Abstract—Self-compacting concrete is a new construction material and its mechanical properties are not completely understood in the literature. Compressive strength is representative of mechanical properties of hardened concrete; hence its prediction at the fresh stage can improve the final performance of structure. Sensitivity of compressive strength to the changes in mixture proportions, curing and environmental conditions together with the heterogeneous nature of the concrete complicates the problem. Considering the capabilities of artificial intelligent systems to discover any consistency between huge amounts of complex data, this study utilizes the ANFIS models to predict the compressive strength of Self-compacting concrete from mixture proportions and slump flow values. The empirical data from previously conducted 55 experiments have been implemented in 18 distinct models in ANFIS. The model including all input data (mixture components and slump flow) gives the best prediction. However eliminating the maximum size and volume of the aggregate from the input data results the least accurate ANFIS model. Any changes in the powder volume, paste content and slump flow also similarly affect the predictedvalue. Particular effect of each input data such as the interaction between the powder volume and the compressive strength are also investigated and compared with the basic concepts of concrete technology. 
 

 

Index Terms—ANFIS, self-compacting concrete, compressive strength, mixture proportion, slump flow, sensitivity, concrete technology

Cite: Behnam Vakhshouri and Shami Nejadi, "Sensitivity of Compressive Strength of Self-Compacting Concrete to Mixture Proportions and Slump Flow in ANFIS Models," International Journal of Structural and Civil Engineering Research, Vol. 6, No. 1, pp. 57-64, February 2017. doi: 10.18178/ijscer.6.1.57-64