Solubility measurements of anthraquione disperse dyestuffs in supercritical carbon dioxide and neural network modeling.

• Solubility of Anthraquione Disperse Dyestuffs SC-CO 2 was measured. • Temperature ranges of (343.15–373.15) K and pressures of (14.0–22.0) MPa. • Correlated and predicted by neural network model with input, hidden, and output layers. • Akaike's Information Criteria (AIC) was used to determine a minimum number of parameters in the neural network model. • Good agreement between the experimental and calculated values was obtained by the neural network model. The solubility of Disperse Violet 4, Disperse Violet 27, and Disperse Blue 72 were measured in supercritical carbon dioxide (SC-CO 2) over the temperature ranges of 343.15 K to 373.15 K and pressures of 14 MPa to 22 MPa. Also, the solubility of Disperse Red 11 in SC-CO 2 was obtained at the temperature of 343.15 K and pressure ranges of 14 MPa to 22 MPa and at the pressure of 14 MPa and temperatures of 343.15 K to 373.15 K, for an extension of the solubility results previously published. The neural network ANN formulated the experimental data in terms of temperature and pressure as input layers, solubility as the output layer, and the number of neurons in the hidden layer. The calculation results show that five different kinds of dyestuff solubility data at the temperature ranges of 343.15 K to 373.15 K, pressures of 14 MPa to 22 MPa as well as the solubility data of Disperse Red 4 and Disperse Red 15 taken from the previous work could be estimated accurately by BP neural network with a fitting level of 0.999. Moreover, the solubility data were correlated with eight different kinds of empirical models proposed previously and compared with those calculated by the ANN. It was found that the ANN model could represent the solubility of anthraquinone dyestuffs in SC-CO 2 better than the empirical models. [ABSTRACT FROM AUTHOR]