Project appraisal - Transport Policy and Planning in Great Britain - page 348

selected (often most critical) locations. Except for narrowly focused schemes it will
be impractical to generate comprehensive travel information for the base year. Instead
sample data will be used to 'calibrate' (i.e. adjust on the basis of local evidence) the
output of mathematical models built using relationships identified in previous studies.
These 'land use-transport models' produce estimates of the volume and patterns of trips
given local data on the pattern of land use, the characteristics of transport networks
and composition of the population. On the assumption that these relationships hold
good through time, the calibrated model is then used in conjunction with forecast
planning data to estimate travel patterns in future years. (A full introduction to land
use-transport modelling is given in TAG unit 3.1.)
Travel forecasts are produced initially for a hypothetical future situation known
as the 'reference case'. This incorporates the effect of exogenous factors, i.e. local
changes in land use and the national projections of GDP etc. within TEMPRO, but
does not reflect conditions on the local transport network. Implicitly the volume
and pattern of demand in the reference case assumes that travel speeds and costs
on all parts of the transport network remain the same as in the base year. In practice
changes in the exogenous factors alter the volume and distribution of trips as between
particular places, modes, routes and times of day. These changes in travel demand alter
conditions on the transport network relative to the base year. The changed pattern of
opportunities then alters the scale and pattern of demand compared with the reference
case, resulting in a new set of transport conditions and so on.
The Department now requires the use of 'variable demand models' to reflect this
interaction between demand and supply until an acceptable level of convergence is
reached with each option examined (TAG unit 2.9.2). In the simplest case the model
will be run to produce a volume and pattern of demand in the forecast year for the 'with
scheme' scenario, the impacts of which are then compared against a 'without scheme'
scenario. An illustration of variable demand forecasting is given in Figure 21.1. In this
example the effect of the scheme is to improve conditions for individual travellers
such that an overall increase in trips occurs relative to the without-scheme scenario.
However the fact that the total shown remains less than the reference case is not a
necessary outcome - in theory it would be possible to improve conditions to the extent
that all initially forecast demand was accommodated and additional travel induced
beyond this. (Historically this was the case with inter-urban road schemes developed
in the era of 'predict and provide', where variable demand modelling was not employed
and the 'reference case' forecasts were used as the basis of scheme design.)
The precise form of the mathematical model used will depend on the nature of
the intervention being examined. For computational reasons it is not practicable
simultaneously to explore interactions within all dimensions of a model if each
of these are represented in detail. Where options for individual schemes are being
examined the spatial features of demand and supply (i.e. the pattern of zoning and the
character of the transport network) have to be represented at a fine level of detail in
order that local travel movements can be depicted accurately. The corollary of this is
that other dimensions of travel demand (e.g. by person-type or trip purpose) will be
represented relatively coarsely and the wider context of transport supply (e.g. the level
of parking charges or the availability of public transport) taken as fixed. Conversely if
more strategic choices are being examined (e.g. different mixes of public and private
transport investment with different regimes of parking or road user charging) then
the spatial dimension can be represented more coarsely - possibly in diagrammatic
or 'cartoon' form only. The other dimensions can then be disaggregated to enable the