Flow regime prediction in air-conditioning units is of great importance for designing evaporator and condensers coils. Most current heat transfer and pressure drop predictions for two-phase flow lack accuracy mainly due to the ignorance of the effect of the flow regime. Because pressure drop and heat transfer are strongly related to two-phase flow regimes, objective and reliable flow pattern maps are needed as a strong basis. Therefore a capacitance sensor was developed for objective flow pattern identification based on the difference in dielectric constant of the vapour and liquid phase. The sensor was tested for air-water flow. Flow patterns were verified using high-speed digital video images. A multivariate analysis with many signal processing parameters was made for investigating the classification potential. A support vector machine was then built based on suitable parameters in amplitude and time domain, in order to statistically classify two-phase flows. A cross-accuracy of 92% was achieved and misclassification only occurs near flow regime transitions.
展开▼
