From Vallvidriera Tunnel in Barcelona in 1991 to the Oresund crossing last year, automatic incident detection (AID) has becomes established as a prime part of tunnel control systems.

In the years spanning these two projects, image processing systems have developed from advances in closed circuit television and information technology. Powerful processors allow the almost instantaneous processing of large quantities of data.

The result is a speeding up of the main operational function of emergency operations: incident detection time.

There are benefits for safety too in risk analysis procedures. Factors that can affect tunnel users include:

  • Black hole effect: difficulty of adapting vision at the entrance due to changing light levels

  • Dazzling: At the tunnel exit the phenomenon is re-peated due to the inverse change in light

  • Wall effect: Drivers tend to move away from the tunnel wall very close to the neighbouring lane, with the consequent risk of collision.

These phenomena are associated with the driver’s vision, but there are factors inherent to all tunnels that can give users a sense of danger or risk, that affect the safety of driving.

  • Geometric characteristics – section plan and side view

  • Pollution/visibility (ventilation)

  • Illumination

  • Signalling

There are also other elements of the carriageway that the user does not notice directly, but that are taken into account by designers and operators:

  • Traffic intensity and characteristics that determine operating procedures

  • Traffic control system with indispensable operational tools

  • Evacuation and assistance equipment

  • Maintenance work etc.

Safety can be improved by the designers and operators with two fundamental acts:

  • A good analysis of risks/safety that allows the best operation procedures to be defined.

  • Installation of control systems that allow any event to be detected and reduce to a minimum the operators’ response times.

The procedure used in recent times by Sainco Tráfico to evaluate risks has been the HAZOP method. HAZOP analysis is carried out intuitively by a multi-disciplinary team of experts from different fields of the safety system. In a series of meetings they review, under the guidance of an independent co-ordinator, each part of the operation critically though creatively.

The search for failings is carried out methodically via a system of key words applied to each sub-system, which allows possible deviations from ideal functioning to be identified in a systematic and ordered way, thus detecting potential failures and human or material damage.

A complementary procedure, “What If”, allows analysis of the effects of those failings, in order to take action on those that present high risks or whose occurrence is undesirable.

The result of using both methods allows:

  • Risks to be identified

  • Discover the causes of possible failures

  • Determine the effects of those failures

  • Provide solutions to minimise the consequences of failures.

Risk minimisation

The most important points that minimise the risk and consequent effect are:

  • Knowledge of what is happening

  • The detection/response time.

The first is of vital importance to solve a risk situation. An operator from a control centre will be able to co-ordinate the appropriate assistance or act on the traffic guidance systems if he has “real” knowledge of what is happening.

Detection/response time is considered in two phases: The time taken in detecting what is happening and the time taken in responding. The operator must be advised by the control systems as quickly as possible to keep the response time to a minimum.

Currently, the most advanced AID technology is video image processing, which is able to detect incidents in just a few seconds by means of TV cameras. Among other data detected are, for example, slow moving or stationery vehicles, or those being driven in the wrong direction.

An AID system consists of analysis of television images originating from cameras located along the carriageway, which allow all anomalies to be detected.

These systems can only detect what the camera is able to see. Therefore the location and position of the cameras is fundamental to avoid blindspots caused by heavy and/or other vehicles.

Image analysis

The field of vision, or image, that will subsequently be analysed by the AID device, is also important. Its optics are closely related to the close coupled device of the camera, making its inclination and height over the carriageway crucially important.

To pick up and act on anomalies shown by the camera, an operator in a control centre requires a TV monitor for each camera or multiple images shown as a mosaic on one screen.

With the level of attention needed to watch the screens and problem of fatigue, operators can only attend to a maximum of six or eight monitors at a time.

In analysing the variation in the pixels of the TV images in real time, the AID system on the one hand takes over the tedious task of analysis and, on the other removes the need to observe all the images in the control centre (only those that have a detected event).

This allows the operator to attend to those events that are produced with the traffic in real time and to react in consequence.

The AID systems, which were initially developed as non-destructive counting and vehicle classification systems, currently allow detection of a stopped vehicle, a sudden speed reduction, congestion and its queue length. They can get occupancy times and other classical traffic parameters, such as speed, number of vehicles and classification.

The first step in the design process will be evaluation of the number of CCTV cameras, their location and characteristics for the specific carriageway, to take best advantage of the technology. Selection of the camera, its position and its optics are parameters that need a detailed analysis.

The act of detecting an incident in this way in a matter of seconds, independent of traffic flow, does not allow the use of algorithms to handle data from more conventional sensors. Such sensors are located along the carriageway such as magnetic loops or infrared detectors. Their degree of precision depends greatly on the flow of vehicles along the carriageway.

In general a video image processing system can only analyse and detect information from the TV cameras. Therefore, all the different aspects must be kept in mind when choosing and locating the cameras.

Accepted

The development in the performance of these sys-tems and the economy of scale has converted AID by image processing into an accepted system on any road, fast becoming regarded as indispensable in tunnels and other strategic points.

And for the future, advances in image processing will offer better solutions to design incident detection systems for tunnels. In recent years Sainco Tráfico has provided more and more AID solutions as sub-systems of its SICOTIE tunnel control system. This assures total coherence of data, alarms and operational procedures, and a new level of automation and operator assistance.



Related Files
AID system