Appendix E Traffic data
E.4 Traffic growth rates
Establishing the traffic growth rate for an HDM analysis is a very important task. This is because it will not only influence the total RUE, but also will influence capacity
improvements and other intervention alternatives.
Traffic grows over time, although the growth can be quite sporadic. Figure E.2 shows the results from traffic counts taken on a road in India over a 10-year period. The data are presented from April and October, which is before and after the monsoon period. It will be noted that the post-monsoon flows are always lower than the pre-monsoon flows. This is due to the major reduction in agricultural activities after the monsoon. While there is a continual growth, the October data from 1992 and 1994 is out of context with the rest of the historical data. This could be due to some unusual event at the site, for example a temporary road
closure or a local catastrophe. If such drops are not recorded elsewhere, they should probably
T raffic Cou nt Data from Surat-Dhu lia Ro ad, Gu jarat, In dia
Figure E.2 Example of variation in traffic over time
The two usual ways of expressing growth are as a geometric or as an arithmetic growth rate.
These are calculated as:
! Geometric
AADTyear base …(E.5)
where:
AADTyear is the traffic volume in the analysis year AADTbase is the traffic volume in year 1
YEAR is the year of analysis GROWTH is the traffic growth rate (%)
HDM uses a geometric growth rate.
The traffic growth rate is usually calculated using historical traffic trends, such as the data illustrated in Figure E.2, economic trends, such as growth in GDP, vehicle ownership trends, or a combination.
Using historical traffic trends is the least accurate method for predicting growth rates. This is because many developing countries are experiencing rapid motorisation and economic growth which will alter the historical trends.
Calculating growth from economic trends is often a more sound method than historical data since there are strong correlations between traffic volume and economic activity, something that a straight historical model will not capture. The underlying approach is to collect data on traffic growth rates and then to do a statistical analysis that fits a model to the data. The variables often used are income, population, fuel use, industrial production, agricultural production, or any other measure of the economy. These lead to elasticities which express the change in traffic as a function of the economic measures. Typical equations would be:
EPOP dPOP EGNP
dGNP
GROWTH= + …(E.6)
EFUEL dGNP
GROWTH= …(E.7)
where:
dGNP is the forecast change in the GNP (%) dPOP is the forecast change in population (%) EGNP is the GNP elasticity
EPOP is the population elasticity EFUEL is the fuel use elasticity
The use of fuel is preferred by some since it gives a good overall view of the historical growth in traffic. By comparing the changes in fuel use with changes in GDP one calculates the effect of GDP changes on fuel use. Coupling this with the forecast GDP gives the total forecast traffic growth. This method has advantages in that it implicitly considers the population growth.
As an example of the elasticities, in India CES (1991) gave values for the GNP elasticity of:
! 1.75 and 1.0 for freight vehicles
! 1.0 and 0.5 for passenger vehicles
in the periods 1994-2005 and after 2005 respectively. Values of 1.43 and 2.71 were used for the population elasticity for freight and passengers respectively. In Nepal (NDLI, 1993) adopted a value of 1.5 for the elasticity of fuel. The use of declining elasticities with time is common as this reduces the impacts of future uncertainties on the predictions.
One important feature of traffic growth is that it is different depending upon the level of a country’s development. Developed countries, with high levels of vehicle densities have markedly different car ownership growth rates than developing countries that are building up their densities. The model that best reflects this is a sigmoidal (‘S’ shaped) model. This is discussed in Button and Ngoe (1991) who present a generalised model with data covering a number of developing countries to use with the model.
WARNING
When forecasting traffic, always consider whether or not the predictions are reasonable. It is easy to adopt what seems to be a relatively low growth rate, such as 5 per cent, but this may mean that after a period of time your facility is at capacity. For this reason it is often prudent to adopt several different growth rates based on short, medium, and long-term considerations, with the rates declining in future years.
There are four principal categories of future traffic which are forecast:
! Normal traffic
The future traffic that can be expected assuming the current trends (for example, historical patterns) remains steady.
! Diverted traffic
The traffic that can be expected to divert to the road because of the improvement.
! Generated traffic
Traffic that would not have existed but is expected because of reduced travel times or diversion from other modes.
! Induced traffic
Traffic expected because of the new development created by better access (for example, building a road into a new area will open it up for development).