From waste to resource: predictive modeling and environmental assessment of rubber–plastic substituted self-compacting concrete

This study investigates the combined use of Recycled Rubber (RR) and Recycled Plastic (RP) as hybrid sand replacements in Self-Compacting Concrete (SCC) and develops an integrated formulation and prediction framework based on Response Surface Methodology (RSM) and Artificial Neural Network (ANN) modeling, validated using k-fold cross-validation. Experimental results indicate that increasing RR content generally reduces workability at high substitution levels; however, moderate RR incorporation (10–20%) led to localized improvements in slump flow and passing ability under optimized mixture-design conditions. In contrast, RP substitution resulted in a milder and more stable reduction in workability across the investigated range. Segregation resistance was enhanced by RP incorporation, with sieve stability improvements of up to 7%, whereas RR addition caused a stability reduction of approximately 8–10%. Compressive Strength (CS) at 28 days decreased markedly with RR content, reaching reductions of 25–30% at 30% replacement, compared with a more moderate 10–15% decrease for RP, highlighting the distinct mechanical roles of the two waste materials. Life Cycle Assessment (LCA) revealed that both substitutions significantly mitigated environmental impacts, with reductions in Climate Change (CC) potential of up to 18%, non-renewable energy demand by 12%, and Human Toxicity (HT) impacts by approximately 15% at higher replacement levels under the adopted system expansion approach. Statistical analysis confirmed the robustness of predictive models (ANOVA, p < 0.01). Comparative modeling demonstrated that ANN consistently outperformed RSM, achieving higher predictive accuracy (R² up to 0.99 for fresh and mechanical properties and 0.98 for environmental indicators).