| This research proposes and develops a novel and timely prototype model to estimate traffic impact under special events. Currently, researchers are using other traffic models that were developed for other purposes instead of traffic impact studies under special events with little success of applications. In this research, a systematic approach is used to study special event impact. It includes building a small network around an activity center based on Geographic Information System, and applying a mathematical model to estimate traffic impact on each link for each time period.; First, this research proposes two mathematical formulations: The Calibration Model and The Full Model. These formulations convert an engineering problem into a mathematical problem. A couple of methods are developed to collect field data for the calibration of the proposed model. The impact patterns in time and spatial dimension are obtained as a result of model calibration. The impact pattern in time dimension is considered as temporal factors and in spatial dimension as spatial factors.; Second, travel time reliability in traffic prediction is studied, which represents measurement of performance of the roadway network. Based on the BPR formula, travel time reliability is calculated and results are analyzed.; At last, the Special Event Assessment Package is developed. It is a GIS application used to set up an area for traffic impact assessment and prediction near the event site. This package is capable of building the special events network of the top 101 cites in the United States with the availability of GIS data.; With field data on I-85 near Georgia Dome in Atlanta City, Georgia, spatial and temporal factors have been estimated and categorized into different groups. According the number of attendees, those factors are then used to generate traffic impact prediction for the next football game. It has been found the estimated traffic impact that measured from field data match very closely within an average error of 20%. Additional tests using data from other places also support this methodology. With further testing and improvement, the modeling approach can be improved and extended potentially for disaster evacuation planning and management for other types of natural emergencies. |
