Central Composite Design and Artificial Neural Network for the Evaluation and Optimization of an Electrochemical HHO Gas Generator

Central composite experimental design (CCD) and artificial neural networks (ANN) is not so popular in electrochemical systems, even though they present several advantages (lower number of experiments, evaluation of variable interactions, and correct analysis of overlapping data). In the present work, CCD and ANN were used for the analysis of an electrochemical cell with dimensionally stable electrodes modified with IrO₂ and an alkaline electrolyte solution, for the evaluation and optimization of an electrochemical HHO gas generator. With the CCD it was possible to maximize HHO gas generation (64.85 g/LNaOH and 10.2 amp) with a low standard deviation (0.004), a R-sq of 98.02%, and a R-sq (pred) of 92.04%, demonstrated that the experimental data fitted properly with the theoretical model. An ANN was trained from a large data set, obtaining recovery values close to 100%, R values close to 1 for the training (0.9894) and test stages (0.9677), and a good fit according to the Chi-squared test for the predictive values in both stages; with this ANN it was possible to solve overlapping issues in the HHO gas production system. The results indicate that CCD and ANN are viable alternatives to increase the performance of an electrochemical HHO gas generator.