Models integration into a large model

Capella
1

Hello,

I would like to use capella by describing a complex system with several models by this way described bellow.

At first, according ISO 15288, a system can be decomposed with sub-systems and system element, each sub-system can be also decomposed into sub-systems and system element.
At the end of all decomposition we’ve only system element ; system element means that it won’t be decomposed after and represent what is expected

image
image444×175 35.6 KB

On my complex System Of Interest (SOS), I described my system with Operational Analysis, System Analysis and Logical Architecture layers.
At the SOS Logical Layer I can modelise different Logical Components (LC1, LC2, LC3, LC4….) each one represent my sub-systems and the function allocated on.

For Each sub-system, I done a model (by using System to Subsystem transition vertical to SA), so for exemple LC2 have is own model with :

  • At System Analysis, a black-box view is inherited from SOS model / Logical Layer by the use of System to subs-System transition : context diagram, functions…
  • At Logical Level, I modelised a white-box view with function and Logical Components : LC2 is decomposed into LC21, LC22, LC23
  • Also (in the future) to add variation point related to the internal design variant of LC2

Recursively, I can use the same approach to modelise LC21 with a new model

At the end I’ve different models, potentially with their own variation points.

I would like to know if their an automatic mechanism in Capella to rebuilt a new detailed model :

  • By starting from the SOS model
  • To replace the black-box LC component LC1, LC2, LC3… by theirs white-box models (or in addition to chose a white-box model in case of several alternaltive white-bow models)
  • In order to have a SOS detailed Logical Architecture Layer with LC11, LC12, LC211, LC212, LC213, LC221, LC222… and the functions allocated onto this detailed Logical Component which represent all my SOS System Elements

After that I want on thiw new / large model to work on Physical Layer, to define where to deployed all detailed Logical Element on Physical nodes

So I’m looking for tips/recommndations or alternative ways and links to documents and demo

Thanks

2

I do not believe that there is something ready to use to achieve what you are trying to achieve with the System to sub-system transition add-on. From there, I see the following options:

  • you can always contact us (Obeo) and fund such a feature.
  • you could do it yourself. Probably the quickest way would be to use the Python4Capella add-on
  • it is possible in Capella to create a model by assembling information coming from other models. The way to do that is to create RECs from your model elements and put them in libraries. Then you can create RPLs in a unified model. I am not aware of users using this strategy combined with the S2SS add-on. But there may be cases out there

Generally speaking:

  • I am not sure it is a good practice to consider that LC1, LC2, LC3… are “sub-systems”. What I mean is that (1) a sub-system may be actually composed of multiple LCs (like “in my project, subsystem A is made of LC1 and LC2”) and (2) more importantly, I would not recommend restricting architects to use only one level of component decomposition (ie “at the SOS level, you’re only allowed to create LC1, LC2… and no further component decomposition is allowed”). I believe this could limit quite significantly the ability to perform good architecting work
  • There are good reasons why you/users decides to use the S2SS add-on and to manage different level of system models in separate Capella projects. But keep in mind that this generates an additional level of complexity in terms of managing model lifecycles/changes and versions. So I would make sure that this is actually needed before going down that path, rather than creating one single model where multiple users can collaborate in real time (with Team for Capella for instance).

I hope this helps.

Stephane

3

Thank you Stéphane for your quick answer. Yes it’s was useful

I planned to use librairies but before I wanted to have advice and if there something already on place