Predicting and optimizing electrochemical degradation of mezcal vinasse with statistical and neural network models
• First ANN-GA model for mezcal vinasse electro-degradation. • 95% COD removal is achieved in 8h, meeting regulatory standards. • Cost is reduced by 12% vs conventional RSM optimization. • Process validation proves reduced eco-toxicity and strong profitability. • Strong economic viability with 13.11% ROI and 1.4-year payback. This study investigates the electrochemical treatment of mezcal vinasse, comparing two modeling approaches to maximize the removal of Chemical Oxygen Demand (COD) and Melanoidins (Mel) while minimizing Specific Energy Consumption (SEC). A Central Composite Design (CCD) with Response Surface Methodology (RSM) and an Artificial Neural Network (ANN) coupled with a Genetic Algorithm (GA) were used to optimize initial pH, current density, and cathode material. The RSM model identified optimal conditions (pH₀ = 6.19, j = 210 mA cm⁻², Ti cathode), achieving 92.44% COD removal, 97.63% Mel degradation, and an SEC of 0.192 kWh (g COD)⁻¹ in 10 hours. The ANN-GA framework, however, proved superior, determining parameters (pH₀ = 6.09, j = 236.2 mA cm⁻², Ti cathode) that yielded 94.76% COD removal, 97.75% Mel degradation, and an SEC of 0.194 kWh (g COD)⁻¹ in only 8 hours. Both methods successfully reduced COD to meet Mexican regulatory limits. The ANN-GA strategy demonstrated a clear advantage, accelerating the process and lowering operational costs (OPEX: USD 1.02 vs. USD 1.16), underscoring its potential for sustainable vinasse management. Through Life Cycle Analysis (LCA) methodology, the global warming potential of the process was 6.53 kgCO 2 eq, which can be reduced by sceneries combining renewable sources of energy and fossil fuels.
Read more