Sugar industry waste produced geopolymer concrete and its compressive strength prediction via statistical analysis and artificial intelligence approach

摘要

AbstractConcrete innovations of the highest caliber have contributed in the development of cementless geopolymer concrete (geocon)that is sustainable and eco-friendly. Various by-products/wastes (such as fly ash (FA), ground granulated blast furnaceslag (GGBS)) generated by various industries (such as coal-fired power generating industry, iron industry) have come to lifein geo-con, appearing to be a hopeful strategy from an environmental standpoint. A lot of research has been done in order todevelop geo-cons from wastes, and one example may be found in sugarcane waste. With increasing demand for sugarcaneby-products, the wastages while processing these by products are also increasing. A major chunk of this wastage is sugarcane bagasse ash (SBA) which is produced by burning sugarcane bagasse in the factory to add up to the environmentalpollution. Keeping these aspects in mind, partial utilization of SBA along with FA and GGBS was considered in this study.Wherein FA and GGBS are well established precursors for the development of geo-con. This study investigated the effectsof SBA (0–15%) and GGBS (0–40%) addition on Fly ash-based geo-con (FA-GPC) as a substitute for FA at different curingtemperatures (60–90 °C) and sodium hydroxide/sodium silicate (NaOH/Na2SiO3) ratios (1:1–1:2.5). The results demonstratedthat the workability of resultant geo-con reduced as percentage of SBA and GGBS increased. Similar trend was alsoobserved with increasing NaOH/Na_2SiO_3 ratios. Furthermore, the compressive strength results reveal that FA to SBA ratiohas negative effect on compressive strength but combined effect of SBA and GGBS ratio along with FA has positive effecton compressive strength. Conduction of experiment for compressive strength is a drudgery process. To eliminate that, twomajor statistical methods, multiple linear regression analysis and polynomial regression analysis, with accuracy of + 5.05 to? 3.05% and + 4.94 to ? 3.40%, respectively, were developed. An artificial neural network model with an accuracy of + 0.44 to? 3.10% was also developed. Statistical parameters such as MSE and R_2 were used to further understand prediction accuracyand compare the models. The results show that the ANN model predicts compressive strength more accurately than othermodels. This study intends to describe a unique process for using sugarcane industrial waste in the form of SBA, with resultsdemonstrating good compressive strength for geo-con. Also, the developed models for compressive strength prediction canbe used effectively to reduce experimental drudgery.