If your country doesn't have a manual then Highway Capacity Manual is a good start. For signal phase, there're plenty commercial software can give you an answer, e.g. this one:

http://www.sidrasolutions.com/Resources/Signals

This is just a practical answer. If you are interested in developing a backbone algorithm then better look at journal papers. 

Dear Saad, you ask very important question and I understand your need. However I am a bit skeptical about some 'definite/universal' guide on the choices regarding signals, phases, etc. In the world of traffic engineering and road safety there are not any universal rules, because there is so many influences. One would need to take them all into account in order to be able to decide on when to install signals or not, what type, number of phases, etc. This is why each agency/state/country has its own guidelines, reflecting site-specific conditions. I am afraid I did not help you, and I apologize for that...

Thanks Le Minh Kieu for the help !

Dear Jiri, I have noticed that actually. I am not a major of traffic engineering, but isn't there some major factors that are always taking into account while defining signals, phases, etc. ?

Thanks for the help !

Hi Saad,

Have you considered the possibility that a roundabout might be better than a signalized intersection?

While signalized intersections are sometimes referred to as the "socialist" way, roundabouts are "market economy"-based. Well, this is a simplification of course, but it illustrates an important point, and in fact roundabouts are cheaper to build (no need to dig down a lot of electronics, for example), and it should be easier to maintain. 

Hi Michael,

I could agree with you on that. But still there is more to it than the economic factor. since roundabouts can be more troublesome if there is an axe that has continuous flow with another with low flow (and it could cause accidents, driver frustrations especially in a third world  country). However, this is still a speculation, and what I am looking for is the factors that would enable me to say that a roundabout is more suited or a signalized intersection.

Thanks for the help :)

Hi Saad,

I looked at some materials regarding traffic signals decisions.

MUTCD (2009) uses 9 warrants:

Warrant 1, Eight-Hour Vehicular Volume

Warrant 2, Four-Hour Vehicular Volume

Warrant 3, Peak Hour

Warrant 4, Pedestrian Volume

Warrant 5, School Crossing

Warrant 6, Coordinated Signal System

Warrant 7, Crash Experience

Warrant 8, Roadway Network

Warrant 9, Intersection Near a Grade Crossing

Czech standards decide based on 4 criteria:

- safety (crash frequency)

- car volume

- pedestrian volume

- fluency of public transport

So I think that in general traffic volume is the most important criterion, then maybe safety as well...

Hi Jiří Ambros,

That's very helpful ! thank you :)

Hi Saad,

I agree with Jiří, and recent HCM 2010 updated several important issues for signalized intersection designs, conflict ratio, vehicle and pedestrian delays etc. Along with above mentioned warrants you can look at delays also and safety issues in terms of severity of turning movement vehicle conflicts.

Hello, Saad.

You mentioned that roundabouts could cause problems in developing countries. Certain countries have introduced traffic lights on larger roundabouts. The lights have addressed certain problems but have also caused others, such as additional delay and incidents.

Hello Yorgos,

I agree with you on that. it's indeed a large scale decision, and should take the population driving habits as a factor. In morocco, one can notice the effect of roundabouts (with no traffic lights) on people, either drivers or pedestrian, especially in zones that have high volume of pedestrian movements. 

And that's why I am looking for the basic factors used to determine which to use in which intersection and see if they are adapted to our case.

Thanks for the information :)

You are right, Saad.

HCM and MUTCD will give the answer for your desired design. These are updated with recent results. Roundabouts have received more attention recently. Another topic that has received attention is the potential to improve safety by reducing excessive signalisation, something of interest in developing countries where intelligence lies with the individual rather than with regulations.

Thank you Yorgos. It has been mentioned to me and I will check them out, maybe it will help me more in my research.

Hi Saad, Following on from your discussion with Michael about the suitability of roundabouts at intersections, there are certainly ample studies on the relative safety benefits of roundabouts compared to signals. Three features of the roundabout make it safer: a correctly designed roundabout produces adequate deflection to induce reduced speed, reducing overall KE of the crash. The deflection along with the central island produces more acute angles as opposed to the 90 degree collisions possible at signals, again affecting the overall outcome. The central island also virtually eliminates the potential for 90 deg collisions, the most severe of impact angles. In addition, conflict points are reduced from  32 to 8, reducing the overall potential of a crash at the intersection. Signals are not considered Safe System compatible.

With respect to traffic movement, it is true signals allow manipulation of traffic, and this is difficult with unsignalised roundabouts. However, a traffic flow advantage at roundabouts is the continuous movement possible during non peak hours. For example, assuming the pivot phase at a signalised intersection is for the major road, vehicles approaching from the minor road are likely to experience some delay even when no other vehicles are present, as the vehicle  would approach on a red light, and so would still need to come to a standstill, wait for the light to turn to green and then progress through the intersection, even with no vehicles present. At a roundabout, when there are no other vehicles present at the intersection, vehicles are able to progress through the intersection directly with no delay. I do believe there are some UK counties that install roundabouts for capacity reasons, maybe worth looking that up. Also, Turbos from The Netherlands have been found to improve capacity by 20% compared to traditional roundabouts.

Please see Traffic Engineering Book by McShane. It shall give you detailed procedure.

Thank Nimmi,

That is actually some very good points you made there, especially concerning the reduction of conflict points and the 90 angle accident. I think roundabouts are more safe than intersection whether signalized or not

As for the delay experienced in signalized intersection (or signalized roundabouts), there is a workaround in that area especially with the advance in adaptive traffic lights and would be considered more pedestrian friendly in my opinion,

Thanks.

Thanks Anwaar, I will give it a look.

It depends on the geometry of the intersection and on the traffic flows on the various legs. One typical configuration of unsignalized intersection has four legs, where one street - the major street - is uncontrolled, while the other street  - the minor street - is controlled by STOP signs. The level of service of this intersection depends on the values of vehicle delay and on the 95th percentile of the queue lengths. In general we can say that, when the delay exceeeds 50 s per vehicle, the intersection needs traffic signal installed. Another solution could be the transformation of the intersection into a roundabout. A thorough discussion on these subjects, and the presentation of the computation methods for both the delay and the queue lengths are in Chapter 19,  Vol. 3 of the Highway Capacity Manual (HCM) 2010.

If I understood your concern clearly, I suggest HCM and MUTCD as a starting point. I also like the idea of simulating the conditions and make a comparison between with and without a signal.

Thanks Hicham and Khaled :)

This is an interesting question. 

There are some criteria that could be used to measure the need of installing a traffic light in a given intersection. Among them: 

-The ratio of accidents occurring in that intersection with respect to the global accidents of a given population (normalized according to the characteristics of the network. 

-The traffic congestion around this intersection,

etc. 

Obviously, the optimal way to determine it would be to analyze (presentially or using sensors/cameras) that intersection during a large amount of time. 

Best regards, 

  Raul

No my speciality, sorry.

Dear Saad, 

Please refer to chapter 4 in the Manual on Uniform Traffic Control Devices (MUTCD) which is published by Federal Highway Administration. In this chapter, you will find the traffic signal warrants. You should check all of these warrants, the traffic control signal should not be installed unless one or more of these warrants are met.

Regards