Professional Documents
Culture Documents
net/publication/360884871
CITATIONS READS
2 113
2 authors:
All content following this page was uploaded by Markus Schreiber on 09 September 2022.
Procedia Procedia
CIRP 00 CIRP
(2017)107
000–000
(2022) 629–634
www.elsevier.com/locate/procedia
École Nationale Supérieure d’Arts et Métiers, Arts et Métiers ParisTech, LCFC EA 4495, 4 Rue Augustin Fresnel, Metz 57078, France
Abstract
*Abstract
Corresponding author. Tel.: +33 3 87 37 54 30; E-mail address: paul.stief@ensam.eu
Traceability systems are widely used in manufacturing processes, mainly for legal reasons. Based on their ability to generate and gather data
Traceability
along processes, systems
theyare
arewidely used in
an excellent basemanufacturing
for creating processes,
performance mainly for legal
indicators. reasons.market
Changing Baseddemands
on their ability
lead to to generate
a rising and gather
amount data
of product
along
variantsprocesses, they are
and decreasing an excellent
batch base higher
sizes causing for creating performance
complexity indicators.
in production Changing
processes. market
In this demands
context lead toavailability
the growing a rising amount
of dataofalongproduct
the
Abstract
variants
value andoffers
chain decreasing batch sizes causing higher complexity in production processes. In this context the growing availability of data along the
new opportunities.
value chain
Based on a offers new opportunities.
manufacturing data set, this paper presents a concept for building a data value chain consisting of a traceability system for data
Ingeneration
today’s
Based on business
aand environment,
manufacturing
acquisition, data the as
set,
as well trend towards
thisa paper
process more aproduct
presents
mining conceptvariety
application forfor and
the customization
building a data of
analysis theischain
value unbroken. Due to
consisting
generated process this
ofdata. development,
a traceability
Firstly, tosystem the need of
for data
determine the
agile and reconfigurable
generation
traceability and
systems production
acquisition,
ability well systems
toasgenerateas relevant emerged
a process todata
mining
process cope forwith
application various
for the
manufacturing, products
analysisand
secondly product
of to families.
thedemonstrate
generated Tothis
process
how design
data and
data. optimize
Firstly,
contributes toproduction
to determine the
data-based
systems as well
traceability
transparency as toability
systems
through choose
processtothe optimal
generate
mining product
relevant
analysis. matches,
process dataproduct
Transparency analysis
forismanufacturing,
essential to methods aretoneeded.
secondly
enable data-based Indeed,
demonstrate
decisions howmost
as thisofdata
well asthe known methods
contributes
improvement aim to
tomeasures
data-basedin
analyze a product
transparency
production. The or oneprocess
through
results product
of familymining
mining
the process on theanalysis
analysis. physical level.
Transparency
are thenDifferent product
is essential
connected families,
tototheenable however,decisions
data-based
specific mayof
configurations differ largely
well asinimprovement
astraceability
the terms
systemof in
theorder
number
measures and
in
to show
nature
the of components.
production. Theand
correlations results This
of thefact
dependencies impedes
process
alongmining
theanentire
efficient
analysis comparison
dataare thenchain.
value and
The choice
connected to the of appropriate
specific
understanding product
ofconfigurations
the data valueoffamily combinations
the traceability
chain from forsystem
system
traceability inthe production
order to to show
process
the correlations
system.
mining Acan
new and dependencies
methodology
empower companies toalong
is proposed
furtherthe entire data
to benefit
analyze value
existing
from chain.
theirproducts The
in understanding
traceability view
system,of their of the datait
functional
by configuring value
and chainthe
to physical
deliver from traceability
architecture.
needed Thesystem is to process
aimtransparency
data-based cluster
these products
mining
and in new
can empower
improvements inassembly
companies oriented product
to further
their production benefit
management. families for the
from their optimization
traceability of existing
system, assemblyit lines
by configuring and the
to deliver thecreation
needed of future reconfigurable
data-based transparency
and improvements
assembly in their
systems. Based onproduction
Datum Flow management.
Chain, the physical structure of the products is analyzed. Functional subassemblies are identified, and
a©©functional
2022 The Authors.
analysis isPublished
performed. by ELSEVIER
Moreover,
2022 The Authors. Published by Elsevier B.V. a B.V.
hybrid functional and physical architecture graph (HyFPAG) is the output which depicts the
This
© is an
2022
similarity
This is an
Theopen
between
open access
Authors. article
Published
product
access under
families
article under the CC
by ELSEVIER
by CC
the BY-NC-ND
providing B.V.
BY-NC-ND license
designlicense
support (https://creativecommons.org/licenses/by-nc-nd/4.0)
to both, production system planners and product designers. An illustrative
(https://creativecommons.org/licenses/by-nc-nd/4.0)
This
example is an
Peer-review
of open
a - access
nail-clipper article
Peer-review is under
under
used to the CC
responsibility
explain BY-NC-ND
the of the
proposed
Peer-review under responsibility of the International Programme committee license
International (https://creativecommons.org/licenses/by-nc-nd/4.0)
methodology. Programme committee
Anofindustrial
the 55th of theon
55th
case Conference
CIRP study twoCIRP
on Conference
product on
of Manufacturing
familiesSystems
Manufacturing steering columns of
Peer-review
Systems
thyssenkrupp -Presta
Peer-review
France under
is thenresponsibility
carried out toofgive the aInternational
first industrial Programme
evaluationcommittee of the 55th
of the proposed CIRP Conference on Manufacturing
approach.
©Systems
2017 TheData
Keywords: value chain;
Authors. traceability
Published system;B.V.
by Elsevier process mining; data-based process transparency; manufacturing processes
Keywords: Data value chain; traceability
Peer-review under responsibility of the scientificsystem; process mining; data-based
committee of the 28thprocess
CIRPtransparency; manufacturing
Design Conference 2018.processes
reliability of input data and to gaining data-based transparency, 1. To present a concept that connects traceability and
the basis to optimizing the coordination of production [6]. process mining from a data perspective in one approach.
Within the “Industrie 4.0” vision of smart factories and smart 2. To demonstrate the data-based connection between
products, automatic identification (autoID) technologies such traceability and process mining in the manufacturing
as RFID are being used to generate data and gain transparency use case.
[7]. Although most manufacturing companies use these so 3. To identify research gaps in the field of traceability as a
called traceability systems due to legal obligations [8] or to data supplier in production.
inventory existing objects, the potential of data acquisition
through autoID technologies is still not fully reached. Research 2. Data Value Chain Concept
in the field of traceability often focuses on the object itself,
giving insights on how to consistently mark objects in A promising concept for companies to adopt the data
production processes, find fitting technologies to track objects perspective and to optimize data usage, can be found in the field
or products [8], or identify effort versus benefit levels that of Big Data by setting up sophisticated data value chains
should be considered when tracking different product (DVC) [16]. Typically, a DVC considers strategically
categories [9]. For successful production management, it has important, value-creating activities [13], and integrate all data-
become crucial to focus on the data application perspective. A affecting steps starting with the generation and collection of
recent research project demonstrated several beneficial use data and ending with the possibility of decision-making based
cases originating from the use of a traceability system and its on data outputs [14,15]. Research shows that the representation
generated data [10]. In this context companies still lack the of DVCs differ widely regarding suggested steps, functions and
knowledge to generate targeted feedback data of their processes its purpose [13,17,16,14,15]
using the traceability system and its ability to locate objects [2]. Figure 1 illustrates a DVC that focuses on six phases that
can be derived from these sources. The general function of each
1.2. Goal phase within the DVC is explained in the corresponding grey
box. The red arrows indicate the application of the DVC
Traceability systems generate process data and can function concept on the manufacturing use case analyzed in this paper.
as an important data supplier in production [9]. From a According to the explanation made in the grey boxes of Figure
theoretical point of view the combination of traceability as a 1 in the DVC, the individual phases are assigned to the
data generating system and process mining as the tool for data traceability system and process mining analysis as follows:
analysis offers great potential for data-based process 1. Data Generation: the traceability system functions as the
transparency [11]. However, researchers have not yet data source in this use case. Which data a traceability system
investigated how a traceability system’s configuration generates depends on the existing configuration of the
influences the availability of traceability data and, in turn, what system in any individual case. The configuration determines
outputs can be revealed using process mining to serve the the place, the time, the quality and the amount of each data
purpose of data-based transparency. This knowledge would point that is generated.
support the challenge to collect the right process data and to 2. Data Acquisition: based on its individual configuration, a
turn them into value-adding information [12]. traceability system records data throughout the production
Based on a manufacturing use case, this paper aims to process and stores them along with meta-information to
accomplish the following:
1. Data Generation 2. Data Acquisition 3. Data Curation 4. Data Preprocessing 5. Data Analysis 6. Data Exploitation
• selection of sources • process of recording • collect, organize and • extract the required • selection of adequate • extract useful output
that generate data and obtaining data preserve data attributes/values data analysis information
• organizations can • use of an inventory to heterogeneous data in and meta-information algorithms • presentation and
consider various data collect data from data a central data from the data • computing power visualization of
sources internally and source(s) warehouse warehouse needs to scalable and results
externally • provide meta- • validate collected • data preparation to considered depending • information base for
information to data including meet the needs for on data amount and decision-making
differentiate data requirements such as analysis: validation, complexity
points meta-information and cleaning, reduction,
quality standards format adjustment,
• ensure data protection aggregation,
integration etc.
Figure 1: Traceability and process mining linked in a data value chain [13–15]
Markus Schreiber et al. / Procedia CIRP 107 (2022) 629–634 631
Markus Schreiber et al./ Procedia CIRP 00 (2019) 000–000 3
distinguish recorded data points and groups. Subsequently, • Workplace: production resource being used by the
there is a certain set of traceability data available. individual manufacturing process
3. Data Curation: the traceability data is (transferred and) • WO Status: indicates whether a partial amount “30” or the
stored systematically in a central data warehouse along with full amount “60” of ordered products is finished; status
further heterogeneous production data. The use of meta- “20” indicates a necessary set-up process of the workplace
information helps to distinguish single data points from • Total Yield: quantity of goods produced in the process step
another. The stored data, IT infrastructure and data for a specific order
protection policies differ from company to company. • Person ID: worker that performs the manufacturing process
4. Data Preprocessing: for process mining analysis the • Material No.: product modules being manufactured in an
extracted dataset from the data warehouse needs to be order
transformed into an event log. Depending on the individual • Begin: start time stamp of the manufacturing process
case the data preparation process might include format • End: end time stamp of the manufacturing process
adjustments, unit standardization, cleansing or the selection
of required data points to create a valid event log. 3.2. Available traceability data (phase 2)
5. Data Analysis: typically, process mining software offers a
wide range of tools for descriptive, diagnostic and In manufacturing, traceability systems typically track
prescriptive analysis. The outcome can be powerful, objects of higher interest such as product [8] or machining tools
especially when domain knowledge is considered within the [18] to record their history of completed production processes.
analysis phase to create meaningful outputs. In the context of the presented DVC, this section aims to
6. Data Exploitation: depending on the available input data understand what specific data points the traceability system
within the event log, the process mining analysis results in generates according to its specific configuration in the use case.
various numbers of outputs that help production managers The available dataset (see Table 1) corresponds to the
to obtain more data-based transparency in their finished data curation (phase 3) of the DVC and requires the
manufacturing processes and to support concrete decision- reverse approach towards the data generation (phase 1). In
making. order to understand the function of the traceability system, the
analysis begins with phase 3 and works backwards towards the
3. Use case analysis – data-based connection between data generation (phase 1) itself. In a first step (towards phase
traceability and process mining 2), the dataset is segmented into primary and secondary data
[19]. The primary data represents all data points that the
3.1. Introduction of the dataset (phase 3) traceability hardware generates. The secondary data covers all
data points that are generated by various sensors.
The use case analysis starts with the company’s extraction
of the dataset from their central data warehouse (phase 3). The
industry dataset represents the production of a manufacturer for
crimping pliers. The structure of the dataset is displayed in
Table 1. An order is divided into several manufacturing steps
using a new line for every new step. Working Operations taking
place within one manufacturing step are represented in a new
line as well.
11.0090- 10 312-100- 20 0 385 2142- 11.12.2018 11.12.2018 Figure 2: Data segmentation from traceability perspective
10 001 15001 10:58:00 12:10:00
11.0090- 10 312-100- 30 25 385 2142- 11.12.2018 11.12.2018
10 001 15001 12:17:00 12:25:00 Figure 2 illustrates the general data segmentation from a
11.0090- 10 312-100- 60 25 385 2142- 11.12.2018 11.12.2018 traceability perspective. It provides an overview of typical
10 001 15001 12:30:00 12:30:00
instances for primary and secondary data. Generally, primary
11.0090- 20 502-100- 60 25 214 2142- 11.12.2018 11.12.2018
10 001 15001 13:00:00 13:34:00 and secondary data can be linked via time stamps. Depending
11.0090- 30 370-200- 60 25 582 2142- 11.12.2018 11.12.2018 on the individual use case, the combination of primary and
10 002 15001 13:49:00 14:26:00
11.0090- 40 110-500- 60 25 376 2142- 11.12.2018 11.12.2018
secondary data should be thoroughly evaluated to ensure it
10 001 15001 16:22:00 17:01:00 serves the intended purpose.
11.0200- 10 312-100- 60 45 385 3784- 11.12.2018 11.12.2018 A good example of combined primary and secondary data is
20 001 24001 11:58:00 13:20:00
the recent trend of companies interested in determining their
products’ individual CO2-footprints. In this case, a product is
The following explains the meaning of each data point:
linked with the time it spent in a machining process (primary
• Order: production order of a customer data) to the machining process data derived from sensors
• Working Operation: sequenced numbering of process steps monitoring electrical power and pressure consumption
within one production order starting with “10”
632 Markus Schreiber et al. / Procedia CIRP 107 (2022) 629–634
4 Markus Schreiber et al. / Procedia CIRP 00 (2019) 000–000
(secondary data): This allows for the generation and allocation assembly represents a coarse data acquisition in comparison to
of a CO2-footprint to the corresponding product [10]. tracking on the individual work station level recording object
movements within the machining, quality or assembly area.
Table 2: Result of conducted data segmentation on the dataset The object granularity specifies whether every part is tracked
as a single part, in a certain batch size or in bulks consisting of
Segmentated Dataset
several batches. The acquisition scope defines the type of
Primary Data Secondary Data
General Traceability Corresponding Corresponding
processes being tracked and can contain transformation
Data Point Name in Dataset
General Data Point
Name in Dataset processes that change the appearance of the object, storage
1 Work Step ID Working Operation 8 Order ID Order processes where the objects are planned to be stored, transport
processes that are important from an intralogistics perspective
2 Work Step Progress WO Status 9 Order Quantity** Total Yield
and handling/ preparation processes that give more detailed
3 Work Station ID Workplace information about necessary supporting activities.
4 Worker ID Person ID
3.4. Process mining analysis (phase 5) and data-based data-based transparency in the company’s manufacturing
transparency (phase 6) processes. All sub-categories of total yield outputs (4.) and
used material no. per order (5.) use exactly one (primary)
Generally, a dataset can be comprised of different datasets traceability data point as input, the remaining outputs use more
or sources. According to the DVC, the dataset (phase 3) needs than one traceability data point as input.
to be preprocessed (phase 4) for the intended process mining
analysis (phase 5). The application of process mining requires 4. Traceability as a data supplier for process mining
a standardized structure of the dataset as input data called an
event log. At minimum an event log must contain at least one The analysis of the manufacturing use case proves the data-
case that includes events (process activities) and a time stamp based connection of traceability and process mining as well as
to set up an occurrence order for the event series. Further the traceability system’s ability to contribute to creating
attributes such as used resources, costs etc. can be included to outputs and transparency. The final result of the analysis is
obtain more detailed input information and hence more outputs. summarized in Figure 4 according to the six phases of the DVC
The event log structure requires the input data to assign events concept. The connections are displayed in a one-to-one
clearly to an individual case and to provide a new line for every relationship between each data point from the traceability
new event within a case [21–23]. system’s configuration in phase 1 to the process mining output
In this specific use case, the company’s dataset does not in phase 6. Phase 1 shows the company’s implemented
require any changes and is suitable for a process mining traceability configuration based on Figure 3Figure 4, the
analysis, so the preprocessing (phase 4) can be skipped. The resulting traceability data points as well as the assignment to
available data points (see Table 1) are assigned to the event log each data point to the dataset based on Table 2. Almost every
as follows: configuration module (phase 1) considered in this use case
• Case: Order generates an individual data point (phase 2) and functions as
• Activity: Workplace data source. However, some configuration modules contribute
• Timestamp Start: Begin to the same data point (e.g. Work Stations and Transformation
• Timestamp End: End to Workstation ID).
• Further attributes: Material No., Working Operation, For a better overview two exemplary process mining outputs
WO Status, Total Yield, Person ID are selected and illustrated in Figure 4. In order to determine
the output total yield depending on individual material
The conducted process mining analysis (phase 5) results in numbers, the traceability system tracks product modules and
a wide range of outputs (phase 6). Those outputs are structured generates product IDs. To create the output workstation
in six output categories with more detailed results in sub- utilization, the configuration modules work station, start point
categories. These outputs do not represent all possible results, and end point are needed.
but only the most relevant outputs for the company that could The large number of process mining outputs mentioned in
be derived from the dataset: chapter 3.4 and the data-based connection of traceability
1. Discovery: (1.1) total, (1.2) by order, (1.3) by material no., system and process mining shown in Figure 4 demonstrate the
(1.4) by orders & material no. traceability system functioning as data supplier and indicate its
2. Conformance check: (2.1) total,
(2.2) by material no., (2.3) by Implemented Available Trans- Process Data-based
Traceability Traceability Dataset formation Mining Process
DVC
high potential to create a vast amount of outputs. In [5] Reinkemeyer, L., 2020. How to Get Started, in: Reinkemeyer, L. (Ed.),
combination with secondary data, the traceability system’s data Process Mining in Action. Springer International Publishing, Cham, pp.
11–14.
contributes strongly to achieving valuable data-based [6] Jahn, M., 2017. Industrie 4.0 konkret. Springer Fachmedien Wiesbaden,
transparency of a company’s production. Wiesbaden, 69 pp.
Before a traceability system is implemented or in the event [7] Bitkom-Gremium - AK Big Data, 2015. Leitlinien für den Big-Data-
the implemented traceability system needs to be adapted in a Einsatz. https://www.bitkom.org/sites/default/files/file/import/150901-
company’s production, numerous configuration options need to Bitkom-Positionspapier-Big-Data-Leitlinien.pdf. Accessed 13 November
2021.
be considered to enable the intended functions. In the use case [8] Wank, A., 2019. Methodik zur Wertstromintegration einer aktiven
analysis, the challenge lay in the lack of a systematic approach Bauteilrückverfolgung in die diskrete Variantenfertigung. Shaker,
to determine the traceability system’s configuration in order to Herzogenrath.
enable the generation of the targeted data points for the desired [9] ZVEI. ZVEI-Traceability-Initiative "Traceability-Levels für
process mining outputs. Further, research should focus on Produktkategorien".
[10] Urnauer, C., Schreiber, M., Bausch, P., Metternich, J., 2021.
developing a systematic configuration model for traceability Anwendungen aktiver Traceability-Systeme: Datennutzung in der
systems that can enable companies to determine their current digitalisierten Produktion. ZWF - Zeitschrift für wirtschaftlichen
configuration and work out necessary alterations for data Fabrikbetrieb 116 (3), 166–170.
availability and targeted output generation. [11] Schreiber, M., Bausch, Phillip, Best, Julian, Metternich, J., 2020.
Datenanalyse in Produktionsprozessen: Potenziale und
Herausforderungen des Process-Mining-Einsatzes in Theorie und
5. Summary and outlook betrieblicher Praxis. ZWF - Zeitschrift für wirtschaftlichen Fabrikbetrieb
115 (5), 309–313.
The generation and usage of data is becoming more relevant [12] G. Schuh, R. Anderl, R. Dumitrescu:A. Krüger:M. ten Hompel, 2020.
in production processes. This paper assesses the potentials of Industrie 4.0 Maturity Index. Die digitale Transformation von
traceability systems and process mining analysis to gain data- Unternehmen gestalten – Update 2020 – (acatech Studie), München, 64
pp.
based transparency. To examine their ability to add data-based [13] Faroukhi, A.Z., El Alaoui, I., Gahi, Y., Amine, A., 2020. Big data
value, the DVC concept is introduced. The DVC adopts the monetization throughout Big Data Value Chain: a comprehensive
data perspective based on six consecutive phases and applies review. J Big Data 7 (1).
them on a company’s manufacturing dataset. [14] Kasim, H., Hung, T., Li, X., 2012. Data Value Chain as a Service
The results show that traceability systems can support the Framework: For Enabling Data Handling, Data Security and Data
Analysis in the Cloud, in: 2012 IEEE 18th International Conference on
challenge to use process mining in production by ensuring the Parallel and Distributed Systems (ICPADS 2012). Singapore, 17 - 19
availability and reliability of data. The combination reveals a December 2012 ; [including workshops. 2012 IEEE 18th International
high potential to enable or at least add to data-based Conference on Parallel and Distributed Systems (ICPADS), Singapore,
transparency in industrial companies. The different phases of Singapore. 17.12.2012 - 19.12.2012. IEEE, Piscataway, NJ, pp. 805–
the DVC concept help to break down the connection of the 809.
[15] Miller, H.G., Mork, P., 2013. From Data to Decisions: A Value Chain
traceability system and process mining on the data point level, for Big Data. IT Professional 15 (1), 57–59.
and promote the understanding from the data source in [16] Jony, I.R., Rony, R.I., Rahman, M., Rahat, A., 2016. Big Data
production to its eventual use. In order to exploit their Characteristics, Value Chain and Challenges. 1st International
connection, research must present a systematic configuration Conference on Advanced Information and Communication Technology
model for traceability systems that helps companies to create 2016, 1–7.
[17] Hu, H., Wen, Y., Chua, T.-S., Li, X., 2014. Toward Scalable Systems for
exactly those data points their production management needs Big Data Analytics: A Technology Tutorial. IEEE Access 2, 652–687.
to achieve specific outputs. [18] Bosch, E., Grosch, T., Abele, E., Metternich, J., Landfried, K.-C.,
Großkurth, D., Hofmann, K., Wieschollek, M., Ebben, A., Schloen, J.,
Acknowledgements Ziegltrum, F., Gutmacher, M., Schwennig, B., 2017. Intelligente
Werkzeuge für die vernetzte Produktion von morgen - SmartTool
Abschlussbericht, Darmstadt.
The research provided in this paper is financed with funding [19] Benfer, M., Gartner, P., Treber, S., Kuhnle, A., Häfner, B., Lanza, G.,
provided by the Ministry of Economics, Energy, Transport and 2020. Implementierung von unternehmensübergreifender Traceability:
Housing in the state of Hessen Germany. The author is Entwicklung, Implementierung und Bewertung von Traceability-
responsible for the content of this publication. Systemen entlang des gesamten Produktlebenszyklus. ZWF - Zeitschrift
für wirtschaftlichen Fabrikbetrieb 115 (5), 304–308.
[20] Ryu, J., Taillard, D., Janssen, C., 2017. GS1 Global Traceability
References Standard: GS1's framework for the design of interoperable traceability
systems for supply chains.
[1] Gottmann, J., 2019. Produktionscontrolling. Springer Fachmedien [21] Leoni, M. de, van der Aalst, W.M., Dees, M., 2016. A general process
Wiesbaden, Wiesbaden, 233 pp. mining framework for correlating, predicting and clustering dynamic
[2] G. Schuh, R. Anderl, R. Dumitrescu:A. Krüger:M. ten Hompel, 2020. behavior based on event logs. Information Systems 56, 235–257.
Der Industrie 4.0 Maturity Index in der betrieblichen Anwendung – [22] Pika, A., van der Aalst, W.M., Wynn, M.T., Fidge, C.J., Hofstede, A.H.
aktuelle Herausforderungen, Fallbeispiele und Entwicklungstrends ter, 2016. Evaluating and predicting overall process risk using event
(acatech Kooperation), München, 44 pp. logs. Information Sciences 352-353, 98–120.
[3] Galic, G., Wolf, M. Delivering Value with Process Analytics Process [23] Suriadi, S., Andrews, R., Hofstede, A.H. ter, Wynn, M.T., 2017. Event
Mining adoption and success factors, 1–36. log imperfection patterns for process mining: Towards a systematic
[4] Flack, C., Dreher, S., Birk, A., Wilhelm, Y., 2020. Process Mining in der approach to cleaning event logs. Information Systems 64, 132–150.
Produktion: Spezifische Herausforderungen bei der Anwendung 115 [24] van der Aalst, W., 2016. Process mining: Data science in action, Second
(11), 1–5. edition ed. Springer, Berlin, Heidelberg, New York, London, 467