Evolving simple-to-apply models for estimating thermal conductivity of supercritical CO₂

ABSTRACT

Today, due to extensive applications of supercritical fluids technology in various chemical engineering process and industrial fields, predicting thermal conductivity of supercritical carbon dioxide is vital. In this research, two simple-to-apply models have been developed to estimate thermal conductivity of supercritical CO₂ as a function of temperature, pressure and density over broad ranges. This research presents a predictive tool based on LSSVM to predict thermal conductivity of supercritical CO₂. Genetic algorithm is employed to determine hyper-variables which are included in the LSSVM approach. In this regard, a set of accessible data containing 745 data points has been gathered from the previous published papers. Estimations are found to be in excellent agreement with reported data. Moreover, statistical analyses have been applied to evaluate the performance of two models. The obtained values of Mean Squared Error and R-Square were 7.415866, 0.9935 and 0.046527, 1.00 for the correlation and LSSVM model, respectively. The developed tools can be of immense practical value for chemical engineers to have a quick check of thermal conductivity of supercritical CO₂ at an extensive range of conditions.

KEYWORDS:

• Supercritical carbon dioxide
• correlation
• least square support vector machine
• genetic algorithm
• predictive tool