On the Estimation of the Density of Brine with an Extensive Range of Different Salts Compositions and Concentrations

Abstract

Accurate determination of brine density is crucial for an efficient designing of various systems through where brine take an undeniable part. The importance of determining brine density is well corroborated by recent ongoing research surge in this scope. Despite existence of several experimental and empirical approaches, a reliable and robust model seems to be requisite for precise determination of brine density. In regard to high performance and great robustness of soft computing approaches for unraveling science and engineering problems, this article proposes LS-SVM and MLP-NN models to determine the brine density. Both models are developed over 1868 data points including both natural and synthetic brines. The proposed models take account an extensive range of input parameters such as temperature, pressure, and concentration. The developed models can significantly estimate the target values with respect to high values of R2, which are 0.999999 and 1.000000 for MLP-NN and LS-SVM models, respectively. Considering high accuracy and swiftness of proposed models, they can be great assets to science and engineering scopes.