Sustainable use of chemically modified tyre rubber in concrete: Machine learning based novel predictive model

To encourage the consumption of crumb rubber (CR), gene expression programming (GEP) has been exercised in this paper to establish empirical models for estimation of mechanical properties of concrete made with NaOH treated CR. An extensive and reliable database of compressive strength of concrete made with NaOH treated CR is established through a comprehensive literature review. Literature review showed that compressive strength of NaOH treated CR concrete is affected by percentage of CR used as a replacement of sand (RS%), concentration of NaOH solution (NC in %), period of NaOH pre-treatment (PTP in hours), water to cement ratio (W/C), quantity of sand (S in kg/m(3)) and quantity of superplasticizer (SP in kg/m(3)). The performance of the established model is evaluated by doing parametric analysis, applying statistical checks and comparing with regression models. The R-values in the testing phase of GEP, linear and non-linear regression (LR and NLR) equations are 0.90 and 0.77 each respectively. Furthermore, objective function (OF) of GEP model is 25%, and 33% better than LR and NLR model. Thus, results reflected that the proposed GEP model is more accurate and possess a high generalization and prediction capability than LR and NLR equations with resolved overfitting issue. The results of this research can boost the re-usage of CR for expansion of green concrete leading to environmental safety and economic advantages.

Automated summary

In this paper, gene expression programming (GEP) was used to establish empirical models for estimating the mechanical properties of concrete made with NaOH treated crumb rubber (CR). Through a comprehensive literature review, a reliable database of compressive strength of concrete made with NaOH treated CR was established. The performance of the GEP model was evaluated and compared with linear and non-linear regression models, showing that the GEP model was more accurate and had better generalization and prediction capabilities. The results of this research can promote the reuse of CR for green concrete, leading to environmental safety and economic advantages.