This repository will provide an open and validated MATSim traffic model for the greater Melbourne area. The project is a collaborative effort between individuals and groups from RMIT University, University of Melbourne, CSIRO Data61, Swinburne University, KPMG Australia, and others.
The first release of the model is expected to be made in Mar 2018.
-
Ensure that you have installed all the pre-requisites as per the section below.
-
Get the project files using git using
git clone https://github.com/agentsoz/matsim-melbourne
-
In the
matsim-melbourne, click on the executablematsim-melbourne-0.0.1-SNAPSHOT-jar-with-dependencies.jar. This will launch the GUI as shown below:
. -
Change the Memory: field to
4000 MBor more, and select the Configuration file: inscenarios/2017-11-scenario-by-kai-from-vista/config.xml. Evidently, you could modify the material of a scenario, in particular the config file. For example, you may want to setlastIterationfirst to 0 and then to 1 until everything is running smoothly. -
Click Start MATSim. This will run the simulation (might take a few minutes) and put the results in
scenarios/2017-11-scenario-by-kai-from-vista/output. -
Now you can visualise the results in Senozon Via. Please refer to the Via user manual, available from https://via.senozon.com, for how to do this. A snapshot of the Via visualisation for the Vista simulation you just ran is below:
.
You will need Java installed on your machine to be able to run the MATSim Melbourne GUI. If unsure, go to https://www.java.com/en/ to find out if you already have Java installed, and/or to get the correct version of Java for your operating system.
If you are not going to be doing any development, but just want to use the tool, you only need the Java Runtime Environemnt (JRE) installed. * Note that this project is compatible with Java 8 and will likely not work for Java 9, as of 07-Feb-18.*
If you are just using the project from the browser, you can ignore this section.
If you want to pull a git clone, this is important: Given that the model will invariably be using large data files from various sources, we will use Git LFS support in GitHub for storing these. The idea is to keep all such files in the ./data directory. LFS is already set up to track any file in this directory, so there is nothing special you have to do, other than ensuring that you install Git LFS. Otherwise when you clone the repository you only receive the pointers to the large files and not the actual data.
The recommended way to make a copy of the project on your machine is to clone it using Git. This will ensure that all large files are correctly imported.
To be able to visualise the results, download Senozon Via together with a Free License. Note that the free version will limit the visualisation to 500 vehicles though.
An alternative is to use Google Earth. It can be a bit crude to use, but will also let you visualise the entire population. Ask us if you want to set this up instead (hopefully we will put instructions for that here soon though).
data contains original data. This is sometimes in MATSim format, because that was the way we received it, but more often it comes in other formats.
scenarios contains generated MATSim scenarios. In consequence, these are files on which a MATSim run can be started. There are often also intermediate files, such as plans after routing, since that accelerates MATSim runs a lot.
results contains output from MATSim runs. One should be a bit careful of putting this into git despite the large file support, but for small (1%) runs it seems feasible and is useful for many results. Maybe something like [http://www.mytardis.org] should be considered for larger outputs.
src contains source code
test contains test input files (for regression tests)
You will need the Java 8 JDK, and Maven to get started. You are a developer so probably know how to do that :)
To build the project, do:
mvn clean install
Find the CreateDemandFromVISTA class somewhere under src/main and look there.
To convert the Synthetic population generated using the latch algorithm to the MatSim syntax, use the command below. It
generates the output file saved as population-from-latch.xml.gz (use --file-format x to save in the xml format)
To parse the entire synthetic population
mvn exec:java -Dexec.mainClass="io.github.agentsoz.matsimmelbourne.demand.latch.CreatePopulationFromLatch" --output-dir . --run-mode
f --file-format z
To parse 100 people only from the synthetic population
mvn exec:java -Dexec.mainClass="io.github.agentsoz.matsimmelbourne.demand.latch.CreatePopulationFromLatch" --output-dir . --run-mode
d --sample-population 100 --file-format z
To convert the Synthetic households generated using the latch algorithm to the MatSim syntax, use the command below. It
generates the output file saved as households-from-latch.xml
mvn exec:java -Dexec.mainClass="io.github.agentsoz.matsimmelbourne.demand.latch.CreateHouseHoldFromLatch"
To generate the MatSim activity plans for the Synthetic population, use the command below.
It generates the output file for certain residence locations (progressive build-up by uploading MTWP files
generated from ABS Table builder and stored under data/census/2011/mtwp/2018-02-16-mtwp-files/)saved as
population-with-home-work-trips.xml.gz
mvn exec:java -Dexec.mainClass="io.github.agentsoz.matsimmelbourne.demand.latch.AssignTripsToPopulationv1"
"NOTE - The file 'AssignTripsToPopulation.java' was developed in an effort to calculate residence workforce proportions to decide the proportionate number of trips to destination locations. Significant number of trips are not assigned in using proportions and tying residence and workforce proportions"
UNUSED - To generate the MatSim activity plans using probabilistic calculation for transport mode and destination
location (considers people between the age of 15 and 84 to be working and, does not factor in person characteristics in
assigning trips), use the command below. It generates the output file saved as population-with-home-work-trips.xml.gz
mvn exec:java -Dexec.mainClass="io.github.agentsoz.matsimmelbourne.demand.latch.AddWorkPlacesToPopulation"
Then to run the built simulation with the generated MATSim population, run MATSimGUI in the matsim-melbourne project. This can
- either be done from the IDE
- or be done from the command line as follows:
mvn exec:java -Dexec.mainClass="org.matsim.gui.MATSimGUI"
Then choose the correct config file (presumably in the `scenarios' directory), then run.
If you are a bit more experienced, you could copy RunMelbourneTemplate.java, adjust to your needs, and run that one.
- Karthikey Surineni, RMIT University
- Claire Boulange, RMIT University
- Dhirendra Singh, RMIT University
- Jonathan Arundel, RMIT University
- Kai Nagel, TU Berlin
- Leorey Marquez, Data61/CSIRO
- Lin Padgham, RMIT University
- Nicole Ronald, Swinburne University
- Renan Grace, KPMG
- Roberto Sabatini, RMIT University
- Sara Moridpour, RMIT University
- Stephan Winter, University of Melbourne
- Zahra Navidi, University of Melbourne
Open source license still to be agreed.