Prediction of Compressive Strength of Fly Ash Concretes using Soft Computing Techniques

Abstract

The concrete compressive strength is one of the crucial parameter with regards to the sustainability of any concrete structure. Emphasizing on cost effectiveness and environmental benefits, it is a well-established practice of transformation of industrial wastes to alternate building materials enhancing the durability properties of concrete at later ages. Fly ash is one such abundantly available pozzolanic material from coal burnt combustion plants which redeems the use of cement for similar strength offsetting the consumption of natural raw materials for cement manufacture. The laboratory procedure of the compressive strength (CS) becomes more tedious and complicated with the usage of higher volumes of mineral admixtures and appropriate chemical admixtures to obtain the required strength grade. The design of mix proportions becomes complicated and the traditional prediction methods fail to perform for new test data with larger number of input parameters. Researchers have recommended for decades many conventional computing methods for solving actual problems which are weighed down due to time consumption and being labor intensive. In trial with human brain attributes, relative computer based soft computing techniques have lately drawn attention to deal with the given actual problem. Soft computing approach employs the irrelevant and ambiguous feature of the problem to yield an approximate solution in comparison to conventional computing techniques which are time consuming and rely on exact solutions. newlineIn the present study, experimental data pertaining to partial replacement of cement with varying percentages of fly ash for concrete manufacture is collected from literature. The approach of the study is predominantly modeling of the fly ash concrete CS using soft computing techniques. From literature, it is evident that there are limited researches carried out in predicting the CS of fly ash concretes with fly ash in large volumes using soft computing techniques (SCT). Various SCTs such as artificial neural network