Accurate density prediction for aqueous Na-K-Mg-Ca-Li-Cl-Br-SO4 solution at high-pressure and high-temperature conditions

Knowledge of water volumetric properties is considered of extreme value in many disciplines of science and engineering, especially those seeking effective management of water resources. This knowledge is normally gathered through experimentation in laboratories within a wide span of time. Traditional methods have already been employed in integrating databases, but these are either afflicted by their limited accuracies or restricted in their applicability ranges. In this communication, density of aqueous electrolyte solution was estimated using adaptive network-based fuzzy inference system. Subtractive clustering was employed to transform crisp input data into fuzzy sets and radius of the clusters were tuned using a hybrid of particle swarm optimization and pattern search algorithms. Results of constructed model were compared to experimental data and excellent accordance was observed yielding an adjusted total correlation coefficient (R2) of 0.9847 and RMSE of 0.0003569. The model was also found to outperform a previously developed multiple regression model. Moreover, using the method of leverage value statistics, suspected outliers were diagnosed from the bulk of data. Results indicate that the model presented in this study can successfully be used for predicting saline water densities for Na-K-Mg-Ca-Li-Cl-Br-SO4-H2O system at temperatures ranging from − 20 to 200 °C, and pressures ranging from 1.013 to 684 bars with acceptable degree of accuracy.