Summary
This paper examines the effect of a mud-filtrate-invaded zone on pressure transients from multiprobe/packer-probe wireline formation testers (WFTs). Invasion zones are modeled as composite zones concentric with the wellbore that have different rock and fluid properties (permeability, porosity, viscosity, and compressibility) from those of the native uninvaded formation.
The results show that for multiprobe wireline testers, the sink- (production) and horizontal-probe pressure responses are highly affected by the invaded-region properties, while the vertical-probe pressures are influenced mainly by the properties of the uninvaded zones. For the packer-probe configuration, similar results are obtained (i.e., the vertical-probe pressures are influenced mainly by the properties of the uninvaded zones, while the packer-interval pressures at early times are influenced by the invaded-zone properties). It is shown that if the invaded zones are incorporated into the interpretation process with a 3D r-???-z single-phase, finite-difference model like the one developed in this work, simultaneous matching of spatially available WFT pressure-data sets using nonlinear regression can provide estimates of both invaded- and uninvaded-zone parameters. A synthetic example of a multiprobe test is presented to confirm the theory and procedures developed in this work.
Introduction
The multiprobe and packer-probe WFTs are used to conduct controlled local production and buildup tests as well as horizontal- and vertical-interference tests (interval pressure transient tests, or IPTT). These tools provide formation-fluid samples and estimates of horizontal and vertical permeabilities and wellbore damage. Further details about WFTs, equipped with packers and multiple probes, can be found in Zimmerman et al. (1990), Goode et al. (1991), Goode and Thambynayagam (1992), Head and Bettis (1993), Pop et al. (1993), Kuchuk et al. (1994), and Onur et al. (2004).
It has always been a concern how the pressure transients from these formation testers are affected by the presence of invaded regions around the wellbore. When an oil/gas well is drilled, some of the borehole fluid (mud filtrate) can leak into the formation, displacing the native formation fluid and creating an invaded zone around the wellbore. The invading fluid usually has a viscosity and compressibility that differ from those of the formation fluid. All WFT drawdown and buildup tests and IPTTs are conducted in an openhole environment, in which formation invasion takes place until the mudcake buildups, after which invasion becomes negligible. The invasion depth may vary from a few inches to a few feet, depending on formation and mud properties, as well as drilling parameters. Therefore, one must know how the invaded zone affects parameter estimates from WFT drawdown and buildup tests and IPTTs. Second, can we estimate some of the parameters of invaded zones? For instance, if we can obtain the permeabilities of both invaded and uninvaded zones, this can give us the endpoint effective permeabilities for both water and oil on a much larger scale than that from cores.
In general, the process of drilling-fluid invasion is quite complicated because it involves solid and solute (and solvent, in the case of oil-based mud) transport and precipitation as well as multiphase flow, capillary pressure hysteresis, wettability alterations, chemical adsorption, and gravity effects (Ferguson and Klotz 1954; Bailey et al. 2000; Civan 2002). One of the main difficulties is how to model the initial condition set by the invasion process before sampling and pressure transients from WFTs. In addition, modeling the effects of invasion zone(s) on fluid sampling during production and on pressure-transient data from WFTs during pumpout and buildup periods is challenging, where we are faced by two distinct problems. However, if one assumes that capillary and gravity effects are negligible and that permeability and porosity impairments caused by solid and solute invasion and precipitation are minimal, then one may tackle the invasion problem by modeling it as miscible or two-phase immiscible flow.
Several researchers (Phelps et al. 1984; Hammond 1991; Akram et al. 1999; Zeybek et al. 2001; Wu et al. 2002) have taken this approach to include the effect of invasion on fluid sampling and pressure-transient data from WFTs. However, all work given in Phelps et al. (1984), Hammond (1991), Akram et al. (1999), Zeybek et al. (2001), and Wu et al. (2002) is limited to a single-layer system. In addition, these references do not address the parameter-estimation problem (mainly permeability) in the presence of invasion. For instance, Akram et al. (1999) numerically simulated the invaded zone by injecting water to create a sharp interface (or shock front) between filtrate and oil in the simulator and then studied the effect of invasion on sampling quality (water cut) vs. time during the pumpout (production) period, which is production from the formation into the wellbore by means of a pump. In the presence of water-based-mud invasion, Zeybek et al. (2001) presented an inversion methodology using a numerical simulator to refine oil/water relative permeabilities by integrating dual-packer pressure and water-cut measurements with openhole array resistivity measurements. During their inversion process, they also updated the absolute horizontal and vertical permeability in the simulator. However, they did not consider vertical-observation probe pressure data in their inversion methodology. Wu et al. (2002) presented miscible and immiscible multiphase sampling modeling for oil- and water-based muds for a dual-probe configuration. Assuming that the capillary pressure and relative permeability curves (and the mud-filtrate characteristics) are known, they presented an inversion methodology of horizontal and vertical permeabilities by matching sampling quality and pressure measurements acquired in the presence of oil- and water-based invaded zones. Their inversion method is based on a neural-network approach coupled with a numerical simulator.
Time-Dependant Responses of High-Definition Induction Log and Case Studies
