE) - System and software qual-
ity models.
J
¨
arvenp
¨
a
¨
a, E., Siltala, N., Hylli, O., and Lanz, M. (2017).
Capability Matchmaking Procedure to Support Rapid
Configuration and Re-configuration of Production
Systems. Procedia Manufacturing, 11:1053–1060.
Semantic Capability Model for the Simulation of Manufacturing Processes
49

Jeleniewski, T., Nabizada, H., Reif, J., K
¨
ocher, A., and Fay,
A. (2023). A Semantic Model to Express Process Pa-
rameters and their Interdependencies in Manufactur-
ing. In 2023 IEEE 32nd International Symposium on
Industrial Electronics (ISIE), pages 1–6. IEEE.
K
¨
ocher, A., Belyaev, A., Hermann, J., Bock, J., Meixner,
K., Volkmann, M., Winter, M., Zimmermann, P.,
Grimm, S., and Diedrich, C. (2023). A reference
model for common understanding of capabilities and
skills in manufacturing. at - Automatisierungstechnik,
71(2):94–104.
Listl, F. G., Fischer, J., and Weyrich, M. (2023). An Archi-
tecture for Knowledge Graph based Simulation Sup-
port. In 2023 IEEE 28th International Conference
on Emerging Technologies and Factory Automation
(ETFA), pages 1–8. IEEE.
Mohagheghi, P. and Dehlen, V. (2009). Existing model met-
rics and relations to model quality. IEEE, Piscataway,
NJ.
Mourtzis, D. (2020). Simulation in the design and operation
of manufacturing systems: state of the art and new
trends. International Journal of Production Research,
58(7):1927–1949.
Naresh, K., Khan, K. A., Umer, R., and Cantwell, W. J.
(2020). The use of X-ray computed tomography for
design and process modeling of aerospace compos-
ites: A review. Materials & Design, 190:108553.
Puntel-Schmidt, P. and Fay, A. (2015). Levels of Detail and
Appropriate Model Types for Virtual Commissioning
in Manufacturing Engineering. IFAC-PapersOnLine,
48(1):922–927.
Reif, J., Jeleniewski, T., and Fay, A. (2023). An Approach
to Automating the Generation of Process Simulation
Sequences. In 2023 IEEE 28th International Confer-
ence on Emerging Technologies and Factory Automa-
tion (ETFA), pages 1–4. IEEE.
Reiter, M., Breitenbucher, U., Kopp, O., and Karastoy-
anova, D. (2012). Quality of data driven simulation
workflows. In 2012 IEEE 8th International Confer-
ence on E-Science, pages 1–8. IEEE.
Saltelli, A., Ratto, M., Andres, T., Campolongo, F., Cari-
boni, J., Gatelli, D., Saisana, M., and Tarantola, S.
(2008). Global sensitivity analysis: The primer. Wi-
ley, Chichester, West Sussex.
ˇ
Sindel
´
a
ˇ
r, R. and Nov
´
ak, P., editors (2016). Ontology-Based
Simulation Design and Integration. Springer Interna-
tional Publishing and Imprint: Springer, Cham, 1st ed.
2016 edition.
Studer, R., Benjamins, V., and Fensel, D. (1998). Knowl-
edge engineering: Principles and methods. Data &
Knowledge Engineering, 25(1-2):161–197.
VDI 3633:1 (12.2014). Simulation of systems in materials
handling, logistics and production Fundamentals.
Vogel-Heuser, B., Fay, A., Schaefer, I., and Tichy, M.
(2015). Evolution of software in automated pro-
duction systems: Challenges and research directions.
Journal of Systems and Software, 110:54–84.
Wenzel, K. (2021). OpenMath-RDF: RDF encodings for
OpenMath objects and Content Dictionaries. In 31st
OpenMath Workshop.
KEOD 2024 - 16th International Conference on Knowledge Engineering and Ontology Development
50