We have found that the u-shape semantic definition is inadequate.
You can only specify an inner edge radius for but in real life profiles the is also an outer edge radius or in practice that edge radius is bigger than flange thickness.
That causes broken solids.
How would it be possible to go forward fixing this?
You can only specify an inner edge radius for but in real life profiles the is also an outer edge radius
Ping @Moult who has recently spent a lot of effort in mapping profile libraries to IFC. Did you also encounter this?
outer edge radius I don’t see an objection on adding a new optional attribute. It’s not an ideal approach, because you don’t really have a way to assert that vendors are not overlooking the attribute (other than detailed unit testing). In that sense it’s better to introduce a new subtype, but given the semantic overlap, that also seems undesirable.
in practice that edge radius is bigger than flange thickness. That causes broken solids.
The constraint is not only related to the flange thickness, the inner radius as well as the FlangeSlope also influence the max outer edge fillet. Not sure if a formula is necessary, but if needed I can derive the equation using a symbolic math toolkit.
In Australia, I don’t think this is an issue, however @jonm is the man for the job being a trained structural engineer in Australia.
Here’s the parallel flange channels used in Australia:
There are also tapered flange beams: