05-10-2016, 10:37 AM
(05-10-2016, 09:48 AM)SammyOES Wrote: "Point is, once a bad behaviour starts (e.g. early merge), it's better to try going with it than correcting it. "
I disagree with this point. Particularly for long-lived bottlenecks like we're talking about here.
You'd pay a 'one-time' price near the start where a relatively small number of cars are inconvenienced (the people that got in line rather than starting a true zipper merge) and after that you get all of the benefits of the zipper merge.
Any time someone jumps to the head of the line, when the early-merged line is moving at the same speed as the traffic through the bottleneck, everyone already in the merge winds up going slower in order to let you in, and then it will pick up until it gets back to the steady-state speed through the bottleneck. None of this moves the useful merge point (where you go from highway speed down to bottleneck speed) any farther forward, even if it creates a secondary, non-useful merge point (where two bottleneck speed lanes merge into one bottleneck speed lane.