Selecting the Best Traffic Scheme for the Bicutan Roundabout: A Microsimulation Approach with Multiple Driver-Agents

Understanding the dynamics of vehicular traffic is important to planning and improving shared public roadways, not only to effectively and efficiently facilitate the controlled movement of vehicles, but also to guarantee the safety of individuals in the vehicles. Because data from real-world traffic is difficult to measure, and conducting replicated experiments on humans has ethical and economical questions, selecting the best scheme out of the proposed vehicular traffic controls has been confined only through simulation. We present the result of our microsimulation study on the effects of six traffic schemes T={t0, t1, ..., t5} on the mean total delay time (Δ) and mean speed (Σ) of vehicles at the non-signalized Bicutan Roundabout (BR) in Upper Bicutan, Taguig City, Metro Manila, with t0 as the current traffic scheme being enforced and ti>0 as the proposed ones. We present first that our simulation approach, a hybridized multi-agent system (MAS) with the car-following and lane-changing models (CLM), can mimic the current observed traffic scenario C at a statistical significance of α=0.05. That is, the respective absolute differences of the Δ and Σ between C and t0 are not statistically different from zero. Next, using our MAS-CLM, we simulated all proposed ti>0 and compared their respective Δ and Σ. We found out using DMRT that the best traffic scheme is t3 (i.e., when we converted the bi-directional 4-lane PNR-PNCC road into a bi-directional 1-lane PNR-to-PNCC and 3lane PNCC-to-PNR routes during rush hours). Then, we experimented on converting BR into a signalized junction and re-implemented all t3 with controlled stops of S={15s, 45s}. We found out that t3 with a 15-s stop has the best performance. Finally, we simulated the effect of increased in vehicular volume V due to traffic improvement and we found out that t3 with 15-s stop still outperforms the others for all increased in V={10%, 50%, 100%}.