Università di Bologna, DICAM-Trasporti SIDT Seminar Venice, October 6 th, 2011 Stefano Angelini Silvia Bertoni Antonio Danesi Marco Donzelli Federico Rupi
Research goals Verification of the relative importance of those variables affecting headway adherence in the case of two main routes of the Bologna transit network Headway adherence: adherence to the bus scheduled travel time, measured evaluating the variation of both the dwell time at bus stops and the running time delay between two consecutive stops Study sample: over 300 bus rides operated by ATC along two of the main bus lines (No. 19 & 27) of the Bologna transit network
Problems created by loosening of headway adherence User side Service quality reduction due to the increase of waiting time (Welding formula) E(w)=[H*(1+Cv^2)]/2 NOTE: if the Cv increases, E(w) increases too Operator side Reduction of veh/km delivered Inefficient use of the vehicular fleet (bunching phenomenon) Operating cost increase Extremely crowded vehicles, and subsequent comfort reduction
Analysis of headway adherence Determinants Demand characteristics (passengers type, bus occupancy factor, etc..) Traffic characteristics (congestion level, flow composition) Route characteristics (No. of lanes, reserved lanes, No. of intersections) Service functioning characteristics (drivers behavior and experience) Vehicles characteristics (loading capacity, No. of doors) Variables Dwell time: time spent at bus stop allowing users getting on and off the bus Running time: travel time between two subsequent stops (it includes the clearance time) Transfer time between two consecutive bus stops
Case study: data collection Collected data Line s number (No. 19 & 27; high frequency bus lines) Vehicle number Arrival time at the bus stop (door opening) Departure time from the bus stop (last door closed) No. of passenger getting on the bus Dwell time Calculated data Running time between two subsequent bus stops Time interval between bus stops i+1 and i (time difference between the arrival at i+1 and i for the same bus n) Effective headway at the generic stop i Period of analysis: May, morning peak hour
Case study: investigated routes Bus lines 19 & 27 (1,600 and 2,200 pass/h served during peak hours, respectively), having a scheduled headway of 4 and 3 30 (14 & 17 rides/h, respectively) Measurement period: 8.00 9.00 (morning peak hour) 5 bus stops considered (total length: 1,200 m)
Dwell time [sec.] Distribution of dwell time High dispersion around the mean value (about 11 s): the variation coefficient is 0.46 Fermi bus stop Bus ride number
dwell time [sec.] Dwell time vs. No. of passengers loaded No. of passenger loaded
Other empirical evidences An empirical relation that links the number of passenger getting on the bus with the stop time (t-test=1.96) was found Accordingly to previous studies, there is a correlation between the no. of passenger getting on the bus and the stop time; anyway, other factors contribute to the variation of the stop time, such as: passenger type (kids, old men/women) stop spacing vehicle crowding and vehicle type driver reaction time on board circulation fare etc.. The results are similar to the ones obtained from a study performed on bus lines 32 and 33
Evaluation of Level of Service at bus stops Accordingly to Transit Capacity Manual, the LOS of the investigated bus stops has been evaluated by computing the coefficient of variation (Cvnj) of the time difference between effective and scheduled headway In the best cases, a LOS equal to E has been observed.
Running times [sec.] Analysis of running times Measured running times are extremely dispersed around the mean value Fermi Laura Bassi Bus ride number
Determinants of running times This dispersion is mainly due to the following factors: No. of intersections along the way traffic conditions & composition No. of signalized intersections presence of parking at street s side obstacles and pedestrians crossing the street roadway narrowing driving style weather etc
Transfer time (1) Both dwell time and running time affect the headway There is a higher dispersion of the total stop times (Cv=0.36) compared to total running times (Cv=0.18) The average distance between two stops is rather short (ca. 287 m), but running time represents the main part of transfer time between two consecutive bus stops Therefore, the running time mainly affects the headway adherence for the investigated rides
Transfer time (2) Variable Mean value [sec.] Standard deviation [sec.] Variation coefficient Dwell time 35,8 12,9 0,36 Running time 201,6 35,9 0,18 Mean value, standard deviation, and coefficient of variation of the total dwell times and the total running times for the investigated route
Investigated rides Transfer time (3) In most measurements (93%), the running time contributes more than 70% to the headway. Running time/transfer time ratio s percentage ranges
Headway variation [sec.] Transfer time (4) For the investigated rides, the relationship between the stop time and the headway shows that the variation of the stop times is statistically insignificant Dwell time variation [sec.]
Headway variation [sec.] Transfer time (5) For the investigated rides, it appears that the running time variation strongly affects the headway, thus increasing the bunching phenomenon to appear. Running time variation [sec.]
Conclusions Headway adherence is affected by several factors; consequently, more than one index should be adopted to evaluate the headway adherence (mean waiting time, variation coefficient of the scheduled headway) The present study found that although the stop time is a variable affecting the headway adherence, its contribution is negligible if compared to the one of the running time for the considered rides