Study On The Characteristics Of Light Scattering For Fire Smoke And Non-fire Particles

Smoke is one of the most important characteristics of early fires. Photoelectric smoke detection technology, based on the principle of light scattering of particles, has been widely used in the field of fire detection. It is important to research light scattering from fire smoke particles by numerical calculation. Traditionally, spherical or spheroid models were used to approximate the shape of smoke particles for light scattering calculations. But actually, smoke particles have a similar fractal structure, which is different from spherical structure. Using the discrete-dipole approximation method, the light scattering Muller matrices were computed for the randomly oriented fractal aggregate, as well as the spherical particle with the same volume of aggregate, and then both normalized Muller matrix elements were compared. The results indicate that the angle distributions of the normalized matrix elements F1 1 (θ) / F11(0),F2 2 (θ) / F11(θ) and F3 4 (θ) / F11(θ) have significant differences between the fractal model and spherical model. We can discriminate spherical non-fire aerosols from smoke particles, using the difference of F2 2 (θ) / F11(θ) between fractal model and spherical model.Because of only sensing the intensity of scattered light by smoke particles for fire alarm, photoelectric smoke detector is often interfered by non-fire smoke particles, such as water vapor, dust and other nuisance. Presently, most study on light scattering from smoke particles are focused on single aggregate. In fact, the smoke has a similar fractal structure. The number of primary particles approximately obeys log-normal distribution, and aggregates are in random orientation. Calculation of light scattering from particles must be averaged statistically in aggregate sizes and spatial orientations. Using the discrete-dipole approximation method, the scattering Muller matrices were computed for the randomly oriented smoke particles with a certain size distribution, as well as the dust particle. Both normalized Muller matrix elements were compared and analyzed. The results show that there are significant differences between the angle distributions of the normalized matrix elements F1 1 (θ) / F11(0), F2 2 (θ)/ F11(θ), F3 3 (θ) / F11(θ), F4 4 (θ) / F11(θ), F1 2 (θ) / F11(θ) and F3 4 (θ) / F11(θ) for smoke particles and dust particle. Matrix elements F2 2 (θ) / F11(θ) can be easily measured in experiment, so we can discriminate dust from smoke particles, using the difference of F2 2 (θ)/ F11(θ) between them. It could be an effective way to reduce false alarm rate of photoelectric smoke detectors.