Prediction of barite scale formation and inhibition in hydrocarbon reservoirs using AI modeling: Focus on different optimization algorithms

• Innovative AI Models : Development and comparison of six intelligent models, including Extreme Learning Machine (ELM) and Support Vector Regression (SVR) with Particle Swarm Optimization (PSO), for predicting barite scale formation and inhibition. • Enhanced Dataset : Utilization of data augmentation techniques, expanding the dataset from 552 to 5501 points, enabling robust model training and validation. • Exceptional Accuracy : The ELM model achieved unprecedented precision (RMSE = 0.0005, R² = 0.99999), outperforming other models in estimating barite scale inhibition and formation. • Pioneering Contribution : Establishes a novel methodology for accurately predicting barite scale behavior, providing practical solutions for improved hydrocarbon reservoir management. Barite scale formation is a persistent and detrimental challenge in the oil industry during water-flooding operations, primarily due to the incompatibility between injected water and formation water. This study introduces a novel and highly effective approach for predicting barite scale formation and inhibition by leveraging advanced artificial intelligence (AI) techniques. Innovative intelligent models, including Random Forest (RF), k-nearest Neighbors (KNN), Extreme Learning Machine (ELM), Support Vector Regression (SVR), Decision Trees (DT), and Multilayer Perceptron (MLP), were developed and optimized for this purpose. Notably, the MLP and SVR models were enhanced through training with the particle swarm optimization algorithm (PSO) and data augmentation techniques. The performance of the models was rigorously evaluated using statistical criteria and graphical tools, demonstrating exceptional accuracy with minimal errors. The ELM model emerged as the most precise (RMSE = 0.0005, R² = 0.999), closely followed by the SVR-PSO model (RMSE = 0.0008, R² = 0.999). Additionally, the ELM model successfully estimated over 95% of the data points, with an applicability domain of 97.82%. Moreover, we conducted an additional validation using a new independent dataset, which showed an excellent correlation coefficient (0.99) and a very low RMSE (0.01), further demonstrating the robustness and generalizability of the proposed models. This work shows the transformative potential of AI-driven modeling in addressing complex challenges such as barite scale formation. The findings underscore the practical implications of AI techniques in optimizing oilfield operations, enhancing efficiency, and reducing costs, thereby offering a significant advancement in the field.