The challenge
A reliable early fire detection system
The reliable early detection of a fire while at its early stages is desirable in tunnels, grain storage facilities, large workshops such as hangers, railway running sheds and stabling yards. Sadly, it is also a challenge best understood by those that have faced it.
Solutions based on thermal detection alone work poorly or not at all in large spaces that may have normal sources of heat. For example, hot engines, hot exhausts and exhaust gases and welding activities are sources of heat that don’t necessarily indicate fire.
Fire detectors work on detecting the presence of smoke or heat. Smoke or smoke-like particles are present in all of the environments described and don’t necessarily indicate the necessity to take fire fighting action. Detection of flame works on the thermal or visual characteristics of fire.
Road tunnels in particular are problematic. Overcoming false positives generated by video analytic and thermal imaging solutions in road tunnels is a challenge. There is a fine margin in setting the sensitivity so that it reliably detects a fire and does not pick up normal sources that may just be outside their maximum heat range.
For video analytics in road tunnels, it is virtually impossible to eliminate all the possible flashing frequencies from the amber lights of road maintenance and recovery vehicles and lane change indicators. While most errors can be overcome by using machine learning/AI, the training and setup costs are very high and the processing power required to analyse such scenes make the cameras expensive.
False positives desensitise operators to an alarm and in most real tunnel monitoring applications it is left up to human operators to make the initial identification of a fire. This works most of the time but in instances where it is not picked up quickly, the results can be disastrous in terms of loss of life and property.
Trantek MST – a company that manufactures, distributes, and installs fault-resilient solutions for control and management of rail and road infrastructure - sought to address the unreliability of these critical early warning systems by working with the University of New South Wales (UNSW). By measuring changes in Wi-Fi radio signals in the environment, more accurate and reliable detection of temperature changes, smoke emissions and flames can be achieved.
The Wi-Fi environmental sensing pilot at Sydney Harbour Tunnel by Trantek MST in collaboration with UNSW Sydney (patent pending), combined existing Wi-Fi hardware, artificial intelligence (AI) and machine learning and offers the potential for significant improvements in the accuracy and timeliness of fire detection. Since Wi-Fi chip-sets are commoditised, each device is about 100 times cheaper than existing specialised fire detection cameras, and Wi-Fi has the potential to pass through non-shielding walls, or have its coverage range extended by simply adding additional low-cost sensors.
The innovative solution reduces, if not eliminates, false positive fire detections; offers more accurate zonal location data of an emerging fire or smoke threat; and provides additional qualitative data on changes in temperature and air quality for first responders.
Our response
Testing a prototype in the Sydney Harbour Tunnel
Supported by an Innovation Connections grant and Innovation Facilitator Andrew Hodge, Trantek and UNSW Sydney undertook a research project to deploy a prototype for field tests.
In conjunction with Trantek MST, the incumbent mission-critical systems vendor for the tunnel, and the tunnel owner/operator, Sydney Harbour Tunnel Company, the researchers set up a series of transmitters and receivers to monitor the environment as a test car prepared for the purpose was detonated and set on fire during a scheduled emergency response training exercise.
This was the first real-life application of this technology.
The results
Improved fire detection for tunnel users
The testing system was a complete success and shown to work in real-time. The Wi-Fi emitter/detectors were able to show real time changes in the reflected wave forms (across different Wi-Fi channels) that indicated the presence of smoke, and the rapid temperature changes that are a signature of developing flames. The test also provided a comparison to existing camera-based video analytic technology, which was also deployed by Trantek MST.
Lionel Ascone, CEO of Trantek MST, said: “The testing that we performed inside the Sydney Harbour Tunnel shows that the Wi-Fi detection technology is capable of detecting a developing fire accurately at the early stages, picking up the heat gradient together with the associated smoke. It did this much earlier than the video analytic flame detection we installed in the tunnel earlier. This technology will not only dramatically reduce false positives but allow for a more informed and less disruptive presentation of alarms to the operators with a ‘watch and act’ warning when smoke and heat rise is detected followed by a full alarm response once a more significant flame is detected. This technology is a game changer in economic and reliable early fire detection of large spaces.”
Trantek has the confidence to continue to refine the system and pursue its commercialisation. The trial has also served as a demonstration of the pilot’s efficacy for supply-chain consideration. Once the Wi-Fi system is successfully commercialised, as well as being more reliable, it promises to be much cheaper and easier to set up and maintain than current systems.
UNSW's work on Wi-Fi for the accurate detection of combustion indicators, is the first time this has been done. This provides opportunities for diverse applications and will change how fire detection is done in the future. We thank the facilitator, Andrew Hodge, who helped us at every stage of the Innovation Connections process.
— Prof. Aruna Seneviratne, UNSW Sydney
Our partner
