A New Strategy for Near Real-Time Prediction of Borehole Stability

Abstract This work presents a sensitivity analysis of the different approaches and parameters that are relevant for modelling borehole instabilities. The study was based on an analytical stability simulator with fully coupled poroelastic formulation. The results showed that rock strength and in situ stresses affect stability more than any other rock property. A new methodology is also presented for correlating rock mineralogy from gamma ray logs to lab static tests, as an alternative to the commonly used acoustic-derived correlations. The chemical interactions between clays and the drilling fluid are predicted based on the shale's mobile/immobile water content, inferred from Nuclear Magnetic Resonance (NMR). Such methods, when integrated with the logging- while-drilling capabilities, allow wellbore instability prediction in near-real-time. Introduction hen a borehole is drilled, the original triaxial stress field is replaced by a fluid pressure typically lower than any of the stresses in the original rock column, and slightly above the static pore pressure at each depth. The rock surrounding the borehole is subjected to a sudden stress concentration and may fail, depending on several parameters extensively discussed by the oil industry. The goal of modelling borehole stability while drilling is to define a safe range for the wellbore pressure, in the context of prevailing mechanical and chemical effects. The drilling fluid is the key to ensuring borehole stability, either by adjusting its density, and thus the wellbore pressure, or by changing its composition, to avoid unfavourable interaction with the rock formation. In competent rocks, the drilling fluid density would be chosen to develop a pressure gradient slightly above the pore pressure in the well, in rder to prevent the pore fluid from entering into the well. Stability concerns, nevertheless, must dictate, almost always, the designed mud weight. Success rate in predicting borehole stability has improved significantly in the last 20 years, with the development of more realistic models and their implementation using improved computer technology. The reliability of the input parameters, however, still limit the accuracy of the model predictions. Most of them cannot be directly obtained in the field. Some affect modelling more than others, as is shown in the sensitivity analysis. Wireline logs from wellbores in the vicinity of a new drilling site or logging-while-drilling (LWD) are usually the only source of input data for stability predictors. Nevertheless, the extensive TABLE 1: Input parameters for different model approaches. Available In Full Paper. use of acoustic-derived mechanical parameters has overshadowed the usefulness of other logs as potential providers of input data for tability analysis. Furthermore correlations based on acoustic logs very often fail because borehole instability is typically a quasi-static process that cannot be associated to dynamic moduli measured from acoustic logs. Parameters for Analyzing Borehole Stability While Drilling Several models are available in the literature to define the proper drilling fluid density for preventing instabilities. These models follow different approaches needing different input parameters but the basic data can be divided into "Initial Assumptions" and "Main Parameters," as is summarized in Table 1.