This study from MIT used geo data collected from cars in Milan, Italy, to check the effectiveness of 30 km/h zones in reducing speed.
The first conclusion is that the signs don't work: 85 percentile speeds are all over the place in 30 km/h zones in Milan, as shown in the figure below:
The second step was finding correlations between speeds and street features extracted from openstreetmap. Results are as expected: narrow, short, curvy sections correlate with lower speeds, as do 1 lane vs more, one way vs 2 ways:
The final step is also interesting: the authors made a model to predict the compliance of 30 km/h speed limit on streets that are 50 km/h at the moment. Useful for urban planning to understand if charging an area to 30 km/h would need structural interventions (like bumps, narrowing of the street...) or not:
There is so much more in the article, I suggest to read it fully.
crossposted from: https://mastodon.uno/users/rivoluzioneurbanamobilita/statuses/114827312307353297
Speed limits (in North America anyway) are determined by what speed the 85th percentile drives at. Of course lowering the speed limit doesn't work, that's literally the reverse of how speed limits are determined.
Drivers determine what speed they feel safe driving at. They take cues from the design of the road. If you want to slow down drivers, you have to redesign the road.