Multiobjective optimization of foamed concrete for sustainable construction using statistical and artificial intelligence approaches

Abstract Foamed concrete is lightweight and insulating, but its mix design is a headache. Strength, durability, and carbon footprint all fight each other. This review looks at how researchers have tried to optimize these trade-offs using methods like RSM, Taguchi, ANN, and genetic algorithms. RSM is easy to interpret but struggles with nonlinear foam behavior. AI models handle nonlinearity better but need lots of data and tell you nothing about why they work. The best path forward seems to be hybrid approaches. On the environmental side, cement drives most of the carbon. Replacing it with SCMs can cut GWP by 20–50%. We put together a conceptual Pareto front showing the sweet spot: 900–1200 kg/m³ density, 8–15 MPa strength, and 350–500 kg CO₂-eq/m³. Big gaps remain. No standard testing protocols. Pore-scale physics is missing from models. Almost nobody has coupled LCA with real optimization. Field validation is rare. We end with a research roadmap: hybrid frameworks, open databases, and proper microstructural inputs.