Simplified travel demand modeling for developing cities : the case of Addis Ababa

Low income cities in developing countries lack road network planning tools. Their roadnetworks are mostly planned and implemented with pure speculation. Travel demandmodelling can be utilized as a strategic planning tool, if it is mainstreamed into the planningprocesses of these municipalities. However, application of the conventional modellingsystem as practiced in developed countries produced unsatisfactory results for developingcities. The focus is, then, on applying simplified modelling technique to developing cities.This thesis, hence, is motivated with the objective of testing the application of simplifiedtravel demand modelling for strategic road network planning purposes in low income cities.The study tested simplified modelling procedure on a selected developing city: Addis Ababa.The conventional travel demand model has four steps: trip generation, trip distribution,modal split, and traffic assignment. The conventional model processes large amount of dataand employs sophisticated software to do the modelling. Simplified modelling, on the otherhand, is about the use of limited data, application of simple model structure without blackbox effect, and getting reliable results.This study defined 35 traffic analysis zones for the city of Addis Ababa; and defined arterialroads network having 137 km length. Required data for the model were gathered from themunicipality. Mobility matrices for three income groups were developed to be refined inputsfor the model. The study then applied the four step process. Trip generation employedregression method by developing a relationship between trips and socioeconomiccharacteristics of zones. Trips were divided into three purpose groups: home-based work,home-based education, and home-based others. Non-home based trips were ignored as theirshare was insignificant. The trips were further stratified into three income groups: low,medium and high income. Trip distribution model employed the gravity model with animpedance matrix of inverse power function of distances between zones. The origindestinationmatrices were calibrated with average distance of the three income groups. Themodal split model employed a unique methodology of utilizing mobility matrices. Totalperson trips were split into trips by modes according to the modal share of trips in themobility matrices, which took into account trip distances. Traffic assignment model wascarried out with a computer program developed as part of this thesis. All-or-nothing andcapacity restraint traffic assignments were carried out within a spreadsheet environment. Allthe four steps were carried out without the use of licensed software.The simplified model replicated trip behavior of the city satisfactorily, considering the orderof magnitude of the unavoidable error margins in the input data for such model. Thecomparison of assigned average daily traffic and counted traffic showed a good fit; averageabsolute deviation being 10 %. With use of limited data and simple model structure, it isdemonstrated that simplified travel demand modelling can indeed be a valuable tool forstrategic road network planning for low income cities. From a number of tests made with themodel, it is concluded that the main focus to get reliable result is concentrated on gettingaccurate zonal trip productions and attractions