Abstract
Sand production is one of the major challenges in the oil and gas industry, so comprehensive geomechanical analysis is necessary to mitigate sand production in mature fields. The absence of crucial well logs is an essential challenge in the oil and gas industry, necessitating geologists and engineers to rely on empirical equations to predict the absence of log intervals. A comprehensive study was carried out on geomechanical modeling using data logs from two wells located in the Nahr Umr formation in Southern Iraq. Furthermore, the geomechanical parameters used by the predictive model were verified through caliper measurements. A machine learning technique was employed to predict the absence of acoustic log in Well-5 instead of using empirical equations. Additionally, two sand management models were developed and compared - one using the empirical Gardner equation and the other employing machine learning techniques. The sand management model based on the Gardner equation predicted the production of sand from the beginning. However, it did not match the actual production data observed in real life. On the other hand, the machine learning-based model indicated no probability of sand generation, which aligned with the observed production data. The findings of this study demonstrate the advantages of using machine learning over traditional empirical equations for geomechanical studies in the particular area under study. Also, these findings suggest that machine-learning techniques might be applied to more basins in southern Iraq. The current research improves our understanding of the impact of machine learning on sand management as well as geomechanical characterization. This study has the potential to enhance procedures for making decisions in the petroleum and natural gas industries and contribute valuable knowledge to improved ways of handling sand production problems.