To whom it may concern:
I am working on a SimTraffic Simulation for a project.

We detected numerous intersections with the 95th %ile Queue Lengths greater than the Maximum Queue Lengths for couple of movements. I have read your feeback on this topic, and understand why it occurs.

I am interested in how to correctly report the 95th% queue length data. Should we:
A. Remove the curvature in the roadways between nodes
B. Remove the amount of nodes between intersections
C. Use directed calculations with the B## values

Also why are the synchro to simtraffic report link lengths differing?

Thank you for your time.
JB

Hi,

You'll want to use caution when using the 95th Percentile queue. It is not a true observation, but a statistical calculation on the average queue. I think you noted you have the definition of average and maximum queue which are based on observed values.

For instance, the 95th %ile queue can be longer then the storage, but the max and average cannot. This is one reason why the 95th can be longer. Also, the maximum is limited to the length of the link (the 95th is not). I believe this is what you are seeing. In cases like this, the 95th is meaninless.

Look at the upstream and storage block times. If these values are high, then the maximum queue calculation is getting limited by storage or link length.

A. You should not need to remove the curvature
B. Remove any unneeded nodes is always a good idea
C. If you have bends, you will likely want to add the B## values. However, this is case dependent on whether you feel the queues on the bend link are created by the downstream intersection.

Thanks,

John.