Fire detection FAQs

A Phase 1 alert is not necessarily an indicator of a fire. When a Silvanet sensor triggers a Phase 1 alert (yellow), it should not be consider a problem. A phase 2 alert (red), however is considered a problem. Phase 1 only states that there is a change in the gas resistance. Phase 1 alerts are used as an optimization to reduce the number of gas scans (which check for fire) and only do that when there is a change in the gas composition of the air.

A Phase 1 alert is used to trigger up to 60 consecutive gas scans which is the real test for a fire. However since gas scans consume a lot of energy, we try to reduce them to a minimum and only start gas scans if there is a change in gas resistance (which indicates some kind of smell, indicated by a change in the gas composition of the air).

A Machine Learning (ML) model is a program that has been trained to recognize certain types of patterns. You can train a model over a set of data, providing it an algorithm that it can use to make predictions and learn from those data. ML models find patterns and make decisions in previously unseen data and make predictions about that data. A ML Model recognizes patterns by having it “trained” with large datasets. Training involves optimizing the machine learning algorithm to find certain patterns or outputs from the dataset. The output - a ML Model - is a computer program with specific rules and data structures.

A VOC (Volatile Organic Compound) is an organic chemical compound that evaporates easily at room temperature. They vaporize into air and dissolve in water, for example smoke from fires. Organic compounds are chemicals that contain carbon and are found in all living things, especially trees. VOCs are also emitted from, for example, oil and gas fields and diesel exhaust which is why the sensors use ML Models to distinguish between VOCs from fires and VOCs from other sources.

The US EPA (Environmental Protection Agency) defines VOCs as follows:

Volatile organic compounds (VOC) means any compound of carbon, excluding carbon monoxide, carbon dioxide, carbonic acid, metallic carbides or carbonates and ammonium carbonate, which participates in atmospheric photochemical reactions, except those designated by EPA as having negligible photochemical reactivity.

The most important measure of fire behavior is fire intensity. Fire intensity (I) represents the heat released per meter of fire front (kW/m of fire front). It is a function of (1) heat yield of fuel (kilojoules/kg), (2) amount of fuel per unit area (kg/m2) and (3) the rate of forward spread of fire front (km/h).

When there is a wildfire directly where the sensors are, they will, of course, be destroyed by the flames and need to be replaced. However, they will continue to run fire detection processes up to the point when they are destroyed.

In general, after a wildfire has been detected by our sensors and firefighting occurs in the area of the sensors, the sensors in the area should be replaced. Also, if the sensors that have been massively soaked in chemicals, such as to extinguish fires, they are most likely going to be contaminated and should be replaced. The sensors may continue to work, but to ensure they function, we recommend replacing them after fire fighting.