Cottbus Scenario

The network of the last section is clearly not realistic. To simulate a real-world scenario, at least some fixed-time signal schedules and junction layouts are required. Data for Zurich, Munich and Berlin was inspected. In part, fixed-time schedules are available.

The matching of traffic signal locations to junction layouts was not covered by all data in a machine processable data format⁷. Within the ADVEST project, fixed-time control schedules and turn pocket layouts were recorded manually⁸ for the work presented in Köhler and Strehler (2010). In the following a scenario is set up around this data.

7The Berlin network has approx. 2000 signalized junctions. Each is controlled by up to 4 fixed-time schedules for weekdays. On top of the fixed-time control some logic programs are set up for many junctions. The layout of junctions is given as technical drawing that is not connected to the control schedules. Converting this data manually is not an option.

8Acknowledgements to Martin Strehler, BTU Cottbus

(a) Cottbus network and municipality borders (b) Cottbus network and municipality borders

The scenario is located in the federal state of Brandenburg, in Germany. It covers the area of the administrative district “Spree-Neiße” that is enclosing the city of Cottbus, plus the City of Cottbus itself.

5.2.1. Network & Population

The network is taken from openstreetmap data⁹. The network is created for an 100 %

Network

sample. In the administrative district “Spree-Neiße” only the main roads are included while for the city of Cottbus also side roads are considered. The network consists of 4’417 nodes and 10’600 links and is depicted in Fig. 5.5a. Fig. 5.5b shows the network on top of the “Corine Land Cover” landuse (European Environment Agency, 2011) provided by European Environmental Agency. In green forests and agricultural areas are depicted.

In the city of Cottbus live around 100’000 inhabitants while approx. 128’000 people

re-Population

side in the administrative district Spree-Neiße. The synthetic population used for the simulation is based on data taken from the German employment agency (Wiethölter et al., 2010). The data contains the number of commuters for each 2-tuple (home–work)

9state 09-2010, seewww.openstreetmap.org, last access 03.10.2013. Technically, the open street map con-version of MATSim is usedOsmNetworkReader.

(c) Home locations (d) Work locations

Figure 5.5.: Synthetic population for the Cottbus scenario, geospatial locations of activ-ities

of municipalities in Germany. Virtual persons in MATSim need a geographic coordi-nate for their activities. If this coordicoordi-nate is drawn randomly, solely based on munic-ipality boarders, home and work activity locations are uniformly distributed over all the area, i.e., most of them in woods and fields. Thus, activity locations are drawn ran-domly in combination with the landuse data. The coordinate has to be in the area of the municipality. In case of a home activity, it must be located in urban fabric areas while in case of a work location, also industrial or commercial areas are allowed. The resulting home activity locations are shown in Fig. 5.5c, while Fig. 5.5d shows activity locations for work.

The work activity must start between 7 and 9 am; initially every commuter starts at a random time in this interval and ends work 8.5 hours later. Work must end before 6 pm, afterwards no further utility is gained by performing an activity of type work.

The typical durations of home and work activity are set to 15.5 hours and 8.5 hours, respectively. The modal split for the area of interest can be taken from the base year of ITP/BVU (2005) and is set to 55% car trips. This results in 33’479 commuters travel-ling by car.

(a) Location within city of Cottbus (b) Signalized area in detail Figure 5.6.: Cottbus network, area with traffic signals

5.2.2. Traffic Signals

The fixed-time control is taken from Köhler and Strehler (2010). Due to the higher

Fixed-time Control

resolution of the transport network, some of the originally recorded fixed-time control schedules are invalid and removed, data for 22 junctions is available. Fig. 5.6 shows their location on the transport network. All signal control plans have a cycle of 90 seconds and run all day (00:00 to 24:00). Green splits are taken from the system that was run in 2009, and offsets are the result of the optimization presented in Köhler and Strehler (2010). The demand used for optimization differs from the commuter demand used in this work. This reflects the typical situation of optimized fixed-time control:

Signals are optimized to a certain demand once, but while the demand changes over time the fixed-time control is not re-adjusted (Bell and Bretherton, 1986).

5.2.3. Base Case

Abase case for further Cottbus simulation runs is computed using the network,

syn-Simulation Setup

thetic population and traffic signal control. The simulation is run with the commuter population until the outcome seems stable, in this case for 500 iterations. In each it-eration, 10 % of the commuters can choose new routes while another 10 % can vary their departure times. The only available mode is car. Then innovation is switched off,

0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000

04:00 08:00 12:00 16:00 20:00 00:00 04:00 08:00 12:00 16:00 20:00

# travelers

time of day [hh:mm]

en-route departures arrivals

(a) 0th iteration

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00

# travelers

time of day [hh:mm]

en-route departures arrivals

(b) 1000th iteration

Figure 5.7.: Cottbus base case: After 1000 iterations a route and time distribution is learned

i.e., another 500 iterations are run where no new plans are generated. Each commuter chooses out of a set of 4 plans using the multinomial logit model, see Sec. 2 for details.

The network is created for a 100 % sample. The demand comprises commuter traffic, but other traffic is not included, e.g., commercial traffic. The scale parametersα_{f low}and αstorageof the traffic flow model are both set to 0.7.

The resulting relaxed transport demand serves as input for the for further simulations. ^Results In contrast to the initial synthetic population the relaxed transport demand features a

route and time distribution that is learned according to the constraints of the transport network and the utility function. The results of this learning process are shown in Fig. 5.7 that depicts the number of travelers departing, arriving or traveling over time of day for first and last iteration.

Clearly, this scenario has many free parameters that are adjusted by rule of thumb only. ^Discussion For other available scenarios, more data and better calibration is available, but neither

signal control data nor junction layouts. For a simulation of traffic signal control the Cottbus scenario is the best available scenario. In comparison to toy networks it features many artefacts of real-world networks.