No abstract

In a probe-type formation test, because of the geometry of the wellbore and the sealing effect of mudcake, the flow pattern is not perfectly spherical. To account for the deviation from spherical flow, several geometric correction factors were proposed for different analysis techniques (Steward and Wittmann 1979;Wilkinson and Hammond 1990;Dussan and Sharma 1992;Goode and Thambynayagam 1992;Proett and Chin 1996). A geometric factor is used in formation rate analysis (FRA) (Kasap et al. 1999), a technique used in analyzing a probe test to estimate formation pressure and permeability. Like other geometric correction factors, the geometric factor is a strong function of permeability anisotropy that is generally unknown before a test. When analyzing the test, we would logically assume an isotropic formation and use the corresponding isotropic geometric factor. Consequently, the FRA-estimated permeability does not represent the true spherical permeability. In contrast, the spherical permeability can be estimated from buildup analysis without prior knowledge of permeability anisotropy. Therefore, there is a discrepancy between the permeability estimates from the two analysis methods. In addition, if considered separately, neither FRA nor buildup analysis can decompose the estimated permeability into its horizontal and vertical components.This paper presents the derived numerical values of several geometric factors. Using these factors, we show that the discrepancy between the permeabilities estimated from FRA and from the conventional buildup analysis is attributable to the permeabilityanisotropy effect. A correct geometric-factor value must be used to estimate permeability correctly. On the basis of the permeabilityanisotropy effect, we present the procedures to estimate horizontal and vertical permeabilities by combining FRA permeability and buildup permeability or by history matching. These procedures are verified with a simulated probe test. Analysis of three actual tests is presented.

No abstract

During the fracturing treatment, fracturing fluid is pumped to generate fractures and then followed with a large amount of proppant to provide enough conductivity for reservoir fluid to flow to the wellbore. The ultimate proppant distribution in the fracture system directly impacts well productivity and production decline rate. However, it is very challenging to predict how far proppants can go and where they will settle because of the complexity of the fracture system. Previous modeling and experimental studies were usually based on simple proppant settling velocity models and limited only to planar fracture cases. In a recent numerical study, proppant transport in different complex fracture geometries was modeled. However, the fracture walls in the model were considered to be perfectly smooth. In this study, proppant transport in complex fracture geometries with different wall roughnesses was investigated using computational fluid dynamics (CFD) model, in which the interaction between proppant particles, the carrying fluid phase, and the rough fracture wall was fully coupled. A planar fracture case with smooth fracture wall was first investigated using a CFD model and benchmarked with results from commercial software. The CFD models were then used to simulate the proppant transport in T-junction and crossing-junction scenarios with different fracture wall roughnesses, which are often seen in unconventional reservoir fracture systems. The results from the CFD models indicate that proppant transport within complex fracture geometries is significantly affected by fracture wall roughness. Rough fracture wall can exert resistant drag force to proppant particles and carrying fluids and hence influence the proppant transport behavior and particle distribution. It is found rough fracture wall decreases both proppant horizontal transport speed and vertical settling speed which can lead to a better vertical coverage of proppant particles in the fracture. However, more pumping energy and time are required to transport the proppant particles to the same fracture length with rough fracture surfaces compared to smooth fracture surfaces. Studies on proppant density show light weight proppant has a better vertical distribution in fractures with rough walls due to more pronounced drag force effect. With high viscous carrying fluids, proppant in both smooth and rough fractures can transport further at the same transport time. Proppant transport models developed in this work fully incorporate the interaction between proppant particles, carrying fluid dynamics, and rough fracture surfaces. This study extends the current understanding of proppant distribution in complex fracture geometries and helps optimize hydraulic fracturing design to improve unconventional well production performance.

Th6 paper was selected for presentation by an SPE Program Commmee Iollowmg rewew 01 mformatlon conlamed In an abstract submmed by the author(s) Contents of the papr, as pm. sented, have not been rewewed by the SIXmty of Petroleum Engineers and are subpct to cor-r~t!on by the author(s) The matercal, as presented, does not necessarily rellect any postmn 01 the SJX@ 01 Petroleum Engineers ,1sofficers m members Papers presented al SPE meet. mgs are subject to pubhcabon revmw by Edltonal Commnlees of the Somety 01 Petroleum Perm#sslon to copy IS resmcted to an abstract 01 not more Ihan 300 words Illustrahons may not be copied The abstracl should contain conspicuous acknowhdgment of where and by whom the paper was presented Write Ltb,armn, SPE, P O u ox 833836, Richardson, TX 75063.3636 U S A Fax 01-214-952.9435 AbstractThere has been a distinct trend during the past 10 years toward developing new multisensor-type acquisition syslems in well logging. In contrast (o corthentional logging tools, the high--definition induction log developed by Western Atlas Logging Services provides multiple measurements with different verlical resolution and depth of investigation. It combines several subarrtiys. each one associated with a specific spacing between a transmitter and receiver that works at mul[iple frequencies covering fi wide dynamic range, Data from the array induction instrument allow (he log analyst [o obtain. more than ever before. a comprehensive pic[ure of the resistivity distribution around the borehole. At the same time, more emphtisis is placed on the data interpretation.Two-dimensional (2-D) inversion techniques represent an excellent way to interpret array induction data. These techniques allow for accurately and simultaneously considering both the 2-D distribution of formation resistivity and the full spectrum of array data in a single, unified interpretation scheme. Results of inversion represent a single earth model consistent with the entire set of array induction data. However, 2-D inversion schemes require a significant computational time, and, in the case of full-spectrum inversion of array induction data, they become impractical. The rapid 2-D inversion method drmtically reduces the computer time requirements and provides a rapid convergence to a valid 2-D earth model.A modificfition of the rapid 2-D inversion method further reduces computer time requirements by use of precalculated sensor responses and allows for interpreting the full-spectrum 6iiltiB *"' 9 . ..SodetyofPetroleumEngineers array data to deliver results at the well site. Results from a number of synthetic examples are presented that show the proposed technique's ability to recover formation parameters in multilayer environments. Results from inversion of field data using the new approach show improved estimates of formation resist ivities.