ABSTRACT
With modern computing capabilities, well test interpretation has evolved from simple straight-line techniques to a methodical interpretaion process making use of a variety of analysis procedures and complex well models. These analysis procedures range from engineering methodology to computer automation. Regardless of which procedure is used, however, there is a fundamental limit to the amount of information that can be extracted from a given well-test. This limit can only be overcome by using information obtained from other sources.
This paper uses a general buildup analysis procedure suggested by A.C. Gringarten as a framework to present an argument as to why there is a limit. The analysis procedure is based on three steps making a diagnostic plot of the data to select a well model, fitting the model to the data, and checking the fit for consistency with all the pressure data. The central point in the argument is if a well model matches a buildup and yields a correct pressure at infinite shut-in time, the model will be found to be consistent with all the pressure data. This results in two limitations which state when all consistency checks are not necessary and when well models cannot be distinguished from one another.
The results from the analysis of a North Sea oil well are presented to illustrate these limitations to well test interpretation. This analysis shows a comparison of results for two diferent bounded reservoir well models and a double-porosity model. Two of the models yield equally good matches of the buildup and drawdown data. Only by using other information about the well and reservoir could a choice be made between the two.
