American Insitute of Mining, Metallurgical, and Petroleum Engineers, Inc.
Although computer solution technique are used universally in reservoir simulation and history matching problems, in the context of well testing one finds an almost exclusive reliance on the conventional graphical techniques (build-up and drawdown plots and type curves). While these graphical methods have been remarkably successful in providing quick answers and valuable providing quick answers and valuable insight into the reservoir structure, they are somewhat limited in scope and impose some strong restrictions on the test schedule. The practical difficulties in applying these methods are that the assumptions on which they are based do not often hold, the flow rates are variable or even unknown during certain phases of the test and finally the estimates of the parameters obtained from different methods may vary widely. It is true that the Horner-buildup and drawndown analyses can be extended to accommodate variable flow rates by a judicious application of the superposition principle to asymptotic analytical solutions [Ridley]; however, the information that can be derived pertaining to the damage near the wellbore is far from complete or accurate. Indeed all that can be inferred about well-damage is in terms of a skin factor and not whether this "skin" is really due to a plugging of the perforations or due to plugging of the perforations or due to mud invasion into the formation and how far into the formation that the damage persists. None of these methods operate satisfactorly if flow data are missing; moreover the assumption of a constant flow rate in place of missing information on flow rate may in fact lead to misleading results. If one has to use type curves, the results are often ambiguous as they are based on very subjective judgements on what is or is not a matching overlay between data and the type curve. In spite of the existence of myriads of such methods, when it comes to the simultaneous estimation of several reservoir parameters, both local and far from parameters, both local and far from the well, there is no unifying procedure that produces a unique answer. This stems partially from the fact that most of these methods are based on the long-time asymptotic solutions of the radial flow equation.