Signalized Intersection Level of Service

[longisland]

Hi,

I'm part of a road maintenance crew assigned to carry out a field traffic count for an existing signalized intersection. The last time I was involved in Traffic Engineering was in College. Anyway, the signal phase for each leg is about 20 to 35 seconds. The whole cycle is about 140 seconds. The longest queue was about 180 yards; the car at the tail end of the queue will normally end up 30 yards before the intersection & will only be able to cross the intersection in the next cycle. Having referred to HCM 2000, the intersection would have a LOS of F (>80Secs). The minor access has no problem clearing the platoon in the 20-second cycle & the LOS is B. How would one rate the LOS of this intersection? Is it based on the longest queue of all turning movements or the average of LOS of all turning movement?

Thanks in advance

 

[ACtrafficengr]

The most common failure mode is excessive delay, and it sounds like you certainly have that.

Intersections can also fail by queuing back through the previous intersection, and causing that one to fail. A former colleague of mine used to tell about experiencing an extreme version of this on Manhattan Island, and having to wait for several hours of it to get all cleared up.

Given that the side road has LOS B, and the intersection is failing, it sounds like some retiming is in order.

"...students of traffic are beginning to realize the false economy of mechanically controlled traffic, and hand work by trained officers will again prevail." - Wm. Phelps Eno, ca. 1928

"I'm searching for the questions, so my answers will make sense." - Stephen Brust

 

[teeman]

As currently defined, signalized intersection level of service represents the total delay experienced on the intersection approaches averaged over the total number of vehicles using the intersection. The weighted calculation helps to keep the focus on the lane groups with major flows.

You don't provide much information regarding signal phasing, but the 140s cycle itself will tend to create long delays and possible cycle failures if the intersection geometry isn't adequate for the demand.

 

[longisland]

Hi,
Thanks for the reply. I think I may have messed up the total cycle time. The green time of the through traffic lane at the major approaches is 35s. The left turn is 25s. The left turn & forward green time is 20s each. Therefore, the full cycle time is the sum of 35s, 25s, 20s & 20s for the turning movements of all approaches amounting to 100s.
I'd appreciate if you can clarify on the definition of delay. Is the delay time defined as the time in queue beginning from the green phase? For example, if a car is in a left turn lane, with a queue of 10 vehicles, is able to cross the intersection in 20s within a green time of 25s, the lane would have a LOS of B. Therefore, the time spent waiting for the green light, may be as long as 75s, is not considered as a delay. As I,ve mentioned, the last car of the main approach was unable to cross the intersection within the green time & was forced to wait until the next cycle of green, therefore, that approach is considered having a LOS of F.

 

[teeman]

You can calculate intersection delay and LOS from an actual delay study at the intersection. However, at signalized intersections the volume of traffic, the difficulty of counting cars in multiple lanes and need to see all the cars in the waiting queue make delay studies difficult if not impossible. The HCM methodology was developed to provide a reasonable approximation of intersection operations with data that is relatively easy to obtain.

In the HCM signalized intersection analysis, approaches are analyzed as "lane groups" which organize the approach traffic based on the type of flow in each lane and the signal phasing. Exclusive lanes have different flow characteristics than lanes that include thru & turning vehicles. Delay is calculated for each lane group and total intersection delay is the sum of the individual lane group delays.

The LOS thresholds are based on what they call average "control delay" per vehicle. It has three components: referred to as d1, d2, and d3.

d1 is essentially the delay associated with uniform (average) arrival rates during the analysis period. At locations where the queues clear every cycle, this may be the only delay present.

d2 is is called "incremental" delay and is the delay experienced by arriving traffic that doesn't clear during the phase green.

d3 is delay related to vehicles arriving during the green encountering a standing queue at the stopline.

For stop controlled intersections I generally prefer to do an actual delay study rather than use the HCM stop control methodology. For signalized intersection the HCM methodology can generally provide an acceptable result.

As with any procedure, when reviewing the results the analyst must ask, "Does the result accurately represent conditions at the intersection?" If the answer is no, then the input data needs to be reviewed and revised as necessary.

 

[longisland]

Hi,
Thanks for the feedback. I'm using the HCM delay study method as well to calculate the average delay on each approach. However, for preliminary assessment, would it be okay to assume an approach is experiencing a LOS of F if the platoon is unable to be cleared within the green phase?
I'll show the level of service of all different approach rather than taking an average because the average LOS does not reflect the worst case. Is taking an average LOS of all approaches a standard practice?
BTW, what's the better approach in accessing a round-about? As you have pointed out, the HCM method may not reflect the actual condition. How would one carry out actual delay study? My guess is to determine an average speed by dividing the distance the last car of the queue traveled to the approach by the time spent. The number of cars in the queue is also recorded. The average delay would be the average time of the cars in the queue to cross the intersection with the average speed computed previously. I hope I'm on the right track.

Thanks

 

[teeman]

The intersection LOS represents the average intersection delay, total intersection delay divided by the total number of vehicles using the intersection. As an MOE it has its uses, but it is not as important as the approach analyses which can be used to refine signal timing. Most software packages automatically calculate intersection LOS. If I were doing an analysis by hand and didn't need it, I might not bother with that calculation.

As I understand it, people familiar with roundabout design do not recommend using HCM for roundabout analyses. RODEL software is recommended for single lane roundabouts. Multilane roundabouts require more sophisticated (and expensive) simulation modelling to account for movement within the circle. New York DOT has a very knowlegable and very helpful roundabout design unit.

In a delay study you essentially record the number of stopped vehicles on an approach or in a lane at set intervals (15s or so). You also have to keep track of the number of vehicles that did not stop. At the end the total of stopped vehicles are multiplied by the sampling interval and then divided by the sum of the stopped and not stopping vehicles to get an average delay for the approach or lane being monitored. At a busy signalized intersection with multiple lane approaches this can be a daunting task, hence the need for a reasonable accurate signalized untersection model such as the HCM.

 

[longisland]

Hi,

Thanks for the advice. It's a very interesting assignment for me. My job on the field is done & the traffic engineer in the design office will take over.
Just out of interest, is there any simple approach to assess an intersection before conducting a full scale count? In my case, the major approaches are having a queue of more than 15 vehicles & normally the last few cars aren't able to cross the intersection in the 35 seconds green phase. One way to overcome is by increasing the green time, what's the threshold value without causing excessive delay to other approaches? The
other way I can think of is to build additional lane or even build a grade separation intersection.

 

[teeman]

Many years ago a gentleman by the name of Greenshield developed a basic algebraic equation to model the discharge of a traffic signal queue. I use a simplified version of this when making signal timing adjustments in the field. The gist of the model is it takes 15 seconds for the first 5 cars in the queue to pass the stopline and 2 seconds for every subsequent vehicle. If you have 15 cars in the queue your "estimated" greentime to service the demand is approximately 15s + (15-5)*2s or 35s. The basic timing for that phase is not far off, and a minor adjustment may clear the approach.

You don't indicate if the intersection is actuated or fixed time. At fixed time signals I want to minimize wasted greentime and will accept failure to clear a queue if it always clears on the next cycle. At actuated signals I try to clear the queue every cycle (if lane group capacity permits).

If I recall, the general guideline for maximum recommended cycle is 120s because delays increase exponentially above that value. My own guideline for unsaturated conditions is the desirable cycle length is the number of phases times 25s. So if I see a plan calling for a 110s cycle at a three phase light, I would expect to find "issues" within the analysis.