Hello,
I am trying to do a study on time headway on an urban road that is preceded with a signalized intersection. After gathering the necessary data, I had to do a sampling process to eliminate the trending and periodicity on the data.
I had to eventually divide the data on 2 distinct categories, which are the phases of the preceeding signalized intersection (the intersection has 2 phases with different durations).
Now I am finding myself with two different patterns. in order to study the Time Headway on different trafic levels. I need to define different increments (number of vehicle arriving in a period). Since I have two categories I would have different resulting hour volumes.
Let me give you an example, Phase 1 has a 36s duration and phase 2 has a 29s duration. So for example If I define an increment of 5-9 vehicles per period for the two categories, I would have a 500-900 veh\h for the samples of phase 1 and 620-1117veh\h for the samples of phase 2.
So I am not sure about this. Do I need to define different intervals for each category ? Did I go to far on dividing the data on two categories ?
Any suggestions are welcome.
P.S : I am not a traffic engineer so I am not familiar with the procedures of time headway study. I am just trying to follow what my fellow researchers did in their papers
Thank you !
Hi,
from what you write, you have already found a first solution to your problem, in terms of statistics; for the moment you do not need anything else, in my opinion.
Dear, Saad Touhbi,
I think you measure now time arrivals as a function of signal time. If these times are different also the number of vehicles in one hour is different. One thing is the number of arrivals (5 or 9 according to your example at different times), another thing is the flow of traffic in one hour. The measurement of the intervals ("it is most commonly measured as the distance from the tip of one vehicle to the tip of the next one behind it, expressed as the time it will take for the trailing vehicle to cover that distance") must be made in terms of time of subsequent vehicle transit. And I think that this measure could be independent of the signal times.
Basic references:
https://en.wikibooks.org/wiki/Fundamentals_of_Transportation/Traffic_Flow
Hi,
in terms of traffic engineering you also have to consider that, against traffic on highways, the vehicles in urban areas usually reach the intersection as a platoon, because of the signalization of the upstream intersection and not randomized. And signal programs do change over the day. So the length of your stages is adapted o the traffic volumes.
Instead of the fundamental diagram the queue theory from Webster and Kimber/Hollis is quite often used for urban traffic.
Traffic engineers use traffic volumes of 1800 to 2000 vehicles per hour for undisturbed streams to calculate the split of the green times. This is a time gap of 2.0 to 1.8 seconds per vehicle for a saturated traffic flow.
Hi,
in order to find stationary time intervals I would suggest you the trend analysius of Luttinen:
http://lib.tkk.fi/Diss/199X/isbn951228474X/isbn951228474X.pdf
In your case, I would try to adapt the trend test to small intervals by modifying the test parameters. If the intervals are small, you can aggregate them (umong different cycles) by respecting some criteria (same flow rate for example)
i agree with Francesco further you can use negative exponential distribution for this purpose
- Badges
- Science topic
- Similar topics
- Engineering
- Civil Engineering
- Transportation Engineering