I'm modeling an urban network and one of the options in the future scenario is to remove every other signal along a heavily congested 8-block corridor (removing 3 signals) and replace them with R-in R-out intersections. The corridor will be 3 lanes NB and 2 lanes SB, intersections about every 440 ft. I've modeled the right turns as channelized rights, with no add lane and a radius of 14 ft. (thr minimum) My problem is that about 10% of my simulation runs are failing because occasionally the car at the head of the queue freezes and will not enter the mainline, even if the block is empty. Has anyone else had problems with channelized rights in this manner (no add lane?)

Thanks in advance.

This message has been edited. Last edited by: Eric Royer,

Eric,

It may be a vehicle in the right that is trying to make left at the downstream intersection. If you get an error message about the link being too short for channelized rights, this could be the issue.

In some cases, you may not be able to use the channelized right (if the downstream links are relatively short with multiple lanes). Or, you might try excluding right turns from turing left at the downstream intersection. You'd do this with the Link OD editor.

Thanks,

John.

quote:

It may be a vehicle in the right that is trying to make left at the downstream intersection. If you get an error message about the link being too short for channelized rights, this could be the issue.

I don't specifically recall getting this warning message recently, but I found the behavior of waiting to cross 2 or 3 lanes causing excessive delay, and had previously set the OD-weights to zero such that the right turns entering the main street make a through movement at the downstream intersection. So this shouldn't be an issue.

currently I've altered the model to remove the channelized rights and modeled them as through-rights and simply not coded any through traffic.

this is less than perfect though because the mainline now leaves the intersection clear and creates longer queues. given the alternative of trying to get more completed runs with what I had (78 runs with some sort of lock up in every one, took me 4 days...) this seems to be a little better.

Eric