• About this site|
• Site index|
• Contact us|
• Jobs

Monitoring and Data Management Protocol: Environmental Indicators for Transport

Acknowledgements

1 Introduction

2 Why Monitor?

3 What to Monitor

4 Vehicle Fleet Composition Indicator

5 Travel Activity Indicator

6 Mode of Journey to Work Indicator

7 Travel Time Indicator

8 Glossary

Appendix A: Sampling Framework

Appendix B: Estimated Precision In Estimating Average AADT

Appendix C: Identifying the Road Use Category (or Traffic Pattern Control Group)

Appendix D: Week Multipliers and Relative Error for Each Road Use Category

References

7 Travel Time Indicator

The environmental pressure of transport is related to the volume of travel, the type of vehicles undertaking that travel and the conditions under which the travel is undertaken. Travelling alone at a given speed on an uninterrupted highway a vehicle will use less fuel, generate less emissions and detritus (from brakes and tyres etc), than a vehicle that has to slow for intersections and curves. As traffic volumes increase and traffic conditions become more and more congested drivers find it more difficult to maintain a steady speed and accelerate and decelerate more. Such actions further increase fuel use, emissions and detritus which all impact on our environment.

An indicator that provides a measure of the actual travel conditions compared to the ideal travel conditions provides analysts with a means of assessing the additional environmental impacts associated with traffic congestion.

7.1 How to monitor

The proposed travel time indicator is based on the congestion indicator (CGI), a standard measure that has been adopted by Austroads (the Association of Australian and New Zealand Road Transport and Traffic Authorities) of which Transit New Zealand is a member. Given that much of the literature associated with this measure refers to CGI, this term is retained to facilitate referencing where necessary.

The travel time indicator (or CGI) is based on measuring the actual travel time (ATT) on a route and comparing this with the nominal travel time (NTT) which is calculated on the assumption that vehicles travel at the posted speed limit without any slowing or stopping. The resulting measure is an indicator of the delay due to congestion reported in minutes per kilometre.

The measurements of actual travel time are undertaken periodically using 'floating car' surveys (described in section 7.5.2) on a sample of the roads which form main road network in New Zealand's main metropolitan centres (Auckland, Wellington, and Christchurch). It is this network where the bulk of urban vehicle travel is undertaken and congestion occurs on a regular basis. The traffic conditions on this sample network are monitored at various times of the day, for five week days, twice a year.

The resulting information gives the average level of congestion on the network.

Note: It is not recommended that this indicator be used to make comparisons between regions due to the difficulty of ensuring consistent collection processes.

7.2 What to monitor

Individual territorial local authorities together with Transit New Zealand have a responsibility for managing the roading network in New Zealand's main cities. Although it is possible for each of these organisations to undertake separate surveys, it is recommended that this indicator be measured and reported on a regional basis and a collaborative approach to data collection adopted Under this system the interested parties would agree the scope and cost of the surveys and contribute to a single regional programme.

Within each region the first step is to define the boundaries of the road network from which the sample of road, the monitoring network, will be drawn. These boundaries occur where the traffic volumes and environment becomes predominantly rural. However, it must take into account travel from satellite settlements that currently or will in future generate significant volumes of traffic.

It is vitally important that when defining these boundaries account is taken of future traffic demands. If the boundaries are constrained to only those areas where congestion is currently experienced, it is likely that in future years the monitoring network may need to be increased. If this is the case the ability to monitor changes over time in the extent of congestion will be compromised.

Figure 7 provides a simple example to illustrate this. In Figure 7 the mean travel speed is plotted for two sections of highway. At first glance it would seem that congestion has not increased on SH2 over the decade 1986 to 1996 and that conditions have improved on the Auckland southern motorway between 2000 and 2001.

Unfortunately neither is true. The start point for the SH2 surveys now lies within the congested area. The indicator registers no change in congestion even though queues now extend a considerable distance back up the road each morning. One way of overcoming this would be to expand the boundary of the survey. However, we can see the effect of such an action by looking at the data for SH 1 Auckland southern motorway.

For this section of highway there appears to be a reduction in congestion between 2000 and 2001. Unfortunately this is the result of extending the survey route to include an additional 10 km section of highway which is less congested. In each case the travel time (or in this case travel speed) indicator cannot provide a reliable history.

It is important that all agencies responsible for managing the road network on the city are involved in defining the boundaries of the study area and the boundaries are defined to take account of the predicted future traffic patterns.

Once the boundaries of the study are have been defined the next step is to identify the principal road network within these boundaries. These are the roads that do and will in future years carry the bulk of traffic. It is the travel conditions on these roads that the indicator will report on. As a first approximation the principal road network should include all motorways and the arterial network. However, it is recognised that different roading authorities describe the road hierarchy for their area in subtly different ways. The general principal is to identify the two highest classes of roads (excluding motorways and urban motorways) which are referred to here as the major and minor arterials but not collector, distributor or local roads.

7.3 Sampling

Although the selection of a representative sample of roads is important, the main value of the indicator is in developing a time history report from year to year. So once the monitoring network is selected, the same road sections will be used each year in the future.

Congestion is a function of both the traffic demand but also the capacity of a road, where the capacity is determined by the road type and the environment through which it passes. There are two methods of selecting a representative sample of roads on which to undertake travel time surveys.

The first is to select a random sample of roads and to monitor these. The second is to select a representative sample. While the former is theoretically better, it is likely to result in a survey programme comprised of a large number of typically short sections of road which are less practical to survey efficiently. The approach adopted by Austroads is to select a representative sample of roads. This approach requires identification of the proportion of the road network that is made up of different types of road, and representative operating conditions on the different road types.

There is no point in only surveying the conditions on the motorways, and then seeking to apply these values to congested arterials passing through suburban shopping areas.

The sections of road that form the principal road network must be categorised according to their key characteristics:

• Road type (motorway, express way, arterial)
• Speed limit
• Number of lanes (not including additional lanes at intersections)
• Median provisions (median divided, flush median, no median)
• Surrounding environment (inner city, outer suburbs, extended shopping area).

Having determined the proportion of road in each category within the road network, it is then necessary to identify the typical operating conditions on these roads and to select a sample of roads that represent these typical conditions and the volume of travel undertaken on these roads.

This is done by carrying out the following calculation, for each of the road categories:

1. Estimate of the volume of travel (i.e. VKT - vehicle kilometres of travel) undertaken on the different types of road. Typically this estimate is based on a traffic model.
2. Select an appropriate representative sample percentage. The Austroads recommendation adopted here is for a minimum sample representing at least 15% of total travel.
3. The estimate calculated in point 1 above is multiplied by the representative sample percentage selected in point 2 above in order to obtain the control representative sample size (VKT).
4. The total length of route (for each category) is also multiplied by the representative sample percentage in order to estimate the required route length to be monitored.
5. Using a traffic model identify the typical operating speeds for each of the road categories by taking the VKT weighted mean speed.
6. For each road type identify those links where the operating speed is generally close to the weighted mean operating speed calculated in point 5 above.
7. The survey routes are then selected to make up required route length as calculated in point 4 above.
8. The selected routes are then checked against the control representative sample size which was calculated in point 3 above.
9. The selected routes are checked to ensure they may be surveyed practically and are representative in terms of the micro conditions (e.g. strip shopping centres), and the spread across the geographical area.

Once selected the individual road sections should be combined into a workable programme of routes. These routes should be 3-5 km in length in the inner city, and approximately 10 km in length in the outer areas and on motorways.

7.4 When to monitor

Traffic conditions on the road network vary throughout the day as shown in Figure 8 but also from day to day and from week to week.

To ensure that the monitoring programme reflects this variation the travel time surveys are undertaken three times per day representing the key periods:

• AM peak
• inter peak
• PM peak.

For each of the five week days the boundaries of the three periods are defined by considering the distribution of traffic volumes throughout the day using the following process:

Step 1 Obtain volume profiles (using standard traffic counts) at a resolution of 15 minutes for two-way traffic along the major roads that will be surveyed.

Step 2 Aggregate all volume profiles.

Step 3 Determine the:

minimum off peak hourly flow, F_min, in the period 9:00 to 17:00.

maximum hourly flow, F_max.

Step 4 Determine the 'daytime' hours as follows:

• start from 6:00
• from 6:00 onwards, eliminate any quarter hour periods where the quarter hour volume is less than a fifth of F_min - i.e. the flow rate in that quarter hour is less than 80% the minimum flow rate in off peak hours
• if the quarter hour flow is not eliminated, then that determines the start of the day
• if the quarter hour is eliminated, then check the next quarter hour to see if it is less than a fifth of F_min
• stop at 7:30
• do the same process backwards from 20:00 stopping at 18:00.

This determines the daytime hours which will have a start time somewhere between 6:00 and 7:30 and an end time somewhere between 18:00 and 20:00.

Example: Figure 9 below plots the quarter hour traffic volume for a single link which is used to illustrate Steps 3 and 4 above.

1. The minimum off peak hourly volume F_min = 455 vehicles/hour and occurs between 13:15 and 14:15.
2. The maximum hourly volume F_max= 933 vehicles/hour and occurs between 16:45 and 17:45.
3. The minimum flow for defining 'daytime' is F_min/5 = 455/5 = 91 vehicles/ 15 minutes
4. The daytime hours are from 06:15 (since the flow in the 15 minutes ending at 06:15 is less than 91 vehicles/15 minutes) until 20:00 (because all 15-minute periods before this have a flow greater than F_min = 91 vehicles/ 15 minutes). The period outside these hours (that below the horizontal line in Figure 9) has a flow below 91 vehicles/15 minutes.

Step 5 Having determined the daytime period the next step is to identify the three time periods. To do this first determine the quarter hourly flow threshold, which will be used to determine the cut-off between peak and off-peak periods, as follows:

Step 6 Determine the start and end of the morning peak as follows:

a. Consider the peak HOUR in the AM as all being part of the morning peak period.

b. Working backwards from the start of this HOUR, the start of the morning peak is the end of the quarter hour period in which the volume first drops below the threshold value.

c. Working forwards from the end of this HOUR, the end of the morning peak is the start of the quarter hour period in which the volume first drops below the threshold value.

d. Do the same process for the evening peak.

An example of this process is shown in Figure 10.

Having identified the periods of interest the next step is to develop a time schedule for the individual surveys. Although the start times for each survey are somewhat arbitrary, it is important that the surveys are not undertaken during the shoulder periods. These are the periods within each peak where the traffic volumes are less than the peak average. By considering the profile of each peak, based on the aggregated traffic counts, identify the time at which the traffic volumes first exceed the average for the peak period. This average is the representative volume for the period of interest.

The ideal time to begin the surveys is when traffic conditions are at the average for the period being considered, although some consideration of the practicality of the survey programme is also required.

The starting time of each travel time run, in each time period, for a particular direction shall be the same. This is illustrated in Table 13. It is important that the individual travel time surveys begin at the same time each day of the week and where multiple agencies are undertaking the survey programme the same peak periods must be used by all agencies.

Table 13: Example of travel time survey schedule

The very first sample of each monitoring period should start on time. A tolerance of ± 5 minutes shall apply to subsequent starting times. This is important to ensure that each successive monitoring activity is undertaken at the same time of day.

Because traffic volumes vary throughout the year it is not sufficient to monitor travel times with only a single weeks surveys. However, Australian experience is that two iterations of the surveys should be undertaken a year. These survey weeks should be scheduled to avoid abnormal events such as road closures, special events, road works, school and public holidays and periods of abnormal traffic. Figure 11 plots the weekly profile of traffic on urban arterial in terms the percentage of annual average daily traffic (AADT) that occurs during each week of the year.

Although the profile is relatively stable, it is clear that during the first few weeks of the year the traffic volumes are below average and similarly in the last few weeks of the year the traffic volumes are higher than average. Ideally the surveys should be undertaken in those weeks where the expected traffic volumes are close to 100% of AADT, i.e. during weeks 17-21 and 31-40. Where accidents or wet weather result in abnormally low travel speeds the surveys should be repeated.

7.5 Survey method

7.5.1 Preparation of travel time surveys

Having identified the road sections and the route, it is essential that these be driven over prior to any surveys. This initial drive over serves three key purposes:

1. To ensure that the routes are indeed representative and practical to survey.
2. Allows the survey team to correctly document the limits of the route, and identify the features that will act as timing points along the route. These should be documented and preferably photographed to ensure that the travel times are recorded at exactly the same points.
3. Identifies where traffic count stations may be set up to obtain a concurrent record of the traffic conditions on the route.

The latter point is most important. Although it is unlikely that traffic counters can be set up on every link in the monitoring network, it is important to have directionally split traffic counts that represent the traffic flow on the monitoring network. The traffic volumes from these traffic count stations are used to develop the weighting factors required in the development of the overall index.

7.5.2 Measuring travel time (floating car method)

The floating car method used to measure the travel time on the selected routes involves a driver and at least one passenger. The driver drives the route at the speed of the surrounding traffic, while the passenger records the travel time over the route. It is particularly important that the driver drives at the ambient speed which can be assessed by recording the number of vehicles which overtake the driver and the number of vehicles that are overtaken.

If the net overtaking rate is kept close to zero we can be confident that the driver did not travel faster, (overtaking more vehicles than overtook the test car), or slower (being overtaken by more vehicles than were overtaken by the test car) than the surrounding traffic. In some cases it is possible for the driver to monitor the overtaking rate while driving, but in difficult circumstances this role should be given to a second passenger. Where the net overtaking rate is significantly different from zero, a correction may be made using traffic count data collected during the survey period.

For example, if the volume of traffic in the direction of travel is recorded by a traffic counter as 2400 vehicles per hour, the average flow is one vehicle every 1.5 seconds. If the survey vehicle over took 10 more vehicles than overtook it the survey vehicle travel time was 1.5 x 10 seconds faster than the traffic stream and should be adjusted by that amount. An example of a recording sheet and the corrections for net overtaking rate is given in Table 14. Each section of the monitoring network is surveyed in each direction (except one way streets). This means that for each section there are effectively two links. It is important to ensure that the route, driver and date are recorded on the travel time recording sheet.

Table 14: Example of travel time recording sheet

Notes

(1) Number overtaken by test car - number overtaking the test car.

(2) In this example the traffic flow was 2400 vehicles per hour so each vehicle represents 1.5 seconds adjustment.

7.6 Data analysis

Having recorded the travel times along the route and adjusted these where the net overtaking rate is expected to make a significant difference, the next step is to aggregate the data to derive the mean travel time and congestion measures. The process is set out in Figure 12.

7.6.1 Calculation of mean link travel time

The mean link travel time for a particular period is calculated as the average travel time for the link, recorded over the five weekday surveys.

where i = travel time sample number (i = 1, 2, ..., n).

The basic formula to calculate the mean travel time (minutes) of a particular link for a particular period is:

n = number of travel time samples collected on link (normally 5)

j = link number (j = 1, 2, ..., X)

X = number of links to be aggregated

t_i = duration of trip (minutes) on a particular link in the ith sample

T_j = mean travel time (minutes) of link j for a particular time period

7.6.2 Calculation of the nominal travel time (NTT)

The nominal travel time is simply the time take to traverse the link, assuming that a vehicle was travelling at the legal speed limit, and is calculated as:

Nj_j = nominal travel time on link j

Lj = length of link j in km

Sj _i = speed limit of link j (km/hr)

The total nominal travel time for the set of links forming the monitored network is simply the sum of the individual values:

7.6.3 Calculation of the aggregate mean travel time

The aggregate mean travel time is a flow weighted average of the mean travel time for the various links. The mean travel time on each link calculated in section 7.6.1 above is entered into the table below, together with the nominal travel time per link (from 7.6.2 above).

View the calculation of the aggregate mean travel time (large table)

The volume of traffic estimated to use the link in the particular period is also entered into the table. This volume is the average period volume for the particular link or set of links and is used to weight the mean to reflect the volume of traffic to which the individual link means apply, as shown in the formulae below:

The Actual Travel Time (ATT_PERIOD) for a set of links for a particular time period is calculated using the following formula:

where ATT_PERIOD = ATT_AM (i.e. ATT for AM peak)

ATT_PM (i.e. ATT for PM peak) or

ATT_OFF (i.e. ATT for off peak).

The actual travel time (ATT) for the whole day is calculated by weighting the individual travel time measurements, i.e. the ATT_PERIOD values by the proportion of daytime traffic that occurs during that particular period using the formula:

where ATT_AM, ATT_PM, ATT_OFF are as calculated above

A, B, C are ratio factors which are described below

(Note: A + B + C = 1)

The ratio factors A, B, C represent the amount of traffic, which occurs in each time period. These ratio factors are calculated for the city as a whole from the output of the regional traffic model. These factors should be updated from time to time to account for peak spreading.

7.6.4 Calculating the time period and whole day values for CGI

The time period for an individual link _J CGI_J is simply the difference between the actual travel time on the link (ATT_PERIOD) and the nominal travel time (NTT_REDIOD) for the link.

Similarly the time period CGI for the whole of the monitored network is the difference between the total actual time and the nominal time:

The CGI is effectively the delay occurring on the link and has units of minutes/kilometres. To provide an overall measure that represents conditions through the whole of the daytime the individual period CGI values are weighted according to the volume of travel that occurs in that period.

The weighting factors are simply the network wide assessment of the volume of traffic that travels in the particular time period divided by the total over the three time periods. Such that A+B+C=1.

7.6.5 Travel time variability

Travel time variability VTT monitors reliability of travel times on the urban arterial road system, and considers routes rather than individual links. Individual routes are comprised of a number of sequential links and should have a total route length of at least 3 km.

The calculation of variability of travel time is basically a two-step process. The variability of travel time is first determined on a selection of routes throughout the metropolitan area. These values are then combined in order to calculate aggregated variability of travel time.

The link travel times from the surveys are summed over the links making up the route to give the route travel time. It is the variation in route travel time measurements with a particular time period that are of interest. Assemble the route travel time information for each floating car survey in a table similar to the one below.

Notes: To obtain a reliable measure of the variability of travel time, it is necessary to carry out more travel time runs than has been proposed for one particular collection period. Therefore, it is proposed that this indicator will only be reported once a year in order that sufficient runs can be taken into account for the calculation of variability of travel time.

Calculate the mean travel time (T_h) and the associated standard deviation SD_h for each route for each period using the following formulae:

where k = travel time sample number (k = 1, 2, ..., p)

p = number of travel time samples collected on route (normally 10-15)

h = route number (h = 1, 2, ..., Y)

Y = number of routes to be aggregated

t_k = duration of trip (minutes) on a particular route in the kth sample

T_h = mean travel time (minutes) of route h for a particular time period

SD_h = standard deviation (minutes) for a route for a particular time period.

The variability of travel time VTT_h for a route for a particular time period is calculated using the following formula:

where VTT_h = variability of travel time for a route for a particular time period

SD_h and T_h are calculated as shown above.

Factor 1.44 has been included above for the following reason. The formula for standard deviation is based on the assumption that the distribution of values of measured travel time will form a normal distribution. Even though this assumption might not be correct, the standard deviation still gives a meaningful measure. The interpretation of the results is that 68% of journeys are within ± one standard deviation of the mean travel time. However, it has been decided that 85% of journeys be taken into account when calculating variability of travel time. In order for this conversion to be carried out the factor of 1.44 is used.

When calculating the travel time variability across all routes for a particular period, it is necessary to weight the contribution of each route by the vehicle kilometres of travel on that route, not just the traffic count on the route. The vehicle kilometres of travel is simply the traffic flow on the route multiplied by the length of the route. The actual values of VKT_h are obtained by summing VKT for the links that make up the route.

The variability of travel time (VTT_PERIOD) for a set of routes for a particular time period is calculated using the following formula:

where VTT_PERIOD = VTT_AM (i.e. VTT for AM peak)

VTT_PM (i.e. VTT for PM peak) or

VTT_OFF (i.e. VTT for off peak)

VTT_h, VKT_h are as calculated above

h = route number (h = 1, 2, ..., Y)

Y = number of routes to be aggregated

As with the previous measures, the whole day variability of travel time (VTT) is calculated by weighting the individual periods by the volume of travel:

Ratio factors A, B, C are the same factors used for aggregating ATT and CGI.

7.7 Reporting

The values in the following table should be reported for each of the two surveys undertaken during the year and annually as indicated