Formation drillability evaluation plays an essential role in the exploitation of shale gas reservoirs. Currently, the common methods for evaluating formation drillability mainly include experimental methods, such as scratch tests and micro-drilling tests, and regression analysis. However, the applications of experimental methods and regression analysis cannot reasonably assess drillability due to a lack of well-core information and describe the strong heterogeneity of shale formations. In this study, a data-driven model based on the constrained K-means (CKM) method was proposed to evaluate shale formation drillability, which can solve the problems related to core sample limitation and heterogeneity of the formation. First, a regular evaluation model was built against the conventional cluster methods, such as K-means, DBSCAN, and GMM. Secondly, the closed pairwise constraints were generated based on domain knowledge. Finally, the evaluation model was established by applying pairwise constraints to the K-means clustering generation process. The value of normalized mutual information (NMI) ranges from 0.86 to 0.93 based on the proposed model, which has been improved by 50% over the conventional cluster methods. The results show that the drillability index of the Longmaxi Formation ranges from 5 to 7, whereas the index value of the Wufeng Formation averages at 8-9, which demonstrates that the Wufeng Formation is more difficult to drill compared to the Longmaxi Formation. The drilling practices in the field demonstrated that the actual drilling rate of the Wufeng Formation was lower than that of the Longmaxi Formation, which is in accord with the model predictions.
As an essential unconventional oil and gas resource, shale gas reservoirs have enormous extraction and economic value. Shale drillability characterizes the resistance of shale to drill, which is a crucial type of information requested for the bit type selection, borehole track design (Chen, 2021; White, 1969), and drilling implementation (White, 1969; Chen et al., 2017). Inaccurate evaluation of rock drillability has resulted in a slow drilling rate and severe drill bit wear, leading to procrastination in construction and significant financial losses. Many investigators considered that geomechanical properties (rock type, rock mass jointing), machine parameters (drilling machine type, bit pressure, rotary speed), and operational process determine the drillability of rock (Saeidi et al., 2013; Yetkin et al., 2016). The methods toward getting drillability are majorly implemented in two ways: indoor rock core experimental and numerical model (Gan et al., 2021).
