[
Are we reaching the limits of existing fire detection technology? Mohamed Cherif Benzerari thinks we may be, and here he sets out an alternative way of quickly and accurately detecting genuine fires by using mass spectrometry techniques.
Mass spectrometry technology may be an alternative and effective method of smoke detection, since it can precisely distinguish molecular compounds of smoke from other particulates not generally produced by fire. The introduction of a unique ID for each material, fixture or product in a building will add substantial sophistication to an addressable fire alarm system by not only locating a fire, but also assessing the burning material in question.
Advances in microelectronics, information technology and fire science have led to substantial progress in fire detection technologies over the last decades. Although unwanted alarm rates have, according to statistics, fallen slightly, it is widely recognised that more progress has to be made.
Various new technologies such as video smoke detection are available, while existing ones such as multi-criteria fire detectors based on advanced signal processing algorithms, have been further developed. Yet what we think of as ‘state-of-the-art’ fire detection technology may be out of date in a few years, if it cannot provide high sensitivity with false alarm management in a cost-effective package. To put it bluntly, existing ionisation and optical smoke detection technologies have been stretched to their limits, and their future development may be subject to diminishing returns. A situation may soon arise where any marginal benefits to be had from these technologies may be outweighed by the investment needed to produce them.
The main challenge of fire detection is that it covers a wide range of scenarios, from smouldering to flaming combustions, and from areas such as a small lift motor room to very large multiple occupancy buildings. Unfortunately, there is no single standard solution that can be applied to that wide range of fire scenarios. Basically, solutions are formulated for a particular fire scenario in a given space, and do not always take into account environmental factors such as corrosion, humidity and wind.
Smoke detectors sense the airborne particulates in flaming or smouldering combustions, either by point detection or, in the case of aspirating systems, drawn through ducts into a sensing unit. But particulates in the air can be made up of dust, powder, steam and fog – as well as from smoke. In the case of optical sensors, scattered light beams measure any increase in the percentage of obscuration, while ionised air type detectors measure any reduction in current.[1] Those variations are measured as a proportion of the total amount of particulates entering the sensing chamber, regardless of the sort of particulates they are. So the fire detector in this case works just like a basic lift monitoring system: when it senses an equivalent weight of, say, 16 people, it generates an overweight warning signal, regardless of how this weight is made up.
There are a number of reasons why these existing sensing methods will soon be out of date:
– Ionisation detectors, which are usually the most suitable to detect the smaller particulates produced by rapidly growing flaming fires, have been banned from some countries due to their radioactive source (Americium 241). There are also severe restrictions on their storage and transportation, which means their use is declining rapidly.[2]
– Optical detectors are more suitable for detecting the larger particles of slow smouldering fires but conversely, they have difficulties in detecting flaming fires with small particles. In addition, non fire-generated particles, such as dust and other nuisance aerosols, may still cause false alarms.
– 95% of deaths in fires are caused during the smouldering phase of a fire, and while the victims where sleeping, while 80% die of toxic smoke inhalation.[3] Both fire detection methods are unable to detect poisonous gases such as carbon monoxide.
Certainly, signal processing algorithms have advanced the intelligence of these traditional methods by incorporating heat sensing into the same smoke detector. But in terms of early detection, the criterion of ‘heat’ comes a bit late in the majority of flaming and smouldering fires. A medium-term solution for this paradox is the use of a combined CO/optical smoke detector developed with advanced signal processing algorithms. This is much more efficient than a combined heat/optical detector and, when necessary, can be upgraded with infra red and/or heat detection capabilities. But for a longer-term approach, a radical solution based on detecting the compounds at a molecular scale (nanoscale) – produced exclusively from a fire signature rather than any other extraneous sources – should be considered.
From particulates to compounds
Mass spectrometry is a well proven and reliable technology, having been applied successfully during the last few decades in many advanced areas of research and application. These include satellites and spacecraft, biotechnology, biochemistry, geology and medical equipment. Incorporating this advanced technology within the future fire detection is fundamental to better fire detection .
A mass spectrometer is a sensitive and an accurate detector of molecules; it can even weigh the molecular mass of any element with an accuracy of 0.01%. With advanced software such that based on Fourier Transform Infra-Red (FTIR)[4, 5] and chromatography process, it can even trace quantities of contaminants and toxins such as CO, CO2, and HCN (hydrogen cyanide), which may be harmful to humans if their threshold levels are exceeded within the surrounding environment. The mass spectrometer has three essential functions or stages: the ion source, mass analyser and the detector (figure 1):
– The ion source In order to measure the characteristics of individual molecules in an air sample, a mass spectrometer converts them to ions, usually cations (positive ions which are molecules or atoms with one electron taken away), so that they can be moved about and manipulated by an external electromagnetic field.
– The mass analyzer The ions are sorted and separated according to their masses and charges. By this means, lighter ions are deflected more than heavier ions and so are separated from each other.
– The detector The positions of separated ions on the detector are a function of their masses; results are displayed in a spectrum graph style, ready to be analysed, as in figure 2.
Application to fire detection
A mass spectrometer based fire alarm would have to be designed and manufactured using an aspirated system architecture style, where the processing unit is extended by ‘sniffers’ or ‘perforated ducts’. In this way, early detection is achieved by drawing the air samples to the processing unit, instead of passively waiting for the air samples to reach the sensing unit, as in the case of a point detector. Artificial intelligence software would manage the comparison of the spectrum of the sensed compounds against its vast spectral database.
Secondly, every liquid and solid has a unique chemical representation. But once a combustible material gets into contact with enough heat and oxygen, ignition takes place. During the very early stage of the material’s combustion, the first quantity of its molecules undergo an initial chemical reaction and are transformed into gases, with new chemical forms and/or representations. This process enables us to set out a unique Material Fire Code (MFC) for each material or furniture in a building. The code could be considered as the ‘DNA’ of the combustible material or furniture, in terms of types and proportions produced during the very early stage of combustion.
Each manufactured material or furniture would be sent to a test centre, where it would be burned, tested and analysed. It would then be labelled with its own MFC in the form of a barcode, describing its unique combustion compounds. The code can then be scanned by the fire alarm system once the material is purchased, placed or fitted in a building; the fire alarm system should then be able to recognise its chemical or gas compounds in the first few seconds of combustion. In order to achieve this, however, the fire alarm system must have sufficient memory capacity to record a full spectral database of all furniture and materials in a building. In this way, the artificial intelligence embedded in a fire alarm system should be capable of sensing the various species of gases relating to the burned furniture or material, and of generating a range analogue values rather than just one value. This would provide information not only that the fire threshold level is reached, but also additional information about the burned material in question, the types of toxic gases, and the location of the combustion within the room.
One drawback with this approach is that it’s likely to be efficient only in the very early stages of a fire. Once the fire reaches a larger size, it wouldn’t be possible to decipher the mixture of all compounds with their proportions, each in relation to its original combustible material.
In conclusion, standards need to be a key part of the process, and manufacturers of all sorts of combustible materials and furniture will have to collaborate with the fire industry to achieve this. If done in this way, manufacturers of fire alarm equipment would be able to produce such equipment at economically viable volumes.
Certainly, the introduction of such coding may seem a long and tedious requirement, and possibly may be considered over-engineered for many low risk premises. So further research, testing and analysis has to be carried out to test the viability of the this approach, then to apply it initially to high tech applications. But the benefits of early and accurate detection, together with a brand new tool for post-fire investigations, are certainly imaginable.
Mohamed Cherif Benzerari Elec.
Eng TMIET is a commissioning engineer
and software developer at Drax (UK).
References
1 Roger Barrett, Fire detectors, FEJ & FP page: 55, June 2004
2 Serio, MA, Bonamno, AS Knight, KS and Newman, JS. An FT-IR Based System for Fire Detection, NIST Annual Conference on Fire Research, Gaithersburg, USA, 1994
3 The detection principles of IQ8Quad, Esser, Art.No: 797989.G0, p8, December 2007.
4 Grosshangler, WL, An Assessment of Technologies for Advanced Fire Detection. Heat and Mass Transfer in Fire and Combustion Systems, HTD-vol. 223, pp. 1-10, ASME, December 1992
5 Serio, M A Bonamno, AS, Knight, KS, and Newman, JS. Fourier Transform Infrared Diagnostics for Improved Fire Detection Systems, NIST Annual Conference on Fire Research, Gaithersburg, USA, 1996
Are we reaching the limits of existing fire detection technology? Mohamed Cherif Benzerari thinks we may be, and here he sets out an alternative way of quickly and accurately detecting genuine fires by using mass spectrometry techniques.
Mass spectrometry technology may be an alternative and effective method of smoke detection, since it can precisely distinguish molecular compounds of smoke from other particulates not generally produced by fire. The introduction of a unique ID for each material, fixture or product in a building will add substantial sophistication to an addressable fire alarm system by not only locating a fire, but also assessing the burning material in question.
Advances in microelectronics, information technology and fire science have led to substantial progress in fire detection technologies over the last decades. Although unwanted alarm rates have, according to statistics, fallen slightly, it is widely recognised that more progress has to be made.
Various new technologies such as video smoke detection are available, while existing ones such as multi-criteria fire detectors based on advanced signal processing algorithms, have been further developed. Yet what we think of as ‘state-of-the-art’ fire detection technology may be out of date in a few years, if it cannot provide high sensitivity with false alarm management in a cost-effective package. To put it bluntly, existing ionisation and optical smoke detection technologies have been stretched to their limits, and their future development may be subject to diminishing returns. A situation may soon arise where any marginal benefits to be had from these technologies may be outweighed by the investment needed to produce them.
The main challenge of fire detection is that it covers a wide range of scenarios, from smouldering to flaming combustions, and from areas such as a small lift motor room to very large multiple occupancy buildings. Unfortunately, there is no single standard solution that can be applied to that wide range of fire scenarios. Basically, solutions are formulated for a particular fire scenario in a given space, and do not always take into account environmental factors such as corrosion, humidity and wind.
Smoke detectors sense the airborne particulates in flaming or smouldering combustions, either by point detection or, in the case of aspirating systems, drawn through ducts into a sensing unit. But particulates in the air can be made up of dust, powder, steam and fog – as well as from smoke. In the case of optical sensors, scattered light beams measure any increase in the percentage of obscuration, while ionised air type detectors measure any reduction in current.[1] Those variations are measured as a proportion of the total amount of particulates entering the sensing chamber, regardless of the sort of particulates they are. So the fire detector in this case works just like a basic lift monitoring system: when it senses an equivalent weight of, say, 16 people, it generates an overweight warning signal, regardless of how this weight is made up.
There are a number of reasons why these existing sensing methods will soon be out of date:
– Ionisation detectors, which are usually the most suitable to detect the smaller particulates produced by rapidly growing flaming fires, have been banned from some countries due to their radioactive source (Americium 241). There are also severe restrictions on their storage and transportation, which means their use is declining rapidly.[2]
– Optical detectors are more suitable for detecting the larger particles of slow smouldering fires but conversely, they have difficulties in detecting flaming fires with small particles. In addition, non fire-generated particles, such as dust and other nuisance aerosols, may still cause false alarms.
– 95% of deaths in fires are caused during the smouldering phase of a fire, and while the victims where sleeping, while 80% die of toxic smoke inhalation.[3] Both fire detection methods are unable to detect poisonous gases such as carbon monoxide.
Certainly, signal processing algorithms have advanced the intelligence of these traditional methods by incorporating heat sensing into the same smoke detector. But in terms of early detection, the criterion of ‘heat’ comes a bit late in the majority of flaming and smouldering fires. A medium-term solution for this paradox is the use of a combined CO/optical smoke detector developed with advanced signal processing algorithms. This is much more efficient than a combined heat/optical detector and, when necessary, can be upgraded with infra red and/or heat detection capabilities. But for a longer-term approach, a radical solution based on detecting the compounds at a molecular scale (nanoscale) – produced exclusively from a fire signature rather than any other extraneous sources – should be considered.
From particulates to compounds
Mass spectrometry is a well proven and reliable technology, having been applied successfully during the last few decades in many advanced areas of research and application. These include satellites and spacecraft, biotechnology, biochemistry, geology and medical equipment. Incorporating this advanced technology within the future fire detection is fundamental to better fire detection.
A mass spectrometer is a sensitive and an accurate detector of molecules; it can even weigh the molecular mass of any element with an accuracy of 0.01%. With advanced software such that based on Fourier Transform Infra-Red (FTIR)[4, 5] and chromatography process, it can even trace quantities of contaminants and toxins such as CO, CO2, and HCN (hydrogen cyanide), which may be harmful to humans if their threshold levels are exceeded within the surrounding environment. The mass spectrometer has three essential functions or stages: the ion source, mass analyser and the detector (figure 1):
– The ion source In order to measure the characteristics of individual molecules in an air sample, a mass spectrometer converts them to ions, usually cations (positive ions which are molecules or atoms with one electron taken away), so that they can be moved about and manipulated by an external electromagnetic field.
– The mass analyzer The ions are sorted and separated according to their masses and charges. By this means, lighter ions are deflected more than heavier ions and so are separated from each other.
– The detector The positions of separated ions on the detector are a function of their masses; results are displayed in a spectrum graph style, ready to be analysed, as in figure 2.
Application to fire detection
A mass spectrometer based fire alarm would have to be designed and manufactured using an aspirated system architecture style, where the processing unit is extended by ‘sniffers’ or ‘perforated ducts’. In this way, early detection is achieved by drawing the air samples to the processing unit, instead of passively waiting for the air samples to reach the sensing unit, as in the case of a point detector. Artificial intelligence software would manage the comparison of the spectrum of the sensed compounds against its vast spectral database.
Secondly, every liquid and solid has a unique chemical representation. But once a combustible material gets into contact with enough heat and oxygen, ignition takes place. During the very early stage of the material’s combustion, the first quantity of its molecules undergo an initial chemical reaction and are transformed into gases, with new chemical forms and/or representations. This process enables us to set out a unique Material Fire Code (MFC) for each material or furniture in a building. The code could be considered as the ‘DNA’ of the combustible material or furniture, in terms of types and proportions produced during the very early stage of combustion.
Each manufactured material or furniture would be sent to a test centre, where it would be burned, tested and analysed. It would then be labelled with its own MFC in the form of a barcode, describing its unique combustion compounds. The code can then be scanned by the fire alarm system once the material is purchased, placed or fitted in a building; the fire alarm system should then be able to recognise its chemical or gas compounds in the first few seconds of combustion. In order to achieve this, however, the fire alarm system must have sufficient memory capacity to record a full spectral database of all furniture and materials in a building. In this way, the artificial intelligence embedded in a fire alarm system should be capable of sensing the various species of gases relating to the burned furniture or material, and of generating a range analogue values rather than just one value. This would provide information not only that the fire threshold level is reached, but also additional information about the burned material in question, the types of toxic gases, and the location of the combustion within the room.
One drawback with this approach is that it’s likely to be efficient only in the very early stages of a fire. Once the fire reaches a larger size, it wouldn’t be possible to decipher the mixture of all compounds with their proportions, each in relation to its original combustible material.
In conclusion, standards need to be a key part of the process, and manufacturers of all sorts of combustible materials and furniture will have to collaborate with the fire industry to achieve this. If done in this way, manufacturers of fire alarm equipment would be able to produce such equipment at economically viable volumes.
Certainly, the introduction of such coding may seem a long and tedious requirement, and possibly may be considered over-engineered for many low risk premises. So further research, testing and analysis has to be carried out to test the viability of the this approach, then to apply it initially to high tech applications. But the benefits of early and accurate detection, together with a brand new tool for post-fire investigations, are certainly imaginable.
Mohamed Cherif Benzerari Elec. Eng TMIET is a commissioning engineer and software developer at Drax (UK).
References
1 Roger Barrett, Fire detectors, FEJ & FP page: 55, June 2004
2 Serio, MA, Bonamno, AS Knight, KS and Newman, JS. An FT-IR Based System for Fire Detection, NIST Annual Conference on Fire Research, Gaithersburg, USA, 1994
3 The detection principles of IQ8Quad, Esser, Art.No: 797989.G0, p8, December 2007.
4 Grosshangler, WL, An Assessment of Technologies for Advanced Fire Detection. Heat and Mass Transfer in Fire and Combustion Systems, HTD-vol. 223, pp. 1-10, ASME, December 1992
5 Serio, M A Bonamno, AS, Knight, KS, and Newman, JS. Fourier Transform Infrared Diagnostics for Improved Fire Detection Systems, NIST Annual Conference on Fire Research, Gaithersburg, USA, 1996