A Data-Infused Methodology for Estimating Swelling Potential in Shales Exposed to Various Completion Fluids

Formation damage in reservoirs poses a recurring challenge throughout the phases of drilling, completion, and production, significantly impeding efficiency and diminishing resource extraction in oil and gas development. This detrimentally affects production capacity, leading to potential reservoir shutdowns and hindering the timely discovery and development of oil and gas fields. The water-based drilling fluids are mixed with various swelling inhibitors; nevertheless, shale swelling could still take place during the completion phase as these fluids do not usually consider this phenomenon. To quantify the swelling inhibition potential of drilling/completion fluids, several laboratory experiments are usually carried out. These experiments are costly, time-consuming, and tedious. This study used machine learning technique to predict the dynamic linear swelling of shale wafers treated with different types of completion fluids containing varying inorganic salts such as NaBr, CaBr₂, and NH₄Q. A comprehensive experimental investigation was conducted to gather datasets suitable for training machine learning model based on various completion fluid constituents. The study involved utilizing a dynamic linear swell meter to quantify swelling inhibition potentials, assessing sodium bentonite clay wafers' responses to all completion fluid solutions through linear swell tests lasting 24 to 48 hours. Additionally, the study measured zeta potential and conductivities across solutions with different concentrations. Leveraging sequential data and memory cell architectures, the research developed an LSTM (Long Short-Term Memory) machine learning model aimed at predicting and comprehending swelling behaviors within specific contexts. This model was trained using input parameters such as zeta potential, salt conductivity, salt concentrations, density, and elapsed time, while the model output represented dynamic linear swelling in percentage. This intelligent technique can be used to guide and streamline laboratory experiments to determine dynamic linear swelling of shales. It can serve as a quick tool to guide fluid engineers at the rig site to delineate shale swelling reasons pre-, post-, and during completion operations. Consequently, operators will be better prepared to deal with unknown swelling issues that lead to NPT in operations.