Innovative Method to Predict H2S Concentration from Advanced Mud Logs

Abstract Hydrogen Sulfide (H2S) presents a hazardous environment while drilling. Measuring H2S concentration real-time while drilling will provide early indication and allows the operator to the necessary safety precautions at the rig site and ensures proper execution of rig operations. There were several attempts in the oil and gas industry to estimate hydrogen sulfide, yet there's no proven methodology to accurately measure H2S concentration real-time while drilling. Mud Logging services, also known as surface data logging can be utilized to predict H2S concentration while drilling. Although Mud Logging does not actually measure H2S directly, some of the actual measurements can be used as proxies to predict H2S concentration. Common practice is to acquire PVT samples and/or perform Drill Stem Test (DST) to quantify fluid properties and have an accurate measurement of H2S volume. Our innovative methodology that will be discussed further revolves around building a predictive model that can estimate H2S concentrations prior to sample acquisition to ensure proper and safer sampling or testing procedure. The model is trained on few wells where DST tests were performed and an actual H2S concentration is known, then the model is deployed on few blind tests to investigate model validity and accuracy. Several iterations were made to select the best inputs from Advanced Mud Logging (AML) logs to best predict H2S % in the reservoir rocks. Furthermore; multiple regression methods were tested to achieve optimum modeling. The CNN (Conventional Neural Networks) model delivered the lowest RMSE (Root Mean Square Error), therefore was identified as best modeling technique. Several inputs from AML logs were selected to build the predictive model: Methane, Total Normalized Gas, Volume of Calcite, Volume of Dolomite and CO2 (Carbon Dioxide). The model indicated that there was a very strong correlation between AML outputs and H2S concentration with Normalized total gas showing the highest correlation followed by C1 gas (Methane). Interpreting potential in-situ H2S can lead to taking the necessary precautions at the rig site. Using advanced mud logging can not only identify zones of interest but can also act as an early warning system for potential drilling hazards. Workflows are placed to identify these potential H2S accumulations that can pose a risk to operations, downhole tools, and forward testing procedures can be planned accordingly.