The transport model performance risks and causes of problems which we seek to address are as follows.
The linked-together transport model will have a worse performance than evident from the calibration of any individual module because:
- new updated planning data is available for the base year, different from that used in transport model estimation and survey expansion,
- the inputs to each transport sub-model are changed from observed to synthetic,
- the trip ends are estimated with synthetic family structure and car ownership model outputs,
- the distribution/mode choice model is operated with with synthetic trip ends and generalised costs fed back from synthetic assignment model outputs,
- assigned flows are based on fully-synthetic trip matrices.
Transport model tuning and validation is aimed at dealing with the error accumulation issues referred to above in connection with:
- changed planning data,
- combined effects of errors in synthetic car ownership, trip ends, distribution/mode choice and generalised costs derived from assignments of the synthetic matrix;
- checking/tuning the fit of the model to observed data;
- testing key transport model sensitivities.
- Analysis of the effects of accumulation of transport model errors
- Validation of the implemented transport model
- Sensitivity tests of the implemented transport model
- Tuning/correcting the transport model