E3S Web of Conferences 471, 04027 (2024) https://doi.org/10.

1051/e3sconf/202447104027
TITDS-XIV-2023

Using digital twins to manage traffic flows

Yaroslav Shamlitsky*, Ovsyankin Aleksey, Evgeny Morozov, and
Tatiana Strekaleva
Reshetnev Siberian State University of Science and Technology, Krasnoyarsk, Russia

Abstract. This paper explores the use of digital twins for traffic flows.
Explores how digital twins can be applied to effectively manage traffic
flows, including improving road safety, optimizing vehicle flow, and
improving overall transportation system performance. Discusses the various
technologies used to create digital twins, as well as the methods and
algorithms that can be used to analyze data and make decisions based on
information obtained from digital twins. The article also discusses the
potential benefits and challenges associated with the use of digital twins for
traffic flows, as well as possible directions for future research. This article
provides useful information and recommendations for professionals in the
field of transport management and technology, as well as for solving
problems in the field of traffic flows.

1 Introduction
Modern cities face increasing challenges in managing traffic flows caused by an increase in
the number of cars on the roads, congestion of public transport and inefficient use of
infrastructure. These factors lead to traffic jams, delays and a poor travel experience for city
residents.
In recent years, digital technologies and the concept of digital twins have become
widespread in various industries, including transportation system management. A digital twin
is a virtual model of a real object or system that reflects its state, behavior and interaction in
real time. The use of digital twins in traffic flows offers new opportunities for effective
management and optimization of transport infrastructure [1].
Digital twins can be used to improve road safety by proactively identifying potential
hazards and preventing accidents. They can also be used to optimize vehicle flow, predict
and avoid congestion and congestion, and improve the overall performance of the
transportation system. Digital twins allow you to analyze a variety of data collected from
various sources, such as sensors, surveillance cameras, mobile devices and transport
infrastructure management systems. This allows you to make informed decisions based on
real data and predict future trends in traffic flows.
However, the use of digital twins for traffic flows also presents challenges and problems.
For example, it is necessary to ensure that the data used to create digital twins is accurate and
up-to-date.

* Corresponding author: [email protected]

© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons
Attribution License 4.0 (https://creativecommons.org/licenses/by/4.0/).
E3S Web of Conferences 471, 04027 (2024) https://doi.org/10.1051/e3sconf/202447104027
TITDS-XIV-2023

In addition, it is necessary to develop effective algorithms and data analysis methods to

extract valuable information from digital twins and make effective decisions.
The purpose of this scientific article is to study the possibilities of a systems approach in
the design of an intelligent transport system in the region as a structural component of traffic
flow management.

2 Materials and methods

Due to the growth of technological progress in large cities, there is a sharp increase in the
number of both personal cars and ground transport in general, which ultimately leads to
overcrowding of the transport network. To avoid such a problem, it is necessary to model a
digital twin of a section of the transport network, describing the behavior on the road of both
individual cars and their groups, called traffic flows.
Next, we will model a section of the road network using the software tool for developing
simulation models Anylogic.

2.1 Modeling a section of the road network

Let's simulate the traffic situation at a road intersection in the AnyLogic environment [2].
Before building the model in the AnyLogic program, a computational and practical
experiment was carried out at the intersection of streets in the city of Krasnoyarsk (Figure 1).

Fig. 1. Intersection of streets in Krasnoyarsk.

2.2 Experimental procedure

To conduct research on the algorithm, an experimental technique was developed [3].
The input parameters of the model are:
1. Intensity of arrival of vehicles from the northern IN, eastern IE, southern IS and
western IW approaches, vehicles/min.
2. Distribution of cars in directions (left, straight, right) for each approach, %:
a. for the northern approach: pNleft, pNstraight, pNright;
b. for the eastern approach: pEleft, pEstraight, pEright;
c. for southern approach: pSleft, pSstraight, pSright;
d. for the western approach: pWleft, pWstraight, pWright.
The test procedure is as follows:
1. First, a test program is drawn up, based on various combinations of sequential changes
in the input parameters of the model.
2. Next, for each test, the durations of the main clock cycles are calculated using the
generally accepted Webster algorithm for hard algorithms.

2
E3S Web of Conferences 471, 04027 (2024) https://doi.org/10.1051/e3sconf/202447104027
TITDS-XIV-2023

2.1. Saturation flux is calculated:

��� ⋅�� straight ��� ⋅�� left ��� ⋅�� right �
� ��⋅�⋅ , (1)
��� ⋅�� straight ��,��⋅�� ⋅�� left ��,��⋅�� ⋅�� right �
where P – traffic lane capacity, units/hour; n – number of lanes open to traffic; ID –
traffic intensity at the considered approach to the intersection, driving units/hour; pDпрям,
pD , pDправ – share of cars traveling in the indicated direction, %.
лев

2.2. The phase coefficient is calculated:

y = ID / M (2)
In each phase, an approach or a dedicated direction with the maximum phase coefficient,
that is, the most loaded, is selected. It is the limiting one [4].
2.3. Calculation of transition cycles is performed t`.
In general, transition intervals should not be scheduled to last less than 3 seconds. If the
found value t´ does not exceed 4 sec., then the transition interval consists of one clock cycle
(yellow signal). At t´ = 5…8 the transition interval must be composed of two auxiliary
measures. To simplify calculations, we assume the duration of transition intervals t` = 3 sec.
2.4. The duration of the regulation cycle is calculated:
��,�⋅����
𝑇𝑇 � (3)
��∑ ��
where L – is the sum of all auxiliary cycles, s; yi – phase coefficients.
2.5. The duration of the main measures is calculated:
���
𝑡𝑡� � �� ⋅ ⋅ 𝑇𝑇 (4)
���,�⋅���
3. The obtained values of the durations of the main measures are compiled into a separate
file.
4. Next, the model is launched and in the execution mode the values of the input
parameters are set in accordance with the experiment number, the operating mode of the
model is “Hard Logic” and the number of the hard control logic is set.
5. After setting the parameters, the maximum execution speed is set and the moment
when the average delay time reaches a stable state is recorded. When a stable state of the
average delay time is reached, its value is taken and recorded in the corresponding cell of the
experimental data processing sheet. Then the model execution mode is exited.
6. Next, the model is restarted, in the execution mode the values of the input parameters
are set in accordance with the experiment number, the operating mode of the model is set to
“Adaptive Logic”, the maximum duration of the green signal is set equal to the maximum
value of the duration of the main clock cycles for the strict control algorithm, the crew time
is set to 4 sec.
7. After setting the parameters, the maximum execution speed is set and the moment
when the average delay time reaches a stable state is recorded. When a stable state of the
average delay time is reached, its value is taken and recorded in the corresponding cell of the
experimental data processing sheet. Then the model execution mode is exited.
8. Then steps 6 and 7 are repeated again with the only difference that in the first repetition
the maximum duration of the green signal is set 2 sec. more, in the second - 2 sec. less and
in the third - 5 sec. less.
9. Then the tests are repeated for each experiment number (steps 4...8). Thus, a complete
study of the model's performance consists of 150 runs.
10. At the end of a series of runs, the difference between the average delays of rigid and
adaptive control is calculated, graphs are drawn, and conclusions are drawn.
With the help of simulation, real-life situations can be easily reproduced, sensitivity
analysis of different scenarios can be performed through simulation experiments, and thus
provide a better understanding of the factors affecting the system.

3
E3S Web of Conferences 471, 04027 (2024) https://doi.org/10.1051/e3sconf/202447104027
TITDS-XIV-2023

3 Results
The AnyLogic 6.9.0 University modeling environment was chosen as the platform for
creating the simulation model [5].
The developed simulation model of work contains the following structural elements:
 four-way intersection with adjacent roads on each side;
 each road has a median strip dividing the road into two carriageways with three
lanes in each direction;
 a system for generating model agents – two types of vehicles: a passenger car and a
truck with a carrying capacity of 2 to 6 tons;
 four blocks of vehicle movement logic (one for each direction);
 models of light signaling elements;
 state diagram modeling the logic of switching traffic light signals based on two-ring
phasing;
 a system dynamics module that models the operation of a local intersection
controller;
 block and elements of model control parameters;
 module for reading rigid control cycle parameters;
 module for collecting statistics of agent parameters, a histogram of vehicle delay
distribution, also showing the average delay time of all vehicles throughout the
experiment.
In the model execution mode, an animation is displayed, which is a two-dimensional plan
of the simulated system with vehicles moving along it (Figure 2).

Fig. 2. Model operation in execution mode.

A simulation model developed using AnyLogic can be used to support decision-making
and covers all stages of conducting research using simulation modeling [6-7].
One of the key features of a transport digital twin is the presence of two-way
communication between the model and the real physical object. Let's try to get data from
traffic lights in real time. The simulated process is controlled using a specially created control
unit shown in Figure 3.

4
E3S Web of Conferences 471, 04027 (2024) https://doi.org/10.1051/e3sconf/202447104027
TITDS-XIV-2023

Fig. 3. Modeling process control unit.

To reverse the change in intersection regulation modes based on the simulation model,
further interaction with city traffic control services is required.

4 Conclusion
In conclusion of the work done, we can say with confidence that the method of computer
modeling of traffic flows, taken as the basis for assessing the throughput of elements of the
street road network, based on the technology of computer modeling of vehicle movement in
various road conditions, made it possible to take a new approach to the formulation and
solution road capacity problems.
This scientific article conducted a study and analysis of the use of digital twins for
managing traffic flows. As a result, the use of digital twins in transport infrastructure
management offers many benefits and opportunities to optimize and improve transport
systems.
Digital twins provide a realistic view of the health and behavior of vehicles and
infrastructure in real time. They allow you to analyze large amounts of data collected from
various sources and use this information to make informed decisions. The use of digital twins
can improve road safety, optimize vehicle flow and improve the overall performance of the
transport system.
Overall, the use of digital twins for traffic flows represents great potential for improving
transport infrastructure and optimizing transport systems. Further research and development
in this area may lead to the development of new innovative solutions and methods that will
help cope with the problems and challenges associated with traffic flows.

References
1. Y. Zhang, Y. Liu, IEEE, 8 (2020)
2. X. Li, D. Wang, X. Zhou, IEEE Transactions on Intelligent Transportation Systems 20,
10 (2019)
3. J. Kim, S. Kim, J. Park, IEEE Transactions on Intelligent Transportation Systems 20, 9
(2019)
4. A. G. Levashev, VINITI, 3 (2004)

5
E3S Web of Conferences 471, 04027 (2024) https://doi.org/10.1051/e3sconf/202447104027
TITDS-XIV-2023

5. J. Banks, J. S. Carson, B. L. Nelson, D. M. Nicol, Discrete-Event System Simulation

(Pearson Education, 2005)
6. A. M. Law, W. D. Kelton, Simulation Modeling and Analysis (McGraw-Hill Education,
2015)
7. M. P. Pechersky, Automated traffic control systems in cities (Transport, 2007)