Hello Everyone. Hope you all are doing well.
I am doing a calibration of VISSIM Network which has (15+ Intersections). I am looking for acceptable ranges for comparison of model output with data obtained from field. i.e. how much difference is acceptable.
Measures of Performace are
1. Travel Time
2. Maximum Queue Length
I have queu length, Travel time and Speed data of every single link.
Thank you very much.
You have to minimize the difference to the smallest. To confirm the significance of the difference(whether it is acceptable or not) you can adopt statistical tests. These tests will help you to know whether you need further tuning up the parameters. This is the case when you are doing manual calibration. If possible, you can adopt AI based calibration. For this you can see the following publication:
https://www.researchgate.net/publication/315716548_VISCAL_Heuristic_Algorithm_Based_Application_Tool_to_Calibrate_Microscopic_Simulation_Parameters
According to Peck (2004), a number of steps are required to build an accurate simulation models: (a) confirmation of the accuracy of the model, that it represents the actual phenomena; (b) analyse the full aspect of subject matter; and (c) to analyse the model using both uncertainty and sensitivity methods to understand the models behaviour. For the simulation to work and to produce reliable results, it requires a lengthy period of trial, errors and comparisons of theory and actual results of physical experiments, allowing for approximations, idealisation, falsification, and additional information (Winsberg, 2003). Peck further suggests that the type of work done using simulation models does not always require data, as in data collection, but it can use it if necessary in the process of testing an hypotheses that otherwise would be impossible to test. Peck makes reference to a simulation model being a real world problem simulated on an artificial world that can be controlled and manipulated by the researcher (Peck, 2004).
PECK, S. L. 2004. Simulation as experiment: a philosophical reassessment for biological modeling. TRENDS in Ecology and Evolution, 19, 5.
WINSBERG, E. 2003. Simulated Experiments: Methodology for a Virtual World. Philosophy of Science 70, 21.