An investigation into the effect of geomechanical properties of reservoir rock on drilling parameters—a case study

Drilling operation is among the most complex engineering activities as it is affected by a wide variety of factors. Many researchers have considered the effects of geomechanical parameters (GPs) of the reservoir rock on the drilling parameters (DPs), so that one can contribute to enhancing the drilling efficiency by building on the knowledge about the interaction between the GPs and the DPs. In this paper, artificial intelligence (AI) methods were used to investigate the association between wide range of the DPs and the GPs and provided new approach to improve drilling operations in carbonate reservoirs. Indeed, as aspect of the types of parameters and also machine learning methods, the study is novel method and seeks coverage in the lake of considering GPs in terms of drilling and its effectiveness. Dataset was collected from two oil wells located in the southwest of Iran. Firstly, 1-D geomechanical models were built to represent two reservoir formations, namely Asmari and Sarvak. Acknowledging the differences in resolution between the mud-logging and petrophysical data, the data was preprocessed (up-scaled) to have it prepared for further analysis. Subsequently, two methods were applied on the preprocessed data: (1) K2 algorithm as an instance of the so-called Bayesian network, showing both direct and indirect associations between the considered variables, and (2) multilayer perceptron (MLP) as an artificial neural network. The results of the K2 algorithm confirmed the direct impact of the unconfined compressive strength (UCS), a GP, and on the rate of penetration (ROP), a DP. It was further found that the internal friction angle and vertical effective stress (GPs) tend to affect the weight on bit (WOB) and torque (DPs). Focusing on the Asmari and Sarvak Formations, the results of the MLP implied that, as far as the effects of GPs were concerned, the rotational speed (RPM) was most greatly affected by the internal friction angle followed by the UCS. Comparative charts were prepared to compare the corresponding GPs and DPs against different geomechanical units (GMUs). Given the non-controllable nature of the DPs, the results showed that the choice of such adjustable DPs as WOB and resultant torque could affect the ROP as indicators of the drilling operation quality. Finally, the results were used to propose optimal values of the adjustable DPs for different classes of rock properties.