Production log analysis and statistical error minimization

The objectives of this dissertation are to develop new production log interpretation methods and minimize the errors in interpreted profiles with statistical methods. After the investigation of all available standard production log interpretation methods, a new interpretation method for determining oil-water flow profiles was developed with temperature and flowmeter logs. Mass and energy conservation are the mathematical bases in the development of this new method. The main difference between the new interpretation method and the conventional two-phase flow interpretation method is that slip velocity and holdup are not needed, thus eliminating uncertainties in these quantities.; In conventional production log interpretation, more than one production log procedure and interpretation method exist in obtaining flow profiles. Errors from the log responses and interpretations are carried in these flow profiles. How to minimize the errors in flow profiles is the goal in the statistical study part of this dissertation work.; In this dissertation, two options for dealing with intermediate parameters are presented to minimize the interpretation errors: one option is to minimize the error for intermediate parameters before final interpretation, while the other option is to minimize the error of the flow profile after standard interpretation. Monte Carlo modeling has been used to investigate which option provides the smaller error in the final flow profiles. Hypothesis tests have been done to check whether the tool responses for down passes and up passes indicate the same variable. Statistical models for unbiased and biased data are developed to minimize errors of one variable represented by multiple data sets.; Error minimization for multi-pass spinner log interpretation has been developed with the application of two methods: conventional variance computation in linear regression and variance computation for the intercept and slope of a regression line. The later has been shown to be a better choice in variance computation of the final result of the flow profile since the slope does not have a significant influence on the values of variance of the flow velocity at each station compared with the former method.