Application of microsimulation models in traffic analysis is a common professional practice. Methodology of calibration of microsimulation models is not finally adopted and various approaches are available. One of the available calibration methods is neural network approach for calibration of microsimulation traffic model. The comparison of the simulated and measured traffic indicators, in real traffic conditions, provides the best insight into the success of the model calibration process. The traffic indicator, used for the calibration of the model, is the travel time between the measuring points at a chosen urban intersection. The calibrated model has predicted the travel time for new sets of measured data at the same intersection with the prediction error smaller than 5%. This paper analyses the simulation results for the traffic indicators that were not used in the model calibration – the queue parameters. The selected queue parameters are the maximum queue at the entrance and number of stops at the intersection entrance. The model has been additionally applied to the other intersection, in order to simulate its queue parameters. This has provided us with an insight into the issue of whether the calibration model is applicable only to the intersection for which the calibration has been done or it can have a wider application. The VISSIM microsimulation traffic model was used for calibration, and two single-lane roundabouts served as the research basis for evaluation of the calibrated traffic microsimulation model by means of queue parameters.

queue parameters; microsimulations; VISSIM; roundabouts