Travel Time and Delay Report

Questions and discussions about operating Tru-Traffic

Moderator: bullock

Travel Time and Delay Report

Postby ela » Mon Nov 15, 2010 8:25 pm

Region of Durham is planning to work on Travel Time Survey project. The arterial road length to be surveyed in both directions of travel is approximately 140 km and we're planning to divide them in 2-2.5 km segments. With 9 runs per each direction both am and pm peak and 3 runs each direction during noon peak we'll be driving in total 3000 km.
> We're planning to collect all data and create some reports using the Tru-traffic software.
> Our arterials will be assessed on the basis of mean speed, Travel Time Index, Buffer Time Index and Delay.
> To calculate the above mentioned parameters we would need the following:
>
> 1. Free Flow Condition Travel Time
> 2.. 95 th Percentile Travel Time
> 3. Average Travel Time
>
> Please advise which parameters (Travel Time and Delay Report) could I use for the above mentioned
> Thanks in advance
> Elizabeth
ela
 
Posts: 1
Joined: Mon Nov 15, 2010 8:20 pm

Re: Travel Time and Delay Report

Postby bullock » Mon Nov 15, 2010 9:20 pm

Great questions, Elizabeth.

1. For Free Flow Condition Travel Time, I would consider either
  • Using TT: Travel Time from previous Node (seconds),
  • Adding a new, user-defined column "MinRT: Minimum Running Time, or Travel Time from previous Node if maintaining Maximum Speed (seconds)" with the formula
    Code: Select all
    TT - MxSD
    , where
    • MxSD: Delay in Maximum-Speed Travel Time from previous Node (seconds) = TT - DL/MxS,
    • MxS is the Maximum Speed reached in Travel from previous Node,
    • DL is the User-specified Design Distance from previous Node, and
    • TT: Travel Time from previous Node (seconds), or
  • Viewing the Speed vs. Distance plots for the set of trip logs, including the instantaneous average speed and the average link speed, and looking at each segment, reading straight off the plot a value that seems reasonable to me for the free flow speed -- a speed that seems to fairly represent the runs when they're not incurring delays due to congestion or incidental traffic control. Divide that speed into the design distance to get a Free Flow Condition Travel Time.
(By the way, I'll probably add MinRT to the report as an intrinsic column so you don't have to create a user-defined column. Off the top of my head, I'm not sure why I haven't done that already. Either it never occurred to me, or it seemed to me there were already too many columns and this one wouldn't be that useful, or it seemed to me that it was easy enough for the user to enter it manually. Anyway, I'll reconsider all of that.)

Furthermore, I'd consider taking the value from my best run ever in the directional Details pages of the TT&D report (e.g., the Northbound Details and the Southbound Details tabs), rather than the average of all the runs in the directional Summary pages of the TT&D report (e.g., the Northbound Summary and the Southbound Summary tabs). Before committing to a particular combination (TT vs. MinRT vs. read-from-speed-plot, and best run vs. average of all runs), I would
  • Consider each segment to decide whether there's anything along that segment to cause the instantaneous Free Flow speed to vary along the segment (for example, a sharp turn, a stop sign, a railroad crossing, a speed bump, rough or slick surface in parts, a change in posted speed limit, etc.). If the Free Flow speed seems like it should ideally be uniform along the entire segment, I'd probably prefer MinRT or read-from-speed-plot over TT, otherwise I'd lean toward TT.
  • View the Speed vs. Distance plots for the set of trip logs, including the instantaneous average speed and the average link speed, and look at each segment. If any run seems to be an outlier along the segment -- especially if its maximum speed seems to spike unusually compared to the other runs -- then I'd consider discarding that run's maximum speed (its TT should still be reliable, but its MxS may be an anomaly that you want to discard). A run can have an outlying speed due to either a driving anomaly during data collection (the driver sped up briefly for some reason) or to a faulty GPS report perhaps due to poor satellite reception (I've seen momentary speed spikes of around 200 mph!). If there are outliers affecting the maximum speed along the link, you can either disable the corresponding point(s) in the trip log, or you can consider the next option below...
  • Lean toward the value calculated from reading a typical speed directly from the Speed vs. Distance plot if the best run ever seems to have an outlying maximum speed and the collection of runs has an appreciable amount of constricted flow (non Free Flow) and disabling the outlying points seems troublesome or undesirable for whatever reason.
  • Be even more likely to prefer the value from my best run, rather than the average of all runs, if I end up choosing TT (over MinRT) and some of the runs show delays that I don't want to design for.

2. For 95th Percentile Travel Time, I'd refer to the directional, link-by-link Summary pages of the TT&D report (e.g., the Northbound Summary and the Southbound Summary tabs), and for each segment, I'd use
Code: Select all
(Average TT) + 1.645*(Std Dev TT),

To get (Std Dev TT), make sure you've got "Standard Deviations from the Averages" included in your report summary (View: Preferences: Layout: TT&D Report: Summary).

I'd use this formula provided that the travel times have a normal (bell curve) distribution. For a normal distribution, 90% of the travel times fall within +/-1.645*(Std Dev TT) of (Average TT) and 95% fall within +/-1.96*(Std Dev TT). For me, this assumption mainly means that there are no traffic signals along the link. If the link does have traffic signals, then -- I'll expose just how clueless I am -- I really don't know what we mean by "95th percentile travel time." Are we referring to the formula above, or are we referring to the time that's greater than or equal to about 95% of all runs? Such phrases are used routinely for signalized arteries, and it's likely that everyone already agrees on their meaning but I'm the one who just hasn't clued in yet. The formula above is accurate only if the travel times fall roughly on a bell curve, but if the link has a traffic signal with a stopped delay of, say, 30 seconds, then this introduces a ~30s gap in the distribution of travel times, which puts a deep dent in that bell curve, making a bimodal distribution. Vehicles either get stopped or they don't, and if they get stopped, their travel times are automatically increased by roughly 30 seconds, with the result that there's no longer a continuum of possible travel times accessible to the vehicles, as the bell curve would suggest. So in general for signalized arteries, I don't know what is meant by a phrase such as "95th percentile of the cumulative (end-to-end) arterial travel time". It seems clear enough, and I'm sorry for my ignorance, but I don't know if that means the above formula (which is readily calculated) or the lowest value greater than or equal to 95% of all travel times (which is more difficult to calculate)?

If you or anyone else would like to clue me in, clarifying the meaning of this for me, then I'm all gratitude.

3. For Average Travel Time, I'd again refer to the directional, link-by-link Summary pages of the TT&D report (e.g., the Northbound Summary and the Southbound Summary tabs), and for each segment, I'd use (Average TT).

Regards.
Greg
bullock
Site Admin
 
Posts: 218
Joined: Thu May 06, 2004 6:51 pm
Location: Pacific Grove, CA


Return to Program Operation

Who is online

Users browsing this forum: No registered users and 7 guests

cron