Influence of Vehicular Composition and Lane Discipline on Delays at Signalised Intersections under Heterogeneous Traffic Conditions R.V. Yogesh Kumar, A. Gowri and R. Sivanandan Centre of Excellence in Urban Transport Transportation Engineering Division Department of Civil Engineering IIT Madras, Chennai National Conference on Urban Mobility Challenges, Solutions and Prospects July 14, 2012
Introduction At intersections, mix-up of vehicles is high and ordered queue and lane disciplines seldom exist Complex vehicle interactions and manoeuvres occur Through and turning vehicles seek to occupy the same physical space Leads to blockage of through vehicles by turning vehicles and vice-versa Intersections become bottlenecks with increasing traffic demand lead to enormous delays Delay also depends on vehicular composition, availability of space and geometric conditions 2
Literature Review Only limited studies on modeling heterogeneous traffic through intersections (e.g.; Maini and Khan, 2000; Arasan and Kashani, 2003; Marwah et al., 2006; Chandra et al.,2009; Mathew and Radhakrishnan, 2010) Literature on influence of composition and lane discipline on fundamental relationships for midblock (Gowri et al., 2012). No specific studies for signalised intersection Study Objectives To study the influence of traffic composition on delays To assess the effect of lane discipline on delays 3
Methodology Identification of study parameters Identification of case study Data Collection Modifications to simulation model Simulation runs Analysis of results Conclusions and recommendations 4
Simulation Model Logics in simulation model (Gowri, 2011) for non-lane following scenario Vehicle Generation Vehicle Placement Vehicle Movement Vehicle Accumulation Vehicle Dissipation Object Oriented Programming concepts Implemented in C++ programming language (Gowri et al., 2009) 5
Data Collection Ashok Nagar Signalised Intersection, Chennai 6
Vehicle Composition and Signal Timings BUS. 0.50% LCV. 1.80% AUTO. 8.11% TRUCK. 0.50% CAR. 19.02% MTW. 70.07% Observed Signal Timings Phase Interval (s) Red 70 Green 64 Amber 4 Cycle time 138 7
Model Validation Parameter used for validation No. of vehicles dissipated per cycle for one hour Dissipation of vehicles per cycle for one hour is calculated from field and simulation model Simulated values are not statistically different from observed values, indicating the validity of the developed model 8
No. of Vehicles Dissipated from Stop Line Green Phase Simulated Vehicles (PCU/Cycle) Observed Vehicles (PCU/Cycle) 1 93.0 76.2 2 63.4 88.1 3 81.5 85.7 4 75.1 72.4 5 80.1 69.9 6 62.6 77.1 7 72.1 85.9 8 70.2 86.8 9 81.3 89.3 10 76.9 73.3 11 78.3 77.0 12 81.2 71.4 9
Modifications to Existing Model - Lane Following Placement logic Vehicles are loaded on appropriate lane based on relative composition of the vehicles Movement logic Vehicles are made to follow lane discipline Imperfect lane movements (vehicle travel between two lanes) are not allowed 10
Influence of Composition on Delay Two-wheeler composition varied from 20% to 80% Effect of composition analyzed for seven different levels Vehicle Composition (%) Composition No. Two wheeler Car Bus Truck LCV Auto- Rickshaw 1 80 7 0.66 0.66 0.66 11 2 70 17 0.66 0.66 0.66 11 3 60 27 0.66 0.66 0.66 11 4 50 37 0.66 0.66 0.66 11 5 40 47 0.66 0.66 0.66 11 6 30 57 0.66 0.66 0.66 11 7 20 67 0.66 0.66 0.66 11 Compositions used for Simulation 11
Control Delay for Non-lane Following Average Control Delay per Vehicle (s/veh) Composition No. 500 veh/h 1000 veh/h 1500 veh/h 2000 veh/h 2500 veh/h 2700 veh/h 1 (TW dominant) 28.8 32.1 33.6 34.7 35.5 35 2 30.1 32 33.1 35.1 35.5 35.6 3 29.7 32.5 34.6 35.5 37.4 37.5 4 30.1 32.9 34.8 35.8 37.8 38.2 5 30.5 33.6 35.2 36.5 38.7 40.1 6 31.3 33.1 35.4 36.2 39.3 41 7 (Car dominant) 31.1 33.8 35.2 36.3 39.5 42 12
Control Delay (s/veh) Delay vs. Vehicle Composition - Non-lane Following For higher volumes & greater car composition, delay to vehicles are higher 45 Lower dissipation rate of cars & restriction in filtering of two wheelers 40 2700 veh/h NLF 2500 veh/h NLF 35 2000 veh/h NLF 1500 veh/h NLF 30 1000 veh/h NLF 500 veh/h NLF 25 1 2 3 4 5 6 7 Vehicle Composition 13
Control Delay for Lane Following Case Average Control Delay per Vehicle (s/veh) Composition No. 500 veh/h LF 1000 veh/h LF 1500 veh/h LF 1 41 41.2 43.9 2 42.03 41.9 44.7 3 40.6 42.2 45.6 4 41.1 41.5 50.2 14
Control Delay (s/veh) Delay vs. Vehicle Composition - Lane Following Composition has little effect on lower volumes 55 50 For higher volumes & greater car composition, delays are higher 1500 veh/h LF 45 40 1000 veh/h LF 35 30 500 veh/h LF 25 1 2 3 4 Vehicle Composition 15
Control Delay (s/veh) Comparison of Control Delay for Lane Following LF) and Non-lane Following (NLF) 55 50 Overall delay is higher for LF Delays are higher by 32-44% for LF 45 1500 veh/h LF 40 1500 veh/h NLF 35 30 25 1 2 3 4 Vehicle Composition Delay vs. Vehicle Composition for 1500 veh/h 16
Control Delay (s/veh) 55 Overall delay is higher for LF Delays are higher by 26-31% for LF 50 45 1000 veh/h LF 40 35 1000 veh/h NLF 30 25 1 2 3 4 Vehicle Composition Delay vs. Vehicle composition for 1000 veh/h 17
Control Delay (s/veh) 55 Overall delay is higher for LF Delays are higher by 36-40% for LF 50 45 500 veh/h LF 40 35 500 veh/h NLF 30 25 1 2 3 4 Vehicle Composition Delay vs. Vehicle composition for 500 veh/h 18
Conclusions For lower volumes (500-1000 veh/h), effect of composition on delays to vehicles are generally not significant for both lane following and non-lane following cases For higher volume level (1500 veh/h), with increase in car composition, delay increases for both lane following and non-lane following cases. For all volume levels, delays to vehicles are higher for lane following cases compared to non-lane following cases 19
Conclusions (contd ) Non-lane following case results in lesser average delays Optimal utilization of available space by various types of vehicles Seepage of smaller vehicles to the front of queue The above insights lead to better traffic management and control strategies at signalised intersections under heterogeneous traffic conditions More case studies and further scenario analysis need to be conducted to generalize the results 20
Acknowledgement The work reported in this paper was made possible partly through a project supported by funds from Ministry of Urban Development (MoUD), GoI, through their sponsorship of Centre of Excellence in Urban Transport at IIT Madras 21
Selected References Arasan, V. T. and S. H. Kashani (2003) Modeling platoon dispersal pattern of heterogeneous road traffic. Transportation Research Board 82 nd Annual Meeting, Washington, D. C., USA. Chandra, S., A. Agrawal and A. Rajamma (2009) Microscopic analysis of service delay at uncontrolled intersections in mixed traffic conditions. Journal of Transportation Engineering, ASCE, 135, 323-329 Gowri, A., K. Venkatesan and R. Sivanandan (2009) Object-oriented methodology for intersection simulation model under heterogeneous traffic conditions. Advances in Engineering Software, 40, 1000-1010. Gowri, A. (2011) Evaluation of turn lanes at signalized intersection in heterogeneous traffic using microscopic simulation model. Ph.D Thesis, IIT Madras. Gowri, A., K. Venkatesan, Karthik K. Srinivasan and R. Sivanandan (2012) Mixed traffic characteristics on urban arterials with significant motorized two-wheeler volumes: role of composition, intra-class variability, and lack of lane discipline. Transportation Research Record, Journal of Transportation Research Board, Washington D.C., USA (Accepted for Publication) Maini, P. and S. Khan (2000) Discharge characteristics of heterogeneous traffic at signalised Intersections. Transportation Research Circular E-C018: 4th International Symposium on Highway Capacity, Maui, Hawaii, 258-270. Marwah, B. R., Raman Parti, and P. K. C. Dev Reddy (2006) Modeling for simulation of heterogeneous traffic at a signalised intersection. Highway Research Bulletin, Indian Road Congress, 74, 81-89. Mathew, T. V. and P. Radhakrishnan (2010) Calibration of microsimulation models for non-lane based heterogeneous traffic at signalized intersections. Journal of Urban Planning and Development, ASCE, 36, 59-66. 22
t-test Values for Control Delays of Different Modes Mode Mean Control Delay (s) t 0 t critical (Two-tail Observed Simulated test) Statistical Inference Two-wheeler 46.5 51.8-2.1652 ±2.5803 Means Equal Car 44.1 49.5-2.5494 ±2.5959 Means Equal Auto-rickshaw 44.2 50.3-2.0947 ±2.6079 Means Equal HV 45.4 51.9-1.2393 ±2.7450 Means Equal 23