Available online at www.sciencedirect.

com

ScienceDirect
Procedia CIRP 50 (2016) 695 – 700

26th CIRP Design Conference

Design Library Solution Patterns in SysML

for Concept Design and Simulation
Benjamin Kruse*, Kristina Shea
ETH Zurich, D-MAVT, Engineering Design and Computing Laboratory, Tannenstrasse 3, 8092 Zurich, Switzerland

* Corresponding author. Tel.: +41 44 632 0891; fax: +41 44 632 1396. E-mail address: bkruse@ethz.ch

Abstract

Object-oriented models in the Systems Modeling Language (SysML) are developed in this paper to support the concept
development phase within engineering design. Generic libraries in SysML for functions, according to the functional basis, and
structural components, are presented in previous work by the authors. This paper extends this work and proposes the use of
multi-solution patterns in SysML that combine a new behavior simulation library together with the previous generic libraries
describing functions and components. These patterns capture coherent solutions to known problems that can be reused in concept
design with the aim to save modeling effort. Since they are based on solution-neutral functions, they also offer multiple potential
solutions at once. The new behavior simulation library and solution patterns are demonstrated in this paper using a 3D printer
case study with two different kinematic solutions.
© 2016
© 2016TheTheAuthors.
Authors. Published
Published by Elsevier
by Elsevier B.V. B.V.
This is an open access article under the CC BY-NC-ND license
Selection and peer-review under responsibility of Professor Lihui Wang.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the 26th CIRP Design Conference
Keywords: Model-Based Systems Engineering; SysML; Patterns; Design Libraries

1. Introduction evolving SysML and supporting MBSE [7]. To provide

additional means of reusing design knowledge while
Current trends in mechanical engineering show an enhancing the complementary usage of all three design
increasing growth of complexity in the systems under libraries, known solutions are described in solution patterns.
development. This includes not only the number of The contributions of design solution patterns are that they
components and interactions but also their functionalities, offer multiple known solutions at once to fulfill a target
which are often enabled by close interactions of different functionality thus taking advantage of solution-neutral
engineering domains [1]. To support multi-disciplinary design functional models. Depending on the selected solution,
in the conceptual design phase, Model-Based Systems extensive data is made available including further functional
Engineering (MBSE) can be applied by means of the general- decomposition of the target functions.
purpose Systems Modeling Language (SysML) of the OMG The used case study to demonstrate the behavior library
[2]. Generic libraries for functions [3], according to the together with the solution patterns is a Fused Deposition
functional basis [4], and structural components [5], are Modeling (FDM) 3D printer, based on the RepRap project [8].
presented in previous work by the authors. This paper The multiple solutions offered by the pattern are two different
contributes by extending this work to include a third kinematic systems for moving the print head. The tool
complementary design library for modeling system behavior Magicdraw (v18.1) is used for modeling in SysML.
in SysML. It reuses knowledge from the database of the After reviewing the related work in the following Section,
multi-physics simulation software Amesim from Siemens Section 3 introduces the general modeling approach. Section 4
LMS [6] to support modeling, while facilitating model and 5 describe the behavior simulation library and the multi-
formality and clear semantics. Model transformations between solution pattern definition in SysML. Afterwards, they are
SysML and Amesim then enable behavior simulation and used in Section 6 in the case study. In Section 7 the results are
evaluation. These aspects were identified to be important for discussed followed by a conclusion.

2212-8271 © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the 26th CIRP Design Conference
doi:10.1016/j.procir.2016.04.132
696 Benjamin Kruse and Kristina Shea / Procedia CIRP 50 (2016) 695 – 700

2. Related work its desired behavior by several aspects. These combined

aspects form a coherent system by correlating with each other
According to INCOSE, MBSE is “the formalized as part of partial models. This definition of solution patterns is
application of modeling to support system [...] design [and] used in this paper. Yet, in its implementation in SysML its
analysis [...] activities beginning in the conceptual design aspects are adapted and it is extended to offer multiple
phase” [9]. One approach to MBSE is SysML that provides a solutions to fulfill the same target functionality.
common and generic modeling language for its Other related work by Canedo and Richter [21] presents an
implementation. Being a general-purpose modeling language, approach to automate the allocation between functions and
SysML enables the modeling of a broad variety of Amesim simulation model components for direct concept
information during product development, e.g. requirements, validation. Yet, it does not specify how the knowledge about
functions, behavior and structure. It is object-oriented, these allocations is stored or how the internal compatibility of
standardized and widely applicable and also enables a more the automatically generated simulation components is
efficient reuse of design knowledge. Yet, it lacks in the ensured. This aspect of ensuring compatibility of partial
inclusion of executable physical relations, e.g. to verify simulation models is achieved by Münzer and Shea in [22],
whether requirements are fulfilled or not. where simulation models based on generated concept model
It is essential for simulation-based design [10] to allow for graphs are automatically generated.
the evaluation of design variants and reduce development Therefore, there is a research opportunity identified to
time while improving system quality and performance improve the computational support of the designer during
through early simulation [10]. One approach for modeling the concept design by providing existing generic knowledge in a
system behavior for simulation is the bond-graph standardized and formal modeling language.
methodology [11] as a graph-based, multi-domain modeling
paradigm that is based on the conservation of energy. One 3. General modeling approach
commercial tool that uses bond-graph representations for its
underlying mathematical model is Amesim [6], which is used The general MBSE approach used in this paper is based on
in this paper. Similar work includes the SysML-Modelica the VDI 2221 [23]. It starts by deriving a functional model
Transformation Specification [12] that allows the creation of from requirements and use cases. The resulting elementary
Modelica models directly in SysML. Its limitations are mostly functions are then realized by behavior and structure elements
regarding usability since it requires complex modeling to to create a principle concept. This is supported by generic
recreate a purpose-built modeling language within a generic libraries for functions, basic structural elements and now
one. Kerzhner [13] uses SysML as a domain specific language behavior elements to enable simulation-based design. The
for design space exploration. Yet here, due to being domain additional multi-solution patterns provide further modeling
specific, the knowledge bases with the complex simulation support by describing concept solutions in the form of partial
knowledge needs to be created for each different problem models that correlate library elements with other aspects to
anew. offer coherent subsystems.
A combination of functional, behavioral and structural
models for design comes from Umeda and Tomiyama [14], 4. Behavior simulation library
who introduce FBS (function-behavior-structure). A solution-
neutral functional model is used in FBS to first identify This Section presents the SysML library that contains the
matching behaviors that are then embodied by components in Amesim [6] simulation components. It makes the knowledge,
the structural model. Combining SysML with function-based which is encapsulated in the component database of a
design, Hutcheson et al. [15] focus on utilizing functional simulation tool, available in SysML. This way it can be used
models in SysML, similar to the modeling approach in this to support the planning of simulation models in SysML, e.g.
work. Functional modeling has the advantage of allowing a to evaluate system behavior already during concept design. To
decomposition of a system’s functionality into more simulate such a model, the capability exists in the simulation
manageable parts to support finding innovative solutions [16]. tool to import internal block diagrams (IBD) from SysML.
To support functional modeling in SysML, previous work by There a manual mapping between the part properties and
the authors [3] includes a generic function library for SysML corresponding elements of the tool’s database takes place. In
containing flows (nouns) and functions (verbs) of the the simulation tool, the model can also be further refined, if
functional basis [4], to be combined to create elementary the SysML model is only partially complete.
functions. This library is used in the following case study too. In the library, two stereotypes are defined: one for the
Patterns are defined as captured expert knowledge in the simulation components, called <<AmesimBlock>>, and
form of reusable solutions to known problems within their another one for the simulation models themselves, called
specific context, like for instance those in software design by <<AMESimSimulationModel>>, which contains simulation
Gamma et al. [17]. According to Cloutier and Verma they are components and further data for the simulation, e.g. its
“time-proven [means] in solving problems similar” [18] to the runtime or the time intervals. An example for a simulation
known problem. For engineering problems, recent work by component is given in Figure 1 with a rotating load element.
Anacker et al. [19, 20] defines solution patterns, which It has two interfaces, modeled as flow ports with reusable
describe domain spanning principle concept solutions. These types shown below. Using flow ports has the advantage of
solutions describe the basic operation mode of the system and better port compatibility checking in the SysML modeling
 Benjamin Kruse and Kristina Shea / Procedia CIRP 50 (2016) 695 – 700 697

tool. The port types have additionally to the flow properties

for the transmitted values for the simulation also a function
flow property that uses the flow elements from the function
library [3]. In Figure 1 the flow properties are the torque in
Nm, the rotary velocity in revolutions per minute, the rotary
angle in degree and the function flow property of rotational
mechanical energy. This function flow supports an
identification of suitable behavior simulation elements by
corresponding to the flows of functions. The AmesimBlock
elements also need a property for their submodel description,
because in the simulation tool a single model element can be
defined by multiple mathematically different submodels.
The relations between seemingly identical simulation
elements that have different submodels are modeled by
generalizations in SysML, with a more generic element as Fig. 1. Excerpt of behavior library: On top is a rotating load element in the
parent to all its possible submodels. The generic parent simulation tool database [6], in the middle is the equivalent AMESimBlock in
element allows the causality of the elements to stay undefined SysML with the two interfaces as flow ports and one displayed property for
in SysML. This is important because of Amesim’s bond-graph its moment of inertia and below is the used port type with its flow properties
origin, where the causality can change when new elements are
added. To avoid this issue there exist rules for the creation of
simulation models in [22] where more details are given. These
rules also ensure compatibility of generically usable partial
simulation models in SysML as in the pattern. An example is
the need for an additional shaft element with its internal spring
and damper between the rotating load element of the 3D
printer’s driving pulley and an added motor. Without this shaft
element there would be no compatible submodels to connect
these two simulation elements.
Fig. 2. “Design Pattern” and “Pattern Solution” stereotypes defined in SysML
To use the library in SysML, AMESimBlock elements that
redefine their inherited properties from library elements are
To model different pattern solutions a second stereotype
connected on IBDs of their AMESimSimulationModel
<<Pattern Solution>> is defined in Figure 2 (right). It contains
elements. The icon of the library element that comes from the
the solution description, resulting context, examples, design
tool database, as highlighted in Figure 1, can be used to
rationale, known uses and the model-based solution itself,
visualize the individual parts with their ports on the IBDs.
consisting of requirements, a functional model, a behavior
model, a structural model and the interfaces. These pattern
5. Multi-solution patterns
solution models cover only their specific solution and are only
partially complete in the context of the whole system.
To facilitate a combined usage of the different design
libraries and to capture more design knowledge about generic
6. Case Study
solutions, patterns are defined in SysML. They describe the
realization of these target functions by several aspects that
The FDM 3D printer case study uses two different
form a coherent system by correlating with each other as part
solutions for the two-dimensional positioning of the print
of the partial models of the pattern. With this they interrelate
head via one or two belts. The belts are driven by stationary
the context, a problem and multiple solutions together with
stepper motors and move the print head on top of a sliding
documentation and usage guidance. They contain among
carriage by means of linear bearings. The Hbot [24] design is
other things solution-dependent further decomposed
simpler with only one belt but an expected higher orthogonal
functional models as well as allocated structural and behavior
load on the linear bearings of the sliding carriage. The
models for simulation. Their aspects follow the frameworks
CoreXY [25] design balances the forces on the print head
for engineering patterns as defined in [19, 20] and in [18].
better but has a more complicated set-up with two crossing
They are refined to allow the inclusion of multiple solutions
belts. Both solutions are known in the RepRap community.
in their implementation in SysML, while utilizing the
Their kinematic schemas are shown in Figure 3. The used “2D
function, behavior and structure libraries. For their
Kinematics” pattern of the case study is partially displayed in
implementation in SysML the package stereotype is extended
Figure 4 with pruned descriptions for the properties defined in
into a new <<Design Pattern>> stereotype, as seen in Figure 2
Figure 2. The inherent modeling elements are also displayed.
(left). The aspects of the pattern are defined by the properties
There are for example the two pattern solutions, the “HBot
of the stereotype for name, alias, keywords, problem context,
Solution” and the “CoreXY Solution”, to fulfill the same
problem description, the target functionality by means of
target functionality of the pattern: positioning the liquid
actions, forces, related patterns, references and the pattern
plastic, measuring its position and converting the needed
solutions.
energy. These functions constitute the pattern.
698 Benjamin Kruse and Kristina Shea / Procedia CIRP 50 (2016) 695 – 700

Fig. 3. CoreXY [25] (left) and HBot (right) kinematic schemas

Fig. 5. Replacement of the pattern’s target functions (top) by

partial functional model (below) at pattern application of HBot solution

6.2. 3D Printer case study results

Fig. 4. Excerpt of used multi-solution pattern in SysML

It was observed by the authors that the use of solution
with two solutions for two-dimensional movement of the print head. patterns allows for faster modeling of different concepts by
A few corresponding model elements are displayed on other figures. making existing knowledge accessible in the form of partial
models that provide predefined sets of elements from the
6.1. Pattern application design libraries for functions, behavior and structure, together
with more case specific data, e.g. requirements. Thus, since
When applying one of the solutions from the pattern, the this knowledge is re-used, fewer modeling steps are needed.
existing models within the pattern are directly copied into the An excerpt of the provided behavior simulation models can
system model. This is shown in Figure 5 with the functional be seen in Figure 6. It shows one pulley as a passive sheave
model since it is used to identify suitable patterns by the target element with its inertia modeled by a rotating load element
functions. In case of matching functions with the right context from Figure 1 together with a zero torque source for its open
in the model, these target functions are first replaced by the end. The sheave is fixed to the printer’s frame represented as
refined and decomposed functional model. a fixed body in the reference frame that serves as a zero
The functional model contains functions that are further acceleration, velocity and displacement source.
decomposed on other diagrams, e.g. “Position Liquid” and To create the simulation model, the default values of the
auxiliary functions, e.g. “Electrical Energy : Distribute”, properties of the library elements are redefined: for example
which are additionally required for the selected solution. the “Rotating Mass” element from Figure 6 is a sub-type of
Finally, the functional model needs to be manually extended the “rotaryload2_02A” element from Figure 1 with a mass
to also include the 3D printer’s third axis of movement, moment of inertia of 0.005 kgm² instead of the default 1
usually realized by moving the print table. The other aspects kgm². This represents a small pulley. To create a complete
of the solution in the pattern, e.g. structure and behavior, are simulation model, a simple controller without feedback and
handled similarly. approximated stepper motor models are added to the partial
kinematic behavior models from the pattern.
 Benjamin Kruse and Kristina Shea / Procedia CIRP 50 (2016) 695 – 700 699

Fig. 7. Simulation results of the HBot and CoreXY solutions, showing the
Fig. 6. Excerpt of simulation model in Amesim (left) and in SysML (right)
orthogonal forces on the linear bearings of the sliding carriage.
for a pulley, as part of both kinematic solutions from pattern

In the simulation, the print head drives in a circular path Comparing it to using only SysML parametric diagrams
for a total of 12 seconds and with time intervals of 0.0005 together with a solver to simulate the system behavior, the
seconds. The simulation results, as shown in Figure 7, reveal library enables knowledge re-use through the existing
that the force on the linear bearings orthogonal to their database of complex simulation elements. In general,
allowed movement is about twice for the HBot solution than integrating multi-physics simulation models early during
for the CoreXY solution. The force peaks for the HBot at concept design enables trade-offs between different design
around ±6 Newton while the CoreXY configuration only parameters as well as the evaluation of different concepts for
reaches ±3 Newton. The graphs appear solid since positive performance and behavior, as shown for the orthogonal load
and negative values are reached almost simultaneously due to on the bearings.
the approximated stepper motors that have inconsistent step- The use of multi-solution patterns in SysML offers
wise provision of angular momentum resulting in very quick multiple advantages compared to existing solution patterns
load changes. [19, 20]. Having multiple solutions in the pattern supports the
As expected, having a higher load on the bearings of the concept that the starting point of the pattern application, a
simpler HBot design results in a need for bearings that are functional model, should be independent of how it is realized.
more robust against orthogonal loads compared to the Therefore, it allows multiple different embodiments. Offering
CoreXY design to reach similar printing performance. Having the designer multiple known solutions at once and including
these results, a first evaluation of the different concepts is basic simulation, can facilitate a selection of a suitable
provided along with a trade-off between different design solution for the given system during concept design.
parameters is possible, e.g. by refining the requirement for Including functional decomposition of the target
“orthogonal load on linear bearings”, as seen in Figure 4. functionality in the pattern has multiple benefits. It enhances
the model consistency by allowing the functional description
7. Discussion to stay on a comparable level of abstraction [20] as the
behavioral and structural aspects after applying the pattern.
The behavior library presented uses existing elements from a This results in more detailed traceability and documentation.
commercial database and thus is more directly applicable to Adding the auxiliary needed functions with their interfaces
integration with a simulation engine, in this case Amesim, but into the pattern can help to select further fitting solutions that,
offers less modeling freedom compared to SysML4Modelica e.g., depend on the existing flow of electrical energy in the
[12]. Compared to creating the simulation models alone in the system. This is also important when trying to merge duplicate
Amesim tool, modeling in SysML offers better traceability by functionalities, for example the import and conditioning of
linking directly to other aspects of the concept model. Yet, the electrical energy at multiple places in the model. Having a
modeling in SysML lacks regarding information about more detailed functional model in general can be used to
compatibility between elements when their causality is identify other functions where a related pattern might apply,
involved. Traceability is achieved by linking the Amesim e.g. to fulfill the now needed conversion of electrical into
elements in SysML with other SysML elements, for example rotational energy by stepper or servo motors.
to relate an electric motor element to its requirements, which Finally, the pattern supports a faster creation of a concept
function it fulfills and which concrete component it is in the model by reusing knowledge in the form of partial models.
structural model. This also helps with the model consistency Compared to having only working principles in [19, 20], the
since object-oriented model elements that are used in different patterns in SysML with their solution-specific partial behavior
places of the SysML model can have only one consistent models can be quickly adapted or refined for simulation, e.g.
definition in the model. The necessary model transformations by adding needed motors and including control algorithms.
between SysML and the Amesim tool are currently an import The effort for creating such patterns is ideally reduced by
of the SysML IBDs into the simulation tool where a manual reusing parts of existing models, which incorporate prior use
mapping is made. of the known solutions that proved to be successful.
700 Benjamin Kruse and Kristina Shea / Procedia CIRP 50 (2016) 695 – 700

For the use of SysML for modeling and the capture of [2] OMG. OMG Systems Modeling Language (OMG SysML). Version 1.4.
patterns themselves, several findings are noted. Being an No. formal/2015-06-03. 2015.
[3] Kruse, B., Münzer, C., Wölkl, S., Canedo, A., and Shea, K. A Model-
object-oriented modeling language, SysML allows extensive Based Functional Modeling and Library Approach for Mechatronic
data to be reused. As a general purpose modeling language, it Systems in SysML. IDETC/CIE 2012. Chicago, IL, USA; 2012.
claims to be powerful enough to cover complex relations [4] Hirtz, J., Stone, R. B., Mcadams, D. A., Szykman, S., and Wood, K. L. A
across multiple domains, e.g. for modeling mechatronic Functional Basis for Engineering Design: Reconciling and Evolving
systems. But, despite being standardized and known in Previous Efforts. No. 1447, Washington, D.C., USA: NIST. 2002.
[5] Wölkl, S. Model Libraries for Conceptual Design. Ph.D. thesis, Munich,
industry, it is often challenging for novice designers to cope Germany: Technische Universität München. 2012.
with such broad modeling freedom. The presented modeling [6] Siemens PLM Software. LMS Imagine.Lab Amesim. 19.01.2015.
approach supports designers with patterns and libraries that http://www.plm.automation.siemens.com/de_ch/products/lms/imagine-
could be used as a starting point for modeling. lab/amesim/
Future work includes extending the modeled solutions [7] Friedenthal, S., and Burkhart, R. M. Evolving SysML and the System
Modeling Environment to Support MBSE. INCOSE, Insight, 2015. 18:2,
patterns as well as including software and control aspects p. 39-41.
within them for mechatronic systems. It also includes [8] Jones, R., Haufe, P., Sells, E., Iravani, P., Olliver, V., Palmer, C., and
demonstration of the approach, libraries and patterns in Bowyer, A. Reprap – the Replicating Rapid Prototyper. Robotica, 2011.
industry that is currently underway. A final step is the semi- 29:01. p. 177-191.
automatic application of the multi-solution patterns in the [9] Haskins, C. Systems Engineering Vision 2020. No. INCOSE-TP-2004-
004-02, INCOSE. 2009.
modeling tool that would enhance their usability, analog to [10] Bossak, M. A. Simulation Based Design. Journal of Materials Processing
the software patterns from Gamma et al. [17] that are already Technology, 1998. 76:1–3. p. 8-11.
implemented in the SysML modeling tool. [11] Borutzky, W. Bond Graph Modelling and Simulation of
Multidisciplinary Systems – an Introduction. Simulation Modelling
8. Conclusion Practice and Theory, 2009. 17:1. p. 3-21.
[12] Paredis, C. J. J., Bernard, Y., Burkhart, R. M., Koning De, H.-P.,
Friedenthal, S., Fritzson, P. A., Rouquette, N. F., and Schamai, W. An
Patterns and libraries are well established approaches in Overview of the SysML-Modelica Transformation Specification.
object-oriented modeling for reusing knowledge. Yet, during INCOSE International Symposium, 2010.
the conceptual phase in engineering design they are rarely [13] Kerzhner, A. A. Using Logic-Based Approaches to Explore System
used, even when they can provide not only single model Architectures for Systems Engineering. Ph.D. thesis, Georgia, USA:
Georgia Institute of Technology. 2012.
elements but also highly-interlinked, multi-viewpoint design [14] Umeda, Y., Takeda, H., Tomiyama, T., and Yoshikawa, H. Function,
knowledge. Therefore, this work contributes by creating a Behaviour, and Structure. IEA/AEI-90. Boston, MA; 1990. 1. p. 177-193.
new library for simulation elements in SysML together with a [15] Hutcheson, R. S., Mcadams, D. A., Stone, R. B., and Tumer, I. Y.
new multi-solution pattern that reuse elements from generic Function-Based Systems Engineering (Fuse). ICED’07. 2007. p. 28-30.
function, behavior and structure design libraries. These multi- [16] Saunders, M. N., Seepersad, C. C., and Hölttä-Otto, K. The
Characteristics of Innovative, Mechanical Products. Journal of
solution patterns contain a selection of known solutions for a Mechanical Design, 2011. 133:2.
certain functionality following the principle that for a [17] Gamma, E., Helm, R., Johnson, R., and Vlissides, J. Design Patterns:
solution-neutral functional model often multiple different Elements of Reusable Object-Oriented Software. ISBN: 0201633612.
realizations exist. The FDM 3D printer case study is Addison-Wesley Professional; 1994.
illustrated by two different kinematic solutions for positioning [18] Cloutier, R. J., and Verma, D. Applying the Concept of Patterns to
Systems Architecture. Systems Engineering, 2007. 10:2. p. 138-154.
the print head. It shows the benefits of reusing components [19] Anacker, H., Dumitrescu, R., and Gausemeier, J. Design Framework for
and their relations as partial models, to improve model the Integration of Cognitive Functions Based on Solution Patterns. In:
consistency and traceability as well as the integration of Design Methodology for Intelligent Technical Systems, Springer,
simulation to provide early evaluation of concepts. Dordrecht; 2014.
[20] Anacker, H., Schierbaum, T., Dumitrescu, R., and Gausemeier, J.
Solution Patterns to Support the Knowledge Intensive Design Process of
Acknowledgements Intelligent Technical Systems. ICED’13. Seoul, Korea; 2013.
[21] Canedo, A., and Richter, J. H. Architectural Design Space Exploration of
The authors thank Clemens Münzer, EDAC, ETH Zurich, Cyber-Physical Systems Using the Functional Modeling Compiler. 24th
for supporting the modeling and providing an initial Amesim CIRP Design Conference. Milano, Italy; 2014.
model of the CoreXY solution for the pattern. [22] Münzer, C., and Shea, K. A Simulation-Based CDS Approach:
Automated Generation of Simulation Models Based from Generated
Concept Model Graphs. IDETC/CIE 2015. Boston, MA, USA; 2015.
References [23] VDI-Gesellschaft. VDI 2221: Methodik Zum Entwickeln Und
Konstruieren Technischer Systeme Und Produkte. No. 03.100.40. 1993.
[1] Beihoff, B., Oster, C., Friedenthal, S., Paredis, C. J. J., Kemp, D., [24] 3D Printers Sp. z o.o. Hbot 3D Printers. 05.02.2016. http://hbot3d.com/
Stoewer, H., Nichols, D., and Wade, J. A World in Motion – Systems [25] Core(X,Y). 05.02.2016. http://corexy.com/
Engineering Vision 2025. INCOSE. 2014.