Computerised Real-time Saturation Flow Measurement for Signalised Traffic Junctions
I Henderson, TRL, UK
Saturation flows are essential for optimising control of signalised traffic junctions. Manual measurement is costly, so an automatic computerised method is being developed to produce real-time values for junction control and monitoring systems.
Computerised real-time saturation flow measurement for signalised traffic junctions
Saturation flow measurements are essential for optimising the control of signalised traffic junctions. Manual measurement of such flows is time consuming when done thoroughly, especially when separate readings are required for every lane. A number of readings may be required where peak flow is tidal, or where traffic composition varies between peak and off-peak periods.
In response to this, a computerised method of saturation flow measurement using inductive loop detectors (or equivalent) is being developed. This automates and simplifies the measurement of saturation flows, tracking changing values over time, and producing real-time saturation flow values that reflect current driving conditions. The method has been designed to be ?stand-alone? and could be implemented within any suitable traffic control or monitoring equipment, or used with junction microsimulation programs.
Results so far have been promising, and preparations are being made for on-street trials. Assuming the method works reliably, the data collected will be extremely valuable in determining future junction design. The data should also be useful in real-time systems, especially when seeking to maximise capacity.
The method is based on stage/phase and detector data in conjunction with junction configuration information. For each lane, the start of green is determined, and vehicle arrivals are recorded until the start of red. The detector data is then analysed to determine the number of vehicles discharging at saturation, and the time those vehicles took to discharge. From this, a saturation flow value is calculated in vehicles per hour, and combined with an exponentially-smoothed average to give an updated value. This can then be recorded and used as befits a given purpose. For instance, in a real-time junction control strategy, the data could be used to assess the end-of-saturation for control purposes.
Measurements of saturation flow can be made at stop-line detectors or detectors positioned further upstream. By default, stop-line detectors are used where available and suitable, reflecting the classic manual method of saturation flow measurement. If there is no stop-line detector, or the stop-line detector fails a consistency check (some stop-line detectors may be set to high-sensitivity or span 2 or more lanes, making them unreliable for counting numbers of vehicles), an upstream detector will be used where available. The method is designed to work with upstream detectors positioned up to 50m from the stop-line.
For a lane with a suitable stop-line detector, the method works as follows:
1. The first vehicle discharging at saturation flow is identified by discounting the first 3 vehicles crossing the stop-line detector after the start of green. (These vehicles are considered to be suffering from lost start-up time.)
2. The last vehicle discharging at saturation flow is identified based on the detection of a critical gap, or the start of leaving amber if the lane is fully saturated. A larger critical gap is used when heavy vehicles are thought to be detected.
3. A saturation flow value is calculated from the number of vehicles detected at saturation, and the time they took to discharge. An estimate of lost start-up time for the first three vehicles is also calculated.
A different method is used for upstream detectors. This is because saturation flow begins later, resulting in fewer vehicles being detected at saturation flow. Two estimates are calculated where possible:
- A ?long queue? estimate is carried out in a similar way to stop-line estimates, with saturation flow being measured at the position of the upstream detector instead of the stop-line. This is less useful in off-peak periods since it requires queues extending significantly beyond the upstream detector.
- A second ?short queue? estimate serves as a back-up for times when there are shorter queues extending not much beyond the upstream detector. This uses upstream detector data to estimate saturation flow at the stop-line, rather than at the upstream detector itself.
An average of the long and short queue estimates is then computed, and weighted towards the more reliable long queue estimate.
The method described above has been implemented and tested using the HUTSIM microsimulation, both on its own and in conjunction with the MOVA junction control system. Several refinements have been made, including:
- The critical gap used by control strategies to determine the end of saturation flow is typically set generously to ensure that saturation flow has ended before action is taken. However, for measurement purposes it was found that a stricter critical gap is required to ensure that saturation flow has not ended before an estimate is made.
- It was found necessary to discount a similar number of vehicles suffering from lost start-up time at an upstream detector as at a stop-line detector.
Comparisons were made between saturation flow measured at stop-line and upstream detectors. Measurements obtained using an upstream detector were found to be within 2% of measurements made at the stop-line in the same lane. Standard error was also well within that found in measurements from real junctions.
Association for European Transport