GB Freight Model - Upgrade report and user guide report

Print Print page   Download PDF PDF image

Executive summary

The Great Britain Freight Model (GBFM) is used in the Department as part of the National Transport Model (NTM) framework. The model is an integrated database and modelling package, which has been developed by MDS-Transmodal. In 2003, an audit of GBFM was carried out by independent modelling experts. The purpose of this project is to implement those of the audit recommendations which the Department sees as the highest priority, in time for inclusion in the next version of NTM. The new model is referred to as GBFM 5.0.

GBFM 5.0 attempts to address the most fundamental issues raised by the audit. The main goals were to create a more intuitive and more conventional structure with no duplication and more consistent definitions and modelling patterns throughout. A clearer user interface has been constructed and greater attention has been placed on ease of use in all areas, including identification of results, log files, calibration and the ability to change model parameters. The addition of a separate base matrix module, provides a greater range of forecasting options, e.g. relating land-use development to traffic generation.

GBFM 5.0 provides a software and data system with two principal components:

  • a matrix generator, based upon a model of the economy, predicting traffic flows between model zones; and
  • predicting path and mode choices.

Each of these components can be operated in two ways. The economic model can be used to build a base year matrix, or to use the relationships calculated for the base matrix to build forecast trip matrices. The transport model can be run to produce a set of results, or it can be run in calibration mode, to calculate weights that will improve the accuracy of the assignments.

The main improvements in GBFM5.0 are to combine the two domestic model stages in previous version into a single multimodal assignment with 2,650 zones, and to integrate the external matrix estimation software into GBFM so that it shares the same supply side inputs (networks and costs).

The software is set up so that the model can be run with an existing base year matrix. Once the model is calibrated for a given year (initially 2004), this matrix will be used frequently without modification until updated and recalibrated in succeeding years to maintain compatibility with latest available port, rail and road tonnage data.

The base matrix can then be re-estimated according to exogenous (i.e. non transport) variables such as growth in the underlying economic sectors of energy, construction and manufacturing. These processes will determine traffic distribution.

Mode split and road assignment are carried out simultaneously within a multimodal network, in which the road links are NTM compatible, so that estimates of HGV kilometres by standard road types and area types can be carried out.

The linkages between the land-use inputs and the O/D matrices are through the survey data collected within the work for the West Midlands Freight Study linking trip generation directly to land use in a framework where intermediate storage of goods (warehousing) is treated like any other stage of a production chain.

Three forecasting approaches are supported in GBFM5:

  • the user can use the base year data to carry out 'what-if' tests relating to a number of policy variables; or
  • trend-based forecasts can be generated; or
  • alternatively changes to the land use and employment inputs can be made.

A graphical user interface has been included for the first time. The emphasis has been placed on allowing the user to set up several model runs in advance and run the jobs as a batch, taking care of all the output file management.

Several model validation exercises were carried out. Some of the comparisons required for validation were against CSRGT sources and others were against road traffic statistics (ROTCES). The problem was that ROTCES survey results, generate significantly higher vehicle kilometres volumes than produced by CSRGT on which the majority of tonnages within GBFM are based. Overall, there was a good match between model outputs and observed flows at national level, but the validation was poor on strategic routes, partly due to the mismatch between ROTCES and CSRGT. Screenline validations varied between -22% to +23%, but the overall screenline validation was -10% which is reasonable given the inconsistency explained above.

In addition, the sensitivity tests that carried out in the audit of GBFM were repeated. The model outputs showed GBFM5.0 produced reasonable results. In addition, two model back-casting were carried out through changes in employment between 1996 to 2004 and through changes in land use between 2000 to 2004. The results showed GBFM5.0 produced better results when land use option was selected.

GBFM 5.0 Upgrade Report (2 Mb)

Report about implementing improvements that were identified in an earlier study of the audit of GBFM.

Published:
01 August 2008

GBFM 5.0 User Guide Report (2 Mb)

The GBFM 5.0 User Guide Report is about data inputs, the running and outputs from the model.

Published:
01 August 2008