[Home] [Session index] [Author index] [Presenter index] th
24 ARRB Conference – Building on 50 years of road and transport research, Melbourne, Australia 2010
THE ACCURACY OF INDUCTIVE LOOP DETECTORS Paul Briedis, Hughes Trueman, Australia Dr Stephen Samuels, Director TEF Consulting, Australia and Visiting Fellow, School of Civil and Environmental Engineering, University of New South Wales, Australia ABSTRACT A significant function of inductive loop detectors at a signalised intersection is to record the number of vehicles travelling through particular lanes. A key application of this data is their use by signal coordination systems to maintain an optimum level of vehicle control and management within a traffic network. Consequently, the accuracy of inductive loop detector data directly affects the quality of the outcomes achieved by the coordination systems. Several factors affect the reliability of inductive loop detector output data such as traffic density, vehicle movements, traffic composition and physical characteristics of the intersection such as pavement condition. The present paper reports an empirically based, limited, investigation of how these factors affect detector accuracy. Analyses and interpretations of detector output data and manually surveyed traffic behaviour data across intersection with the Australian Capital Territory, Australia, indicated that considerable variations in detector accuracy occurred over a range of scenarios. The paper will identify those factors that directly influenced the observed detector accuracies and consider the nature of these influences. Outcomes of this investigation might be applied over a range of scenarios of vehicle flow regimes and the physical conditions of an intersection in ascertaining the likely accuracy of inductive loop detector data.
INTRODUCTION There is evidence to suggest that there can be errors in the vehicle counts recorded by inductive loop detectors. Moreover, these errors, which can be considerable, appear to vary with different traffic scenarios. According to Smaglik, Bullock, Urbanik and Bryant (2006), while the outputs from both inductive loop and pneumatic tube detectors are reliable where traffic is free flowing, this is not the case at signalised intersections. May, Clayford, Coifman and Merrit (2003), Martin (2008), and Kalaee, Esmaeili and Hatami (2008) support these assertions. On the basis of the findings of all these authors, the following factors were identified:
The number of lanes in a traffic stream appears to influence detector error
Areas where clear or distinct traffic lane markings were not present seem to influence detector error
Traffic composition also appears to influence detector error, with inductive loop detectors apparently consistently over counting.
The empirically based investigation reported in this paper was aimed firstly at examining and quantifying inductive loop detector error. Secondly, it attempted to determine if and how four relevant factors may have contributed to any observed errors. Specifically, the investigation focused on the following four factors:
1. vehicle flow per traffic lane 2. pavement condition 3. specific vehicle movements 4. heavy vehicle content of the traffic.
© ARRB Group Ltd and Authors 2010
1
th 24 ARRB Conference – Building on 50 years of road and transport research, Melbourne, Australia 2010
The investigation was undertaken by the primary author, whilst an undergraduate student, for his final year thesis, under the supervision of the secondary author. Therefore, the investigation was somewhat limited due to the usual budgetary and logistic constraints associated with such thesis projects.
INDUCTIVE LOOP DETECTORS Brief overview A loop detector has been defined as ‘a detector that senses a change in inductance of its inductive loop sensor caused by the passage of a vehicle near the sensor’ (FHWA 1990, Martin 2008). Inductive loop detectors (ILDs) are the most common form of vehicle detection at signalised intersections. The components of an ILD system are as follows.
One or more turns of insulated loop wire wound in a shallow slot built into the road pavement, typically 50mm below the road surface
A lead-in cable connecting the loop wire to the kerb-side junction box
Detector electronics unit (loop vehicle detector), housed in the junction box (placed behind the kerb within the intersection).
Furthermore, each of the detector electronics units within a signalised intersection must be connected to the single signal controller box. ILDs are capable of recognising a passing vehicle in one of two ways (Austroads 2003). 1. Presence Mode. The sensor unit produces a continuous output while a vehicle is within the detection zone (i.e. detector loop) and is influenced by the length and speed of the vehicle. This mode can detect both moving and stationary vehicles. 2. Passage Mode. The sensor unit produces a brief pulse identifying when a vehicle has passed over the detector zone, thus detecting only moving vehicles regardless of their length or speed. Stationary or slow moving vehicles do not trigger further pulses while in the detector zone. Some typical loop configurations are shown below in Figure 1.
© ARRB Group Ltd and Authors 2010
2
th 24 ARRB Conference – Building on 50 years of road and transport research, Melbourne, Australia 2010
Figure 1. Typical loop configurations (After Austroads 2003).
DETECTOR ACCURACY There have been a number of documented studies dealing with the possible causes of detector errors. Martin (2008) found that variations exist between the accuracy of ILDs, with an expected accuracy range between detector and manual surveys of 98-92% (in this instance, a detector accuracy