03. What is SUMO - dataandcrowd/GlasgowLEZ_SUMO GitHub Wiki

What is SUMO?

This project is using Simulation of Urban Mobility (SUMO), an open-source simulation platform which handles large road networks in a continuous geographic space 19(https://www.zotero.org/google-docs/?mWa8iq). This section explains the basics of SUMO and why we try to answer our questions using the platform.

SUMO

SUMO was developed by the German Aerospace Center in 2002 under a General Public License, with the goal of promoting transportation innovation across various sectors and communities 22, 23(https://www.zotero.org/google-docs/?XLtVE9). SUMO's primary function is to simulate traffic flows in cities or identify the routes of individual vehicles. However, what makes this model special is its ability to incorporate code from different libraries into its own model. In our study, we focused on two models for pollution emissions: HBEFA and PHEM. HBEFA, or The Handbook Emission Factors for Road Transport (3rd edition), provides pollution factors such as CO2, CO, HC, NOx, and PMx by vehicle category and fuel type. PHEM (Passenger Car and Heavy Duty Emission Model), on the other hand, determines vehicle emissions according to European emission standards.

Why is SUMO selected?

We chose this model for both technical and financial reasons. Firstly, the graphical user interface (GUI) makes it easy to import roads in shapefile (.shp) format, allowing us to quickly check for any issues like disconnection points (as shown in Figure X). Typically, we obtain street data from OpenStreetMap, which includes nodes, links, and turn signals. Secondly, SUMO has the ability to collaborate with various models, including an emission model. iTETRIS, a model developed by the German government to reduce emissions from vehicles 23, 24(https://www.zotero.org/google-docs/?Flr6nC), is one such example. Additionally, since SUMO is written in C++ and Python, vehicle emissions can be calculated on a High-Performance Computing such as the cluster located in our unit at the University of Glasgow. Lastly, it's important to note that both SUMO and iTETRIS are freely available software packages that can be downloaded and used on Windows, Linux, or Macintosh operating systems. Alternatively, we are also considering using GRAL (Graz Lagrangian Model), which takes into account the city's meteorological conditions and morphology.

Example of SUMO GUI (Left), and a map of CO2 emission (Right) (Krajzewicz et al. 2015)

How does SUMO work?

The overall process involves three main steps: network generation, demand generation, and simulation (as shown in Figure X). The first step is network generation, where the digital road map is imported and checked for completeness. The netconvert function can read codes from other traffic simulators, such as Visum, Vissim, or MATSim. OpenStreetMap is now the most commonly used format in SUMO 22(https://www.zotero.org/google-docs/?1kyjQ1). The second step is demand generation, which involves creating vehicles and defining their routes through the network. Each vehicle is given a unique identifier, departure time, and route. SUMO has built-in route functions, allowing it to run simulations with a large number of vehicles without the need for manual traffic demand control. Origin-Destination matrices are commonly used in city-scale analyses, providing estimates of the number of vehicles moving between specific origins and destinations per hour. There are two versions of SUMO for simulation: a pure command line version and a command line version with a graphical user interface (GUI). The GUI allows for customization of the visualization, such as adjusting car speeds and signal waiting times or tracking individual vehicles. In the simulation stage, SUMO uses a time-discrete simulation, with a default length of 1 second that can be scaled down to 1 microsecond. The maximum duration of one simulation run is about 49 days.

During the simulation, each vehicle's speed is controlled by considering various factors, such as the distance to the leading vehicle, junctions, and lane changes. These car-following models were first created by Brockfeld et al. 25(https://www.zotero.org/google-docs/?7Y7J5A) and have since been expanded to more sophisticated models, including the intelligent driver model (IDM), Wiedemann model, and three-phase model. Results are exported in XML format as time steps, single vehicles, or aggregated measures of all streets.

Overall procedure of SUMO modelling

Plans to mount SUMO with dispersion models

To ensure accurate results, this project will first verify that the road network is properly connected, as fragmented nodes can lead to traffic congestion and introduce errors in the simulation. Once the road network is verified, each vehicle will be assigned an emissions function based on its fuel type and engine displacement. SUMO provides a built-in emissions function called PHEM-lite (or PHEM for the paid version). After setting up the emissions model, we will integrate it with a dispersion model. We will use a dispersion model called R-Line, which was developed by the Community Modelling Analysis System (CMAS) team at the University of North Carolina at Chapel Hill and is available for free. R-Line is a steady-state plume dispersion model that integrates point sources along roads 26(https://www.zotero.org/google-docs/?sSb1NH), and accounts for vertical wind effects and low-wind situations based on the Monin-Okukhov equation. We note that R-Line is a research model and not a regulatory model 26(https://www.zotero.org/google-docs/?tmZpIb).