Development of an automated prototype for biodiesel production applying production improvements with genetic algorithms

Biofuels are crucial in the decarbonisation of cities and economies and are especially sustainable if they are generated from food waste. Despite their importance, the technology for their production poses challenges, and their use in the domestic environment is rarely pursued. This study explores the production of biodiesel from used vegetable oils through an automated small-scale prototype with operational improvements using genetic algorithms (GA). The structure was vertically arranged, with control and response devices interfaced via an Arduino UNO R3 board. The transesterification process was programmed in C++ using free Arduino software. Eleven trials were conducted using a rotational composite central response surface design, including three replicates at the centre point and 1.5 L of waste oil per trial. The maximum biodiesel yield was 33.42 %, achieved with 8.79 g of potassium hydroxide (KOH) and 620.05 ml of ethanol, applying a GA population of 500 individuals over 100 generations with crossover and mutation probabilities of 0.8 and 0.1, respectively. The biodiesel was characterised and met ASTM D6751 and EN 14214 standards. The prototype, tested under low loads, demonstrated sufficient performance. The learning algorithms optimised composition, improving both performance and physicochemical properties of the biodiesel. • Developed an automated prototype for small-scale biodiesel from waste cooking oil via transesterification. • Arduino-based automation integrated sensors and feedback to control temperature, mixing and timing. • Conducted 11 experiments using central composite design to optimise biodiesel process variables. • Achieved 33% higher ethanolysis efficiency using genetic algorithms with 500 individuals, 100 gens. • Genetic algorithm improved repeatability by refining reactant ratios and enhancing process stability.