Prediction of compressive strength of self-compacting concrete by ANFIS models

Publication Type:
Journal Article
Neurocomputing, 2018, 280 pp. 13 - 22
Issue Date:
Filename Description Size
1-s2.0-S0925231217317654-main.pdfPublished Version2.6 MB
Adobe PDF
Full metadata record
© 2017 Elsevier Ltd Many studies predict the compressive strength of conventional concrete from hardened characteristics; however, in the case of self-compacting concrete, these investigations are very rare. There is no study to predict the compressive strength of self-compacting concrete from mixture proportions and slump flow. This paper designs ANFIS models to establish relationship between the compressive strength as output, and slump flow and mixture proportions as input in eighteen combinations of input parameters. The applied dada is taken from 55 previously conducted experimental studies. Effect of each parameter on the compressive strength and its importance level in the developed model has been investigated. Based on the error size in each combination analysis, weighting factor and importance level of each parameter is evaluated to apply the correction factors to get the most optimized relationship. Obtained results indicate that the model including all input data (slump flow and mixture proportions) gives the best prediction of the compressive strength. Excluding the slump flow from combinations affects the prediction of compressive strength, considerably. However it's not as much as the effect of the maximum aggregate size and aggregate volume in the mixture design. In addition, different values of powder volume, aggregate volume and paste content in the mixture reveal different ascending and descending effects on the compressive strength.
Please use this identifier to cite or link to this item: