applicability of sysml to the early definition phase of space ......methodology: a mbse methodology...
TRANSCRIPT
1 Challenge the future
Applicability of SysML to the Early
Definition Phase of Space Missions
in a Concurrent Environment
Dorus de Lange (TU Delft) Jian Guo (TU Delft) Hans-Peter de Koning (ESA) CSDM 2011 Conference 2011-12-09
2 Challenge the future
Presentation Content
• Introduction
• Study Context
• Thesis Approach
• Engineering Approaches
• Concurrent Engineering
• CDF
• MBSE
• SysML
• Methodology
• Model
• Case Study
• Evaluation
• Conclusions
3 Challenge the future
Study Context
Majority of SE community sees Model Based Systems Engineering (MBSE) using
Systems Modeling Language (SysML) as the next step in managing
complexity of design projects.
Main Thesis Question:
“Can SysML add value to the studies performed in the CDF”
My experience at the Concurrent Design Facility (CDF)
• 4 month stage, assistant SE in 2 CDF studies: familiarize with CDF
• 3 month literature study: familiarize with subject
• 6 month thesis: perform research
4 Challenge the future
Thesis Approach
The following main tasks were performed:
• Study MBSE, SysML and MBSE Methodologies
• Prepare a methodology that fits the CDF environment
• Prepare a model using SysML containing a template to be used
for CDF studies
• Perform a case study (real CDF study)
• Perform evaluation
5 Challenge the future
Traditional Engineering Approaches
Centralised design
Sequential Design (“over-the-fence” approach)
6 Challenge the future
Concurrent Engineering
• structured process, • physical environment enabling the design team to co-locate, interact, and
communicate effectively, • dedicated multi-disciplinary team comprised of space vehicle designers, mission
specialists, subcontracts, pricing, quality assurance, manufacturing and production operations, all with well defined responsibilities,
• software and hardware infrastructure, • an Integrated Design Model, • active involvement of the customer.
Concurrent design
7 Challenge the future
Concurrent Design Facility
8 Challenge the future
Pre -
Ph. A SPEC. ’ SPEC. ’ SPEC. ’ s FDIR
Lessons Learned
Review Review Pre -
Ph. A
Phase A Phase B Phase C/D
SPEC. ’ s SPEC. ’ s CDR
Lessons Learned
…
…
Evolving Applications
Launch
= CDF application / quantity
Review Review
150+ 25 25 3 2
I n d u s t r y
E S A
Focus of CDF
9 Challenge the future
Concurrent Design Facility
10 Challenge the future
Concurrent Design Facility • Conducted in sessions
• Plenary meeting where representatives of all space engineering domains participate from early phases (requirement analysis) to end of design (costing)
• 6 to 10 sessions / study — 4 hours / session bi-weekly frequency
• Team leader co-ordination • Customer participation
• Model driven • On-line design • Highly co-operative & interactive • Iterations • Design options comparison and trade-offs
• Critical element in the decision making process of the agency!
11 Challenge the future
Model Based Systems Engineering
Key aspects:
• One central information model containing all design details,
• Potentially executable,
Commonly Expected Benefits:
• Re-usability of models results in a decrease in project time,
• Increase in consistency is observed,
• More precise communication resulting in more to-the-point,
more streamlined and faster discussions,
• Increased problem/systems understanding,
• Faster and more consistent means of documenting.
12 Challenge the future
MBSE Taxonomy Overview
SysML
MagicDraw
Developed in Thesis
13 Challenge the future
SysML
SysML
14 Challenge the future
SysML
SysML in short:
– SysML is a graphical modelling language for the modelling of combined
soft/hardware systems and all aspects involved in the life cycle of
systems,
– SysML is a standardized language by OMG, from the specification:
• "SysML is intended to be supported by two evolving interoperability
standards including the OMG XMI 2.1 model interchange standard for
UML 2 modelling tools and the ISO 10303 STEP AP233 data
interchange standard for systems engineering tools.“
– Based on UML 2,
– Uses three main “things”: Model elements, relations and diagrams.
15 Challenge the future
SysML – Model “Things”
Diagrams
Model Elements
Relations
16 Challenge the future
SysML
–All model elements and relations are defined once, but shown
many times in different diagrams to create different views on the
system,
–SysML contains nine diagram types.
17 Challenge the future
Methodology
Methodology: A MBSE methodology can be characterized as the collection of related processes, methods and tools used to support the discipline of systems engineering in a “model-based” or “model-driven” context*. *[J. Estefan, 2008]
18 Challenge the future
Methodology Definition Process A very important aspect in effectively applying MBSE is the definition and usage of a well
defined MBSE methodology.
Used approach:
• Study existing MBSE methodologies
• INCOSE OOSEM (Object-Oriented Systems Engineering Method)
• IBM Telelogic Harmony-SE
• MBSSE (Model-Based methodology to support the Space System Engineering)
• Evaluate potential for CDF environment
• Phase 0
• Concurrent
• Current infrastructure (tools)
• Define CDF methodology
• Use activities from existing methodologies
• Add CDF specific activities
19 Challenge the future
Methodology Overview
20 Challenge the future
Model
...also known as the CDF MBSE SysML Study Model (CMSSM)
21 Challenge the future
Model Modules
• Methodology: Description of methodology activities
• Profile: Profile to SysML, specific to the CDF
• Library: Storage of re-usable model elements
• Template: Standard structure for future CDF studies
22 Challenge the future
Model - Profile Profile contains model elements and relations, specifically for use in CDF.
Comparable to a dialect in a normal language.
Profile contains:
• 20 Model element stereotypes
• 3 Relation stereotypes
Trade Option
Study Files
Function
23 Challenge the future
Model - Library
Container for re-usable elements, studies and other useful items
24 Challenge the future
Model - Template
25 Challenge the future
Case Study - NEMS
Case study is used as a means to evaluate the developed Methodology and Model. Methodology, Model and case study are a means to evaluate the use of MBSE using SysML in the CDF.
Near-Earth Exploration Minimum System (NEMS) CDF Study used as case study.
• Mission Statement: Transfer a crew of 3 astronauts from the Earth to an
accessible Near-Earth target beyond the Earth-Moon system (e.g. NEO), perform a
rendezvous with the target, stay for 5 to 15 days to perform EVAs and safely return
the crew back to the Earth before the end of 2030.
• Kick-Off at 23rd of March, Final Presentation at 20th of April,
• Modelling work done as “System Modeller”, next to the team leader, SE and
assistant SE.
26 Challenge the future
• 200+ Diagrams
• 10.000+ Model elements
• 125+ Requirements
• 5 Spacecraft Modules on conceptual level
Case Study - NEMS
27 Challenge the future
Evaluation - Method
Evaluation Methods:
• Evaluation based on a questionnaire, distributed to the CDF Team,
• Evaluation based on modelling experience,
• Analysis of the statistics of the model and the modelling itself,
• Analysis of difficulties in the modelling process.
28 Challenge the future
Evaluation - Questionnaire
–The questionnaire contained an introduction into the thesis work and
SysML, after which interviewees were guided through the model
(using MagicDraw Reader) that was made for the case study and
were asked to provide scores to statements, including rationale.
– Included 19 statements/questions
–Focus of the questionnaire:
–Using the model in communication to the team
–Readability of the model
–SysML in general
–Tool speed and graphics
–Methodology: what to model and what not
– Identifying next steps for the CDF
–Completed by 7 of the CDF section’s SEs/trainees
29 Challenge the future
Evaluation – Questionnaire
Example (1/2)
30 Challenge the future
Evaluation – Questionnaire
Example (2/2)
Average score: 3 Standard deviation: 1.3 Summary of Rationale: It is useful that the information is linked but in communicating to the team PowerPoint is better
0
1
2
3
4
Nr
of
rep
lies
1 2 3 4 5
PowerPoint better - > SysML Model better
Example Question Scores
31 Challenge the future
Evaluation - Processing Questionnaire Results (1/2)
32 Challenge the future
Evaluation - Processing Questionnaire Results (2/2)
33 Challenge the future
Evaluation - Summary Opinions in general very varied (large standard deviation in score) Summary of evaluation: • Trade-offs, study assumptions and decisions including rationale can be captured in the model to
provide the relation to the requirements and increase traceability, thus adding significant value, • Modelling the study/mission/system provides the modeller with an overall better understanding
due to structured and systematic approach. • Transition phase to SysML will be difficult/requires considerable effort, • SysML does not provide sufficient support in variant modelling, • Graphical presentation provided by MagicDraw is limiting, • Model is less effective in communication to the team as compared to PowerPoint, • Diagrams export to the traditional report is not easy. Not necessarily bad, same problem for other
models (CAD, thermal etc). • As the IDM is present in the CDF, no added value is achieved by manually modelling the physical
architecture properties, as long as no automatic link between IDM/OCDS and SysML exists, • The significant effort of introducing executable SysML likely will not be justified by the resulting
added value for the Phase 0 studies performed in the CDF. Likely more suitable from Phase A onwards.
34 Challenge the future
Conclusions
MBSE using SysML is in line with the concurrent
engineering approach.
The MBSE methodology using SysML has potential but for now
has only limited applicability to Phase 0 studies. A
significant development effort of the software infrastructure
could increase the applicability.
For now, MBSE using SysML can add value by using it next to
the IDM to model the requirements and trade-offs. When the
graphical capabilities of MagicDraw improve, it could slowly be
used in communicating to the team, thus making the other
activities and diagrams worth modelling.
35 Challenge the future
Recommendations
• A more in-depth study on the potentials of executable SysML
for the CDF is highly recommended,
• Evolution of methodology and model/template on the
basis of the current work,
• A second case study focussing on requirements and trade
space modelling, by one of the CDF SE’s is recommended,
• Continued involvement in SysML by ESA and in particular
the CDF. It is a young language and now is the time to shape
the language.
37 Challenge the future
Back-up Slides
38 Challenge the future
Back-Up Content
• SysML and concurrent engineering
• SysML and Phase 0
• SysML Evaluation
• MagicDraw Evaluation
• Modeling History
• Requirement Status
• Relation Stereotypes
39 Challenge the future
SysML and concurrent engineering
Application of MBSE using SysML will strengthen to some extend three of these main elements, namely: "structured process", "integrated design environment" and to some extend "software and hardware infrastructure".
MBSE using SysML provides for a standardized language that ensures clearer, more streamlined communication. Also, modeling the study provides the modeler with a better understanding of the problem and of the design process. This supports the “structured process” element.
The model acts as a central information model and can potentially be integrated to the domain specific tools to orchestrate these tools. This results in an improvement to the “integrated design environment“ CE element.
Finally, the "hardware and software infrastructure" element of CE is strengthened by the integration of the SysML model as central information model with interfaces to the domain specific analysis tools.
All-in-all, MBSE using SysML is very much in line with the concurrent
engineering approach.
40 Challenge the future
SysML and Phase 0 (1/2) For MBSE using SysML five benefits were identified (slide 11) • The benefit of re-usability highly depends on the variance in the studies that
are performed. For the CDF, the variance in the types of studies is very large, thus decreasing the potential for re-use. Some of the recurring model elements can be re-used which will provide significant time savings. Re-use of executable model parts is less likely as these parts are in general more detailed. For a company performing quick feasibility studies with more similarities between the studies (on subsystem level for example), the benefit of re-use could be stronger and executable SysML could pay off.
• Consistency is always an issue in Phase 0 studies. Design options can be identified, named and eliminated within in an hour. It is unlikely the MBSE using SysML can improve this aspect for this particular design phase. In general, Phase 0 A can be seen as a long, chaotic brainstorm session. Attempting to structure this better by applying MBSE using SysML will likely be unsuccessful, resulting in too much time spend on modeling.
• As for the more precise communication provided by the SysML language. This is likely true, but can not be confirmed by the thesis work as the modeling of the case study was mostly a one-man job.
41 Challenge the future
• Although not measurable, it is concluded that the modeling activity itself resulted in an increased understanding of the customer's needs, the requirements, the mission, the system and the solution space. Also, from the model itself it became clear what the weak points of the study were and where more work was to be done.
• The final benefit that is mentioned is a faster and more consistent means of documenting. This can be true can not be confirmed based on the case study itself as only a part of the study was modeled. A model has the potential to be a more consistent means of documenting, when it is fully integrated with all other design tools. Only when the model is used as the central information model, the consistency aspect is true. This requires integration with all team members and their tools, which will likely not be reached in the near future for large studies with over 20 experts. As for the benefit of having faster documentation, this could be true when the modeling activity is highly automated and no traditional report is to be produced manually as well. Not having a traditional report will likely take more than a few decades as the engineering culture is based on it. Management will likely always want to have a physical written report. Thus requiring automated report generation from the model itself.
It can be concluded that the MBSE methodology has potential but for now has only limited applicability to Phase 0 studies. A significant development effort of the software infrastructure could increase the applicability.
SysML and Phase 0 (2/2)
42 Challenge the future
SysML Evaluation
• The language has a steep learning curve, significantly penalizing the
transition to MBSE,
• Language is flexible and allows to model all the systems aspects,
• Language is standardized, resulting in an increase of consistency in
complex projects and better communication,
• Requirements diagram quickly becomes useless when dealing when
dealing with 100+ requirements, and becomes useless when dealing with
only a few requirements when the relations are obvious.
• Relations could be made more readable. A relation such as "Allocate"
could be renamed "Is allocated to" to make the relations more readable for
non-SysML experts. The "Satisfy" relation could also be interpreted in twofold,
namely as "Satisfies" and "Should Satisfy".
It can be concluded that SysML is a good step towards the standardization of
MBSE, but it for sure is not the last step.
43 Challenge the future
MagicDraw Evaluation
• Tables and matrices need to be improved by adding more flexibility by
adding more options for the user. Sorting, zooming, fixing the export function
of the generic table which sometimes leaves out some of the attributes,
• Overall graphics need to be improved, especially the rendering of the
text in the symbols,
• Introduce more basic drawing skills, starting with functionality similar to
the PowerPoint "AutoShapes". This allows for making better drawings and
thus being able to use the model in communicating to the team,
• Quick layout functions needs to include more user options (for
example, distance between symbols) to provide for faster modeling,
• Reasonably stable tool with in general an acceptable speed.
Although there are still some weaknesses, MagicDraw is a mature tool for
SysML modeling. However, the identified weaknesses need to be diminished
before it can be used to its full potential for MBSE in the CDF.
44 Challenge the future
Modelling History
45 Challenge the future
Back-Up Content – Requirement Status
46 Challenge the future
Back-Up Content – Relation Stereotypes