TY - GEN

T1 - The Impact of Data Type on the Accuracy and Speed of Equilibrium Flow Assignment Techniques

AU - Крылатов, Александр Юрьевич

AU - Король, Максим Максимович

AU - Раевская, Анастасия Павловна

PY - 2025/8/29

Y1 - 2025/8/29

N2 - The equilibrium flow assignment problem is an optimization problem, which allows transportation engineers to estimate the usability and congestion on routes of an available network. Equilibration techniques appeared to be among the very first ideas on solving the problem due to their simplicity. Indeed, transferring pieces of flow between the available routes in order to shift a flow from a route with a longer period of travel time to a route with a shorter period of travel time seems to be quite a natural way for solving the equilibrium flow assignment problem. The approach demonstrates itself to be fruitful and easy to apply from a computational perspective. As a computational result, this method returns assignment patterns that provide equal travel times on used routes between origin-destination pairs. However, when we require the highest level of accuracy, we face multiple re-assignments of very small pieces of flow that leads to needs in additional computation efforts. In other words, the higher precision, the slower flow equilibration. In the present paper, we provide the computational results of a computational strategy that combines two equilibration procedures we developed for different data types. The first procedure obtains a fast but not so accurate result, while the second one uses the output of the first as the initial solution to achieve higher precision. We show that the combination of procedures demonstrates better performance compared to their separate usage. We believe our findings can give fresh insights to transportation engineers.

AB - The equilibrium flow assignment problem is an optimization problem, which allows transportation engineers to estimate the usability and congestion on routes of an available network. Equilibration techniques appeared to be among the very first ideas on solving the problem due to their simplicity. Indeed, transferring pieces of flow between the available routes in order to shift a flow from a route with a longer period of travel time to a route with a shorter period of travel time seems to be quite a natural way for solving the equilibrium flow assignment problem. The approach demonstrates itself to be fruitful and easy to apply from a computational perspective. As a computational result, this method returns assignment patterns that provide equal travel times on used routes between origin-destination pairs. However, when we require the highest level of accuracy, we face multiple re-assignments of very small pieces of flow that leads to needs in additional computation efforts. In other words, the higher precision, the slower flow equilibration. In the present paper, we provide the computational results of a computational strategy that combines two equilibration procedures we developed for different data types. The first procedure obtains a fast but not so accurate result, while the second one uses the output of the first as the initial solution to achieve higher precision. We show that the combination of procedures demonstrates better performance compared to their separate usage. We believe our findings can give fresh insights to transportation engineers.

KW - Wardrop principles

KW - equilibrium flow assignment

KW - user-equilibrium

UR - https://www.mendeley.com/catalogue/41831546-835a-33a6-8b33-008e7431246b/

U2 - 10.1007/978-3-031-96759-7_22

DO - 10.1007/978-3-031-96759-7_22

M3 - Conference contribution

SN - 9783031967580

T3 - Lecture Notes in Networks and Systems

SP - 325

EP - 333

BT - Artificial Intelligence and System Engineering (CoMeSySo 2024)

T2 - The Computational Methods in Systems and Software

Y2 - 12 October 2024 through 14 October 2024

ER -