Machine-learning (ANN) and regression-based response surface models were compared for modelling and optimising Lactobacillus paracasei fermentation of black soldier fly farm waste for chitin production. ANN combined with multi-objective genetic optimisation (MOGA) achieved higher predictive accuracy for both deproteinisation (DP%) and demineralisation (DD%) and revealed substantially different parameter effects compared with classical RSM.
A catalyst-free in-situ subcritical ethanol transesterification process was developed and optimised for biodiesel (FAEE) production from insect-derived waste, achieving high fuel yield and compliant fuel properties, and was conceptually integrated with anaerobic digestion and techno-economic assessment for circular bioenergy recovery
This study evaluates three fish waste valorisation pathways: enzymatic hydrolysis, subcritical water processing and anaerobic digestion. Aspen Plus simulations using NRTL thermodynamics were applied to model mass and energy balances. Products include protein hydrolysates, fatty acids and biogas. Capital and operating costs were estimated with scaling factors and cost indices. Economic performance was assessed using NPV, payback period and return indicators. Sensitivity and risk analyses evaluated uncertainty. Results show all routes are feasible and robust
