Strength reduction of crumb rubber concrete: microstructures and mesoscale modelling

• A mesoscale model for strength simulation of crumb rubber concrete (CRC). • Nanoindentation tests on coarse aggregate-mortar, rubber-mortar interfacial transition zones (ITZ). • Parametric study on rubber content, particle size, shape, distribution, ITZ, mortar grade effect. • Rubber particle effect is equivalent to that of same content pores. • Theoretical prediction model for rubber-induced strength reduction rate of CRC. This paper presents a mesoscale model to investigate the rubber particle effect on the compressive strength of crumb rubber concrete (CRC) through considering varying rubber particle contents, particle sizes, distributions, and shapes. The mesoscale model was simulated through a five-phase internal structure composed of rubber particle, coarse aggregate, mortar, coarse aggregate-mortar interfacial transition zone (A-M ITZ), and rubber particle-mortar interfacial transition zone (R-M ITZ). The modulus of R-M ITZ and A-M ITZ were 41%-58% and 70%-85% of the mortar phase with corresponding thickness of 30-70µm and 30-60µm, respectively, as measured from nanoindentation tests. Mesoscale modelling was carried out on 100*200mm CRC cylinder samples with 6%-18% rubber percentage by sand volume and compared with the average results of three tested samples. The parametric analysis indicated that the rubber content was the main factor affecting the strength reduction of CRC. The effect of ITZ on the CRC strength was relatively low and negligible. Rubber particle size effect was evaluated through analysing CRC samples with three different rubber sizes, i.e., 1.77mm, 3mm, and 7mm. For each size, six CRC models containing different rubber particle distributions were generated. CRC models with the same rubber content presented similar average compressive strengths even though with rubber particles of varied sizes. However, larger rubber particle size increased the strength variation among CRC models. Further simulation showed that replacing the rubber particles in the CRC with the same content of pores did not change compressive strength or the damage pattern of the CRC samples. Finally, a prediction formula for strength reduction rate of CRC in terms of rubber content was developed. The numerical results were in good agreement with experimental tests in the current study and previous studies from literature.