NCDOT has relied on trip estimates based on the ITE Trip Generation Handbook for years; however, the number of fueling positions at contemporary fueling centers typically exceeds the range presented by ITE. This project aimed to quantify the effects of contemporary fueling stations by looking into other previous literature and investigating various analysis methods, such as linear and multi-variable linear regression models, on actual trip count data.
The fueling center is known to be one of the most difficult land uses to quantify trip ends because many site characteristics affect whether or not a fueling center is chosen by a consumer. The research team conducted an extensive literature review and thorough email surveys prior to conducting data collection at thirty case sites. Most states use traditional methods based on the ITE Trip Generation Handbook for predicting trip demand. However, a couple of states employ multilinear regression models with limited success. The results from these models were either flawed or had low R2 values, and therefore were not useful.
The project team conducted a two-fold analysis of the data collected at each of the sites. The first analysis used traditional methods such as those employed by ITE. Linear and log-linear plots were generated for a number of various independent variables. The second analysis method looked at the potential for a multi-variable model. The team ran many different models including various combinations of independent variables. Based on our analysis of the data, the proposed recommendation is to use two multi-variable linear regression models for AM and PM peak trip estimations.
