An Automated Drilling Geomechanics Simulator Using Machine-Learning Assisted Elasto-Plastic Finite Element Model

Abstract

The petroleum industry has long relied on pre-drilling geomechanics models to generate static representation of the mud weight limits. These models rely on simplifying assumptions such as linear elasticity, a uniform wellbore shape, and generalized failure criteria to predict failure and determine a safe mud weight, which lead to inaccurate results. Thus, this study’s main objective is to improve the process for predicting the wellbore rock failure while drilling. Wellbore failure prediction is improved through the use of a novel modelling scheme that involves an elastoplastic finite-element model (FEM), machine learning (ML) algorithms, as well as imaging data that accurately describes rock failure. Geomechanics data are modelled in the FEM code, which are then used to train the ML algorithms in conjunction with imaging data. The produced integrated model of FEM and ML is used to predict failure limits for new wells. This improved failure prediction can be updated with the occurrence of different drilling events such as induced fractures and wellbore enlargements. The integrated modelling scheme was first applied to lab experimental results to provide a proof-of-concept and validation. This application showed improvement in rock-failure prediction when compared to conventional failure criteria such as Mohr-Coulomb. Also, offset-well data from wireline logging and drilling records are used to train and build a field-based integrated model, which is then used to showcase different field applications of the model. The field applications presented exhibit the advantages of this modelling scheme where integrating a physics model such as the FEM with a ML algorithm can significantly improve failure prediction while requiring significantly smaller datasets for training than those required for purely data driven ML models. The applications also show that the integration process lead to a better understanding of how failure takes place, which can be used to re-assess the mathematical formulation of the failure surface. From a practical perspective, the improvement in failure prediction can help avoid non-productive time (NPT) events such as wellbore enlargements, hole cleaning issues, pack-offs, stuck-pipe, and lost circulation. This efficiency is to be achieved by a real-time implementation of the model and by taking advantage of available data that are not routinely utilized by drilling.