System needs definition

Hello everyone.
I cope with a difficulty while working on a project with the Arcadia method.
During the SA, I have the feeling that all the parameters that allows us to choose between two alternatives are linked with the preconceptions on the two architecture that we have.
The capability trade off that one has to make is performed considering the effect that the choice will have on the system.
However, how to judge those effets if the LA have not be made ?
My first thought will be to make two ‘rough’ LA, in order to see the structural differences between the two alternatives, and then make the choice. But it can be very time consumming.

I suppose that the two alternatives that you mention are design choices, right ?
If so, maybe your SA has not the good contents : SA should only describe customer expectations on the system, such as expected services to be delivered to the users in different use cases (missions and required capabilities). Therefore, it should in no means be subject to any influence of design choices.

Hello Jean-Luc.
In your book ‘Model Based System engineering with the Arcadia method’, part 6.2, you gave the example of the choice between an automatic and a manual level crossing. If the operator of the system is considered as integrated in the system, isn’t that a design choice ?
For exemple in this case, if the customer’s expectation is both related to the cost and the safety of the system, how can one assess that an automatic will be safer than a manual crossing level, considering that those elements are actually extracted from the architecture of the system ?
Thank you for your quick reply !

Hello J,
Thanks for reading !
This is a good question, emphasizing differences between system need and solution, as supported by System Need Analysis (SA) and Logical/physical Architecture (LA/PA) respectively.
It all depends on the customer need contents: if he considers that a human supervision is requested (maybe due to regulation constraints), then this is part of his need, so you will have to express this in SA, notably with an external actor. But in this case, you still have to discuss with the customer the distribution of tasks between the system and the operator, including automatic Vs manual behavior; this will drive functional allocation to the system and the operator in SA. This is what is illustrated in the book.
Another example is that today, regulations require two pilots in the cockpit for commercial Aircraft, no room for optimization or automation (yet).
In some cases, the customer may leave you free to define not only the system, but also the role of possible operators, if any (this is not considered in the book example). In this case, you should probably not define any external operator in SA, and only introduce an operator in LA or PA. And here, you are right, this would be a pure design decision, thus not appearing before (in SA).
For the aircraft example, if you have to design a surveillance aircraft, then the crew concept and the number/roles of operators might be a pure design decision, depending on the kind and number of on board sensors and tasks required by the mission, among others.

I suppose that what I have to do nom is ask my customer whether he wants the best system (operators in LA) or the best system with its constraints (operators in SA).
Thank you for your answer !

Copyright © Eclipse Capella, the Eclipse Capella logo, Eclipse and the Eclipse logo are Trademarks of The Eclipse Foundation.