A new methodology to calibrate the congestion wave for the Cell Transmission Model for Urban Traffic
Even though trraffic simulation models offer the possibility to select entry parameters based on the available transport literatures, it is necessary to calibrate these models according to a specific type of transport network. Such calibration procedure involves the regulation of the parameters until there is a correct relation between the data collected by direct observation and those reproduced by the model. This calibration process requires a great effort in terms of time and resources. It is also possible to solve the calibration problems by optimization method, where a combination of values that best satisfies an objective function is searched. It is obvious that the microscopic simulation requires greater efforts in calibration than the macroscopic one. Studies on calibration
are focused on minimizing the error between the observed data and the simulation.
Originally, the Cell Transmission Model (CTM) has been designed to provide a simple representation of traffic on a highway. To extend and to apply the CTM in the urban contest is necessary to set the variables of the fundamental diagram appropriately. We present a new model to calibrate the supply constraints of each link. This model allows the variation of the congestion wave speed maintaining the same values of max capacity, free flow speed, critical and jamming density of the triangular fundamental diagram defined . In the following figure is possibile note the differtent type of calibration of the congestionwave speed for the triangular fundamental diagram.
The calibration of a new congestion speed wave is obtained by an optimization method. Via optimization we obtain the minimization of the error between the congestion wave by CTM and real data (or microscopic model). We can calibrate accurately the traffic values that represent the congestion wave for each link. The results demonstrate that the application of the calibration method improves the behavior of macroscopic model obtaining an error under 1% in comparison with a representation made with a microscopic model.
More details about a new methodology to calibrate the congestion wave can be found on the following paper:
