Analysis of Travel Time Reliability on Indiana Interstates

Travel-time reliability is a key performance measure in any transportation system. It is a measure of quality of travel time experienced by transportation system users and reflects the efficiency of the transportation system to serve citizens, businesses and visitors. Travel-time reliability (the variability in travel times on the same route at the same time from one day to the next) is critical to travelers, shippers, receivers and carriers for trip decisions and on-time arrivals at destinations. Thus, understanding travel time reliability is important to travelers in order for them to plan their trips effectively as well as shippers for them to plan and select routes appropriately. The first objective of this study is to formulate a methodology to obtain travel time data using Bluetooth technology on a freeway segment and collect travel time data. Bluetooth technology enables to collect real travel time data with a high sampling rate of up to 10% of the traffic flow. It also eliminates that need to use complex and often inaccurate algorithms use to calculate travel time from point speed data. Another benefit of this technology is it’s the fact that it is relatively inexpensive to implement; every station where travel time is desirable needs to be equipped with a processing unit, power source and Bluetooth dongle. The second objective is to observe daily and inter-daily variations as well as those due to poor weather conditions and estimate econometric models to predict travel time and variability. Within the context of this study, travel times were collected for two sections of freeway that experience heavy congestion during the peak hours. These travel times were then used to estimate three econometric models that predict travel time as a function of traffic flow parameters including speed and volume. The first model, which is linear regression with lagged dependent variable terms, aims to predict individual travel times during all times of day. The second model, a survival model, seeks to evaluate the probability of the trip lasting any specified length of time. In addition, it can predict the probability of exiting the freeway segment given that the vehicle has been traversing the segment up to that time. The third model seeks to describe travel time and the variability of travel time using the seemingly unrelated regression equations.