A combination of high overpressures and weak rocks can quickly narrow the range of available mud weights for operators drilling in the challenging environment of the Gulf of Mexico. Basic geomechanical modeling can assist in mitigating these challenges by helping explain the pressures at which boreholes fail and/or fractures propagate from the well bore. But occasionally the basic models come up short when determining operating mud weight ranges.

Figure 1. Figure 1a shows typical cavings from well bores that have failed in shear. They are typically splintery in shape. Figure 1b shows cavings from Shenzi that appear more blocky or tabular in shape, indicating that failure was more likely caused by anisotropic strength than shear failure of the wellbore wall. (All figures courtesy of GMI)
A case in point is the Shenzi field in Green Canyon blocks 653 and 654. Early exploration and appraisal drilling in this deepwater (>3,000 ft or 915 m) field resulted in significant wellbore instabilities and lost circulation in the subsalt formations.

After the initial well bores were drilled, the drilling team believed the instabilities could be explained by the uncertainty in the pre-drill models. Adjustments were made to key parameters such as pore pressure, rock strength and stress magnitudes, and operations resumed. However, drilling problems in subsequent wells persisted, and the team concluded that other considerations needed to be addressed before drilling any future wells.

While drilling the Shenzi wells, the team noted that borehole instabilities were more severe when drilling down-dip at low angles of attack to the bedding planes; however, instabilities were almost non-existent when drilling up-dip at angles nearly perpendicular to the bedding planes. Operations also experienced significant lost circulation when mud weight was increased to compensate for instabilities in the down-dip well bores. The high mud weights required to limit borehole collapse only increased the risk of losses in low-pore-pressure sands. These operational issues resulted in narrow mud windows that constrained the casing design, which in turn resulted in severe cost overruns.


Complex geomechanical model needed

The Shenzi observations presented a convincing argument that previous boreholes had experienced anisotropic failure due to an anisotropic strength along the bedding planes. In addition to wellbore attack angles making a difference in the stability of the well, samples collected at the rig were characterized as blocky and platy. This is not typical of normal wellbore instability due to shear failure. Based on these facts, anisotropic rock strength was diagnosed as the root cause of the wellbore instability, and the operator concluded that a more complex geomechanical model was needed to predict mud weight ranges for future wells.

In the Gulf of Mexico, failure of the well bore is usually characterized as isotropic due to shear stresses acting on the wellbore wall. When the mud weight isn’t high enough, the rock can fail on the side of the well bore that is in compression. For any given trajectory, the size of these compressive failures, called breakouts, depends on the magnitude of the stresses acting on the wellbore wall, the mud weight seen by the rock, the rock’s mechanical properties and the wellbore orientation.


Figure 2. The image shows two of the calculations made at different well locations. The different colors represent the mud weight needed to keep the borehole stable for different orientations at a fixed depth. A vertical well would plot at the center of the circles, and horizontal wells would plot along the outside of the largest circle. Well azimuth is represented by moving clockwise around the circles. In this case, the colors are offset from the center, largely due to the high bedding dips at those locations with formations tending to dip towards the northwest.

ilure due to anisotropic strength is not often observed in drilling environments. Usually, rock strength is isotropic, and when loaded in a particular direction, the failure plane is independent of the bedding planes. But when anisotropic rocks are loaded in the same manner, they tend to fail in planes dictated by the fabric of the rock at stresses lower than that would cause an intact rock to fail.

Ideally, high-resolution image data is collected for anisotropic failure modeling so the shape of the borehole failure can be observed. But in the Shenzi well bores, high-resolution image data was not possible to obtain due to the heavy oil-based muds used. Instead, several oriented multi-arm caliper logs were available and, while not as good as image logs, these data can be used to help identify the presence of breakouts. Also, some of the breakouts took on a “two-lobed” shape when plotted in a rose diagram frequency plot. This further suggested that failure due to anisotropic strength was occurring in the Shenzi downdip well bores.


Calibration of the model

To validate the geomechanical model, verification that the model can predict what actually happened in previous wells must be accomplished. Actual mud weights used to drill previous wells were inserted into the model and then the forward models to predict the amount of failure were compared to the drilling experience and caliper logs. Good correlation between the predicted failure, caliper data and reported borehole problems provided confidence that the new model would be adequate for predicting failure in future wells.

Figure 3. The image shows an example of an operating mud window determined by evaluating borehole stability constraints together with fracture gradient constraints.
Early Shenzi wells had several problems due to geomechanical instabilities. However, not all boreholes experienced problems. The geologic sidetracks that were drilled updip experienced nearly no borehole instabilities with similar mud weights. To help explain this, all wells were evaluated at the same true vertical depth to see how wellbore orientation affected the stability calculations. Because of different formation dips at various well locations, comparisons at each of the Shenzi well locations were done to ensure the geomechanical model was accurately predicting borehole failures for the entire field.

Understanding the borehole failure mechanism and mud weights for hole stability was only part of the calibration process for the model. To successfully use the model in future operations, the “fracture gradient” also needed to be understood. In order to do this, knowledge of what was happening to the wellbore wall when lost circulation occurs was also needed.

Typically, lost circulation occurs when the downhole pressure exceeds both the near-borehole and far-field stress magnitudes and a hydraulic fracture can be propagated away from the wellbore wall. The near-borehole effects are highly dependent on the borehole orientation relative to the stress field and the existence of natural or induced fractures.

The Shenzi field presented a puzzle. How were some of the wells experiencing lost circulation with certain mud weights, while others were not? In this case, the answers were found by evaluating the pressure-while-drilling (PWD) data.

In this data, spikes in the PWD, most likely due to packing-off events, were seen just before lost circulation was experienced. This was more than enough excess pressure to overcome the near-wellbore stress effects. Lost circulation was experienced following this pressure spike, and fluid was lost to the formation until the equivalent downhole pressure was reduced to magnitudes less than the far-field minimum stress magnitude. Once the annular equivalent pressure dropped below the minimum horizontal stress, the fractures could close, preventing any further lost circulation.


Application and results

The purpose of this complex wellbore stability modeling was to gain an understanding of the mud weights needed for success when drilling future Shenzi field wells. Once the failure mechanism was properly identified, calculated mud weights for both up- and down-dip boreholes were obtained. This, coupled with knowledge of how the near-borehole and far-field stress can affect the fracture gradient, gave a clear understanding of the allowable operating mud weights.

In some cases, the modeling revealed that current casing design did not allow for sufficient difference between the stability mud weight and fracture gradient; therefore, changes in the equivalent downhole pressure, such as ECD, could not be adjusted for completely.

While the Shenzi field experienced multiple wellbore stability and lost circulation problems during the early stages of exploration, the use of geomechanical modeling assisted significantly in turning difficult drilling problems into manageable ones. The geomechanical model has since been included in all the planning aspects of Shenzi appraisal and future development wells. Developments well paths have been designed to “geomechanically” increase the mud window, and in some cases it was even possible to drop a planned casing string.