Quantitative Analysis Of Hydrocarbon Gases Based On Genetic Algorithm Using Raman Spectroscopy

Gas logging plays an important role for the monitoring in the oil field. According tocertain cycle Gas logging measurements hydrocarbon gases which includ methane, ethane,propane, butane, isobutane, pentane and isopentane and other gases. The real-time speciesdetection and quantitative analysis of hydrocarbon gases dissolved in drilling fluids in theoil field can provide abundance information of stratigraphy, such as types of waterformation, the location of the oil-water boundary. Gas logging needs not only ofhydrocarbon detection, but also need to analysis various hydrocarbon composition.Ramanspectroscopy has more directly, rapid and simple advantages, Which identifise anddistinguishes different material from the molecular vibration spectrum of material. In thisthesis, To demand the hydrocarbon detection and quantitative analysis, some experimentalinvestigations into the Raman spectroscopy have been carried out with hydrocarbon gassamples in the laboratory, In order to solve the problem difficult to extract spectralinformation, genetic algorithm is used in the quantitative analysis.The thesis begins with an overview of Raman and its potential significanceinQuantitative Analysis of Hydrocarbon Gases Based on Genetic Algorithm. A detailedreview on the development of quantitative analysis using Raman is given in Chapter2toshow the basis of Raman, various methods used in quantitative analysis. The Chapter3isthe description of the relevant instrumentation and samples used in the experimentalinvestigations. The bulk of author’s contribution was to perform the results which aredescribed in Chapters4and5.Chapter4deals with the detection of hydrocarbon gases, different detection schemesutilizing laboratory platform and the local platform for methane, ethane, propane, butaneelemental gas have constructed. It was found that the Raman spectra characteristic ofmulticomponent hydrocarbon gases is too intensive and overlapping, which cannot contentthe needs of hydrocarbon gas detection in the local detection.532nm excitation wavelength was selected as the detection laser. With laboratory detection platform, quantitative analysisof hydrocarbon gases using Raman was studied.To address the exceptionally serious problem problem of Raman peaks overlap ofhydrocarbon gases, Genetic Algorithm for the quantitative analysis of multicomponentgases is given in Chapter5. In order to validate the performance of the quantitative analysisof genetic algorithm for Raman spectroscopy of multicomponent hydrocarbon gased,concentration tests of fitting mixtures is designed According to the actual concentrationratio of the fitting mixtures, Raman spectra of fitting mixtures makes up Raman spectra ofelemental gases. The results show that regardless of the Raman spectra of overlappingmixtures and non-overlapping mixtures, predictive results of Genetic Algorithm of fittingmixtures, which the error close to zero, has the high accuracy and good repeatability. It issuccessfully appliedd to quantitative analysis of Genetic Algorithm. For the Raman spectraof non-overlapping measuring mixtures, the obtained results suggested precision of therelative error of quantitative analysis of Genetic Algorithm is within13.38%. For the Ramanspectra of overlapping measuring mixtures, the precision of the relative error of quantitativeanalysis of Genetic Algorithm is within6.7%.The difference of fitting mixtures and actual measurement mixtures has direct influenceon the predictive accuracy of Genetic Algorithm. For the sake of minimize error, Ramanspectra preprocessing of wavelet transform is applied. The preprocessing Raman spectra ofwavelet transform import to the Genetic Algorithm. For the Raman spectra ofnon-overlapping measuring mixtures, the predicted of quantitative analysis results from13.38%to0.16%. For the Raman spectra of overlapping measuring mixtures, the relativeerror of preprocessin spectra of wavelet transform and original spectra has little difference.Finally, a summery of the work so far the author has done and some suggestions forpossible future development are given in Chapter6.