ANN-based prediction of solid-vapor equilibria for N₂-CO₂ system relevant to cryogenic carbon capture

Cryogenic carbon capture (CCC) has been recognized as an efficient method to mitigate carbon dioxide emission to the atmosphere. The capture efficiency is dictated by the phase equilibria and interphase mass transfer rate. This paper focus on the desublimation-based CCC from a N₂-CO₂ mixture that is taken as a representative flue gas. The phase equilibria are conventionally predicted using equations of state (EoS), for which the computational time is generally large depending on the complexity of the EoS. Artificial neural network (ANN)- based models are often used to speed up the phase equilibrium predictions. Therefore, we have developed an ANN-based multilayer perceptron (MLP) model, to predict the solid- vapor equilibria (SVE) associated with the CCC by desublimation from a N₂-CO₂ mixture that is taken as a representative flue gas. We found that our ANN model not only reduced the computational time significantly but also gave more accurate SVE predictions than EoS.