Optimization of operational conditions for adipate ester synthesis in a stirred tank reactor

摘要: Esterification of adipic acid and oleyl alcohol in a solvent-free system featuring stirred tank reactor containing commercially immobilized Candida antarctica lipase B was performed. The process carried out using an artificial neural network (ANN) trained by the Levenberg-Marquardt (LM) algorithm. effects four operative variables, temperature, time, amount enzyme, impeller speed, on reaction yield were studied. By examining different ANN configurations, best found to consist seven hidden nodes hyperbolic tangent sigmoid transfer function. values coefficient determination (R2) root mean squared error (RMSE) between actual predicted responses determined be 1 0.0058178 for training 0.99467 0.622540 testing datasets, respectively. These results imply that developed model capable predicting esterification yield. variables affected yield, their order contribution as follows: time > enzyme temperature speed. A high percentage (95.7%) obtained low level (2.5% w/w), speed 66.5°C, 354 min (about 6 h), 500 rpm, simple protocol efficient substrate conversion evidenced stability is indicative successful ester synthesis.