Methods for estimating congestion by the use of GPS data

Student thesis: Master thesis (including HD thesis)

  • Niels Thorup Andersen
  • Kristian Skoven Pedersen
With the purpose of describing and confining the initiative problems regarding estimation of congestion by the use of GPS data the consequences of congestion is initially described. Additionally, a review of both traditional and more recent methods for collecting data on the area of traffic engineering has been conducted. Through literature studies it is sought to elucidate diverse definitions of congestion with the selection of a generally applicable definition in mind, which will be used in this project. This has resulted in the following definition of congestion, which focuses on the phenomenon of congestion and not on what causes it or the consequences hereof: “Congestion reflects the inconveniences motorists impose on each other in the form of reduced freedom of movement when they move about the traffic system.” Furthermore, a selection of national as well as international projects that are concerned with estimating congestion by the use of data from fleets of vehicles have been examined with focus on fleet sizes and logging frequencies, among other things. Based on the above the overall problem formulation is as follows: How can a general method for registration, analysis and quantification of congestion be developed so that it on background of GPS data from a fleet of vehicles can be utilised for any given road system? In order to answer this question the project’s analyses has been constructed as before and after studies. The overall analytical method thereby consists of a comparison of a situation with congestion and a situation without congestion – a free-flow situation. This comparison is carried out in practice by using a number of software programmes that have been developed in co-operation with a computer science student. These programmes have been designed from the consideration that traffic consists of a number of flows. Thus, an overall geographical limitation for the collection of data is selected along with two boxes whereby the programmes aggregate the individual loggings to trips in the flows between the defined boxes. On the basis of the selected trips actual cruising speed or travel time can be determined for the individual flows as indicators for the actual situations for which the degree of congestion is wished determined. As a direct consequence of the above-mentioned definition of congestion the situation without congestion – the free-flow situation – must be defined as a situation where the individual driver is not influenced by additional traffic. Hence, different methods for approaching such situations are described, evaluated and tested, which have resulted in that free-flow speeds for this project are determined as average cruising speeds in the time period from 8 PM to 6 AM where the possibility of free-flow situations are greatest. In order to avoid influences from signal control systems, which can be characterised as indirect influences of additional traffic, all trips that contain one or more speed loggings on 0 km/h are segregated. Furthermore, a method by which all speed loggings that exceed the actual speed limit are lowered to match this speed limit. By using this method it is avoided that violations of the actual speed limit can raise the calculated free-flow speed and thereby also the recorded congestion and the socio-economic costs hereof. In the execution of a validation of the GPS data set it is concluded that the analyses’ degree of detail with a data set this size should be limited to an estimation over time periods of three hours, which are summarised over all weekdays in the period in which data has been collected. The validation has been carried out as a comparison between GPS data for a selected section of Østre Allé and a mechanical conducted stationary traffic count from the aforementioned section. Additionally, a comparison between delays determined on the basis of the GPS data set and delays calculated with the software programme Dankap has been conducted. When there is an expression for cruising speed in situations with and without congestion respectively the congestion related expenditure of time – the delay – can be determined if the length of the distance travelled is known. By using traffic counts for the analysed periods of time the calculated delays can be scaled up to represent the total traffic in the analysed flows. Thus, the total socio-economic costs associated to the congestion in the analysed periods of time are estimated by introducing socio-economic uniform prices for expenditure of time. The above described procedure can scale up the calculated delays so that it represents the total intensity of traffic on a stretch of road or in an intersection. By applying the consideration of traffic as flows to a whole road system the described methods could also measure the total delays and costs of a whole road system.
Publication date2007
Number of pages128
Publishing institutionAalborg Universitet
ID: 10587417