Collect-R cameras were placed strategically to monitor traffic flow, average speed, headway, vehicle occupancy, concentration and class
Traffic flow has been improved, congestion eased and safety increased - by a system of 'stepped speed limits' introduced to UK roadworks. URS Scott Wilson principal consultant Jamie Uff reports
Studies have shown that road users approach roadworks with differing behaviours. Some motorists brake gradually while others maintain a consistent speed before braking sharply, often in response to visible enforcement of a reduction in speed limit. The result of these different reactions is a tendency for increased lane changing, breakdown in traffic flow and a build up of queues, increasing risk of accidents on the approach to roadworks.
One possible solution to this problem is a system of stepped speed limits (SSL), developed by consultants
Typically, SSL involves implementing an additional mandatory speed limit for a short distance on the approach to roadworks 10mph (16 kph) more than the limit through the works. For example, an additional 60mph (96 kph) speed limit would be applied prior to a 50mph (80 kph) speed limit through the roadworks.
Schematic of SSL implementation
SSL is not a solution which should be introduced to all roadworks. It is suggested that it is only implemented on major roads where at least one of the following statements is correct:• There is likely to be an increase in the number of accidents due to the roadworks;
• Where there has been a history of unsatisfactory speed limit compliance; or,
• The existence of roadworks will increase queuing on the approach.
On the schemes where SSL is implemented, significant benefits for all stakeholders are expected. Initial micro simulation modelling of SSL indicated that the concept could yield many benefits including reduction in average travel time (and its reliability and predictability), increased flow throughput and easing of congestion.
Due to the inherent limitations of modelling (its basis upon assumptions, hard coded behaviour, no true accounting for real interactions and that safety benefits could not be measured), it was decided that on-road trials were required to validate the results.
The Trial
Opportunity to apply an SSL trial arose in 2009 when the SSL project team was approached by the Highways Agency's Major Projects directorate and contractorThe speed limit through the 3km long roadworks had been reduced from 70mph to 40mph for the duration of the viaduct replacement project for safety reasons and on account of the temporary road layout.
Consequently, it was not unusual for significant queues to form on the approach to the A34 scheme, particularly in the mornings when traffic flows into Oxford are at their highest. Therefore, a trial of SSL was proposed for the southbound A34 carriageway, which takes drivers into Oxford from the M40. The SSL would set a step down of speed limit from 70mph to 50mph to 40mph, with the 50mph restriction established for a distance of 1.1km before the 40mph limit at the start of the 3km of roadworks.
One challenge faced by the team was how to avoid slowing down traffic unnecessarily outside of the morning peak time. The answer to this came in the form of SmartBlinds from
These remotely operable signs enabled the SSL to be switched on and off depending upon demand. However, to ensure a robust dataset from this trial, the project team decided that SSL should be activated consistently, covering the morning peak hours between 6am and 10am each day.
The primary data collection device used to capture driver behaviour metrics was the Collect-R video based system from
The cameras had very low power requirements so could be powered by battery if renewable (solar/wind) or mains power could not be sourced; and they were significantly cheaper per unit than alternatives capable of capturing all of the metrics required.
Three Collect-R cameras were installed to capture a baseline of data before SSL was introduced and then remained in place for six weeks with SSL in operation.
The cameras were placed strategically to capture driver behaviour at different points. Camera 1 was placed at a significant distance ahead of the speed restrictions to monitor 'normal' unrestricted behaviour of the road users; Camera 2 was located half way through the 1.1km of 50mph speed restriction; Camera 3 was sited 1km inside the 40mph section, alongside an existing average speed enforcement camera.
Throughout the monitoring periods the Collect-R units were measuring traffic flow, average speed and headway, and vehicle occupancy, concentration and class.
All data captured for each of these parameters was averaged over every 15 minute period and stored to on-board memory for later daily retrieval via a remote wireless connection.
Data Processing
There were five main stages involved in the processing of the trial data. The first involved merging all downloaded data into one amalgamated database for ease of analysis. Stages two and three were implemented to ensure that any events or unexpected data values which could distort the final results were removed.Firstly this involved removal of data collected on days where incidents, such as road traffic accidents, were known to have occurred nearby. This was necessary as any incident on the surrounding transport network was likely to have an effect upon the flow characteristics at the trial site. Secondly, outliers beyond expected values and indicating an error of some sort were also removed. For example, where extreme or zero values of headway occurred the data was removed to avoid the results becoming skewed towards the outlier.
The primary aim of processing stage four was to isolate data captured on comparable days and periods of time.
Accordingly, only weekdays within school terms were assessed to ensure the flow characteristics were comparable. Data collected during public holidays, weekends and school holidays was filtered out of the dataset for analysis. Further segmentation removed data captured outside the targeted time periods of between 6am to 10am when SSL was operational.
Finally, processing stage five involved analysis of the final datasets, to identify changes of behaviour before and after SSL implementation. For example, the graph above shows how the flow and average speed data captured by the Collect-R sensors were used to calculate the quarterly weighted percentage change in average speed before and after SSL.
The dotted line at -30% indicates the theoretical percentage drop expected at Camera 2 due to the speed limit changing from 70mph to 50 mph.
The results
Following analysis of the modelling and trial results, SSL has been shown to improve journey times through roadworks. The A34 trial showed that up to 29 seconds (representing a 7% journey time improvement over the measured section) could be saved on average per vehicle during the peak hour; traffic queuing approaching the roadworks was reduced during peak periods; vehicle headway approaching and within the roadworks improved by up to 14m; and speed limit compliance increased.Greater headway between vehicles in particular is expected to significantly improve safety at roadworks. The difference this may make becomes apparent when considering that recent research by TRL found that 22% of all road traffic accidents in roadworks are caused by close following.
