Sustainable optimization of AISI 1045 steel heat treatment using PSO-ANN, TOPSIS, MOALO algorithm for energy efficiency and hardness

Abstract The study evaluates the interactive effects of quenching factors using different quenching media on hardness and on machinability energy usage of AISI 1045 steel. In conducting this research, Response Surface Methodology (RSM) Design of Experiments (DoE) was undertaken. The samples were subjected to various heat treatment conditions. When machined, hardness was measured as the primary post quench property and energy consumption was measured when dry cutting the specimens. Water quenching produced the highest surface hardness (approximately 56.7 HRC) but also the steepest hardness gradient towards the core and the highest normalized energy consumption during machining (0.55, dimensionless value after normalization). Cooling medium dominated: 76.56% H variance, 58.95% Ec variance; water max H/highest Ec; synthetic oil min H/lowest Ec. In total, a hybrid Particle Swarm Optimization–Artificial Neural Network (PSO–ANN) model was developed that predicted the responses with good accuracy; correlation coefficients were above 0.98. Multi-objective optimization using the Multi-Objective Ant Lion Optimizer (MOALO) was conducted to find the Pareto front of non-dominated solutions. TOPSIS selected the optimal balance: high hardness and low energy. Vegetable oil provided this balance. It offers high sustainability, good wear resistance, and machinability. The study provides evidence of a framework where experimental approaches, AI-based modeling, and bio-inspired optimization can be integrated to optimize quality of materials and energy efficiency in heat treatment.