| Owing to natural or man-induced factors, oil sludge or oily sludge is a by-product generated in numerous processes, such as oil production, transportation, storage, and refining. It is composed of water, oil, and suspended solids and forms stable yet complex water-in-oil (W/O) emulsions. According to its sources, the water content, the oil content, and the solids content vary in the ranges of30-85%,15-50%, and5-46%, respectively. Due to its high concentration of petroleum hydrocarbon and heavy metals, oil sludge is regarded as a hazardous waste and required non-hazardous treatment, while it is also a considerable potential energy resource, provided that the oil is recoverable technically and economically. In order to evaluate the economy of selected treatment methods and the economic value of oil sludge samples, it is necessary to have an accurate, rapid method for simultaneous determination of water and oil contents in oil sludge. Currently, the drying method, azeotropic distillation based on ASTM95, thermal analysis method, and volumetric Karl Fisher titration are employed for water analysis in oil sludge, while the Soxhlet method, spectrophotometry, and Dean-Stark azeotropic distillation are used for oil analysis. The deficiency with these methods is that it requires large amounts of solvents and is time-consuming. Moreover, the results provided by some of these methods are inaccurate. And the above two values are gained separately. With the advantages of accuracy, good repeatability, solvent-free, and time-saving, a non-destructive method, called low-field nuclear magnetic resonance (LF NMR) is proposed in this thesis for simultaneous determination of water and oil contents in oil sludge samples. The main contents and conclusions are summarized as follows:(1) In order to make the analysis of water and oil in oil sludge samples go well, it is necessary to have the prior information of sample physical-chemical characteristics. At first, the physical-chemical characteristics of oil and solids extracted from oil sludge samples are analyzed. The H/C molecular ratio of oil from oil sludge is comparable with those of vacuum residua, and is a little higher than those of oil sands bitumen. Due to the different sources, the content of saturates, aromatics, resins, and asphaltene has great variations.So do the uniformity coefficients of the solid particle size distribution. According to the XRD analysis, the main components of solids are calcium carbonate, quartz, and iron-contained oxides. The contact angle analysis indicates that the solids surface is water-wetted. The microscopic photo further verified that the water in the oil sludge samples exists in the state of W/O emulsions.(2) Different analysis methods of water in oil sludge, drying method, volumetric Karl Fischer titration, and Dean-Stark azeotropic distillation, are assessed in order to provide helpful suggestions for water analysis method selection. Furthermore, the optimum volume ratio of methanol and chloroform is determined. The model oil sludge with known mass of water, oil and solids is used to evaluate the feasibility of water and/or oil content determination by volumetric Karl Fischer titration and Dean-Sark azeotropic distillation. The experimental results indicate that the optimum volume ratio of methanol and chloroform is between2:1and4:1. Taking the accuracy of results by volume Karl Fischer titration and the best recovery rate of water, oil, and solids from model oil sludge samples, it can be concluded that the volume Kail Fischer titration is suitable for water analysis in oil sludge samples. And the Dean-Stark azeotropic distillation can be used as reference method for water and oil analysis in the future experiments. The drying method isn’t suitable for the determination of water content in oil sludge because of the volatilization of low boiling point of petroleum hydrocarbons and water together. The comparable results can be attained between Dean-Stark azeotropic distillation and volumetric Karl Fischer titration.(3) The problem of superimposed signals from water and oil in the T2distribution curves is solved by the addition of MnCl2·4H2O solution. According to T2distribution curves, the quantification methods for water and oil content in oil sludge by using LF NMR is built with the construction of calibration curves and the introduction of amplitude area index (AI). In comparison with reference method, Dean-Stark azeotropic distillation (AD), the good correlation (The correlation coefficient R2is higher than0.98) and acceptable errors (The standard deviation is less than2.9%) indicate that the results from LF NMR is identical to those from AD, and thus the accurate results can be achieved. In conclusion, the LF NMR is suitable for the determination of oil and water content in oil sludge samples.(4) Since99%of variables have intercorrelations higher than0.9for LF NMR data, the general calibration model for water and oil analysis in oil sludge was developed with full cross-validation on the data obtained by Dean-Stark azeotropic distillation as a reference. The results show that the correlation coefficients are R2=0.9657and R2=0.9785for both water and oil. And the root mean square errors of cross-validation (RMSECV) are2.73%and2.22%for water and oil. The calibration model for the sample HZ-OS is selected for verifying the predictability of different batches of samples with the same source. For validation set, the correlation coefficients R2=0.9141and0.9247for both water and oil. The root mean square errors of prediction (RMSEP) are1.85%and2.04%for water and oil, which are close to the values of the root mean square error of calibration (RMSEC)(1.29%and1.12%for water and oil respectively). This shows that the general calibration model including all kinds of oil sludge samples is steady and good predictability. What’s more, the models are built respectively for every kind of oil sludge samples. The results indicate that calibration is satisfactory with the correlation coefficients R2=0.9948and0.9940for water and oil, while the root mean squared errors of cross-validation (RMSECV) are0.78%and0.82%for the sample HZ-OS; for the sample ZS-B, the correlation coefficients R2=0.9907and0.9904, while the RMSECV is0.63for both water and oil; for the sample ZS-D, the correlation coefficients R2=0.9857for both water and oil, while the RMSECV are1.31%and1.29%.(5) The classification model is developed by using raw magnetization decay data obtained by LF NMR. The two dimensional scatter is plotted with the first and second principal components, and indicate that it is a good clustering analysis for classification varieties of oil sludge samples. The36randomly selected oil sludge samples from three varieties are used to develop classification models, while the rest of9unknown samples are employed to validate this model. The high recognition rate indicates that the classification models have good discriminant analysis ability. Therefore, the goal of rapid and accurate classification of oil sludge samples can be achieved by using LF NMR. |
PDF Full Text Request