Threshold of borehole failure: Breaking in before breaking out, Mississippi fan, Gulf of Mexico

Borehole breakouts in hard rocks have been used extensively for the interpretation of stress orientations and magnitudes (Zoback, 2007). Recent borehole images, applied to shallowly buried, poorly consolidated sediment, also show breakouts, from which stress directions and magnitudes have been estimated (McNeill et al., 2004; Goldberg and Janik, 2006; Weinberger and Brown, 2006; Moore et al., 2009; Tobin et al., 2009; Chang et al., 2010). These examples of breakouts are superficially similar to those observed in hard rocks.

Here we also document breakouts from the Gulf of Mexico (Fig. 1) similar to those traditionally recognized; these are zones of lower resistivity on opposite sides of the borehole, apparently due to wellbore enlargement and/or damage or invasion, where the low-resistivity borehole fluid decreases the overall resistivity signal (e.g., Barton et al., 1997; Goldberg and Janik, 2006). We also recognized zones of high resistivity on opposite sides of the borehole with the same orientation and widths as the traditional breakouts. This paper focuses on the state of stress in the well, the origin of these previously unrecognized high-resistivity features, and their geomechanical connection to the traditional breakouts.

We analyzed image logs, related logs, and cores from a hole penetrating sediments of the Ursa Basin, which comprises part of the Mississippi fan, ∼100 km south of the mouth of the Mississippi River (Fig. 1) (Normark et al., 1986; Winker and Booth, 2000; Flemings et al., 2006; Sawyer et al., 2007). Specifically we analyzed borehole failure at Integrated Ocean Drilling Program (IODP) Site U1322, which was initiated 1319.5 m below sea level and penetrated to 238 m below seafloor (mbsf). Sediments recovered at Site U1322 consist of late Pleistocene–Holocene muds and muddy turbidites. These sediments formed as levee deposits and mud drape and/or distal turbidites sourced from distributary channels of the Mississippi fan system (Expedition 308 Scientists, 2006; Sawyer et al., 2007). These sediments are rapidly deposited (∼3.5 m/k.y.) silty claystones younger than 65 ka (Fig. 2). The sedimentary section includes a number of mass transport deposits, or slumps, that rework the sediments (Fig. 2) (Sawyer et al., 2009a). At the depths of borehole failure (90–216 mbsf), porosities range from 43% to 50% (Expedition 308 Scientists, 2006).

Overall the sediments are highly plastic at Site U1322. The sediment liquid limit is ∼65% and the plastic limit is ∼35%, yielding a plasticity index of 30% (Expedition 308 Scientists, 2006; Urgeles et al., 2007). The clay minerals of the silty clays average 73% illite-smectite, ∼80% of which is smectite (Day-Stirrat et al., 2010).

During the past decade logging while drilling (LWD) resistivity at the bit (RAB) imaging has provided remarkable borehole images in sedimentary environments virtually inaccessible by traditional wireline imaging techniques. Using the RAB technique, resistivity images are continuously collected around the borehole. The RAB azimuthal buttons measure resistivities 56 times per bit rotation, which equates to a measurement every ∼1.5 cm in a 25.1 cm borehole (Bonner et al., 1994, 1996). The stated vertical resolution of the RAB tool is 5–8 cm (Bonner et al., 1996). However, at Site U1322 the relatively low penetration rate of 30 m/h and a rotation rate of 60 rpm imaged the borehole approximately once for every 1 cm of penetration, overlapping and improving the resolution over images collected at higher rates of penetration. The pixel size varies with the display parameters in the software used to analyze the images. The unwrapped images of the borehole (Figs. 2, 3, and 4) show variations in resistivity related to changes in lithology, porosity, fracturing, and borehole enlargement.

Breakouts are concentrated between 90 and 120 mbsf but continue downhole to 186 mbsf. Starting at ∼90 mbsf resistivity images show paired areas of lower resistivity 180° apart, which is the characteristic signature of breakouts (Goldberg and Janik, 2006) (Figs. 2 and 3). The resistivity tool does not measure the distance to the borehole wall. However, the decreased resistivity is believed to result from increased distance between the resistivity tool and the borehole wall due to failure and spalling into the well. The increased proportion of drilling fluid between the tool and the borehole wall decreases the resistivity. At Site 1322 the borehole fluid is seawater with a resistivity of 0.2 Ωm (Rider, 1996), whereas the sediment at this depth has a resistivity of 1–2 Ωm (Fig. 2). A decreased resistivity indicates borehole enlargement. This type of borehole failure is properly designated a breakout because material has been removed by spallation into the borehole caused by stress concentrated over an azimuthally limited portion of the borehole wall. These images are typical of breakouts at a number of localities where IODP has conducted RAB imaging (McNeill et al., 2004; Goldberg and Janik, 2006; Weinberger and Brown, 2006; Moore et al., 2009; Tobin et al., 2009; Chang et al., 2010).

From 134 to 216 mbsf breakouts are interspersed with relatively high-resistivity features that share similar symmetry, azimuthal orientations, and widths (Figs. 4 and 5). These higher resistivity anomalies occur in both higher and lower resistivity beds. The occurrence of higher resistivity anomalies in both higher and lower resistivity stratigraphic layering is consistent with a locally smaller borehole diameter bringing sediments, more resistive than seawater, closer to the RAB electrodes. The vertical zones of high resistivity are locally flanked and defined by narrow parallel stripes or edges of low resistivity (Fig. 4).

Ultrasonic caliper measurements indicate that the hole diameter is larger where breakouts are developed and smaller where high-resistivity features occur. The tool provides a measurement of average standoff of the borehole wall for each 15 cm of penetration. The ultrasonic caliper shows an average standoff of ∼2 cm where the breakouts are best developed (90–120 mbsf) and <1 cm where the high-resistivity features are common (Fig. 5C).

We interpret the high-resistivity features at Site U1322 as incipient failure, in which sediment has extruded into the borehole, and not spalled or fallen away. We call these extrusions into the borehole “breakins.” In order to understand this extrusive failure process, we examine the conditions of borehole failure and geotechnical experiments on samples from the borehole.

Understanding the change in state of stress from the undisturbed sediment to the conditions of in the borehole provides insights on the nature of the borehole failure processes and a basis for comparisons to geotechnical experiments on core samples. Site 1322, a vertical borehole in the Gulf of Mexico, is in an extensional tectonic environment (Urgeles et al., 2007; Sawyer et al., 2009a) characterized by active landslides (Fig. 1); the vertical stress is assumed to maximum principal stress (Tables 1 and 2). The data used in the following analysis are derived from both the IODP holes and nearby industry boreholes (Tables 1 and 2).

Borehole failure is caused by a combination of environmental parameters, including overburden stress (σ_v), formation pore pressure (Po), borehole fluid pressure (Pmud), intermediate and minimum principal stresses (σ_2, σ₃), and material properties that resist failure, for example, coefficient of friction (∝) and cohesion (c). These parameters were used to distinguish conditions in which breakouts are dominant (100 mbsf) from conditions at which extrusive failure is common (195 mbsf) (Tables 1 and 2).

Boreholes concentrate the surrounding Earth or far-field stresses (Table 2; Figs. 6A, 6B). Typically stress concentrations caused by boreholes are estimated using elastic models (Zoback, 2007). We have used a version of the Kirsch equations (Zoback, 2007, p. 174) and the far-field stresses to calculate the maximum hoop stress (σ_θθ), colocated vertical stress (σ_zz), and radial stress (σ_rr) acting on three orthogonal planes along the borehole wall (equations and calculated magnitudes are listed in Table 2; Figs. 6A, 6B). These elastic solutions predict that the borehole is significantly into failure at both the 100 and 195 mbsf levels (Figs. 6C, 6D), consistent with our observations (Figs. 3 and 4). The Mohr diagrams (Figs. 6C, 6D) are drawn at the point of maximum hoop stress, at the azimuth of S_hmin, and where compressive borehole failure is consistent with breakouts. High pore pressure (Flemings et al., 2008), low mud weight (Table 1), and low sediment strength (Urgeles et al., 2007; Dugan and Germaine, 2009) all promote borehole failure in these intervals. Because the section is significantly overpressured and the borehole fluid is water, the difference (σ_rr) between the borehole pressure (Pmud) and the formation pressure (Po), defining σ_rr, is negative, exacerbating failure (see Table 2).

Borehole failures recorded in the RAB images occurred quickly, ∼6 min after the hole was cut (penetration rate 30 m/h, RAB tool 3 m above the bit). Excluding the start of the hole (<10 mbsf) and the ending of the hole (200–238 mbsf), the penetration rate was very uniform (29.4 m/h with a standard deviation of 2.4 m/h). Breakouts and extrusive failure are recorded from 90 mbsf. This is the depth at which the liquidity index measured in samples of the Ursa Basin approaches zero [liquidity index = (natural water content – water content at plastic limit)/(water content at liquid limit – water content at plastic limit)] (Lambe and Whitman, 1969); i.e., the natural water content of the samples taken from the borehole is close to or at the plastic limit, denoting a transition from a plastic to a more brittle behavior (Urgeles et al., 2007). Because the muds at Site U1322 are very low permeability (10⁻¹⁷ to 10⁻¹⁸ m²; Binh et al., 2009; Schneider et al., 2008), the failure, occurring within minutes of cutting the hole, is undrained. The fluid pressure in the borehole (Pmud), measured while drilling, approximates a hydrostatic gradient (Table 1). This is expected as seawater was used during drilling of U1322.

The borehole conditions (Table 1) indicate that the sediments in the Site U1322 borehole were in an overconsolidated state during borehole failure. That is, once the hole was drilled the minimum principal stress magnitude at the borehole wall changed from the far-field Earth stress (approximated by S_hmin) to the radial stress (σ_rr), equivalent to the pore pressure less the borehole fluid pressure. The low permeability of the sediments and the short time between cutting the hole and imaging indicate approximately undrained conditions associated with borehole failure.

Experimental deformation of samples from Sites U1322 and U1324 provides clues to the mechanical processes of failure (Urgeles et al., 2007; Dugan and Germaine, 2009). We present the previously unpublished plots of stress, strain, and fluid pressure of four isotropically consolidated undrained triaxial experiments of the nine summarized in Urgeles et al. (2007) (also see Supplemental Table File¹ for details of experiments). Three experiments (Figs. 7A, 7B, 7C) were conducted at lower stress magnitudes than our estimates of in situ stress conditions (Tables 1 and 2). In the fourth experiment (Fig. 7D) the sample was consolidated to an effective stress approximating vertical effective stress in the borehole; thus, it was approximately normally consolidated at the inception of the experiment. Although at an adjacent drill site, the experimental samples in Figures 7A and 7B were selected as they represent the approximate depths of the prominent zones of borehole failure at Site U1322. In Figures 7A and 7B samples show very large strains (∼14%–15%) before the deviator stress stops increasing and the sample is fully in plastic failure (see ^{footnote 1}). These samples exhibit strain softening of 0%–4%. The photographs of the samples show no discrete failure surface, but rather a prominent lateral bulge, and the sample retains cohesion. In Figure 7C, the results are similar to Figures 7A and 7B except that the sample shows a discrete brittle failure surface. In each of these experiments the fluid pressure increases initially, then declines. The decline in fluid pressure precedes the peak deviator stress by ∼9%–10% strain, perhaps indicating the onset of dilatancy. All of the foregoing examples (Figs. 7A, 7B, and 7C) were deformed in an overconsolidated state; the effective stress was equivalent to about half of the preconsolidation pressure, which was determined in consolidation tests from nearby samples (Urgeles et al., 2007).

In contrast, the experiment in Figure 7D was conducted under normally consolidated conditions; in this circumstance, the peak deviator stress and subsequent plateau are reached at a strain 4%–5% less than in Figures 7A, 7B, and 7C; the sample in Figure 7D also shows ∼20% strain softening toward the end of the experiment, unlike the other experiments. In the experiment in Figure 7D, pore pressure remains high and the sample photograph shows a discrete planar failure surface. The angular relationships between the principal stress and any failure surfaces in all experiments are generally consistent with the friction angle of 28°, conforming to the Coulomb failure criterion (Urgeles et al., 2007).

None of the experiments described above showed a loss of cohesion. However, the sample deformed at normally consolidated conditions (Fig. 7D) showed the best developed failure plane, which would likely provide the best surface for later spallation.

The experiments outlined here replicate some of the deformation in the borehole in that they show ∼10%–15% strain in elastic-plastic deformation prior to plastic failure. The lateral bulging of the experimental samples may mimic the extrusion of the sediments into the borehole at Site U1322. In one case (Fig. 7C), the lateral bulging is associated with a discrete failure surface. In the overconsolidated experiments (Figs. 7A–7C) the decline of pore pressure after an initial increase is consistent with the sample undergoing internal dilation. The experiments are undrained, so monitoring pore pressure is a means of inferring small changes in volume. Dilation increases the internal effective normal stress and frictional strength, causing dilatancy hardening (Rudnicki, 1979). Dilatancy hardening can temporarily keep the sample from strain softening, in contrast to the sample in Figure 7D that remains at high pore pressure and weakens near the end of the experiment. The sample in Figure 7D shows a more discrete failure surface than any of the other samples, and in the case of the borehole, is more likely to undergo spallation and formation of a typical breakout.

The development of discrete shear surfaces with associated bulging may explain the low-resistivity zones at the edges of some of the high-resistivity borehole failures (Fig. 4). The low-resistivity edges may be the projection of discrete shears flanking the high-resistivity zones (Fig. 8). The shears, associated fractures, and any incipient spallation would be more likely to be invaded by salt water, therefore lowering their resistivity.

The extended plastic deformation of sediment leading to an initial stage of bulging at Site U1322 is consistent with the more than 50% clay-sized particles (Sawyer et al., 2009b), the abundance of smectitic clays (Day-Stirrat et al., 2010), and the consequent high plasticity (Urgeles et al., 2007). These are the most clay-rich sediments encountered at continental margins where breakouts have been observed in IODP holes. The highly cohesive nature of the clays apparently both resists discrete failure and the disaggregation of the sample along discrete shear planes.

The extrusive borehole failure interpreted at Site U1322 is similar to some failure behavior observed in industry boreholes (Moos et al., 2007). Based on observations and models, Moos et al. (2007) showed that sediments with a higher residual strength form shallower breakouts; the less that the strength drops after initial failure, the shallower the breakout. Moos et al. (2007) did not observe the bulging noted here, probably due to the lesser plasticity of the rocks they were studying. However, the role of residual strength in retardation of spallation is consistent with our interpretations of the bulging sediment as an incomplete or unspalled conjugate failure.

The extrusive failure at Site U1322 indicates volume expansion in the brief interval between cutting and imaging the borehole. Such volume expansion could locally lower the pore pressure, momentarily, and strengthen the sediment. Both the breakouts and the extrusive borehole failure occur in mass transport deposits (Fig. 2) which, in this borehole, have the lowest porosities (35%–40%) and lowest predicted permeabilities (10⁻¹⁷ to 10⁻¹⁸ m²) based on known porosity-permeability relationships (Binh et al., 2009). The lower porosities of the mass transport complexes, relative to the normally sediments deposits, imply a higher overconsolidation state and further potential for dilation during borehole failure. Mass transport complexes in industry wells off Angola also appear to be more susceptible to borehole enlargement (Elieff et al., 2008).

Borehole failure in the upper part of the hole is dominated by breakouts. Extrusive failure is more common in the lower part of the borehole. The occurrence of extrusive failure adjacent to breakouts is consistent with extrusive failure as an initial state of breakout formation. We have no good explanation for the concentration of breakouts high in the borehole and the greater incidence of extrusive failure or breakins at depth. Porosity, gamma ray, resistivity, clay content, and pore pressure do not vary significantly through the section with breakouts and breakins below (Fig. 2).

Apparently, RAB images from U1322 catch an early extrusive stage of breakout formation. As these images were taken only 6 min after opening of the borehole, it is possible they record the initial stage of breakout formation. Given more time the borehole failure process could have become more complete, resulting in full breakout formation and potentially more extensive failure throughout the borehole.

At Site U1322 breakouts are marked by low-resistivity vertical zones 180° apart across the borehole. Vertical zones or stripes of high resistivity follow the azimuthal trends of breakouts and occur intermixed with breakouts. Caliper measurements show that the borehole is larger in diameter where the breakouts occur and smaller in diameter where the high-resistivity zones occur. The high-resistivity zones apparently represent intervals of sediment extrusion into the borehole. Analysis of the state of stress in the vicinity of the borehole predicts that sediments failed in an overconsolidated state once the hole was drilled. Geotechnical experiments on samples from the borehole show a bulging failure, especially in overconsolidated samples; this may explain the extrusion of sediment in the borehole. The occurrence of discrete shear failure in some samples could explain the narrow low-resistivity zones flanking some of the high-resistivity features in the borehole. Borehole fluid invasion along the incipient shears could produce the narrow low-resistivity zones adjacent to the apparently extrusive failures. During the experiments the samples retain cohesion, even along discrete shear surfaces. The cohesive nature of the clay-rich, highly plastic samples may retard spalling along the failure surfaces, maintaining the bulge into the borehole. The more traditional appearance of the breakouts from elsewhere in the borehole can be explained by spallation or disaggregation after initial extrusive failure. Maintenance of cohesion and strength may be due to the clay-rich nature of the sediment and local dilatancy hardening that lowers fluid pressure and temporarily strengthens the sediment during borehole failure. We name the extrusive failures “breakins,” in contrast to the traditional breakouts into which they evolve.

This paper is dedicated to the memory of Bill Normark, pioneer in many fields, including the Gulf of Mexico. We acknowledge the U.S. Science Support Program for funding during and after Integrated Ocean Drilling Program Expedition 308. Geomechanics International provided a generous academic discount on their image interpretation software that was instrumental in post-cruise data analysis. We thank Nicholas Davatzes and an anonymous reviewer for thoughtful insights on the original manuscript and Dave Scholl and Dennis Harry for editorial advice.