ACKNOWLEDGMENTS The authors express their gratitude to Robert F. Holub, Bureau of Mines, Denver Mining Research Center, for his assistance in developing the mathematical prediction techniques in this paper. GENERAL CONSIDERATIONS When a fire breaks out in a mine, smoke can become a major hindrance in the safe evacuation of operating personnel. The thick smoke that fills the mine openings induce a visual "whiteout" and reduce visibility to near zero. Men trapped in these smokes may wander aimlessly, eventually succumbing to confusion, exhaustion, or panic. No determinations have been made as to visibility requirements for escaping from fire through smoke. It is generally believed, however, that a visibility of 15 to 25 meters is necessary to escape from fires in such places as department stores and underground shopping streets (6). A visibility of 3 to 5 meters is adequate with people well acquainted with escape routes. Probably, these minimum visibility requirements are also valid for mine fires. Because visibility in the thick smokes developed in mines is too low, both escape from the mine and quick, effective rescue are difficult. Rescue teams that enter the mines must work through the same smoke that prevents escape. Some instrument or technique is needed whereby better visibility through smoke can be obtained. Efforts to achieve this increased visibility by means of more powerful lights have generally failed because the light must travel through the smoke, and scattering of the light by particulate matter in the smoke prevents visual observation of desired targets. Infrared imagers were evaluated for use in smoke-filled openings; their use does not require a lamp to illuminate the target. Rather, they receive only the infrared energy (radiant energy) emitted by the target in the spectral range of the imager. By eliminating the use of a lamp, the travel path between the observer and the target becomes one-way, and transmission problems are reduced. In addition, scattering of the radiant energy by particulate matter is reduced because the infrared wavelengths are longer and are less affected by the small particles of the smoke. INSTRUMENTATION Two different imagers were used during this investigation. Both instruments develop images from the electrical analogs of radiation in their fields of view, providing a thermal picture of the targets observed. Both have a spectral region of 3 to 5 micrometers (3 to 5 um). Although these instruments have been described in detail in other papers (12), a brief description of each is presented for the reader's convenience. Both are approved by the Bureau of Mines for use in coal mines, and both were capable of detecting men in dense smoke. The first of these instruments is small in size (3.5 by 5.5 by 10 inches), lightweight (6 pounds), and powered by a battery attached to the operator's belt. Consequently, it is extremely portable and normally can be used in any location accessible to the operator. Thermoelectrically cooled, lead selenide detectors and associated electronics provide a system sensitivity of less than 0.2° C. A real-time image is developed using light-emitting diodes for the display. Design specifications require that the imager detect a man at 50 meters (in clear air) with sufficient resolution that the target, when viewed on the display, can be identified as a man. In actual usage, the instrument exceeds this requirement. Figure 1 is a photograph of this instrument. The second instrument used in the study also has excellent portability. It, too, is small in size (4 by 9 by 8 inches) and has a total operational weight of 7.5 pounds. The battery power supply for this instrument is inside the case, thereby eliminating external cables and increasing convenience in operation. Indium antimonide detectors are used with a 10-sided rotating mirror (and associated electronics) in a 10 to 1 interlace with light-emitting diodes to form a real-time image. Rated detector sensitivity is less than 0.2° C; the detectors are cooled by expansion of argon gas. Figure 2 is a photograph of this instrument. Once the capability of infrared imagers to see through coal smoke, at least to some degree, was established, a mathematical analysis was made (1) to determine whether or not serious range limitations were created by the smoke and, if so, (2) to develop a technique for predicting the effectiveness of infrared techniques in individual applications. A literature search revealed that data pertaining to the size, shape, numbers, composition, and distribution of particles in smoke from combustion of in situ coal were not available. In addition, variations in coal volatile content, sulfurization, ash content, and natural moisture would influence the amount and nature of Nevertheless, a quantitatively broad solution to the the smoke generated. "smoke vision" phenomena was developed by using average values based on measurements of smokes from coal-fired powerplants. Decrease in visibility due to the presence of smoke is the result of energy attenuation caused by the scattering and adsorbing characteristics of the interference medium (smoke). Feldman and Coy (4) have shown that when particulate diameters are at or near radiation wavelengths, scattering (Rayleigh) is most pronounced. However, Jamieson and others (5) have shown that Rayleigh scattering is unimportant at wave lengths greater than 1.5 μm, and Rayleigh scattering can be disregarded in dealing with wavelengths of 3 to 5 μm. The mathematical analysis theory best adapted to the spectrum and particle sizes associated with this investigation is the "Mie" theory; it is suitable for evaluation of the synergetic effects involved in this study. For the purposes of calculations that would aid in overall apparatus evaluation, a "simplified" smoke was created on paper. This smoke vehicle, representative of a synopsis of available data, was as follows: 1. The size distribution of the smoke was assumed to be log-normal (3); this parameter was used as best representing the polydispersed system under study. 2. The mean geometric particle radius was assumed to be between 0.5 and 1.0 um.4 3. The particulates would be unburned carbon with adsorbing hydrocarbonic volatiles. These coated carbon particles would tend to agglomerate to "soot" particulates, but subsequent mathematical calculations would ignore these latter phenomena (3). 4. CO2 would comprise no more than 12 wt-pct of the interfering particulate mass. 5. Fly ash would comprise no more than 5 wt-pct of the interference medium (1). 6. The refractive index of the smoke particulates was taken to be 2.00.661, this being the index of amorphous carbon. 7. The authors recognize the existence of coatings (tars, etc.) on particles in the smoke. Because their effect on the optical properties of the smoke is unknown, they were not included in the calculations. Calculations Pilat and Ensor (3, 8-9) have derived a "smoke plume opacity" equation that has found widespread application to the transmittancy of visual light *Recent work done under the supervision of Joseph Singer and Martin Hertzberg at the Bureau of Mines Safety and Research Center in Pittsburgh, Pa., indicates that the average particle size in mine coal smoke may be slightly smaller than the 0.5- to 1.0-um range. If so, scattering affects should be less severe. The 0.5- to 1.0-um assumption provides a safety factor; visibility in actual mine conditions should be at least as good as, and probably better than, that derived from mathematical analysis. through smokestack emissions. This electromagnetic radiation transmission function was used in determining the infrared transmittancy through coal smoke. This function was as follows:. and W = mass concentration of particles, in grams per cubic centimeter, = particulate density, QE = particle light extinction efficiency factor, the total light flux scattered and absorbed by a particle divided by the light flux incident on the particle, W, the mass concentration, was calculated for the analysis of infrared through coal smoke (2). The form of the W function is. where K = specific particulate volume/extinction coefficient ratio, = = intensity of incident light, length in meters in which the entire white light of a miner's hatlamp is 99 pct obscured. For emergency situations, where rapid determination of relative mass density of the interfering smoke is needed, the rescue team need only pace the distance into the smoke necessary for the light from a miner's cap lamp to be obliterated. This paced distance (approximately 1 meter/pace) will be sufficient for reference to figure 3 for the infrared transmittancy. QE is the light extinction efficiency factor for a single particle; it is a function of the wavelength of the transmitted light, the radius of the particle and the refractive index. The coefficient is a series calculation involving a spherical Bessel function with a complex argument. This portion of the work was handled by an IBM subroutine by Dave (2). QE's in agreement with Dave's were achieved, and the subroutine was incorporated into a larger computer program for the computation of the infrared transmittance. The mass |