A modified model for signal intersection capacity considering road geometric conditions

When throughput demand exceeds intersection capacity in urban road networks, queue overflow causes congestion. This paper investigates limitations of the Webster algorithm in initial signal timing calculations and proposes an improved capacity estimation model for signalised intersections, considering traffic flow characteristics, lane number, length and road geometry. The model development involves establishing a computational framework based on lane number and flow at each entrance, correcting and analysing lane number and traffic flow reduction coefficients. Road geometry and actual traffic flow effects are studied, leading to a modified model. Using the Jinshan Avenue in Fuzhou, China as an example, the highway capacity manual (HCM) method verifies the model. Results show: (a) no significant difference between modified model inflow and actual observed flow; (b) the model exhibits an average error of 8.44 veh/h for calibrated flow during off-peak periods and 13.67 veh/h during peak hours. In both scenarios, it better matches real-world traffic conditions than HCM and Synchro; (c) in the case study, a single-factor sensitivity analysis was conducted for parameter α. Results show that this parameter exerts a notable influence on the model.