Initial matrix effects on estimation of origin destination matrix in transport planning

Transportation Demand Models (TDM) are traditionally constructed using the 4-step model which is trip generation, trip distribution, modal split and trip assignment. The step of trip distribution in transportation planning is explained by the Origin and Destination matrix (O-D) that shows the distribution of passengers or goods between zones of the planning area. O-D matrices are the essential input of any transportation demand model, subjected to modal split and trip assigned to the transportation network. As it can be understood the most important input of the 4-step model is O-D matrices, and the accuracy level in the predictions for the operation of the model increases depending on the precision level of the base year matrices during the calibration of the model. Base year O-D matrices; It is also obtained by using information from data sources such as household surveys, roadside surveys, vehicle tracking systems (AVL), mobile device passenger tracking (GPRS or station based). In addition to these methods, O-D matrices can be estimated from the vehicle or passenger counts on the transportation network. In this study, the effect of the first input matrix and the analysis tools used in the estimation method on the estimation of O-D matrices, which are accepted as the basic input for many analyzes while creating the Transportation Models, with roadside and traffic counts are evaluated. As a result of the studies, it has been obtained that the convergence of the first input matrix to the O-D matrix is of great importance and the L-BFGS-B method gives the best results in the analysis tools, followed by the CG, trust-constr, SLSQP and TNC methods.