A Hybrid Physics Augmented Predictive Model for Friction Pressure Loss in Hydraulic Fracturing Process Based on Experimental and Field Data

Abstract

Abstract During the hydraulic fracturing process, fracturing fluid leads to excessive frictional losses within the casing and formation which can be impediment to effectiveness of the operations and may result in screenout. Total pressure losses during hydraulic fracturing include wellbore pressure losses in pipe and the near wellbore pressure losses that includes both perforation and formation tortuosity losses. Wellbore friction pressure losses can be determined experimentally using flow loop test at different friction reducer concentrations and flow rates. Practically, fracturing engineers need to estimate and monitor the wellhead pressure and total friction losses during the operation and adjust the fluid, proppant, and friction reducer concentrations during the job. In this study, physics augmented data-driven models are used to predict the pressure losses in pipe using the flow loop test results and the well head pressures and total friction losses. Total friction losses used are based on field data of hydraulic fracturing stages from several wells. The flow loop test is modeled for specific friction reducer types at different concentrations and flow rates. The process of this model relies on analyzing the data obtained from the tests and establishing a correlation between the concentrations of the friction reducers, flow rate and their corresponding pressure losses. The second part is the modeling total pressure losses using field data considering fluid properties and the modeled lab results. Modeling approach incorporates all the key parameters such as fluid properties, proppant concentrations, friction reducer concentrations, slurry rates, perforation (Diameter and Density) and formation characteristics to accurately predict the frictional losses. Both models are tested using validation data. Outcome of the total friction pressure loss model and the forecasted wellhead pressure results exhibited a close agreement between the predicted and the recorded pressure values, indicating that the results can be used in actual field operations. This approach provides valuable information for optimizing hydraulic fracturing design and improving operational practices to minimize the down time. Utilization of the flow loop tests results eliminates the need for repetitive experiments, enabling parametric studies and sensitivity analysis and thus leading to more efficient operations. This unique approach, using the combination of physics-based and data-driven methodologies, represents a major step in predicting pressure losses in hydraulic fracturing operations. Integration of these models not only enhances understanding of the physical processes and the causalities, but also provides practical solutions for improving overall operational efficiency of the process.