Linear Fluid-Barrier Detection by Well Pressure Measurements (includes associated discussion and reply)

Abstract

Abstract In recent years considerable attention has been given to the use of unsteady-state pressure measurements in wells as a means of investigating thenature of petroleum reservoirs both near to and away from wells. These methodsare based on sound theory; for, theoretically at least, pressure build-up anddrawdown curves reflect the nature of the reservoirs from which the data aretaken. One of the more common and prominent features of reservoirs which can bedetected in some instances by pressure measurements is the linearfluid-barrier, e.g., a linear sealing fault. There has been some misuse of thistechnique in engineering practice, and it was therefore thought worthwhile todiscuss the criteria for valid pressure tests which indicate the presence andlocation of linear fluid-barriers. Introduction The present discussion will be concerned with the detection of linear fluid-barriers in ideal reservoirs, thus establishing optimum conditions fortheir detection in real reservoirs. The ideal reservoir considered is ahomogeneous, isotropic, porous bed of uniform thickness saturated with aslightly compressible fluid. Bounded on one side by a reasonably extensivelinear fluid barrier, the reservoir is assumed to be of such extent thatdiscontinuities other than the linear barrier have no appreciable affect on themeasured well pressures within the duration of the test. it is further assumedthat there are no appreciable pressure transients at the test well as caused byflow rate changes in other wells tapping the reservoir. A skin effect, ifpresent in the test well, is assumed to be invariant with flow rate, pressureand time. Discussion The theory of linear barrier detection using well pressure measurements has been treated by Horner, Jones, Dolan, Einarsen and Hill and Davis. For thepressure drawdown case in an ideal reservoir with constant rate of fluid entryinto the wellbore, this theory predicts pressure decline curves such as shownin Fig. 1. Here the dimensionless pressure drop p, assuming no skin effect, isplotted vs the natural logarithm of dimensionless time t for several values ofdimensionless distance d between the test well and the linear fluid-pressurebarrier. These dimensionless groups are defined in the Nomenclature. The plotsof Fig. 1 show a curved section which covers approximately a 25-fold timerange, i.e., the ratio of the time at the end of the curved section to that atthe beginning. Stated otherwise, the curved section covers about three naturallog cycles on the time scale or about 1.4 common log cycles, beginning at 0.35and ending at 9. JPT P. 1077^