Keywords: levulinic acid production, response surface methodology, RSM, artificial neural networks, ANNs, biomaterials, glucose, empty fruit bunch, kenaf, lignocellulose biomass, hydrolysis
Comparison of response surface methodology and artificial neural network for optimum levulinic acid production from glucose, empty fruit bunch and kenaf
Levulinic acid (LA) is one of the versatile chemicals that can be produced from lignocellulosic biomass. In this study, response surface methodology (RSM) and artificial neural network (ANN) were applied to optimise LA yield from glucose, empty fruit bunch (EFB) and kenaf. The effect of process variables namely reaction temperature, reaction time and acid concentration on LA production was investigated. Application of ANN for modelling technique was more reliable since better data fitting and prediction capabilities were obtained. Next, both biomass were pre–treated with 1–ethyl–3–methylimidazolium chloride ionic liquid, [EMIM][Cl] to degrade the biomass and to observe its effect on the hydrolysis process. Thermal gravimetric analysis results revealed kenaf degraded to cellulose and hemicellulose more easily than EFB. Higher LA yield was obtained from both pre–treated samples, but the yield from kenaf (39.5 wt%) was more prevalent than EFB (31.6 wt%).