Traffic jams have become a sad part of everyday life for many people even beyond large cities. Of course, jams are the symptom of lack of throughput capacity of streets and roads. But sometimes the reason for congestion is just poor traffic flows management – in this case wasting time becomes especially disappointing.
The typical case of poor traffic flows management is inefficient signaling. Many Russian drivers can easily remember the situation when precious green seconds are on when nobody can use them while suddenly turned on red light stops a whole bunch of approaching cars. If traffic lights work like this, simple driving ahead through wide street can turn into a nightmare.
The unstable, “stop-and-go”, traffic flow is not only annoying for drivers but also has a negative impact on environment as acceleration requires 2-3 times more fuel than maintaining certain speed.
The solution is to adjust traffic light phases considering not only the single crossing but the whole chain of crossroads over the main street. This type of signaling is called a “green wave”. The principle of a green wave is the following: a buch of cars is formed at the first crossing and passes the whole chain of crossroads on green light that is turned on at each following crossing with a certain delay.
Knowing the average flow rate and speed, it is, of course, possible to estimate the signal phases (duration and time offsets) for a green wave. But such estimation will not consider the following significant factors:
- Flow speed changes under the impact of cars turning to main street from joining streets
- Non-linearity of flow density and speed on acceleration (for example, during the first 5 seconds of green light much less cars leave the crossing than from 20th to 25th second)
- Lane changings during the movement through streets
- Difference in cars speed and acceleration rates
All these factors (an some others) are considered in the simulation model below. The model was built for reasoning of green wave implementation on Dimitrovgradsky ave. in Ulyanovsk, Russia. The model was built on the real layout of the area.
Model used search through signaling durations and offsets to find the phases allowing to maximize average daily speed through the area. The real traffic flow data was used as input data to the model. Current traffic light phases were chosen as a base (“AS-IS”) case for estimation the effect of green wave implementation.
Modeling showed that by optimizing traffic light signal timings (particularly, implementation of green wave), it is possible to increase the average speed on Dimitrovgradsky ave. in Ulyanovsk by more than 10% (from 28.4 to 31.6 kph) and decrease the average stops count from 1.9 до 1.1 stops
Launch the simulation model made for this research and check the results yourself:
Navigation in the model:
- Use CTRL + mouse wheel for zooming the map
- Use mouse with right button for panning the map
- Use area scheme at top right corner to quickly jump to views of crossroads
Before launching the experiment you can choose “AS-IS” or “TO-BE” phases or set any phases you like manually. The interface for signaling setup is shown below:
Each phase is determined by duration of red and green lights as well as offset from time of model start.
Demo model summary
|Name||Optimization of Traffic Light Signal Timings via Simulation|
|Industry||Automotive traffic, transportation|
|Business problem statement||To what extent is it possible to increase the average speed on Dimitrovgradsky ave. in Ulyanovsk, Russia by optimizing the traffic light signal timings?|
|Model input||Road plan of the area, vehicle flow rates in all directions, cars’ speed performances, traffic lights’ timings|
|Model output||Average speed of passing the area, average stops count while passing the area|
|Modeling method||Agent-based simulation|
|Modeling result||By optimizing traffic light signal timings (particularly, implementation of green waves) it is possible to increase the average speed on Dimitrovgradsky ave. in Ulyanovsk, Russia by more than 10% (from 28.4 to 31.6 kph) and decrease the average stops count from 1.9 до 1.1 stops|