Closed gioagu closed 8 years ago
Im agree. The relation composedOf (ThermalBoundary -> ThermalComponent) should be optional, and a new (optional) attribute area (Type Area) should be introduced in ThermalBoundary. Additionally, we can formulate a conformance requirement that this new attribute shall not be used in case a ThermalBoundary is related to one or more ThermalComponents.
Agree also.
Delete Boolean attributes in ThermalComponent, change cardinality to 0..* Add optional attribute "area" in ThermalBoundary. Mention base attribute "relativeToTerrain" in guidelines.
Implemented in latest version of the UML model
So far, each ThermalBounday MUST have at least one ThermalComponents. If I use a single ThermalComponent, I can define the area and - e.g. by Construction - how it is made. If I use two ThermalComponents, I may define how many m^2 are wall and how many are glazing.
However, if I do not have this information, but just the ThermalBoundary, I still need to generate at least one ThermalComponent for each ThermalBoudary. This makes sense if I am using the data as input for simulation software, but (sometimes) not if I simply want to transfer data.
Imagine I have a city with 1000000 ThermalBoundaries and no other information. My XML file needs 1000000 ThermalComponents to be valid. But do I really need them? I will generate the ThermalComponents depending on the type of library I am using (e.g. from where I get the window_to_wall ratio and the construction) in order to feed the simulation tool, but do I really need to store this information in advance?
Maybe the cardinality from ThermalBoundary to ThermalComponent could be relaxed to 0..n. If so, I would also add an area attribute also to the ThermalBoudary object.