Caracterização e aplicação de sensores à fibra ótica no setor petroquímico

Abstract

This work shows the applicability of long period gratings as transducers in the petrochemical industry. Results are presented in order to illustrate the phases involved in the fabrication, characterization, encapsulation and application of the long period gratings. Three long period gratings were produced using the point-by-point technique based on the ultraviolet irradiation. An encapsulation methodology was proposed to improve the long period grating robustness and versatility. The encapsulated long period grating was applied to the fuel blends and solvents analysis. It was shown that the optical fiber device sensitivity for changes in the external refractive index is not affected by the encapsulation. In addition to the extra protection to external damages added by the encapsulation, the device reproducibility was considerably improved. For the employed samples this improvements range from 40 % to 78 %. Finally the transducer performance was assessed in the analysis of ethanol concentration in ethanol-water mixture. It was found that the long period grating response to the ethanol concentration is non-linear and presents an ambiguous behavior to samples with ethanol concentration ranging from 60 % to 100 %. Four different methods based on artificial neural networks and fuzzy inference systems were proposed to determine the ethanol concentration in ethanol-water mixtures. The method with the best performance was based on artificial neural networks, and implemented in order to correlate the LPG response with the sample’s sound velocity obtained to samples with different ethanol concentration. The maximum deviation observed between the sample’s ethanol concentration and the value provided by the artificial neural network was 0,55%, with mean squared error of 0.13 %.