Improvement Strategy Modeling For Transit Passengers' Travel Time Reliability

With the continuous development of shared transportation,intelligent transportation,and three-dimensional transportation,passengers are getting to concentrate more on service level and reliability of traditional transit transportation.However,the bus services in many large cities have problems of poor reliability due to factors such as traffic congestion,heavy weather,or traffic accidents.It has become a common phenomenon of uncertain waiting time,bus-buching,and congestions.In peak hours,the contradiction between the short-term gathering transit demand with high reliability requirements and the unreliable transit network services by various uncertain factors has become the main constraints for transit system development.The method to improve bus passenger travel time reliability is the core work of the dissertation.Based on quantitive analysis of travel time reliability,the optimization model is proposed.The methodology consists two levels of methods including exclusive bus lane(EBL),which can improve bus on-road riding environment,and demand transfer in hot area by flexible lines.The main work of the dissertation is as follows:(1)Quantification methods of different passenger travel time reliability are compared.Based on the requirment of optimization modeling,the travel time reliability is quantified by buffer time.Travel time reliability of different travel sections on transit network is expressed by the combination index of objective fluctuation and passenger reliability preference.(2)Based on the bus trajectory data collected by GPS,the estimation method of passenger's travel time reliability including in-vehicle time and waiting time is determined.By data pre-processing,over five hundred thousand pieces of passenger unit in-vehicle time data and bus arrival interval data at stops are calculated.Passenger in-vehicle time and waiting time are fitted and analyzed,respectively.For the in-vehicle time,compared with Gaussian fit,Weibull fit,and Laplace fit,Lognormal distribution is best fit,for the R-square is closest to 1 and the Root Mean Square Error(RMSE)is the smallest.It can be considered as the best to explain the distribution of the passenger unit in-vehicle time.The expectation and standard deviation of in-vehicle time are calculated.The results show that due to the influencing of EBL,for bus lines which are running on main trafiic corridors,the passenger travel time variability is the lowest in peak hour,which is only 48%of that for conventional bus lines.Then correlation relationship between waiting time and bus arrival intervals is determined to estimate passenger waiting time fluctuation Based on the actual interval data,the bus arrival interval distribution under different headways(1 min-10 min)is fitted.The results show that compared with Gaussian fit,Weibull fit,and Lognormal fit,Laplace distribution has the best fitting result.As the bus headway is increased from 1 minute to 10 minutes,the fluctuation of passenger waiting time rises from 1.74 minutes to 9.58 minutes.(3)The value of passenger travel time reliability is analyzed.The cost of in-vehicle and waiting time is depicted by travel arc.The additional waiting delay caused by the crowded boarding and alighting process is considered.The transit assignment model is established based on user equilibrium.The arc-based depth-first search algorithm and the linearization algorithm are developed,respectively.The multi-nonlinear problem is transformed into the linear programming problem and solved then.For the numerical example,the influence of travel time reliability on passenger travel choice and network demand distribution is quantitatively evaluated by passenger total reliability cost index CRn.The expectation of passenger in-vehicle time,waiting time,and corresponding volatility values are determined by real test data and statistical result,instead of traditional empirical rules or hypothetical distribution characteristics.(4)The basic theoretical framework for planning EBL to improve the on-road environment of buses,then enhance passenger reliability is constructed.According to the characteristics of different classification of bus lines,a bi-level optimization model for EBL considering travel time reliability is established.The trade-off among travel time cost of bus and car passengers,and construction cost is considered in upper model.The dynamic distribution of the passenger flow under the environment of EBL is analyzed in lower model.According to the statistical modeling and fitting result of the actual trajectory data of bus lines on EBL in Beijing,expectation and standard deviation of the passengers of are 1.59 min/km and 1.15 min/km,respectively.A two-step algorithm based on "feedback-loop" is designed considering the method of linearization.Finally,the numerical example is conducted to verify the feasibility of the model and algorithm.The travel time reliability will be improved significantly when EBL is considered by the cost index of CRn.Due to the phenomenon of passenger demand transferring,the optimization effect can be extended in different classification of lines.Although EBL is designed on arterial road to serve the arterial bus lines directly,the passenger travel time reliability on branch lines still can be improved indirectly.(5)The theoretical basis is analyzed to relax the OD demand in the hotspots area by new flexible bus,thus to reduce the load of transit network and improve passenger travel time reliability.A transit demand transfer optimization model for improving passenger travel time reliability is established.The transferable demand and the corresponding demand delivery route are applied as decision variables.Considering the nonlinear programming problem with a multiplication of integer decision variable and 0-1 decision variable in the model,a linear alternative algorithm is designed.By numerical study,the strategy of transferring demand has a dramatic effect on reduction of the total cost of passenger travel time reliability.(6)A case study is conducted in Zhongguancun Software Park and the surrounding local transit network in Beijing.The methods of EBL and demand transfer can reduce passengers total travel time reliability costs by 5.5%and 13.6%,respectively.The optimized percentage is 18.3%when the two strategies are applied together.According to the results,the two kinds of optimization methods have their own applicability.For small areas in the case study,travel demand is concentrated and passengers is highly sensitive to travel time reliability.The effect of demand-level optimization method by flexible bus is more effective.The EBL is proved to be more suitable for a wide-ranging transit network or commuting corridor.