SOUZA, A. O.; http://lattes.cnpq.br/4289620474946265; SOUZA, Andrey Oliveira de.
Résumé:
For many years the risk of explosion and consequent legal requirements for area classification has been a global concern. The Brazilian standard for area classification by explosive atmospheres is a faithful translation of the international standard. Many authors defends that these standard has not a scientific basis, what can causes many mistakes, because of excess or a false impression of safety. Therefore, the present work has as objective to develop a reliable alternative to area classification. For that, was developed a CFD model, by software ANSYS CFX 16.1, validated and parameterized to a great interval of sonic leak. To set random leak conditions to be simulated, it was used a statistical technique of multidimensional sampling (Latin Hipercubo), varying pressure and temperature of reservoir, orifice diameter, gas properties and leak direction relative to gravity and wind. The results show that the calculation domain should be parameterized in 8 meters length for millimeter in diameter from the source of release. The mesh must be parameterized in elements 50 along the hole diameter, while maintaining the hexahedral structure throughout the domain. Gravity proved not interfere in the extension and volume results of explosive atmosphere in sonic leaks. The deviation from ideality at reservoir conditions observed by applying the equation of Soave Redlich Kwong also not significantly influenced the extension and volume of explosive atmosphere. The analysis of simulations of random leaks conditions allowed the development of simple and practical integral equation for reliable determination of explosive atmosphere extension. Consideration of wind direction in the simulations show that the volume of the explosive atmosphere could not be directly related to its length. Finally, the relationship between the hypothetical volume of explosive atmosphere and its extension is applied as a criterion to define the risk of ignition in a proposal for a more reliable area classification, which takes into account the effects of dispersion.