Good to hear.

So basically, everything we've been doing to meet the last information (make the roads as parallel as possible through the junctions) has resulted in us creating proper geometry for this to work.

Nice that the angles of the cross street do not matter, so two 90s or a 45 and a 135 will still function.

So both of these intersections will properly restrict the U-turn for the NB approach....

KuniaKid wrote:This segment is in question as allowing an illegal U-turn. I can't find anything with the segment that would allow the U-turn. It seems to conform to the criteria set forth in the Junction Style Wiki. Can somebody put a fresh set of eyes on it and see what I'm missing?

It looks like it should be blocking the u-turn, however, it is a u-turn PENALTY.... not restriction <groan>

If traffic is heavy, or the routing server does not see another route that takes less time than the penalty, it will route the illegal turn.

We really need to know what this penalty is, or the penalty needs to be changed to a restriction.

PesachZ wrote:Without the advantage of scripts running now I can't assess the turn angles, but by eyeballing it it seems the angle may be out of range. I have a few questions here to troubleshoot

The curve in the road makes it look off, but on the north side I've got an even 179, and 179.7 on the south side.
Doesn't get much closer than that.

North side (southbound) 88.6 + 90.4
South side (northbound) 93.9 + 85.8

So we are within 1 degree of 180 on both and we're (barely) within 5% of 90 degrees on the northbound first turn, well within 5% on the rest.... but I've seen a number that were outside of 5% of 90 (by a lot) but still about 180 total and did not route a u-turn.

PesachZ wrote:

taco909 wrote:North side (southbound) 88.6 + 90.4
South side (northbound) 93.9 + 85.8

So we are within 1 degree of 180 on both and we're (barely) within 5% of 90 degrees on the northbound first turn, well within 5% on the rest.... but I've seen a number that were outside of 5% of 90 (by a lot) but still about 180 total and did not route a u-turn.

Does this type a U-turn consistently in live map and app? Or only during severe traffic?

That needs to go back to KuniaKid.
Everything looks good to me and I suspect it is an issue with traffic (or the legal route) overriding the penalty.

This "penalty" needs to be changed to "restriction"

PesachZ wrote:

taco909 wrote:

PesachZ wrote: Does this type a U-turn consistently in live map and app? Or only during severe traffic?

That needs to go back to KuniaKid.
Everything looks good to me and I suspect it is an issue with traffic (or the legal route) overriding the penalty.

This "penalty" needs to be changed to "restriction"

Try to test it in livemap, can you get livemap to route the U-turn? I can't check now, so I ask.

I can't get a pair of segments that will route a U-turn cleanly on the south side.

I do get one routed on the north side, but the trace from Livemap does not overlay the segment in WME... the angle of the approachy is significantly different.
It shows KuniaKid edited the segment yesterday, I suspect an adjustment was made either after posting (and before I checked the angles yesterday) or before posting and KK was looking for confirmation.

I'm thinking it WAS outside of the allowed angles to enable to penalty, but should be OK now.

Would be nice if KK would confirm so we aren't spinning our wheels for the next 2 pages.

PesachZ wrote:

spookyx wrote:From my tests and experience, if a segment is 15 meters it will not alway do a U turn.
If, however is 16, than you will get a U turn on 100% of cases.

You must take over account rounding in wme. Which is why I always recommend using <14 or >16. Do you have examples?

Exactly.
I'll use 14 or 16, just like I try to avoid turn angles between 41 and 49 degrees.

That fits the rules for U-Turns.
Aside from traffic, it could also be a penalty-based issue.
If there are no "around the block" options, it may still override the U-Turn penalty logic.
We don't know exactly what this penalty is.

If they used the same logic as segment name inheritance, then the limit would be 3 segments.
We had an offramp arrangement where the road name changed on the overpass, and when the overpass was comprised of 4 segments, the turns did not inherit the name of the road on the opposite side of the ramp.

Joining the junctions of the two loop ramps into the center of the overpass dropping it to 3 segments restored inheritance.

So < 4 segments and < 400m would be a logical point to begin testing.