A statistical method for flow pattern and efficiency prediction using pressure drop in a two-stage gas liquid cylindrical cyclone

An experimental investigation is conducted in a novel gas-liquid separator called Two-stage Gas Liquid Cylindrical Cyclone (TGLCC) with gas-liquid ratio of 25-600, which enables flow pattern observation and separation efficiency/pressure drop measurement. The experimental results show that three inlet flow patterns are observed in the vertical pipe before the TGLCC inlet, which are continuous flow, intermittent flow and annular flow. The presence of vertical pipe before inlet can affect the evolution process of inlet flow patterns, and both intermittent flow and continuous flow have 3 forms of evolution process. Based on the experimental results, a statistical method called box plot is used to characterize the pressure drop fluctuation for different flow pattern evolutions, including the maximum, minimum, percentiles, mean and median. Based on pressure drop data analysis relating to corresponding flow patterns, a statistical method of predicting flow patterns is proposed using 5 criteria obtained from the box plot analysis. For all experimental data, the prediction accuracy is 98%. Then the separation efficiencies of gas riser and drain pipe are analyzed, which is found to be heavily affected by flow pattern evolutions. Therefore, the statistical method is further used to predict the separation efficiency of TGLCC with accuracy of 93.4%.

Automated summary

An experimental investigation on a novel gas-liquid separator called Two-stage Gas Liquid Cylindrical Cyclone (TGLCC) is conducted to observe flow patterns and measure separation efficiency/pressure drop. The presence of the vertical pipe before the TGLCC inlet affects the evolution process of inlet flow patterns. A statistical method using box plot analysis is proposed to predict flow patterns based on pressure drop data, with an accuracy of 98%. The analysis also shows that the separation efficiency of TGLCC is heavily influenced by flow pattern evolutions, and the statistical method predicts the separation efficiency with an accuracy of 93.4%.