Strength prediction based on ultrasonic property of fractal gangue cemented rockfill reinforced by carbon nanotubes

Producing cemented rockfill for supplying building material and underground filling material is currently the most effective approach to recycle solid mineral waste. Investigating its ultrasonic property and constructing a model for predicting its strength are of great significance to its on-site assessment and engineering stability. Ultrasonic detection test was carried out on the fractal gangue cemented rockfill reinforced by carbon nanotubes (CNTs) to study the effects of the curing time, CNT dosage and aggregate size distribution on its ultrasonic property. The relation between ultrasonic pulse velocity (UPV) and uniaxial compressive strength (UCS) was established through uniaxial compression test, and the model I for predicting the strength of cemented rockfill was built based on this relationship. The coupling relationship between the curing time, CNT dosage, particle size distribution (PSD) fractal dimension of aggregates, UPV and the UCS of cemented rockfill was established by genetic algorithm (GA), and the strength prediction model II was constructed based on this coupling relationship. The difference between both prediction models and the matching degree between predictive strength and experimental data were discussed, a more accurate method for predicting the strength of cemented backfill was proposed. The results show that the relationship between CNT dosage and UPV can be described by the quadratic function, it believes that the optimal CNT dosage is between 0.05% and 0.10% for cemented rockfill. The relationship between PSD fractal dimension and UPV can be characterized by the quadratic function, it considers that the optimal PSD fractal dimension is in the range of 2.4150 and 2.6084 for the aggregates in cemented rockfill, and this optimal aggregate size distribution is easily affected by other conditions. The exponential function sigma(1c) = xi(10)e(xi 11v) - xi(10) can be used to describe the relationship between UPV and UCS, the prediction model I based on this relationship for predicting the strength of cemented rockfill has the determination coefficient of 0.9164 matching with the experimental data. The prediction model II based on the coupling relationship combining material parameters with UPV correction by using GA performs the reliable prediction accuracy with 4.5223% of mean absolute percentage error (MAPE) and 0.9513 of determination coefficient better than model I.

Automated summary

Investigating the ultrasonic properties of cemented rockfill and constructing predictive models based on genetic algorithm can accurately predict its strength. The optimal CNT dosage and aggregate size distribution can be determined through these models, improving the accuracy of strength prediction for cemented backfill.