DISAGGREGATED TRANSPORTATION DEMAND MODEL FOR HOME BASED TRIP GENERATIONS

Abstract

Comprehesive traffic and transportation plans are now being prepared for several metropolitan cities in India, in recognition of the continually increasing demand for better transportation facilities. The demand for travel is basically estimated through the traditional urban transportation planning system utilising the aggregated information on socio-economic, location and trip-making characteristics of geographical zones determined in an arbitrary manner. The zonal methods, as prac ticed and further recommended for predicting the trip generation from home are found to have many important shortcomings. Models based on disaggregated or 'individual1 data inputs are observed to be capable of removing most of these drawbacks. The main outcome of this research is presented in two parts. In the first part, a modelling procedure has been formulated that is directly related to the individual trip making, but articulated to the traditional urban transportation system. In the second part a procedure has been evolved that is based on individual behaviour on travel decisions but treats the outcome in a probablistic manner. Both the procedxires relate to the Indian environment, particularly the requirements of home to work trip journeys. Comprehensive home interview surveys are conducted in Roorkee town to develop essential data in the formulation of - ii - hypothesis for the disaggregated traditional trip generation models. Most of the data available from metropolitan cities are found to be deficient in their utility for the develop ment of disaggregated behavioural models. An elaborate procedure has therefore been followed to collect essential data from the home intervie\tf surveys of Ahmedabad metropolitan city, the Ahmedabad Municipal Transport Service/ The municipality, and L.D. College of .Engineering, Ahmedabad. Finally the data are constructed for every individual commuter with regard to variables like his travel options and level of service. Other surveys included are the attitude survey, the longitudinal survey, the social status survey etc. All these data are processed and compiled through data processing system and stored on punched cards. A hypothesis is formulated to disaggregate the population into a number of homogeneous groups after a.careful study of the socio-economic conditions and the travel requirements of Indian commuters. MANOVA confirmed this disaggregation policy. An alternative method termed as Individual trip generation model has been proposed. This method fits Into traditional urban transportation planning package, but does not incorporate transportation system parameters. The problems associated with the estimation of these models are examined leading to the PACTOR ANALYSIS of the test battery on socio-economic variables. It is found that - iii - only a few factors could conveniently represent the full information contained by a large number of variables and considerable economy could be achieved by redesigning the Home Interview Survey formats in this direction. The Individual model hypothesis is tested with the full scale home interview survey data of Roorkee town. It is shown that the proposed method treating individual as the descriptive unit of travel, is superior to other conventional models and are capable of predicting the trips nearest to the replicating sample. Problems arising out of the trip rates for the essential trips are discussed and the findings are sxipported through attitudinal and longitudinal surveys. Implications of individual modelling on the predictive capability vis-a-vis other models are presented. A suggested method of application of this methodology within the traditional transportation planning package is also presented. The absence of any formal mode-split procedure, the introduction of new types of public transportation systems, and the emergence of fOil Crisis' have posed the need for studying the consequences of changes in transportation systems and policies. Sequential disaggregated behavioural models appear to be the most suitable for this purpose.Complexities in composition of traffic, and a wide range of socio-economic - iv - conditions have created problems in the effective development of such models for developing countries. A hypothesis is therefore proposed for modelling work-trip travel demand based on Indian experience. The basic approach recommended is to further stratify the groups of population proposed in the traditional 'individual' models, into persons having car, scooter, bicycle and walk modes, and to treat each person as having two choices-namely his own mode, and alternatively other transportation facility. Similarly, the choices of worktrip frequency are also defined. Subjective and Objective measures of system attributes have been argued out, recommending the objective estimates for regular commuting trip decisions. Mode choice probability.model, and trip frequency choice probability models are developed on eight different groups of people. Successful calibration was achieved by maximum likelihood techniques, and it is brought out that all the logit models so developed are satisfactory. Useful information has been derived from these models. Of particular importance, are the derivation of values of travel time to car, scooter, bicycle and walk mode people, and also the elasticities and micro-elasticities of mode split and trip frequency probabilities to a variety of policy decisions. One set of curves for mode choice probabilities and another set for trip frequency probabilities for a wide variation of level of transportation service are developed. - v - Problems associated with the introduction of new types of transportation system in an urban area are discussed and the usefulness of behavioral models is illustrated. Further a procedure is suggested to illustrate the utilisation of behavioural models in conjunction with the traditional models or as an independent system by itself. Both the methods, namely the traditional and behavioural disaggregated models,can be fully utilized if only the indepen dent causal variables involved in the model building arc capable of being predicted at the horizon year. Hence a procedure based on discriminant analysis for predicting vehicle ownerships and a detailed flow chart to obtain remaining variables through standard modelling procedures like Cohort survival models and Garin Lovrey models are provided. i