| Gas-liquid two-phase flow widely exists in many industries, such as chemical, pharmaceutical, petroleum, energy and power engineering, etc. Online measurement of flow pattern and void fraction are very important for operation monitoring, process controlling and safety assurance of gas-liquid two phase flow system. Due to the complexity of two-phase flow, flow pattern identification and void fraction measurement are two difficult problems which have not been well solved both in scientific research field and industrial application. Capacitively Coupled Contactless Conductivity Detection (C4D) is a new conductivity detection technique, which may provide a promising way for parameter measurement of gas-liquid two-phase flow. This dissertation aims to study the feasibility of applying C4D to flow pattern identification and void fraction measurement of gas-liquid two-phase flow, and then based on C4D and combine with modern information processing technique, to propose new methods for flow pattern identification and void fraction measurement of gas-liquid two-phase flow.The main innovation points and contributions of this dissertation are listed as follows:1. Based on series resonance principle, a new C4D sensor which is suitable for parameter measurement of gas-liquid two-phase flow is developed. Research results show that:based on series resonance principle, the new C4D sensor overcome the unfavourable influence of the coupled capacitances on conductivity detection, improves the detection range and resolution, and expands the application of C4D to conductivity detection in millimeter-scale pipes (the maximum inner diameter of the new C4D sensor is7.8mm, while conventional C4D technique are mainly used in capillaries). Research results also show that the conductivity detection accuracies of the new C4D sensor in larger pipes are satisfactory. In four pipes (the inner diameters are1.8mm,3.3mm.5.5mm and7.8mm, respectively), the maximum relative errors of conductivity detection are all less than3.5%. The new C4D sensor has the advantages of wide detection range, high resolution, simple construction, good stability and anti-interference ability, and provides an effective mean for parameter measurement of gas-liquid two-phase flow.2. Based on the conductivity signals obtained by the new C4D sensor, flow pattern identification of gas-liquid two-phase flow are studied combining with the modern information processing technique. According to different feature extraction methods, two new flow pattern identification methods are proposed and compared:1) use statistical analysis and Fourier analysis methods to extract the features of conductivity signals, then input the extracted features to the flow pattern classifier (which is built by SVM technique) for classification, finally the flow patterns are obtained.2) use statistical analysis and Wavelet analysis methods to extract the features of conductivity signals, then input the extracted features to the flow pattern classifier (which is built by SVM technique) for classification, finally the flow patterns are obtained. Research results show that applying C4D to flow pattern identification of gas-liquid two-phase flow is feasible, and the proposed two flow pattern identification methods are both effective. In five pipes with the inner diameters of1.8mm,2.8mm,4.0mm,6.1mm and7.8mm, respectively, the identification results for five typical flow patterns are all satisfactory. The flow patten identification method based on the combination of statistical analysis, Wavelet analysis and SVM techniques are better (The identification results using the combination of statistical analysis. Wavelet analysis and SVM techniques are all above91%, while the identification results using the combination of statistical analysis, Fourier analysis and SVM techniques are only above87%.). Research results also show that combining the feature extraction and the SVM technique is effective for flow pattern identification. The extracted features can effectively reflect the information of flow pattern, and SVM technique can succeffully implement the multi-classification of flow patterns.3. Based on the conductivity signals obtained by the new C4D sensor, a new method for void fraction measurement of gas-liquid two-phase flow is proposed combine with the SVM regression technique. This method develops the void fraction measurement model for each typical flow pattern on the basis of SVM regression technique. In practical measurement process, firstly, the conductivity signals of gas-liquid two-phase flow are obtained by the developed C4D sensor, then, the flow pattern of gas-liquid two-phase flow are identified, finally, according to the identification results, the suitable void fraction measurement model is selected to calculate the void fraction value. Research results show that applying C4D to void fraction measurement of gas-liquid two-phase flow is feasible, and the proposed void fraction measurement method is effective. In the pipe with the inner diameter of7.0mm, the void fraction measurement results are satisfactory. For stratified flow (wavy flow), bubble flow, slug flow and annular flow, the maximum measurement errors are all less than7.0%. Research results also show that using the combination of feature extraction and SVM regression technique to develop the void fraction measurement model is successful. The extracted features (the mean value and the standard deviation) of conductivity signals can effectively reflect the information of void fraction, and the SVM regression technique can overcome the nonlinearity between the conductivity signals and the void fraction, and can successfully develop the void fraction measurement models. |
PDF Full Text Request