DEVELOPMENT OF INTEGRATED PUBLIC TRANSPORT SYSTEM FOR DELHI

Abstract

Delhi the capital city of India, has been facing a tremendous growth of travel demand due to increased urban population and economic growth. Patronage of the public transport system is only the solution to the current major problems like reduced Level of Service (LoS), reduced mobility and pollution. Commuters prefer to travel by metro as main mode in their multimodal journey as it provides safe, affordable, quicker, comfortable and environmentally sound means of mass transport. At the same time service area of metro is limited due to guided movement. It can be improved by making accessibility focused coordinated service with other modes of public transport or intermediate para-transit. Different road and rail based public transport modes are used by the commuters in Delhi but are not integrated in true sense. To achieve seamless travel through various transit modes where metro is main mode; optimized design of feeder routes from the metro station to various destinations with the associated frequency is very important. This study has been intended to develop an integrated transport system using public modes which is optimized in terms total cost consists of user and operator cost. In this research sequence of models has been evaluated to develop proposed methodology. The proposed solution has been divided into five categories: 1) Evaluation of transport system and identification of study area, 2) Data collection and analysis, 3) Data base preparation and travel demand modelling, 4) Model development for FBNOP using GA and 5) Application of model to the study area (Study application) for planning an effective integrated transit system (ITranS). The evaluation of existing road and rail based transport system has been done to get an idea about the problems related to that creating the need of integration and also about the reasons for inefficiency of current metro and feeder bus integrated service. After identifying study corridor as Line I of Delhi metro which was fully operational, study area was defined by demarcating zone and zone boundaries to demonstrate the effectiveness of the proposed methodology. To develop an optimized feeder network in the study area it is important to know flow pattern which can be OD trips from origin to destination. To evaluate OD trips of horizon years (2011 and 2021) travel demand models have been developed from the base year OD trips and forecasted for future years. Four stage travel demand models involve huge data. The required data were identified and collected by defining in two broad categories: Primary and Secondary data. Household survey data was the major source of demographic, economic and trip characteristics of commuters and collected for study area. SP survey was planned to collect data related to modal choice. Feeder route survey was carried out at metro stations having feeder services. The secondary data collected from Census of India to know population, from DTC to have operational characteristics of bus, from DMRC to get operational characteristics of metro and feeder service. The databases are prepared in GIS platform to have ease in handling, analysing and better understanding due to visualization capability. The network consists of links and nodes was developed along with the link characteristics. By providing dummy links such as connectors and transfer links, integrated network was developed to have multimodal journey which exists but without coordination. The generated data bases were used to develop travel demand models using GIS-T. First of all, four models: Trip generation, Trip distribution, Mode choice and Trip assignment were evaluated for the study area and validated with the observed data. Validated models were further used to forecast modal share and trips for horizon years. Data from the household survey with other planning variables were used to generate trip productions and attractions. Calibrated trip productions and attractions were further used as an input to develop trip distribution model along with the impedance (time or cost). Impedance function was evaluated based on thejourney time or cost required to travel between origin and destinations. Trip distribution model was validated using observed trip length distribution curve. At trip distribution stage productions and attractions were converted into trips between origin and destinations. Mode choice model was developed using SP survey data to evaluate percentage share of modes (public and private). The modal share of base year was compared with the existing (observed) share to validate the generated utility equations. The developedOD matrix at the stage of trip distribution was further divided into mode wise OD matrices according to percentage modal share using mode choice modelling. Mode wise OD matrices are further used to supply the flow onto the network. Trip assignment model loads the matrices onto the network. That means first three models evaluate demand in the study area and last model supplies the evaluated demand onto the road and rail based network. Trip assignment model was performed in two levels: transit assignment and traffic assignment. Interaction between the assignments has been done using preload to take care of already assigned public transport flow onto the network (here route system). Iterative process of this interaction continues until a convergence criterion is satisfied. The generated model is calibrated by comparing flow on road and metro network. Validated four stage models are used to forecast trips for horizon years. High capacity trunk system such as metro should be integrated with feeder system through appropriate hub-spoke arrangement that enable higher ridership on such trunk system. The development of an integrated transit system for a trunk line includes many sub-problems such as generation of optimized feeder route network, optimized scheduling between the modes of integration, vehicle movement on route etc. Model developed for optimized route and schedule has been presented here as a feeder bus network optimization problem (FBNOP). The FBNOP explains designing of optimized feeder bus and IPT network to provide access to an existing metro line. Here feeder bus also represents IPT that means where ever the word feeder bus is used it is also for the combination of feeder bus and IPT. The model for FBNOP developed to provide various service options namely; procedure for assignment algorithm (PAA), procedure for selection of destination nodes (PSDN), procedure for initial candidate route set (skeleton) generation (PICRSG), procedure for network analysis (PNA) and Simultaneous Route and Frequency based Heuristic SearchAlgorithm (SRFHSA). Basic data required for network optimization for solving this problem of routing and scheduling is origin-destination (OD) flow matrix. This can be achieved by many ways. PAA explains about the development of OD matrix. In this study it is obtained by converting zone to zone (zonal) OD matrix into passenger demand (between hub and nodes) on influencing nodes. Four different assignment alternatives: inverse of square, exponential, negative exponential and equal distribution are presented for this problem. The computational test on the algorithms has been carried out using data from a multi-modal transport network of Delhi to compare relative efficiency of these methods. Identification of destination nodes is also important for generation of feeder network. Five different concepts based on demand, distance, overlapping nodes, maximum route flow, tree node are explained and based on these destination nodes are decided in PSDN. The algorithmic approach in PICRSG consists of two algorithms: Dijkstra's (1959) shortest path and Sheir's (1979) label setting algorithm for k-shortest path which also incorporate knowledge of user and expertise of transit planner through defining maximum route constraint. The minimum route constraint was already applied at the stage of deciding destination nodes. This algorithmic procedure decides all candidate routes for the given transit network with fixed demand. As discussed, there can be number of feasible solutions (candidate routes) to FBNOP and it is equally important to measure the quality of each proposed solution. This can be possible using evaluation of the objective function which is in terms of mathematical formulation of user and operator cost. The PNA is an iterative process to analyze alternative network structures and to determine associated route frequencies for route network configuration generated by the PICRSG. It also evaluates various transit design parameters such as fleet size, vehicle size, and transit network system through various system performance measures that can reflect the quality of in service which is in terms of user and operator cost. The evaluation of performance measures and objective function is concurrent procedure and that why discussed with the SRFHSA. The SRFHSA is applied to address FBNOP by formulating model for feeder route generation and frequency assignment. FBNOP has been aimed to develop an optimized feeder network which minimizes combined cost of user and operator. Due to combinatorial nature of the problem solution search space is very large and not easily solved with the traditional methods. 1) Objective function: The widely used objective function is minimization of combined cost. 2) Demand: For simplicity most of the researchers pursue the optimal transit network with fixed demand. 3) Constraints: Feasibility constraints often include: minimum operating frequencies; a maximum and/or minimum load factor; a maximum allowable bus fleet size; maximum and minimum limits of route length; maximum number of routes; maximum route ridership volume and maximum allowable percentage unsatisfied demand. Out of all these mentioned constraints maximum fleet size; maximum and minimum load factor and maximum allowable percentage unsatisfied demand are considered in the analysis. 4) Decision Variables: Network route configuration and assigned service frequency are decision variables. 5) Passenger Behaviour: Transit trip assignment models can be divided into two groups namely: single path assignment and multiple path assignment models. Multiple-path trip assignmentmodel is developed to assign trips on network. 6) Solution Methodology: Most of the approaches rely on practical guidelines and guided by genetic algorithm based search procedures. SRFHSA is developed using C++ programming language and combined with Genetic Algorithm (GA) to find optimal solution. The developed five stage model is applied to real size network of part of Delhi near by the Metro Line 1 to develop ITranS and thus to test developed methodological framework and model. For feeder modes, bus as a "fixed-demand and fixed-schedule", and auto-rickshaw and taxi as "demand-based" service are considered. Optimized feeder network is generated by considering two different seating capacity of bus using GA. Different feeder networks are generated for different hours of a day by freezing feeder routes. Based on the optimized frequency, coordinated schedule is generated for feeder bus in synchronized manner with existing metro schedule. Evaluated optimized feeder routes are further considered and transit assignment has been evaluated by considering feeder as an additional mode. The assignment with and without integration scenarios has demonstrated potentiality of an ITranS as a considerable ridership increasing strategy for public transport modes. Increased ridership in case of integration scenario is due to shifting of passengers from road based vehicles, which reduced demand of vehicles on road. Reduction in demand of vehicles leads to reduction in fuel demand and also reduction in accident rate and total number of accidents. Developed ITranS has been compared with other road iv based transport system and ITranS has been analysed as an efficient system. Flexibility and successful application of the developed model on study area shows potentiality of its application on any typical Indian City. Study applications of the developed model to address FBNOP have proven the universality of methodology and the developed model both. Key Words: Integration, Public Transport, Feeder Routes, Schedules, Travel Demand Modelling, Geographic Information Systems, Optimization, Genetic Algorithm,