علی رنجبر

The instability of wells is one of the most important challenges in the drilling industry. Determining the stability of the well is highly dependent on the evaluation and calculation of the in_situ stress. For this purpose, in this study, after calculating the stress with the help of petrophysical data, stress estimation is done through two machine learning techniques. Because obtaining the amount of stress at different depths requires a lot of logs and also high costs, we can obtain the stress values at the desired depths from the drilling data always available in the fields and without the need for additional costs. Therefore, in this study, stress estimation is introduced with the help of machine learning and using daily drilling data as input parameters. The machine learning algorithms used in this article include random tree (RF) and support vector machine (SVM) and the drilling data includes mud input rate (Mwin), mud output rate (Mwout), flow rate (Q ), torque (T) and round per minute (RPM). According to the algorithms used, RF is more accurate than the SVM method for modeling.