Available online at www.sciencedirect.com Available online at www.sciencedirect.com
ScienceDirect ScienceDirect Available online atonline www.sciencedirect.com Procedia CIRP 00 (2019) 000–000 Available at www.sciencedirect.com
ScienceDirect ScienceDirect
Procedia CIRP 00 (2019) 000–000
www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia
Procedia CIRP 00 (2017) 000–000 Procedia CIRP 81 (2019) 441–446
52nd CIRP Conference on Manufacturing Systems 52nd CIRP Conference on Manufacturing Systems
www.elsevier.com/locate/procedia
Benefit evaluation of digital assistance systems for assembly workstations 28thof CIRP Designassistance Conference, May 2018, Nantes, France Benefit evaluation digital systems fora assembly workstations a a a Thimo Keller *, Christian Bayer , Phillip Bausch , Joachim Metternich a a Thimo Keller Christian Bayer ,Tools Phillip Bausch ,Str. Joachim Metternich a methodology toa*, analyze the functional and aphysical architecture Institute for Production Management, Technology and Machine (PTW), Otto-Berndt 2, 64287 Darmstadt, Germany
A new of a Institute for Productionfor Management, Technology and Machine Tools (PTW),product Otto-Berndt Str.family 2, 64287 Darmstadt, Germany existing products an assembly oriented identification
* Corresponding author. +49-6151-16-20289; E-mail address:
[email protected].
Paul Stief *, Jean-Yves Dantan, Alain Etienne, Ali Siadat
* Corresponding author. +49-6151-16-20289; E-mail address:
[email protected].
Abstract École Nationale Supérieure d’Arts et Métiers, Arts et Métiers ParisTech, LCFC EA 4495, 4 Rue Augustin Fresnel, Metz 57078, France Abstract
diversification of products results in more challenging jobs for employees *The Corresponding author. Tel.: +33 3 87 37 54 30; E-mail address:
[email protected]
working within the production processes. Digital assistance is expected to increase the performance and flexibility of employees. This paper presents a framework to analyse the The diversification of products results in more challenging jobs for employees working within the production processes. Digital influence of digital assistance systems on essential target variables in a production system. For this purpose, functions of the digital assistance is expected to increase the performance and flexibility of employees. This paper presents a framework to analyse the assistance system, effects on the employee and key performance indicators are defined. From this, hypotheses are derived to influence of digital assistance systems on essential target variables in a production system. For this purpose, functions of the digital Abstract describe the effect of digital assistance systems on employee performance. Subsequently, a concept for the verification of the assistance system, effects on the employee and key performance indicators are defined. From this, hypotheses are derived to framework and the hypotheses is presented. The approach allows to quantify the benefits of digital support in production. This the effect of digital the assistance systems onproduct employee performance. Subsequently, a concept fordevelopment, the verification of the Indescribe today’s business environment, trend towards more variety and customization is unbroken. Due to this the need of supports acceptance and enables decision-makers to integrate digital technologies more benefit-oriented. framework and the hypotheses presented. Thetoapproach quantify benefits of digital support production. This agile and reconfigurable productionissystems emerged cope withallows varioustoproducts andthe product families. To design andin optimize production supports and enables decision-makers to integrate digital technologies more benefit-oriented. systems as acceptance well as to choose the optimal product matches, product analysis methods are needed. Indeed, most of the known methods aim to © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license analyze product or one product family on the physical level. Different product families, however, may differ largely in terms of the number and © 2019aThe Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/3.0/) © 2019 The Authors. Published by Elsevier Ltd. This iscomparison an open(http://creativecommons.org/licenses/by-nc-nd/3.0/) access the CC BY-NC-ND license nature ofan components. This fact impedes anBY-NC-ND efficient and article choice under of appropriate product family combinations for the production This is open access article under the CC license Peer-review under responsibility of the scientific committee of the 52nd CIRP Conference on Manufacturing Systems. (http://creativecommons.org/licenses/by-nc-nd/3.0/) system. A newunder methodology is proposed to analyze existing products in view their functional and physicalSystems. architecture. The aim is to cluster Peer-review responsibility of the scientific committee of the 52nd CIRPof Conference on Manufacturing Peer-review responsibility of the scientific committee of the 52nd CIRPofConference on Manufacturing Systems. these productsunder in new assembly oriented product families for the optimization existing assembly lines and the creation of future reconfigurable Keywords: flexible assembly, employee support, experimental setup assembly systems. Based on Datum Flow Chain, the physical structure of the products is analyzed. Functional subassemblies are identified, and assembly, employee support,a experimental setup and physical architecture graph (HyFPAG) is the output which depicts the a Keywords: functionalflexible analysis is performed. Moreover, hybrid functional similarity between product families by providing design support to both, production system planners and product designers. An illustrative example of a nail-clipper is used to explain the proposed methodology. An industrial studyofondigitisation two product and families of steering columns of 1. Motivation In the case context "Industrie 4.0", experts thyssenkrupp Presta France is then carried out to give a first industrial evaluation of the proposed approach. predict great potential for improvement. For example, Motivation In the context of digitisation and "Industrie 4.0", savings experts ©1.2017 The Authors. Published by Elsevier B.V. Industrial production faces a multitude of challenges which in inventory, manufacturing and logistics as well as predict great potential for improvement. For example, savings Peer-review under responsibility of the scientific committee of the 28th CIRP Design Conference 2018.
have a massive impactfaces on awork in industrial production complexity, quality and maintenance costs are assumed Industrial production multitude of challenges which in inventory, manufacturing and logistics as well [3]. as environments [1]. As a result of increasing flexibilisation and However, the question of what specific effects technologies Keywords: Assembly; Design Family in identification have a massive impactmethod; on work industrial production complexity, quality and maintenance costs are assumed [3]. digitisation, work of of employees can be characterised and applications have onofproduction-relevant key technologies performance environmentsthe[1]. As tasks a result increasing flexibilisation and However, the question what specific effects as follows [2]: indicators (KPIs) mostly remains unanswered. The digitisation, the work tasks of employees can be characterised and applications have on production-relevant key performance as follows Frequent product changes quantification of benefits is an important step towards the [2]: indicators (KPIs) mostly remains unanswered. The 1. Introduction of the product rangetechnologies. and characteristics manufactured and/or Short-cycle, more flexible change of work task acceptance of new To address this research gap Frequent product changes quantification of benefits is an important step towards the assembled this system. In this context, the main challenge in Highly flexible deployment of employees in the fieldinof worker assistance systems and to quantify Short-cycle, more flexible change of work task acceptance ofdigital new technologies. To address this research gap to independent the fast work development in the domain of modelling and analysis is nowa not onlymeasurement to cope withconcept single Due Cycle tasks the added value for companies, suitable Highly flexible deployment of employees in the field of digital worker assistance systems and to quantify communication and an solving ongoingandtrend of digitization and products, a limited product range or existing product families, Increase in problem monitoring tasks is introduced. Cycle independent work tasks the added value for companies, a suitable measurement concept digitalization, manufacturing enterprises are facing important but For alsothis to be able to in analyze and to compare products to define purpose, this article Increase in problem solving and monitoring tasks is introduced. challenges in today’s environments: a continuing new productfunctions families.ofItdigital can beassistance observed systems that classical existing As a result of thesemarket developments in production, higher For specific are described this purpose, in this article tendency towards reductiondirectly of product development times and product families are regrouped in function of clients or features. and distinguished (section 2), performance requirements arise for the employees. In As a result of these developments in production, higher specific functions of digital assistance systems are described shortened product lifecycles. In addition, there iscomplexity an increasing However, assembly oriented families are hardly to find. and matching target values of product a2), production system are identified order to meet these requirements make distinguished (section performance requirements directlyand arise for the the employees. of In demand of customization, being at the same time inthem. a global On the product family level, products differ mainly in two (section 3 and 4), the processes manageable, it is necessary to support This matching target values of a production system are identified order to meet these requirements and make the complexity of competition with all oversystems the world. This trend, main characteristics: can be done by competitors digital assistance supporting the (section 3 and 4), (i) the number of components and (ii) the the processes manageable, it is necessary to support them. This which is inducing the task. development from macro to micro type of components (e.g. mechanical, electrical, electronical). employee for the given can be done by digital assistance systems supporting the markets, results in diminished lot sizes due to augmenting Classical methodologies considering mainly single products employee for the given task. product varieties (high-volume to low-volume production) [1]. or solitary, already existing product families analyze the 2212-8271 © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license To cope with this augmenting variety as well as to be able to product structure on a physical level (components level) which (http://creativecommons.org/licenses/by-nc-nd/3.0/) 2212-8271 © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license an efficient definition and identify possible optimization potentials in the existing causes regarding Peer-review under responsibility of the scientific committee of the 52nd CIRP Conference on difficulties Manufacturing Systems. (http://creativecommons.org/licenses/by-nc-nd/3.0/) production system, it is important to have a precise knowledge comparison of different product families. Addressing this Peer-review under responsibility of the scientific committee of the 52nd CIRP Conference on Manufacturing Systems. 2212-8271 © 2019 The Authors. Published by Elsevier Ltd. This is an©open article Published under theby CC BY-NC-ND 2212-8271 2017access The Authors. Elsevier B.V. license (http://creativecommons.org/licenses/by-nc-nd/3.0/) Peer-review under responsibility of scientific the scientific committee theCIRP 52ndDesign CIRPConference Conference2018. on Manufacturing Systems. Peer-review under responsibility of the committee of the of 28th 10.1016/j.procir.2019.03.076
Thimo Keller et al. / Procedia CIRP 81 (2019) 441–446 Keller et al. / Procedia CIRP 00 (2019) 000–000
442 2
a causal model is created by connecting the functions and target values to describe the impact of a digital assistance system on a production process (section 5), possible test scenarios are categorised in order to evaluate digital assistance systems in their respective application context (section 6), an approach to verify the causal model by using a multivariate analysis method is presented (section 7). In the end of this article a short summary and an outlook on the further research activities will be given. 2. Functions of digital assistance systems To describe a digital assistance system, its properties must be clearly defined. According to DIN EN ISO 6385 the working system is characterized by the interaction between the worker and the work equipment in order to fulfill the work task under the given conditions [4]. Assistance systems, as part of the work equipment in the work system, support the employee for the given task to be performed. By now, different scientific approaches exist to categorise and differentiate assistance systems. In general, these can be divided into sensory, cognitive and physical assistance systems depending on the type of support [5,6]. Others consider sensory assistance as a part of cognitive assistance [7] or informative assistance is introduced as a new category [8]. A logical approach leads to a subdivision into the following types of assistance: Sensory assistance supports perception (data acquisition), cognitive assistance provides the worker with instructions and analyses (data processing) and physical assistance assists in the execution of action. Since the focus of digital assistance systems lies in the sensory and cognitive support of employees, the physical assistance is not addressed within this scientific approach. Table 1. Functions of digital assistance systems.
Based on the existing literature, the structure as presented in Table 1 can be developed [9, 10]. Cognitive assistance is further divided into two categories. “Instructions for unknown processes” enhance the standard functions as soon as the process is no longer known. Considering this classification, all relevant functions of a digital assistance system can be circumscribed. Thus, also different distinct digital assistance systems can be described and clearly distinguished by their functionalities. 3. Effects of the assistance system on the employee‘s performance The assistance system as part of the work equipment is intended to support the employee in the performance of his work tasks. It directly influences the employee and thus only has an indirect influence on the production process (Fig. 1). In order to describe the benefits of the digital assistance system on the production process in detail, the effects on the employee must be described first. Subsequently, KPIs are defined to quantify the benefits of digital assistance systems. A distinction must be made between different target categories in which effects of digital assistance can be expected. For the definition of such a target system, the highly accepted target triangle consisting of the categories quality, cost and time is applied [11]. The target triangle is enhanced by the category flexibility. This adaptation is supported by a number of other authors and useful because it is a key challenge for the manufacturing industry [12, 13]. Within the scope of a literature research, expected effects of digital assistance systems are gathered and subsequently clustered for all mentioned target categories. In the field of quality, the use of digital assistance leads to a decreasing error rate (Q1), improved correction (Q2) and detection (Q3) of errors. In terms of time, the use of cognitive assistance mainly influences the effort for handling information (T1) and communication (T2). Flexibility is primarily ensured by the qualification level of the employee (F2) and the qualification requirement of the task (F1). The qualification requirements of employees can be influenced in particular by intuitive guidance and the appropriate provision of information. By using a digital assistance system, it is also possible to make selected learning content available to employees and thus increase the qualification level. No effects to reduce costs were included as the cost advantages result from other effects. Separate costrelated effects were not identified.
Type of assistance
Function
Definition
Perception (sensory assistance)
Documentation
Documentation of machine-, product-, process- and employee-related data
Monitoring
Monitoring of machine, product and process (target/actual-comparison )
Context recognition
Sensory recognition of process progress
Initiation of processes
Triggered start of a process
Information output
Output of information (signal, text, image, video)
Learning and practicing
System provides educational content
Decisionmaking support
Support in problem solving through analysis results
Functions of the digital assistance system
Communication
Contacting people in charge
Fig. 1. Influence of the assistance system on the production process.
Instruction for known processes (cognitive assistance)
Instructions for unknown processes (cognitive assistance)
Effects on the efficiency of the production process (KPIs) Effects on the employee‘s performance
Thimo Keller et al. / Procedia CIRP 81 (2019) 441–446 Keller et al. / Procedia CIRP 00 (2019) 000–000
The aim of quantifying the benefits of digital assistance in mind, the used parameters regarding quality are defined as follows: 𝑄𝑄1: 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 = 𝑞𝑞𝐸𝐸 =
𝑥𝑥𝐸𝐸𝐸𝐸 𝑥𝑥𝐸𝐸𝐸𝐸
(1)
𝑄𝑄2: 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 = 𝑞𝑞𝐸𝐸𝐸𝐸 = 𝑄𝑄3: 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 = 𝑞𝑞𝐸𝐸𝐸𝐸 =
𝑥𝑥𝐸𝐸𝐸𝐸
𝑥𝑥𝐸𝐸𝐸𝐸 𝑥𝑥𝐸𝐸𝐸𝐸 𝑥𝑥𝐸𝐸𝐸𝐸
(2) (3)
In this case 𝑥𝑥𝐸𝐸𝐸𝐸 is the number of errors after the observed process, 𝑥𝑥𝐸𝐸𝐸𝐸 the total number of different errors made by all employees, 𝑥𝑥𝐸𝐸𝐸𝐸 the number of corrected errors, 𝑥𝑥𝐸𝐸𝐸𝐸 the number of detected errors and 𝑥𝑥𝐸𝐸𝐵𝐵 the number of errors before the observed process has started. The time-related effects are quantified by recording the corresponding times for handling information 𝑡𝑡𝑖𝑖 (T1) and for communication 𝑡𝑡𝑐𝑐 (T2). The qualification requirement 𝑄𝑄𝑛𝑛 is described by using the defined allocated time 𝑡𝑡𝑎𝑎 and the time actually required 𝑡𝑡𝑤𝑤 :
𝐹𝐹1: 𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 = 𝑄𝑄𝑛𝑛 = 𝑡𝑡𝑤𝑤 − 𝑡𝑡𝑎𝑎
(4)
The quantification of changes in the qualification level (F2) of employees is much more complicated and possible through a long-term evaluation of competence development only. To sum up this section, suitable effects for direct recording at the workplace are listed in the following: Error rate 𝑞𝑞𝐸𝐸 Error correction rate 𝑞𝑞𝐸𝐸𝐸𝐸 Error detection rate 𝑞𝑞𝐸𝐸𝐸𝐸 Times for handling information 𝑡𝑡𝑖𝑖 Times for communication 𝑡𝑡𝑐𝑐 Qualification requirement 𝑄𝑄𝑛𝑛 4. Effects on the efficiency of the production system
It is expected that the integration of digital assistance impacts different levels of the production system. A distinction is made between workplace, process and company level. At the workplace level the usage of a digital assistance system has an effect on specific time requirements (processing time, reworking time, etc.) and produced quantities (total quantity, scrap parts, reworking parts). These effects can be assigned to the categories quality and time. The goal of developing a measurement concept requires directly measurable and clearly defined key performance indicators that can be assigned to the effect of the digital assistance system. Consequently, the following selection of indicators is limited to the workplace level because KPIs at the process or company level are not directly measurable, are more complex in their composition and causal relationships, can be assigned directly to the digital assistance system to a limited extent only and can be used to a limited extent in practice only, due to the high effort required for recording.
443 3
Suitable KPIs for direct recording at the workplace are: number of scrap parts 𝑥𝑥𝑠𝑠 number of reworked parts 𝑥𝑥𝑟𝑟 total number of produced parts 𝑥𝑥𝑡𝑡 total time 𝑡𝑡 time for work task 𝑡𝑡𝑤𝑤 time for reworking 𝑡𝑡𝑟𝑟 training period 𝑡𝑡𝑡𝑡 setup time 𝑡𝑡𝑠𝑠 downtime 𝑡𝑡𝑑𝑑 know-how carrier quota 𝑞𝑞𝑘𝑘
The quantities xs, xr and xt as well as the time periods tw, tr, tt, ts and td are recorded directly at the workplace and serve as KPIs in the categories quality and time. The know-how carrier quota 𝑞𝑞𝑘𝑘 serves as an indicator for the more flexible deployment of employees. It is defined as the ratio of employees with know-how carrier status for the examined workplace (e.g.: product group) to the total number of employees [14]. The mentioned KPIs are the foundation for the benefit assessment of digital support in production processes. By aggregating this data it is additionally possible to calculate key performance indicators at process or company level. Examples are the reject rate, rework rate or the first pass yield for quality. For a holistic evaluation of the benefits of digital assistance at the process or company level, the overall equipment effectiveness (OEE) is an appropriate tool. The OEE includes essential temporal effects as well as the quality of the process in one KPI and can be used as an indicator for the economic efficiency of the process. As shown in the following formula, the OEE can be calculated using the data recorded at the workplace under consideration of the target number of produced parts 𝑥𝑥𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 [15]: 𝑂𝑂𝐸𝐸𝐸𝐸 =
𝑥𝑥𝑡𝑡 −𝑥𝑥𝑠𝑠 −𝑥𝑥𝑟𝑟 𝑥𝑥𝑡𝑡
∗
𝑡𝑡−𝑡𝑡𝑑𝑑 −𝑡𝑡𝑠𝑠 𝑡𝑡
∗
𝑥𝑥𝑡𝑡
𝑥𝑥𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡
(5)
Thus, the selection of indicators meets the requirement to quantitatively describe the benefits of digital assistance systems in the target categories quality, time, flexibility and costs (economic efficiency). 5. Description of the developed causal model After identifying and defining the functions of digital worker assistance as well as the effects and (process-) KPIs, a causal model for the benefit assessment of digital assistance systems can be derived, by linking the functions (Section 2), effects (Section 3) and KPIs (Section 4) in context. A causal model is defined as follows: “Causal models are mathematical models representing causal relationships within an individual system (…). They facilitate inferences about
Thimo Keller et al. / Procedia CIRP 81 (2019) 441–446 Keller et al. / Procedia CIRP 00 (2019) 000–000
444 4
Functions of the assistance system
Effects on the employee
KPI‘s production process
Quality Scrap Parts [ ]
Q1 Error Rate [ ]
F1 Documentation [0/1]
Time
Reworked Parts [ ]
Total Number Of Parts [ ]
Q2 Error Correction Rate
F2 Control [0/1]
Time For Work Task [ ]
Q3 Error Detection Rate [ ]
F3 Context Identification [0/1]
Time For Reworking [ ]
T1 Times For Handling Information ]
F4 Initiation Of Processes [0/1]
Perception Hypothesis
Flexibility
Training Period [ ]
Setup Time [ ]
T2 Times For Communication [ ]
F5 Information Output [0/1]
Downtime [ ]
F1 Qualification Requirement [
F6 Learn & Practice [0/1]
F7 Communication [0/1]
Instructions for action for known processes
Know-How Carrier Quota [
F2 Qualification Level [Competence; Long-term]
F8 Decision Support [0/1]
Instructions for action for unknown processes
Fig. 2. Causal model to describe the effects of digital assistance on the production process.
causal relationships from statistical data. They can teach us a good deal about the epistemology of causation, and about the relationship between causation and probability.“ [16] The aim of the model design is the illustration of interdependencies between the functions of the assistance system, the identified effects and the KPIs in the form of 31 hypothetical assumptions. The developed causal model is shown in Fig. 2. The cause-effect relationships (hypotheses; red arrows) presented in the model were logically derived on the basis of current literature and are the foundation for the verification of the model. The aim is to quantify the assumed relationships and also to identify missing ones. Thus, in addition to the superior research question - What are the quantifiable benefits of a digital assistance system on the efficiency of assembly employees? - subordinate questions, such as the influence of individual functions can also be examined. For example, it is expected that the implementation of the "Control" function will significantly reduce the number of rejects due to the effect of a higher defect detection rate. The examination is conducted by means of a pre-/postcomparison. The same activity is carried out by comparable experimental groups under conditions as equal as possible and once with and once without digital assistance. This approach allows to identify the benefits of digital assistance systems on the performance of employees and key performance indicators. 6. Categorisation of different test scenarios Depending on the observed employee and the process to be performed, very different results are expected. When testing the hypotheses formulated in the causal model, it is important to distinguish between various application scenarios/use contexts for which different forms of benefit effects are expected. This is why different test scenarios have to be defined
in order to allocate the expected different results. The process as well as the executing employee are considered as further elements of the test scenarios. This follows the understanding of a production system as a socio-technical system [17]. To differentiate between processes, the difficulty of the task is evaluated. A specific parameter is the entropy of the work task, which is based on the number of components of the considered product as well as the number of elements required for the description of the task according to MTM (MethodsTime Measurement) [18]. In business practice, the information required to calculate the entropy correctly is usually not available and can be determined with enormous effort only. Therefore, this paper follows a simplified approach in which the work task is evaluated on the foundation of product variance and the scope of the work task. On basis of these criteria a differentiation between low, medium and highly complex processes is possible (see Table 2). A process is classified as sufficiently complex if one of the two criteria reaches the value range of the higher complexity level. Table 2. Differentiation of processes. Process complexity
Product variance
Scope of work task
low
< 30
< 5 min
medium
31-500
5 min-60 min
high
> 500
> 60 min
The distinction between different employees is based on employee competencies. Competence is defined as the contextspecific use of skills and knowledge by the employee [19]. The employee's experience with the task is to be regarded as a decisive condition for correct context-specific action of the employee. According to this understanding, a distinction is
Thimo Keller et al. / Procedia CIRP 81 (2019) 441–446 Keller et al. / Procedia CIRP 00 (2019) 000–000
Employee competence 1.2 Process experienced employees – low process complexity
2.2 Process experienced employees – medium process complexity
3.2 Process experienced employees – high process complexity
1.1 Inexperienced employees – low process complexity
2.1 Inexperienced employees – medium process complexity
3.1 Inexperienced employees – high process complexity
Process complexity Fig. 3. Test scenarios to differentiate between use contexts.
made between process experienced and process inexperienced employees. An employee who has not yet performed the process is considered to be inexperienced. Recording the respective competencies of the employee is not practicable. By variation of process complexity and employee competence, six different standard applications of digital assistance can be derived to obtain comparable results (Fig. 3). These will be the considered test scenarios within the framework of the experimental execution. 7. Verification of the theoretical model To verify the model (Fig. 2) and quantify the relationships (hypotheses) a statistical evaluation of several experiments is necessary. The holistic approach for the verification is shown in Fig. 4 and will be explained below. After specifiying the test scenario the statistical experiment has to be prepared. Structural multivariate analysis methods can be used to verify theoretical models. For the presented case the regression analysis is a suitable solution to describe and explain the assumed relationships quantitativly [20]. The regression function describes the mathematical relationship between a target quantity (Y) and all related independent variables (x): 𝑌𝑌 = 𝑏𝑏0 + 𝑏𝑏1 𝑥𝑥1 + 𝑏𝑏2 𝑥𝑥2 + . . . + 𝑏𝑏𝑗𝑗 𝑥𝑥𝑗𝑗 + . . . + 𝑏𝑏𝐽𝐽 𝑥𝑥𝐽𝐽
(6)
As the influence of the assistance system on the production process is to be determined via the effects on the employee, a two-stage regression is necessary. In the first step the relationships between the functions of the assistance system and the effects on the employee are examined. To analyse the results from the pre-/post-tests described in section 5, the data must be defined as independent
1. Define test scenario (process & employee)
2. Define independent variables using existing functions
3. Reduce number of independent variables (clustering) or tests to be performed (DoE)
445 5
or dependent variables. Within the framework of regression, employee competence, process complexity and the functions of the assistance system form the regressors (independent variables). Employee competence and functions are treated as "dummy variables". This means that a function is regarded as “on” (value 1) or “off“ (value 0). Equivalently, the employee's competence is considered to exist (1) or not to exist (0). Process complexity is divided into low complexity (0), medium complexity (1) and high complexity (2) using the ordinal scale shown in table 2. The effects achieved are the target variables (dependent variables). In a second step, the relationships between the effects and the KPIs are verified. In this case the effects form the influencing variables (independent variables) and the indicators the target variables (dependent variables). This allows to describe and quantify the influence of the effects on the recorded indicators mathematically. As a result of this procedure, a regression function is created for each effect and each KPI (see Fig. 4). Even missing causal relationships can be identified and supplemented by testing the hypotheses. If a significant improvement in individual indicators can be observed without being attributable to a hypothesis, a further as yet unidentified effect can be assumed. This new relation must then be investigated by means of specified experiments. The full factorial verification of the theoretic model is not practicable due to the large number of variables. If all possible combinations of functional scope, employee competence and process complexity are to be tested, a total of 1,536 experiments must be carried out. For this reason, two approaches to simplify the model are possible: Clustering or Design of Experiments (DoE). Clustering is proposed for field tests or laboratory experiments whereas DoE is proposed only for laboratory experiments only, since the required adjustments of the assistance system are not feasible in a field test. If a field test in a company is to be carried out, it is not recommended to consider individual functions of the assistance system. The defined functions (F1-F8) are combined to one single independent variable "Digital Assistance System". All identified effects and KPIs (see section 3 and 4) are recorded automatically or manually and subsequently evaluated. The independent variables of employee competence and process complexity are varied according to the possibilities in the company considered. Now the used assistance system can be described on the basis of the existing functions and distinguished from other systems. In this way the influence of the used digital assistance system can be quantified and compared with other results. Possible differences can be explained by the functions which are used during the test.
ͳǣ
ǣ ʹǣ ǣ
Fig. 4. Procedure for the execution of the statistical evaluation.
+ +
4. Execute regression with test results
446 6
Thimo Keller et al. / Procedia CIRP 81 (2019) 441–446 Keller et al. / Procedia CIRP 00 (2019) 000–000
Within the framework of laboratory experiments it is possible to switch individual functions on or off in order to pursue more detailed questions. However, a reduction of the number of necessary tests is also mandatory. There are two different approaches: On the one hand, functions can be clustered into higher-level functions in order to reduce the number of independent variables. Thus, the defined functions can be clustered according to the type of support resulting in the three clusters "perception”, “instruction for known processes" and "instruction for unknown processes" (see Table 1). This way independent variables in the model are reduced from eight to three. On the other hand, the number of experiments to be performed can be significantly reduced with the help of statistical experiment planning (DoE). 8. Summary and outlook With the aim of quantifying the benefits of digital assistance systems for production processes the functions of digital assistance systems are classified and defined within this paper. Subsequently, these functions are faced with expected effects on the employee’s performance and the defined KPIs. As a result, a complete causal model to illustrate the benefits of digital assistance on the production process is derived. Within the next step, the causal model has to be verified. On the basis of the test scenarios described in section 6 pre-/posttrials can be planned and carried out to investigate the causal model. Using a regression analysis, the cause-effect relationships will be quantified in order to make a well-founded statement on the benefits of digital assistance systems in different production environments. Within the framework of the research project IntAKom, which is funded by the BMBF and ESF, pre- and post-tests will be carried out in different production environments. The process learning factory CiP at the TU Darmstadt, which is operated by the Institute for Production Management, Technology and Machine Tools, offers a suitable research environment for comprehensive laboratory tests. The results from field and laboratory are then compared to identify systematic differences. All performed experiments will be classified into one of the scenarios defined in section 6 in order to describe the differences between the defined categories. Thus, the presented causal model forms the basis for a comprehensive understanding of the topic as well as for the further investigation of the interactions between digital assistance systems, employees and production systems. Acknowledgements The project IntAKom is financed with funding provided by the Federal Ministry of Education and Research and the European Social Fund under the "Future of work" programme. It is implemented by the Project Management Agency Karlsruhe (PTKA). The author is responsible for the content of this publication.
References [1] acatech. Kompetenzentwicklungsstudie Industrie 4.0 – Erste Ergebnisse und Schlussfolgerungen. München, 2016. p. 9. [2] Dombrowski U, Riechel C, Evers M. Industrie 4.0 - Die Rolle des Menschen in der vierten industriellen Revolution. In: Kersten W, Koller H, Hermann L. Industrie 4.0. Wie intelligente Vernetzung und kognitive Systeme unsere Arbeit verändern. Berlin: Gito, 2014. p. 129–153. [3] Bauernhansl T. Die Vierte industrielle Revolution. Der Weg in ein wertschaffendes Produktionsparadigma. In: Bauernhansl T, ten Hompel M, Vogel-Heuser B. Industrie 4.0 in Produktion, Automatisierung und Logistik. Wiesbaden: Springer Vieweg, 2014. p. 31. [4] DIN EN ISO 6385 2016. Grundsätze der Ergonomie für die Gestaltung von Arbeitssystemen, Deutsche Fassung EN ISO 6385, 2016. p. 6 [5] Apt W, Schubert M, Wischmann S. Digitale Assistenzsysteme – Perspektiven und Herausforderungen für den Einsatz in Industrie und Dienstleistungen. Berlin: Institut für Innovation und Technik, 2018. p. 2021. [6] Müller R, Vette M, Mailahn O, Ginschel A, Ball J. Innovative Produktionsassistenz für die Montage - Intelligente Werkerunterstützung bei der Montage von Großbauteilen in der Luftfahrt. Düsseldorf: wtonline, 2014. p. 552-560. [7] Reinhart G, Bengler K, Dollinger C, Intra C, Lock C, Popova-Dlogosch S, Rimpau C, Schmidtler J, Teubner S, Vernim S. Der Mensch in der Produktion von Morgen. In: Reinhart G. Handbuch Industrie 4.0. München: Carl Hanser Verlag; 2017. [8] Lewin M, Wallenborn M, Küstner D, Erdelmeier D, Fay A. Auf dem Weg zu innovativen Assistenzsystemen–Systematische Klassifizierung für die Praxis. Wissenschaft trifft Praxis, 11; 2017. p. 12. [9] Spath D, Ganschar O, Gerlach S, Hämmerle M, Krause T, Schlund S,. Produktionsarbeit der Zukunft - Industrie 4.0, Stuttgart: Fraunhofer Verlag; 2013. [10] Klapper J. Digitale Assistenzsysteme – auch in Ihrer Produktion?. 2018, retrieved from: https://blog.iao.fraunhofer.de/digitale-assistenzsystemeauch-in-ihrer-produktion, 12.12.2018. [11] Kletti J, Schumacher J. Die perfekte Produktion - Manufacturing Excellence durch Short Interval Technology (SIT). Berlin: Springer Vieweg; 2015. [12] Ward PT, McCreery JK, Ritzman LP, Sharma D. Competitive Priorities in Operations Management. Decis Sci 1998; 29:1035–1046. [13] Thun JH, Drüke M, Grübner A. Empowering Kanban through TPSprinciples – an empirical analysis of the Toyota Production System. Int J Oper Prod Manag 2010; 48:7089–7106. [14] Havighorst F. Personalkennzahlen, Düsseldorf: Hans-Böckler-Stiftung, 2006. [15] May C, Koch A. Overall Equipment Effectiveness (OEE) - Werkzeug zur Produktivitätssteigerung. Zeitschrift der Unternehmensberatung 2008, p. 245–250. [16] Hitchcok C. Causal Models. 2018, retrieved from: https://plato.stanford.edu/archives/fall2018/entries/causalmodels/#IntrSEMs, 12.12.2018. [17] Ullrich C, Aust M, Blach R, Dietrich M, Igel C, Kreggenfeld N, Kahl D, Prinz C, Schwantzer S. Assistenz- und Wissensdienste für den Shopfloor, In: Rathmayer S, Pongratz H, editors. Proceedings of DeLFI Workshops 2015 co-located with 13th e-Learning - Conference of the German Computer Society. München; 2015. p. 47–55. [18] Jeske T, Schlick CM, Mütze-Niewöhner S. Unterstützung von Lernprozessen bei Montageaufgaben. In: Schlick CM, Moser K, Schenk M. Flexible Produktionskapazität innovativ managen. Berlin: Springer Vieweg; 2014. p. 163-192. [19] Abele E, Chryssolouris G, Sihn W, Metternich J, ElMaraghy H, Seliger G, Sivard G, ElMaraghy W, Hummel V, Tisch M, Seifermann S. Learning factories for future oriented research and education in manufacturing, CIRP Annals 2017; 66:803–826. [20] Backhaus K, Erichson B, Plinke W, Weiber R. Multivariate Analysemethoden. Berlin: Springer Gabler, 2018. p. 58.