Research Of Measurement Method And Model For Multiphase Flow In Vertical Wells

In the process of crude oil extraction in vertical well, the three-phase flow phenomenon of oil-water-gas is widespread. Multiphase flow has its own special discipline compared with the single-phase fluid flow. Studying the flowing discipline of multiphase flow and the measurement methods of water content in the crude oil has important guiding significance on the production and development of oil. This paper, using the three-phase flow simulation well and the coaxial line water detector of daqing petroleum institute to collect data by configured the rate of oil, gas and water. Those collected data is the response of the coaxial line water detector, which is about the water content in the crude oil with three-phase flow characteristics of oil, gas and water. Through deeply research the collected data, this paper finally establish the prediction model of water content in the crude oil which has three-phase flow characteristics in the vertical well.Firstly, this paper proposed two kinds of experimental methods. The first One is processing the experiment by taking different flow rate in response to different water content. Considering the gas influence on the result of the measurement, the other one is using the bubbling umbrella and the common slip umbrella to process the gas-water two-phase experiments. In order to overcome the signal noise of the data caused by sensor of the coaxial line water detector, this paper using a 4-layer wavelet decomposition of sym6 wavelet de-noising method for de-noising the data processing. The MATLAB simulation results indicated that this method can effectively resolve the problem of signal noise interference on the result of experiment.Secondly, because of the non-linear mapping relation between the rate of water content and impact factors of crude oil, two models of predicting the rate of water content in crude oil which are based on BP neural network and WNN (wavelet neural network) respectively were proposed. The material arithmetic of the network is presented and by using this arithmetic the water content in crude oil is predicted. Results of the simulation in MATLAB indicated that the way of WNN has better convergent rate, prediction precision, learning ability and generalization ability than the traditional BP neural network. The method of WNN can predict the water content in crude oil with high precision and it owns much more powerful theoretical guide and much better application effects. At last, this paper analyzed the results of the experiment used bubbling umbrella and common slip umbrella in contrast. The comparative results indicated that the experiment which used bubbling umbrella has reduced the gas influence on the measuring result largely. At the same time, the thesis proposed the predict model of water content based on WNN by using the data which is got from the experiment used the bubbling umbrella. The result of simulation shows that the model based on the state of gas-liquid separation has better prediction precision than the model based on the normal state.This paper set up a forecasting model about crude oil based on wavelet neural network. This model provides a solution for the measurement of the water content in crude oil in the vertical well. This method meets the basic requirements of the measurement on oil-water-gas three-phase flow. It will have a broad application and research in the future.