Horizontal Well Damage Characterization and Removal

Abstract

Summary This paper presents conceptual and mathematical descriptions of the damage along and normal to a horizontal well. Expressions for the skin effect that take permeability anisotropy into account are developed. Stimulation methodology, respecting the shape of damage, is presented. Treatment effectiveness, partial stimulation, and corresponding volume and time requirements are quantified. Introduction The distribution of damage surrounding a horizontal well is neither radial nor distributed evenly along the horizontal well. Permeability anisotropy necessarily would generate an elliptical shape normal to the well. The time of exposure during drilling and completion would result in a truncated elliptical cone with the larger base near the vertical section of the well. Also, the pressure profile in the well during production, would imply the largest pressure gradient normal to the well trajectory nearer the vertical section. Production-induced damage also would be elliptical. This characteristic of the damage would have an impact on the effectiveness of a matrix stimulation treatment. In sandstones, damage removal will be elliptical, and if coiled tubing is used (which is recommended), the treatment would result in a stimulated "collar" around the well. For carbonate stimulation, where the treatment relies not only on fluid flow but also on the actual rock dissolution, damage removal probably will be radial. Simulation of injection has demonstrated clearly the characteristic shape of damage. The larger the permeability anisotropy, the more lopsided the ellipse is. Production rate forecasts show the effect of damage distribution. Dissolution patterns and their sizes (which depend on the volume of stimulation fluid injected) significantly affect post-treatment production. Background In early horizontal wells, failures could be attributed to the selection of inappropriate reservoirs. Significant works1–7 identified the important formation characteristics that would contribute to the success of a well. Generally, horizontal wells have targeted relatively thin formations with good vertical permeability. Also reservoirs with coning problems are good candidates. With these issues relatively settled, the appropriate stimulation of horizontal wells becomes the critical issue. Several publications (not referenced here) describing case studies either have not mentioned stimulation or have compared the performances of fully stimulated vertical wells and totally unstimulated horizontal wells. This is highly inappropriate. Economides et al. 8 showed that the steady-state PI ratio is less than 2:1 for a 1,000-ft [328-m] horizontal well with a skin effect equal to 20 in a formation with kH=10 md, h=50 ft [15.2 m], and I ani=3, whereas a fully stimulated well could deliver several times that amount. The same work showed that long horizontal wells may be good candidates for matrix stimulation (assuming that it is done), rather than hydraulic fracturing, in reservoirs with reasonably low permeabilities (0.5 to 1.5 md) and small thicknesses (<50 ft [<15.2 m]). Except for this small range, however, it has been shown3 that reservoirs where vertical wells ordinarily are candidates for hydraulic fracturing also require hydraulically fractured horizontal wells. More important, though, it has also been shown8 that reasonably modest matrix stimulation treatments (i.e., lower volumes per unit length compared with the 150 gal/ft [1.86 m3/m] routinely pumped in vertical wells) can lead to successful wells. Placement and distribution of the stimulation fluids along the horizontal well are critical. Bullheading the fluids into the horizontal portion is highly unsuccessful. Several reports on such treatments (again a not referenced) and presentations of post-treatment production logs showed that only a small portion of the horizontal well (the one closest to the vertical section) contributes to flow. In one case where bullheading was used, production logs from more than 10 wells in a particular field have shown that 5% to 10% of the total length contributed to flow. This, of course, makes drilling the remaining portion of the well questionable. Coiled tubing, chemical diverters, mechanical isolation devices, and even deliberate blanking of well portions (i.e., leaving certain intermediate lengths unperforated) have been used in attempts to distribute the stimulation fluids more evenly.8 Thus, it is necessary to examine the distributin and shape of damage and to estimate its effectiveness after a partial treatment (i.e., incomplete damage removal). The latter case would describe a collar of stimulation surrounded by nonremoved damage. Characterization of Damage Zone Around a Horizontal Well The shape and distribution of damage around a horizontal well would reflect the vertical-to-horizontal permeability anisotropy and the time of exposure to drilling, completion, and workover fluids. During production, the larger pressure gradient nearer the vertical section would result in a similar shape of production-induced damage. During drilling, it is obvious why mud filtrate penetration would generate a truncated cone with the larger base near the vertical section of the well. Assuming that a 1,000-ft [328-m] horizontal well is to be drilled and that the rate of drilling is 100 ft/D [33 m/d], then the first 100 ft [33 m] would be exposed to 10 days of filtrate invasion, the next 100 ft [33 m] to 9 days, etc., until the last 100-ft [33-m] section is exposed to only 1 day of filtrate invasion. The bases of this cone would be radial in the case of permeability isotropy (Iani= kH/kV=1) and elliptical in all other cases. If the vertical permeability were much smaller than the horizontal permeability (typical anisotropy), then the cone would be elliptical, with the larger axis of the ellipse being horizontal. In the rare case when the vertical permeability may be larger, then the larger axis of the ellipse would be vertical. To demonstrate this concept, the filtrate invasion was simulated with a reservoir simulator with grid refinement and perpendicular bisection grid, as described by Heinemann et al.9 and Heinemann and Brand10 and as applied to horizontal wells by Economides et al.7 To model an example application with this simulator, a 1,000-ft [328-m] horizontal well was discretized into 10 sections, 100 ft [33 m] each, assuming that each section is drilled in 1 day. The injection fluid (filtrate) was water because the filter cake should absorb most of the fines and additives of the mud. Table 1 contains all other pertinent reservoir and well variables. The filter cake was modeled by a flow efficiency reduction (equivalent to a skin effect of 50) for each wellblock as the well was drilled. Hence, each wellblock was exposed to the mud filtrate for at least 1 day.