A Statistical Evaluation of Methods Used To Predict Pressure Losses for Multiphase Flow in Vertical Oilwell Tubing

Abstract

Several methods for predicting pressure losses for upward, concurrent, continuous, multiphase flow in vertical oilwell tubing were tested against measured pressure losses from 726 well tests. This comparison of well known pressure-loss correlations reveals the relative strengths and weaknesses of each correlation and should be useful in selecting satisfactory methods for various applications. Introduction Several correlations have been published that can be used to predict pressure losses in vertical oilwell tubing for the simultaneous, upward, concurrent, continuous flow of oil, water, and gas. Because of the extreme complexity of multiphase flow, the proposed correlations are by necessity highly empirical. The validity of the correlations is then somewhat limited to the quality and scope of the data upon which they are based. Therefore, some correlations perform quite well for cases in the range of the data used in developing the correlation but fail for other applications.The purpose of this study is to ascertain the accuracy of several pressure-loss prediction methods in terms of flow variables familiar to the practicing engineer. The results of the study should assist the petroleum production engineer in selecting the most petroleum production engineer in selecting the most accurate method for his problem and also should indicate the general accuracy to be expected from the methods.The correlations included in the study are those of Poettmann and Carpenter, Baxendell and Thomas, Poettmann and Carpenter, Baxendell and Thomas, Duns and Ros, Fancher and Brown, Hagedorn and Brown, and Orkiszewski. Each of these correlations was proposed specifically for predicting pressure losses in vertical oilwell tubing for the upward flow of multi-phase well fluids.The pressure-loss prediction methods were programmed for the IBM 360 computer and tested programmed for the IBM 360 computer and tested against 726 well tests from field and experimental wells. A statistical analysis was made to find the most acceptable method for different ranges of flow variables and also the method having the best over-all performance for predicting the measured pressure performance for predicting the measured pressure losses. Pressure-Loss Prediction Correlations Pressure-Loss Prediction Correlations The six correlations chosen for study will be discussed briefly. A review of the history and content of multi-phase flow literature and a more detailed discussion of individual correlations may be found in Brown.The correlation of Poettmann and Carpenter is a relatively simple and practical one to predict pressure losses in flowing wells. However, its very pressure losses in flowing wells. However, its very simplicity limits its accuracy. The method was included in this study for comparison, and because it served as a starting point for later correlations.Poettmann and Carpenter correlated the irreversible energy losses of 49 well tests with a Fanning-type friction term. The friction term was related to the numerator of the Reynolds number for the well fluid mixture. No attempt was made to account for liquid holdup (volume fraction liquid in a pipe section), but rather an average density of produced fluids corrected for down-hole conditions was used. The correlation reproduced the measured pressure gradients to an average deviation of 1.8 percent and a standard deviation from the average of 8.3 percent. JPT P. 903