An enhanced model for calculating delay as a function of offset

Abstract

This thesis presents the enhancements made to an existing model for calculating delay as a function of offset between the traffic signals of a link. The delay-difference-of offset technique is a signal control concept used for strategic optimization of offsets in a fixed time signal timing plan for an arterial or a closed network. A delay estimation model was developed using this delay-difference-of-offset technique for theoretically estimating delay incurred by traffic because of queuing at the traffic signal on the downstream end of the link. This model is applicable under certain assumptions. One significant limitation was that it worked only for two-phase operation. Today, most arterial signal systems operate under multi-phase operation. So, under this work, this delay model was expanded for multi-phase operation. First, the model was expanded for three phase operation to work for diamond interchanges. The results were validated against PASSER 111. PASSER III is a software package used to analyze diamond interchange signalization. The expanded model was named DDOFF. It was found that with certain enhancements, DDOFF was replicating PASSER IR results. The delay results from DDOFF were almost the same as PASSER Ill. Next, DDOFF was expanded for four phase operation or for a generic link. The results were validated against NETSIM simulations. NETSIM is a computer simulation program which reflects field conditions closely. Also, an arbitrary link was established to make NETSIM runs. It was found that DDOFF gave delay estimates significantly similar to NETSIM. DDOFF does not account for blockage or spill back delays experienced in real time for high volume traffic. So, the difference in NETSIM and DDOFF delays were larger for short links and high VIC cases. It was concluded that DDOFF performs well given its limitations being a deterministic model. Also, it was recommended that DDOFF be enhanced further to account for spillbacks and dispersion in long links. Finally, DDOFF could be expanded to work for oversaturated cases making it a more powerful tool.