With the development of supercritical fluids technology, more and more attention is being paid to the study of flow and heat transfer under supercritical conditions. In order to better understand the influence of gravity on the flow and heat transfer of supercritical fluids, with the CFD software FLUENT 6.0, a numerical study has been conducted for the developing laminar flow and heat transfer of water in a smooth vertical tube under supercritical pressure. The flow and heat transfer have been studied with the inlet Reynolds number ranging from 500 to 2000, Grashof number er ranging from1

在线识别的关键问题在于如何利用最短时间历程的参数波动过程完成由参数波动的特征空间到流型空间的映射。 CPN神经网络能够对自组织映射的结果进行有导师的训练，因此可以为流型的自动、客观和在线识别提供有效的手段。文中结合压力波动过程的快速傅立叶变换系数，对U形管垂直上升段内空气-水两相流的流型进行自动识别，在线性指标达到了8.2s。对如何提高流型快速识别的在线性指标和识别率等作了相关的探讨。

实验研究了螺管内绝热和沸腾汽液两相流动时压力波动特性，分析了热流密度对压力波动的影响规律。结果表明沸腾时波动的均方根和分形参数（分维数、关联维数和Kolmogorov熵）与绝热两相流明显不同，加热热流密度的大小影响各特征参数的大小及其随干度的变化规律。未考虑流密度影响的流型图或流型转变准则只能在小热流密度时使用。

基于油气水多相流混合物在管道内流动时的多传感器信号——压力和压差信号，建立特定流型的识别规则，采用学习矢量量化模式分类器作为未知流型的分类器，根据数据融合的技术思路获得了油气水多相流的流型在线识别技术。通过系统集成的手段，利用微处理装置研制出了流型识别系统。该识别技术具有置信度高、容错性好、性能稳定、降低了对单个传感器的性能要求等优点。该识别系统结构简单、无运动部件、不改变管道的截面结构，可以处理非快速变化的瞬态流动，测量误差小于10%，并具有连续工作、定时打印、信号输出等功能。

In the present investigation, both pressure fluctuations and differential pressure signals were measured for oil-gas-water multiphase piping flow in wide ranges of flow parameters in horizontal pipe. The signal features were extracted and analyzed, which included time domain features, frequency domain features and fractal features. Both liquid velocity and oil fraction have tremendous influence on the feature analysis, which all the feature have bad clustering and bad separability for the same flow pattern or different flow pattern respectively. It is proved that the recognition methods, RMS of the pressure, PDF or PSD of differential pressure fluctuation, are not able to be employed widely in the industrial applications. The pressure fluctuations exhibit chaotic characters and are influenced by flow patterns. The regulations to identify the special flow pattern are obtained basing on the joint feature analysis. The online intelligent recognition of flow patterns is proposed which is a fusion method combining the special regulations and the pattern recognition. The proposed method was examined in the oil-gas-water loop and the results showed that bubble flow, stratified flow, intermittent flow and annular flow could be identified with more than 90% accuracy.