Estimation of Bottom Hole and Formation Temperature by Drilling Fluid Data: A Machine Learning Approach

Abstract

Two of the most important measurements during and after drilling a geothermal well are the bottom hole circulating temperature (BHCT) and the static formation temperature (SFT). BHCT is critical for bottom hole assembly (BHA), drilling fluid and cement slurry designs. SFT is significantly essential since it directly correlates with the amount of renewable energy power that can be produced from the well. Currently, these data are obtained from various equipment such as measurement while drilling (MWD), logging while drilling (LWD) and temperature logs. The measurements are used for calculations related to geothermal power plant construction as well as the drilling and completion designs of subsequent wells in the field. However, data from MWD is not always available. On the other hand, the process of taking a temperature log is time-consuming and expensive. In this paper, we are proposing a machine learning approach to predict the BHCT and SFT in real-time using drilling fluid data (mud weight, rheological properties, flow in and out temperatures and circulation time). For the analyses, data from various wells were obtained from a large independent operator. Data available includes casing and drillstring design, daily mud reports, flow rate, bottom hole temperature readings from MWD and the temperature log data obtained subsequent to drilling and testing the wells. This data was used for training and testing the machine learning algorithms. Two different models (random forest and XGBoost) were trained. 80% of data was used for training while 20% of data was used for testing the performance of the algorithm. A perfect match with the trained models and testing dataset was observed with mean absolute percentage errors (MAPE) of less than 1% for both algorithms. The trained models are able to provide both BHCT and SFT with extremely high accuracies using the drilling fluid data which can be recorded on the surface in real-time. This paper presents a novel approach to estimate geothermal well temperatures and is believed to be very beneficial for practicing engineers to save a large amount of time and cost in geothermal development projects.