Below is a question I just emailed John Albeck. I know he's a busy man, so I figured I would try it out on the forum.
When I get a reply from John, I'll post it (assuming John hasn't resolved it here first).
Thanks in advance.
~ellis



John,
I’m slightly confused by information contained in the Queue info-sheets you had forwarded me.
In “Queues in SimTraffic from Training Manual”, the yellow sidebar says, “In SimTraffic, the queue length reported for a movement IS limited by the turn pocket length, just like in the real world. SimTraffic simply observes the maximum queue. So, if you have a left turn pocket of 250', then you could not achieve a SimTraffic left turn queue of greater than 250'. Any spillover from the left turn pocket would be reported in the adjacent lane queue length.”
However, in the “Queue Block Time SimTraffic” info-sheet, the snap shot of the SimTraffic MOE shows a (observed) Maximum Queue for EB LT as 351ft, even though the Storage Bay Distance is clearly indicated as 325ft; thus a situation in which the observed queue is in fact longer than the “left turn pocket”. Is this because SimTraffic also considers a vehicle in the taper? If so, this would indicate why the Maximim Queue is only 26 ft (roughly 1 veh.) longer than the Storage Bay Distance.

The reason I am trying to figure out such a little detail is because I am responsible for determining left-turn queues along a corridor, and whether the existing storage bays are of sufficient length. My initial methodology was to build the synchro network, simulate, average the multiple runs, and compare the 95th%-tile queues with the existing storage bay lengths. However, it seems that such an approach will not work because an overflowing, left-turn queue will just become part of the adjacent through-lane queue and not be properly represented via Maximum Queue. And even though the 95th %-tile queue is not necessarily ever observed, it is calculated upon inaccurate (or at least incomplete) data.

Is there an alternative method that you recommend in ascertaining the simulated 95th %-tile queue for a left-turn movement, especially for situations when such a queue spills over beyond its storage bay length?
I hope that this makes sense, and that I have framed my question upon a logical AND accurate assessment of the SimTraffic methodologies.
Thanks again,
~ellis

Hi Ellis,

Thanks for the post and for allowing me to clarify. On the first point, SimTraffic is considering the taper length. SO, with a 250' pocket, you could get a max queue of 275' +/-. But, it will not be any longer than that.

What you note about the 95th Queue in SimTraffic is also true. The 95th %ile queue may never really be observed. It is based on the average queue (average of 2-minute maximums). Since the maximums are limited by the storage, the 95th may also be limited.

I would suggest starting with the Synchro queue as an initial test. However, if you see a ~ symbol, then the model is over capacity and this value is an underestimate. But, with no ~, you could start with the 95th in Synchro. Supplement this with your knowledge of the area (ie, ROW needs, adjacent storage, driveways, etc).

Next, run some simulations with SimTraffic to test your design. I would look at the storage block time rows in the report. If you have an EBL with a 20% block time, this indicates that the EBL is backing up to the end of the storage 20% of the time. This might indicate to you that you should (if feasible) make the storage length longer. Note: if the EBT storage block time is 10%, this indicates that the EBT is blocking the EBL storage lane 10% of the time.

Another approach that some users have taken is to enter a zero storage for the turn lane. Then, the lane would be the full length of the link. Doing this will allow vehicles to enter the turn lane unimpeded by the storage. Then, you can get a good idea of how long the queues will be and design based on this.

My note of caution is that by using a zero storage, you could get unwanted lane positioning at the upstream end of the link. Be sure to check the vehicle paths (Options>Map-Settings) to ensure there are no artifical problems caused by using this approach.

My final note is that queue lengths in both models are very sensitive to timing settings. So, you'll want to make sure to use reasonable timing values for your area. For instance, you would not want to use a cycle of 40 seconds if this would never really be used.

Thanks,

John.