Improving the Measurement of Congestion: Final Report

1. The DTLR has a target to reduce road congestion and has chosen average delay per vehicle kilometre as its indicator to measure progress against the target.

2. We believe that the DTLR approach is sound (and helpful if accompanied by effective measures to improve network performance), but should be supplemented by information which is more meaningful to motorists by:

• measuring performance of the road network in terms of the recognisable effects of congestion (eg impacts on journey speeds, times and reliability) on specific links/routes/locations,
• ensuring that motorists have access to information on the predictability of journey times, which enables them better to plan the way in which they use the road network.

3. Techniques for data collection and analysis on journey times already exist or are being developed in the UK, through the public and private sectors (eg Highways Agency, Trafficmaster).

4. Practical and cost-effective complementary indicators may illustrate aspects of the road network not captured by the Department's measure. Annex 1 gives two examples of what is currently possible - from TRL research commissioned by the Highways Agency and from data collected by one UK private sector provider of traffic information, ITIS.

5. However, further work is needed to ensure that these and other possible approaches to measurement are robust. Our understanding is that completing the necessary work could take until 2004/05 at the earliest.

6. We recommend therefore that, as part of the outcomes from the forthcoming review of its 10 Year Plan, the DTLR commits to:

• Adopt by 2003 (in principle, and subject to further work) additional indicators of congestion for both the strategic and non-strategic network to supplement its current form of measurement, with a view to making these operational by 2005. Two broad options appear worth pursuing:
• an indicator which shows the percentage of time spent travelling at speeds below an "agreed" threshold of performance in conditions where a significant proportion of vehicle speeds are appreciably below the speed limit. Further work is needed (through proper consultation with motoring organisations and others) to agree "acceptable" baselines of performance, below which the measured part of the network is deemed to be congested.
• an indicator of journey time reliability, which can be expressed technically as a coefficient of variation (defined as the standard deviation divided by the mean, as a percentage) - subject to meaningful outcomes from current DTLR research into the relationship between congestion and reliability. Further work is also needed to establish the robustness of such an approach, given the distorting effect that relatively few road sections with abnormally low average traffic speeds may have.
• Work with motoring organisations and key private and public sector players to promote the provision of information on journey time predictability by 2005 (if not earlier), which enables motorists better to plan their journeys.
• Establish the desirable level of disaggregation of information for indicators of congestion and predictability, through consultation with motoring organisations and key private and public sector players, eg:
• geographically - potentially specific links on the motorway and strategic trunk road network (ie stretches between junctions), and key urban areas where motorists most often encounter congestion on a regular basis.
• temporally - potentially peak and off peak.
• Work with key players in the public and private sectors to improve the data collection needed to ensure the robustness of the proposed indicators as they relate to:
• The strategic road network for which the Highways Agency is responsible. By early 2005, the busiest 30% of the motorway network will be covered by inductive loops capable of generating robust journey time data under the MIDAS (Motorway Incident Detection and Automatic Signalling) project. The Highways Agency should be encouraged to develop options for data collection on its network not covered by MIDAS, which might include use of TrafficMaster infra-red sensors, GPS-generated information and installation of more induction loops (beyond those already due as part of plans for a national Traffic Control Centre).
• Non-strategic roads. GPS-generated information can generate significant amounts of data. In the case of ITIS, for example, over 100 million data values daily are generated, of which approximately 60% relate to movements on the non-strategic network. While anecdotally this particular set of data appears to provide accurate information about road conditions, further work is needed to establish whether the level of data and nature of its generation (via HGV and coach movements) are sufficiently robust.
• Build on existing research into the reaction of motorists to the expression and use of any congestion indicators to establish the user-friendliness and accuracy of any selected new indicators. For example:
• on journey time reliability, should a coefficient of variation prove technically suitable as an indicator, the coefficient itself may not be the most useful way of expressing the indicator to motorists. Another option might be to establish an index of extra time needed to be 90-95% (say) confident of completing a journey within a certain time as compared with some baseline (similar in concept to the illustrative presentation of data on predictability arising from the HA/TRL research indicated in the annex).
• on journey speeds, the ITIS example illustrated in the annex indicates what might be possible, subject to establishing the robustness of such an approach. If a significant proportion of vehicle speeds on a route or in an area is below a specified proportion of the speed limit, then that road would be considered to be congested. Further work is needed to establish whether it is necessary or desirable to use baselines using the same proportion, when measuring performance on different parts of the road network.
• Identify how the package of measures in the 10 Year Plan (or any further iteration of it) is likely to affect road performance, as measured by any new indicators. This would replicate projected changes in congestion as measured by the DTLR's current indicator, included in the 10 Year Plan. Further work would be needed to establish which, if any, of the suggested new, disaggregated indicators might be adopted by the DTLR to measure the impact of delivering the 10 Year Plan.
• Continue to monitor changes in congestion using the DTLR's current approach, until such time as any new indicators can be measured alongside it. One option may be to use an interim additional indicator (for example, related to the motorway network covered by MIDAS) until more extensive arrangements can be put in place.

Annex 1: Highways Agency/TRL research into journey time predictability

The HA, via a contract with TRL, has investigated the repeatability of journey times by time of day, day of week and season on congested sections of the motorway network, and the practicality of making such information available to the public. Detailed journey time data from HA owned Automatic Number Plate Recognition (ANPR) cameras on the M6 and M5 around Birmingham has been collected and analysed.

An example of a first attempt at a method of presentation is shown below. This illustrates the broad journey speeds/times experienced for 90% of trips made on the measured link at specific points in the day. The darkest bars indicate periods of highest probability of experiencing most congestion.

The study has suggested that there is sufficient predictability of journey times on congested sections of motorway to consider further the promulgation of the information to the public. The Agency is now to consider if and how it might develop the idea.

90th percentile speeds/journey times on the southbound M6 from J13/12 to J4

[Driver Information on Journey Time Variability Generated Using ANPR Data - B Frith and D Pearce, Proceedings of the 11th International Conference on Road Transport Information and Control, March 2002].

ITIS traffic data analysis:

ITIS operates a Floating Vehicle Data network comprising 130,000 units (mainly high usage HGVs & coaches) generating data records by GPS which are currently aggregated by road and day/time category for a number of applications. The monitoring is capable of capturing over 100 million data values covering various parts of the road network. The data can then be presented in a variety of ways, one of which is given below.

Central Manchester Postcodes (M5/M6). Congestion shown as number of records below 50% of the speed limit.