SysML’ – incorporating component properties in early design phases of automated production systems

1 Introduction and motivation

In an increasingly competitive plant manufacturing sector, companies must seize every opportunity to increase efficiency in engineering and operation for their customers. The characteristics of gear-drives as well as motors significantly impact the overall system performance in many application examples, e.g., continuous press, paper production machines, or rolling mills, to name only three. If the drive system with its gear and motor fails, the entire plant experiences downtime.

Models of automated Production Systems (aPS) typically include descriptions of mechanical parts, electrical and electronic parts (automation hardware) and software, which are all closely interwoven. They are characterized by their integration of sensor and actuator subsystems and thus represent a special class of mechatronic systems. The integration of gear-drives models as well as temperature and altitude constraints of motors in aPS is still insufficiently researched with a special focus on interdisciplinary interfaces and environmental constraints, and their efficient design poses significant challenges for the industry [1].

A gear-drive system is a constructive unit responsible for the machine’s movement and it includes mechanics, sensors, actuators, the motor, and information processing. The behavior of gear-drives as well as electronic components like sensors, and bus couplers depend on environmental conditions such as temperature, and altitude of the plants premise. Furthermore, the wear of the drive components has a crucial influence on their possible performance. Comprehensive knowledge of the components of the drive system, the automation components, including the software and the communication network including all physical properties is particularly needed for the development of future aPS and the optimization of system performance [1]. This paper proposes a Systems Modeling Language (SysML) Profile to ease the modeling of these related dependencies.

SysML was introduced in 2001 by the Object Management Group to support the development of complex systems from a system engineering point of view. It is a graphical modeling language that defines a set of diagrams to describe systems’ behavior and structure/composition. SysML as a general-purpose language provides a semantic that is not rich enough to capture the variety of attributes that effects the systems behavior of an aPS.

The SysML’ profile for aPS modeling, specified by a metamodel is the main contribution of this paper. The SysML’ profile includes selected environmental characteristics by using references to existing standards like ECLASS and REXS, as well as suppliers’ documentation. The metamodel documents the abstract syntax of reference and its internal transformation mechanism. Additionally, a Business Process Modeling Notation is used to illustrate the benefits of SysML’ in an interdisciplinary workflow. To demonstrate the applicability, this paper models a very high-level industrial application example involving co-rotating conveying belts driven by gear-drives.

The remainder of this paper is structured as follows. To motivate the modeling of aPS in a multi-disciplinary way, the state of the art and its limitations using SysML is presented in Section 2. In Section 3 the usage of the different diagrams of SysML’ is demonstrated through the introduction of an application example. A metamodel of the SysML’ profile, is introduced in Section 4. Section 5 demonstrates the role of SysML’ in the engineering process, while Section 6 summarizes the contribution and provides an outlook for further work.

2 State of the art in modeling mechatronic systems

In the german joint research project SPES, leading computer scientists from academia and industry initially developed a seamless methodology and analysis techniques for embedded systems resulting in a modeling framework. Since 2012, automation scientists further extended SPES for process automation, automobile, and aviation to meet higher requirements regarding robustness and longevity. The SPES modeling framework allows seamless, model-based development of complex embedded software by defining structures, contents, and concepts used in artifacts [2]. It features four different viewpoints: requirements, functional, logical, and technical. The requirements viewpoint supports the elicitation, documentation, and management of requirements. The functional viewpoint focusses on the functionality of the developed system and describes its intended behavior. The logical viewpoint decomposes functions into logical components while the technical viewpoint deals with the deployment of software to hardware components. Orthogonal to the viewpoints the framework suggests using multiple granularity levels to allow for a divide-and-conquer approach [2]. This led to a variety of different domain-specific models, description languages and proceedings.

Schröck et al. [3] introduced an approach of using principles developed in SPES for the engineering of automated plants. They identified that several factors that influence the automation system, such as standards, environmental conditions (e.g., weather conditions), and business processes. Although most of the principles developed in SPES can already be applied in the automation software engineering of machines and plants, they pointed out that seamless model-based engineering is not fully applied to the automated plant business so far because of using different models and views in the domain [3]. The different viewpoints and granularity levels inspire the approach introduced in this paper.

SysML [4] is a general-purpose modeling language that has found widespread usage in systems engineering. It proposes the Block Definition Diagram (BDD) and the Internal Block Diagram (IBD) for structural modeling. BDDs define “Blocks” that have a name, properties (characteristics) and methods (behavior). Blocks can inherit all properties and methods from a generalized Block and extend them. Blocks can further be composed of a variable number of other Blocks. SysML supports references to other blocks in BDDs, however the information contained within those blocks cannot be linked to a property as needed. BDDs thus provide a structural overview of a system. IBDs in contrast provide a static view of the (internal) connections of a Block. These internal connections can describe physical flows such as signal-, or energy-flows depending on the concrete purpose of the created diagram. For the modeling of dynamic processes, activity diagrams, state charts, and sequence diagrams are used in the design process. Closely related to sequence diagrams are Unified Modeling Language (UML) timing diagrams, which are another type of interaction diagram that instead shows state changes of interaction partners resulting from events occurring [8]. UML Timing diagrams place a greater focus on the synchronicity and order of events as they occur, making them more suitable for modeling aPS behavior.

Additionally, UML and SysML include a profile mechanism that enables the extension and adaptation of the original UML metamodel. This profile mechanism facilitates the representation and thus reuse of domain-specific elements through the addition of stereotypes, constraints, and other syntax elements. Within the domain of aPS, different domain-specific metamodels and profiles based on them have been proposed. To include information of synchronized movements or motion profiles Süß and Diedrich proposed to store motion trajectories of a six-axis robot in a property of a SysML block that describes the robot [9]. For this reason, static properties were enlarged to represent pre-defined dynamic behavior. Barbieri et al. proposed to use SysML to couple the discipline-specific simulation models and tools [10]. SysML4Mechatronics explicitly included mechanical, electrical and software views and used discipline specific ports with properties that allow consistency checks [11]. These properties are limited to minimal and maximal values that might describe environmental constraints, ignoring the dynamic changes in technical properties in response to environmental changes, e.g. derating of motor.

The systematic SysML mapping study of Wolny et al. [12] revealed that most of the SysML papers focus on the design phase. The study selected 10 papers that extract SysML models for Simulink simulations, and identified lifecycle support and the modeling of hybrid systems as two required research directions: SysML’ addresses both needs.

To support collaboration on a system design model across different companies, departments, or engineers, a standardized classification of component characteristics is required, including the measuring unit and measuring environments in which these characteristics are valid. Classification systems like ECLASS [13] or REXS [7] seek to support cross disciplinary and cross-company suppler-customer collaboration especially by selecting a standardized set of properties for components used in the design phase to describe component characteristics. ECLASS serves as a classification system for products and services, with mapping to the United Nations Standard Products and Services Code (UNSPSC) [14], to enhance global procurement and supply chain management of components. REXS, on the other hand, is a standardized interface for the exchange of gear unit data. They have been introduced to develop an agreed upon terminology with well-defined semantics. Unfortunately, the semantics for the same properties are inconsistent across the standards. The standards have different scopes: Gears are not included in ECLASS, but in REXS, whereas motors are included in ECLASS historically but currently not in REXS.

The characteristics of gears in aPS are often determined for differently defined and non-comparable environmental conditions. In order to generate comparable values, definitions and boundary conditions must be defined as comprehensively as possible. As mentioned above, the standards to be met have different scopes and different sets of defined properties. An overview of the gearbox and component definition systems is presented in Table 1. For example, REXS encloses properties for a whole gearbox, while GDE [6] covers detailed geometry properties only for one gear. The gear’s main geometry may be given in GDE format, because many manufacturing machines and many quality control measurement centers work with that format natively. Geometry of shafts as well as type and details of bearings may be described only in REXS, since the other standards do not include those data and several software suits can read REXS. Finally, parameters like load and speed cycles may not be defined in a standard format at all. Overall, a full model of a gear pair including all specifics may consist of several hundred parameters, not all of which are defined in the same standard. Despite all possibilities the standards for mechanical components in a gearbox offer, the biggest drawback is their limitation to common designs. Most include cylindrical gears, some also bevel and worm gears, but there is no standard definition for cycloidal gears, strain wave gears or even more exotic ones. However, these designs are commonly used in aPS.

None of the previously described SysML modeling notations or the standards for describing aPS component characteristics as properties include reduced effectiveness of components in case of wear or environmental conditions like temperature or humidity. Recent developments in the field of digital twins make engineering information from datasheets readily available for modeling [15]. Consequently, SysML’ should also include information from datasheets. Datasheets enable to derive additional information like characteristic changes depending on temperature. So far, usage of SysML for aPS modeling in industry has been mostly limited to the design phase, except for the automotive and aerospace sectors. A SysML profile integrating datasheets and performance characteristics depending on environmental conditions may allow it to be used in later phases of lifecycle, for consistency checking and maintenance planning.

The conventional sequential design phase for aPS used to complete mechanical engineering first, followed by electrical and software engineering. Forwarding the designed documents and files like Automation ML from one discipline to the subsequent discipline. Of course this has much improved in recent years nevertheless interdisciplinary collaborative designs may still suffer from cross-disciplinary inconsistencies [16]. This leads to reengineering, prolonging the design process and reduces efficiency. As a countermeasure, Business Process Model and Notation (BPMN) 2.0 is a standardized method for representing business processes, which enables the depiction of workflow and information exchange for cross-departmental collaboration through graphical notation and is understandable across disciplines [17]. Zou et al. elaborated the general benefit of SysML-based systems engineering workflow [18] compared to a sequential document-oriented workflow still used in the one or the other company. They neither focus on the support in early design phases nor on components properties.

3 Adaption of SysML to SysML’ profile

Addressing the challenges from a class of continuous processes like paper production machines, continuous rolling mills or nonwoven lines that require a master-slave motion control, classical SysML needs to be enlarged to incorporate crucial aspects of gear-drives and selected extra-functional properties. Such extra-functional properties like temperature and altitude need to be include in mechatronic system models because the characteristics of gear-drives significantly impact the overall system performance throughout the system’s lifecycle. Classical SysML provides neither the necessary semantics to define the gear-drive model parts nor the temperature and altitude or humidity dependencies of motors in the system as required in machine or processes automation with master slave motion control as described above. Consequently, an adequate description must be included in a system model starting in early design phases as described in the following. To model these dependencies, SysML’ proposes to include gear and motor characteristics like REXS for gears, ECLASS for motors as SysML properties. These characteristics may for example also be series of curves. Often, these motor characteristics are not defined as ECLASS properties, but are only accessible from supplier documentations like operating instructions or online catalogs. Additionally, the UML, timing diagram is included in the concept of SysML’ to model the behavioral aspect of switching control modes as required in master slave control. The proposed SysML enrichment is targeting to SysML V1.6. It is described using a metamodel as proposed by SysML4mechatronics [11] and is introduced in Section 4. Additionally, a potential interdisciplinary design workflow is proposed in Section 5, modeled using BPMN. An overview of the proposed enrichments to SysML, the mechanism by which they were implemented, and the intended purpose is given in Table 2.

According to Kernschmidt et al. [11] different views of structural models are proposed: functional view (early design phase), electrical view (electrical engineering) and mechanical view (mechanical engineering).

3.1 Application example continuous production process with switching master slave control

To demonstrate the usage and validate the notation a continuous thermohydraulic press for timber products is introduced in the following. Two co-rotating conveying belts (labelled upper conveyor and lower conveyor in Figure 1) are controlled by separate electric motors that are synchronized through communication with a central processing unit one being the master and others being the slave that follows the control strategy of the master. The conveying belts can be controlled on either RPM (Revolutions per Minute) or momentum, depending on material infeed (raw mixture, left hand side, Figure 1) to avoid defects on the very thin metal conveying belts. If the material is not yet in the press the conveying belts are synchronized using Revolution per minute (RPM) to avoid surface damages due to contact of both belts. Upon detecting material in the infeed of the press, the control strategy is changed to momentum control to ensure the same momentum on the material from the upper and lower conveying belt. When the material exits the press output, the control mode returns to RPM control. This is a simplified example of two cooperating belts. The separate motors and gears for each belt need to be synchronized according to the control strategy (RPM or momentum) and selected in the design process considering temperature to name only one environmental constraint. Temperature is even more crucial here, as a continuous thermohydraulic press operates at about 240 °C, resulting in a temperature of about 50 °C at the motor position.

Figure 1:

Motivating application example continuous press with co-rotating conveying belts [19]. SC and PC are abbreviations for speed controller and pressure controller.

The functional view models the behavior of the continuous thermohydraulic press (cp. Figure 2) and aids the engineers in the early design phase. The electrical view focusses on the signal flow the delays occurring in the different components from the reading of the motor encoder to the motor set values (cp. Figure 3) and thereby the real-time characteristics. The mechanical view (cp. Figure 5) includes effects of selected environmental constraints like altitude and temperature that are also relevant in the electrical view, but only accessible as curves. A reference framework containing the proposed transformation mechanism, including information extraction, decomposition, and reconstruction, is included in the metamodel of SysML’ (cp. Figure 6).

Figure 2:

UML timing diagram detailing the switching control modes of the co-rotating conveying belts as the material (mat) reaches different parts of the press [19]. The solid arrows represent the alignment of time points at which the system components change control modes, while the dashed arrows represent the mode duration.

Figure 3:

IBD of electrical view of a motor and encoder showing the delay chain from measuring to operation according to [20]. Temperature and altitude effects are not modeled here but are shown in Figure 4 (left) and as references in the block properties of Figure 5 (Blue arrows added manually, indicating the delay time involved in the components in the delay chain).

3.3 Electrical view and transition to mechanical view

While the UML timing diagram (cp. Figure 2) describes the necessary control logic for switching operation modes from a functional point of view, it needs to be enlarged by the signal chain from sensor to actuator as basis for the synchronized switching of control modes. The sensor-actuator response time must be considered in the electrical view needs to include also the activation time of the mechanical parts of the actuator (cp. Figure 3).

The sensor-actuator response time describes the communication delay between detecting a changed process signal using a sensor and influencing the technical process with an actuator [20]. To control the conveying belts as described in Section 3.1, the motor output driving the conveying belts must be set depending on the material position in the press. The slave drive needs to precisely follow the encoder value of the master drive. The encoder value detection time includes not only the sensing time, but also the network delay from the bus coupler, the delay from the bus itself and the drive controller or PLC computation time (cp. Figure 3). The calculated set-value signal to the motor is similarly delayed. When validating the real time behavior of the mode switching of the master slave control the sensor-actuator response time must be considered. This can be done by indicating the relevant communication delays in an IBD of the electrical view.

The electrical components, as well as the gearbox components translating motor output to conveying belt movement are affected by environmental conditions. While the effect of humidity as an environmental condition remains relevant especially in decentralized control architectures, the focus of this paper will be mainly on the effect of temperature and altitude in central control architectures.

Different manufacturer provide different characteristics describing components’ environmental dependencies. An existing approach that attempts to solve this issue is ECLASS, which defines categories for many electronic/electrical components, each with a list of properties which manufacturers are expected to populate with their product specific data. In this way, comparisons between electrical components are supported, and design and operational constraints of the technical and functional views can be checked for consistency.

Unfortunately, in the ECLASS standard there are limitations on the included properties. The ECLASS advanced classification for the Low-voltage-three-phase current asynchronous motor, squirrel-cage rotor (27-02-21-01) specifies a category of properties for motor performance. Some properties defined here include “rated torque”, “rated rotation speed” and “rated power”, all describing the expected performance of the motor. However, ECLASS does not support linked properties, meaning that the quantification of the effect of temperature and altitude on torque or rotation speed is not included in the standard.

To still quantify the motors performance at high temperatures, manufacturer specific information like datasheets and catalogs must be consulted. In Figure 4 (left), two curves extracted from manufacturers design catalog [21] are shown. Factors f _T and f _H describe the motor power reduction, also commonly referred to as derating, when the motor is installed in an environment with temperature T and altitude H respectively. As an example, the installation of a motor in Mexico City is visualized in red. For Mexico City situated at an altitude of 2250 m with an ambient temperature of 50 °C in the summer, f _T is 0.87 and f _H is 0.88. For a rated Power P _N  = 22 kW the power must then be reduced to P _N,red  = 0.87 × 0.88 × 22 kW = 16.8 kW.

Figure 4:

Left Motor power reduction (derating) depending on ambient temperature (T) and installation altitude (H) [21]. Right: Mean friction factor for circumferential velocities, and wear per transferred work for different fluid greases [22].

The expression for calculating reduced motor power at higher temperatures and altitudes is only valid for longer time periods if maintenance instructions provided by the manufacturer are followed. For the motor introduced above, regular inspection interval every 10,000 operating hours is required according to the maintenance manual, which includes the cleaning of cooling air ducts [23]. If not followed, the buildup of dust in the cooling air ducts would result in higher operating temperatures which causes motor performance decay and eventual degradation of the windings and encoder due to increased temperatures. As such, maintenance requirements in the form of additional information need to be included in SysML’.

Since ECLASS properties are not available or insufficient for gears, the REXS standard is used to define all mechanical components inside the gearbox. That includes gears, shafts, bearings, sealings, lubricant and materials. Not only geometric values but also input values like load, speed, and output results like contact pressure are described.

The REXS standard has similar limitations to ECLASS, wherein static properties are not sufficient to describe behavior depending on the environment or long-term wear. Just as before, the REXS properties must be supplemented by gear curves from manufacturer datasheets. For an exemplary gearbox, the friction coefficient in tooth contact of the gears is described using a curve from a datasheet (cp. Figure 4, centre right), where the mean friction coefficient µ _m changes with the circumferential velocity at the pitch circle v _t . Additionally, µ _m depends on the type of lubricant, its temperature etc. The friction curve describes the relationship for a sump temperature of 90 °C. The combined wear rate (Figure 4 right) of two gears in contact does not only depend on the transferred work, but also on the viscosity of the lubricant (fluid grease) and the operating temperature (60 °C for the given curve). The friction coefficient at different temperatures as well as the wear rate of a gear mesh at different values of transferred work are available as curves in the suppliers’ datasheets. In general, the value pairs of the characteristic curves given in product datasheets are transferred in arrays of REXS properties. While the REXS specification allows to compare mechanical components, the encoding of an entire characteristic curve within properties is still a limitation.

Since using properties defined in the ECLASS or REXS standard is insufficient, the focus will instead be on integrating curves extracted from manufacturer catalogs into SysML.

3.4 Mechanical view including electric components

In SysML, BDDs are used to show system components and their properties. Following this intention, BDDs in SysML’ should include references to ECLASS and REXS properties, as well as information that cannot be found in either standards but has to be extracted from suppliers’ documentation, like the drive design and drive selection catalog, or maintenance policies.

A BDD showing the mechanical components of a motor drive for the co-rotating conveying belts (cp. Figure 5) has to include a) component characteristics that can be found in the ECLASS or REXS classification and b) referenced from a supplier information like design documents, datasheets and maintenance manual. As shown above, besides static property like minimal and maximal values, some properties that are environment or wear dependent require a reference to a gear curve or a power reduction curve (cp. Figure 5, highlighted blue boxes). In the block “Motor”, the property “reduced_power(Pn,red)” is influenced by both temperature and altitude as introduced in Section 3.3 (cp. Figure 5, right blue box). Similarly for the gearbox, the properties “mean_friction_factor” and “linear_wear_coefficient” have dependencies which are expressed in gear curves provided by manufacturers (cp. Figure 5, left blue box). For both examples, the set of curves are referenced in the mechanical BDD to refine the components’ mechanical description.

Figure 5:

BDD of mechanical view of drive system including frequency converter, motor and encoder (Highlighting added manually).

4 Metamodel of SysML’ – specifying the SysML’ profile

As an extension of SysML, SysML’ incorporates interdisciplinary artifacts (cp. Table 2), along with references to ECLASS, REXS (only for gears), and suppliers’ documentation, such as datasheets, maintenance handbooks and component catalogs. Its metamodel (cp. Figure 6) enlarges the metamodel of SysML4Mechatronics [11] by following the distinction in SysML4Mechatronics among different disciplinaries, additionally allowing access to referenced information like maintenance policies thus supporting collaborative design with increased consistency.

Figure 6:

Metamodel of SysML’ profile with omission of details in three views (left) and details of information transformation mechanism between suppliers’ documentation and ECLASS in reference (right).

Instances of the omitted details can be found in the figures in the preceding sections: “BBD Mechanical View” (cp. Figure 5), “IBD Electrical View” (cp. Figure 3), “Timing Diagram” (cp. Figure 2). The dependencies between packages of different views provide support for cross-disciplinary collaboration during the design phase, while the reference provides a source of information for mechatronic component design including constraints like curves or tables documented in suppliers’ documentation.

In the Advanced View of ECLASS, the different qualities of the product are organized in a hierarchical structure. For example, “Performance”, “Information for Use” and “Permissible Environment Conditions” (cp. Figure 6) predefined categories at the same hierarchy, whose subclasses contain detailed information that we use as a reference. The properties belonging to the different lifecycle phases of the product, i.e. the motor, are available in the various types of suppliers’ documentation. For example, the “Characteristic Curves” and “Rated Technical Properties” in Figure 6 belong to the product design phase, while the “Maintenance Policies” is affiliated with the product operation phase. Within the Package “Reference”, if the component to be specified can be identified in ECLASS, connections between suppliers’ documentation and ECLASS are established. Precisely, as in the aforementioned examples, between the properties of the product in its different lifecycle phases and the predefined classifications within ECLASS, either through

1. Direct access, the original information from the supplier is read and stored in a mapping-like manner, without transformation. The mapping form is utilized in case the nomenclature of ECLASS is inconsistent with the nomenclature of the same property in the suppliers’ documentation, such as from “Rated Technical Properties” in suppliers’ documentation to “Performance” in ECLASS; or through

2. Transformation, the original information from the supplier is transformed into a predefined format by utilizing a specific method of the transformer. For example, the motor’s derating curve (cp. Figure 4, left) found in “Characteristics Curves” in suppliers’ documentation is transformed into “SupplierName_ProductID_Transformed Derating Table” (an instance of “Tables Transformed from Curves”) and subsequently integrated into “Permissible Environment Conditions” within ECLASS.

Notably, the maintenance intervals and methods for a multitude of mechatronic components are considerable dependent on the operating environment, with temperature being a critical factor. Considering that such maintenance policies in suppliers’ documentation may be presented in a mixture of textual explanations, diagrams and tables, transformation methods may be required. For example, the instance “SupplierName_ProductID Relubrication Intervals” of “Maintenance Policies” in Figure 6 is considered to be a mixture of text and table to describe the re-lubrication policy of bearing, where the transformation mechanism should be a hybrid of “Text_to_table()” and “Table_to_table()” in this case. The format of the tables translated from the curves and maintenance policies is varied due to the different ECLASS classifications they need to be introduced into.

The Content Development Platform of ECLASS allows creation of product and edition of the product properties by users. Thus, various information extracted from different suppliers’ documentation can be integrated into standardized ECLASS classifications. Conversely, if components are not included in any ECLASS classification, the extraction of the necessary information and the decomposition of the curves need to be performed in the same transformation mechanism without being integrated into ECLASS to enhance machine readability. The tabular form of data enables the utilization of environmental indicators in a direct way, such as temperature and humidity, obtained through sensors in the aPS, to obtain accurate readings of environment-related properties.

This paradigm of information extraction, decomposition, and reconstruction for referencing in SysML’ enables the introduction of curve-based environment-related properties in a standardized, machine-readable and -useable format. Difficulties in information filtration and comparison arising from the diverse presentation conventions of suppliers can consequently be overcome due to the ECLASS-based reference, easing the mechatronic component selection. The export of standardized structural data from ECLASS benefits the design phase of model-based system engineering for aPS. Furthermore, the transformed and integrated information can be centrally managed and reused via ECLASS in the future, hence radiating the benefits of the proposed referencing mechanism over the entire lifecycle.

5 Usage of SysML’ profile for component selection in early design phase

Compared to the conventional design workflow, characterized by a lack of communication and knowledge among different disciplines, the model-based workflow using SysML’ shall address these shortcomings. The proposed SysML reference mechanism standardizes mechatronic component descriptions throughout different design disciplines, thereby supporting mechatronic component selection by ensuring design decisions of all disciplines are made on the basis of the same information. It can also alleviate the burden of inconsistency management, as the properties of mechatronic components in a discipline-specific model are derived by referencing a central SysML’ artifact repository.

Following Zou et al. approach [18] the benefit of the proposed SysML’ in the design workflow of mechatronic systems in the multi-disciplinary team is modeled using BPMN 2.0 (cp. Figure 7). Because SysML’ aims to support component selection in the early design phase by incorporating properties the BPMN model focusses on this engineering task.

Figure 7:

BPMN of SysML’ usage in design phase of aPS including requirements viewpoint, functional viewpoint, technical viewpoint consisting to SPES [2]. The task “Referencing” (orange in bottom) serves as a detailed explanation for the package “Reference” in Figure 6 in the form of workflow. Artifacts expanded in SysML’ are painted in different colors.

In engineering tasks, all artifacts and the reference mechanism in SysML’ make component functionalities and properties transparent to all views, effectively alleviating the limitations of a mono-disciplinary horizon. The typical functional requirement to switch between different modes of master slave control considering real time aspects is highlighted using the UML timing diagram. UML timing diagrams provide a more explicit blueprint for design work and BDD creation in the mechanical view. In the next step the real time aspects are examined by analyzing the signal delay chain modeled in the IBD of the electrical view. Once the requirements can be achieved the environmental constraints and components properties are in focus.

Before starting the design process, each discipline needs to perform the task “Referencing”, which includes two phases: check and information processing. In the check phase, the following steps need to be completed: check available artifacts, check ECLASS classifications and check suppliers’ documentation (green cylinders). Afterwards, the useful information obtained from the check phase needs to be processed according to a predefined mechanism, i.e. direct access or transformation. The way of storage and utilization of the processed information depends on the results of the task “Check Classification in ECLASS”. The final output of this referencing task is the referenced information in a standardized format, which serves as the basis for the subsequent design process, especially the component selection task.

After the mechanical view has generated the component list, if necessary, in cooperation with the electrical view, the electrical view can start working on the corresponding electronic component design, where the reference mechanism provides crucial assistance.

The artifacts from the view’s work steps are refined and updated on top of the existing SysML’ model. For example, the BDD and IBD produced in each work step are immediately saved in the SysML’ model and provided to all disciplines as the basis for their subsequent tasks (cp. Figure 7, the bolded black dashed arrows). More precisely, blocks designed by other disciplines can be easily inherited and refined when working in another discipline, owing to the fact that they are designed with the same standardized reference of SysML’.

As the constraint design task involves the properties of system components being of interest from mechanical, electrical and functional views, it is completed on the premise of interdisciplinary conversations illustrated by the hexagon (UML conversation symbol). The UML conversation symbol is used to model coordination and alignment processes with the different participants. They might include iterations until the constraints from all components involved are fulfilling the requirements. It is worth mentioning that with the SysML’ model, communication between disciplines exists not only in interdisciplinary conversations during the constraint design but is readily achieved through interaction with SysML’.

6 Summary and outlook

The selected mechatronic components of an aPS, particularly gear-drives and motors, impact the overall equipment effectiveness significantly. If the drives fail the aPS is down. Inspite of their profound influence, gear-drives’ properties and environmental dependencies of motors as two components of a drive are insufficiently considered from a systems engineering perspective in the early design phase of aPS. The properties of mechatronic components like gear and motor appear in a wide variety of semantic conventions and data formats some only in the suppliers’ documentation, which makes it challenging to consider them during the early design phase. Although with ECLASS a standard classification exists, it still lacks important environment-related characteristics, which are often documented in the form of curves.

To address these challenges, this paper proposes the profile SysML’ to assist with incorporating mechatronic component properties in the early design phase of aPS. SysML’ extends SysML in three ways: An extension with a referencing mechanism to import component properties from ECLASS, REXS, and suppliers’ documentation enables the inclusion of critical environmental constraints. Using this approach selected gear-drive characteristics and environmental dependencies of motors are available for interdisciplinary decisions in the early design phase. The described referencing mechanism aims to provide the referenced data machine-readable and semantically compliant with ECLASS. Additionally, SysML’ repurposes IBDs and BDDs for interdisciplinary design: by making these models available across different disciplines, interdisciplinary cooperation is supported and inconsistencies resulting from disjointed early design phases can be avoided. Finally, UML-timing diagrams are included to describe the required synchronized behavior of drives. Their inclusion enables to communicate important requirements from the functional view to technical views in an early design phase. The deployment of SysML’ in an interdisciplinary design workflow is demonstrated using a BPMN model, manifesting the potential of the reference mechanism for component selection assistance and inconsistency avoidance.

In future work, the automation of the proposed reference mechanism, especially in case of semantic ambiguities among classifications systems (e.g., ECLASS, REXS, GDE, and STEPS) and suppliers’ documentation, is planned. In addition, code generation and interfaces of SysML’ to other engineering platforms, like simulation software will be focussed. The use of machine-readable attribute tables obtained by reference in conjunction with physical constraints and sensed environmental indicators during system operation need to be implemented. Overall, subsequent work will focus on ensuring that SysML’ covers multiple disciplines of aPS design more comprehensively, leveraging the benefits it can bring in different phases of the system lifecycle, and validation the proposed approach by applying it to several application examples.