Development of the Dutch National and Regional Models



Development of the Dutch National and Regional Models

Authors

R Smit, Ministry of Public Works and Water Management, NL; P Mijer, 4-Cast, NL; A Schoemakers, Oranjewoud, NL

Description

The paper will discuss the latest generation of strategic transport models in The Netherlands. It will focus on the realization of the base matrices, and will describe the process of developing the calibration procedure.

Abstract

Over the last 20 years the strategic transport models of the Ministry of Public Works and Watermanagement were built using a hierarchical approach with a National Model and Regional Models using the same modeling principles. Due to freedom in several modeling options there are more differences between the existing models than nowadays desirable. The paper describes the building of the new generation of the National and regional models using maximum consistency. It will start with the motivation why we use a system of consistent models at national and regional level in the Netherlands. Then it will shortly describe the main principles of the models (marginal approach using Base matrices and a Pivot procedure based on multi modal discrete choice models with TOD choice). The focus of the paper will be on the realization of the base matrices, how consistency between the models was achieved, which choices, and why, where made during the process of developing the base matrices.
In order to achieve consistency several handbooks were developed. General handbooks describing the model system as a whole, and specific handbooks on how to set up the zoning system and socio economic data and how to build the networks. Several national datasets on socio economic data, traffic counts, networks and the national travel survey were realized. Also a calibration handbook was written prior to the project in which the base matrices were built. The main innovations with regard to the base matrices were that car and freight are calibrated simultaneously and that the peak hour matrices are not defined by strict time frames (ie 7:00-9:00 and 16:00-18:00) but contain all the peak related traffic. The paper will describe this principle in more detail.
During the calibration process the information of the a-priori matrices (being the synthetic matrices from the discrete choice pivot models for car travel, and freight matrices from the so called Regional Freight model), the national travel survey and traffic counts have to be brought together and are used realize calibrated base matrices that, after assignment, meet all the preconditions defined by these observations. In that process much attention is paid to the structure of the matrices. Is the structure of the a-priori matrices maintained, are the 24-hour matrices still symmetric etcetera. During the building of the base matrices several options were further investigated that led to adaptation of the calibration handbook. The focus of the paper will be on this aspect: What was detected during the calibration process, how was it analyzed, why did we decide to change the procedure, and what was the final result?

Publisher

Association for European Transport