CHOICE OF URBAN TRANSPORTATION MODES FOR WORK TRIPS

Abstract

In the urban transportation planning process, one of the basic requirements is the prediction of trips made by various modes of travel. In Western countries, many mathema tical models have been developed to accomplish the estimation of mode choice, but these models cannot directly be used for Indian cities because the travel habits of Indian urban population are quite different from those of the advanced countries. In India, the urban transportation issue does not involve choice between transit and auto driving, rather it deals with balancing of choices among variety of modes that are available, namely, public modes like bus, mini-bus and train, intermediate public transport (IPT) like three-wheeler, taxi, rickshaw, tonga (horse drawn carriage) etc., and the private modes like cycle, scooter, motor-cycle, car and walk. Large variations in socio-economic conditions of urban commuters and use of variety of viable modes make the choice problem more complex in Indian cities than in the advanoed countries. This necessitates a probe into the socio-economic and behavioural characteristics of the commuters influencing choice of travel modes. The present research study deals with the problem of identifying the various factors which influence the mode choice for home based work trips in typical mid-size city of the country so that optimal spatial distribution of transport modes can be worked out for urban centres based on the socio-economic settlements. ii In contrast to the procedures of the past two to three decades which dealt the problem on the basis of zonal aggre gation, this study has adopted the individual-choice approach, which is based on an understanding of the decision making process of the individual traveller. Since very little information was available about commuters behavioural atti tudes towards selection of travel modes, an exploratory survey was conducted. Patna city which is capital of the state of Bihar was selected for data collection. Various modes avail able in Patna are typical of those which are available in other mid-size cities of the country. Quantitative information on individual's choice and attitude towards different modes based on the evaluation of mode characteristics was achieved through a structured questionnaire. Personal interview method was adopted for data collection from the subjects at their work places. As many as 14 work places were selected for the survey. Questionnaire was designed to seek information on (i) number of factors related to socio-economic conditions of the respondent (ii) characteristics of transportation modes in use (iii) respondent's opinion, attitude and evaluation of various modes and (iv) distance and access time of respondents. Choice-based sampling procedure was adopted to have adequate sample size of the users of all modes. In all, 418 commuters were interviewed. iii The 'universe' constituted only of those persons who actually made journey to work by the usual transportation modes in the study area. Persons commuting within Patna and also the persons regularly commuting to Patna for work from outside using the above modes constituted the universe. Cross-classification technique has been adopted for analysis of survey data. This technique has provided an insight into the factors that affect travel demand of the commuters. The sample collected has been stratified into various income groups ranging from Rs.400 and below to Rs.2500 and above. Since each mode was found to have a characteristic lead distance upto which it was preferred the most, mode choice data for each income group was further disaggregated into various trip lengths, ranging from less than 1 km to as long as 25 km. In the survey, information related to household and mode characteristics and use of alternative modes in substi tution of primary mode and the satisfaction derived from them were also collected. Survey data has been analysed systemati cally t# quantify the usage of various modes. Income as well as distance are found to be the two most significant parame ters affecting the choice of a particular mode for work trips. Trip matrix of various modes disaggregated with income has been prepared which can be directly used for establishing the travel demand of the study area in terms of various modes. Biases in the estimates were removed by accounting for actual iv proportions of people in different socio-economic groups in the universe. Analysis of probe questions asked in the interview to identify mode captive respondents indicates that the commu ters did not have much knowledge about the characteristics• of the alternative modes available to them. If users of a particular mode had to select an alternative mode,they showed greater tendency to select the alternative modes from within the same category as of the primary mode or from the next akin category. Commuters attach certain value to their travel time and tend to trade off time with cost.while choosing a parti cular mode. Determination of value of time is useful to find out disutility of a Particular mode. Beesley's method has been used for computing average value of time of commuters using primary modes. These values have been determined with respect -to walk as an alternative mode of travel. Mann-Whitney test has been applied to examine the influence of income on type of mode used. The study has indicated strong dependency of type of mode used on the value of time at 5 per cent signi ficance level. For example, it is found to vary from 4 per cent of income for train users to 117 per cent for rickshaw users. Having identified the factors which influence the choice of mode, an attempt has been made to formulate a mode-choice model to predict the market share value of each mode based on the evaluation of disutility of the mode in terms of generalized cost. The lat-tfer was expressed as a linear function of cost incurred upon travel time and out of pocket cost. In order to predict whether or not a particular alternative mode is chosen, the values of utilities have been transformed into probability values with the help of Gumbel distribution function. The latter facilitates the analysis of the multi-mode choice that may be available to the commu ters of various stratified groups. Another technique to fur ther simplify the modelling process is to represent the multi modal choice situation by an equivalent binary hierarchy (EBH) The mode choice model in each approach was calibrated by maxi mum likelihood technique. This technique is quite complex and difficult to perform, especially when the number of explana tory variables are large and also when the number of alter native modes increases. To simplify the estimation process, only one explanatory variable i.e. generalized cost has been considered in this study. Mode choice model has been calibrated using the survey data of Patna city. Computer programmes for MNL and EBH models have been developed for use on DEC 2050 Computer of Regional Computer Centre of the University of Roorkee. On an average about 10 iterations were needed in the case of MNL model while about 5 iterations were required in EBH model to achieve vi the desired level of convergence. The program also yields results of statistical tests in terms of t-scores and likeli hood ratios. In both the approaches, the model calibration is found to be significant at 95 per cent confidence level. However, the null hypothesis that the generalized cost coeffi cient is not significantly different from zero could not be rejected in either approach, which tends to indicate that within the specified income and trip length group, the influence of generalized cost on mode choice is not significant. This situation may be the result of modal captivity which may have been left undetected during the survey analysis. Thus in each case the model calibration has given values of mode specific constants which are used in the formulation of expo nents of e to reproduce the observed usage of modes. Unlike MNL model, the EBH model, does not involve inversion of large matrices for its calibration, which can be easily carried out on programmable calculators, thus avoiding the need of high speed digital computers. EBH model has, therefore, a consider able potential and its usefulness for prediction of mode choice of urban commuters should be fully explored. Future research should also concentrate on development of suitable procedure for identification and prediction of modal captivity of urban commuters.