I’m currently modelling a large and complex transport system.
So it doesn’t become unwieldy, I’d like to model different parts of the system in different projects and create functional links between them.
However, this doesn’t seem to be an option.
Does anyone know if (and ideally how) it may be possible to do a large system over multiple projects?
Thanks
Stu
This has the potential to actually be a huge question
First, from a technical point if view:
You cannot have direct links between Capella projects. Meaning, you cannot have a function in a project and its allocating component in another one.
The only possibity to have direct references is from projects to libraries, and it is about referencing data and interface elements.
The only way to share more than data and interfaces is through the REC-RPL mechanism (synchronized copies)
Then, from a methodological standpoint
When I read your question, I think first about the main reasons why you want to spread the design of your system across several projects referencing each others. Do you have different contributors for each part? Do you have separate lifecycles for each part? One possible, typical answer to your question is to use the “system to subsystem transition” add-on: the idea is to have a certain level of detail in a global system model, and then having an automated transition towards subsystem models that contain further details about the subsystem. The rationale is that at each engineering level, we find only what is necessary and not more.
There are other possible answers to your question, but not necessarily easy to explain in a forum post. For example, we see more and more the need to synchronize system and subsystem models when the latter ones already exist and cannot be just the result of a system to subsystem transition. There are ways to implement that in Capella to some extent, with schemes involving REC/RPL and library mechanisms.
Thanks for the response Stephane
I will have a look at the use of libraries and rec-rpl components, I can see there is a prototype to allow model elements to be moved from a project to a library which would be useful. And I assume the functionality of Blackbox, Constrained Reuse and Inheritence will work when using replicable elements from a library? Will the link through the library mean changes to a replicable component are automatically made in another project i.e in the case that I
I create a project (project 1) and model a system
I add an element of that system to a library as a replicable element
I use that library in a different project (project 2) and use the replicable component (and I’ve allowed inheritance)
Add a new interface to the replicable component in project 2
Is the new interface now visible in project 1?
In answer to your methodology questions
Yes multiple contributors and multiple lifecycles but also scale
For example taking the Seven Samurai of Systems Engineering model -
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.457 .3253&rep=rep1&type=pdf
If I want to model the various systems (S1 to S7) doing this in one model will quickly end up with 100’s of model components and functions, and 100’s of viewpoints. The Intervention System will likely include over a 100 subsystems
A single project which contains all of this will quickly become difficult to navigate and understand
Thanks for the help
Stu
About REC-RPL
I believe we need to make a video/tutorial for people to have a better idea of the features
The solution allows for synchronization between RECs-RPLs
The solution has anticipated different kinds of compliance relationships, but actually only the validation rules are only implemented for Blackbox is implemented.
The promotion of a set of elements in a project as REC in a library (making the original set of element become an instance of the REC) is a feature we would like to develop but does not exist yet.
Point 3 means you need to create things in library first (unless you start playing with diff/merge, etc.)
About the 7 Samourai
The content of each model might be pretty different, the relationships between the models would typically be close to generic traceability. Capella does not allow this kind of links between models. We have seen a project that has developed an add-on to do that, but to my knowledge, this is not a public development. Typically, this is the kind of link that could be managed by of more global workbench to which Capella would be plugged.
Hi Stephane,
Has a video/tutorial been made to demonstrate these features?
Rav
Hello,
No, there is no new video or tutorial. You might find a sneak peek in this webinar:
https://youtu.be/I28EhAXe-i8
Stephane Bonnet wrote on Wed, 24 October 2018 10:44
About REC-RPL
The content of each model might be pretty different, the relationships between the models would typically be close to generic traceability. Capella does not allow this kind of links between models. We have seen a project that has developed an add-on to do that, but to my knowledge, this is not a public development. Typically, this is the kind of link that could be managed by of more global workbench to which Capella would be plugged.
As i know there is Eclipse Carpa project with the purpose of general tracebility between EMF models
Project
https://projects.eclipse.org/projects/modeling.capra
Demonstration
https://www.youtube.com/watch?v=61sUdObyKUk&feature=yout u.be&t=703