Motorway Traffic Flow Optimization: From Theory to Practice

TII collects significant quantities of data which can be used to understand traffic patterns and driver behavior on the M50 motorway. Figure 1 provides an overview of some of the ITS equipment measuring data on the M50. This data includes speed, flow, occupancy, headway, and vehicle lengths recorded every 20 s from inductance loops located at 500m intervals along the full 40 km length of the motorway. In addition to this, there are a number of weather stations recording real-time weather information related to the road conditions. Camera technologies provide both CCTV and Automatic Number Plate Recognition (ANPR) capabilities covering the whole length of the M50 and provide detailed information on traffic conditions and journey times. Detailed logs of the displayed speed limits are also recorded along with details of incidents on the motorway, or any ongoing or planned roadworks.

In order to develop appropriate response strategies to mitigate the effects of congestion caused by general traffic flow breakdown, it is important to understand the locations at which the congestion typically initiates [5]. A software tool was developed to allow the loop data to be processed and plotted on a time-space heatmap. This visual representation of traffic data allowed congestion patterns on individual days to be examined, and also allowed other influences such as incidents, or adverse weather conditions to be overlaid to identify external causes of congestion. Figure 2 illustrates how the heatmaps can be used to visually identify the typical congestion seed points (Fig. 2(a)), and the congestion associated with incidents (Fig. 2(b)). Days where congestion is caused by incidents or extreme weather can then be separated from days where general congestion occurred due to excess demand causing flow breakdown.

Bild vergrößern

While Fig. 2 shows congestion patterns on two single individual days, when examining recurrent congestion seed points it was necessary to combine the data from multiple days. Generating heatmaps of the average speeds for specific days of the week (with incidents, holidays and adverse weather removed), allowed recurring seed points to be clearly identified so that further investigation of the causes of congestion could be carried out. Detailed reviews of each seed point, using a combination of traffic data from inductance loops and CCTV footage allowed the source of congestion to be identified. Typically, flow breakdown initiated near junctions, where merging traffic, driver behavior or road geometry resulted in a localized capacity reduction. Understanding the nature and source of congestion build-up at each seed point, allowed appropriate speed management plans to be developed to optimize traffic flows and to mitigate the effects of congestion on the M50.

In order to optimize traffic flows through the application of VSL in a proactive manner, speed limits must be reduced in advance of flow breakdown occurring. Control room operators have traditionally relied on visual interpretation of the data feeds, along with CCTV footage. However, given the dynamic nature of traffic conditions, it was found that visual observation alone was not sufficient to allow operators to accurately identify the onset of congestion before flow breakdown occurred, meaning that without additional guidance, speed plans would often be implemented after congestion had already started to form, in a more reactive manner. In order to identify the traffic conditions which are likely to initiate the onset of congestion and facilitate a proactive response, historical loop data was leveraged to establish the fundamental relationships between speed, flow and density/occupancy along the motorway. This provided a measure of the capacity being achieved at each seed point, and an understanding of the traffic conditions being experienced directly in advance of flow-breakdown. Figure 3(a) illustrates the flow-occupancy relationship for one location near junction 6, based on 5 years of data from 1^st January 2017 to 31^st December 2022. It can be seen that typically, a capacity of approximately 1,600 veh/hr/lane is achieved, however there is a large spread in the data points. This is indicative of the various factors which influence traffic behavior and hence the achieved capacity (e.g. time of day, weather conditions, time of year, traffic lanes). More details on establishing the fundamental diagrams using measured traffic data can be found in [7].

Bild vergrößern

The application of traffic flow theory to establish the fundamental diagrams for each seed point provided a general understanding of the capacity being achieved, and the various factors influencing capacity. However, in order to develop suitable alerts to assist operators with the implementation of proactive speed management plans, a detailed study was carried out to examine the traffic parameters directly in the lead up to flow breakdown occurring. A number of metrics were studied, and it was found that the 5-min rolling average speed was a reliable indicator of the onset of congestion. Suitable trigger locations and threshold values for the 5-min rolling average speeds were identified for each seed point and audible alerts were configured in NIMS to warn control room operators of traffic conditions when the thresholds were breached. Upon receiving an alert from one of the trigger locations, operators implement reduced speed limits along the relevant section of the M50, following pre-defined plans. Figure 3(b) shows the trigger locations and geographical extent of the speed reductions implemented by operators for each trigger during the morning peak, for a portion of the Southbound carriageway. It is noted that the speed limit of this part of the motorway is 100km/h, and the operators reduce the speeds to 80km/h following a trigger, and further reduce the speeds to 60km/h when the data indicates that flow breakdown has occurred.