Proceedings of the 2016 Winter Simulation Conference

T. M. K. Roeder, P. I. Frazier, R. Szechtman, E. Zhou, T. Huschka, and S. E. Chick, eds.

SIGNAL PHASE TIMING IMPACT ON TRAFFIC DELAY AND QUEUE LENGTH-A

Xiaobing Li, Asad J. Khattak, Airton G. Kohls

Department of Civil and Environmental Engineering, The University of Tennessee, Knoxville

ABSTRACT

Traditional intersection traffic signal control strategy is pre-determined signal with certain phase timing
length for each circle. Studies focusing on adaptive traffic signal strategy have somewhat achieved the
goal of reducing traffic system delay to some extent. However, few of them capture the benefit of using
the queue length as the criteria under the connected vehicle environment, and this paper focuses on firstly
identifying the potential saving of average system delay with agent-based simulation modeling, and sec-
ondly finding out the relationship between average system delay and average queue length for traffic ap-
proaching the signalized intersections. Through applying the agent-based simulation modeling approach
in AnyLogic, findings show that average system delay could be reduced using optimized parameters (e.g.
arrival rate, signal phase length, etc.), specifically, 5.29% saving of total average system time, 4%-28%
traffic queue reduction for different traffic lanes, and a positive relationship between average system de-
lay and the average traffic queue length is detected.

1 INTRODUCTION
Current situation of the traffic control strategies at the intersection has evolved and been improved
through applying the advanced control technologies during the last decades, especially for signalized in-
tersections around urban areas. Under the vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I)
connection environment, it is necessary to conduct an in-depth analysis about how these techniques could
be applied to improve the overall efficiency and safety conditions at intersections. This study is intended
to present an in-depth analysis about how to use the information from the connected vehicles to simulate
the real world situation, so that an optimal traffic signal control solution can be achieved at intersections.
Therfore, the research question raised in this study is whether it is more efficent to cross the intersections
under connected vehicle (CV) environment in terms of system delay and traffic queue length.

2 METHODOLOGY
The data used in this study is from City of Detroit Connected vehicle data environment, which can be
downloaded from Research Data Exchange website https://www.its-rde.net/. It was collected during a
queue length estimation field test being conducted in the Southeast Michigan test bed, during the 2014 In-
telligent Transportation Systems World Congress. The primary goal of this field test is to support a queue
estimation algorithm while using connected vehicles in a CV environment. This Data Environment in-
cludes 4 data sets: 1) Vehicle Situation Data, 2) Intersection Situational Data, 3)Traveler Situation Data,
and 4) the Queue Length Data sets. Agent-based simulation modeling is the adopted method towards
studying the traffic flow at intersections. By using the performance indicators such as intersection traffic
system delay, and queue length, we may have a general clue of how and to what extent the connected ve-
hicle information will impact on the intersection operations. Statistical data descriptive analysis is another
fundamental analysis method towards analyzing the basic frequentist and distribution information of the

3722
 Li, Khattak & Kohls

traffic flow and the output from statistical analysis will be used as the input parameters into the simulation
model, which will eventually affect the outcome of the simulation results.

3 ANALYIS RESUTLS
Based on the fundamental traffic flow theory, the relationship between flow, speed and density is shown
as 𝑞 = 𝑘 × 𝑣. Where 𝑞 = traffic flow, vehicle per hour per lane; 𝑘 = flow density, vehicle per mile per
lane; and 𝑣 = speed, miles per hour. In order to determine the traffic flow rate, density and speed is need-
ed. Statistical analysis results show that jam density is finally calculated to be 170 vehicles per mile per
lane. Using Greenshields’s linear model, 𝑣 = 𝑣𝑓 (1 − 𝑘⁄𝑘𝑗 ), where 𝑣𝑓 denotes free-flow speed, the post-
ed speed limit is 25mph, and 𝑘𝑗 is jam density. Then, 𝑞 = 𝑘𝑗 (1 − 𝑣 ⁄𝑣𝑓 ) × 𝑣 = 170(𝑣 − 𝑣 2 ⁄25).
Based on available data, the speed follows the triangle distribution in AnyLogic, the arrival rate is
calculated as Triangle (0, 517, 1063) representing number of vehicles coming to the intersection per hour
with minimum 0, maximum 1,063 and mean 517. Signal phase and timing (SPAT) is analyzed and shown
in Table 1.
Table 1: Signal phase and timing descriptive statistics.
Time to change(s)
Current signal state Frequency Min Max Mean Std. Deviation
0 195788 0.1 63.7 25.291 24.2264
1 / green 969970 0.1 48 8.616 10.1704
2 / yellow 260459 0.1 75 7.638 14.0866
4 / red 1418919 0.1 93.2 15.313 15.104
16 / left arrow 25253 7 50 45.97 2.8865

The simulation is based on intersection of Shelby and Larned road. Optimization procedures are
based on total system average delay, and traffic arrival rate are initially set as 15, 15, 20 and 6 vehi-
cles/min from top, bottom, left and right entrance into the system as shown in Figure 1. Through optimi-
zation procedure with 500 runs, new values for signal phase parameters are obtained, and results shows
that adaptive signal control can be further optimized with optimal control signal phase parameters, and to-
tal average time saving percentage is around 5.29%, and queue length reduction percentage ranges from
4% up to 28% under dynamic traffic flow environment.

Figure 1 Traffic flow simulation logic and signal control scheme.

4 CONCLUSION
By simulating the intersection and optimizing traffic signal parameters, I found that there is still potential
system average time savings and traffic queue length reduction, due to application of CV techniques.

REFERENCES

Maslekar, N., et al., CATS: An adaptive traffic signal system based on car-to-car communication. Journal
of Network and Computer Applications, 2013. 36(5): p. 1308-1315.
Guler, S.I., M. Menendez, and L. Meier, Using connected vehicle technology to improve the efficiency of
intersections. Transportation Research Part C: Emerging Technologies, 2014. 46: p. 121-131.

3723