So I thought about adding a filter for dual_carriageway, but it doesn’t seem to be used (almost) at all in Belgium (overpass turbo). Is there a reason for that? Any alternative suggestions for filtering out such cases?
I can highly recommend using this tagging, both as a mapper and especially as a data user. It offers many advantages in data analysis that mappers don’t always consider. In Berlin, we are increasingly using it to obtain more precise evaluations, e.g. for the cycling network (see above), for more precise street length calculations and also for precise renderings.
Hasn’t been discussed before AFAIK. I guess most people who cared thought it should be feasible to do this on the data user side. It’s not trivial but not impossible to identify most dual carriageways automatically, I’d think.
It’s certainly useful tagging for data consumers, because there’s no easy way to tell if a highway way is part of a dual carriageway without doing spatial queries. So I’d say by all means add it. Also consider using the expressway=yes tag in places where that applies.
Edge case time! Should these get dual_carriageway=yes or not?
Unholy N34 madness (I hold that this road was designed specifically to wreak havoc on the dual_carriageway tag)
N34/N34z in Nieuwpoort-Bad: topologically they definitely form a dual carriageway, but there’s another street in between. When you made up your mind, continue to the next level:
N34/N34y in De Haan: same story, but this time there’s an entire settlement in between.
For me, a good indicator for dual_carriageway=yes is, that you don’t need a oneway sign to know, that you are only allowed to go in one direction on each side.
So Case 1 is absolutely no dual carriageway, since there aren’t two adjacent, structurally separated carriageway. Only west of Vosseslag it’s a dual carriageway, because the lanes are located next to each other. Case 2 and 3, on the other hand, IMHO are dual carriageway segments in the OSM sense, because they represent one side of a road mapped in several structurally separated segments in OSM.
One criterium could be distance (maybe varying depending on the category of the road?), another one - is there anything not related to transport in-between the carriageways, that could be a trip source/destination on it’s own? So if there’s only a tram line, a parking, a cycle path or a De Lijn kiosk, then it’s a dual carriageway. If there are residential buildings, a basketball field or a church then it’s not?
Note that the case in the original post (contraflow biking) applies more to tertiary and residential roads than roads of higher classes that typically have bike paths or don’t allow bikes. That is also the case I’m interested in.