Modeling of the effects of changes in the sizes of traffic analysis zones in transport demand forecast model to travel assingment results

Traffic congestion continues to grow in mid-sized and large cities in both developed and developing countries with the growth of transportation demand day by day. Approach to urban transportation issue have passed through various stages from the very beginning. As a result of this, in many cities in the world, the rising of comfort and accessibility expectations has increased the use of private vehicles and caused reduction in the use of public transport. Increment in vehicle supply, in response to this, lack of road network has resulted in traffic congestion. Alternative scenarios have been generated with travel demand forecast models to find solutions for the traffic congestion. In those scenarios, future forecasts are carried out based on user, travel and zonal characteristics using land-use configurations and transportation infrastructures. Institutions dealing with transportation constantly develop new model infrastructures to improve decision-making process. In this study, demand forecast models for İstanbul have been examined according to purposes of travel. Demand forecast model applied for İstanbul and structures of traffic analysis zones have been examined. Selection criteria of the traffic analysis zones have been addressed and the effects of increasing the sensitivity of zones for travel start and end points on model forecast capability have also been examined by minimizing zones. The effects of dividing 540 traffic analysis zones in Istanbul into smaller areas on model consistency have been addressed. Demand forecast model infrastructure has been recreated for 1788 traffic analysis zones by separating zones from each other which show similar land use characteristics considering neighborhoods which are the smallest administrative boundaries. The three most important components of this infrastructure which are O-D matrices, highway and public transport networks, and traffic analysis zones' attributes have been revised for the 1788 traffic analysis zones. Regulations which have been carried out for the settlement of İstanbul into smaller zones have been discussed and the effects of these improvements on network assignment algorithms have been examined. As an example, general operational information for Kadıköy-Kartal Metro Line has been provided and the consistency of the outputs of the demand forecast model has been studied according to actual values for traffic analysis zones of different sizes along the Metro Line. The benefits of the demand forecast model has also been analyzed. When the public transport network assignments for the Kadıköy-Kartal Metro Line for the 540 and 1788 traffic analysis zones have been compared, it has been observed that both models have been quite representative; however there has been significant improvements in model accuracy when smaller zones are created. Station-based rides have been evaluated by the mean square root error (RMSE) method and comparisons have been performed. The error rate for the 1788 zones has been reduced by 23%. Population density and employment density have been evaluated as independent variables for the determination of proper radius for working area and exponential model has been applied. Attention should be paid to the characteristics of the working area while choosing the most suitable model to use during calculating radius of zones. It has been observed that it is possible to achieve %26 increment in R2 value by converging zone radiuses calculated using population density and employment intensity with the average radiuses used in the model within 1 km2 area.