Predicting pure and binary gas adsorption on activated carbon with two-dimensional cubic equations of state (2-D EOSs) and artificial neural network (ANN) method

ABSTRACT

In this study, four two-dimensional equations of state (2-D EOSs) including Redlich–Kwong (RK), Soave–Redlich–Kwong (SRK), Peng–Robinson (PR), and Modified Mohsennia–Modarress–Mansoori (M4) were applied to evaluate pure and binary gas adsorptions on activated carbons as the absorbent. The 2-D EOSs were derived from the respective cubic (or three-dimensional, 3-D) forms of the EOSs. The non-ideality of the equilibrium phases, the adsorbed phase and the gas phase, is described by using the above-mentioned 2-D EOS. A general adsorption isotherm equation along with a relation for predicting multicomponent gas adsorption is presented based on the generalised form of the 2-D EOSs. The parameters of 2-D EOS are determined from pure component adsorption data and are used to predict the binary adsorption parameters. The obtained results were compared with the available experiment data and it was concluded that the performance of M4 EOS ( and ) is better than the other three 2-D EOSs (RK, SRK, and PR). In continuation of this study by applying the ANN method, pure and binary gas adsorptions on activated carbon were predicted. The optimum architectures of ANN models for these predictions were 7:6:1 for pure and 9:7:1 for binary gas adsorptions. The calculated error in the results ( and ) indicated that the ANN model predicted the adsorption with better accuracy compared with the 2-D EOSs.

KEYWORDS:

• Adsorption isotherm
• Gibbs adsorption isotherm
• two-dimensional equation of state
• artificial neural network

Disclosure statement

No potential conflict of interest was reported by the authors.